Review

Machine Learning Techniques in Structural Wind Engineering:

A State-of-the-Art Review
Karim Mostafa 1,*, Ioannis Zisis 1 and Mohamed A. Moustafa 2

1 CEE, College of Engineering and Computing, Florida International University, Miami, FL 33199, USA;
[email protected]
2 CEE, College of Engineering, University of Nevada, Reno, NV 89557, USA; [email protected]

* Correspondence: [email protected]

Abstract: Machine learning (ML) techniques, which are a subset of artificial intelligence (AI), have
played a crucial role across a wide spectrum of disciplines, including engineering, over the last dec-
ades. The promise of using ML is due to its ability to learn from given data, identify patterns, and
accordingly make decisions or predictions without being specifically programmed to do so. This
paper provides a comprehensive state-of-the-art review of the implementation of ML techniques in
the structural wind engineering domain and presents the most promising methods and applications
in this field, such as regression trees, random forest, neural networks, etc. The existing literature
was reviewed and categorized into three main traits: (1) prediction of wind-induced pressure/ve-
locities on different structures using data from experimental studies, (2) integration of computa-
tional fluid dynamics (CFD) models with ML models for wind load prediction, and (3) assessment
of the aeroelastic response of structures, such as buildings and bridges, using ML. Overall, the re-
view identified that some of the examined studies show satisfactory and promising results in pre-
dicting wind load and aeroelastic responses while others showed less conservative results com-
pared to the experimental data. The review demonstrates that the artificial neural network (ANN)
Citation: Mostafa, K.; Zisis, I.; is the most powerful tool that is widely used in wind engineering applications, but the paper still
Moustafa, M.A. Machine Learning identifies other powerful ML models as well for prospective operations and future research.
Techniques in Structural Wind
Engineering: A State-of-the-Art
Keywords: machine learning; neural networks; wind engineering; wind-induced pressure;
Review. Appl. Sci. 2022, 12, 5232.
aeroelastic response; computational fluid dynamics
https://doi.org/10.3390/app12105232

Academic Editors: Tianyou Tao,

Yong Chen and Haiwei Xu

Received: 2 April 2022 1. Introduction

Accepted: 19 May 2022 Artificial intelligence (AI) has evolved rapidly since its realization in the 1956 Dart-
Published: 22 May 2022 mouth Summer workshop and has attracted significant attention from academicians in
Publisher’s Note: MDPI stays neu-
different fields of research [1]. Machine learning (ML), which is a form and subset of AI,
tral with regard to jurisdictional is used widely in many applications in the area of engineering, business, and science [2].
claims in published maps and institu- ML algorithms are capable of learning and detecting patterns and then self-improve their
tional affiliations. performance to better complete the assigned tasks. In addition, they offer a vantage for
handling more complex approach problems, ensuring computational efficiency, dealing
with uncertainties, and facilitating predictions with minimal human interference [3].
Meanwhile, the ML capabilities in performing complex applications with large-scale and
Copyright: © 2022 by the authors. Li- high-dimensional nonlinear data have been enhanced over the years due to the expansion
censee MDPI, Basel, Switzerland. of computational capabilities and power [4].
This article is an open access article There are four main types of learning for ML algorithms: supervised learning, unsu-
distributed under the terms and con- pervised learning, semi-supervised learning, and reinforcement learning [5,6]. In super-
ditions of the Creative Commons At-
vised learning, the computer is trained with a labeled set of data to develop predictive
tribution (CC BY) license (https://cre-
models through a relationship between the input and the labeled data (i.e., regression and
ativecommons.org/licenses/by/4.0/).
classification). In unsupervised learning, which is more complex, the computer is trained

Appl. Sci. 2022, 12, 5232. https://doi.org/10.3390/app12105232 www.mdpi.com/journal/applsci

Appl. Sci. 2022, 12, 5232 2 of 28

with an unlabeled set of data to derive the structure present in the data by extracting gen-
eral rules (i.e., clustering and dimensionality reduction). In semi-supervised learning, the
computer is trained with a mixture of labeled and unlabeled sets. In reinforcement learn-
ing, which is so far the least common learning type, the computer acquires knowledge by
observing the data through some iterations that require reinforcement signals to identify
the predictive behavior or action (i.e., make decisions) [3,7].
ML is becoming more prevalent in civil engineering with numerous studies publish-
ing reviews and applications of ML in this field. While this paper focuses only on struc-
tural wind applications as explained later, a few key general summary studies or reviews
are listed first for the convenience of the readers interested in broader applications. Adeli
in [8] reviewed the applications of artificial neural networks (ANN) in the fields of struc-
tural engineering and construction management. The study presented the integration of
neural networks with different computing paradigms (i.e., fuzzy logic, genetic algorithm,
etc.). Çevik et al. [9] reviewed different studies on the support vector machine (SVM)
method in structural engineering and studied the feasibility of this approach by providing
three case studies. Similarly, Dibike et al. [10] investigated the usability of SVM for classi-
fication and regression problems using data for horizontal force initiated by dynamic
waves on a vertical structure. Recently, Sun et al. [4] presented a review of historical and
recent developments of ML applications in the area of structural design and performance
assessment for buildings.
More recently, ML applications have been involved in predicting catastrophic natu-
ral hazards. Recent studies investigated the integration of real-time hybrid simulation
(RTHS) with deep learning (DL) algorithms to represent the dynamic behavior of nonlin-
ear analytical substructures [11,12]. A comprehensive review was also provided by Xie et
al. [13] on the progress and challenges of ML applications in the field of earthquake engi-
neering including seismic hazard analysis and seismic fragility. Mosavi et al. [14] demon-
strated state-of-the-art ML methods for flood prediction and the most promising methods
to predict long- and short-term floods. Likewise, Munawar et al. [15] presented a novel
approach for detecting the areas that are affected by flooding through the integration of
ML and image processing. Moreover, ML applications were implemented in many other
fields related to civil engineering generally and structural engineering particularly [16–
25], structural damage detection [26–29], structural health monitoring [30–33], and ge-
otechnical engineering [34–39]. In addition, ML techniques, such as Gaussian regression,
can be used for numerical weather predictions [40]. Taking into consideration the above
efforts to summarize ML techniques and their applications for different civil engineering
sub-disciplines, no previous studies focused on structural wind engineering. Thus, the
objective of this paper is to fill this important knowledge gap by providing a thorough
and comprehensive review of ML techniques and implementations in structural wind en-
gineering.
To better relate ML implementations, a brief overview of typical structural wind en-
gineering problems is provided first. Bluff body aerodynamics is associated with a high
level of complexity due to the several ways that wind flow interacts with civil engineering
structures. Wind flow at the bottom of the atmosphere is influenced by the roughness of
the natural terrain as well as by the built environment itself. As a result, eddies are formed
that vary in size and shape and travel with the wind creating the well-known atmospheric
boundary layer (ABL) flow characteristics [41]. Studying and understanding the behavior
of wind and its interaction with buildings and other structures is critical in the analysis
and design process. Generally, ABL wind tunnel testing is still the most reliable tool to
assess the aerodynamics of any structure and provide an accurate surface pressure and/or
aeroelastic response. Computational fluid dynamics (CFD) tools became more popular
and can perform well in predicting mostly mean, and in some cases peak, wind flow char-
acteristics and corresponding loads on structures. To address larger problems, ML tech-
niques were recently introduced in different applications in wind engineering but mostly
to support and expand experimental and numerical wind engineering studies.
Appl. Sci. 2022, 12, 5232 3 of 28

Based on the above introduction and the witnessed increased interest to incorporate
ML techniques in structural wind engineering, a state-of-the-art review of the existing lit-
erature is beneficial and timely, which motivates this study. The goal of this paper again
is to present an overview of the state of knowledge for commonly used ML methods in
structural wind engineering as well as try to identify prospective research domains. We
focus on the different ML methods that were used mainly for predicting wind-induced
loads or aeroelastic responses. Therefore, eight major ML methods that were commonly
used in the previous studies are the core of this review. These are: (1) artificial neural
networks (ANN), (2) decision tree regression (DT), (3) ensemble methods (EM) that in-
clude: random forest (RF), gradient boosting regression tree (GBRT) or alternatively re-
ferred to as gradient boosting decision tree (GBDT), and XGboost, (4) fuzzy neural net-
works (FNN), (5) the gaussian regression process (GRP), (6) generative adversarial net-
works (GAN), (7) k-nearest neighbor regression (KNN), and (8) support vector regression
(SVR).
The review and discussion following this introduction are divided into four sections.
The first section goes over the different ML methods that were previously used through
an overview of the formulation and the theoretical background for each method. This is
to provide a fair context before discussing their applications for prediction and classifica-
tion purposes. The second section is the core of this paper, which focuses on reviewing
the previous studies that are categorized and presented through three main applications:
(1) the prediction of wind-induced pressure/speed on different structures using data from
experimental models, (2) integration of CFD models with ML models for wind loads pre-
diction, and (3) assessment of the aeroelastic responses for two major types of structures,
i.e., buildings and bridges. The third section provides a summary of the ML assessment
tools and error estimation metrics based on the reviewed studies. The provided summary
includes a list of assessment equations that are provided for the convenience of future
researchers. The last section provides an overall comparison of the methods and recom-
mendations to pave the path for using ML techniques in addressing future challenges and
prospective research opportunities in wind engineering. It is important to note that this
study did not review the ML implementation in non-structural wind applications such as
wind turbines wake modeling, condition monitoring, blade fault detection, etc.

2. ML Methods Used in Structural Wind Engineering

This section discusses a brief theoretical background and an overview formulation
for the commonly used ML methods in structural wind engineering. The discussion in-
cludes the eight classes that are mentioned before: ANN, FNN, DT, EM, GPR, GAN, KNN
and SVM. It is noted that ANN methods are found to be the most commonly used meth-
ods in the area of focus, therefore, ANN is discussed in this section in more detail com-
pared to the other methods.

2.1. Artificial Neural Network (ANN)

The concept of ANN is derived from biological sciences, where it mimics the com-
plexity of the human brain in recognizing patterns through biological neurons, and thus
imitates the process of thinking, recognizing, making decisions, and solving problems
[42,43]. ANN was the most popular method found in the reviewed literature to predict
wind-induced pressures compared to other neural network methods (e.g., CNN or RNN).
ANN is robust enough to solve multivariate and nonlinear modeling problems, such as
classification and predictions. ANN is a group of layers that comprise multiple neurons
at each layer and is also known as a feed-forward neural network (FFNN). It is composed
of input layers, where all the variables are defined and fed into the hidden layers which
are weighted and fed into the output layers that represent the response of the operation.
The ANN architecture could be written as x-h-h-y which defines x number of inputs (var-
Appl. Sci. 2022, 12, 5232 4 of 28

iables), h number of hidden layers, and y number of outputs (responses) as shown in Fig-
ure 1. Each hidden layer comprises a certain number of neurons that gives a robust model,
and this could be achieved by training and trials.

Figure 1. Feed-forward neural network architecture.

The hidden layers are composed of activation functions that apply different weights
to the input layer and transfer them to the output layers. The most common activation
functions are the nonlinear continuous sigmoid, the tangent sigmoid, and the logarithmic
sigmoid [44]. The weights are multiplied with the inputs and calibrated through a training
process between the input and output layers to reduce the loss. The training process is
applied using the Levenberg–Marquardt backpropagation algorithm, which belongs to
the family of the Multi-Layer Perceptron (MLP) network [45] and was originally proposed
by Rumelhart et al. [46]. It consists of two steps: feed-forward the values to calculate the
error, and then propagate back the error to previous layers [47,48]. The repeated iteration
process (epochs) of backpropagation network error continues and it keeps adjusting the
interconnecting weights until the network error is reduced to an acceptable level. Once
the most accurate solution is formed during the training process, the weights and biases
are fixed, and the training process stops. The Levenberg–Marquardt is a standard numer-
ical method which achieves the second-order training speed with no need to compute the
Hessian matrix and was demonstrated to be efficient with training networks up to a few
hundred weights [47,49]. Figure 2 shows the output signal for a generic neuron j in the
hidden layer h defined in Equation (1), where 𝑤 is the weight that connects the ith neu-
ron of the current layer to the jth neuron of the following layer, xi is the input variable, b
is the bias associated with the jth neuron to adjust the output along with the weighted
sum, and f is the activation function that is usually adapted as either a tangent sigmoid or
a logarithmic sigmoid, Equations (2) and (3), respectively. The (RBF-NN) that was used
first by [50] is a function whose response either decreases or increases with the distance
from a center point [51,52].

𝑦 =𝑓( 𝑤 𝑥 +𝑏 ) (1)

2
𝑓 (𝑢) = −1 + (2)
(1 + 𝑒 )
1
𝑓 (𝑢) = (3)
(1 + 𝑒 )
Appl. Sci. 2022, 12, 5232 5 of 28

Figure 2. The generic model of neuron j in hidden layer h.

During the training process of BPNN, usually, the training is terminated when one
of the following criteria is first met: (i) fixing the number of epochs to a certain number,
(ii) the training error is less than a specific training goal, or (iii) the magnitude of the train-
ing gradient is less than a specified small value (i.e., 1.0 × 10−10). The training error is the
error obtained from running the trained model back on the data used in the training pro-
cess, while the training gradient is the error calculated as a direction and magnitude dur-
ing the training of the network that is used to update the network weight in the right
direction and amount.

