8.-Wavelet Decomposition and Convolutional LSTM Solar Irradiance Forecasting-2018

Article
Wavelet Decomposition and Convolutional LSTM

Networks Based Improved Deep Learning Model for
Solar Irradiance Forecasting
Fei Wang 1,2,3,*, Yili Yu 2, Zhanyao Zhang 2, Jie Li 2, Zhao Zhen 2,* and Kangping Li 2
1 State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources,
North China Electric Power University, Baoding 071003, China
2 Department of Electrical Engineering, North China Electric Power University, Baoding 071003, China;
[email protected] (Y.Y.); [email protected] (Z.Z.); [email protected] (J.L.);

[email protected] (K.L.)
3 Hebei Key Laboratory of Distributed Energy Storage and Micro-grid, North China Electric Power
University, Baoding 071003, China

* Correspondence: [email protected] (F.W.); [email protected] (Z.Z.);
Tel.: +86-139-0312-5055 (F.W.); +86-136-6339-9230 (Z.Z.)
Received: 8 July 2018; Accepted: 24 July 2018; Published: 1 August 2018
Abstract: Solar photovoltaic (PV) power forecasting has become an important issue with regard to
the power grid in terms of the effective integration of large-scale PV plants. As the main influence
factor of PV power generation, solar irradiance and its accurate forecasting are the prerequisite for
solar PV power forecasting. However, previous forecasting approaches using manual feature
extraction (MFE), traditional modeling and single deep learning (DL) models could not satisfy the
performance requirements in partial scenarios with complex fluctuations. Therefore, an improved
DL model based on wavelet decomposition (WD), the Convolutional Neural Network (CNN), and
Long Short-Term Memory (LSTM) is proposed for day-ahead solar irradiance forecasting. Given
the high dependency of solar irradiance on weather status, the proposed model is individually
established under four general weather type (i.e., sunny, cloudy, rainy and heavy rainy). For certain
weather types, the raw solar irradiance sequence is decomposed into several subsequences via
discrete wavelet transformation. Then each subsequence is fed into the CNN based local feature
extractor to automatically learn the abstract feature representation from the raw subsequence data.
Since the extracted features of each subsequence are also time series data, they are individually
transported to LSTM to construct the subsequence forecasting model. In the end, the final solar
irradiance forecasting results under certain weather types are obtained via the wavelet
reconstruction of these forecasted subsequences. This case study further verifies the enhanced
forecasting accuracy of our proposed method via a comparison with traditional and single DL
models.
Keywords: solar irradiance forecasting; wavelet decomposition; convolutional neural network;

recurrent neural network; long short term memory
1. Introduction
1.1. Background and Motivation

With the global attention to environmental issues, the solar photovoltaic (PV) power has been
increasingly regarded as an important kind of renewable energy used to supply clean energy for the
power grid [1]. Nearly 60% of power generated in 2040 is projected to come from renewables, which
Appl. Sci. 2018, 8, 1286; doi:10.3390/app8081286 www.mdpi.com/journal/applsci

Appl. Sci. 2018, 8, 1286 2 of 30
wind and solar PV accounts for more than 50%. Additionally, International Energy Agency (IEA)
reported that the installed solar PV capacity has already reached more than 300 GW by the end of
2016 [2]. The annual market of solar PV power has increased by nearly 50%. The top five countries,
led by China, accounted for 85% of additions [3]. The above phenomena verified that solar PV power
was the world’s leading source of renewables in 2016.
However, the high dependence of solar PV power on geographical locations and weather
conditions can lead to the dynamic volatility and randomness characteristics of solar PV output
power. This unavoidable phenomenon makes PV power forecasting become an important challenge
for the power grid in terms of the effective integration of large-scale PV plants, because accurate solar
PV power forecasting can provide expected future PV output power, which provides good guidance
for the system operator to design a rational dispatching scheme and maintain the balance between
supply and demand sides. At the same time, scheduling PV power and other power reasonably may
be helpful for effectively addressing the problems, such as system stability and electric power balance
[4]. Therefore, accurate solar PV forecasting is essential for the sustainable and stable operation of the
whole power system.
In the actual PV stations, its final PV output is affected by a variety of meteorological factors,
such as solar irradiance [5], moisture, ambient temperature, wind velocity and barometric pressure.
There are two categories of the existing PV forecasting approaches: direct forecast and step-wise
forecast. Direct forecast creates a map between historical power data and power forecast values [6,7].
Differently, the step-wise forecast is comprised of two steps. In the first step, each meteorological
factor is predicted at the target time. In the next step, these predicted meteorological factors are then
utilized to create a map that can reflect the relationship between these meteorological factors and PV
power forecast value. In sum, the reliable information of the relevant meteorological factors is the
key to PV power forecasting. Therefore, as the main influence factor of PV power generation, the
solar irradiance and its accurate forecasting are the prerequisite for solar PV power forecasting.
1.2. Literature Review

With the fast advancement of forecasting theories [8,9], solar physics [10], stochastic learning
[11], and machine learning [12], the relevant technology of the solar irradiance forecasting research
area has also developed rapidly. In general, the existing various forecasting models are
correspondingly designed for solar PV prediction with different time horizon. For example, the
forecasting horizon of Numerical Weather Prediction (NWP) forecasting models is from several
hours to several days [13]. Time series forecasting models generate forecast outputs with a time scale
that ranges from 5 min to 6 h [14]. Statistical forecasting models based on cloud motion images and
satellite information can generate PV forecast value with a time sclerosis of 6 h [15]. In this paper, we
focus on day-ahead solar irradiance forecasting which the forecasting horizon is 24 h.
Among the previous studies, solar irradiance forecasting approaches can be generally divided
into several categories: statistical approaches, physical approaches and machine learning approaches
and ensemble approaches. In physical approaches, three kinds of basic methods are NWP forecasting
model [16], Total Sky Imagery (TSI) [17] and cloud moving based satellite imagery models, which
can also help to estimate the output power of distributed PV system [18]. These kinds of physical
based forecasting models require additional information about the sky image.
As for the statistical approaches, persistence forecasting, time series, and Model Output Statistics
(MOS) models [19] are involved. In this model, it is supposed that the forecasting data at time t+1 is
equal to the historical data at time t.
Time series approaches primarily aim at the modeling of long-term solar irradiance forecast,
which includes Moving Average (MA), Autoregressive (AR) [20], Autoregressive Moving Average
(ARMA) [21], and Autoregressive Integrated Moving Average (ARIMA) [22] models. The time series
forecasting model only requires historical irradiance data, in which the relevant meteorological
factors are not involved. In addition, time series approaches can merely capture linear relationships
and require stationary input data or stationary differencing data.
Appl. Sci. 2018, 8, 1286 3 of 30
In recent years, machine learning based forecasting methods have also been successfully applied
in many fields [23–26]. Machine learning models that have been done widely applied in solar
forecasting field are non-linear regression models such as Artificial Neural Network (ANNs) [27,28],
the Support Vector Machine (SVM) [29], and the Markov chain [30]. These nonlinear regression
models are also frequently used together with the classification models [31].
Regarding the ensemble approach, this kind of integrated model consists of multiple trained
forecasting sub-models. Additionally, all the outputs of these forecasting sub-models are taken into
consideration to determine the best output of the ensemble model. This method can well leverage the
advantages of different forecasting sub-models to achieve the performance optimization of the
ensemble model to provide better forecasting results for application [32,33].
Based on the abovementioned forecasting theories, many researchers have carried out important
research work in the field of solar irradiance forecasting and PV power forecasting (both referred to
as “solar forecasting” in what follows). Considering this abundant literature on solar forecasting,
Yang et al. [34] have conducted an adequate literature review work on the history and trends in solar
irradiance and PV power forecasting through text mining. Furthermore, Wan et al. [35] have also
reviewed the state-of-the-art of PV and solar forecasting methodologies developed over the past
decade. Regarding the forecasting of grid-connected photovoltaic plant production, Ferlito et al. [36]
implemented a comparative analysis of eleven forecasting data-driven models online and offline. The
above eleven models include: (1) simple linear models, such as Multiple Linear Regression; (2)
nonlinear models, such as Extreme Learning Machines and weighted k-Nearest Neighbors; and (3)
ensemble methods, such as Random Forests and Extreme Gradient Boosting. To improve real-time
control performance and reduce possible negative impacts of PV systems, Yang et al. [37] proposed
a weather-based hybrid method for 1-day ahead hourly forecasting of PV power output with the
application of Self-organizing Map (SOM), Learning Vector Quantization (LVQ) and Support Vector
Regression (SVR). Gensler et al. [38] used auto-encoder to reduce the dimension of historical data,
and employed LSTM to forecast solar power.
In the field of solar forecasting, a few researchers have also paid attention to the prediction of
solar irradiance due to its important influence on PV power output. For example, Hussain et al. [39]
applied a simple and linear statistical forecasting technique named ARIMA to day ahead hourly
forecast of solar irradiance for Abu Dhabi, UAE. In another relevant study, five novel semi-empiric
models for hourly solar radiation forecasting are developed and then compared with the Angstrom-
Prescott (A-P) type models [40]. Differently, a multi-level wavelet decomposition is applied by Zhen
et al. [41] to preprocess the solar irradiance data in order to further improve the day-ahead solar
irradiance forecasting accuracy. In Zhen’s another paper, a new day-ahead solar irradiance ensemble
forecasting model was developed based on time-section fusion pattern classification and mutual
iterative optimization [42]. With the emergence of deep learning (DL) models, Qing et al. [43] turned
to Long Short Term Memory (LSTM) to catch the dependence between consecutive hours of daily
solar irradiance data.
In general, the DL algorithm is more promising compared to the abovementioned traditional
machine learning. Recently, DL approaches have been not only successfully applied in image
processing [44], but also utilized to address the classification and regression issues of one-
dimensional data [45]. In the DL system, there are various branches, including LSTM, Convolutional
Neural Networks (CNN), and Recurrent Neural Network (RNN) and so on. In spite of the superior
performance of DL algorithms, few studies have applied the DL methods in the day-ahead solar
irradiance forecasting. Researchers need to validate whether the introduction of DL can improve the
solar irradiance forecasting accuracy. Moreover, there are various versions of DL models just like
those mentioned above. Different DL models have their own advantages and disadvantages.
Therefore, in the practice of solar irradiance forecasting, three important issues should be taken into
consideration, namely how to select the rational DL models, how to well combine them, and how to
further improve the performance of the hybrid DL model.
Appl. Sci. 2018, 8, 1286 4 of 30
1.3. The Content and Contribution of the Paper

