energies

Article
Effective Fault Detection and Diagnosis for Power Converters
in Wind Turbine Systems Using KPCA-Based BiLSTM
Zahra Yahyaoui 1 , Mansour Hajji 1 , Majdi Mansouri 2,3, *, Kamaleldin Abodayeh 3 , Kais Bouzrara 4
and Hazem Nounou 2

1 Research Unit Advanced Materials and Nanotechnologies (UR16ES03), Higher Institute of Applied Sciences
and Technology of Kasserine, Kairouan University, Kasserine 1200, Tunisia
2 Electrical and Computer Engineering Program, Texas A&M University at Qatar, Doha P.O. Box 23874, Qatar
3 Department of Mathematical Sciences, Prince Sultan University, Riyadh 12435, Saudi Arabia
4 Laboratory of Automatic Signal and Image Processing, National Engineering School of Monastir,
Monastir 5035, Tunisia
* Correspondence: [email protected]

Abstract: The current work presents an effective fault detection and diagnosis (FDD) technique in
wind energy converter (WEC) systems. The proposed FDD framework merges the benefits of kernel
principal component analysis (KPCA) model and the bidirectional long short-term memory (BiLSTM)
classifier. In the developed FDD approach, the KPCA model is applied to extract and select the
most effective features, while the BiLSTM is utilized for classification purposes. The developed
KPCA-based BiLSTM approach involves two main steps: feature extraction and selection, and fault
classification. The KPCA model is developed in order to select and extract the most efficient features
and the final features are fed to the BiLSTM to distinguish between different working modes. Different
simulation scenarios are considered in this study in order to show the robustness and performance
Citation: Yahyaoui, Z.; Hajji, M.;
of the developed technique when compared to the conventional FDD methods. To evaluate the
Mansouri, M.; Abodayeh, K.; effectiveness of the proposed KPCA-based BiLSTM approach, we utilize data obtained from a healthy
Bouzrara, K.; Nounou, H. Effective WTC, which are then injected with several fault scenarios: simple fault generator-side, simple fault
Fault Detection and Diagnosis for grid-side, multiple fault generator-side, multiple fault grid-side, and mixed fault on both sides. The
Power Converters in Wind Turbine diagnosis performance is analyzed in terms of accuracy, recall, precision, and computation time.
Systems Using KPCA-Based BiLSTM. Furthermore, the efficiency of fault diagnosis is shown by the classification accuracy parameter. The
Energies 2022, 15, 6127. https://doi. experimental results show the efficiency of the developed KPCA-based BiLSTM technique compared
org/10.3390/en15176127 to the classical FDD techniques (an accuracy of 97.30%).
Academic Editors: Davide Astolfi
and Abu-Siada Ahmed Keywords: wind energy conversion (WEC); fault detection and diagnosis (FDD); kernel PCA (KPCA);
bidirectional long short term memory (BiLSTM)
Received: 29 March 2022
Accepted: 31 May 2022
Published: 23 August 2022

Publisher’s Note: MDPI stays neutral 1. Introduction

with regard to jurisdictional claims in
Wind energy is one of most essential substitute energies due its competitive cost and
published maps and institutional affil-
maturity of technology. According to the World Wind Energy Association (WWEA), the
iations.
total capacity of all wind farms worldwide reached 744 GW in 2020.
Due to the development of wind power production, enhancement of the control of
wind energy conversion (WEC) systems is required. For this reason, manufacturers’ efforts
Copyright: © 2022 by the authors.
have been focused on the improvement of these systems’ lifetimes and the decrease of
Licensee MDPI, Basel, Switzerland. operation breakdowns (downtime maintenance process), leading to continuous energy
This article is an open access article production with high power quality [1,2].
distributed under the terms and Wind energy conversion (WEC) systems are composed of various interconnected
conditions of the Creative Commons electrical and mechanical elements. However, unexpected failures usually accompany the
Attribution (CC BY) license (https:// operation of these systems. When a fault in a system occurs, it can have an adverse effect on
creativecommons.org/licenses/by/ the system’s availability, in addition to the production rate. Indeed, many components of
4.0/). wind turbines (WT) can fail due to harsh environmental and operating conditions, resulting

Energies 2022, 15, 6127. https://doi.org/10.3390/en15176127 https://www.mdpi.com/journal/energies

Energies 2022, 15, 6127 2 of 19

in lengthy downtime maintenance periods [3,4]. The most common failures are related to
blades [5,6], generators [7,8], power converters [1,9], and gearboxes [10,11]. As a crucial
component and the heart of these systems, the power converter plays a significant role
in transferring the generated power to the grid. It converts electrical energy that varies
according to the wind speed to energy with a constant frequency complying with grid
specifications [12]. It was indicated in [13] that 21% of 25% of the total failures in WEC
converters (WECC) are caused by the semiconductor. In order to avoid the WECC collapse,
these failures should be detected and diagnosed at an early stage. Therefore, fault detection
and diagnosis (FDD) is viewed as essential means to achieve these goals [14]. The authors
of [15,16] considered multiple faults in the same-side converter. They address multiple
faults in both converter sides at once. The authors of [17] have studied multiple faults by
modeling both converter sides as a state space equation. In [18], the authors examined two
open-switch faults in one sub-module and also addressed the detection of multiple faults
in random sub-module elements. However, the linking effects between generator-side
and grid-side converters are not taken into account, which could affect considerably the
system behavior. The authors in [14] focused on simple faults in both converter sides. This
current work deals with faults in both converter sides, taking into consideration all possible
fault scenarios such as simple fault generator-side, simple fault grid-side, multiple fault
generator-side, multiple fault grid-side, and mix fault on both sides. Each scenario affects
the system behavior in a different way, accordingly, considering each of it is a crucial task.
Generally, FDD approaches can be categorized into two main classes: the model-based
and the data-driven methods. Model-based FDD uses observers and system identification
models of the processes; it demands a precise mathematical model, which is compli-
cated to acquire in reality. Its performance is dramatically impacted by uncertainties
and unmodeled noises [19,20]. Data-driven methods aim to extract information from the
measured signals to train the model, and then use the information for diagnosis in the
testing phase [21–23]. Numerous studies based on machine learning approaches have been
employed in WEC FDD, such as decision tree (DT) [24], naive Bayes (NB) [25], support
vector machine (SVM) [26], K-nearest neighbors (KNN) [27], and random forest (RF) [14].
In [2], a WEC fault diagnosis technique based on an RF and kernel principal component
analysis (KPCA) approach is developed. In this proposal, KPCA is applied to extract the
most informative features from data, with the aim of improving the classification results
using an RF classifier. In [24], the authors introduce five-stage statistical process control
and machine learning methods to diagnose wind turbine faults (rotary blades, gearboxes,
generators, and hydraulic oil systems) and predict maintenance demands. The five adopted
analytical tools in statistical process control are: (1) check lists, (2) Pareto charts, (3) cause
and effect diagrams, (4) scatter plots, and (5) control charts. Firstly, the check list comprises
information such as the type of wind turbine faults, the duration of faults, causes, and re-
pair events. Authors have classified the repair events by frequency of anomalies in the
dataset. Secondly, a Pareto chart is developed based on the classified check list items and
presents the repair events with regard to cumulative percentage. Thirdly, an analytical
tool, that is, the cause and effect diagram, is presented in order to distinguish the essential
causes of principal mechanical issues and produce recommendations to technicians for
maintenance. Fourthly, scatter plots are applied to investigate the relationship between
features and determine abnormal data. Lastly, control charts are applied to show changes
and variation in the observed data over time. After that, a density-based spatial clustering
of applications with noise (DBSCAN) approach is used to represent the relationship be-
tween the entire amount of wind generation and the five attributes, in addition to ranking
normal and abnormal data. Finally, two machine learning techniques—decision tree and
random forest—are applied in order to construct a predictive maintenance models for
anomalies. The inherent disadvantages of traditional ML-based approaches make them
ineffective at representing complex functions due to their unsatisfactory performance and
their generalization capabilities. With the explosion of deep learning (DL) algorithms in
artificial intelligence (AI) applications, technology has shown a strong ability to surpass
Energies 2022, 15, 6127 3 of 19