2.2. Fuzzy Neural Network (FNN)

The FNN approach combines the capability of neural networks with fuzzy logic rea-
soning attributes [53,54]. The architecture of FNN is composed of an input layer, a mem-
bership layer, an inference layer, and an output layer (defuzzification layer), as shown in
Figure 3. The membership and inference layers replace the hidden layers in the ANN. The
input layer consists of n number of variables and the inference layer is composed of m
number of rules, and accordingly n × m numbers of neurons exist in the membership layer.
The activation function adopted in the membership layer is a Gaussian function as shown
in Equation (4) and illustrated in Figure 3.
( )
𝑢 = exp(− ), 1≤ i ≤ n, 1 ≤ j ≤ m (4)

where uij is the value of the membership function of the ith input corresponding to the jth
rule, mij and σij are the mean and the standard deviation of the Gaussian function.
Appl. Sci. 2022, 12, 5232 6 of 28

Figure 3. The architecture of the four-layer fuzzy neural network.

2.3. Decision Tree (DT)

The DT method is one of the supervised ML models where the algorithm assigns the
output through testing in a tree of nodes and by filtering the nodes (decision nodes) down
within the split sub-nodes (leaf nodes) to reach the final output. The decision trees may
differ in several dimensions such as the test might be multivariate or univariate, or the
test may have two or more outcomes, and the attributes might be numeric or categorical
[55–57].

2.4. Ensemble Methods (EM)

The EM methods include: (1) bagging regression tree that is also referred to as the
random forest (RF) algorithm, (2) gradient boosting regression tree (GBRT) or decision
tree (GBDT), and (3) extreme gradient boosting (XGB). All EM methods could be defined
as a combination of different decision trees to overcome the weakness that may occur in a
single tree such as sensitivity to training data and unstableness [58]. The forest generated
by the RF algorithm is either trained through bagging, which was proposed by Breiman
[59], or through bootstrap aggregating [60]. RF splits in each node n features among the
total m features, where n is recommended to be or √2𝑚 [61]. It reduces the overfitting
of datasets and increases precision. Overfitting is overtraining the model which causes it
to be particular to certain datasets and lose the generalized aspect desired in ML models.
The DR and RF methods are commonly used in classification and regression problems.
GBRT, also known as GBDT as mentioned above, was first developed by Friedman
[62] and is one of the most powerful ML techniques deemed successful in a broad range
of applications [63,64]. GBDT combines a set of weak learners called classification and
regression tree (CART). To eliminate overfitting, each regression tree is scaled by a factor,
called learning rate (Lr) which represents the contribution of each tree to the predicted
values for the final model. The predicted values are computed as the sum of all trees mul-
tiplied by the learning rate [65]. Lr with maximum tree depth (Td) determines the number
of regression trees for building the model [66]. Previous studies proved that smaller Lr
decreases the test error but increases computational time [63,64,67]. A subsampling pro-
cedure was introduced by Friedman [60] to improve the generation capability of the
model using subsampling fraction (Fs) that is chosen randomly from the full date set to fit
the base learner.
Another popular method from the EM family is the XGBoost, or XGB as defined
above, which is similar to the random forest and was developed by Chen and Guestrin
[68]. XGB has more enhancement compared to other ensemble methods. It can penalize
more complex models by using both LASSO (L1) and Ridge (L2) regularization to avoid
Appl. Sci. 2022, 12, 5232 7 of 28

overfitting. It handles different types of sparsity patterns in the data, and it uses the dis-
tributed weighted quantile sketch algorithm to find split points among weighted datasets.
There is no need to specify in every single run the exact number of iterations as the algo-
rithm has built-in cross-validation that takes care of this task.

2.5. Gaussian Process Regression (GPR)

The GPR is a supervised learning model that combines two processes: (1) prior pro-
cess, where the random variables are collected, and (2) posteriori process, where the re-
sults are interpolated. This method was introduced by Rasmussen [69] and developed on
the basis of statistical and Bayesian theory. GPR has a stronger generalization ability, self-
calculates the hyper-parameters in GPR, and the outputs have clear probabilistic meaning
[70]. These advantages make the GPR preferable compared to BPNN, as it could handle
complex regression problems with high dimensions and a small sample size [69,71]. Back-
ground theory and informative equations can be found in detail in the literature [69,70].

2.6. Generative Adversarial Networks (GAN)

The GAN technique was proposed by Goodfellow et al. [72], which is based on a
game theory of a minimax two players game. The GAN has attracted worldwide interest
in terms of generative modeling tasks. The purpose of this approach is to estimate the
generative models via an adversarial process. The approach is achieved by training two
models; first, a generative model G that capture all the distribution in the data, and sec-
ond, a discriminative model D that estimates the probability of a sample to come from the
training data rather than G. The G model defines the p model (x) and draws samples from
the distribution of p model. The input is placed as vector z and the model is defined by a
prior distribution p(z) over the vector z as a generator function G(z:θ(G)), where θ(G) is a
set of learnable parameters that define the generator’s strategy in the game [73]. More
details about the GAN models can be found in [72,73].

2.7. K-Nearest Neighbors (KNN)

The KNN algorithm is a supervised non-parametric classification machine learning
algorithm that was developed by Fix and Hodges [74]. The KNN does not perform any
training or assumption for storing the data, but it assigns the unseen data to the nearest
set of data used in the training process. According to the value of K, the algorithm started
to determine the class for the point to be assigned to according to the value K. For instance,
if K is 1, the unseen point will be assigned to a certain class according to the class of the
nearest point, or to the nearest five points in the case of K is 5, etc. The KNN is one of the
simplest ML classification algorithms and more details can be found in [75].

2.8. Support Vector Machine (SVM)

The SVM is a supervised learning method used for the purpose of classification and
regression that use kernel functions. The SVM algorithm is based on determining a hy-
perplane in an N-dimensional space depending on the number of features that classify the
dataset. The optimum hyperplane for classification purposes is associated with the maxi-
mum margin between the support vectors which are composed of the dataset nearest to
that hyperplane [76]. SVM was developed by Vapnik [77] and is considered to be one of
the most simple and robust classification algorithms. More details about SVM can be
found in [78].

3. Prior Studies on Applying ML Techniques in Structural Wind Engineering

A broad range of studies is summarized in this section based on the three categories
mentioned before, i.e., (1) prediction of wind-induced pressure/speed on different struc-
tures using data from experimental models, (2) integration of CFD models with ML mod-
els for wind loads prediction, and (3) assessment of the aeroelastic responses for buildings
Appl. Sci. 2022, 12, 5232 8 of 28

and bridges. Like several ML trends, the number of studies applying or implementing ML
for wind engineering has been increasing significantly, specifically in the last couple of
years. This reveals the future potential within the wind engineering community where
ML techniques continue to gain more attention and interest from academicians and re-
searchers. More than 50% of the total number of studies that were considered in this sur-
vey and started in the past 30 years were published only in the last two years (Figure 4),
which elucidates the importance of implementing ML techniques in this important and
critical domain.

30
Number of studies

25
20
15
10
5
0
Before 2000 2004–2000 2009–2005 2014–2010 2019–2015 2020–2022
Years
Figure 4. Number of published ML-related studies with wind engineering applications.

3.1. Prediction of Wind-Induced Pressure

Wind-induced pressure prediction forms an essential area in structural wind engi-
neering. In addition to field studies, different tools can be used for estimating wind loads
and pressure coefficients on surfaces, such as atmospheric boundary layer wind tunnels
(ABLWT) or CFD simulations. Both ABLWT and CFD are commonly used but in some
cases may require significant time, cost and expertise [79]. As in other fields of civil engi-
neering, studies using ML techniques have gained some momentum and wind engineers
have shown interest in identifying a reliable approach to predict wind speeds and or
wind-induced pressures for common wind-related structural applications. A summary of
the key attributes and ML implementation in the studies that were included in the review
related to the first category, i.e., the prediction of wind-induced pressures and time series
from experimental testing or databases, is first provided in Table 1, then each study is
discussed in more details in this section. The input variables used in each study are sig-
nificant to the desired output needed from training the ML model. It depends mainly on
the architecture of the model and the different inclusive parameters for each dataset. For
predicting surface pressure it may depend mainly on either the coordinates of the pres-
sure taps, or slope of the roof, wind direction, or building height. While for the aeroelastic
responses of bridges, the input variables mainly depend on parameters such as displace-
ment, velocity and acceleration response for the bridges. One of the studies used the di-
mension between the buildings as input variables (Sx, Sy) to predict the interference effect
on surface pressure.
Appl. Sci. 2022, 12, 5232 9 of 28

Table 1. Summary of studies reviewed for wind-induced predictions.

Study no. Ref. Surface Type Source of Data Input Variables Output Variables ML Algorithm
Experimental data Sampling time
1 [80] Flat roof Pressure time series ANN
from BLWT series
Experimental data
2 [81] Gable roof x, y, z, and (θ) 𝐶𝑝 and 𝐶 ANN
from BLWT
Previous experimental
3 [82] Tall buildings Sx, Sy and h Interference effect RBFNN
studies
𝐶𝑝, 𝐶 and power
Experimental data
4 [53] Flat roof x, y, z, and (θ) spectra of fluctuatingANN
from BLWT
wind pressures
Experimental data
5 [83] Gable roof x, y, z, (θ), and (β) 𝐶𝑝 and 𝐶 ANN
from BLWT
High-rise Experimental data x, y, z and sampling𝐶𝑝, 𝐶 and pressure
6 [84] POD-ANN
building from BLWT time series time series
Flat, gable and
NIST database, and
7 [85] hip roofs and D/B, (θ) and (β) 𝐶 ANN
TPU database
walls
Experimental data
8 [86] Flat roof Terrain turbulence 𝐶𝑝, 𝐶 and 𝐶 ANN
from BLWT
x, y, z, (θ) and
Experimental data 𝐶𝑝, 𝐶 and pressure
9 [87] Flat roof sampling time GPR
from BLWT time series
series
Re, Ti and cylinder
Previous experimental DT, RF, and
10 [66] Circular cylinders circumferential 𝐶𝑝 and 𝐶
studies GBRT
angle
DT, RF,
High-rise
11 [88] TPU database (Sx and Sy) and (θ) 𝐶𝑝 and 𝐶 GANN, and
building
XGBoost
C-shaped Experimental data R/D, D/B, d/b and
12 [89] 𝐶𝑝 GMDH-NN
building from BLWT D/H
NIST database and
Gable roof and
13 [90] DesignSafe-CI x, y, z, and (θ) 𝐶𝑝 and 𝐶 ANN
walls
database
Experimental data Time series, power ANN-GANN-
14 [91] Tall buildings (θ)
from BLWT spectra and 𝐶𝑝 WNN
𝐶𝑝, 𝐶 and 𝐶 and
15 [92] Gable roof TPU database CA, (θ) GBDT
time series

Many methods can be used for predicting and interpolating multivariate modeling
problems, such as linear interpolation and regression polynomials. However, linear inter-
polation cannot solve nonlinear problems and regression polynomials are common to ob-
tain empirical equations, but these empirical equations lack the generality to be used with
other data and a large number of variables [81]. Therefore, ML models generally and ANN
particularly have the advantages over the latter methods in complex problems.
Most of the studies have adopted the three-stage evaluation process of training, test-
ing and validation (TTV), which was proposed by [93] to build a robust ML model. The
cross-validation process comprises two steps: first, the dataset is randomly shuffled and
is divided into k subsets of similar sizes, then k − 1 sets are used for training and one set is
used as the testing set to assess the performance of the model. The stability and the accu-
racy of the validation method depend mainly on the k value. Hence, the cross-validation
Appl. Sci. 2022, 12, 5232 10 of 28

method is usually referred to as k-fold cross-validation [19,94] and is illustrated in Figure

5. Many of the reviewed studies used the 10-fold CV method following Refaeilzadeh et
al.’s [95] recommendation of using k = 10 as a good estimate.

Figure 5. Illustration of the k-fold cross-validation method.

