Solar Energy 163 (2018) 189–199

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

A decomposition-clustering-ensemble learning approach for solar radiation T

forecasting
⁎
Shaolong Suna,b,c, Shouyang Wanga,b,d, , Guowei Zhanga,b, Jiali Zhenga,b
a
Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100190, China
b
School of Economics and Management, University of Chinese Academy of Sciences, Beijing 100190, China
c
Department of Systems Engineering and Engineering Management, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong
d
Center for Forecasting Science, Chinese Academy of Sciences, Beijing 100190, China

A R T I C L E I N F O A B S T R A C T

Keywords:
Solar radiation forecasting
Decomposition-clustering-ensemble learning
approach
Ensemble empirical mode decomposition
Least square support vector regression
A decomposition-clustering-ensemble (DCE) learning approach is proposed for solar radiation forecasting in this
paper. In the proposed DCE learning approach, (1) ensemble empirical mode decomposition (EEMD) is used to
decompose the original solar radiation data into several intrinsic mode functions (IMFs) and a residual com-
ponent; (2) least square support vector regression (LSSVR) is performed to forecast IMFs and residual component
respectively with parameters optimized by gravitational search algorithm (GSA); (3) Kmeans method is adopted
to cluster all component forecasting results; (4) another GSA-LSSVR method is applied to ensemble the com-
ponent forecasts of each cluster and the final forecasting results are obtained by means of corresponding cluster’s
ensemble weights. To verify the performance of the proposed DCE learning approach, solar radiation data in
Beijing is introduced for empirical analysis. The results of out-of-sample forecasting power show that the DCE
learning approach produces smaller NRMSE, MAPE and better directional forecasts than all other benchmark
models, reaching up to accuracy rate of 2.96%, 2.83% and 88.24% respectively in the one-day-ahead forecasting.
This indicates that the proposed DCE learning approach is a relatively promising framework for forecasting solar
radiation by means of level accuracy, directional accuracy and robustness.

1. Introduction
With the continuous consumption of fossil fuels, energy and en-
vironmental problems become increasingly severe. It is urgent to find a
solution to solve energy and environmental problems and achieve
sustainable development. Therefore, renewable energy has attracted
much attention all over the world and been rapidly developed in recent
years.
The solar energy is considered to be one of the cleanest and most
promising renewable energy sources, which makes solar power an im-
portant direction of exploring renewable energy. The solar power
generation is widely used in developed countries and considerable de-
veloping countries at present, and has partially replaced the traditional
power generation. In recent years, China has gained more than 25%
growth rate every year in the development and utilization of renewable
energy. The solar radiation plays an important role in solar photovoltaic
power generation. With the development of solar technologies, de-
mands for solar radiation data with high accuracy are increasing.
Consequently, the solar radiation forecasting has become one of core
contents of solar photovoltaic power generation.
The scholars and researchers so far have conducted in-depth re-
search for solar radiation forecasting theory and put forward a number
of feasible and efficient methods to predict actual solar radiation in-
tensity, which has achieved satisfactory results. These methods can be
divided into three categories: traditional mathematical statistics, nu-
merical weather forecasting and machine learning. The traditional
mathematical statistics includes: regression analysis (Trapero et al.,
2015), time series analysis (Huang et al., 2013; Voyant et al., 2013),
gray theory (Fidan et al., 2014), fuzzy theory (Chen et al., 2013), wa-
velet analysis (Mellit et al., 2006; Monjoly et al., 2017) and Kalman
filter (Akarslan et al., 2014), etc.; numerical weather forecasting solves
thermal fluid dynamics equations with weather evolution by computers
with high performance to predict the solar radiation intensity of the
next period mainly based on the actual situation of the atmosphere
(Chow et al., 2011; Mathiesen and Kleissl, 2011; Mathiesen et al.,
2013), which is complicated and time-consuming. With the rise of big
data mining, machine learning techniques currently have attracted
much attention, for example, artificial neural networks (ANN)
(Amrouche and Le Pivert, 2014; Benmouiza and Cheknane, 2013; Niu
et al., 2015; Paoli et al., 2010), support vector machines (SVM) (Gala

⁎
Corresponding author at: Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100190, China.
E-mail address: sywang@amss.ac.cn (S. Wang).