According to the literature review work, we have found that the previous forecasting
approaches using manual feature extraction (MFE), traditional modeling and single DL models could
not satisfy the performance requirements in partial solar irradiance forecasting scenarios with
complex fluctuations. In this paper, we proposed an improved DL model to achieve the performance
improvement of day-ahead solar irradiance forecasting. This proposed model is named the DWT-
CNN-LSTM model. It should be noted that the historical daily solar irradiance curve always presents
high variability and fluctuation since the solar irradiance is influenced by the non-stationary weather
conditions. Therefore, the forecasting accuracy of day-ahead solar irradiance strongly depends on the
weather statuses no matter what kinds of forecasting models we choose. Given this fact, the DWT-
CNN-LSTM models are independently constructed for four general weather types (i.e., sunny,
cloudy, rainy, and heavy rainy days). This is because a single forecasting model cannot well reflect
the temporal relationships between historical and future solar irradiance under different weather
conditions. In other words, classification modeling could reduce the complexity and difficulty of
intro-class data fitting to improve the corresponding forecasting accuracy [1,28].
The basic pipeline framework behind data-driven DWT-CNN-LSTM models consists of three
major parts: (1) Discrete Wavelet Transformation (DWT) based solar irradiance sequence
decomposition, (2) a CNN-based local feature extractor, and (3) an LSTM based sequence forecasting
model. In solar irradiance forecasting under certain weather types, the raw solar irradiance sequence
is decomposed into several subsequences via discrete wavelet transformation. Then, each
subsequence is fed into the CNN-based local feature extractor, which leverages the advantage of
CNN to automatically learn the abstract feature representation from the raw subsequence data. Since
the extracted features are also time series data, they are individually transported to LSTM to construct
the subsequence forecasting model. In the end, the final solar irradiance forecasting results under
certain weather types are obtained via the wavelet reconstruction of these forecasted subsequences.
Compared to the existing studies for solar irradiance forecasting, the contributions of this paper can
be summarized as follows:
(1) Discrete wavelet transformation is applied in our proposed DWT-CNN-LSTM model to
decompose the raw solar irradiance sequence data of certain weather types into several stable
parts (i.e., low-frequency signals) and fluctuant parts (i.e., high-frequency signals). These
decomposed subsequences have better behaviors (e.g., more stable variances and fewer outliers)
in terms of regularity than the raw solar irradiance sequence data. Such wavelet decomposition
(WD) is helpful for precision improvement of the solar irradiance forecasting model.
(2) The CNN and LSTM are perfectly combined in our proposed DWT-CNN-LSTM model, in which
the abstract feature representation from the raw subsequence data is effectively extracted by
CNN and then these features are fed into LSTM. CNN is good at automatically extracting
abstract features from its input, and LSTM is able to find the long dependencies of the time series
input.
(3) The validity of the proposed DWT-CNN-LSTM model is verified based on the two measured
dataset, namely the dataset of Elizabeth City State University and Desert Rock Station.
The rest of paper is constructed as follows. Section 2 illustrates the three main parts of the
proposed DWT-CNN-LSTM model, including DWT based solar irradiance sequence decomposition,
the CNN-based local feature extractor, and the LSTM based sequence forecasting model. In Section 3,
the details of the experimental simulation are introduced and the relevant analysis results are
discussed. Finally, conclusions are drawn in Section 4.
2. Improved Deep Learning Model for Day-Ahead Solar Irradiance Forecasting

The historical daily solar irradiance curve always presents high variability and fluctuation since
solar irradiance is influenced by non-stationary weather conditions. This makes the forecasting
accuracy of day-ahead solar irradiance strongly depend on the weather statuses no matter what kinds
of forecasting models we choose.
Appl. Sci. 2018, 8, 1286 5 of 30
Therefore, as shown in Figure 1, the solar irradiance forecasting models are independently
constructed for four general weather types, because according to different weather types,
classification modeling could reduce the complexity and difficulty of intro-class data fitting so as to
improve the corresponding forecasting accuracy.
In terms of the proposed model (i.e., DWT-CNN-LSTM model) for day-ahead solar irradiance
forecasting, its integrated framework is illustrated in Figure 2. The basic pipeline framework behind
data-driven DWT-CNN-LSTM models consists of three major parts: (1) DWT based solar irradiance
sequence decomposition; (2) CNN based local feature extractor; and (3) LSTM based sequence
forecasting model. As for certain weather types, the raw historical solar irradiance sequence is
decomposed into approximate subsequence and several detailed subsequences. Then each
subsequence is fed to the CNN based local feature extractor, which leverages the advantage of CNN
to automatically learn the abstract feature representation from the raw subsequence data. Since the
features extracted by the CNN are also time series data that have rich temporal dynamics, then they
are input to LSTM to construct the subsequence forecasting model. In the end, the final solar
irradiance forecasting results under certain weather types are obtained through the wavelet
reconstruction of these forecasted subsequences. More details about three major parts above are
respectively illustrated in Sections 2.1–2.3.
Raw solar irradiance sequence
The classification of weather types

based on the closeness and correlation between surface and extraterrestrial solar irradiance
Four general weather types Extraterrestrial solar irradiance Surface solar irradiance
Solar irradiance
1500 1500 1500 1500

(W/m2)
1000 1000 1000 1000

500 500 500 500
0 0 0 0
0 30 60 90 0 30 60 90 0 30 60 90 0 30 60 90
Sunny day Cloudy day Rainy day Heavy rainy day
Day-ahead solar Day-ahead solar Day-ahead solar Day-ahead solar

irradiance forecasting irradiance forecasting irradiance forecasting irradiance forecasting
model model model model
(DWT-CNN-LSTM) (DWT-CNN-LSTM) (DWT-CNN-LSTM) (DWT-CNN-LSTM)
of sunny day of cloudy day of rainy da of heavy rainy day
Solar irradiance Solar irradiance Solar irradiance Solar irradiance

forecasting results of forecasting results of forecasting results of forecasting results of
sunny day cloudy day rainy day heavy rainy day
Figure 1. The flowchart of the day–ahead solar irradiance forecasting for four general weather types.
The DWT-CNN-LSTM forecasting model is based on discrete wavelet transformation (DWT),
convolutional neural network (CNN) and long short term memory (LSTM) network.
Appl. Sci. 2018, 8, 1286 6 of 30
Raw solar irradiance sequence under certain weather type
Normalization and k-level discrete wavelet transformation (DWT)
Approximate Detailed Detailed Detailed

subsequence Ak subsequence D1 subsequence D2 ……
subsequence Dk
CNN based local CNN based local CNN based local CNN based local
feature extractor feature extractor feature extractor ……
feature extractor
LSTM based LSTM based LSTM based LSTM based

sequence sequence sequence …… sequence
forecasting model forecasting model forecasting model forecasting model
Forecasting results Forecasting results Forecasting results Forecasting results

of sequence Ak of sequence D1 of sequence D2 ……
of sequence Dk
Wavelet reconstruction and anti-normalization
Final forecasting results of solar irradiance under certain weather type
Figure 2. The detailed framework of DWT-CNN-LSTM day-ahead forecasting model for solar
irradiance under certain weather type. The DWT-CNN-LSTM forecasting model is based on discrete
wavelet transformation (DWT), convolutional neural network (CNN) and long short term memory
(LSTM) network.
2.1. Discrete Wavelet Transformation Based Solar Irradiance Sequence Decomposition

In general, solar irradiance sequence data always presents high volatility, variability and
randomness due to its correlation to non-stationary weather conditions. Therefore, the raw solar
irradiance sequence probably includes nonlinear and dynamic components in the form of spikes and
fluctuations. The existence of these components will undoubtedly deteriorate the precision of the
solar irradiance forecasting models. In practice, high-frequency signals and low-frequency signals
are contained in solar irradiance sequence data. The former primarily results from the chaotic nature
of the weather system. The latter is caused by the daily rotation of the earth. As for each signal with
certain frequency, it is easier for a specific sequence forecasting model to predict the corresponding
outliners and behaviors of that signal. Given the above considerations, DWT is employed here to
decompose the raw solar irradiance sequence data into several stable parts (i.e., low-frequency
signals) and fluctuant parts (i.e., high-frequency signals). These decomposed subsequences have
better behaviors (e.g., more stable variances and fewer outliers) in terms of regularity than the raw
solar irradiance sequence data, which is helpful for the precision improvement of the solar irradiance
forecasting model [46].
In numerical analysis, DWT is a kind of wavelet transform for which the wavelets are discretely
sampled. The key advantage of DWT over Fourier transforms is that DWT is able to capture both
frequency and location information (location in time). In addition, DWT is good at the processing of
multi-scale information processing [47]. These superiorities make DWT an efficient tool for complex
data sequence analysis. In wavelet theory, the original sequence data are generally decomposed into
two parts called approximate subsequence and detailed subsequence via DWT. The approximate
subsequence captures the low-frequency features of the original sequence, while the detailed
subsequence contains the high-frequency features. This process is regarded as wavelet
decomposition (WD), and the approximate subsequences obtained from the original sequence can
Appl. Sci. 2018, 8, 1286 7 of 30
also be further decomposed by WD process. Then the high-frequency noise in the forms of the
fluctuation and randomness in original sequence can be extracted and filtered through WD process.
Given a certain mother wavelet function  (t ) and its corresponding scaling function  (t ) , a
sequence of wavelet  j,k ( t ) and binary scale-functions  j,k ( t ) can be calculated as follows:
j
 j,k ( t ) =2 2 ( 2 j t − k ) (1)
j
 j,k ( t ) =2 2  ( 2 j t − k ) (2)
in which t , j and k respectively denote the time index, scaling variable and translation variable.
Then the original sequence os ( t ) can be expressed as follows:
n J n
os ( t ) =  c j ,k j,k ( t ) +  d j ,k j,k (t ) (3)
k =1 j =1 k =1
in which c j ,k is the approximation coefficient at scale j and location k , d j , k denotes the detailed
coefficient at scale j and location k , n is the size of the original sequence, and J is the
decomposition level. Based on the fast DWT proposed by Mallat [48], the approximate sequence and
detailed sequence under a certain WD level can be obtained via multiple low-pass filters (LPF) and
high-pass filters (HPF).
Figure 3 exhibits the specific WD process in our practical work. During a certain k-level WD
process, the raw solar irradiance sequence of certain weather types is first decomposed into two parts:
approximate subsequence A1 and detailed subsequence D1. Next, the approximate subsequence A1
is further decomposed into another two parts namely A2 and D2 at WD level 2, and continues to A3
and Ds at WD level 3, etc. Therefore, as shown in Figure 2, the approximate subsequence Ak and
detailed subsequences D1 to Dk can be individually forecasted by various time sequence forecasting
models (i.e., our proposed CNN-LSTM model, autoregressive integrated moving average model,
support vector regression, etc). Then the final forecasting results of solar irradiance sequence can be
obtained through the wavelet reconstruction on the forecasting results of Ak and D1 to Dk.
Level 1 Level 2 Level 3 Level k