ANN is the most commonly used technique employed in the reviewed studies (see
Table 1). A study by Chen et al. [81] predicted the pressure coefficients on a gable roof
using ANN. This was one of the most important and early studies for implementing ML
models to predict wind-induced pressure on building surfaces. Later, Chen et al. [96] in-
terpolated pressure time series from existing buildings to different roof height buildings,
and then successfully extrapolated to other buildings with different dimensions and roof
slopes using ANN.
Zhang and Zhang [82] evaluated the interference wind-induced effects, that were ex-
pressed by interference factor (IF) among tall buildings using radial basis function neural
networks (RBF-NN). The RBF-NN is a feed-forward type neural network, but the activa-
tion function is different from those that are commonly used (i.e., tangent sigmoid or a
logarithmic sigmoid). The RBF-NN was used first by [50] and it is a function whose re-
sponse either decreases or increases with the distance from a center point [51,52]. It was
found that the predicted IF values were in very good agreement with the experimental
counterparts. The interference index due to shielding between buildings was predicted
from experimental data from wind tunnels using neural network models by English [97].
The study found that the neural network model was able to accurately predict the inter-
ference index for building configurations that have not been tested experimentally. The
interference index can be calculated by subtracting 1 from the shielding (buffeting) factor.
Bre et al. [85] predicted the surface-averaged pressure coefficients of low-rise build-
ings with different types of roofs using ANN. The predicted mean pressure coefficients,
using the Tokyo Polytechnic University (TPU) database [98] as input data, were reasona-
ble when compared to the “M&P” parametric equation [99] and the “S&C” equation [100].
Those two equations are provided here (Equations (5) and (6), respectively) for conven-
ience.
Appl. Sci. 2022, 12, 5232 11 of 28

𝐷 𝑎 + 𝑎 𝐺 + 𝑎 𝜃 + 𝑎 𝜃 + 𝑎 𝐺𝜃
𝐶𝑝 𝜃, = (5)
𝐵 1 + 𝑏 𝐺 + 𝑏 𝜃 + 𝑏 𝜃 + 𝑏 𝐺𝜃
𝐷 𝜃
𝐶𝑝 𝜃, = 𝐶𝑝(0°) ln[1.248 − 0.703 sin − 1.175 sin (𝜃)
𝐵 2
𝜃 𝜃
+ 0.131sin (2𝐺𝜃) + 0.769 cos + 0.07𝐺 sin (6)
2 2
𝜃
+ 0.717cos
2
where ai and bi are adjustable coefficients, θ is the wind angle, D/b is the side ratio, G =
ln(D/B), and 𝐶𝑝(0°) is assumed by Swami and Chnadra [100] equal to 0.6 independent
from D/B.
Hu and Kwok [66] successfully predicted the wind pressures around cylinders using
different ML techniques for Reynolds numbers ranging from 104 to 106, and turbulence
intensities levels ranging from 0% to 15% using several data from previous literature. In
this particular study, the RF and GBRT performed better than the single regression tree
model. Fernández-Cabán et al. [86] used ANN to predict the mean, RMS and peak pres-
sure coefficients on low-rise building flat roofs for three different scaled models. The pre-
dicted mean and RMS pressure coefficient show a very good agreement with the experi-
mental data, especially for the smaller-scale model. Hu and Kwok [88] investigated the
wind pressure on tall buildings under interference effects using different ML models. The
models were trained by different portions of the dataset ranging from 10% to 90% of the
available data. The results showed that the GANs model could predict wind pressures
based on 30% training data only, which may eliminate 70% of the wind tunnel test cases
and accordingly decrease the cost of testing. In addition, RF exhibited a good performance
when the number of grown trees, the n number of features and the maximum depth of
the tree were set to 100, 3 and 25, respectively. Likewise, Vrachimi [101] predicted wind
pressure coefficients for box-shaped obstructed building facades using ANN with a ±0.05
confidence interval for a confidence level of 95%.
Tian et al. [90] focused on predicting the mean and the peak pressure coefficient on a
low-rise gable building using a deep neural network (DNN). This study presented a strat-
egy to predict peak pressure coefficients which is considered a more challenging task
when ML models are used. The strategy is used to predict first the mean pressure coeffi-
cient and then use the predicted mean pressure as an input with other input variables to
predict peak pressure coefficients. This strategy is a reflection of the ensemble methods
idea [58], which is an effective method for solving complex problems with limited inputs.
FNN models were also successfully used in several studies [53,54,102] to predict mean
pressure distribution and power spectra of fluctuating pressures. The most significant fea-
ture of FNN models is the capability of approximating any nonlinear continuous function
to a desired degree of accuracy. Thus, this family of methods can capture the non-linearity
relationship between the different input variables such as wind pressures, wind direc-
tions, and coordinates of pressure taps.
Another technique that is based on the methodology of applying ANN was used by
Mallick et al. [92] in predicting surface mean pressure coefficients using equations for the
group method of data handling neural networks (GMDH-NN)—a derivative method
from ANN. The GMDH-NN is a self-organized system that provides a parametric equa-
tion to predict the output and can solve extremely complex problems [103]. This ML algo-
rithm was established using the GMDH shell software [104] and it is based on the princi-
ple of termination [104–106] to find the nonlinear relation between pressure coefficients
and the input variables. Termination is the process where the parameters are seeded,
reared, hybridized, selected, and rejected to determine the input variables. The study in-
vestigated in detail the effect of curvature and corners on pressure distribution and ob-
tained an equation with different variables to predict the mean pressure coefficients. One
Appl. Sci. 2022, 12, 5232 12 of 28

major difference between ANN and GMDH-NN is that the neurons are filtered simulta-
neously based on their ability to predict the desired values, and then only those beneficial
neurons are fed forward to be trained in the following layer, while the rest are discarded.
One other method to predict wind-induced pressures and full dynamic response, i.e.,
time history on high-rise building surfaces, was proposed by Dongmei et al. [84] using a
backpropagation neural network (BPNN) combined with proper orthogonal decomposi-
tion (POD-BPNN). POD was utilized by Armitt [107] and later by Lumley [108] to deal
with wind turbulence-related issues. The advantage of the POD-BPNN method over the
ANN is its capability to predict pressure time series for trained data with time parameter
t. POD is an approach that is based on a linear combination of a series of orthogonal load
modes, where the spatial distributed multivariable random loads can be reconstructed
through it and loading principle coordinates [109]. The orthogonal load modes are space-
related and time-independent, while the loading principal coordinates are time-varying
and space-independent. Before applying the BPNN, the wind loads were decomposed
using POD where the interdependent variables are transformed into a weighted superpo-
sition of several independent variables. More details about the POD background theory
can be found in the literature [110–112]. The training algorithm applied in that study was
the improved global Levenberg–Marquardt algorithm, which can achieve a faster conver-
gence speed [113,114]. A similar study by Ma et al. [87] investigated the wind pressure-
time history using both gaussian process regression (GPR) and BPNN on a low-rise build-
ing with a flat roof. The study concluded that GPR has high accuracy for time history
interpolation and extrapolation.
The wind pressure time series and power spectra were again recently simulated and
interpolated on tall buildings by Chen et al. [91] using three ML methods: BPNN, genetic
algorithm (GANN), and wavelet neural network (WNN). The WNN produced the most
accurate results within the three methods. The WNN combines the advantages of ANN
with wavelet transformation, which has time-frequency localization property and focal
features which are different from neural networks that have self-adaptive, fault tolerance,
robustness and strong inference ability [115]. The reviewed literature showed that the de-
veloped BPNN models could generalize the complex, multivariate nonlinear functional
relationships among different variables such as wind-induced pressures and locations of
pressure taps. Predicting pressure time series at different roof locations was achieved us-
ing ANN and the robustness of the models was able to overcome the problems associated
with linear interpolation for low-resolution data.
A recent study [92] developed an ML model to predict the wind-induced mean and
peak pressure for non-isolated buildings, considering the interference effect of neighbor-
ing structures using GBDT combined with the grid search algorithm (GSA). The study
used wind tunnel data from TPU for non-isolated buildings. The data were split by a ratio
of 9:1, where 90% of the dataset was used for training and 10% of the dataset was used for
testing. Four hyperparameters were considered in developing the ML model, two hy-
perparameters for CART (i.e., maximum depth, d, for each decision tree, and a minimum
number of samples to split an internal node), and two hyperparameters for a gradient
boosting approach, i.e., learning rate (Lr) and number of CART models. The developed
method was shown to be a robust and accurate method to predict the wind-induced pres-
sure on structures under the interference effects of neighboring structures. Zhang et al.
[116] predicted the typhoon-induced response (TIR) on long-span bridges using quantile
random forest (QRF) with bayesian optimization instead of the traditional FE analysis.
The QRF with bayesian optimization was able to provide adequate probabilistic estima-
tions to quantify the uncertainty in predictions.

3.2. Integration of CFD with Machine Learning

Several studies integrated CFD simulations with ML techniques to predict either the
wind force exerted on bluff bodies or the aeroelastic response of bridges and other flexible
structures [117–122]. Chang et al. [123] predicted the peak pressure coefficients on a low-
Appl. Sci. 2022, 12, 5232 13 of 28

rise building using 12 output data types from a CFD model such as mean pressure coeffi-
cient, dynamic pressure, wind speed, etc. as input variables in the ANN model. The pre-
dicted peak pressures were in good agreement with the wind tunnel data. Similarly,
Vesmawala et al. [124] used ANN to predict the pressure coefficient on domes of different
span to height ratios. The data were generated from the CFD model by developing a dome
and a wind flow through the model. The predicted mean pressure coefficients were used
for training the ML model with a maximum number of epochs of 50,000 to achieve the
specified error tolerance. There were three main inputs: the span/height ratio, the angle
measured vertically with respect to the vertical axis of the dome to the ring beam, and the
angle measured horizontally with respect to wind direction. The study used neuroscience
software in the model training and testing, and it was found that the BPNN predicted the
mean pressure coefficients accurately through different locations along the dome.
Bairagi and Dalui [125] investigated the effect of a setback in tall buildings by pre-
dicting pressure coefficients along the building’s face. The study used ANN and Fast Fou-
rier Transform (FFT) to validate the wind-induced pressure on different setback buildings
predicted by CFD simulation models. The predicted wind pressures were validated before
using similar experimental data. The study showed that CFD was capable to predict sim-
ilar pressure coefficients to experimental data and showed that ANN was capable to pre-
dict and validate these pressure coefficients. The Levenberg–Marquardt algorithm was
used as the training function, starting with 500 training epochs which were increased until
the correlation coefficient exceeded the 99th percentile. The model was trained using
MATLAB neural network toolbox [126].
A recent study [127] proposed a multi-fidelity ML approach to predict wind loads on
tall buildings by integrating CFD models with ML models. The study combined data from
a large number of wind directions using the computationally efficient Reynolds-averaged
Navier–Stokes (RANS) model with a smaller number of wind directions using the more
computationally intense Large Eddy Simulation (LES) method to predict the RMS pres-
sure coefficients on a tall building. The study utilized four types of ML models: linear
regression, quadratic regression, RF, and DNN, with the latter being the most accurate. In
addition, a bootstrap algorithm was used to generate an ensemble of ML models with
accurate confidence intervals. This study used the Adam optimization algorithm [128] and
Rectified Linear Unit (ReLU) activation function [129,130] with a learning rate of 0.001 and
regularization strength of 0.01 to avoid overfitting. That was contrary to other studies that
used the Levenberg–Marquardt algorithm and tangent sigmoid or logarithmic sigmoid
activation functions and this is because the other studies used the ANN method of two or
less hidden layers, while the latter study used a DNN with three hidden layers.
To conclude this section, a summary of the attributes of the reviewed previous stud-
ies that integrate ML applications with CFD is provided in Table 2.

Table 2. Summary of studies reviewed for integrating ML models with CFD simulation.

Study no. Ref. Surface Type Source of Data Input Variables Output Variables ML Algorithm
1 [123] Flat roof 12 parameters 𝐶 ANN
span/height ratio, П
2 [124] Spherical domes 𝐶𝑝 ANN
and φ
Flutter and
Disp., velocities, and
3 [131] Box-girder bridge buffeting ANN
CFD simulation accelerations
responses
Response time Motion-induced
4 [132] Bridges ANN
histories forces
𝐶𝑝 along the face,
5 [125] Setback building (θ) drag and lift ANN
coefficients
Appl. Sci. 2022, 12, 5232 14 of 28

6 [133] Bridges Displacements Deck vibrations LSTM

Vortex induced
7 [120] Circular Cylinders M (θ), U and L DT, RF and GBRT
vibrations
8 [127] Tall buildings (θ) 𝐶 LR-QR-RF-DNN
Different nodes on RF-GP-LR-KNN-
9 [134] Tall building 𝐶𝑝
the surface DT-SVR

3.3. Aeroelastic Response Prediction Using ML

The prediction of aeroelastic responses for buildings and structures by using ML
models is also of interest to this review. The input that was used for the prediction of these
responses is either CFD simulations (Table 2) or physical testing databases (Table 3). Sim-
ilar to the previous two sections, Table 3 is meant to provide a summary of the attributes
of the key studies reviewed in this section that is concerned with using ML for aeroelastic
response prediction.
Chen et al. [135] used a BPNN that was built from a limited dataset of already existing
dynamic responses of rectangular bridge sections. The results indicated that the ANN
prediction scheme performed well in the prediction of dynamic responses. The authors
claimed that such an approach may reduce cost and save time by not using extensive wind
tunnel testing, especially in the preliminary design. Wu and Kareem [131] developed a
new approach utilizing ANN with cellular automata (CA) scheme to model the hysteretic
behavior of bridge aerodynamic nonlinearities in the time domain. This approach was
developed because the ANN is time-consuming until the ideal number of hidden layers
and neurons between the input and output are determined. By embedding the CA
scheme, which was originally proposed by [136] and later developed by [137] with ANN,
the authors of that study aimed to improve the efficiency of the ANN models. The CA
scheme is an approach that dynamically evolves in discrete space and time using a local
rule belonging to a class of Boolean functions. This scheme is appealing as it could simu-
late very complicated problems with the simple local rule which is applied to the system
consistently in space and time. The activation function used in the ANN training was bi-
polar sigmoid as shown in Equation (7). The CA scheme is an indirect encoding scheme
that is based on the CA representative and could be designed using two cellular systems,
i.e., the growing cellular system and the pruning cellular system [138]. The ANN config-
uration based on the CA scheme was examined using a fitness index that is defined in
Equation (8), which is a function of learning cycles and connections of ANN [139].
2
𝑓 (𝑢) = −1 (7)
1 + exp(−𝑢)
1
𝑓 = (8)
𝑐𝑜𝑛𝑛 + 𝑐𝑜𝑛𝑛 𝑐𝑦𝑐

Table 3. Summary of studies reviewed for aeroelastic response.

Study no. Ref. Surface Type Source of Data Input Variables Output Variables ML Algorithm
Experimental data Flutter derivatives
1 [135] Bridges D/B ANN
from BLWT (H1 and A2)
Experimental data Vb and top floor
2 [140] Tall buildings Column strains CNN
from BLWT displacements
H, B, L, Vb and Across wind shear
3 [141] Tall buildings IndianWind Code ANN
TC and moment
Cross spectral
4 [142] Long span bridge Full scale data Buffeting response ANN and SVR
density
Appl. Sci. 2022, 12, 5232 15 of 28

Vertex
Experimental data SVR, ANN, RF
5 [143] Box girders coordinates (mi, Flutter wind speed
from BLWT and GBRT
ni )
Rectangular Previous Crosswind DT-RF-KNN-
6 [144] Ti, B/D and Sc
cylinders experimental studies vibrations GBRT
Experimental data
Vertical
7 [145] Cable roofs from BLWT and 11 parameters ANN
displacements
(FEM)
Terrain rough-
Crosswind force
8 [146] Tall buildings WERC database-TU ness, aspect ratio LGBM
spectra
and D/B.