https://doi.org/10.1016/j.solener.2018.02.006
Received 28 July 2017; Received in revised form 26 December 2017; Accepted 1 February 2018
0038-092X/ © 2018 Elsevier Ltd. All rights reserved.
S. Sun et al.
et al., 2016; Lauret et al., 2015) and heuristic intelligent optimization
algorithms (Jiang et al., 2015; Niu et al., 2017; Niu et al., 2016a; Wang
et al., 2015) have been widely used in solar radiation forecasting.
According to the literature above, Trapero et al. (2015) applied
dynamic harmonic regression model (DHR) to predict short-term direct
solar radiation and scattered solar radiation in Spain for the first time in
2015. Huang et al. (2013) took autoregressive model to forecast the
solar radiation in the framework of meteorological factors-dynamic
adjustment system in 2013, which increased the accuracy by 30% than
general neural network or random models. Fidan et al. (2014) predicted
hourly solar radiation in Izmir, Turkey by integration of Fourier
transform and neural network. Mellit et al. (2006) applied infinite
impulse response function to filter time series of the total solar radia-
tion in Algeria from 1981 to 2000, and then substituted the filtered data
into adaptive wavelet neural network model to forecast the total solar
radiation in 2001, where the error percentage is less than 6% per-
forming better than conventional neural network models and classical
statistical methods. However, Akarslan et al. (2014) first used the
multidimensional linear predictive filtering model to forecast the solar
radiation, which surpasses the two-dimensional linear predictive fil-
tering model and the traditional statistical forecasting method by means
of empirical analysis. Amrouche and Le Pivert (2014) adopted the idea
of spatial modeling and artificial neural networks (ANNs) to forecast
the daily solar total radiation of four sites in the United States, in which
the empirical results show that the forecasting results of this proposed
model satisfy the expected accuracy requirements. Benmouiza and
Cheknane (2013) classified the input data by K-means, then modeled
different class by using nonlinear autoregressive neural network, and
finally predicted the solar radiation of test data by the corresponding
model.
In recent years, the integrated model has grown rapidly. Paoli et al.
(2010) used the integrated model to predict the total solar radiation in
three sites in France. First, they pretreated the original total solar ra-
diation sequence by using the seasonal index adjustment method. And
then they used the multi-layer perceptron neural network (MLPNN) for
daily solar radiation prediction. The results show that the mean abso-
lute percentage error of the multi-layer perceptual neural network
model is about 6%, which is superior to the ARIMA, the Bayesian,
Markov chain model and K-nearest neighbor algorithm. Lauret et al.
(2015) used three different methods which are artificial neural network
(ANN), Gaussian process (GP) and support vector machine (SVM) to
forecast global horizontal solar irradiance (GHI). Through analyzing the
actual data in three different places in France, the three machine
learning algorithms proposed in this paper are found to be better than
the linear autoregressive (AR) and persistence model. Gala et al. (2016)
used a combination of support vector regression (SVR), gradient
boosted regression (GBR), and random forest regression (RFR) to pre-
dict three-hour accumulated solar radiation of 11 regions in Spain.
Wang et al. (2015) proposed a new integrated model to predict hourly
solar radiation in 2015. First, they used the network structure of multi-
response sparse regression (MRSR), leave-one-out cross validation, and
extreme learning machine (ELM), then used the cuckoo search (CS) to
optimize its weight and threshold, and finally analyzed six sites in the
US. The prediction results show that this combination model is stronger
than the ARIMA, BPNN and optimally pruned extreme learning ma-
chine (OP-ELM).
The main innovation of this paper is to propose a novel decom-
position clustering ensemble (DCE) learning approach integrating
EEMD, K-means and LSSVR to improve the performance of solar ra-
diation forecasting by means of forecasting accuracy and robustness,
and to compare its forecasting power with some popular existing
forecasting models. The rest of the paper is organized as follows.
Section 2 describes the formulation process of the proposed DCE
learning approach in detail. Related methodologies are illustrated in
Section 3. The empirical results and effectiveness of the proposed ap-
proach are discussed in Section 4. Finally, Section 5 provides some
190
Solar Energy 163 (2018) 189–199
conclusions and indicates the direction of further research.
2. Decomposition-Clustering-Ensemble (DCE) learning approach
According to the work of TEI@I methodology (Wang et al., 2005;
Wang, 2004), which is based on integration (@ Integration) of text
mining (T) plus econometrics (E) plus intelligence techniques (I). Yu
et al. (2008) proposed a decomposition ensemble learning approach for
crude oil price forecasting. Recently, this learning approach has been
applied in many fields, including financial time series forecasting (Yu
et al., 2009), nuclear energy consumption forecasting (Tang et al.,
2012; Wang et al., 2011), hydropower consumption forecasting (Wang
et al., 2011), crude oil price forecasting (He et al., 2012; Yu et al., 2014;
Yu et al., 2015; Yu et al., 2016), etc.
By analyzing the literature above in detail, there are three main
steps involved in the decomposition and ensemble learning approach,
i.e. decomposition, single forecasting and ensemble forecasting. Firstly,
some decomposition algorithms can be used to decompose the original
time series into a number of meaningful components. Secondly, some
forecasting methods are applied to forecast all components respectively.
Finally, these forecasting results of each component are combined to
generate an aggregated output as the final forecasting result using some
ensemble methods. It can be concluded that ensemble learning is cri-
tical to the final forecasting results. Sub-component forecasting has
different attributes in different time. If ensemble weights are the same
all the time, different attributes cannot be captured. Therefore, DCE
learning approach is proposed in this paper, which employs clustering
scheme to cluster the sub-component forecasting results. By using dif-
ferent ensemble weights in different forecasting time, a better perfor-
mance can be obtained by compared with the fixed ensemble weights.
The framework of DCE learning approach is shown in Fig. 1. It can
be seen from Fig. 1, DCE learning approach contains four steps:
(1) Decomposition. Decomposition method is introduced to decompose
the original time series into some relatively simple and meaningful
component series.
(2) Individual forecasting. Various forecasting models are employed to
forecast each component series.
(3) Clustering. A clustering method is used to cluster the individual
forecasting results.
(4) Ensemble forecasting. An ensemble method is applied to calculate
the ensemble weights of different cluster. Then the corresponding
clusters’ ensemble weights are used for component forecasts to
obtain the ensemble forecasting results.
3. Related methodologies
3.1. Ensemble empirical mode decomposition
The empirical mode decomposition (EMD) is an adaptive signal
decomposing method which is proposed by Huang et al. (1998). It can
be used to decompose the signal into several intrinsic mode functions
(IMFs) and a residual series.
Each IMF has a zero local mean value and its number of extreme
values and zero crossings are equal or differ at most by one. Different
IMFs have different frequency ranges and represents different kinds of
natural oscillation modes embedded in the original signal. Thus, dif-
ferent IMFs may highlight different details of the signal. Compared with
other methods, the key advantage of EMD is that the basis function is
derived directly from the original signal rather than a priori fixed basis
function. It is especially suitable for analyzing non-linear and non-sta-
tionary signals.
The ensemble empirical mode decomposition (EEMD) is proposed as
an improved version of EMD which overcomes the mode mixing pro-
blem of EMD. The main idea of EEMD is to add white noise into the
original signal with many trials. Then the EMD is applied to the noisy
S. Sun et al. Solar Energy 163 (2018) 189–199

Fig. 1. General framework of decomposition-clustering-ensemble (DCE) learning approach.

signal with additional white noise rather than the original signal to 3.3. Least square support vector machine
obtain the IMFs. For further details on these methods, please refer to
(Huang et al., 1998; Wu and Huang, 2009; Zhang et al., 2008). Support vector machine (SVM) is firstly proposed by Vapnik based
on statistical learning theory and the principle of structural risk mini-
mization which possess good performance even for small samples
3.2. Kmeans clustering (Vapnik, 2013). However, it is time-consuming and leads to a high
computational cost when deal with large-scale problem. Therefore,
Kmeans is an unsupervised learning algorithm for clustering, which least square support vector regression (LSSVR) is proposed by Suykens
was proposed by MacQueen (1967). The basic idea of Kmeans algo- and Vandewalle (1999).
rithm is to split the historical dataset into several subsets where each The basic idea of support vector regression is to map the original
subset has its own unique characteristics. Assuming that training set is data into a high-dimensional feature space and make a linear regression
X = {x1,x2,…,x n} , the detail steps of Kmeans clustering are as follows: in the space. It can be formulated into:
f (x ) = wT φ (x ) + b (1)
(1) Initialize cluster centroids (c1,c2,…,ck ) randomly.
(2) If ‖x i−ck ‖ ⩽ ‖x i−cp ‖,p = 1,2,…,K ,k ≠ p, x i is assigned to cluster k. where φ (x ) is a nonlinear mapping function, f (x ) is the estimation
1 N
(3) Update cluster centroids by ci = N ∑ j =i 1 x ij,i = 1,2,…,k . value, wT and b are the weights.
i
k N It can be transformed into an optimization problem:
(4) Compute the least square error of clustering J = ∑i = 1 ∑ j =i 1 ‖x ij−ci ‖2 .
(5) Repeat step 2–4 until J convergence. T
1 T
min 2
w w + C ∑ (ξt + ξt∗)
t=1
From the above processes, it can be seen that the number of clusters
T
K and the election of the initial cluster center are critical to the final ⎧ w φ (x t ) + b−yt ⩽ ε + ξt ,(t = 1,2,…,T )
⎪
results of clustering. For further details on this method, please refer to s. t . y −(wT φ (x t ) + b) ⩽ ε + ξt∗,(t = 1,2,…,T )
⎨ t ∗
(MacQueen, 1967). ⎪ ξt ,ξt ⩾ 0
⎩ (2)