conventional intelligent algorithms [28], whose problems include their dependence on

hand-designed feature, as well as their difficulty in understanding sequential data. Thus,
many researchers have opted to use DL modes instead of traditional classifiers in fault
diagnosis. In fact, the major distinction between AI models and DL models is that the latter
can automatically learn precious features directly from raw data. Considering the rapid
rise of DL, many architecture have been developed, such as convolutional neural networks
(CNN), deep belief networks (DBN), and recurrent neural networks (RNN).
Authors in [29] propose an ensemble transfer CNN driven by multi-channel signals
for fault diagnosis of rotating machinery. In this case, modified CNNs based on stochastic
pooling and leaky rectified linear unit (LReLU) are pre-trained using multi-channel sig-
nals. Then, the target CNN is initialized using the learned parameter knowledge of each
individual source CNN with the help of parameter transfer. Lastly, in order to achieve the
comprehensive result, a new decision fusion procedure is constructed to flexibly fuse each
individual target CNN. An FDD approach based on the convolutional neural network long
short-term memory attention mechanism (CNN-LSTM-AM) for anomaly recognition and
fault detecting of wind turbine is suggested in [30]. The CNN is used to extract features of
state space from wind turbine, LSTM is applied to improve the time characteristics fusion
of different part states, and AM is used to help the model make more accurate judgments
through mapping weight and parameter learning. The authors of [31] propose an approach
to regularize the discriminant structure of the deep network with both intrinsic and ex-
trinsic generalization goals in order to improve the learning of robustness features and to
generalize to unseen domains. In [32], the authors develop an improved RNN techniques
for fault detection and diagnosis for wind energy conversion (WEC) systems. In the begin-
ning, a reduced RNN-based hierarchical K-means clustering is adopted in order to simplify
the complexity of the model in terms of training and computation time. It is used to treat
the correlations between samples and extract a reduced number of observations from the
training data matrix. Then, two reduced RNN-based interval-valued-data methods are
developed for classification purposes.
With the RNN, sequence inputs of variable length can be handled due to the recurrent
hidden states, whose activation at any particular time is dependent on that of the previous
moment. Other research proposes long short-term memory (LSTM) to directly learn features
and time-series data [33]. In fact, the recursive behavior of the LSTM gate architecture
allows it to capture long-term dependencies and efficiency figures without the gradient
vanishing problem of recurrent neural networks (RNNs) [34].
In the current work, we propose an innovative fault diagnosis paradigm using KPCA-
based BiLSTM. In fact, the previous studied LSTM-based fault diagnosis approaches were
applied directly to raw data without taking account the impact on the extracted and
selected features on the classification accuracy, as well as the nonlinear behavior of features.
To address these issues, a KPCA-based bidirectional LSTM (KPCA-based BiLSTM) FDD
approach is proposed to detect the faults and distinguish between the working modes in
the WTC systems. The KPCA model is able to deal with noisy, nonlinear, multivariate,
and statistical features [35]. In comparison to other nonlinear techniques, KPCA has
the advantages of not involving nonlinear optimization, requiring no prior specification
of reduced space dimensions, and being able to handle a wide range of nonlinearities
due to its ability to use different kernels [36]. Therefore, in this work, the KPCA feature
extraction/selection paradigm and the BiLSTM classification model are applied to detect
and classify the WTC faults. The proposed approach makes full use of the KPCA for
powerful feature extraction/selection and BiLSTM for fault diagnosis, which can solve the
problem of nonlinear, statistical, and multivariate feature extraction and fault diagnosis in
WTC systems.
This paper is organized as follows: Section 2 is dedicated to a brief description of
the KPCA tool used in feature extraction and selection and of the BiLSTM technique for
classification purposes. Section 3 presents the application of the developed methodology
for fault detection and diagnosis. Finally, the conclusions are illustrated in Section 4.
Energies 2022, 15, 6127 4 of 19

2. Model
2.1. Bidirectional LSTM Description
LSTM was derived from recurrent neural networks (RNN) in 1997 by Hochreiterand
Schmidhuber [28]. It was developed to tackle the vanishing gradient issue witnessed in
RNNs. Hence, to achieve this target, the architecture of the LSTM has three gates: the input
gate, forget gate, and output gate. Figure 1 illustrates the LSTM cell with input gate (it ),
forget gate ( f t ), and output gate (ot ), which are denoted by the following equations:

Figure 1. Architecture of LSTM.