Raw solar irradiance sequence S
A1 A2 A3 Ak
LPF LPF LPF LPF
D1 D2 D3 Dk
HPF HPF HPF HPF
Figure 3. The detailed process of k-level wavelet decomposition. A1 to Ak are the approximate
subsequences, and D1 to Dk are the detailed subsequences. All of these subsequences can be
forecasted individually using some kind of time sequence forecasting models.
2.2. Convolutional Neural Networks Based Local Feature Extractor

Generally speaking, the historical solar irradiance sequence data is the most important input
that contains abundant information for forecasting the day-ahead solar irradiance. In our proposed
DWT-CNN-LSTM model, the original solar irradiance sequence under certain weather type is
decomposed through DWT into several subsequences. These subsequences also include relevant and
significant information that is useful for the later forecasting of subsequences. Therefore, the effective
Appl. Sci. 2018, 8, 1286 8 of 30
extraction of local features that are robust and informative from the sequential input is very important
for enhancing the forecasting precision. Traditionally, many previous works primarily focused on
multi-domain feature extractions [49], including statistical (variance, skewness, and kurtosis)
features, frequency (spectral skewness) features, time frequency (wavelet coefficients) features, etc.
However, these hand-engineered features require intensive expert knowledge of the sequence
characteristics and cannot necessarily capture the intrinsic sequential characteristic behind the input
data. Moreover, knowing how to select these manually extracted features is another big challenge.
Unlike manual feature extraction, CNN is an emerging branch of DL that is used for automatically
generating useful and discriminative features from raw data, which has already been broadly applied
in image recognition, speech recognition, and natural language processing [50].
As for application, the subsequences decomposed from solar irradiance sequence can be
regarded as 1-dimensional sequences. Thus 1-dimensional CNN is adopted here to work as a local
feature extractor. The key idea of CNN lies in the fact that abstract features can be extracted by
convolutional kernels and the pooling operation. In practice, to address the sequences, the
convolutional layers (convolutional kernels) firstly convolve multiple local filters with the sequential
input. Each feature map corresponding to each local filter can be generated by sliding the filter over
the whole sequential input. Subsequently, the pooling layer is utilized to extract the most significant
and fixed-length features from each feature map. In addition, the convolution and pooling layers can
be combined in a stacked way.
First of all, the most simply constructed CNN with only one convolutional layer and one pooling
layer is introduced to briefly show how the CNN directly process the raw sequential input. It is
assumed that K filters with a window size of m are used in the convolutional layer. The details
of the relevant mathematical operation in these two layers are presented in the following two
subsections.
(1) Convolutional Layer
Convolution operation is regarded as a specific linear process that aims to extract local patterns
in the time dimension and to find local dependencies in the raw sequences. The raw sequential input
S and filter sequence FS is defined as follows. Here vectors are expressed in bold according to the
convention.
S = [s1 ,s2 ,s3 ,  ,sL ] (4)
FS = [ w1 , w2 , w3 , , wK ] (5)
m 1
in which si  R is the single sequential data point that is arrayed according to time, and w j  R
is one of the filter vectors. L is the length of the raw sequential input S , and K is the number of
total filters in the convolutional layer. Then the convolution operation is defined as a multiplication
operation between a filter vector wj and a concatenation vector representation si:i+m-1 .
si:i+m-1 =si  si+1  si+ 2      si+m-1 (6)
in which  is the concatenation operator, and si:i+m-1 denotes a window of m continuous time
steps starting from the i-th time step. Moreover, the bias term b  R should also be considered
into the convolution operation. Thus, the final calculation equation is written as follows.
(
ci = f wjT si:i+m-1 +b ) (7)
T
in which w j represents the transpose of a filter matrix w j , and f is a nonlinear activation function.
In addition, index i denotes the i-th time step, and index j is the j-th filter.
The application of activation function aims to enhance the ability of models to learn more
complex functions, which can further improve forecasting performance. Applying suitable activation
function can not only accelerate the convergence rate but also improve the expression ability of
Appl. Sci. 2018, 8, 1286 9 of 30
model. Here, Rectified Linear Units (ReLu) are adopted in our model due to their superiority over
other kinds of activation functions [51].
(2) Pooling layer
In the above subsection, the given example only introduces the detailed convolution operation
process between one filter and the input sequence. In actual application, one filter can only generate
one feature map. Generally, multiple filters are set in the convolution layer in order to better excavate
the key features of input data. Just as assumed above, there are K filters with a window size of m
in the convolutional layer. In Equations (5) and (7), each vector w j represents a filter, and the sing
value c i denotes the activation of the window.
The convolution operation over the whole sequential input is implemented via sliding a filtering
window from the beginning time step to the ending time step. So the feature map corresponding to
that filter can be denoted in the form of a vector as follows.
Fj = [c1 ,c2 ,c3 , ,cL−m+1 ] (8)
in which index j is the j-th filter, and the elements in Fj corresponds to the multi-windows as
s1:m ,s2:m ,  ,sl-m+1:L  .
The function of pooling is equal to subsampling as it subsamples the output of convolutional
layer based on the definite pooling size p . That means the pooling layer can effectively compress the
length of feature map so as to further reduce the number of model parameters. Based on the max-
pooling applied in our model, the compressed feature vector Fj-compress can be obtained as follows.
In addition, the max operation takes a max function over the p consecutive values in feature map
Fj .
Fj-compress = [h1 ,h2 ,h3 , ,h L-m ]

+1 (9)
p
( )
in which h j = max c( j -1) p ,c( j -1) p +1 ,  ,c jp -1 .
In the application in our solar irradiance forecasting, the solar irradiance sequence input is a
vector with only one dimension. The subsequences that are decomposed from the solar irradiance
sequence are also a vector with only one dimension. Therefore, the size of the input subsequences in
the convolution layer is n L  1 . n L is the length of the
is the number of data samples and
subsequences. The size of the corresponding outputs after the pooling layer is n  ( ( L -m ) / p+1)  K
. It can be obviously noted that the length of the input sequence is compressed from L to
(( L-m) / p+1) .
In sum, the CNN based feature extractor can provide more representative and relevant
information than the raw sequential input. Moreover, the compression of the input sequence’s length
also increases the capability of the subsequent LSTM models to capture temporal information.
To give a brief illustration, the framework for the CNN-based local feature extractor is shown in
Figure 4. Additionally, in the actual application, some important parameters need to be set according
to the specific circumstances. These parameters include the number of the convolutional and pooling
layers, the number of filters in each convolution layer, the sliding steps, the size of sliding window,
the pooling size, etc.
Appl. Sci. 2018, 8, 1286 10 of 30
Sliding Feature Feature map after

window map compressed
Time steps LSTM based
...
...
...
...
...
...
sequence forecasting
...
...
...
...
model
Convolution Non- linear Pooling

Raw sequence input layer
layer activation
Figure 4. The picture shows the framework of the CNN based local feature extractor. The convolution
layer consists of different filters marked by yellow, green and grey colors. Each filter can generate a
specific feature map to extract the key information of the raw sequence input through sliding the
corresponding windows. The activation function is used to enhance the ability of models to learn
more complex functions. The function of pooling is equal to subsampling as it subsamples the output
of convolutional layer based on the definite pooling size.
2.3. Long Short Term Memory Based Sequence Forecasting Model (from RNN to LSTM)
In the previous works, some sequence models (e.g., Markov models, Kalman filters and
conditional random fields) are commonly used tools to address the raw sequential input data.
However, the biggest drawback of these traditional sequential models is that they are unable to
adequately capture long-range dependencies. In the application of day-ahead solar irradiance, many
indiscriminative or even noisy signals that exist in the sequential input during a long time period
may bury informative and discriminative signals. This can lead to the failure of these above
sequences models. Recently, RNN has emerged as one effective model for sequence learning, which
has already been successfully applied in the various fields, including image captioning, speech
recognition, genomic analysis and natural language processing [52].
In our proposed DWT-CNN-LSTM model, LSTM that overcomes the problems of gradient
exploding or vanishing in RNN, is adopted to take the output of CNN based local feature extractor
to further predict the targeted subsequences. As mentioned in Section 2.1, these subsequences are
decomposed from solar irradiance data. In the following two subsections, the principle of RNN is
simply introduced and the construction of its improved variant (i.e., LSTM) is then illustrated in
detail.
2.3.1. Recurrent Neural Network

The traditional neural network structure is characterized by the full connections between
neighboring layers, which can only map from current input to target vectors. However, RNN has the
ability to map target vectors from the whole history of the previous inputs. Thus RNN is more
effective at modeling dynamics in sequential data when compared to traditional neural networks. In
general, RNN builds connections between units from a directed cycle and memorizes the previous
inputs via its internal state. Specifically speaking, the output of RNN at time step t−1 could influence
the output of RNN at time step t. This makes RNN able to establish the temporal correlations between
present sequence and previous sequences. The structure of RNN is shown in Figure 5.
Appl. Sci. 2018, 8, 1286 11 of 30
y = [y(0),y(1),y(2)]
y(0) y(1) y(2)
W
V
V S(0) V V
S(1) S(2)
S
RNN
= RNN
W
RNN
W
RNN
U U U U
X= [x(0),x(1),x(2)] x(0) x(1) x(2)
Figure 5. The structure of Recurrent Neural Network.
In Figure 5, the sequential vectors X =  x ( 0 ) ,x (1) ,x ( 2 )  are passed into RNN one by one
according to the set time step. This is obviously different from the traditional feed-forward network
in which all the sequential vectors are fed into the model at one time. The relevant mathematical
equation can be described as follows.
S(t )=σ(U  x(t)+W  S(t-1)+b) (10)
y(t)=σ(V  s(t)+c) (11)
in which x (t ) is the input variable at t time step, W , U and V are weight matrixes, b and
c are the biases vectors, σ is activation functions, and y (t ) is the expected output at t time step.
Although RNN is very effective at modeling dynamics in sequential data, it can suffer from the
gradient vanishing and explosion problem in its backpropagation based model training when
modeling long sequences [53]. Considering the inherent disadvantages of typical RNN, its improved
variant named LSTM is adopted in our work, which is illustrated in the following subsection.
2.3.2. Long-Short-Term Memory

LSTM network proposed by Hochreiter et al. [53] in 1997 is a variant type of RNN, which
combines representation learning with model training without requiring additional domain
knowledge. The improved construction of LSTM is helpful for the achievement of avoiding gradient
vanishing and explosion problems in typical RNN. This means that LSTM is superior at capturing
long-term dependencies and modeling nonlinear dynamics when addressing the sequential data
with a longer length. The structure of LSTM cell is shown in Figure 6.
y(t)
Forget gate
c(t-1) c(t)
Input gate
tanh h(t)
f(t) g(t) i(t)

o(t)
sigmoid tanh sigmoid sigmoid
Output gate
h(t-1)
Point-wise
LSTM Cell multiplication
Addition
x(t)
Figure 6. The structure of Long Short-Term Memory Cell.