where C is the penalty parameter, ξt and ξt∗ are the nonnegative slack

191
S. Sun et al.
Fig. 2. The framework of EEMD-LSSVR-K-LSSVR learning approach.
variables.
It is time-consuming to solve the above problem and LSSVR is
proposed to transform the problem into:
T (3) Kmeans is introduced to cluster the sub-series forecasting results of
1 T
min 2
w w + C ∑ et2
t=1
s. t . yt = wT φ (x t ) + b + et ,(t = 1,2,…,T )
where et is the slack variable.
Generally speaking, the parameters of SVR and LSSVR have a great
influence on the accuracy of the regression estimation. Thereby, grav-
itational search algorithm (GSA) is employed to automatically choose
the optimal parameters of SVR and LSSVR in this paper. For more in-
formation about GSA, please refer to (Niu et al., 2016b; Rashedi et al.,
2009).
3.4. EEMD-LSSVR-K-LSSVR learning approach
An EEMD-LSSVR-K-LSSVR learning approach is obtained by using
EEMD as the decomposition method, Kmeans as the clustering method,
LSSVR as the forecasting and ensemble method. The framework of
EEMD-LSSVR-K-LSSVR learning approach is illustrated in Fig. 2.
In Fig. 2, the proposed EEMD-LSSVR-K-LSSVR learning approach
consists of four steps:
(1) EEMD is applied to decompose the original series into the IMFs and
Solar Energy 163 (2018) 189–199
residual series.
(2) Each component series is modeled by using optimized LSSVR re-
spectively.
the in-sample to obtain cluster centers.
(4) Another optimized LSSVR is employed to calculate each cluster’s
(3) ensemble weights. By using corresponding clusters’ ensemble
weights, final forecasting results are obtained.
4. Empirical analysis
In this section, Beijing daily solar radiation data are carried out to
evaluate the performance of the proposed DCE approach. The data
description and evaluation criteria are first introduced in Section 4.1.
Then empirical results are further discussed in Section 4.2.
4.1. Data description and evaluation criteria
The daily solar radiation data used in this paper are obtained from a
meteorological station of Beijing with 3103 observations ranging from
January 1, 2009 to June 30, 2017, which is shown in Fig. 3. The da-
tasets are divided into in-sample subset and out-of-sample subset. The
in-sample subset is used for model training with data from January 1,
2009 to June 30, 2015, while out-of-sample subset is used for model
testing with data from July 1, 2015 to June 30, 2017. The detailed data
192
S. Sun et al.
Fig. 3. The original daily solar radiation series in Beijing.
are not listed here, but can be obtained from the authors.
The m-step-ahead forecasting horizons are employed to evaluate the
performance of the proposed DCE learning approach. Given a time
̂ m:
series x t ,(t = 1,2,…,T ) , we make m-step-ahead forecasting for x t +
x t +̂ m = f (x t ,x t − 1,…,x t − (l − 1) ),t = 1,2,…,T .
̂ m is the m-step-ahead forecast value at time t , x t is the actual
where x t +
value at time t , and l denotes the lag orders which is chosen by auto-
correlation and partial correlation analysis (Lewis, 1982).
In addition, in order to assess the level forecasting accuracy and the
directional forecasting accuracy of the proposed DCE learning approach
with some other benchmark models. Three main evaluation criteria are
employed to compare the in-sample and out-of-sample forecasting
performance (Coimbra et al., 2013). For level forecasting accuracy
evaluation, the normalized root mean squared error (NRMSE) and the
mean absolute percentage error (MAPE) are selected:
T
100 1 T 1 x t −x t ̂
NRMSE =
x T
∑t=1 (xt −xt ̂)2 ,MAPE = T
∑ xt
× 100%
t=1
where T is the number of observations in the out-of-sample subset.
The performance to forecast direction of movement can be mea-
sured by a directional symmetry (DS) as follows:
T decomposition results of solar daily radiation data in Beijing. All IMF
1 1 if (x t −x t − 1)(x t ̂−x t − 1) > 0
DS = ∑ dt × 100%, where dt = ⎧
T t=2
⎨
⎩ 0 otherwise
Furthermore, in order to further evaluate the level forecasting per-
formance and the directional forecasting performance from the statis-
tical view, the Diebold-Mariano (DM) statistic and Pesaran-
Timmermann (PM) statistic are adopted to test the statistical sig-
nificance of all models. The main process of the DM test and the PM test
can be found in (Diebold and Mariano, 2002; Pesaran and
Timmermann, 1992; Sun et al., 2017).
4.2. Empirical results
For comparison purposes, some other traditional forecasting
methods are regarded as benchmark models to be compared with the
proposed DCE learning approach in terms of forecasting performance.
According to previous literature review, the most popular univariate
forecasting models are introduced for solar radiation series, i.e., the
typical ARIMA model and the most popular AI techniques of ANN and
SVR. In particular, these three tools of ARIMA, ANN and SVR might be
the most popular time series forecasting models, which have widely
been used as single forecasting models and component forecasting
model for decomposition ensemble learning approach. Therefore, these
three typical univariate forecasting tools are introduced as single
benchmark models in this study. Furthermore, we also introduce LSSVR
as the single benchmark model since LSSVR is selected as the compo-
nent forecasting tool for extracted components in formulating the
proposed DCE learning approach. In the decomposition stage, EMD, the
original form of EEMD, is regarded as benchmark decomposition
method.
Solar Energy 163 (2018) 189–199
Therefore, in order to verify the forecasting performance of the
proposed DCE learning approach, four single forecasting models, i.e.,
single ARIMA, ANN, SVR and LSSVR models and five hybrid ensemble
learning approaches, i.e., EMD-LSSVR-ADD, EMD-LSSVR-LSSVR, EMD-
LSSVR-K-LSSVR, EEMD-LSSVR-ADD and EEMD-LSSVR-LSSVR are con-
(4)
sidered as benchmark models.
For consistency, all LSSVR models, as both component forecasting
and nonlinear ensemble forecasting, are the same as LSSVR of single
benchmark model in the determination method, as mentioned above.
Generally speaking, the parameter specification is crucial for model
performance. For single SVR and LSSVR models, the Gaussian RBF
kernel function is selected and GSA algorithm is used to determine
values of optimal parameters γ , σ 2 and ε in terms of the smallest error in
the in-sample subset (Niu et al., 2016b; Rashedi et al., 2009). For single
ANN forecasting technique, input neurons is set to 6 chosen by auto-
correlation function (ACF) and partial correlation function (PCF),
hidden neuron is set to 7 and output neuron is set to 1 (Niu et al.,
2016b). ANN models are iteratively run 10,000 times to train the model
(5) (Yu et al., 2009). In ARIMA (p-d-q) model, the optimal form is esti-
mated by Schwarz Criterion (SC) minimization (Yu et al., 2009).
Six decomposition ensemble learning approaches are performed.
For EEMD decomposition, the ensemble member is set to 100 and the
standard deviation of added white noise is set to 0.2. Fig. 4 shows the
components are listed from the highest frequency to the lowest fre-
(6)
quency and the last one is the residual term.
Therefore, four steps are involved in empirical results analysis: (1)
four single models are performed to forecast solar radiation in Beijing to
select the best single model; (2) the forecasting performance of the
proposed EEMD-LSSVR-K-LSSVR approach is compared with other en-
semble benchmark approaches to ensure that the proposed DCE ap-
proach can be statistically superior to all other benchmark approaches;
(3) the robustness of the DCE approach is analyzed; (4) some conclu-
sions are summarized from the empirical analysis.
4.2.1. Performance comparison of single models
The performances of four single forecasting models are analyzed by
means of the NRMSE, MAPE and DS evaluation criteria as shown in
Figs. 5–7. Tables 1–4 demonstrate the results of DM test and PT test
with respect to different forecasting horizons.
From Figs. 5–7, it can be concluded that: (1) the performance of
LSSVR is the best single benchmark model, followed by other AI tech-
niques, and ARIMA ranks the last, by means of forecasting accuracy and
statistical test; (2) it may be difficult for the ARIMA, as a traditional
linear model, to capture the complexity and nonlinearity of solar ra-
diation data, while AI techniques are more appropriate for exploring
the nonlinear patterns; (3) the forecasting performances of SVR and
ANN are quite similar in different forecasting horizons.
The DM test and PT test are employed to test the difference of four
single benchmark models in terms of level accuracy and directional
accuracy, and the results are shown in Tables 1–4. According to the
above empirical results, it can be obtained that (1) all the single models
are almost ineffective in solar radiation forecasting; (2) the level
193
S. Sun et al. Solar Energy 163 (2018) 189–199