The forget gate ( f t ) indicates what information of the previous state (Ct−1 ) will be
forgot or kept by looking at the values of the current input vector (xt ) and hidden state
(ht−1 ), as given in the following equation:
 
f t = σ W f [ h t −1 , x t ] + b f (1)

where W f and b f represent the weight matrix and the bias term, respectively.
In the same way and in order to update the cell state, the input gate (it ) decides how
much information from the input (xt ) and (ht−1 ) must pass, expressed as:

it = σ (Wi [ht−1 , xt ] + bi ) (2)

C̃t = tanh(Wc [ht−1 , xt ] + bc ) (3)

where C̃t denotes an immediate condition.
The updated state of the cell (when deciding which information to reserve and which
to forget) is presented as follows:

Ct = f t Ct−1 it C̃t (4)

where Ct represents the long term state and the symbol denotes element-wise vector
multiplication. The output gate (ot ) checks the flow of information from the current cell
state to the hidden state.
ot = σ (Wo [ht−1 , xt ] + b0 ), and (5)
ht = ot tanh(Ct ) (6)
where ht denotes the output. LSTM exists in several architectures [37] and it might be used
in the following forms: vanilla LSTM, stacked LSTM, CNN-LSTM, encoder–decoder LSTM,
and bidirectional LSTM. The last of these is the focus of this study.
In 2005, Graves and Schmidhuber developed the bidirectional LSTM by fusing the
BRNN with the LSTM cell. The sequential data have strong temporal dependencies in
machine disease monitoring systems [38]. Thus, it is important to take into consideration
the future situation [39]. Accordingly, the BiLSTM is an essential means of handling this
case. Figure 2 illustrates the general concept of BiLSTM architecture.
Energies 2022, 15, 6127 5 of 19

Figure 2. Architecture of BiLSTM.

The architecture for classification purposes is shown in Figure 3. It is composed of

an input layer, a BiLSTM layer followed by a fully connected layer, and a softmax layer at
the output.
BiLSTM can learn input in both directions: forward and backward. The forward LSTM
→ →
treats data from left to right and its hidden state can be expressed as ht = LSTM( xt , ht−1 ),
while the backward LSTM treats information in the opposite direction, and its hidden state
← ←
can be presented as ht = LSTM( xt , ht+1 ). Finally, concatenate the forward and backward
states to generate the BiLSTM output, as presented in the following equation:
→ ←
ht = [ ht , ht ] (7)

Take into consideration that the final hidden state h f encodes the most features from
the input signal and uses this as input to the fully connected layer, which aims to convert it
into a vector in which the length is equal to the class number. A softmax layer is approved
for fault classification. The probability distribution is given as:
 
Ỹ = softmax Ws h f + bs (8)

where Ws and bs indicate the weight and bias, respectively.

exp(zi )
softmax(zi ) = (9)
k
∑ exp(z j )
j =1

where zi is the ith element of the input vector z. The BiLSTM model is trained by minimizing
the error between the predicted Ỹ and actual Y.
Energies 2022, 15, 6127 6 of 19

Figure 3. BiLSTM architecture for fault classification.

2.2. Kernel Principal Component Analysis

KPCA extends conventional PCA to handle nonlinear data [40]. In fact, KPCA in-
cludes two main steps: (1) mapping the data into a higher dimensional feature space, and
(2) performing the linear PCA in that space.

2.2.1. KPCA-Based Feature Extraction

Consider the data matrix X = [ X1 X2 · · · Xm ] ∈ < N ×m , where N denotes the number
of samples and m represents the number of variables. The mapped data in the new feature
space is organized as follows:

X = [φ( x1 ) φ( x2 ) · · · φ( x N )] T ∈ < N ×h (10)

where h >> m is the dimension of the feature space. The following Eigenvector expression
is used to compute the kernel principal components (KPCs):

Kα = λα (11)

This equation indicates that α and λ are the Eigenvectors and Eigenvalues of the kernel
matrix K. The kernel matrix K is declared as:

K = XX T
···
 
k ( x1 , x1 ) k ( x1 , x N )
 . . 
  (12)
=
 . ··· . 

 . . 
k ( x N , x1 ) ··· k( x N , x N )

2.2.2. KPCA-Based Features Selection

The feature selection function intends to select the smallest group with most relevant
and expressive features by eliminating all the irrelevant and redundant features. The given
equation represents the kernel matrix Eigenvector [41]:

v = λ −1 X T α (13)
Energies 2022, 15, 6127 7 of 19

The matrix P = [v1 , . . . , v` ] denotes the matrix of the ` retained principal loading of
the KPCA in the feature space. Referring to Equation (2), the matrix P can defined as:

P = [ λ1 X T α1 , . . . , λ1 α` ]
" 1 ` #
−1/2 T ∗ −1/2 T ∗
= λ1 X α1 , . . . , λ ` X α` (14)
1
= X T P∗ Λ− /2

where P∗ = [α1∗ , . . . , α∗` ] and Λ = diag(λ1 , . . . , λ` ) are the ` principal Eigenvectors and
Eigenvalues of K, respectively.
Moreover, the kernel principal components are computed as:

t = Λ−1/2 P∗T k( x ) (15)

The selection of the number of kernel principal components (KPCs) has been subjected
to various studies; Ref. [42] details some of them. In this work, the cumulative percent
variance (CPV) criterion is used to select the first KPCs in the KPCA model. The features
extracted from the KPCA model are the first retained KPCs.

3. Proposed Approach and Case Study Experiment

3.1. Proposed Approach
The proposed methodology includes two major steps comprising feature selection and
extraction, and fault classification. The developed approach is discussed in such a way that
the kernel principal component analysis (KPCA) method is applied for feature extraction
goals and the BiLSTM classifier is used for fault diagnosis. The goal of this methodology
is to reduce the complexity of the proposed classifier. The first step of the proposed
methodology includes the gathering of WECC data. Then, the KPCA is applied to the data
in the interest of extracting and selecting the most effective and relevant features. In the
next step, the final features subset is considered as input to the BiLSTM tool to classify faults
and distinguish between the different operating mode. To summarize, the current paper
presents an intelligent fault diagnosis approach based on the KPCA model and the BiLSTM
classifier. In fact, the classical BiLSTM-based fault diagnosis techniques were previously
utilized directly on the raw data without considering the impact on the feature extraction
and selection phase in the diagnosis performance. To deal with these issues, a multivariate
KPCA-based bidirectional LSTM classifier approach is presented to detect and identify the
faults in WTC systems. In the developed FDD approach (so-called KPCA-based BiLSTM),
the KPCA model is applied to extract nonlinear, multivariate, and statistical features, and
BiLSTM is utilized for fault classification purposes. The proposed KPCA-based BiLSTM
approach makes full application of the KPCA for feature extraction/selection and BiLSTM
for fault classification, which can solve the problem of nonlinear, statistical and multivariate
feature extraction and fault diagnosis in WTC systems. The flowchart of the developed
KPCA-based BiLSTM algorithm is illustrated in Figure 4. Algorithm 1 shows the principal
steps of the developed KPCA-based BiLSTM algorithm.
Energies 2022, 15, 6127 8 of 19

Training data Testing data

Data pre-processing Data pre-processing

Kernel matrix computation

Covariance matrix computation Kernel vector computation

Eigen decomposition execution

Feature extraction and selection Feature extraction and selection

using KPCA model using KPCA model

Classify faults using BiLSTM Softmax classification

Classification model Predected model

Fault classification results

Figure 4. Architecture of the proposed approach.