The dynamic response of tall buildings was studied by Nikose and Sonparote
[141,147] using ANN and the proposed graphs were able to predict the along- and across-
wind responses in terms of base shear and base bending moments according to the Indian
Wind Code (IWC). Both studies found that the back propagation neural network algo-
rithm was able to satisfactory estimate the dynamic along- and across-wind responses of
tall buildings. Similarly, different ML models were applied by Hu and Kwok [144] based
on DT, KNN regression, RF, and GBRT to predict four types of crosswind vibrations (i.e.,
over-coupled, coupled, semi-coupled and decoupled) for rectangular cylinders. The data
used in training and testing processes were extracted from wind tunnel data. It was found
that GBRT can accurately predict crosswind responses which can supplement wind tun-
nel tests and numerical simulation techniques. One of the input variables used in that
study was the Scruton number (Sc).
Oh et al. (2019) [140] studied the wind-induced response of tall buildings using CNN
and focused on the structural safety evaluation. The trained model predicted the column
strains using wind tunnel data such as wind speed and top floor displacements. The ar-
chitecture of the trained model is composed of the input layer, two convolutional layers,
two pooling layers, one fully connected layer, and the output layer. The input map forms
the convolutional layer through convolution using the kernel operator. The ML-based
model was utilized to overcome the uncertainties in the material, geometric properties
and stiffness contribution of nonstructural elements which make it difficult to construct a
refined finite element model.
Li et al. [133] used LSTM—originally proposed by Hochreiter and Schmidhuber
[148]—to predict the nonlinear unsteady bridge aerodynamic responses to overcome the
increasing difficulties that exist in the gradient-based learning algorithm in the recurrent
neural network (RNN) face. The RNN was developed to introduce the time dimension
into the network structure, and it was found to be capable of predicting a full-time series
where nonlinear relation exists between input and output. The study used displacement
time series as input variables, and by weighting these time series, both the acceleration
and velocity were obtained. The LSTM model was able to calculate the deck vibrations
(i.e., lift displacement and torsional angle) under the unsteady nonlinear wind loads. Hu
and Kwok [136] investigated the vortex-induced vibrations (VIV) of two circular cylinders
with the same dimensions but staggered configurations, using three ML algorithms: DT,
RF, and GBRT. The two cylinders were modeled first into a CFD simulation, and the mass
ratio, wind direction, the distance between cylinders, and wind velocity were used as in-
put variables. The GBRT algorithm was the most accurate in predicting the amplitude of
the upstream and downstream vibration. Abbas et al. [132] employed ANN to predict the
aeroelastic response of bridge decks using response time histories as the input variables.
The predicted forces were compared with CFD findings to evaluate the ANN model. The
ANN model was also coupled with the structural model to determine the aeroelastic in-
stability limit of the bridge section, which demonstrated the potential use of this frame-
work to predict the aeroelastic response for other bridge cross-sections.
Appl. Sci. 2022, 12, 5232 16 of 28

More recently, surrogate models have been used widely in different areas related to
structural wind engineering [149–152]. One type of surrogate model is using the aid of
finite element models (FEM) to obtain an output that can be used as an input in the trained
model of the ML. Chen et al. [153] used a surrogate model in which the ANN was applied
to the FE model to update the model parameter for computing the dynamic response of a
cable-suspended roof while using the wind loads from full-scale measurements for three
typhoon events in three consecutive years from 2011 to 2014. Luo and Kareem [154] pro-
posed a surrogate model using a convolutional neural network (CNN) for systems with
high dimensional inputs/outputs. Rizzo and Caracoglia [145] predicted the wind-induced
vertical displacement of a cable net roof using ANN. The trained model used wind tunnel
pressure coefficient datasets and FEM wind-induced vertical displacement datasets. The
surrogate model showed that it can successfully replicate more complex geometrically
nonlinear structural behavior. Rizzo and Caracoglia [155] used surrogate flutter derivate
models to predict the flutter velocity of a suspension bridge. The ANN model was trained
using the critical flutter velocities dataset by calculating the flutter derivatives experimen-
tally. The model successfully generated a large dataset of critical flutter velocities. In ad-
dition, surrogate modeling could analyze the structural performance of vertical structures
under tornado loads by training fragilities using ANN [156,157].
Lin et al. [146] used a light gradient boosting machine (LGBM) method, which is an
optimized version of the GBDT algorithm proposed by Ke et al. [158], with a clustering
algorithm to predict the crosswind force spectra of tall buildings. This optimized algo-
rithm combined two techniques in training the models: the gradient base one side sam-
pling (GOSS) and the exclusive feature bundling (EFB). The results showed that the pro-
posed method is effective and efficient to predict the crosswind force spectrum for a rec-
tangular tall building.
Liao et al. [143] used four different ML techniques (i.e., SVR, ANN, RF, and GBRT)
to predict the flutter wind speed for a box girder bridge. The ANN and GBRT models
accurately predicted the flutter wind speed for the streamlined box girders. The buffeting
response of bridges can be predicted analytically using buffeting theory. However, some
previous studies [159–163] have shown inconsistency between full-scale measured re-
sponse and buffeting theory estimates. Thus, Castellon et al. [142] trained two ML models
(ANN and SVR) to estimate the buffeting response speed using full-scale data from the
Hardanger bridge in Norway. The two ML models predicted the bridge response more
accurately than the buffeting theory when compared to the full-scale measurement. Fur-
thermore, the drag force of a circular cylinder can be reduced by optimizing the control
parameter such as feedback gain and the phase lag using neural networks by minimizing
the velocity fluctuations in the cylinder wake [164].

4. Summary of Tools of Performance Assessment of ML Models

The performance of the ML models in wind engineering applications throughout the
reviewed literature was assessed through at least one or more forms of different standard
statistical error and standard indices. It is important for any ML model to evaluate the
performance of the model using some error metrics or factors. Thus, this section aims to
provide future researchers with a summary of all the tools and equations that have been
used up to this date in structural wind engineering ML applications along with an assess-
ment of which tools are more appropriate for the applications at hand. The compiled list
of metrics, or factors, calculates the error to evaluate the accuracy between the ML pre-
dicted data and a form of ground truth such as experimental data or independent sets of
data that were not used in training among others. There is always a lack of consensus on
the most accurate metric that can be used. Nonetheless, this section attempts to provide
more guidance on which methods are preferred based on the surveyed studies.
Several error metrics were used throughout the reviewed literature which include:
Akaike information criterion (AIC), coefficient of efficiency (Ef), coefficient of determina-
tion (R2), Pearson’s correlation coefficient (R), mean absolute error (MAE), mean absolute
Appl. Sci. 2022, 12, 5232 17 of 28

percentage error (MAPE), mean square error (MSE), root mean square error (RMSE), scat-
ter index (SI), and sensitivity error (Si). For the convenience of the readers and for com-
pleteness, the equations used to express each of these error metrics for assessing predicted
data (pi) against measured data (mi) are summarized below (Equations (9) to (18)). For N
number of data points (e.g., N could be the number of pressure tabs used to provide ex-
perimental data), some of the error calculation equations also use average or mean values
for predicted data (𝑝̅ ) as well as measured data (𝑚).

1
𝐴𝐼𝐶 = 𝑁 log( (𝑝 − 𝑚 ) ) + 2𝑘 (9)
𝑁

∑ |𝑚 − 𝑝 |
𝐸 =1− (10)
∑ 𝑝
∑ (𝑚 − 𝑚)(𝑝 − 𝑝̅ )
𝑅 = (11)
∑ (𝑚 − 𝑚) ∑ (𝑝 − 𝑝̅ )

∑ (𝑚 − 𝑚)(𝑝 − 𝑝̅ )
𝑅= (12)
∑ (𝑚 − 𝑚) ∑ (𝑝 − 𝑝̅ )

1
𝑀𝐴𝐸 = |𝑚 − 𝑝 | (13)
𝑁

1 𝑝 − 𝑚
𝑀𝐴𝑃𝐸 = × 100% (14)
𝑁 𝑝

1 𝑝 − 𝑚
𝑀𝑆𝐸 = ( ) (15)
𝑁 𝑚

1
𝑅𝑀𝑆𝐸 = (𝑝 − 𝑚 ) (16)
𝑁

∑ [(𝑝 − 𝑝̅ ) − (𝑚 − 𝑚)]
𝑆𝐼 = (17)
∑ (𝑚 )

𝑋
𝑆 = × 100; where 𝑋 = 𝑓 (𝑥 ) − 𝑓 (𝑥 ), (18)
∑ 𝑋
where 𝑓 (𝑥 ) and 𝑓 (𝑥 ), are the corresponding maximum and minimum values for
the predicted output over the ith input factor while using the mean values for the other
factors.
In general, MSE was employed in most of the studies and is considered one of the
most common error metrics for pressure distribution prediction, but it is not always an
accurate error metric. The MSE accuracy decreases when the pressure among the walls is
included in the prediction because walls might introduce a pressure coefficient near zero
which may cause a great rise as the normalizing denominator [90]. Nevertheless, MSE is
generally stable when used in RF models when the number of trees reaches 100 [88]. The
RMSE is not affected by the near-zero pressure coefficient as with MSE because it does not
include a normalization factor in the calculation. Nevertheless, the lack of normalization
is considered a limitation for this metric in the cases where the scale of pressure coeffi-
cients changes [90]. The accuracy of some metric errors increases when their values ap-
proach one (i.e., coefficient of determination-R2), which means that the predicted data are
close to the experimental data, and the accuracy of some others increases when their val-
ues are close to zero (i.e., root mean square error-RMSE).
Appl. Sci. 2022, 12, 5232 18 of 28

The correlation coefficient, R, is considered a reliable approach for estimating the

prediction accuracy by measuring how similar two sets of data are, but its limitation is
that it does not reflect the range nor the bias between the two datasets. The coefficient of
efficiency, E, corresponds to the match between the model and observed data and can
range from −∞ to 1, and a perfect match corresponds to E = 1 [89]. AIC is a mathematical
method used to evaluate how the model fits the trained data and this is an information
criterion used to select the best-fit model. One other error metric that has not been com-
monly used in the literature is the SI normalized measure of error, where a lower SI value
indicates better performance for the model. Besides the error metrics that assess the per-
formance level of the model, other factors are used to indicate the effect of input variables
on the output. The most common example is the sensitivity analysis error percentage (Si)
(Equation (18)) which computes the contribution of each input variable to the output var-
iable [165–167]. The Si is an important factor to determine the contribution of each input
value, especially when different inputs are used in the ML model training, which could
be of great significance for informing and changing the assigned weight of neurons in
neural networks.
Overall, it is important to note that each error metric or factor usually conveys spe-
cific information regarding the performance of the ML model, especially in the case of
wind engineering applications (due to variation of wall versus roof pressures for in-
stance), and most of these metrics and factors are interdependent. Thus, our recommen-
dation is to consider the following factors together: (1) use R2 to assess the similarity be-
tween the actual and predicted set; (2) use MSE when the model includes the prediction
of roof surface pressure coefficients only without walls, but use either MAPE or RMSE
when pressure coefficients for walls’ surfaces are included in the model; (3) use AIC to
select the best fit model in case of linear regression. This recommendation is to stress the
fact that using several metric errors together is essential to assess the performance of ML
models for structural wind engineering as opposed to only relying on a single metric.