Fig. 4. The IMFs and a residual for Beijing solar radiation data decomposed by EEMD.

LSSVR SVR ANN ARIMA LSSVR SVR ANN ARIMA

19.25 19.61
25 20
22.36 18.03
20.09
20 19.18
15 14.35 14.23
13.63
12.89
14.88 15.19 15.07 11.76
NRMSE (%)

14.03 10.54 11.17

MAPE (%)

15 13.25 13.15 10.83

12.34 12.38 9.69
11.26 10
10

5
5

0 0
One-step-ahead forecasting Three-step-ahead forecasting Six-step-ahead forecasting One-step-ahead forecasting Three-step-ahead forecasting Six-step-ahead forecasting

Fig. 5. Performance comparison of different single models in terms of NRMSE criteria. Fig. 6. Performance comparison of different single models in terms of MAPE criteria.

194
S. Sun et al. Solar Energy 163 (2018) 189–199

LSSVR SVR ANN ARIMA EEMD-LSSVR-K-LSSVR EMD-LSSVR-K-LSSVR EEMD-LSSVR-LSSVR

70 EMD-LSSVR-LSSVR EEMD-LSSVR-ADD EMD-LSSVR-ADD
59.51 59.09 58.14 10
60 56.91 56.49 54.86 55.68 54.99 54.04 8.96
8.14 8.46
7.65 7.74
50 45.83 44.32 8 7.02 7.09
43.78
6.54 6.24 6.41
40 5.89 5.73 5.86

NRMSE (%)
DS (%)

6 5.14 4.87
4.53
30
4 3.15
2.96
20

10 2

0 0
One-step-ahead forecasting Three-step-ahead forecasting Six-step-ahead forecasting One-step-ahead forecasting Three-step-ahead forecasting Six-step-ahead forecasting

Fig. 7. Performance comparison of different single models in terms of DS criteria. Fig. 8. Performance comparison of different hybrid ensemble approaches in terms of
NRMSE criteria.

Table 1
DM test results for single models in one-step-ahead forecasting. EEMD-LSSVR-K-LSSVR EMD-LSSVR-K-LSSVR EEMD-LSSVR-LSSVR
EMD-LSSVR-LSSVR EEMD-LSSVR-ADD EMD-LSSVR-ADD
Tested model Benchmark model 10

SVR ANN ARIMA

8 7.54 7.58 7.86
6.91 6.89 6.76
6.07 5.89 6.06 6.23
LSSVR −0.4121 (0.3401) −0.5129 (0.3040) −3.3627 (0.0004)

MAPE (%)
6 5.11 5.26 5.04
SVR −0.1182 (0.4530) −3.5628 (0.0001) 4.68
ANN −3.7143 (0.0001) 4.01 4.12
4 3.06
2.83

2
Table 2
DM test results for single models in three-step-ahead forecasting. 0
One-step-ahead forecasting Three-step-ahead forecasting Six-step-ahead forecasting
Tested model Benchmark model
Fig. 9. Performance comparison of different hybrid ensemble approaches in terms of
SVR ANN ARIMA MAPE criteria.