Algorithm 1 KPCA-based BiLSTM Algorithm

Input: N × m data matrix Xi , i = 1, 2, . . . , n
Open-circuit
Training phase
Short-circuit
1. Standardize the training data set;
2. DetermineGate
the driver
KPCA model;
failure High voltage collapse
3. Extract and select the more relevant features using the KPCA model;
4. Classify the faults through the BiLSTM classifier;
5. Ascertain the classification model. Dynamic latch-up
Solder exhaust
Testing phase Second breakdown

1. Standardize the testing data set;

High temperature via
2. Calculate the kernel
Bond vector;
wire lift-off power dissipation,
3. Extract and select features using the KPCA Impactmodel;
ionization
4. Classify the faults through the BiLSTM classifier;
5. Establish the prediction model;
6. Achieve the fault diagnosis results.

3.2. System Description

In this paper, a variable speed wind turbine based on a squirrel cage induction genera-
tor (SCIG) is considered, as illustrated in Figure 5. This structure offers unlimited variable
speed operation. No matter the rotation speed of the machine, the voltage created is recti-
fied and converted into direct current and voltage. Accordingly, the grid-side converter
command assists in giving an alternating voltage with a constant frequency referring to
that of the grid. The maximum power generated by the turbine is determined by the
nominal power of the generator. For this configuration, the grid-side generator is based on
an insulated gate bipolar transistor (IGBT), the structure of which is the same as that of the
grid-side converter. The wind turbine parameters are shown in Table 1.
Energies 2022, 15, 6127 9 of 19

Figure 5. Variable speed wind turbine based on SCIG [14].

Table 1. Wind turbine parameters.

Parameters Nomenclature Values

Nominal power of turbine Ptn 15 kW
Moment of inertia of turbine Jt 1000 kgm2
Stator resistance Rs 0.087 Ohm
Stator leakage inductance Is 0.8 mH
Rotor resistance Rr 0.228 Ohm
Rotor resistance Rr 0.228 Ohm
Rotor leakage inductance Ir 0.8 mH
Magnetizing inductance Lm 34.7 mH
Number of poles P 4
Moment of inertia of generator Jg 0.2 kgm2

The power converters are a crucial component in WEC systems. The authors of [43]
proved that 21% of the faults in power converters are attributed to semiconductors (IGBT,
diode), as shown in Figure 6.

Figure 6. Common catastrophic failures of IGBT.

The usual faults in power switches involve two type of failures: wear-out failures and
catastrophic failures. The first type ensues from long time degradation, while catastrophic
faults generally happen due to one overstress incident. This paper concern only open-
circuits and short-circuits, which cause irretrievable harm to the converter system. In fact,
open-circuit faults of IGBT do not cause serious damage to the converter, but influence the
performance of the other-side converter and the feedbacks in the control loop. Figure 7
classifies IGBT catastrophic failures into open-circuit and short-circuit statuses arising from
various failure mechanisms.
 Classification model Predected model

Fault classification results

Energies 2022, 15, 6127 10 of 19

Open-circuit Short-circuit

Gate driver failure High voltage collapse

Dynamic latch-up
Solder exhaust
Second breakdown

High temperature via

Bond wire lift-off power dissipation,
Impact ionization

Figure 7. Common catastrophic failures of IGBT.

In the wind chain, the power converter topology exists on two levels. Each converter is
composed of three arms. Each arm includes a high and a low IGBT, (as shown in Figure 8).

Figure 8. Converter topology [14].

3.3. Data Collection

In order to construct a data base to perform FDD, a test bench must be designed under
realistic conditions. As detailed in Figure 9, the test setup should be positioned to stress the
IGBT modules as they would be in a real wind turbine application. For the sake of injecting
short-circuits and open-circuits, we add a controlled switch either in parallel or in serial.
This paper deals with several fault scenarios and each scenario comprises different cases,
as shown in Table 2.
• First scenario: This denotes simple faults that concern just one IGBT on the generator-
side converter (SFGS);
• Second scenario: This denotes simple faults that concern just one IGBT on the grid-
side converter (SFGrS);
• Third scenario, forth scenario: Practically, there can be more than one fault on the
same converter side; in this paper, we consider multiple faults on the generator side
(MFGS) and grid side (MFGrS) separately;
• Fifth scenario: In the real word, faults may happen on both the converter sides
simultaneously; for that reason, we consider mixed faults (MxF);
• Sixth scenario: In order to monitor the system in all its states, we combine all the
above scenarios.
Energies 2022, 15, 6127 11 of 19

Figure 9. Illustration of failure mode distribution.

Table 2. Construction of database for the fault diagnosis system.

Type of Fault Symbol Fault Scenario

Case 1: Short circuit (SC-A1HG) (IGBTA1HG ).
Simple fault
SFGS Case 2: Open circuit (OC-A2LG) (IGBTA2LG ).
generator-side
Case 3: Open circuit (OC-A3HG) (IGBTA3HG ).
Case 4: Open circuit (OC-A1LGr) (IGBTA1LGr ).
Simple fault
SFGrS Case 5: Short circuit (SC-A2HGr) (IGBTA2HGr ).
grid-side
Case 6: Short circuit (SC-A3LGr) (IGBTA3LGr ).
Case 7: Short circuit (SC-A1LG) (IGBTA1LG ) and Open circuit
(OC-A2LG) (IGBTA2LG ).
Multiple faults Case 8: Short circuit (IGBTA2LG ) and Open circuit (SC-A2LG
MFGS
generator-side and OC-A3HG) (IGBTA3HG ).
Case 9: Short circuit (IGBTA2HG ) and Short (SC-A2HG and
SC-A2HG) (IGBTA2HG ).
Case 10: Short circuit (IGBTA1LGr ) and Open circuit (SC-
A1LGr and OC-A2HGr) (IGBTA2HGr ).
Multiple faults Case 11: Open circuit (IGBTA1LGr ) and Short circuit (OC-
MFGrS
grid-side A1LGr and SC-A2HGr) (IGBTA2HGr ).
Case 12: Open circuit (IGBTA2LGr ) and Open circuit (OC-
A2LGr and OC-A3LGr) (IGBTA3LGr ).
Case 13: Short circuit (IGBTA1LG ) and Open circuit (SC-A1LG
and OC-A2HGr) (IGBTA2HGr ).
Case 14: Open circuit (IGBTA1HG ), Short circuit (IGBTA2LG )
Mixed fault and Short circuit (OC-A1HG, SC-A1LGr and SC-A2LG)
MxF
both sides (IGBTA1LGr ).
Case 15: Short circuit (IGBTA3LG ), Open circuit (IGBTA1HGr ),
Short circuit (IGBTA2LGr ) and Open circuit (SC_A3LG, OC-
A3HGr, OC-A1HGr, SC-A2LGr) (IGBTA3HGr ).