Appl. Sci. 2018, 8, 1286 12 of 30
LSTM is explicitly designed to overcome the problem of gradient vanishing, by which the
correlation between vectors in both short and long-term can be easily remembered. In LSTM cell,
h ( t ) can be considered as a short-term state, and c ( t ) can be considered as a long-term state. The
significant characteristic of LSTM is that it can learn what needs to be stored in the long-term, what
( )
needs to be thrown away and what needs to be read. When c t-1 point enters into cell, it first goes
through a forget gate to drop some memory; then, some new memories are added to it via an input
gate; finally, a new output y t () that is filtered by the output gate is obtained. The process of where
the new memories come from and how these gates work is shown below.
(1) Forget
This part reveals how LSTM controls what kinds of information can enter into the memory cell.
After h t-1( ) and x t() has passed through sigmoid function, a value f t () between 0 and 1 is
( )
generated. The value of 1 means that h t-1 will be completely absorbed in the cell state c t-1 . On ( )
( )
the contrary, if the value is 0, h t-1 will be abandoned by cell state c t-1 . The formula of this( )
process is shown below.
f ( t ) =σ(W f  [h ( t-1) ,x ( t )]+b f ) (12)
in which W f weight matrix, b f is biases vectors, and σ is activation function.
(2) Store
This part shows how LSTM decides what kinds of information can be stored in the cell state.
( )
First, h t-1 passes through sigmoid function, and a value i t () between 0 and 1 is then obtained.
Next, h ( t-1) passes through tanh function and then a new candidate value g t () is obtained. In the
end, the above two steps can be integrated to update the previous state.
i ( t ) =σ(Wi  [h ( t-1) ,x ( t )] + bi ) (13)
g ( t ) =tanh(Wg  [h ( t-1) ,x ( t )]+bg ) (14)
Then the previous cell state c t-1 ( ) considers what information should be abandoned and
()
stored and then creates a new cell state c t . This process can be formulated as follows.
c ( t ) =f ( t )  c ( t-1) +it  g t (15)
(3) Output
()
The output of LSTM is based on the updated cell state c t . First of all, we employ the sigmoid
function to generate a value o t () to control the output. Then tanh and the output of sigmoid
function o t () ()
are further utilized to generate the cell state h t . Thus we can output y t () after the
above process as shown in the following two steps.
o ( t ) = σ(Wo  [h ( t-1) ,x ( t )]+bo ) (16)
y ( t ) =h ( t ) =o ( t ) *tanh(C ( t )) (17)
Appl. Sci. 2018, 8, 1286 13 of 30
The training process of LSTM is called BPTT (backpropagation through time) [54].
3. Case Study
3.1. Data Source and Experimental Setup

The historical irradiance data applied in the above proposed solar irradiance forecasting models
is based on the dataset of Elizabeth City State University and Desert Rock Station. The first irradiance
dataset in our simulation is downloaded from the National Renewable Energy Laboratory (NREL),
which is measured by the Elizabeth City State University at Elizabeth City from 2008 to 2012 [55].
There are 1817 days of solar irradiance data available with 5 min time resolution. The second
irradiance dataset in our simulation is downloaded from the National Oceanic & Atmospheric
Administration (NOAA) Earth System Research Laboratory website, which is measured by the
Surface Radiation station at Desert Rock from 2014 to 2017 [56]. There are 1196 days of solar irradiance
data available with 1min time resolution.
To meet the international standard of short-period solar irradiance forecasting, the irradiance
data should be further transformed to be the data with 15 min time resolution by taking the average
of irradiance points data in the span of every 15 min. Therefore, there are total 96 irradiance data
points in one day. Considering the earliest sunrise time and the latest sunset time in three years, we
only use daily data points that range from 18th to 78th. As for the forecast periodicity, we use the
historical irradiance data from the previous three days to predict the irradiance value for the next
day. Therefore, in the solar irradiance forecasting model, the input variable is the historical irradiance
data from the previous three days and the output variable is the predicted irradiance value for the
next day.
All experimental platforms are built on high-performance Lenovo desktop computer equipped
with the Win10 operating system, Intel(R) Core(TM) i5-6300HQ [email protected], 8.00 GB RAM, and
NVIDIA GeForce GTX 960M GPU. We use Python 3.6.1 with Keras [57] and Scikit-learn [58] to
establish the DWT-CNN-LSTM forecasting models for day-ahead solar irradiance.
3.2. Model Training and Hyperparameters Selection

In the DL based forecasting models, the mean square error (MSE) is chosen as loss function, and
Adam Optimization is selected as an optimizer. During the deep learning training process, weight
initialization and bias initialization play a vital role. Therefore, we choose the data from truncated
normal distribution with 0 mean and 0.05 standard deviation as weight initialization method of CNN
and fully connected layer. This method is the recommended initializer for neural network weights
and filters. Orthogonal method, a popular initialization way, is selected as weight initializer for LSTM
block. The bias for all hidden layers is set as 0.1. The learning rate is 0.001, the batch size is 24 and the
epoch is 200.
In addition, for two dataset, the numbers of training set and the testing set are different under
four general weather types. The training set is used for training forecasting model, the testing set for
evaluating forecasting result. All the above mentioned details of the division of training and testing
sets, as well as parameter setting of DWT-CNN-LSTM model, are listed in Tables 1 and 2.
Table 1. The division detail of samples sets under four general weather types.
Weather Types Elizabeth City Desert Rock Station

Sunny type The number of training set: {288} The number of training set: {412}
The number of validation set: {32} The number of validation set {46}
The number of testing set: {80} The number of testing set: {115}
Cloudy type The number of training set: {504} The number of training set: {230}
The number of validation set {56} The number of validation set{25}
The number of testing set: {140} The number of testing set:{65}
Rainy type The number of training set: {366} The number of training set: {147}
Appl. Sci. 2018, 8, 1286 14 of 30
The number of validation set {40} The number of validation set {14}
Heavy Rainy type The number of training set: {153} The number of training set: {72}
The number of validation set {16} The number of validation set {10}
Table 2. The parameter setting detail of DWT-CNN-LSTM model.
Option Parameter Setting

Training Method Adam Optimizer
Learning rate {0.001}
Batch size {24}
Epoch {200}
Training stop strategy {early stopping}
Loss Function MSE
We set the split proportion of training set, validation set and testing set as 0.7:0.1:0.2. The training
set is used to train the solar irradiance forecasting models. The validation set is used to adjust the
hyper-parameters of these DL forecasting models. The testing set is used to verify the model
performance.
For the proposed model, we first design two CNN layers with 64 filters, and the filter size and
pooling size are both set to 3. Then, two LSTM layers are connected to CNN output with 100 neurons.
The outputs of LSTM are fed into two fully connected layers with linear activation function. The Relu
activation function is applied to CNN and LSTM layers. To overcome the overfitting problems in
models, dropout method with 0.2 parameter is applied after CNN and LSTM layers. In addition, early
stopping method is also applied. In addition, the output data format of the input layer, each
intermediate layer, and the output layer are accordingly shown in Table 3. Additionally, Table 4
illustrates the structure of the other forecasting models used as benchmarks.
Table 3. The output data format of the input layer, each intermediate layer, and the output layer in
DWT-CNN-LSTM model.
Layer Output Shape

Input layer (180,1)
Conv-1D layer (180,64)
Max-Pooling layer (60,64)
Conv-1D layer (60,64)
Max-Pooling layer (20,64)
LSTM layer (20,100)
LSTM layer (100)
Fully connected Layer (100)
Output layer (60,1)
Table 4. The structure of the other forecasting models used as benchmarks.
Forecasting Models Structure

Convolutional layer (64 filters + 3 filter size) + maxpooling (3
CNN pooling size) + convolutional layer (64 filters + 3 filter size)
MaxPooling (3 pooling size) + Fully connected layer (100 neurons)
LSTM 2 LSTM layers (100 neurons)
ANN 2 fully connected layers (100 neurons)
ARIMA Determined by the minimum AIC of each input sample
Appl. Sci. 2018, 8, 1286 15 of 30
3.3. Performance Criterion

To evaluate the performance of solar irradiance forecasting models, we employ three effective
error indexes that are Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and
Correlation Coefficient (R). The smaller RMSE and MAE, together with the higher R denote the good
performance of a forecasting model. The mathematical calculation methods of these three error
indexes are shown in the following equations in turn.
N
 (yt -yˆ t )2 (18)
t= 1
RMSE=
N
N
|yt -yˆ t| (19)
MAE= t=1
N
Cov ( y,yˆ )
R= (20)
V ( y ) V ( yˆ )
in which yˆ t , yt are, respectively, the forecasting value and actual value at time t. y refers to the
mean value of the whole yt , and N is the sample size of the test set.
3.4. Model Performance Analysis for DWT-CNN-LSTM Model with Different WD Level
In the proposed DWT-CNN-LSTM model, the first step is to decompose the raw solar irradiance
sequence of certain weather type into several approximate subsequences and detailed subsequences.
The key of this step is the determination of decomposition level. As for the solar irradiance forecasting
based on certain dataset, both the higher and lower WD level are not conducive to the performance
improvement of subsequent forecasting models. Therefore, in this part, the performance comparison
of DWT-CNN-LSTM model with different WD level is conducted using two different datasets,
namely the dataset of Elizabeth City State University and Desert Rock Station. The detailed results
are respectively shown in Tables 5 and 6. As shown in Table 5, under the sunny weather type, the
DWT-CNN-LSTM model without WD performs better than that with WD level 1 to 4. This is mainly
because the solar irradiance curve of sunny days is smooth and less fluctuating. Therefore, the
application of WD will not bring very obvious improvement of the forecasting performance.
Table 5. The performance comparison of DWT-CNN-LSTM model at different WD levels using the
dataset of Elizabeth City State University.
Weather Error Wavelet Decomposition (WD) Level

Types Index Level 1 Level 2 Level 3 Level 4 without WD
MAE 23.174 23.474 24.213 24.848 22.560
Sunny RMSE 36.548 36.363 40.323 41.244 36.226
R 0.991 0.991 0.989 0.989 0.992
MAE 86.313 81.466 83.547 88.731 86.754
Cloudy RMSE 121.506 118.645 124.364 126.149 121.922
R 0.926 0.928 0.925 0.919 0.925
MAE 95.1758 89.503 93.126 93.695 93.694
Rainy RMSE 145.219 139.133 143.919 142.998 142.194
R 0.748 0.757 0.741 0.741 0.743
MAE 41.234 38.642 39.981 42.774 43.435
Heavy
RMSE 68.742 67.574 68.981 70.885 70.410
rainy
R 0.628 0.641 0.634 0.611 0.615
Appl. Sci. 2018, 8, 1286 16 of 30
Table 6. The performance comparison of DWT-CNN-LSTM model at different WD levels using the
dataset of Desert Rock Station.
Weather Error Wavelet Decomposition (WD) Level