LSSVR −0.4017 (0.3440) −0.3892 (0.3486) −4.1026 (0.0000)

EEMD-LSSVR-K-LSSVR EMD-LSSVR-K-LSSVR EEMD-LSSVR-LSSVR
SVR −0.2053 (0.4187) −3.8968 (0.0000)
ANN −4.1236 (0.0000) EMD-LSSVR-LSSVR EEMD-LSSVR-ADD EMD-LSSVR-ADD
100
88.24
80.31 81.94 82.08 80.03
80 75.51 72.23 75.11
Table 3 68.39 69.91 71.55
65.12 64.52
DM test results for single models in six-step-ahead forecasting. 64.71 60.05
62.11 58.82
60 55.81
DS (%)

Tested model Benchmark model

40
SVR ANN ARIMA
20
LSSVR −0.4173 (0.3382) −0.3156 (0.3762) −4.6027 (0.0000)
SVR −0.3319 (0.3700) −5.5361 (0.0000) 0
ANN −4.9586 (0.0000) One-step-ahead forecasting Three-step-ahead forecasting Six-step-ahead forecasting

Fig. 10. Performance comparison of different hybrid ensemble approaches in terms of DS

criteria.
Table 4
PT test results for single models.
4.2.2. Performance comparison of hybrid ensemble learning approaches
Horizon Value Single model The forecasting performances of six hybrid ensemble approaches are
discussed in this section. Figs. 8–10 show the comparison results of
ARIMA ANN SVR LSSVR
NRMSE, MAPE and DS evaluation criteria. The results of DM test and
One-step-ahead forecasting SPT 0.3126 −0.8926 −0.5923 1.4018 PT test with different forecasting horizons are reported in Tables 6–9.
p-Value 0.7546 0.3721 0.5536 0.1610 The cluster numbers of the proposed DCE learning approach are set to 3
Three-step-ahead SPT 0.3125 −1.6017 −0.9137 1.9896 which can be interpreted as the rainy, cloudy and fine day respectively
forecasting p-Value 0.7547 0.1092 0.3609 0.0466
Six-step-ahead forecasting SPT −0.4027 −1.5129 −0.6674 1.7123
(Benmouiza and Cheknane, 2013). Table 5 describes the cluster centers
p-Value 0.6872 0.1303 0.5045 0.0868 and member numbers of component forecasting results by using
Kmeans method.
According to above comparison results, it can be concluded that the
accuracy and directional accuracy of LSSVR forecasting model are the proposed DCE learning approach significantly outperforms all other
optimal among all the single models. Hence, the LSSVR is considered as hybrid ensemble approaches in both level accuracy and directional
the individual forecasting model and ensemble forecasting method in accuracy for solar radiation forecasting. The results of DM test indicate
decomposition ensemble learning approach and the proposed DCE that the proposed DCE learning approach statistically confirms its su-
learning approach. periority at the 95% confidence level. Furthermore, the results of PT
test reveal that the DCE learning approach is more powerful than all

195
S. Sun et al. Solar Energy 163 (2018) 189–199

Table 5
Cluster centers and member numbers of EEMD-LSSVR-K-LSSVR ensemble approach.

Horizon Cluster Cluster center Number

One-step-ahead forecasting
Three-step-ahead forecasting
Six-step-ahead forecasting
1 (0.228, −0.180, −0.178, −0.194, −0.173, −0.001, −0.815, −0.122, 0.037, −0.010, 8.445) 715
2 (0.010, 0.137, 0.028, −0.001, 0.052, −0.394, −0.348, −0.016, 0.061, 0.010, 8.609) 989
3 (−0.087, −0.260, 0.206, −0.074, −0.290, −0.369, 0.478, 0.086, −0.008, −0.010, 8.701) 662
1 (−0.147, −0.0.088, 0.028, 0.146, −0.204, 0.144, −0.465, −0.093, 0.034, 0.008, 8.446) 726
2 (−0.449, 0.209, 0.217, 0.001, −0.006, 0.217, −1.050, 0.006, 0.041, 0.009, 8.560) 982
3 (0.535, −0.208, 0.317, −0.004, −0.317, −0.350, 0.494, 0.086, −0.008, −0.010, 8.700) 656
1 (0.096, 0.002, −0.078, −0.045, −0.167, −0.366, 0.272, −0.026, 0.029, −0.008, 8.450) 695
2 (0.385, 0.038, −0.226, −0.091, 0.064, −0.050, 0.097, −0.017, 0.067, 0.010, 8.612) 987
3 (−0.118, −0.058, 0.306, 0.048, −0.402, −0.218, 0.596, 0.087, −0.011, −0.009, 8.703) 679

Table 6
DM test results for hybrid ensemble approaches in one-step-ahead forecasting.

Tested model Benchmark model

EMD-LSSVR-K-LSSVR EEMD-LSSVR-LSSVR EMD-LSSVR-LSSVR EEMD-LSSVR-ADD EMD-LSSVR-ADD

EEMD-LSSVR-K-LSSVR −1.9124 (0.0279) −2.0125 (0.0221) −2.0964 (0.0180) −2.9974 (0.0014) −3.8961 (0.0000)
EMD-LSSVR-K-LSSVR −1.7029 (0.0443) −1.2385 (0.1078) −1.6951 (0.0450) −3.5264 (0.0002)
EEMD-LSSVR-LSSVR −1.3346 (0.0910) −1.0263 (0.1524) −1.5788 (0.0572)
EMD-LSSVR-LSSVR −1.5968 (0.0552) −2.9816 (0.0014)
EEMD-LSSVR-ADD 0.8463 (0.1987)

Table 7
DM test results for hybrid ensemble approaches in three-step-ahead forecasting.

Tested model Benchmark model

EMD-LSSVR-K-LSSVR EEMD-LSSVR-LSSVR EMD-LSSVR-LSSVR EEMD-LSSVR-ADD EMD-LSSVR-ADD

EEMD-LSSVR-K-LSSVR −2.2354 (0.0127) −2.9025 (0.0019) −3.4458 (0.0002) −4.5543 (0.0000) −4.0056 (0.0000)
EMD-LSSVR-K-LSSVR −1.8974 (0.0289) −2.5961 (0.0047) −3.2594 (0.0005) −4.2674 (0.0000)
EEMD-LSSVR-LSSVR 0.7842 (0.2165) −1.0268 (0.1523) −1.3642 (0.0863)
EMD-LSSVR-LSSVR −1.1269 (0.1299) −1.6874 (0.0458)
EEMD-LSSVR-ADD 0.4419 (0.3293)

Table 8
DM test results for hybrid ensemble approaches in six-step-ahead forecasting.