Figures 10 and 11 show the behavior of some electrical and mechanical variables in
different faulty cases.
Energies 2022, 15, 6127 12 of 19

Healthy case Faulty case 12 Faulty case 2

2000

Mechanical torque, Cm
1000

-1000
0 0.5 1 1.5 2 2.5 3 3.5
Sample time
4
10 Healthy case Faulty case 12 Faulty case 2
2
Output power, P out

-1

-2

0 0.5 1 1.5 2 2.5 3 3.5

Sample time 4
10

Figure 10. Input torque and output power for different cases.
Generator current phase a, isag

Healthy case Faulty case 12 Faulty case 2

500

-500
0 0.5 1 1.5 2 2.5 3 3.5
Sample time 10 4
Healthy case Faulty case 12 Faulty case 2
500
Grid current phase a, i sar

-500
0 0.5 1 1.5 2 2.5 3 3.5
Sample time 10 4

Figure 11. Generator current and grid current for different cases.

Figures 10 and 11 clearly demonstrate that faults do not affect the system behavior in
the same way. In fact, some fault scenarios do not significantly affect the behavior of the
system, in which case service can be maintained until the fault is isolated, as illustrated in
Case 12. For example, the output power in healthy mode is almost constant, while when
the fault is injected, the same level of power is found with some oscillations. Other types of
faults that considerably affect the behavior of the system are considered serious. In Case 2,
for instance, the generator current reaches around 500A, which is an insupportable current
for the system, and in this situation the system must be taken out of service immediately.
Energies 2022, 15, 6127 13 of 19

4. Results and Discussions

4.1. Performance Metrics
In order to evaluate and compare performance, the approved criteria are: accuracy
(%), which indicates the rate of observations correctly predicted over the total number
of observations; recall (%), which indicates, in the pertinent class, the rate of positive
observations correctly predicted to observations; precision (%), which indicates the number
of positive observations correctly predicted divided by the number of total predicted
positive observations; F1 score (%), which indicates the weighted average of precision
and recall; and computation time (CT(s)), which represents the time required to carry out
the algorithm.
TP + TN
Accuracy = (16)
TP + TN + FP + FN
TP
Recall = (17)
TP + FN
TP
Precision = (18)
TP + FP
precision × recall
F1 Score = 2 × (19)
precision + recall
where TP (true positive) is properly classified positive observations, FP (false positive)
is mis-classified positive observations , TN (true negative) is correctly classified negative
samples, and FN (false negative) is misclassified negative observations.

4.2. Parameters Setting

In this work, the 95% cumulative variance criterion is applied to select the retained
KPCs where 32 KPCs are maintained. Sampling noise can appear during the training
process due to the complex relationships among inputs and outputs of neural networks,
leading to overfitting, which decreases the predictive capability of the model [44]. In order
to avoid this issue, optimal hyperparameters are used in this paper (as shown in Table 3),
such as the Adam optimization algorithm, which is used in order to decrease the error in
each iteration. Actually, Adam exceeds other optimization algorithms due to its relatively
low memory requirement [45], as well by using dropout, which is a method that evades
extracting same features over and over again to reduce the risk of overfitting [46]. For the
NN, FFNN, CFNN, and RNN classifiers, the number of selected hidden layers is equal
to 10 and the number of neurons in the hidden layers is 50. For the CNN classifier, we
used a convolution layer, ReLU function, pooling layer, fully connected layer, and softmax
layer. Furthermore, to train the neural network, CNN uses the cross-entropy loss function.
Moreover, Adam optimization algorithm is applied.

Table 3. Hyperparameter settings.

Hyperparameters Values
Optimizer Adam
Loss function Cross-entropy
Dropout 0.2
Learning rate 0.001
Regularizer L2
Maximum epochs 20
Mini-batch size 250
BiLSTM layer nodes 50

4.3. Fault Classification Results

For the purpose of performing the different experiments for FDD purposes, ten vari-
able measurements are gathered, as listed in Table 4. These variables represent 1 healthy (at-
Energies 2022, 15, 6127 14 of 19

tached to class C0 ) and 15 faulty operating conditions of WECC (attached to Ci ; i = 1, . . . , 15),

as shown in Table 5. Each mode behavior is adequately described over 2000 10-spaced
samples, with 20 KHz as the sampling frequency for the training phase. We used of the
samples 80% for the training phase and 20% for the testing phase.

Table 4. Labeling and description of the measured and monitored system variables.

Variables Descriptions
x1 Cm : Mechanical torque ( Nm )
x2 Ng : Generator speed (tr/m)
x3 isag : Generator current phase a ( A)
x4 isbg : Generator current phase b ( A)
x5 iscg : Generator current phase c ( A)
x6 VDC : Bus voltage (V )
x7 Pout : Output power (W )
x8 isar : Grid current phase a ( A)
x9 isbr : Grid current phase b ( A)
x10 isbr : Grid current phase b ( A)

Table 5. Creation of database for fault diagnosis system.

Classes Mode Training Data Testing Data

C0 Healthy 10,000 2500
C1 10,000 2500
C2 SFGS 10,000 2500
C3 10,000 2500
C4 10,000 2500
C5 SFGrS 10,000 2500
C6 10,000 2500
C7 10,000 2500
C8 MxF 10,000 2500
C9 10,000 2500
C10 10,000 2500
C11 MFGS 10,000 2500
C14 10,000 2500
C12 10,000 2500
C13 MFGrS 10,000 2500
C15 10,000 2500

In this paper, various classifiers are applied and the best classifier is selected on the
basis of its classification accuracy. Table 6 illustrates the global performance accuracy.
In Scenario 1, the faults occur in the grid-side converter, which do not seriously affect
the behavior of the wind system. In this case, all the developed techniques have showed
high diagnosis performance except for CNN. However, in Scenario 2, different faults are
presented in the generator-side converter that considerably affect the behavior of the system.
This affects the diagnosis performance of the applied FDD techniques. In cases of Scenarios
3, 4, and 5, the FDD techniques showed good results, with the exception of CNN. When
dealing with all fault scenarios, it is clear from Tables 6 and 7 that the BiLSTM classifier
provides better classification performance when compared to the classical methods.
Energies 2022, 15, 6127 15 of 19

Table 6. Performance comparison of conventional techniques.