Types Index Level 1 Level 2 Level 3 Level 4 without WD
MAE 17.131 17.379 18.249 18.498 16.573
Sunny RMSE 34.299 34.429 35.844 36.477 33.101
R 0.992 0.991 0.989 0.987 0.993
MAE 62.144 66.499 67.425 68.552 66.661
Cloudy RMSE 91.099 95.377 96.374 98.551 96.641
R 0.965 0.963 0.958 0.957 0.959
MAE 131.384 130.194 136.847 138.257 132.83
Rainy RMSE 181.392 180.079 184.963 187.241 182.97
R 0.865 0.866 0.847 0.832 0.857
MAE 68.212 62.160 64.161 65.840 63.448
Heavy
RMSE 96.490 94.977 97.203 103.880 96.373
rainy
R 0.657 0.663 0.651 0.619 0.647
Nevertheless, for other three weather types (i.e. cloudy, rainy and heavy rainy) shown in Table 5,
DWT based solar irradiance sequence decomposition does enhance the corresponding forecasting
performance to a different extent. This can be explained by the fact that the solar irradiance curve of
cloudy, rainy and heavy rainy days presents higher volatility, variability and randomness than that
of sunny days. Therefore, the raw solar irradiance sequence of cloudy, rainy and heavy rainy days
probably includes nonlinear and dynamic components in the form of spikes and fluctuations. The
existence of these components will undoubtedly deteriorate the precision of the solar irradiance
forecasting models. Additionally, the application of WD can mitigate the above problems.
To summarize the information provided in Table 5, WD cannot effectively improve the
forecasting performance of sunny days. Under the other three weather types, DWT-CNN-LSTM
model performs best at WD level 2 when using the dataset of Elizabeth City State University. The
results of performance comparison shown in Table 6 are different. Specifically speaking, DWT-CNN-
LSTM model of cloudy days performs best at WD level 1 rather than WD level 2 when using the
dataset of Desert Rock Station. Therefore, we can draw the conclusion that the influence of WD on
forecasting performance, as well as the best WD level, generally varies under different weather types
and validation datasets.
3.5. Performance Comparison Analysis of Different Solar Irradiance Forecasting Models

The proposed DWT-CNN-LSTM forecasting model is different from the previous traditional
solar irradiance forecasting models. The key characteristics of the DWT-CNN-LSTM forecasting
model are the perfect combination of the following parts: (1) DWT based solar irradiance sequence
decomposition; (2) CNN based local feature extractor; and (3) LSTM based sequence forecasting
model. In addition, the solar irradiance forecasting models are individually established under sunny,
cloudy, rainy and heavy rainy days. Given this fact, the relevant performance comparison analysis is
also shown and discussed under the above four weather types. The involved three error indexes (i.e.,
RMSE, MAE, and R) are considered as the basis of the following performance comparison analysis of
different forecasting models.
3.5.1. Comparison Analysis of Sunny Days

As previously shown in Table 5, the DWT-CNN-LSTM forecasting model of sunny days
performs best at WD level 1 among different WD levels. So in this part, the DWT-CNN-LSTM model
at WD level 1 is compared with six solar irradiance forecasting models, namely CNN-LSTM (i.e., our
proposed model without WD), artificial neural network (ANN), and manually extracted features
Appl. Sci. 2018, 8, 1286 17 of 30
(ANN, persistence forecasting, CNN and LSTM). As for the manually extracted features-ANN model,
the relevant statistical features and their corresponding expressions are shown in Table 7.
Table 7. The list of manually extracted features.
Statistical Features Expression

2
Z var = (1 / n ) i =1 ( zi −  )
n
Variance 1
Maximum Zmax = max ( z)

Zskew = E ( ( z- ) /  ) 
3
Skewness
 
Zskew = E ( ( z- ) /  ) 
4
Kurtosis
 
Zaver = (1 / n ) i =1 zi
n
Average
1 zi is the solar irradiance data point at time i during the whole day. z is the data point set of {z 1, z2,…, zn}
The performance comparisons of different sunny days’ forecasting models using the dataset of
Elizabeth City State University and Desert Rock Station are respectively shown in Tables 8 and 9. In
Table 8, the prediction accuracy of DWT-CNN-LSTM (WD level 1) is worse than the single CNN-
LSTM without WD. The corresponding conclusion can be drawn that the application of DWT based
solar irradiance sequence decomposition does not improve the forecasting performance. The reason
behind this phenomenon has already been explained in Section 3.5.
Table 8. The performance comparison of different sunny day’s forecasting models using the dataset
of Elizabeth City State University.
Forecasting Models MAE RMSE R

DWT-CNN-LSTM (WD level 1) 23.174 36.548 0.991
CNN-LSTM 22.560 36.226 0.992
CNN 22.773 36.763 0.992
LSTM 24.497 37.049 0.990
Manually extracted features-ANN 43.045 54.796 0.985
ANN 23.533 36.888 0.989
Persistence forecasting 30.271 41.742 0.987
ARIMA 32.148 40.174 0.988
As for our proposed model without WD (i.e., CNN-LSTM), it is superior to manually extracted
features-ANN. This further verifies the ability of CNN to automatically and effectively extract
representative and significant information from the raw input data. Additionally, ANN, persistence
forecasting, and ARIMA models perform worse than CNN-LSTM, which also validates the
advisability of applying the combined DL models in solar irradiance forecasting. By comparing
among CNN-LSTM, CNN and LSTM, the comparing results also verify the reasonableness of the
tandem connection of CNN and LSTM, because the performance evaluation (based on MAE, RMSE
and R) results of CNN-LSTM are all better than those of CNN and LSTM. The above similar results
can also be found in Table 9. Figure 7 shows the actual and forecasted solar irradiance curve on sunny
day pattern using dataset of Elizabeth City State University.
Table 9. The performance comparison of different sunny day’s forecasting models using the dataset
of Desert Rock Station.

CNN-LSTM 16.573 32.411 0.993
Appl. Sci. 2018, 8, 1286 18 of 30
CNN 16.222 33.178 0.993

LSTM 17.032 33.294 0.992
ANN 17.869 34.783 0.990
ARIMA 20.433 37.781 0.987
Figure 7. Actual and forecasted solar irradiance on sunny day pattern using dataset of Elizabeth City
State University.
3.5.2. Comparison Analysis under Cloudy Day

Based on the dataset of Elizabeth City State University and Desert Rock Station, the performance
comparisons among different cloudy day’s forecasting models are presented in Tables 10 and 11,
respectively. As previously discussed in Table 5, the DWT-CNN-LSTM model of cloudy days has the
highest forecasting precision at WD level 2 when using the dataset of Elizabeth City State University.
Therefore, as shown in Table 10, the proposed DWT-CNN-LSTM model with WD level 2 is selected
to make comparisons with the other kinds of forecasting models.
Table 10. The performance comparison of different cloudy days’ forecasting models using the dataset

CNN-LSTM 86.754 121.922 0.925
CNN 87.043 122.042 0.923
LSTM 87.997 122.479 0.921
ANN 89.743 123.532 0.917
ARIMA 110.334 207.694 0.772
Table 11. The performance comparison of different cloudy days’ forecasting models using the dataset

CNN-LSTM 63.661 96.641 0.959
Appl. Sci. 2018, 8, 1286 19 of 30
CNN 64.339 95.373 0.961

LSTM 66.752 97.523 0.954
ANN 69.522 100.811 0.950
ARIMA 89.543 150.192 0.865
First of all, it should be noted that all the error index values of DWT-CNN-LSTM (WD level 2)
model is better than that of single CNN-LSTM. This result indicates that the DWT based solar
irradiance sequence decomposition has the capability to further improve the forecasting performance
of combined CNN-LSTM models. As discussed in Section 3.5, the obvious performance improvement
can be attributed to the fact that the solar irradiance curve of cloudy days presents high volatility,
variability and randomness. Therefore, the cloudy day’s solar irradiance sequence includes nonlinear
and dynamic components in the form of spikes and fluctuations. The existence of these components
will undoubtedly deteriorate the precision of the solar irradiance forecasting models. Additionally,
the application of WD could well mitigate the above problems.
When compared to the manually extracted features-ANN, as well as the traditional forecasting
models (i.e., ANN, persistence forecasting and ARIMA), the comparison results verify our proposed
model’s advantages in the following two respects. One is the ability to automatically extract
representative and significant information from the raw input data, and the other is the ability to
capture the long dependencies among the time series input data. In addition, the performance
improvement of CNN-LSTM over CNN and LSTM also reveals the benefits of the combination of
them. A similar discussion can also be made according to Table 11. Figure 8 shows the actual and
forecasted solar irradiance curve on cloudy day pattern using dataset of Elizabeth City State
University.
Figure 8. Actual and forecasted solar irradiance on cloudy day pattern using dataset of Elizabeth City
State University.
3.5.3. Comparison Analysis under Rainy Days

In terms of the rain day, it is discussed in Section 3.5 that the corresponding DWT-CNN-LSTM
model performs best at level 2 whether using the dataset of Elizabeth City State University or Desert
Rock Station. Therefore, as shown in Tables 12 and 13, the DWT-CNN-LSTM (WD level 2) is
compared with other forecasting models.
Appl. Sci. 2018, 8, 1286 20 of 30
Table 12. The performance comparison of different rainy days’ forecasting models using the dataset

CNN-LSTM 93.694 142.194 0.743
CNN 94.773 143.072 0.737
LSTM 95.089 142.877 0.741
ANN 97.894 147.818 0.736
ARIMA 132.066 181.681 0.656
Table 13. The performance comparison of different rainy days’ forecasting models using the dataset