Tested model Benchmark model

EMD-LSSVR-K-LSSVR EEMD-LSSVR-LSSVR EMD-LSSVR-LSSVR EEMD-LSSVR-ADD EMD-LSSVR-ADD

EEMD-LSSVR-K-LSSVR −2.9156 (0.0018) −2.6671 (0.0038) −3.1146 (0.0009) −6.1573 (0.0000) −6.2105 (0.0000)
EMD-LSSVR-K-LSSVR −1.7849 (0.0371) −2.5963 (0.0047) −4.5267 (0.0000) −4.5249 (0.0000)
EEMD-LSSVR-LSSVR 1.0284 (0.1519) −1.6157 (0.0531) −6.4892 (0.0000)
EMD-LSSVR-LSSVR −1.8816 (0.0299) −3.1257 (0.0008)
EEMD-LSSVR-ADD −0.3674 (0.3567)

Table 9
PT test results for hybrid ensemble approaches.

Forecasting horizons Value Hybrid ensemble approach

EMD-LSSVR EEMD-LSSVR

ADD LSSVR K-LSSVR ADD LSSVR K-LSSVR

One-step-ahead forecasting SPT 1.6587 2.2206 3.4106 1.9687 2.5861 4.0129

p-Value 0.0972 0.0264 0.0006 0.0490 0.0097 0.0001
Three-step-ahead forecasting SPT 1.5543 2.0017 3.0146 1.9217 2.6349 3.8874
p-Value 0.1201 0.0453 0.0026 0.0546 0.0084 0.0001
Six-step-ahead forecasting SPT 1.5021 1.8954 2.8914 1.7416 2.4417 3.6253
p-Value 0.1331 0.0580 0.0038 0.0816 0.0146 0.0002

196
S. Sun et al. Solar Energy 163 (2018) 189–199

other hybrid ensemble approaches in terms of directional accuracy. 4.2.4. Discussion

From Figs. 8–10, certain interesting findings can be summarized: (1) To evaluate the forecasting performance of our proposed DCE ap-
the proposed EEMD-LSSVR-K-LSSVR outperforms all other benchmark proach against other multiscale hybrid models we applied to this pro-
models in different forecasting horizons, which implies that the EEMD- blem the models proposed in Monjoly et al. (2017). In that paper the
LSSVR-K-LSSVR is a powerful learning approach for solar radiation authors proposed an EMD-hybrid model, an EEMD-hybrid model and a
forecasting in both level accuracy and directional accuracy. In contrast, WD-hybrid model to forecast hourly solar radiation. Table 12 lists the
the forecasting performances of EEMD-LSSVR-ADD and EMD-LSSVR- results obtained when we applied those models to our problem. The
ADD are poor. The main reason may be that the addition ensemble is results show that the proposed DCE approach obtained better results
very difficult to identify the characteristic of component forecasting than the multiscale hybrid models in Monjoly et al. (2017). The main
results; (2) it clearly shows that the hybrid ensemble approach with reason why DCE approach performs better is that the clustering
EEMD is much better than the one with EMD by means of level accuracy leverage improves the final forecasting performance of the decom-
and directional accuracy, which reveals that EEMD is a more effective position ensemble learning approach.
decomposition algorithm; (3) according to Kmeans clustering method, Meanwhile, according to above comparisons and statistical test re-
it can be found that the hybrid ensemble approach with clustering sults, the following six main conclusions can be summarized:
outperform the one without clustering in both level accuracy and di-
rectional accuracy, which indicates that the clustering scheme is a (1) The proposed DCE learning approach significantly outperforms all
simple and effective technique for hybrid ensemble learning approach; other benchmark models by means of level accuracy and directional
(4) the forecasting performance of hybrid ensemble approach is sig- accuracy, which indicates that the clustering strategy can effec-
nificantly better than single model. The possible reason is that the de- tively improve the performance of hybrid ensemble learning ap-
composition can dramatically improve the forecasting performance of proach.
single models. (2) The performance of LSSVR is the best in comparison with other
Similarly, the DM test and PT test are performed to statistically single models.
evaluate the difference of hybrid ensemble approaches from both level (3) The hybrid ensemble learning approaches perform better than the
accuracy and directional accuracy. The comparison results are shown in single models in terms of different forecasting horizons, which
Tables 6–9, which demonstrates the same conclusions as above. imply that the scheme of decomposition-ensemble can effectively
Therefore, the empirical results show that the proposed DCE improve the forecasting performance of solar radiation.
learning approach significantly improves the performance of decom- (4) The proposed EEMD-LSSVR-K-LSSVR learning approach is a stable
position ensemble learning approach, which also improves the fore- and effective forecasting framework with respect to robustness
casting performance of single models. analysis.
(5) Due to the intrinsic complexity of solar radiation data, nonlinear AI
techniques with powerful self-learning ability are much more ap-
4.2.3. Robustness analysis propriate for solar radiation forecasting than linear ARIMA model.
The robustness of single models and hybrid ensemble learning ap- (6) The proposed DCE learning approach is a promising framework for
proaches are evaluated in this section. The ARIMA, ANN, SVR and forecasting solar radiation in terms of level accuracy, directional
LSSVR models produce different forecasting results with different initial accuracy and robustness.
settings. Therefore, we run all single models and hybrid ensemble
learning approaches for twenty times, and analyze their robustness in 5. Conclusions
terms of standard deviation of NRMSE, MAPE and DS. The results are
shown in Tables 10 and 11, and some interesting findings can be This paper proposes a decomposition-clustering-ensemble (DCE)
summarized: (1) it is obvious that LSSVR is the most stable model in all learning approach for solar radiation forecasting. The DCE learning
single models, since its standard deviations of NRMSE, MAPE and DS approach employs EEMD for decomposition, Kmeans for clustering
are far less than other single models; (2) EEMD-LSSVR-K-LSSVR is the analysis and LSSVR for ensemble learning. For further verification, the
most robust hybrid ensemble learning approach, and its standard de- proposed EEMD-LSSVR-K-LSSVR learning approach is applied to fore-
viations of NRMSE, MAPE and DS are far less than 0.01; (3) ADD-based cast solar radiation. The empirical results show that the EEMD-LSSVR-
ensemble forecasting are most unstable among all the hybrid ensemble K-LSSVR learning approach can significantly improve the forecasting
learning approaches. performance and outperform some other popular forecasting methods
in terms of forecasting accuracy and robustness analysis, which in-
dicates that the EEMD-LSSVR-K-LSSVR learning approach is a relatively
promising approach for solar radiation forecasting problems.
Table 10
Besides solar radiation forecasting, the proposed CNE learning ap-
Robustness analysis for single models.
proach can be applied to solve other complex and difficult forecasting
Horizon Std. Single model problems, including wind speed forecasting, container throughput
forecasting, crude oil price forecasting, traffic flow forecasting, etc.
ARIMA ANN SVR LSSVR Furthermore, this paper only focuses on univariate time series
One-step-ahead forecasting Std. of 0.5123 0.2035 0.1147 0.0876 analysis, while other factors affecting solar radiation are not taken into
NRMSE consideration. If those factors can be introduced into the proposed DCE
Std. of MAPE 0.1022 0.0875 0.0625 0.0421 learning approach, the forecasting performance may be better. In ad-
Std. of DS 0.6756 0.2074 0.1169 0.0876 dition, the kernel function of LSSVR is selected only by means of the
Three-step-ahead Std. of 0.5374 0.2194 0.1873 0.0967
forecasting NRMSE
experience and the selected kernel function may not be optimal for the
Std. of MAPE 0.1124 0.0954 0.0714 0.0461 specific problem. Hence, the optimal selection of kernel function for a
Std. of DS 0.6811 0.0975 0.1008 0.0926 specific problem is left for a further study.
Six-step-ahead forecasting Std. of 0.5486 0.2367 0.1982 0.0923
NRMSE
Acknowledgement
Std. of MAPE 0.1326 0.1059 0.0829 0.0562
Std. of DS 0.6938 0.1041 0.1073 0.0964
This research was supported by the National Natural Science
The bold values denotes the best performance of robustness analysis. Foundation of China (Project No: 71373262). The Authors would like to