Global Performance
Fault Side Techniques
Accuracy Recall Precision F1 Score CT (s) PL
SFGS 72.26 71.87 64.82 68.6 0.21 **
SFGrS 96.94 96.94 96.95 96.94 0.18 ***
MFGS NN 87.41 88.22 88.54 88.38 0.20 ***
MFGrS 97.16 97.39 97.88 97.63 0.21 ****
MxF 93.79 94.4 95.42 94.71 0.18 ***
All faults 59.62 58.53 59.87 59.19 0.35 *
SFGS 76.94 76.94 78.89 77.90 0.15 **
SFGrS 87.89 87.89 89.74 88.80 0.18 ***
MFGS FFNN 84.81 84.81 88.35 86.54 0.20 ***
MFGrS 85.35 85.35 85.59 85.47 0.14 ***
MxF 95.14 95.14 95.93 95.53 0.15 ***
All faults 49.46 45.72 44.71 45.21 0.34 *
SFGS 75.11 75.11 75.3 75.20 0.18 **
SFGrS 96.75 96.75 96.79 96.67 0.18 ***
MFGS CFNN 90.46 90.46 91.32 90.88 0.14 ***
MFGrS 87.78 87.78 87.81 87.79 0.24 ***
MxF 95.18 95.18 95.96 95.57 0.15 ***
All faults 59.87 59.56 56.75 58.12 0.18 *
SFGS 70.07 70.04 61.14 65.25 0.15 ****
SFGrS 95.73 95.73 95.73 95.73 0.17 ***
MFGS RNN 80.56 80.56 75.86 78.14 0.17 ***
MFGrS 86.23 86.23 87.82 87.2 0.14 ***
MxF 94.23 94.47 95.47 95.47 0.17 ***
All faults 47.50 47.55 40.43 43.65 0.34 *

Table 7. Performance comparison of deep learning techniques.

Global Performance
Fault Side Techniques
Accuracy Recall Precision F1 Score CT (s) PL
SFGS 41.16 40.33 60.43 48.37 0.95 *
SFGrS 44.06 44.06 62.01 51.52 0.74 *
MFGS CNN 52.85 52.85 53.10 52.97 0.7 *
MFGrS 38.18 38.18 38.74 38.46 0.77 *
MxF 42.55 42.55 44.89 43.68 0.95 *
All faults 16.43 14.39 11.98 13.07 1.29 *
SFGS 75.08 75.08 75.21 75.14 0.71 **
SFGrS 86.90 86.9 88.96 87.92 0.55 ***
MFGS LSTM 88.40 88.40 89.48 88.94 0.67 ***
MFGrS 91.83 91.83 93.18 92.50 0.59 ***
MxF 100.0 100.0 100.0 100.0 0.81 ****
All faults 73.70 73.58 73.70 73.64 1.32 **
SFGS 72.92 65.77 72.5 72.60 0.78 **
SFGrS 95.71 95.71 95.72 88.24 0.7 ***
MFGS BiLSTM 88.49 88.49 89.66 89.07 0.64 ***
MFGrS 90.71 90.71 93.19 91.93 0.78 ***
MxF 100.0 100.0 100.0 100.0 0.75 ****
All faults 79.0 79.54 81.34 81.34 1.62 **
Energies 2022, 15, 6127 16 of 19

In order to improve further the above results, a novel FDD approach is proposed using
a KPCA-based BiLSTM, in which the most informative features are extracted and selected
using KPCA and then fed to the BiLSTM for fault classification purposes.
As shown in Table 8, the developed KPCA-BiLSTM approach reached an accuracy
rate of 97.20%. This result demonstrates its enhanced classification performance when
compared to the standard BiLSTM.

Table 8. Performance comparison of different techniques.

Global Performance
Fault Side Techniques
Accuracy Recall Precision F1 Score CT (s) PL
All faults BiLSTM 79.0 79.54 81.34 80.43 1.62 **
All faults KPCA-BiLSTM 97.20 97.20 97.30 97.25 2.56 ****

To better assess the efficiency of the proposed approach, the testing classification results
are illustrated in Table 9 using the confusion matrix (CM). The CM illustrates the correctly
classified samples and misclassified ones for the healthy case (C0 ) and faulty cases (C1 to
C15 ). For example, for the healthy case (C0 ), the KPCA-based BiLSTM approach determined
2320 observations among 2500 (true positive). For this class, 7.2% were misclassifications
(false alarms). In the faulty operating modes (C5 , C8 , C9 , C11 , C12 , C13 ), the precision was
100% and the recall was 100%, with 0.0% misclassification.

Table 9. Confusion matrix of KPCA-based BiLSTM in testing phase.

Predicted Classes
True Classes C0 C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 C15 Recall
C0 2320 0 0 0 110 0 0 0 0 0 0 0 0 0 0 0 92.80
C1 23 2308 23 32 0 2 0 105 0 7 0 0 0 0 0 0 92.32
C2 0 0 2387 0 8 22 22 5 2 0 41 13 0 0 0 0 95.48
C3 9 4 0 2428 0 18 0 0 0 0 0 5 8 9 0 19 97.12
C4 129 0 0 0 2309 0 0 0 0 0 0 0 0 0 0 62 92.36
C5 0 0 0 0 0 2500 0 0 0 0 0 0 0 0 0 0 100.0
C6 3 0 0 0 0 0 2487 0 0 0 0 0 0 0 0 10 99.48
C7 213 0 0 0 14 1 0 2185 0 0 0 0 0 0 0 87 87.40
C8 0 0 0 0 0 0 0 0 2500 0 0 0 0 0 0 0 100.0
C9 0 0 0 0 0 0 0 0 0 2500 0 0 0 0 0 0 100.0
C10 0 3 0 0 0 0 0 0 0 0 2497 0 0 0 0 0 99.88
C11 0 0 0 0 0 0 0 0 0 0 0 2500 0 0 0 0 100.0
C12 0 0 0 0 0 0 0 0 0 0 0 0 2500 0 0 0 100.0
C13 0 0 0 0 0 0 0 0 0 0 0 0 0 2500 0 0 100.0
C14 0 0 0 0 0 0 0 0 0 0 0 0 1 0 2499 0 99.96
C15 29 0 0 0 11 0 0 0 0 0 0 0 0 0 0 2500 100.0
Precision 85.10 99.69 99.04 98.69 94.16 98.30 99.12 95.20 99.92 99.72 98.38 99.28 99.64 99.64 100 93.35 97.30