CNN-LSTM 132.831 181.973 0.857
CNN 132.755 183.076 0.857
LSTM 133.007 184.332 0.855
ANN 138.045 186.553 0.829
ARIMA 177.053 210.119 0.772
When CNN-LSTM and DWT-CNN-LSTM (WD level 2) are compared, the results and the
reasons for them are similar to those discussed in Section 3.5.3. Specifically, the MAE is lowered from
93.694 in CNN-LSTM to 89.503 in DWT-CNN-LSTM. The RMSE is lowered from 142.194 in CNN-
LSTM to 139.133 in DWT-CNN-LSTM. At the same time, the R has also been improved from 0.743 in
CNN-LSTM to 0.757 in DWT-CNN-LSTM. The lower MAE and RMAE denote smaller differences
between forecasted and true solar irradiance data, and the higher R also represents that the forecasted
solar irradiance curve is closer to the true one. Therefore, the application of the DWT based sequence
decomposition also helps the improvement of forecasting performance. Additionally, the combined
CNN-LSTM shows better forecasting performance than the rest models (i.e., single DL models and
traditional forecasting models). This indicates that the reasonable combination of DL models can
better take advantage of the CNN and LSTM.
In sum, the improved DL models (i.e., DWT-CNN-LSTM) not only leverages the advantages of
DWT to obtain subsequences with good behavior (e.g., more stable variances and fewer outliers) in
terms of regularity, but also absorbs the superiority of CNN-LSTM to automatically extract abstract
features and find long dependencies. Similar results can also be found in Table 13. Figure 9 shows
the actual and forecasted solar irradiance curve on rainy day pattern using dataset of Elizabeth City
State University.
Appl. Sci. 2018, 8, 1286 21 of 30
Figure 9. Actual and forecasted solar irradiance on rainy day pattern using dataset of Elizabeth City
State University.
3.5.4. Comparison Analysis under Heavy rainy Days

Regarding the weather type of rainy days, the corresponding simulation result in Section 3.5
reveals that the DWT-CNN-LSTM model can reach the best precision at WD level 2. Therefore, the
DWT-CNN-LSTM (WD level 2) is adopted once again to be compared with other forecasting models.
Similar to the cloudy and rainy days, the solar irradiance data under heavy rainy days is also volatile
and fluctuates. The introduction of DWT based sequence decomposition is able to mitigate the
adverse influence of fluctuation on forecasting models. This idea is in accordance with comparison
results shown in Tables 14 and 15.
Table 14. The performance comparison of different heavy rainy days’ forecasting models using the
dataset of Elizabeth City State University.

CNN-LSTM 43.435 70.410 0.616
CNN 45.775 73.377 0.611
LSTM 44.373 74.086 0.611
ANN 48.956 77.034 0.589
ARIMA 63.848 110.735 0.388
Additionally, the great performance improvement is also achieved via automatic feature
extraction and long dependency identification, especially under unstable weather conditions. This
can also be verified by the following results shown in Table 14. For example, the MAE is reduced a
lot from 64.416 in persistence forecasting to 38.642 in DWT-CNN-LSTM (WD level 2). The RMSE is
reduced a lot from 107.290 in persistence forecasting to 67.574 in DWT-CNN-LSTM (WD level 2).
Additionally, the R is enhanced from 0.401 in persistence forecasting to 0.641 in DWT-CNN-LSTM
(WD level 2). The performance improvement achieved by DWT-CNN-LSTM (WD level 2) can also
be found when compared with other forecasting models shown in Table 14.
Moreover, it should be noted the applicability degree of DWT-CNN-LSTM model in different
weather conditions is different. For instance, as mentioned in Section 3.5.1, the MAE of sunny days’
forecasting is decreased little with 30.271 in the persistence forecasting model and 23.174 in the DWT-
CNN-LSTM model. Nevertheless, in Table 12, the MAE of heavy rainy’ forecasting is reduced a lot
Appl. Sci. 2018, 8, 1286 22 of 30
from 64.416 in the persistence forecasting model to 38.642 in the DWT-CNN-LSTM model. This
further indicates that our proposed model is more applicable for the solar irradiance forecasting of
extreme weather conditions. Similar results can also be found in Table 15. Figure 10 shows the actual
and forecasted solar irradiance curve for rainy day pattern using dataset of Elizabeth City State
University.
Table 15. The performance comparison of different heavy rainy days’ forecasting models using the
dataset of Desert Rock Station.

CNN-LSTM 63.448 95.374 0.647
CNN 64.743 96.774 0.640
LSTM 65.014 97.096 0.641
ANN 66.312 99.863 0.615
ARIMA 79.473 120.744 0.477
Figure 10. Actual and forecasted solar irradiance on heavy rainy day pattern using dataset of
Elizabeth City State University.
3.6. Simulation Discussion

In this paper, an improved DL model (i.e., DWT-CNN-LSTM) based on WD, CNN, and LSTM
is proposed for day-ahead solar irradiance forecasting. In the actual simulation based on two datasets,
the model performance of DWT-CNN-LSTM model with Different WD Level is assessed for four
general weather types (i.e., sunny, cloudy, rainy, and heavy rainy). At the same time, the DWT-CNN-
LSTM model with certain WD Level is also compared with other DL models (e.g., CNN and LSTM)
and traditional forecasting models (e.g., ANN, persistence forecast and ARIMA) for each weather
type. The information previously shown in Tables 5–15 is vividly described in the following Figures
11–14, which is conducive to further summary. The changing trends of bars in these four figures are
similar, which can be summarized as follows.
First of all, it can be concluded that the influence of WD on forecasting performance, as well as
the best WD level, generally varies under different weather types and validation datasets.
Additionally, the introduction of certain WD level can improving the forecasting performance of
DWT-CNN-LSTM model for cloudy, rainy and heavy rainy days, excluding sunny day. The
conclusions are revealed by the fact in Figures 11–14 that the heights of all the blue bars (represent
Appl. Sci. 2018, 8, 1286 23 of 30
DWT-CNN-LSTM models with different WD Level) of sunny day are higher than the dark green bars
(represents CNN-LSTM model). This can be explained by the fact that the solar irradiance curve of
cloudy, rainy and heavy rainy days presents higher volatility, variability and randomness than that
of sunny days. Therefore, the raw solar irradiance sequence of cloudy, rainy and heavy rainy days
probably includes nonlinear and dynamic components in the form of spikes and fluctuations. The
existence of these components will undoubtedly deteriorate the precision of the solar irradiance
forecasting models. Additionally, the application of WD could mitigate the above problems.
Figure 11. The MAE of different forecasting models for sunny, cloudy, rainy and heavy rainy days
using the dataset of Elizabeth City State University.
Figure 12. The RMSE of different forecasting models for sunny, cloudy, rainy and heavy rainy days
using the dataset of Elizabeth City State University.
Appl. Sci. 2018, 8, 1286 24 of 30
Figure 13. The MAE of different forecasting models for sunny, cloudy, rainy and heavy rainy days
using the dataset of Desert Rock Station.
Figure 14. The RMSE of different forecasting models for sunny, cloudy, rainy and heavy rainy days
using the dataset of Desert Rock Station.
Secondly, the proposed DWT-CNN-LSTM models with suitable WD Level are always superior
to other DL models (e.g., CNN and LSTM) and traditional forecasting models (e.g., ANN, persistence
forecast and ARIMA) for cloudy, rainy and heavy rainy days. For sunny days, the CNN-LSTM model
without WD also performs better than other DL models and traditional forecasting models. The
performance enhancement can be attributed to the application of WD and the reasonable tandem
connection of CNN and LSTM. WD is used to decompose the raw solar irradiance sequence data of
certain weather types into several subsequences with better behaviors (e.g., more stable variances
and fewer outliers). CNN is good at automatically and effectively extracting representative and
significant information from the raw subsequence data. As shown in Figure 15, the sequential
characteristics with low and high frequency are well captured by CNN. LSTM is able to find the long
dependencies of the time series input.
Appl. Sci. 2018, 8, 1286 25 of 30
Figure 15. The visualization of feature maps extracted by CNN from the raw subsequence data. (a)
the original data before convolution operation; (b) The first feature map yielded by convolution
operation; (c) the second feature map yielded by convolution operation; and (d) the third feature map
yielded by convolution operation.
In the end, it should be noted that the applicability degree of DWT-CNN-LSTM model under
the different weather is not the same. Specifically speaking, the height differences of bars under
different weather types reveal that our proposed DWT-CNN-LSTM model obviously performs better
than traditional forecasting models (e.g., ARMIA) under cloudy, rainy and heavy rainy days. In other
words, our proposed model is more applicable for the solar irradiance forecasting of extreme weather
conditions. However, as shown in Figures 7–10, there still exists a certain deviation between the
actual solar irradiance value and the predicted value. This may be explained by the fact that the DWT-
based decomposition of raw solar irradiance data may miss part of the information. It is an important
problem needed be overcome in the next research stage.
4. Conclusions
The nature of the volatility and randomness characteristics of the output power of solar PV
generation causes serious difficulty for the real-time power balance of the interconnected grid. This
makes PV power forecasting become an important issue to the power grid in terms of the effective
integration of large-scale PV plants. As the main influence factor of PV power generation, the solar
irradiance and its accurate forecasting are prerequisites for solar PV power forecasting. Therefore,
this paper proposes an improved DL model to enhance the accuracy of day-ahead solar irradiance
forecasting. It should be noted that the DWT-CNN-LSTM model is individually established under
four general weather types (i.e., sunny, cloudy, rainy and heavy rainy) due to the high dependency
of solar irradiance on weather status.
The basic pipeline framework behind the data-driven DWT-CNN-LSTM model consists of three
major parts: (1) DWT based solar irradiance sequence decomposition; (2) the CNN-based local feature
extractor; and (3) the LSTM-based sequence forecasting model. In the solar irradiance forecasting
under certain weather types, the raw solar irradiance sequence is decomposed into several
subsequences via discrete wavelet transformation. Then each subsequence is fed to the CNN-based
local feature extractor, which leverages the advantage of CNN to automatically learn the abstract
feature representation from the raw subsequence data. Since the extracted features are also time series
data, they are individually transported to LSTM to construct the subsequence forecasting model. In
the end, the final solar irradiance forecasting results under certain weather types are obtained via the
wavelet reconstruction of these forecasted subsequences.
In the case study using two datasets of Elizabeth City State University and Desert Rock Station,
the performance of the proposed DWT-CNN-LSTM model is compared with another six solar
irradiance forecasting models, namely, CNN-LSTM (i.e., our proposed model without WD), ANN,
manually extracted features-ANN, persistence forecasting, CNN, and LSTM. Based on three error
Appl. Sci. 2018, 8, 1286 26 of 30
indexes (i.e., RMSE, MAE, and R), the simulation results indicate that DWT-CNN-LSTM model has
high superiority in the solar irradiance forecasting, especially under extreme weather conditions. This
mans the proposed DL technique-based day-ahead solar irradiance forecasting model has high
potential for future practical applications.
Author Contributions: All authors have worked on this manuscript together and all authors have read and
approved the final manuscript. F.W., Y.Y. and Z.Z. (Zhanyao Zhang) conceived and designed the experiments;
Y.Y. and Z.Z. (Zhanyao Zhang) performed the experiments; J.L., K.L., Z.Z. (Zhao Zhen) analyzed the data; F.W.
and Y.Y. wrote the paper.
Funding: This work was supported by the National Key R&D Program of China (2018YFB0904200), the National
Natural Science Foundation of China (51577067), the Beijing Natural Science Foundation of China (3162033), the
Hebei Natural Science Foundation of China (E2015502060), the State Key Laboratory of Alternate Electrical
Power System with Renewable Energy Sources (LAPS18008), the Science and Technology Project of State Grid
Corporation of China (SGCC)(NY7117020), the Open Fund of State Key Laboratory of Operation and Control of
Renewable Energy & Storage Systems (China Electric Power Research Institute) (5242001600FB), and the
Fundamental Research Funds for the Central Universities (2018QN077).
Conflicts of Interest: The authors declare no conflict of interest.
Nomenclature
PV photovoltaic
DL deep learning
WD wavelet decomposition
CNN convolutional neural network
LSTM long short-term memory
IEA international energy agency
NWP numerical weather prediction
TSI total sky imagery
MOS model output statistics
MA moving average
AR autoregressive
ARMA autoregressive moving average
ARIMA autoregressive integrated moving average
ANN artificial neural network
SVM support vector machine
SOM self-organizing map
LVQ learning vector quantization
SVR support vector regression
A-P Angstrom-Prescott
RNN recurrent neural network
MFE manual feature extraction
DWT discrete wavelet transformation
LPF low-pass filters
HPF high-pass filters
ReLu rectified linear units
RMSE root mean squared error
MAE mean absolute error
R correlation coefficient
 (t ) mother wavelet function
 (t ) scaling function
 j,k ( t ) a sequence of wavelet at time index t