197
S. Sun et al. Solar Energy 163 (2018) 189–199

Table 11
Robustness analysis for hybrid ensemble approaches.

Horizon Std. Ensemble approach

EMD-LSSVR EEMD-LSSVR

ADD LSSVR K-LSSVR ADD LSSVR K-LSSVR

One-step-ahead forecasting Std. of NRMSE 0.2049 0.0509 0.0512 0.3671 0.1327 0.0256
Std. of MAPE 0.0106 0.0076 0.0029 0.0087 0.0095 0.0032
Std. of DS 0.1956 0.0791 0.0113 0.1652 0.0592 0.0098
Three-step-ahead forecasting Std. of NRMSE 0.2149 0.0583 0.0664 0.3792 0.1428 0.0329
Std. of MAPE 0.0153 0.0107 0.0059 0.0114 0.0137 0.0065
Std. of DS 0.2011 0.0831 0.0139 0.1798 0.0625 0.0119
Six-step-ahead forecasting Std. of NRMSE 0.2218 0.0623 0.0704 0.3884 0.1568 0.0395
Std. of MAPE 0.0163 0.0125 0.0071 0.0182 0.0152 0.0085
Std. of DS 0.2217 0.0892 0.0149 0.1859 0.0782 0.0128

The bold values denotes the best performance of robustness analysis.

Table 12
Forecasting performance using the hybrid model proposed in Monjoly et al. (2017).

Evaluation criteria Models One-step-ahead forecasting Three-step-ahead forecasting Six-step-ahead forecasting

NRMSE (%) EMD-hybrid model 3.27 3.45 4.65

EEMD-hybrid model 4.38 4.13 5.28
WD-hybrid model 4.76 5.08 5.92
MAPE (%) EMD-hybrid model 3.09 3.19 4.23
EEMD-hybrid model 3.56 3.74 5.15
WD-hybrid model 4.42 4.38 5.69
DS (%) EMD-hybrid model 86.46 87.55 78.66
EEMD-hybrid model 82.76 80.57 77.98
WD-hybrid model 79.07 79.34 73.87

express their sincere appreciation to Jan Kleissl (the subject editor), method, LSSVR as component forecasting method and LSSVR as
Hugo Pedro (the associate editor), and two anonymous referees in ensemble method
making valuable comments and suggestions to this paper. Their com- EEMD-LSSVR-K-LSSVR – the model using EEMD as decomposition
ments and suggestions have improved the quality of the paper im- method, LSSVR as component forecasting method, Kmeans as clus-
mensely. tering method and LSSVR as ensemble method
ELM – extreme learning machine
Conflict of interest EMD – empirical mode decomposition
EMD-LSSVR-ADD – the model using EMD as decomposition method,
The authors declare that there is no conflict of interests regarding LSSVR as component forecasting method and ADD as ensemble
the publication of this paper. method
EMD-LSSVR-LSSVR – the model using EMD as decomposition
Appendix A. List of abbreviations method, LSSVR as component forecasting method and LSSVR as
ensemble method
Here, all terms mentioned in this paper and their definitions are EMD-LSSVR-K-LSSVR – the model using EMD as decomposition
listed in alphabetical order: method, LSSVR as component forecasting method, Kmeans as clus-
tering method and LSSVR as ensemble method
ACF – autocorrelation function GBR – gradient boosted regression
AI – artificial intelligence GHI – global horizontal solar irradiance
ANN – artificial neural network GP – Gaussian process
AR – autoregressive model GSA – gravitational search algorithm
ARMA – autoregressive moving average IMFs – intrinsic mode functions
ARIMA – autoregressive integrated moving average LSSVR – least square support vector regression
BPNN – back propagation neural network MAPE – mean absolute percentage error
CS – cuckoo search algorithm MLPNN – multi-layer perceptron neural network
DCE – decomposition-clustering-ensemble learning approach MRSR – multi-response sparse regression
DHR – dynamic harmonic regression model NRMSE – normalized root mean squared error
DM – Diebold-Mariano test OP-ELM – optimally pruned extreme learning machine
DS – directional symmetry PCF – partial correlation function
EEMD – ensemble empirical mode decomposition PM – Pesaran-Timmermann test
EEMD-LSSVR-ADD – the model using EMD as decomposition RFR – random forest regression
method, LSSVR as component forecasting method and ADD as en- SC – Schwarz criterion
semble method SVM – support vector machines
EEMD-LSSVR-LSSVR – the model using EMD as decomposition SVR – support vector regression