5. Discussion and Conclusions

As in any other application, the quantity and the quality of data is the main challenge
in successfully implementing ML models in the broader area of structural wind engineer-
ing. It is important to mention that the quality of the dataset used for training is as im-
portant as the quantity of data. The measurements usually may involve some anomalies
such as missing data or outliers, thus removing the outliers is essential for the accuracy
and robustness of the model [168,169]. ML algorithms are data-hungry processes that re-
quire thousands if not millions of observations to reach acceptable performance levels.
Bias in data collection is another major drawback that could dramatically affect the per-
formance of ML models [170]. To this end, some literature recommends that the number
of datasets shall not be taken less than 10 times the number of independent variables,
according to 10 events per variable (EPV) [171]. Meanwhile, K-means clustering was used
in many different studies due to its ability to analyze the dataset and recognize its under-
lying pattern. Most of the ML techniques need several trials and experiments through the
validation process to develop a robust model with high accuracy prediction levels. For
instance, whenever ANN is used, several trials are conducted for training purposes in
terms of choosing the number of hidden layers and the number of neurons in each layer.
The ANN method is not recommended for datasets with a small sample size because
this would achieve double the mean absolute error (MAE) compared to other ML tech-
niques [134]. ANN is capable of learning and generalizing nonlinear complex functional
relationships via the training process, but there is currently no theoretical basis for deter-
mining the ideal neural network configuration [81]. The architecture of ANN and training
parameters cannot be generalized even within data of similar nature [141]. Generally, one
hidden layer is enough for most problems, but for very complex, fuzzy and highly non-
linear problems, more than one hidden layer (node) might be required to capture the sig-
nificant features in the data [172]. The number of hidden nodes is determined through
Appl. Sci. 2022, 12, 5232 19 of 28

trials and in most cases, this number is set to no more than 2n + 1, where n is the number
of input variables [173]. In addition, a study by Sheela and Deepa [174] reviewed different
models for calculating the number of hidden neurons and developed a proposed method
that gave the least MSE compared to the other models. The proposed approach was im-
plemented on wind speed prediction and was very effective compared to other models.
Furthermore, a general principle of a ratio of 3:1 or 3:2 between the first and second hidden
nodes provides a better prediction performance compared to other combinations [175].
Generally, a robust neural network model can be built of two hidden layers and ten neu-
rons and will give a very reasonable response.
ANN also appears to have a significant computational advantage over a CFD-based
scheme. In ANN, the computational work is mainly focused on identifying the proper
weights in the network. Once the training phase is completed, the output of the simulated
system could be obtained through a simple arithmetic operation with any desired input
information. On the other hand, in the case of a CFD scheme, each new input scenario
requires a complete reevaluation of the fluid–structure interaction over the discretized
domain.
From the review of the literature, it was also apparent that ANN has weighted ad-
vantages over other ML methods. However, there are some challenges accompanying im-
plementing ANN in certain types of wind engineering applications. ANN is problematic
in predicting the pressure coefficients within the leading corner and edges due to the sep-
aration which is accompanied by high rms pressure coefficient values and corner vortices.
This may be eliminated by training datasets of full- or large-scale models that contain
high-resolution pressure tapped areas. It is important to note that whenever the data are
fed into a regression model or ANN model (training, validation or testing process), all the
predictors are normalized between [−1, 1] to condition the input matrix. In the case of
implementing ANN models, the Levenberg–Marquardt algorithm and tangent sigmoid
or logarithmic sigmoid activation functions shall be used. On the contrary, the Adam op-
timization algorithm and Rectified linear unit activation function shall be used whenever
a DNN model (i.e., three or more hidden layers) is used as the ML technique.
The literature review revealed that there are selected ML techniques that might not
be as popular as ANN yet but with potential for future wind engineering applications and
specific structural wind engineering problems. Less common ML methods, such as the
wavelet neural network (WNN), are gaining increasing attention due to their advantage
over ANN and other models in terms of prediction accuracy and good fit [176]. In addi-
tion, wavelet analysis is becoming popular due to its capacity to reveal simultaneous spec-
tral and temporal information within a single signal [177]. Other ML techniques such as
DL can be used as a probabilistic model for predictions based on limited and noisy data
[178]. GANs models can be used in structural health monitoring for damage detections in
buildings using different images for damage that occurred during an extreme wind event.
BPNN and GRNN were used to acquire the missing data due to the failed pressure sensors
while testing [179]. The GPR has high accuracy for time history interpolation and extrap-
olation and in the same context, the WNN predicts the time series accurately compared to
other methods. Surrogate models were proved to be a powerful tool to integrate both FEM
with ML models which could solve complex problems, such as the dynamic response of
roofs and bridges while using the wind loads from physical testing measurements and
can replicate more complex geometrically nonlinear structure behavior.
Ensemble methods have shown good results in predicting wind-induced forces and
vibrations of structures. Due to the time-consuming and cost-prohibitive nature of con-
ducting a lot of wind tunnel testing, ML models such as DT, KNN, RF and GBRT are found
to be efficient [144], and in turn, recommended for accurately predicting crosswind vibra-
tions. The GBRT specifically can accurately predict crosswind responses when it is needed
to supplement wind tunnel tests and numerical simulation techniques. ANN and GBRT
are found to be the ideal ML models for wind speed prediction. Moreover, RF and GBRT
are found to predict wind-induced loads more accurately when compared to DT. GBDT
Appl. Sci. 2022, 12, 5232 20 of 28

is preferable to be used over ANN in the case of a small amount of input data, as ANN
requires a large amount of input data for an accurate prediction as explained above. Pre-
dicting wind gusts, which has not been a common application in the reviewed work in
this study, can be achieved accurately using ensemble methods or neural networks and
logistic regression [180–185].
If only wind tunnel testing is considered, the wind flow around buildings, which
provides deep insight into the aerodynamic behavior of buildings, is usually captured
using particle image velocimetry (PIV). However, measuring wind velocities at some lo-
cations is a challenge due to the laser-light shielding. In such cases, DL might be used to
predict these unmeasured velocities at certain locations as proposed in previous work
[186]. Tropical cyclones and typhoons' wind fields can be predicted using ML models us-
ing the storm parameters such as spatial coordinates, storm size and intensity [187,188].
Overall, it was demonstrated through this review that ML techniques offer a power-
ful tool and were successfully implemented in several areas of research related to struc-
tural wind engineering. Such areas that can extend previous work and continue to benefit
from ML techniques are mostly: the prediction of wind-induced pressure time series and
overall loads as well as the prediction of aeroelastic responses, wind gust estimates, and
damage detection following extreme wind events. Nonetheless, other areas that can also
benefit from ML but are yet to be explored more and recommended for future wind engi-
neering research include the development and future codification of ML-based wind vul-
nerability models, advanced testing methods such as cyber-physical testing or hybrid
wind simulation by incorporating surrogate and ML models for geometry optimization,
wind-structure interaction evaluation, among other future applications. Finally, the phys-
ics-informed ML methods could provide a promising way to further improve the perfor-
mance of traditional ML techniques and finite element analysis.

Author Contributions: Conceptualization, K.M.; methodology, K.M.; validation, K.M., I.Z. and
M.A.M.; formal analysis, K.M.; investigation, K.M.; resources, K.M.; writing—original draft prepa-
ration, K.M.; writing—review and editing, I.Z. and M.A.M.; supervision, I.Z. and M.A.M. All au-
thors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations
Nomenclature
x Machine learning input variable
y Machine learning output
h Neural network hidden layer
𝑥 Input for a generic neuron
𝑤 Weight of a generic connection between two nodes
𝑏 Bias of a generic neuron
𝑦 Output for a generic neuron
𝑓 (𝑢) Transfer function
𝑢 Value of membership function
mij Mean of the Gaussian function
σij Standard deviation of the Gaussian function
L1 LASSO regularization
L2 Ridge regularization
𝑝 Predicted output
𝑚 Measured output
𝑆 Normalized measure for error
Appl. Sci. 2022, 12, 5232 21 of 28

θ Wind direction

β Roof slope

D/B Side ratio

x, y, z Pressure taps coordinates
Re Reynolds number
Ti Turbulence intensity
Sx, Sy Interfering building location
R/D Curvature ratio
d/b Side ratio without curvature
D/H Height ratio
h Building height
Sc Scruton number
M Mass ratio
L Distance between the centerline of the cylinders
U Reduced velocity
H1 Flutter Derivatives (vertical motion)
A2 Flutter Derivatives (torisonal motion)
mi, ni Vertex coordinates
L Length of the building
Vb Wind velocity
TC Terrain category
𝐶 Mean pressure coefficient
𝐶 Peak pressure coefficient
𝐶 Root mean square pressure coefficient
φ The angle measured horizontally with respect to wind direction
The angle measured vertically with respect to the vertical axis of the dome to
П
the ring beam.
CA Neighboring area density
Abbreviations
ABLWT Atmospheric boundary layer wind tunnel
AIC Akaike information criterion
ANN Artificial neural network
CFD Computational fluid dynamics
CNN Convolutional neural networks
DL Deep learning
DNN Deep neural network
DT Decision tree regression
Ef Coefficient of efficiency
FFNN Feed-forward neural network
FNN Fuzzy neural networks
GAN Generative adversarial networks
GANN Genetic neural networks
GBRT Gradient boosting regression tree
GMDH-NN Group method of data handling neural networks
GPR Gaussian process regression
KNN K-nearest neighbor regression
LES Large eddy simulation
Lr Learning Rate
LSTM Long short-term memory
MAE Mean absolute error
MAPE Mean absolute percentage error
ML Machine learning
MSE Mean square error
POD-BPNN Proper orthogonal decomposition-backpropagation neural network
R Pearson’s correlation coefficient
R2 Coefficient of determination
Appl. Sci. 2022, 12, 5232 22 of 28

RANS Reynolds-averaged Navier–Stokes

RBF-NN Radial basis function neural networks
ReLU Rectified liner unit
RF Random forest
RMS Root mean square
RMSE Root mean square error
RNN recurrent neural networks
RTHS Real-time hybrid simulation
SI Scatter index
SVM Support vector machine
VIV Vortex induced vibration
WNN Wavelet neural network

References
1. Solomonoff, R. The time scale of artificial intelligence: Reflections on social effects. Hum. Syst. Manag. 1985, 5, 149–153.
https://doi.org/10.3233/HSM-1985-5207.
2. Mjolsness, E.; DeCoste, D. Machine Learning for Science: State of the Art and Future Prospects. Science 2001, 293, 2051–2055.
https://doi.org/10.1126/science.293.5537.2051.
3. Murphy, K.P. Machine Learning: A Probabilistic Perspective; MIT Press: Cambridge, MA, USA, 2012.
4. Sun, H.; Burton, H.V.; Huang, H. Machine learning applications for building structural design and performance assessment:
State-of-the-art review. J. Build. Eng. 2020, 33, 101816. https://doi.org/10.1016/j.jobe.2020.101816.
5. Saravanan, R.; Sujatha, P. A State of Art Techniques on Machine Learning Algorithms: A Perspective of Supervised Learning
Approaches in Data Classification. In Proceedings of the 2018 Second International Conference on Intelligent Computing and
Control Systems (ICICCS), Madurai, India, 14–15 June 2018; pp. 945–949. https://doi.org/10.1109/iccons.2018.8663155.
6. Kang, M.; Jameson, N.J. Machine Learning: Fundamentals. Progn. Health Manag. Electron. 2018, 85–109.
https://doi.org/10.1002/9781119515326.ch4.
7. Hastie, T.; Tibshirani, R.; Friedman, J. The Elements of Statistical Learning; Springer Series in Statistics; Springer: Berlin/Heidel-
berg, Germany, 2001.
8. Adeli, H. Neural Networks in Civil Engineering: 1989–2000. Comput. Civ. Infrastruct. Eng. 2001, 16, 126–142.
https://doi.org/10.1111/0885-9507.00219.
9. Çevik, A.; Kurtoğlu, A.E.; Bilgehan, M.; Gülşan, M.E.; Albegmprli, H.M. Support vector machines in structural engineering: A
review. J. Civ. Eng. Manag. 2015, 21, 261–281.
10. Dibike, Y.B.; Velickov, S.; Solomatine, D. Support vector machines: Review and applications in civil engineering. In Proceedings
of the 2nd Joint Workshop on Application of AI in Civil Engineering, Cottbus, Germany, 26–28 March 2000; pp. 45–58.
11. Bas, E.E.; Moustafa, M.A. Real-Time Hybrid Simulation with Deep Learning Computational Substructures: System Validation
Using Linear Specimens. Mach. Learn. Knowl. Extr. 2020, 2, 469–489. https://doi.org/10.3390/make2040026.
12. Bas, E.E.; Moustafa, M.A. Communication Development and Verification for Python-Based Machine Learning Models for Real-
Time Hybrid Simulation. Front. Built Environ. 2020, 6, 574965.https://doi.org/10.3389/fbuil.2020.574965.
13. Xie, Y.; Ebad Sichani, M.; Padgett, J.E.; Desroches, R. The promise of implementing machine learning in earthquake engineering:
A state-of-the-art review. Earthq. Spectra 2020, 36, 1769–1801. https://doi.org/10.1177/8755293020919419.
14. Mosavi, A.; Ozturk, P.; Chau, K.-W. Flood Prediction Using Machine Learning Models: Literature Review. Water 2018, 10, 1536.
https://doi.org/10.3390/w10111536.
15. Munawar, H.S.; Hammad, A.; Ullah, F.; Ali, T.H. After the flood: A novel application of image processing and machine learning
for post-flood disaster management. In Proceedings of the 2nd International Conference on Sustainable Development in Civil
Engineering (ICSDC 2019), Jamshoro, Pakistan, 5–7 December 2019; pp. 5–7.
16. Deka, P.C. A Primer on Machine Learning Applications in Civil Engineering; CRC Press: Boca Raton, FL, USA, 2019.
https://doi.org/10.1201/9780429451423.
17. Huang, Y.; Li, J.; Fu, J. Review on Application of Artificial Intelligence in Civil Engineering. Comput. Model. Eng. Sci. 2019, 121,
845–875. https://doi.org/10.32604/cmes.2019.07653.
18. Reich, Y. Artificial Intelligence in Bridge Engineering. Comput. Civ. Infrastruct. Eng. 1996, 11, 433–445.
https://doi.org/10.1111/j.1467-8667.1996.tb00355.x.
19. Reich, Y. Machine Learning Techniques for Civil Engineering Problems. Comput. Civ. Infrastruct. Eng. 1997, 12, 295–310.
https://doi.org/10.1111/0885-9507.00065.
20. Lu, P.; Chen, S.; Zheng, Y. Artificial Intelligence in Civil Engineering. Math. Probl. Eng. 2012, 2012, 145974.
https://doi.org/10.1155/2012/145974.
21. Vadyala, S.R.; Betgeri, S.N.; Matthews, D.; John, C. A Review of Physics-based Machine Learning in Civil Engineering. arXiv
2021, arXiv:2110.04600.
22. Salehi, H.; Burgueño, R. Emerging artificial intelligence methods in structural engineering. Eng. Struct. 2018, 171, 170–189.
https://doi.org/10.1016/j.engstruct.2018.05.084.
Appl. Sci. 2022, 12, 5232 23 of 28