5. Conclusions and Future Works

In this paper, an enhanced KPCA-based BiLSTM method was presented for wind
energy conversion (WEC) system fault detection and diagnosis (FDD). The proposed FDD
approach was addressed in such a way that the extracted and selected features using the
KPCA model are introduced as input for the BiLSTM for classification purposes. In fact,
the effectiveness of the proposed classifier was validated by comparing it with several
classical methods, including NN, FFNN, CFNN, RNN, and CNN. In order to evaluate the
performance of the developed KPCA-based BiLSTM approach, we used data obtained from
healthy WEC converters (WECC) that were then injected with several fault scenarios of fault:
simple fault generator-side, simple fault grid-side, multiple faults generator-side, multiple
faults grid side, and mixed faults both side. The obtained results showed the effectiveness
and robustness of the proposed FDD approach in terms of accuracy, recall, precision, and
computation time. The fault diagnosis accuracy when using the proposed tools showed
Energies 2022, 15, 6127 17 of 19

some missed detection and false alarm results, and some faults were not correctly classified.
Thus, one future research direction is to develop adaptive BiLSTM-based tools to update
the model in order to reduce missed classification results. Another future direction is to
develop adaptive BiLSTM-based approaches dealing with uncertainties in WTC systems
using interval-valued data representation. Additionally, ensemble-based models will be
developed using multiple models in order to enhance decision-making accuracy. Ensemble-
based models merges multiple learning models in order to produce one optimal predictive
model that gives effective diagnosis results. Furthermore, in this study, we considered a
wind profile where the mean value of the speed, as well as the pitch angle, is constant.
In the real world, the wind has a variable profile according to climatic conditions. Thus,
one future research direction is to implement an FDD approach while taking into account
wind variations.

Author Contributions: Investigation, K.A.; Methodology, Z.Y., M.H. and M.M.; Supervision, M.H.,
M.M., K.B. and H.N.; Validation, M.H. All authors have read and agreed to the published version of
the manuscript.
Funding: This work was supported by the Qatar National Library through the Qatar National
Research Fund (QNRF) Research Grant.
Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations

FDD Fault Detection and Diagnosis

FES Feature Extraction and Selection
PCA Principal Component Analysis
KPCA Kernel PCA
LSTM Long Short Term Memory
BiLSTM Bidirectional LSTM
CNN Convolutional Neural Network
WEC Wind Energy Conversion
WECC WEC Converters
WT Wind Turbine
CPV Cumulative Percentage of Variance
CT Computation Time
CM Confusion Matrix

References
1. Kouadri, A.; Hajji, M.; Harkat, M.F.; Abodayeh, K.; Mansouri, M.; Nounou, H.; Nounou, M. Hidden Markov model based
principal component analysis for intelligent fault diagnosis of wind energy converter systems. Renew. Energy 2020, 150, 598–606.
[CrossRef]
2. Fezai, R.; Dhibi, K.; Mansouri, M.; Trabelsi, M.; Hajji, M.; Bouzrara, K.; Nounou, H.; Nounou, M. Effective Random Forest-Based
Fault Detection and Diagnosis for Wind Energy Conversion Systems. IEEE Sens. J. 2020, 21, 6914–6921. [CrossRef]
3. Faulstich, S.; Hahn, B.; Tavner, P.J. Wind turbine downtime and its importance for offshore deployment. Wind Energy 2011,
14, 327–337. [CrossRef]
4. Zhang, Y.; Zheng, H.; Liu, J.; Zhao, J.; Sun, P. An anomaly identification model for wind turbine state parameters. J. Clean. Prod.
2018, 195, 1214–1227. [CrossRef]
5. Ravikumar, K.; Subbiah, R.; Ranganathan, N.; Bensingh, J.; Kader, A.; Nayak, S.K. A review on fatigue damages in the wind
turbines: Challenges in determining and reducing fatigue failures in wind turbine blades. Wind Eng. 2020, 44, 434–451. [CrossRef]
6. Mishnaevsky, L. Root Causes and Mechanisms of Failure of Wind Turbine Blades: Overview. Materials 2022, 15, 2959. [CrossRef]
7. Singh, G.; Sundaram, K. Methods to improve wind turbine generator bearing temperature imbalance for onshore wind turbines.
Wind Eng. 2021, 46, 0309524X211015292. [CrossRef]
8. Xu, Y.; Nascimento, N.M.M.; de Sousa, P.H.F.; Nogueira, F.G.; Torrico, B.C.; Han, T.; Jia, C.; Rebouças Filho, P.P. Multi-sensor
edge computing architecture for identification of failures short-circuits in wind turbine generators. Appl. Soft Comput. 2021,
101, 107053. [CrossRef]
9. Xiao, C.; Liu, Z.; Zhang, T.; Zhang, X. Deep learning method for fault detection of wind turbine converter. Appl. Sci. 2021,
11, 1280. [CrossRef]
Energies 2022, 15, 6127 18 of 19