 j,k ( t ) binary scale-functions at time index t
os ( t ) the original sequence at time index t
Appl. Sci. 2018, 8, 1286 27 of 30
c j ,k the approximation coefficient at scale j and location k

d j ,k the detailed coefficient at scale j and location k
S the raw sequential input
FS the filter sequence
L the length of raw sequential input
K the number of total filters in the convolutional Layer
m window size
 the concatenation operator
si:i+m- 1 a concatenation vector representation of si  si+1  si+ 2    si+m-1
f,σ a nonlinear activation function
wj the j -th filter matrix
Fj the feature map of j -th filter
p pooling size
Fj-compress the compressed feature vector from the pooling layer
X the sequential vectors

x (t ) the input variable at time step t
y (t ) the expected output at time step t
W ,U ,V weight matrixes
b,c biases vectors
h (t ) a short-term state
c (t ) long-term state
References
1. Wang, F.; Zhen, Z.; Mi, Z.; Sun, H.; Su, S.; Yang, G. Solar irradiance feature extraction and support vector
machines based weather status pattern recognition model for short-term photovoltaic power forecasting.
Energy Build. 2015, 86, 427–438, doi:10.1016/j.enbuild.2014.10.002.
2. World Energy Outlook 2016. Available online: https://www.iea.org/newsroom/news/2016/november/
world-energy-outlook-2016.html (accessed on 20 June 2018)
3. Renewables 2017: Global Status Report. Available online: http://www.ren21.net/gsr-2017/ (accessed on 20
June 2018)
4. Inman, R.H.; Pedro, H.T.C.; Coimbra, C.F.M. Solar forecasting methods for renewable energy integration.
Prog. Energy Combust. Sci. 2013, 39, 535–576, doi:10.1016/j.pecs.2013.06.002.
5. Yona, A.; Senjyu, T.; Funabashi, T.; Mandal, P.; Kim, C.-H. Optimizing Re-planning Operation for Smart
House Applying Solar Radiation Forecasting. Appl. Sci. 2014, 4, 366–379, doi:10.3390/app4030366.
6. Sun, Y.; Wang, F.; Wang, B.; Chen, Q.; Engerer, N.A.; Mi, Z. Correlation feature selection and mutual
information theory based quantitative research on meteorological impact factors of module temperature
for solar photovoltaic systems. Energies 2017, 10, 7, doi:10.3390/en10010007.
7. Wang, J.; Li, P.; Ran, R.; Che, Y.; Zhou, Y. A short-term photovoltaic power prediction model based on the
Gradient Boost Decision Tree. Appl. Sci. 2018, 8, 689, doi:10.3390/app8050689.
8. Baños, R.; Manzano-Agugliaro, F.; Montoya, F.G.; Gil, C.; Alcayde, A.; Gómez, J. Optimization methods
applied to renewable and sustainable energy: A review. Renew. Sustain. Energy Rev. 2011, 15, 1753–1766,
doi:10.1016/j.rser.2010.12.008.
9. Sharma, A.; Kakkar, A. Forecasting daily global solar irradiance generation using machine learning. Renew.
Sustain. Energy Rev. 2018, 82, 2254–2269, doi:10.1016/j.rser.2017.08.066.
10. Christensen-Dalsgaard, J. Physics of Solar-Like Oscillations. Highlights Astron. 2005, 13, 397–402,
doi:10.1017/S1539299600022711.
11. Marquez, R.; Coimbra, C.F.M. Forecasting of global and direct solar irradiance using stochastic learning
methods, ground experiments and the NWS database. Sol. Energy 2011, 85, 746–756,
Appl. Sci. 2018, 8, 1286 28 of 30
doi:10.1016/j.solener.2011.01.007.
12. Li, J.; Ward, J.K.; Tong, J.; Collins, L.; Platt, G. Machine learning for solar irradiance forecasting of
photovoltaic system. Renew. Energy 2016, 90, 542–553, doi:10.1016/j.renene.2015.12.069.
13. Diagne, M.; David, M.; Lauret, P.; Boland, J.; Schmutz, N. Solar irradiation forecasting: state-of-the-art and
proposition for future developments for small-scale insular grids. Renew. Sustain. Energy Rev. 2013, 27, 65–
76, doi:10.1016/j.rser.2013.06.042.
14. Reikard, G. Predicting solar radiation at high resolutions: A comparison of time series forecasts. Sol. Energy
2009, 83, 342–349, doi:10.1016/j.solener.2008.08.007.
15. Lorenz, E.; Hammer, A.; Heinemann, D. Short term forecasting of solar radiation based on satellite data. In
Proceedings of the EuroSun 2004 ISES Europe Solar Congress, Freiburg, Germany, 20–23 June 2004; pp. 841–
848.
16. Arbizu-Barrena, C.; Ruiz-Arias, J.A.; Rodríguez-Benítez, F.J.; Pozo-Vázquez, D.; Tovar-Pescador, J. Short-
term solar radiation forecasting by advecting and diffusing MSG cloud index. Sol. Energy 2017, 155, 1092–
1103, doi:10.1016/j.solener.2017.07.045.
17. Wang, F.; Zhen, Z.; Liu, C.; Mi, Z.; Hodge, B.M.; Shafie-khah, M.; Catalão, J.P.S. Image phase shift
invariance based cloud motion displacement vector calculation method for ultra-short-term solar PV
power forecasting. Energy Convers. Manag. 2018, 157, 123–135, doi:10.1016/j.enconman.2017.11.080.
18. Wang, F.; Li, K.; Wang, X.; Jiang, L.; Ren, J.; Mi, Z.; Shafie-khah, M.; Catalão, J.P.S. A Distributed PV System
Capacity Estimation Approach Based on Support Vector Machine with Customer Net Load Curve Features.
Energies 2018, 11, 1750.
19. Verzijlbergh, R.A.; Heijnen, P.W.; de Roode, S.R.; Los, A.; Jonker, H.J.J. Improved model output statistics
of numerical weather prediction based irradiance forecasts for solar power applications. Sol. Energy 2015,
118, 634–645, doi:10.1016/j.solener.2015.06.005.
20. Bacher, P.; Madsen, H.; Nielsen, H.A. Online short-term solar power forecasting. Sol. Energy 2009, 83, 1772–
1783, doi:10.1016/j.solener.2009.05.016.
21. Huang, R.; Huang, T.; Gadh, R.; Li, N. Solar generation prediction using the ARMA model in a laboratory-
level micro-grid. In Proceedings of the 2012 IEEE Third International Conference Smart Grid
Communications, Tainan, Taiwan, 5–8 November 2012.
22. Perdomo, R.; Banguero, E.; Gordillo, G. Statistical Modeling for Global Solar Radiation Forecasting in
Bogotá. In Proceedings of the 2010 35th IEEE Photovoltic Specialists Conference, Honolulu, HI, USA, 20–25
June 2010; pp. 2374–2379.
23. Wang, F.; Li, K.; Liu, C.; Mi, Z.; Shafie-khah, M.; Catalao, J.P.S. Synchronous Pattern Matching Principle
Based Residential Demand Response Baseline Estimation: Mechanism Analysis and Approach Description.
IEEE Trans. Smart Grid 2018, 3053, 1–13, doi:10.1109/TSG.2018.2824842.
24. Chen, Q.; Wang, F.; Hodge, B.-M.; Zhang, J.; Li, Z.; Shafie-Khah, M.; Catalao, J.P.S. Dynamic Price Vector
Formation Model-Based Automatic Demand Response Strategy for PV-Assisted EV Charging Stations.
IEEE Trans. Smart Grid 2017, 8, 2903–2915, doi:10.1109/TSG.2017.2693121.
25. Wang, F.; Xu, H.; Xu, T.; Li, K.; Shafie-Khah, M.; Catalao, J.P.S. The values of market-based demand
response on improving power system reliability under extreme circumstances. Appl. Energy 2017, 193, 220–
231, doi:10.1016/j.apenergy.2017.01.103.
26. Wang, F.; Zhou, L.; Ren H.; Liu, X.; Shafie-khah, M. Multi-objective Optimization Model of Source-Load-
Storage Synergetic Dispatch for Building Energy System Based on TOU Price Demand Response. IEEE
Trans. Ind. Appl. 2018, 54, 1017–1028.
27. Maier, H.R.; Dandy, G.C. Neural networks for the prediction and forecasting of water resources variables:
A review of modelling issues and applications. Environ. Model. Softw. 2000, 15, 101–124, doi:10.1016/S1364–
8152(99)00007–9.
28. Wang, F.; Mi, Z.; Su, S.; Zhao, H. Short-Term Solar Irradiance Forecasting Model Based on Artificial Neural
Network Using Statistical Feature Parameters. Energies 2012, 5, 1355–1370, doi:10.3390/en5051355.
29. Zeng, J.; Qiao, W. Short-term solar power prediction using a support vector machine. Renew. Energy 2013,
52, 118–127, doi:10.1016/j.renene.2012.10.009.
30. Shakya, A.; Michael, S.; Saunders, C.; Armstrong, D.; Pandey, P.; Chalise, S.; Tonkoski, R. Using Markov
Switching Model for solar irradiance forecasting in remote microgrids. In Proceedings of the 2016 IEEE
Energy Conversion Congress and Exposition, Milwaukee, WI, USA, 18–22 September 2016; pp. 895–905.
31. Wang, F.; Zhen, Z.; Wang, B.; Mi, Z. Comparative Study on KNN and SVM Based Weather Classification
Appl. Sci. 2018, 8, 1286 29 of 30
Models for Day Ahead Short Term Solar PV Power Forecasting. Appl. Sci. 2017, 8, 28,
doi:10.3390/app8010028.
32. Gala, Y.; Fernández, Á.; Díaz, J.; Dorronsoro, J.R. Hybrid machine learning forecasting of solar radiation
values. Neurocomputing 2016, 176, 48–59, doi:10.1016/j.neucom.2015.02.078.
33. Wang, F.; Zhou, L.; Ren H.; Liu, X. Search Improvement Process-Chaotic Optimization-Particle Swarm
Optimization-Elite Retention Strategy and Improved Combined Cooling-Heating-Power Strategy Based
Two-Time Scale Multi-Objective Optimization Model for Stand-Alone Microgrid Operation. Energies 2017,
10, 1936, doi.org/10.3390/en10121936.
34. Yang, D.; Kleissl, J.; Gueymard, C.A.; Pedro, H.T.C.; Coimbra, C.F.M. History and trends in solar irradiance
and PV power forecasting: A preliminary assessment and review using text mining. Sol. Energy 2018, 168,
60–101, doi:10.1016/j.solener.2017.11.023.
35. Wan, C.; Zhao, J.; Song, Y.; Xu, Z.; Lin, J.; Hu, Z. Photovoltaic and solar power forecasting for smart grid
energy management. CSEE J. Power Energy Syst. 2015, 1, 38–46, doi:10.17775/CSEEJPES.2015.00046.
36. Ferlito, S.; Adinolfi, G.; Graditi, G. Comparative analysis of data-driven methods online and offline trained
to the forecasting of grid-connected photovoltaic plant production. Appl. Energy 2017, 205, 116–129,
doi:10.1016/j.apenergy.2017.07.124.
37. Yang, H.-T.; Huang, C.-M.; Huang, Y.-C.; Pai, Y.S. A Weather-Based Hybrid method for one-day ahead
hourly forecasting of PV power output. IEEE Trans. Sustain. Energy 2014, 5, 917–926,
doi:10.1109/TSTE.2014.2313600.
38. Gensler, A.; Henze, J.; Sick, B.; Raabe, N. Deep Learning for solar power forecasting-An approach using
AutoEncoder and LSTM Neural Networks. In Proceedings of the 2016 IEEE International Conference on
Systems, Man, and Cybernetics, Budapest, Hungary, 9–12 October 2016.
39. Hussain, S.; Alili, A. Day ahead hourly forecast of solar irradiance for Abu Dhabi, UAE. In Proceedings of
the 2016 IEEE Smart Energy Grid Engineering (SEGE), Oshawa, ON, Canada, 21-24 August 2016.
40. Akarslan, E.; Hocaoglu, F.O.; Edizkan, R. Novel short term solar irradiance forecasting models. Renew.
Energy 2018, 123, 58–66, doi:10.1016/j.renene.2018.02.048.
41. Zhen, Z.; Wan, X.; Wang, Z.; Wang, F.; Ren, H.; Mi, Z. Multi-level wavelet decomposition based day-ahead
solar irradiance forecasting. In Proceedings of the 2018 IEEE Power Energy Society Innovative Smart Grid
Technologies Conference (ISGT), Washington, DC, USA, 19–22 February 2018; pp. 1–5.
42. Wang, F.; Zhen, Z.; Liu, C.; Mi, Z.; Shafie-Khah, M.; Catalão, J.P.S. Time-section fusion pattern classification
based day-ahead solar irradiance ensemble forecasting model using mutual iterative optimization. Energies
2018, 11, 184, doi:10.3390/en11010184.
43. Qing, X.; Niu, Y. Hourly day-ahead solar irradiance prediction using weather forecasts by LSTM. Energy
2018, 148, 461–468, doi:10.1016/j.energy.2018.01.177.
44. Llamas, J.; Lerones, P.M.; Medina, R.; Zalama, E.; Gómez-García-Bermejo, J. Classification of Architectural
Heritage Images Using Deep Learning Techniques. Appl. Sci. 2017, 7, 992, doi:10.3390/app7100992.
45. Almeida, A.; Azkune, G. Predicting Human Behaviour with Recurrent Neural Networks. Appl. Sci. 2018, 8,
305, doi:10.3390/app8020305.
46. Yoo, Y.; Baek, J.-G. A Novel Image Feature for the Remaining Useful Lifetime Prediction of Bearings Based
on Continuous Wavelet Transform and Convolutional Neural Network. Appl. Sci. 2018, 8, 1102. 1,
doi:10.3390/app8071102.
47. Panapakidis, I.P.; Dagoumas, A.S. Day-ahead natural gas demand forecasting based on the combination of
wavelet transform and ANFIS/genetic algorithm/neural network model. Energy 2017, 118, 231–245,
doi:0.1016/j.energy.2016.12.033.
48. Mallat, S.G. A Theory for Multiresolution Signal Decomposition: The Wavelet Representation. IEEE
Computer Soc. 1989, 11, 674–693.
49. Zhao, R.; Yan, R.; Wang, J.; Mao, K. Learning to Monitor Machine Health with Convolutional Bi-Directional
LSTM Netw. Sens. 2017, 17, 273.
50. Gu, J.; Wang, Z.; Kuen, J.; Ma, L.; Shahroudy, A.; Shuai, B.; Liu, T.; Wang, X.; Wang, G.; Cai, J.; et al. Recent
advances in convolutional neural networks. Pattern Recognit. 2018, 77, 354–377,
doi:10.1016/j.patcog.2017.10.013.
51. Nair, V.; Hinton, G.E. Rectified linear units improve restricted boltzmann machines. In Proceedings of the
27th International Conference on Machine Learning, Haifa, Israel, 21–24 June 2010; pp. 807–814.
Appl. Sci. 2018, 8, 1286 30 of 30
52. Längkvist, M.; Karlsson, L.; Loutfi, A. A review of unsupervised feature learning and deep learning for
time-series modeling. Pattern Recognit. Lett. 2014, 42, 11–24, doi:10.1016/j.patrec.2014.01.008.
53. Hochreiter, S.; Schmidhuber, J. Long Short-Term Memory. Neural Comput. 1997, 9, 1735–1780.
54. Bengio, Y.; Simard, P.; Frasconi, P. Learning Long-Term Dependencies with Gradient Descent is Dicfficult.
IEEE Trans. Neural Netw. 1994, 5, 157–166.
55. US Department of Energy, NREL, National Renewable Energy Laboratory. Available online:
https://rredc.nrel.gov/solar/new_data/confrrm/bs/ (accessed on 20 June 2018).
56. US Department of Commerce, NOAA, Earth System Research Laboratory. Available online:
https://www.esrl.noaa.gov/gmd/grad/surfrad/ (accessed on 20 June 2018).
57. Keras Documentation. Available online: https://keras.io/ (accessed on 20 June 2018).
58. Scikit-learn: Machine Learning in Python. Available online: http://scikit-learn.github.io/stable (accessed on
20 June 2018).
© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