23. Dixon, C.R. The Wind Resistance of Asphalt Roofing Shingles; University of Florida: Gainesville, FL, USA, 2013.
24. Flood, I. Neural Networks in Civil Engineering: A Review. In Civil and Structural Engineering Computing: 2001; Saxe-Coburg
Publications: Stirlingshire, UK, 2001; pp. 185–209. https://doi.org/10.4203/csets.5.8.
25. Rao, D.H. Fuzzy Neural Networks. IETE J. Res. 1998, 44, 227–236. https://doi.org/10.1080/03772063.1998.11416049.
26. Avci, O.; Abdeljaber, O.; Kiranyaz, S. Structural Damage Detection in Civil Engineering with Machine Learning: Current State
of the Art. In Sensors and Instrumentation, Aircraft/Aerospace, Energy Harvesting & Dynamic Environments Testing; Springer: Cham,
Switzerland, 2022; pp. 223–229. https://doi.org/10.1007/978-3-030-75988-9_17.
27. Avci, O.; Abdeljaber, O.; Kiranyaz, S.; Hussein, M.; Gabbouj, M.; Inman, D.J. A review of vibration-based damage detection in
civil structures: From traditional methods to Machine Learning and Deep Learning applications. Mech. Syst. Signal Process. 2021,
147, 107077. https://doi.org/10.1016/j.ymssp.2020.107077.
28. Hsieh, Y.-A.; Tsai, Y.J. Machine Learning for Crack Detection: Review and Model Performance Comparison. J. Comput. Civ. Eng.
2020, 34, 04020038. https://doi.org/10.1061/(asce)cp.1943-5487.0000918.
29. Hou, R.; Xia, Y. Review on the new development of vibration-based damage identification for civil engineering structures: 2010–
2019. J. Sound Vib. 2020, 491, 115741. https://doi.org/10.1016/j.jsv.2020.115741.
30. Flah, M.; Nunez, I.; Ben Chaabene, W.; Nehdi, M.L. Machine Learning Algorithms in Civil Structural Health Monitoring: A
Systematic Review. Arch. Comput. Methods Eng. 2020, 28, 2621–2643. https://doi.org/10.1007/s11831-020-09471-9.
31. Smarsly, K.; Dragos, K.; Wiggenbrock, J. Machine learning techniques for structural health monitoring. In Proceedings of the
8th European Workshop On Structural Health Monitoring (EWSHM 2016), Bilbao, Spain, 5–8 July 2016; Volume 2, pp. 1522–
1531.
32. Mishra, M. Machine learning techniques for structural health monitoring of heritage buildings: A state-of-the-art review and
case studies. J. Cult. Heritage 2021, 47, 227–245. https://doi.org/10.1016/j.culher.2020.09.005.
33. Li, S.; Li, S.; Laima, S.; Li, H. Data-driven modeling of bridge buffeting in the time domain using long short-term memory
network based on structural health monitoring. Struct. Control Health Monit. 2021, 28, e2772. https://doi.org/10.1002/stc.2772.
34. Shahin, M. A review of artificial intelligence applications in shallow foundations. Int. J. Geotech. Eng. 2014, 9, 49–60.
https://doi.org/10.1179/1939787914y.0000000058.
35. Puri, N.; Prasad, H.D.; Jain, A. Prediction of Geotechnical Parameters Using Machine Learning Techniques. Procedia Comput.
Sci. 2018, 125, 509–517. https://doi.org/10.1016/j.procs.2017.12.066.
36. Pirnia, P.; Duhaime, F.; Manashti, J. Machine learning algorithms for applications in geotechnical engineering. In Proceedings
of the GeoEdmonton, Edmonton, AL, Canada, 23–26 September 2018; pp. 1–37.
37. Yin, Z.; Jin, Y.; Liu, Z. Practice of artificial intelligence in geotechnical engineering. J. Zhejiang Univ. A 2020, 21, 407–411.
https://doi.org/10.1631/jzus.a20aige1.
38. Chao, Z.; Ma, G.; Zhang, Y.; Zhu, Y.; Hu, H. The application of artificial neural network in geotechnical engineering. IOP Conf.
Ser. Earth Environ. Sci. 2018, 189, 022054. https://doi.org/10.1088/1755-1315/189/2/022054.
39. Shahin, M.A. State-of-the-art review of some artificial intelligence applications in pile foundations. Geosci. Front. 2016, 7, 33–44.
https://doi.org/10.1016/j.gsf.2014.10.002.
40. Wang, H.; Zhang, Y.-M.; Mao, J.-X. Sparse Gaussian process regression for multi-step ahead forecasting of wind gusts combin-
ing numerical weather predictions and on-site measurements. J. Wind Eng. Ind. Aerodyn. 2021, 220, 104873.
https://doi.org/10.1016/j.jweia.2021.104873.
41. Simiu, E.; Scanlan, R.H. Wind Effects on Structures: Fundamentals and Applications to Design; John Wiley: New York, NY, USA,
1996.
42. Haykin, S. Neural Networks: A Comprehensive Foundation, 1999; Mc Millan: Hamilton, NJ, USA, 2010; pp. 1–24.
43. Nasrabadi, N.M. Pattern recognition and machine learning. J. Electron. Imaging 2007, 16, 049901.
https://doi.org/10.1117/1.2819119.
44. Haykin, S. Neural Networks and Learning Machines, 3/E; Pearson Education India: Noida, India, 2010.
45. Waszczyszyn, Z.; Ziemiański, L. Neural Networks in the Identification Analysis of Structural Mechanics Problems. In Parameter
Identification of Materials and Structures; Springer: Berlin/Heidelberg, Germany, 2005; pp. 265–340.
46. Rumelhart, D.E.; Hinton, G.E.; Williams, R.J. Learning representations by back-propagating errors. Nature 1986, 323, 533–536.
https://doi.org/10.1038/323533a0.
47. Hagan, M.T.; Menhaj, M.B. Training feedforward networks with the Marquardt algorithm. IEEE Trans. Neural Netw. 1994, 5,
989–993. https://doi.org/10.1109/72.329697.
48. Marquardt, D.W. An Algorithm for Least-Squares Estimation of Nonlinear Parameters. J. Soc. Ind. Appl. Math. 1963, 11, 431–441.
https://doi.org/10.1137/0111030.
49. Demuth, H.; Beale, M. Neural Network Toolbox for Use with MATLAB; The Math Works Inc.: Natick, MA, USA, 1998; pp. 10–30.
50. Broomhead, D.S.; Lowe, D. Radial Basis Functions, Multi-Variable Functional Interpolation and Adaptive Networks; Royal Signals
and Radar Establishment Malvern: Malvern, UK, 1988.
51. Park, J.; Sandberg, I.W. Universal Approximation Using Radial-Basis-Function Networks. Neural Comput. 1991, 3, 246–257.
https://doi.org/10.1162/neco.1991.3.2.246.
52. Bianchini, M.; Frasconi, P.; Gori, M. Learning without local minima in radial basis function networks. IEEE Trans. Neural Net-
works 1995, 6, 749–756. https://doi.org/10.1109/72.377979.
Appl. Sci. 2022, 12, 5232 24 of 28

53. Fu, J.; Liang, S.; Li, Q. Prediction of wind-induced pressures on a large gymnasium roof using artificial neural networks. Comput.
Struct. 2007, 85, 179–192. https://doi.org/10.1016/j.compstruc.2006.08.070.
54. Fu, J.; Li, Q.; Xie, Z. Prediction of wind loads on a large flat roof using fuzzy neural networks. Eng. Struct. 2005, 28, 153–161.
https://doi.org/10.1016/j.engstruct.2005.08.006.
55. Nilsson, N.J. Introduction to Machine Learning an Early Draft of a Proposed Textbook Department of Computer Science. Mach.
Learn. 2005, 56, 387–399.
56. Loh, W. Classification and regression trees. Wiley Interdiscip. Rev. Data Min. Knowl. Discov. 2011, 1, 14–23.
57. Loh, W.-Y. Fifty Years of Classification and Regression Trees. Int. Stat. Rev. 2014, 82, 329–348. https://doi.org/10.1111/insr.12016.
58. Zhou, Z.-H. Ensemble Methods: Foundations and Algorithms; CRC Press: Boca Raton, FL, USA, 2012.
59. Breiman, L. Bagging predictors. Mach. Learn. 1996, 24, 123–140.
60. Hastie, T.; Tibshirani, R.; Friedman, J. Unsupervised learning. In The Elements of Statistical Learning; Springer: Berlin/Heidelberg,
Germany, 2009; pp. 485–585.
61. Breiman, L. Random forests. Mach. Learn. 2001, 45, 5–32.
62. Friedman, J.H. Greedy function approximation: A gradient boosting machine. Ann. Stat. 2001, 29, 1189–1232.
https://doi.org/10.1214/aos/1013203451.
63. Persson, C.; Bacher, P.; Shiga, T.; Madsen, H. Multi-site solar power forecasting using gradient boosted regression trees. Sol.
Energy 2017, 150, 423–436. https://doi.org/10.1016/j.solener.2017.04.066.
64. Natekin, A.; Knoll, A. Gradient boosting machines, a tutorial. Front. Neurorobot. 2013, 7, 21.
65. Elith, J.; Leathwick, J.R.; Hastie, T. A working guide to boosted regression trees. J. Anim. Ecol. 2008, 77, 802–813.
https://doi.org/10.1111/j.1365-2656.2008.01390.x.
66. Hu, G.; Kwok, K. Predicting wind pressures around circular cylinders using machine learning techniques. J. Wind Eng. Ind.
Aerodyn. 2020, 198, 104099. https://doi.org/10.1016/j.jweia.2020.104099.
67. Zhang, Y.; Haghani, A. A gradient boosting method to improve travel time prediction. Transp. Res. Part C Emerg. Technol. 2015,
58, 308–324. https://doi.org/10.1016/j.trc.2015.02.019.
68. Chen, T.; Guestrin, C. Xgboost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD International Con-
ference on Knowledge Discovery and Data, San Francisco, CA, USA, 13–17 August 2016; pp. 785–794.
69. Rasmussen, C.E. Gaussian processes in machine learning. In Summer School on Machine Learning; Springer: Berlin/Heidelberg,
Germany, 2003; pp. 63–71.
70. Rasmussen, C.E.; Williams, C.K.I. Model Selection and Adaptation of Hyperparameters. In Gaussian Processes for Machine Learn-
ing; MIT Press: Cambridge, MA, USA, 2005. https://doi.org/10.7551/mitpress/3206.003.0008.
71. Ebden, M. Gaussian Processes: A Quick Introduction. arXiv 2015, arXiv:1505.02965.
72. Goodfellow, I.; Pouget-Abadie, J.; Mirza, M.; Xu, B.; Warde-Farley, D.; Ozair, S.; Courville, A.; Bengio, Y. Generative adversarial
nets. In Proceedings of the Advances in Neural Information Processing Systems 27 (NIPS 2014), Montreal, QC, Canada, 8–11
December 2014.
73. Goodfellow, I.; Pouget-Abadie, J.; Mirza, M.; Xu, B.; Warde-Farley, D.; Ozair, S.; Courville, A.; Bengio, Y. Generative Adversarial
Networks. Commun. ACM 2020, 63, 139–144. https://doi.org/10.1145/3422622.
74. Fix, E.; Hodges, J.L. Discriminatory Analysis. Nonparametric Discrimination: Consistency Properties. Int. Stat. Rev. Int. Stat.
1989, 57, 238–247.
75. Zhang, Z. Introduction to machine learning: K-nearest neighbors. Ann. Transl. Med. 2016, 4, 218–218.
https://doi.org/10.21037/atm.2016.03.37.
76. Noble, W.S. What is a support vector machine? Nat. Biotechnol. 2006, 24, 1565–1567.
77. Cortes, C.; Vapnik, V. Support-vector networks. Mach. Learn. 1995, 20, 273–297.
78. Wang, L. Support Vector Machines: Theory and Applications; Springer Science & Business Media: Berlin/Heidelberg, Germany,
2005; Volume 177.
79. Cóstola, D.; Blocken, B.; Hensen, J. Overview of pressure coefficient data in building energy simulation and airflow network
programs. Build. Environ. 2009, 44, 2027–2036. https://doi.org/10.1016/j.buildenv.2009.02.006.
80. Chen, Y.; Kopp, G.; Surry, D. Interpolation of wind-induced pressure time series with an artificial neural network. J. Wind Eng.
Ind. Aerodyn. 2002, 90, 589–615. https://doi.org/10.1016/s0167-6105(02)00155-1.
81. Chen, Y.; Kopp, G.; Surry, D. Prediction of pressure coefficients on roofs of low buildings using artificial neural networks. J.
Wind Eng. Ind. Aerodyn. 2003, 91, 423–441. https://doi.org/10.1016/s0167-6105(02)00381-1.
82. Zhang, A.; Zhang, L. RBF neural networks for the prediction of building interference effects. Comput. Struct. 2004, 82, 2333–
2339. https://doi.org/10.1016/j.compstruc.2004.05.014.
83. Gavalda, X.; Ferrer-Gener, J.; Kopp, G.A.; Giralt, F. Interpolation of pressure coefficients for low-rise buildings of different plan
dimensions and roof slopes using artificial neural networks. J. Wind Eng. Ind. Aerodyn. 2011, 99, 658–664.
https://doi.org/10.1016/j.jweia.2011.02.008.
84. Dongmei, H.; Shiqing, H.; Xuhui, H.; Xue, Z. Prediction of wind loads on high-rise building using a BP neural network com-
bined with POD. J. Wind Eng. Ind. Aerodyn. 2017, 170, 1–17. https://doi.org/10.1016/j.jweia.2017.07.021.
85. Bre, F.; Gimenez, J.M.; Fachinotti, V. Prediction of wind pressure coefficients on building surfaces using artificial neural net-
works. Energy Build. 2018, 158, 1429–1441. https://doi.org/10.1016/j.enbuild.2017.11.045.
Appl. Sci. 2022, 12, 5232 25 of 28