10. López-Uruñuela, F.J.; Fernández-Díaz, B.; Pagano, F.; López-Ortega, A.; Pinedo, B.; Bayón, R.; Aguirrebeitia, J. Broad review of
“White Etching Crack” failure in wind turbine gearbox bearings: Main factors and experimental investigations. Int. J. Fatigue
2021, 145, 106091. [CrossRef]
11. Wang, L.; Zhang, Z.; Long, H.; Xu, J.; Liu, R. Wind turbine gearbox failure identification with deep neural networks. IEEE Trans.
Ind. Inform. 2016, 13, 1360–1368. [CrossRef]
12. Dao, P.B.; Staszewski, W.J.; Barszcz, T.; Uhl, T. Condition monitoring and fault detection in wind turbines based on cointegration
analysis of SCADA data. Renew. Energy 2018, 116, 107–122. [CrossRef]
13. Qiao, W.; Lu, D. A survey on wind turbine condition monitoring and fault diagnosis—Part I: Components and subsystems. IEEE
Trans. Ind. Electron. 2015, 62, 6536–6545. [CrossRef]
14. Mansouri, M.; Fezai, R.; Trabelsi, M.; Nounou, H.; Nounou, M.; Bouzrara, K. Reduced Gaussian process regression based random
forest approach for fault diagnosis of wind energy conversion systems. IET Renew. Power Gener. 2021, 15, 3612–3621. [CrossRef]
15. Kamel, T.; Biletskiy, Y.; Chang, L. Fault diagnoses for industrial grid-connected converters in the power distribution systems.
IEEE Trans. Ind. Electron. 2015, 62, 6496–6507. [CrossRef]
16. Kamel, T.; Biletskiy, Y.; Chang, L. Real-time diagnosis for open-circuited and unbalance faults in electronic converters connected
to residential wind systems. IEEE Trans. Ind. Electron. 2015, 63, 1781–1792. [CrossRef]
17. Jlassi, I.; Estima, J.O.; El Khil, S.K.; Bellaaj, N.M.; Cardoso, A.J.M. Multiple open-circuit faults diagnosis in back-to-back converters
of PMSG drives for wind turbine systems. IEEE Trans. Power Electron. 2014, 30, 2689–2702. [CrossRef]
18. Duan, P.; Xie, K.g.; Zhang, L.; Rong, X. Open-switch fault diagnosis and system reconfiguration of doubly fed wind power
converter used in a microgrid. IEEE Trans. Power Electron. 2010, 26, 816–821. [CrossRef]
19. Abid, A.; Khan, M.T.; Iqbal, J. A review on fault detection and diagnosis techniques: Basics and beyond. Artif. Intell. Rev. 2021,
54, 3639–3664. [CrossRef]
20. Zeroual, A.; Harrou, F.; Sun, Y.; Messai, N. Integrating model-based observer and Kullback–Leibler metric for estimating and
detecting road traffic congestion. IEEE Sens. J. 2018, 18, 8605–8616. [CrossRef]
21. Nai-Zhi, G.; Ming-Ming, Z.; Bo, L. A data-driven analytical model for wind turbine wakes using machine learning method.
Energy Convers. Manag. 2022, 252, 115130. [CrossRef]
22. Simani, S.; Farsoni, S.; Castaldi, P. Data-driven techniques for the fault diagnosis of a wind turbine benchmark. Int. J. Appl. Math.
Comput. Sci. 2018, 28, 247–268. [CrossRef]
23. Pashazadeh, V.; Salmasi, F.R.; Araabi, B.N. Data driven sensor and actuator fault detection and isolation in wind turbine using
classifier fusion. Renew. Energy 2018, 116, 99–106. [CrossRef]
24. Hsu, J.Y.; Wang, Y.F.; Lin, K.C.; Chen, M.Y.; Hsu, J.H.Y. Wind turbine fault diagnosis and predictive maintenance through
statistical process control and machine learning. IEEE Access 2020, 8, 23427–23439. [CrossRef]
25. Wu, Y.; Fu, Z.; Fei, J. Fault diagnosis for industrial robots based on a combined approach of manifold learning, treelet transform
and Naive Bayes. Rev. Sci. Instrum. 2020, 91, 015116. [CrossRef]
26. Tuerxun, W.; Chang, X.; Hongyu, G.; Zhijie, J.; Huajian, Z. Fault diagnosis of wind turbines based on a support vector machine
optimized by the sparrow search algorithm. IEEE Access 2021, 9, 69307–69315. [CrossRef]
27. Lu, J.; Qian, W.; Li, S.; Cui, R. Enhanced K-nearest neighbor for intelligent fault diagnosis of rotating machinery. Appl. Sci. 2021,
11, 919. [CrossRef]
28. Jalayer, M.; Orsenigo, C.; Vercellis, C. Fault detection and diagnosis for rotating machinery: A model based on convolutional
LSTM, Fast Fourier and continuous wavelet transforms. Comput. Ind. 2021, 125, 103378. [CrossRef]
29. He, Z.; Shao, H.; Zhong, X.; Zhao, X. Ensemble transfer CNNs driven by multi-channel signals for fault diagnosis of rotating
machinery cross working conditions. Knowl.-Based Syst. 2020, 207, 106396. [CrossRef]
30. Xiang, L.; Wang, P.; Yang, X.; Hu, A.; Su, H. Fault detection of wind turbine based on SCADA data analysis using CNN and
LSTM with attention mechanism. Measurement 2021, 175, 109094. [CrossRef]
31. Han, T.; Li, Y.F.; Qian, M. A hybrid generalization network for intelligent fault diagnosis of rotating machinery under unseen
working conditions. IEEE Trans. Instrum. Meas. 2021, 70, 1–11. [CrossRef]
32. Mansouri, M.; Dhibi, K.; Hajji, M.; Bouzara, K.; Nounou, H.; Nounou, M. Interval-Valued Reduced RNN for Fault Detection and
Diagnosis for Wind Energy Conversion Systems. IEEE Sens. J. 2022. [CrossRef]
33. Mansouri, M.; Trabelsi, M.; Nounou, H.; Nounou, M. Deep learning based fault diagnosis of photovoltaic systems: A comprehen-
sive review and enhancement prospects. IEEE Access 2021, 9, 126286–126306. [CrossRef]
34. Lei, J.; Liu, C.; Jiang, D. Fault diagnosis of wind turbine based on Long Short-term memory networks. Renew. Energy 2019,
133, 422–432. [CrossRef]
35. Lee, J.M.; Yoo, C.; Choi, S.W.; Vanrolleghem, P.A.; Lee, I.B. Nonlinear process monitoring using kernel principal component
analysis. Chem. Eng. Sci. 2004, 59, 223–234. [CrossRef]
36. Zhang, Y. Enhanced statistical analysis of nonlinear processes using KPCA, KICA and SVM. Chem. Eng. Sci. 2009, 64, 801–811.
[CrossRef]
37. Thara, D.; PremaSudha, B.; Xiong, F. Epileptic seizure detection and prediction using stacked bidirectional long short term
memory. Pattern Recognit. Lett. 2019, 128, 529–535.
38. Zhao, R.; Yan, R.; Wang, J.; Mao, K. Learning to monitor machine health with convolutional bi-directional LSTM networks.
Sensors 2017, 17, 273. [CrossRef]
Energies 2022, 15, 6127 19 of 19

39. Faust, O.; Shenfield, A.; Kareem, M.; San, T.R.; Fujita, H.; Acharya, U.R. Automated detection of atrial fibrillation using long
short-term memory network with RR interval signals. Comput. Biol. Med. 2018, 102, 327–335. [CrossRef]
40. Hoffmann, H. Kernel PCA for novelty detection. Pattern Recognit. 2007, 40, 863–874. [CrossRef]
41. Dhibi, K.; Mansouri, M.; Bouzrara, K.; Nounou, H.; Nounou, M. An Enhanced Ensemble Learning-Based Fault Detection and
Diagnosis for Grid-Connected PV Systems. IEEE Access 2021, 9, 155622–155633. [CrossRef]
42. Valle, S.; Li, W.; Qin, S.J. Selection of the number of principal components: The variance of the reconstruction error criterion with
a comparison to other methods. Ind. Eng. Chem. Res. 1999, 38, 4389–4401. [CrossRef]
43. Yang, Z.; Chai, Y. A survey of fault diagnosis for onshore grid-connected converter in wind energy conversion systems. Renew.
Sustain. Energy Rev. 2016, 66, 345–359. [CrossRef]
44. Hameed, Z.; Garcia-Zapirain, B. Sentiment classification using a single-layered BiLSTM model. IEEE Access 2020, 8, 73992–74001.
[CrossRef]
45. Aslan, M.F.; Unlersen, M.F.; Sabanci, K.; Durdu, A. CNN-based transfer learning–BiLSTM network: A novel approach for
COVID-19 infection detection. Appl. Soft Comput. 2021, 98, 106912. [CrossRef]
46. Xia, M.; Li, T.; Xu, L.; Liu, L.; De Silva, C.W. Fault diagnosis for rotating machinery using multiple sensors and convolutional
neural networks. IEEE/ASME Trans. Mechatron. 2017, 23, 101–110. [CrossRef]