8.-Wavelet Decomposition and Convolutional LSTM Solar Irradiance Forecasting-2018

Uploaded by

Copyright:

Available Formats

8.-Wavelet Decomposition and Convolutional LSTM Solar Irradiance Forecasting-2018

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

8.-Wavelet Decomposition and Convolutional LSTM Solar Irradiance Forecasting-2018

Uploaded by

Copyright:

Available Formats

Article

Wavelet Decomposition and Convolutional LSTM

[email protected] (Y.Y.); [email protected] (Z.Z.); [email protected] (J.L.);

University, Baoding 071003, China

Received: 8 July 2018; Accepted: 24 July 2018; Published: 1 August 2018

Keywords: solar irradiance forecasting; wavelet decomposition; convolutional neural network;

1.1. Background and Motivation

Appl. Sci. 2018, 8, 1286; doi:10.3390/app8081286 www.mdpi.com/journal/applsci

1.2. Literature Review

1.3. The Content and Contribution of the Paper

2. Improved Deep Learning Model for Day-Ahead Solar Irradiance Forecasting

Raw solar irradiance sequence

The classification of weather types

1500 1500 1500 1500

1000 1000 1000 1000

Day-ahead solar Day-ahead solar Day-ahead solar Day-ahead solar

Solar irradiance Solar irradiance Solar irradiance Solar irradiance

Raw solar irradiance sequence under certain weather type

Normalization and k-level discrete wavelet transformation (DWT)

Approximate Detailed Detailed Detailed

LSTM based LSTM based LSTM based LSTM based

Forecasting results Forecasting results Forecasting results Forecasting results

Wavelet reconstruction and anti-normalization

Final forecasting results of solar irradiance under certain weather type

2.1. Discrete Wavelet Transformation Based Solar Irradiance Sequence Decomposition

Level 1 Level 2 Level 3 Level k

2.2. Convolutional Neural Networks Based Local Feature Extractor

si:i+m-1 =si  si+1  si+ 2      si+m-1 (6)

Fj = [c1 ,c2 ,c3 , ,cL−m+1 ] (8)

Fj-compress = [h1 ,h2 ,h3 , ,h L-m ]

Sliding Feature Feature map after

Time steps LSTM based

Convolution Non- linear Pooling

2.3.1. Recurrent Neural Network

X= [x(0),x(1),x(2)] x(0) x(1) x(2)

Figure 5. The structure of Recurrent Neural Network.

2.3.2. Long-Short-Term Memory

f(t) g(t) i(t)

Figure 6. The structure of Long Short-Term Memory Cell.

f ( t ) =σ(W f  [h ( t-1) ,x ( t )]+b f ) (12)

in which W f weight matrix, b f is biases vectors, and σ is activation function.

i ( t ) =σ(Wi  [h ( t-1) ,x ( t )] + bi ) (13)

g ( t ) =tanh(Wg  [h ( t-1) ,x ( t )]+bg ) (14)

c ( t ) =f ( t )  c ( t-1) +it  g t (15)

o ( t ) = σ(Wo  [h ( t-1) ,x ( t )]+bo ) (16)

3.1. Data Source and Experimental Setup

3.2. Model Training and Hyperparameters Selection

Weather Types Elizabeth City Desert Rock Station

Table 2. The parameter setting detail of DWT-CNN-LSTM model.

Option Parameter Setting

Layer Output Shape

Table 4. The structure of the other forecasting models used as benchmarks.

Forecasting Models Structure

3.3. Performance Criterion

Weather Error Wavelet Decomposition (WD) Level

Weather Error Wavelet Decomposition (WD) Level

3.5. Performance Comparison Analysis of Different Solar Irradiance Forecasting Models

3.5.1. Comparison Analysis of Sunny Days

Table 7. The list of manually extracted features.

Statistical Features Expression