86. Fernández-Cabán, P.L.; Masters, F.J.; Phillips, B. Predicting Roof Pressures on a Low-Rise Structure From Freestream Turbu-
lence Using Artificial Neural Networks. Front. Built Environ. 2018, 4, 68. https://doi.org/10.3389/fbuil.2018.00068.
87. Ma, X.; Xu, F.; Chen, B. Interpolation of wind pressures using Gaussian process regression. J. Wind Eng. Ind. Aerodyn. 2019, 188,
30–42. https://doi.org/10.1016/j.jweia.2019.02.002.
88. Hu, G.; Liu, L.; Tao, D.; Song, J.; Tse, K.; Kwok, K. Deep learning-based investigation of wind pressures on tall building under
interference effects. J. Wind Eng. Ind. Aerodyn. 2020, 201, 104138. https://doi.org/10.1016/j.jweia.2020.104138.
89. Mallick, M.; Mohanta, A.; Kumar, A.; Patra, K.C. Prediction of Wind-Induced Mean Pressure Coefficients Using GMDH Neural
Network. J. Aerosp. Eng. 2020, 33, 04019104. https://doi.org/10.1061/(asce)as.1943-5525.0001101.
90. Tian, J.; Gurley, K.R.; Diaz, M.T.; Fernández-Cabán, P.L.; Masters, F.J.; Fang, R. Low-rise gable roof buildings pressure predic-
tion using deep neural networks. J. Wind Eng. Ind. Aerodyn. 2019, 196, 104026. https://doi.org/10.1016/j.jweia.2019.104026.
91. Chen, F.; Wang, X.; Li, X.; Shu, Z.; Zhou, K. Prediction of wind pressures on tall buildings using wavelet neural network. J.
Build. Eng. 2021, 46, 103674. https://doi.org/10.1016/j.jobe.2021.103674.
92. Weng, Y.; Paal, S.G. Machine learning-based wind pressure prediction of low-rise non-isolated buildings. Eng. Struct. 2022, 258,
114148. https://doi.org/10.1016/j.engstruct.2022.114148.
93. Reich, Y.; Barai, S. Evaluating machine learning models for engineering problems. Artif. Intell. Eng. 1999, 13, 257–272.
https://doi.org/10.1016/s0954-1810(98)00021-1.
94. Browne, M.W. Cross-Validation Methods. J. Math. Psychol. 2000, 44, 108–132. https://doi.org/10.1006/jmps.1999.1279.
95. Refaeilzadeh, P.; Tang, L.; Liu, H. Cross-validation. Encycl. Database Syst. 2009, 5, 532–538.
96. Chen, Y.; Kopp, G.A.; Surry, D. Interpolation of pressure time series in an aerodynamic database for low buildings. J. Wind Eng.
Ind. Aerodyn. 2003, 91, 737–765. https://doi.org/10.1016/s0167-6105(03)00006-0.
97. English, E.; Fricke, F. The interference index and its prediction using a neural network analysis of wind-tunnel data. J. Wind
Eng. Ind. Aerodyn. 1999, 83, 567–575. https://doi.org/10.1016/s0167-6105(99)00102-6.
98. Yoshie, R.; Iizuka, S.; Ito, Y.; Ooka, R.; Okaze, T.; Ohba, M.; Kataoka, H.; Katsuchi, H.; Katsumura, A.; Kikitsu, H.; et al. 13th
International Conference on Wind Engineering. Wind Eng. JAWE 2011, 36, 406–428. https://doi.org/10.5359/jawe.36.406.
99. Muehleisen, R.; Patrizi, S. A new parametric equation for the wind pressure coefficient for low-rise buildings. Energy Build.
2013, 57, 245–249. https://doi.org/10.1016/j.enbuild.2012.10.051.
100. Swami, M.V.; Chandra, S. Correlations for pressure distribution on buildings and calculation of natural-ventilation airflow.
ASHRAE Trans.1988, 94, 243–266.
101. Vrachimi, I. Predicting local wind pressure coefficients for obstructed buildings using machine learning techniques. In Proceed-
ings of the Building Simulation Conference, 14 December 2017, San Francisco, CA, USA. pp. 1–8.
102. Gavalda, X.; Ferrer-Gener, J.; Kopp, G.A.; Giralt, F.; Galsworthy, J. Simulating pressure coefficients on a circular cylinder at Re=
106 by cognitive classifiers. Comput. Struct. 2009, 87, 838–846. https://doi.org/10.1016/j.compstruc.2009.03.005.
103. Ebtehaj, I.; Bonakdari, H.; Khoshbin, F.; Azimi, H. Pareto genetic design of group method of data handling type neural network
for prediction discharge coefficient in rectangular side orifices. Flow Meas. Instrum. 2015, 41, 67–74.
https://doi.org/10.1016/j.flowmeasinst.2014.10.016.
104. Amanifard, N.; Nariman-Zadeh, N.; Farahani, M.; Khalkhali, A. Modelling of multiple short-length-scale stall cells in an axial
compressor using evolved GMDH neural networks. Energy Convers. Manag. 2008, 49, 2588–2594. https://doi.org/10.1016/j.encon-
man.2008.05.025.
105. Ivakhnenko, A.G. Polynomial Theory of Complex Systems. IEEE Trans. Syst. Man Cybern. 1971, SMC-1, 364–378.
https://doi.org/10.1109/tsmc.1971.4308320.
106. Ivakhnenko, A.G.; Ivakhnenko, G.A. Problems of further development of the group method of data handling algorithms. Part
I. Pattern Recognit. Image Anal. C/C Raspoznavaniye Obraz. I Anal. Izobr. 2000, 10, 187–194.
107. Armitt, J. Eigenvector analysis of pressure fluctuations on the West Burton instrumented cooling tower. In Central Electricity
Research Laboratories (UK) Internal Report; RD/L/N 114/68; Central Electricity Research Laboratories: Leatherhead, UK, 1968.
108. Lumley, J.L. Stochastic Tools in Turbulence; Courier Corporation: Chelmsford, MA, USA, 2007.
109. Azam, S.E.; Mariani, S. Investigation of computational and accuracy issues in POD-based reduced order modeling of dynamic
structural systems. Eng. Struct. 2013, 54, 150–167. https://doi.org/10.1016/j.engstruct.2013.04.004.
110. Chatterjee, A. An introduction to the proper orthogonal decomposition. Curr. Sci. 2000, 78, 808–817.
111. Liang, Y.; Lee, H.; Lim, S.; Lin, W.; Lee, K.; Wu, C. Proper Orthogonal Decomposition and Its Applications—Part I: Theory. J.
Sound Vib. 2002, 252, 527–544. https://doi.org/10.1006/jsvi.2001.4041.
112. Berkooz, G.; Holmes, P.; Lumley, J.L. The proper orthogonal decomposition in the analysis of turbulent flows. Annu. Rev. Fluid
Mech. 1993, 25, 539–575.
113. Fan, J.Y. Modified Levenberg-Marquardt algorithm for singular system of nonlinear equations. J. Comput. Math. 2003, 21, 625–
636.
114. Fan, J.; Pan, J. A note on the Levenberg–Marquardt parameter. Appl. Math. Comput. 2009, 207, 351–359.
https://doi.org/10.1016/j.amc.2008.10.056.
115. Wang, G.; Guo, L.; Duan, H. Wavelet Neural Network Using Multiple Wavelet Functions in Target Threat Assessment. Sci.
World J. 2013, 2013, 632437. https://doi.org/10.1155/2013/632437.
Appl. Sci. 2022, 12, 5232 26 of 28

116. Zhang, Y.-M.; Wang, H.; Mao, J.-X.; Xu, Z.-D.; Zhang, Y.-F. Probabilistic Framework with Bayesian Optimization for Predicting
Typhoon-Induced Dynamic Responses of a Long-Span Bridge. J. Struct. Eng. 2021, 147, 04020297.
https://doi.org/10.1061/(asce)st.1943-541x.0002881.
117. Zhao, Y.; Meng, Y.; Yu, P.; Wang, T.; Su, S. Prediction of Fluid Force Exerted on Bluff Body by Neural Network Method. J.
Shanghai Jiaotong Univ. 2019, 25, 186–192. https://doi.org/10.1007/s12204-019-2140-0.
118. Miyanawala, T.P.; Jaiman, R.K. An efficient deep learning technique for the Navier-Stokes equations: Application to unsteady
wake flow dynamics. arXiv 2017, arXiv:1710.09099.
119. Ye, S.; Zhang, Z.; Song, X.; Wang, Y.; Chen, Y.; Huang, C. A flow feature detection method for modeling pressure distribution
around a cylinder in non-uniform flows by using a convolutional neural network. Sci. Rep. 2020, 10, 4459–10.
https://doi.org/10.1038/s41598-020-61450-z.
120. Gu, S.; Wang, J.; Hu, G.; Lin, P.; Zhang, C.; Tang, L.; Xu, F. Prediction of wind-induced vibrations of twin circular cylinders
based on machine learning. Ocean Eng. 2021, 239, 109868. https://doi.org/10.1016/j.oceaneng.2021.109868.
121. Raissi, M.; Wang, Z.; Triantafyllou, M.S.; Karniadakis, G.E. Deep learning of vortex-induced vibrations. J. Fluid Mech. 2018, 861,
119–137. https://doi.org/10.1017/jfm.2018.872.
122. Peeters, R.; Decuyper, J.; de Troyer, T.; Runacres, M.C. Modelling vortex-induced loads using machine learning. In Proceedings
of the International Conference on Noise and Vibration Engineering (ISMA), Virtual, 7–9 September 2020; pp. 1601–1614.
123. Chang, C.; Shang, N.; Wu, C.; Chen, C. Predicting peak pressures from computed CFD data and artificial neural networks
algorithm. J. Chin. Inst. Eng. 2008, 31, 95–103. https://doi.org/10.1080/02533839.2008.9671362.
124. Vesmawala, G.R.; Desai, J.A.; Patil, H.S. Wind pressure coefficients prediction on different span to height ratios domes using
artificial neural networks. Asian J. Civ. Eng. 2009, 10, 131–144.
125. Bairagi, A.K.; Dalui, S.K. Forecasting of Wind Induced Pressure on Setback Building Using Artificial Neural Network. Period.
Polytech. Civ. Eng. 2020, 64, 751–763. https://doi.org/10.3311/ppci.15769.
126. Demuth, H.; Beale, M. Neural Network Toolbox: For Use with MATLAB (Version 4.0); The MathWorks Inc.: Natick, MA, USA, 2004.
127. Lamberti, G.; Gorlé, C. A multi-fidelity machine learning framework to predict wind loads on buildings. J. Wind Eng. Ind. Aer-
odyn. 2021, 214, 104647. https://doi.org/10.1016/j.jweia.2021.104647.
128. Kingma, D.P.; Ba, J.L. Adam: A method for stochastic optimization. In Proceedings of the 3rd International Conference on
Learning Representations, San Diego, CA, USA, 7–9 May 2015; pp. 1–15.
129. Agarap, A.F. Deep Learning Using Rectified Linear Units (ReLU). 2018. pp. 2–8. Available online:
http://arxiv.org/abs/1803.08375 (accessed on 1 March 2022).
130. Schmidt-Hieber, J. Nonparametric regression using deep neural networks with ReLU activation function. Ann. Stat. 2020, 48,
1875–1897. https://doi.org/10.1214/19-aos1875.
131. Wu, T.; Kareem, A. Modeling hysteretic nonlinear behavior of bridge aerodynamics via cellular automata nested neural net-
work. J. Wind Eng. Ind. Aerodyn. 2011, 99, 378–388. https://doi.org/10.1016/j.jweia.2010.12.011.
132. Abbas, T.; Kavrakov, I.; Morgenthal, G.; Lahmer, T. Prediction of aeroelastic response of bridge decks using artificial neural
networks. Comput. Struct. 2020, 231, 106198. https://doi.org/10.1016/j.compstruc.2020.106198.
133. Li, T.; Wu, T.; Liu, Z. Nonlinear unsteady bridge aerodynamics: Reduced-order modeling based on deep LSTM networks. J.
Wind Eng. Ind. Aerodyn. 2020, 198, 104116. https://doi.org/10.1016/j.jweia.2020.104116.
134. Waibel, C.; Zhang, R.; Wortmann, T. Physics Meets Machine Learning: Coupling FFD with Regression Models for Wind Pressure
Prediction on High-Rise Facades; Association for Computing Machinery: New York, NY, USA, 2021; Volume 1.
135. Chen, C.-H.; Wu, J.-C.; Chen, J.-H. Prediction of flutter derivatives by artificial neural networks. J. Wind Eng. Ind. Aerodyn. 2008,
96, 1925–1937. https://doi.org/10.1016/j.jweia.2008.02.044.
136. Schwartz, J.T.; Von Neumann, J.; Burks, A.W. Theory of Self-Reproducing Automata. Math. Comput. 1967, 21, 745.
https://doi.org/10.2307/2005041.
137. Wolfram, S. Universality and complexity in cellular automata. Phys. D Nonlinear Phenom. 1984, 10, 1–35.
138. Galván, I.M.; Isasi, P.; López, J.M.M.; de Miguel, M.A.S. Neural Network Architectures Design by Cellular Automata Evolution;
Kluwer Academic Publishers, Norwell, MA, USA, 2000.
139. Gutiérrez, G.; Sanchis, A.; Isasi, P.; Molina, M. Non-direct encoding method based on cellular automata to design neural net-
work architectures. Comput. Inform. 2005, 24, 225–247.
140. Oh, B.K.; Glisic, B.; Kim, Y.; Park, H.S. Convolutional neural network-based wind-induced response estimation model for tall
buildings. Comput. Civ. Infrastruct. Eng. 2019, 34, 843–858. https://doi.org/10.1111/mice.12476.
141. Nikose, T.J.; Sonparote, R.S. Computing dynamic across-wind response of tall buildings using artificial neural network. J. Su-
percomput. 2018, 76, 3788–3813. https://doi.org/10.1007/s11227-018-2708-8.
142. Castellon, D.F.; Fenerci, A.; Øiseth, O. A comparative study of wind-induced dynamic response models of long-span bridges
using artificial neural networks, support vector regression and buffeting theory. J. Wind Eng. Ind. Aerodyn. 2020, 209, 104484.
https://doi.org/10.1016/j.jweia.2020.104484.
143. Liao, H.; Mei, H.; Hu, G.; Wu, B.; Wang, Q. Machine learning strategy for predicting flutter performance of streamlined box
girders. J. Wind Eng. Ind. Aerodyn. 2021, 209, 104493. https://doi.org/10.1016/j.jweia.2020.104493.
144. Lin, P.; Hu, G.; Li, C.; Li, L.; Xiao, Y.; Tse, K.; Kwok, K. Machine learning-based prediction of crosswind vibrations of rectangular
cylinders. J. Wind Eng. Ind. Aerodyn. 2021, 211, 104549. https://doi.org/10.1016/j.jweia.2021.104549.
Appl. Sci. 2022, 12, 5232 27 of 28

145. Rizzo, F.; Caracoglia, L. Examination of artificial neural networks to predict wind-induced displacements of cable net roofs.
Eng. Struct. 2021, 245, 112956. https://doi.org/10.1016/j.engstruct.2021.112956.
146. Lin, P.; Ding, F.; Hu, G.; Li, C.; Xiao, Y.; Tse, K.; Kwok, K.; Kareem, A. Machine learning-enabled estimation of crosswind load
effect on tall buildings. J. Wind Eng. Ind. Aerodyn. 2021, 220, 104860. https://doi.org/10.1016/j.jweia.2021.104860.
147. Nikose, T.J.; Sonparote, R.S. Dynamic along wind response of tall buildings using Artificial Neural Network. Clust. Comput.
2018, 22, 3231–3246. https://doi.org/10.1007/s10586-018-2027-0.
148. Hochreiter, S.; Schmidhuber, J. Long short-term memory. Neural Comput. 1997, 9, 1735–1780.
149. Micheli, L.; Hong, J.; Laflamme, S.; Alipour, A. Surrogate models for high performance control systems in wind-excited tall
buildings. Appl. Soft Comput. 2020, 90, 106133. https://doi.org/10.1016/j.asoc.2020.106133.
150. Qiu, Y.; Yu, R.; San, B.; Li, J. Aerodynamic shape optimization of large-span coal sheds for wind-induced effect mitigation using
surrogate models. Eng. Struct. 2022, 253, 113818. https://doi.org/10.1016/j.engstruct.2021.113818.
151. Sun, L.; Gao, H.; Pan, S.; Wang, J.-X. Surrogate modeling for fluid flows based on physics-constrained deep learning without
simulation data. Comput. Methods Appl. Mech. Eng. 2019, 361, 112732. https://doi.org/10.1016/j.cma.2019.112732.
152. Peña, F.L.; Casás, V.D.; Gosset, A.; Duro, R. A surrogate method based on the enhancement of low fidelity computational fluid
dynamics approximations by artificial neural networks. Comput. Fluids 2012, 58, 112–119.
https://doi.org/10.1016/j.compfluid.2012.01.008.
153. Chen, B.; Wu, T.; Yang, Y.; Yang, Q.; Li, Q.; Kareem, A. Wind effects on a cable-suspended roof: Full-scale measurements and
wind tunnel based predictions. J. Wind Eng. Ind. Aerodyn. 2016, 155, 159–173. https://doi.org/10.1016/j.jweia.2016.06.006.
154. Luo, X.; Kareem, A. Deep convolutional neural networks for uncertainty propagation in random fields. Comput. Civ. Infrastruct.
Eng. 2019, 34, 1043–1054. https://doi.org/10.1111/mice.12510.
155. Rizzo, F.; Caracoglia, L. Artificial Neural Network model to predict the flutter velocity of suspension bridges. Comput. Struct.
2020, 233, 106236. https://doi.org/10.1016/j.compstruc.2020.106236.
156. Le, V.; Caracoglia, L. A neural network surrogate model for the performance assessment of a vertical structure subjected to non-
stationary, tornadic wind loads. Comput. Struct. 2020, 231, 106208. https://doi.org/10.1016/j.compstruc.2020.106208.
157. Caracoglia, L.; Le, V. A MATLAB-based GUI for Performance-based Tornado Engineering (PBTE) of a Monopole, Vertical Struc-
ture with Artificial Neural Networks (ANN). 2020. Available online: https://designsafeci-
dev.tacc.utexas.edu/data/browser/public/designsafe.storage.published/PRJ-2772%2FPBTE_ANN_User_manual.pdf (accessed
on 14 May 2020).
158. Ke, G.; Meng, Q.; Finley, T.; Wang, T.; Chen, W.; Ma, W.; Ye, Q.; Liu, T.Y. Lightgbm: A highly efficient gradient boosting decision
tree. Adv. Neural Inf. Process. Syst. 2017, 30, 3146–3154.
159. Bietry, J.; Delaunay, D.; Conti, E. Comparison of full-scale measurement and computation of wind effects on a cable-stayed
bridge. J. Wind Eng. Ind. Aerodyn. 1995, 57, 225–235. https://doi.org/10.1016/0167-6105(94)00110-y.
160. Macdonald, J. Evaluation of buffeting predictions of a cable-stayed bridge from full-scale measurements. J. Wind Eng. Ind. Aer-
odyn. 2003, 91, 1465–1483. https://doi.org/10.1016/j.jweia.2003.09.009.
161. Cheynet, E.; Jakobsen, J.B.; Snæbjörnsson, J. Buffeting response of a suspension bridge in complex terrain. Eng. Struct. 2016, 128,
474–487. https://doi.org/10.1016/j.engstruct.2016.09.060.
162. Xu, Y.-L.; Zhu, L. Buffeting response of long-span cable-supported bridges under skew winds. Part 2: Case study. J. Sound Vib.
2005, 281, 675–697. https://doi.org/10.1016/j.jsv.2004.01.025.
163. Fenerci, A.; Øiseth, O.; Rønnquist, A. Long-term monitoring of wind field characteristics and dynamic response of a long-span
suspension bridge in complex terrain. Eng. Struct. 2017, 147, 269–284. https://doi.org/10.1016/j.engstruct.2017.05.070.
164. Fujisawa, N.; Nakabayashi, T. Neural Network Control of Vortex Shedding from a Circular Cylinder Using Rotational Feedback
Oscillations. J. Fluids Struct. 2002, 16, 113–119. https://doi.org/10.1006/jfls.2001.0414.
165. Barati, R. Application of excel solver for parameter estimation of the nonlinear Muskingum models. KSCE J. Civ. Eng. 2013, 17,
1139–1148. https://doi.org/10.1007/s12205-013-0037-2.
166. Gandomi, A.H.; Yun, G.J.; Alavi, A.H. An evolutionary approach for modeling of shear strength of RC deep beams. Mater.
Struct. 2013, 46, 2109–2119. https://doi.org/10.1617/s11527-013-0039-z.
167. Mohanta, A.; Patra, K.C. MARS for Prediction of Shear Force and Discharge in Two-Stage Meandering Channel. J. Irrig. Drain.
Eng. 2019, 145, 04019016. https://doi.org/10.1061/(asce)ir.1943-4774.0001402.
168. Zhang, Y.-M.; Wang, H.; Bai, Y.; Mao, J.-X.; Xu, Y.-C. Bayesian dynamic regression for reconstructing missing data in structural
health monitoring. Struct. Health Monit. 2022, 14759217211053779. https://doi.org/10.1177/14759217211053779.
169. Wan, H.-P.; Ni, Y.-Q. Bayesian multi-task learning methodology for reconstruction of structural health monitoring data. Struct.
Health Monit. 2018, 18, 1282–1309. https://doi.org/10.1177/1475921718794953.
170. Halevy, A.; Norvig, P.; Pereira, F. The Unreasonable Effectiveness of Data. IEEE Intell. Syst. 2009, 24, 8–12.
https://doi.org/10.1109/mis.2009.36.
171. Peduzzi, P.; Concato, J.; Kemper, E.; Holford, T.R.; Feinstein, A.R. A simulation study of the number of events per variable in
logistic regression analysis. J. Clin. Epidemiol. 1996, 49, 1373–1379. https://doi.org/10.1016/s0895-4356(96)00236-3.
172. Khanduri, A.; Bédard, C.; Stathopoulos, T. Modelling wind-induced interference effects using backpropagation neural net-
works. J. Wind Eng. Ind. Aerodyn. 1997, 72, 71–79. https://doi.org/10.1016/s0167-6105(97)00259-6.
173. Teng, G.; Xiao, J.; He, Y.; Zheng, T.; He, C. Use of group method of data handling for transport energy demand modeling. Energy
Sci. Eng. 2017, 5, 302–317. https://doi.org/10.1002/ese3.176.
Appl. Sci. 2022, 12, 5232 28 of 28

174. Sheela, K.G.; Deepa, S.N. Review on Methods to Fix Number of Hidden Neurons in Neural Networks. Math. Probl. Eng. 2013,
2013, 425740. https://doi.org/10.1155/2013/425740.
175. Maier, H.; Dandy, G. The effect of internal parameters and geometry on the performance of back-propagation neural networks:
An empirical study. Environ. Model. Softw. 1998, 13, 193–209. https://doi.org/10.1016/s1364-8152(98)00020-6.
176. Wei, S.; Yang, H.; Song, J.; Abbaspour, K.; Xu, Z. A wavelet-neural network hybrid modelling approach for estimating and
predicting river monthly flows. Hydrol. Sci. J. 2013, 58, 374–389. https://doi.org/10.1080/02626667.2012.754102.
177. Nourani, V.; Alami, M.T.; Aminfar, M.H. A combined neural-wavelet model for prediction of Ligvanchai watershed precipita-
tion. Eng. Appl. Artif. Intell. 2009, 22, 466–472. https://doi.org/10.1016/j.engappai.2008.09.003.
178. Luo, X.; Kareem, A. Bayesian deep learning with hierarchical prior: Predictions from limited and noisy data. Struct. Saf. 2020,
84, 101918. https://doi.org/10.1016/j.strusafe.2019.101918.
179. Ni, Y.-Q.; Li, M. Wind pressure data reconstruction using neural network techniques: A comparison between BPNN and GRNN.
Measurement 2016, 88, 468–476. https://doi.org/10.1016/j.measurement.2016.04.049.
180. Sallis, P.; Claster, W.; Hernández, S. A machine-learning algorithm for wind gust prediction. Comput. Geosci. 2011, 37, 1337–
1344. https://doi.org/10.1016/j.cageo.2011.03.004.
181. Cao, Q.; Ewing, B.T.; Thompson, M. Forecasting wind speed with recurrent neural networks. Eur. J. Oper. Res. 2012, 221, 148–
154. https://doi.org/10.1016/j.ejor.2012.02.042.
182. Li, F.; Ren, G.; Lee, J. Multi-step wind speed prediction based on turbulence intensity and hybrid deep neural networks. Energy
Convers. Manag. 2019, 186, 306–322. https://doi.org/10.1016/j.enconman.2019.02.045.
183. Türkan, Y.S.; Aydoğmuş, H.Y.; Erdal, H. The prediction of the wind speed at different heights by machine learning methods.
Int. J. Optim. Control. Theor. Appl. 2016, 6, 179–187. https://doi.org/10.11121/ijocta.01.2016.00315.
184. Wang, H.; Zhang, Y.; Mao, J.-X.; Wan, H.-P. A probabilistic approach for short-term prediction of wind gust speed using en-
semble learning. J. Wind Eng. Ind. Aerodyn. 2020, 202, 104198. https://doi.org/10.1016/j.jweia.2020.104198.
185. Saavedra-Moreno, B.; Salcedo-Sanz, S.; Carro-Calvo, L.; Gascón-Moreno, J.; Jiménez-Fernández, S.; Prieto, L. Very fast training
neural-computation techniques for real measure-correlate-predict wind operations in wind farms. J. Wind Eng. Ind. Aerodyn.
2013, 116, 49–60. https://doi.org/10.1016/j.jweia.2013.03.005.
186. Kim, B.; Yuvaraj, N.; Preethaa, K.S.; Hu, G.; Lee, D.-E. Wind-Induced Pressure Prediction on Tall Buildings Using Generative
Adversarial Imputation Network. Sensors 2021, 21, 2515. https://doi.org/10.3390/s21072515.
187. Snaiki, R.; Wu, T. Knowledge-enhanced deep learning for simulation of tropical cyclone boundary-layer winds. J. Wind Eng.
Ind. Aerodyn. 2019, 194, 103983. https://doi.org/10.1016/j.jweia.2019.103983.
188. Tseng, C.; Jan, C.; Wang, J.; Wang, C. Application of artificial neural networks in typhoon surge forecasting. Ocean Eng. 2007,
34, 1757–1768. https://doi.org/10.1016/j.oceaneng.2006.09.005.