198
S. Sun et al.
References
Akarslan, E., Hocaoğlu, F.O., Edizkan, R., 2014. A novel MD (multi-dimensional) linear
prediction filter approach for hourly solar radiation forecasting. Energy 73, 978–986.
Amrouche, B., Le Pivert, X., 2014. Artificial neural network based daily local forecasting
for global solar radiation. Appl. Energ. 130, 333–341.
Benmouiza, K., Cheknane, A., 2013. Forecasting hourly global solar radiation using hy-
brid k-means and nonlinear autoregressive neural network models. Energ. Convers.
Manage. 75, 561–569.
Chen, S.X., Gooi, H.B., Wang, M.Q., 2013. Solar radiation forecast based on fuzzy logic
and neural networks. Renew. Energ. 60, 195–201.
Chow, C.W., Urquhart, B., Lave, M., Dominguez, A., Kleissl, J., Shields, J., Washom, B.,
2011. Intra-hour forecasting with a total sky imager at the UC San Diego solar energy
testbed. Sol Energy 85 (11), 2881–2893.
Coimbra, C.F., Kleissl, J., Marquez, R., 2013. Overview of solar-forecasting methods and a
metric for accuracy evaluation. Solar Resource Assess. Forecast.
Diebold, F.X., Mariano, R.S., 2002. Comparing predictive accuracy. J. Bus. Econ. Stat. 20
(1), 134–144.
Fidan, M., Hocaoğlu, F.O., Gerek, Ö.N., 2014. Harmonic analysis based hourly solar ra-
diation forecasting model. IET. Renew. Power. Gen. 9 (3), 218–227.
Gala, Y., Fernández, Á., Díaz, J., Dorronsoro, J.R., 2016. Hybrid machine learning fore-
casting of solar radiation values. Neurocomputing 176, 48–59.
He, K., Yu, L., Lai, K.K., 2012. Crude oil price analysis and forecasting using wavelet
decomposed ensemble model. Energy 46 (1), 564–574.
Huang, J., Korolkiewicz, M., Agrawal, M., Boland, J., 2013. Forecasting solar radiation on
an hourly time scale using a coupled autoregressive and dynamical system (CARDS)
model. Sol. Energy 87, 136–149.
Huang, N.E., Shen, Z., Long, S.R., Wu, M.C., Shih, H.H., Zheng, Q., Yen, N., Tung, C.C.,
Liu, H.H., 1998. The empirical mode decomposition and the Hilbert spectrum for
nonlinear and non-stationary time series analysis. Proc. Roy. Soc. Lond. A: Math.,
Phys. Eng. Sci. 454, 903–995.
Jiang, H., Dong, Y., Wang, J., Li, Y., 2015. Intelligent optimization models based on hard-
ridge penalty and RBF for forecasting global solar radiation. Energ. Convers. Manage.
95, 42–58.
Lauret, P., Voyant, C., Soubdhan, T., David, M., Poggi, P., 2015. A benchmarking of
machine learning techniques for solar radiation forecasting in an insular context. Sol.
Energy 112, 446–457.
Lewis, C.D., 1982. International and Business Forecasting Methods. Butter-Worths,
London.
MacQueen, J., 1967. Some methods for classification and analysis of multivariate ob-
servations. Proc. Fifth Berkeley Sympos. Math. Stat. Probab. 1 (14), 281–297.
Mathiesen, P., Collier, C., Kleissl, J., 2013. A high-resolution, cloud-assimilating numer-
ical weather prediction model for solar irradiance forecasting. Sol Energy 92, 47–61.
Mathiesen, P., Kleissl, J., 2011. Evaluation of numerical weather prediction for intra-day
solar forecasting in the continental United States. Sol Energy 85 (5), 967–977.
Mellit, A., Benghanem, M., Kalogirou, S.A., 2006. An adaptive wavelet-network model for
forecasting daily total solar-radiation. Appl. Energ. 83 (7), 705–722.
Monjoly, S., André, M., Calif, R., Soubdhan, T., 2017. Hourly forecasting of global solar
radiation based on multiscale decomposition methods: a hybrid approach. Energy
119, 288–298.
Niu, M.F., Gan, K., Sun, S.L., Li, F.Y., 2017. Application of decomposition-ensemble
learning paradigm with phase space reconstruction for day-ahead PM2.5 concentra-
tion forecasting. J. Environ. Manage. 196, 110–118.
Niu, M.F., Sun, S.L., Wu, J., Yu, L., Wang, J.Z., 2016b. An innovative integrated model
199
Solar Energy 163 (2018) 189–199
using the singular spectrum analysis and nonlinear multi-layer perceptron network
optimized by hybrid intelligent algorithm for short-term load forecasting. Appl.
Math. Model. 40 (5), 4079–4093.
Niu, M.F., Sun, S.L., Wu, J., Zhang, Y.L., 2015. Short-term wind speed hybrid forecasting
model based on bias correcting study and its application. Math. Probl. Eng. 2015.
http://dx.doi.org/10.1155/2015/351354.
Niu, M.F., Wang, Y.F., Sun, S.L., Li, Y.W., 2016a. A novel hybrid decomposition and
ensemble model based on CEEMD and GWO for short-term PM2.5 concentration
forecasting. Atmos. Environ. 134, 168–180.
Paoli, C., Voyant, C., Muselli, M., Nivet, M., 2010. Forecasting of preprocessed daily solar
radiation time series using neural networks. Sol. Energ. 84 (12), 2146–2160.
Pesaran, M.H., Timmermann, A., 1992. A simple nonparametric test of predictive per-
formance. J. Bus. Econ. Stat. 10 (4), 461–465.
Rashedi, E., Nezamabadi-Pour, H., Saryazdi, S., 2009. GSA: a gravitational search algo-
rithm. Inform. Sci. 179 (13), 2232–2248.
Sun, S.L., Qiao, H., Wei, Y., Wang, S.Y., 2017. A new dynamic integrated approach for
wind speed forecasting. Appl. Energ. 197, 151–162.
Suykens, J.A., Vandewalle, J., 1999. Least squares support vector machine classifiers.
Neural Process. Lett. 9 (3), 293–300.
Tang, L., Yu, L., Wang, S., Li, J., Wang, S.Y., 2012. A novel hybrid ensemble learning
paradigm for nuclear energy consumption forecasting. Appl. Energ. 93, 432–443.
Trapero, J.R., Kourentzes, N., Martin, A., 2015. Short-term solar irradiation forecasting
based on dynamic harmonic regression. Energy 84, 289–295.
Vapnik, V., 2013. The Nature of Statistical Learning Theory. Springer science & business
media.
Voyant, C., Paoli, C., Muselli, M., Nivet, M., 2013. Multi-horizon solar radiation fore-
casting for Mediterranean locations using time series models. Renew. Sust. Energ.
Rev. 28, 44–52.
Wang, J., Jiang, H., Wu, Y., Dong, Y., 2015. Forecasting solar radiation using an opti-
mized hybrid model by Cuckoo Search algorithm. Energy 81, 627–644.
Wang, S., Yu, L., Tang, L., Wang, S.Y., 2011. A novel seasonal decomposition based least
squares support vector regression ensemble learning approach for hydropower con-
sumption forecasting in China. Energy 36 (11), 6542–6554.
Wang, S.Y., Yu, L., Lai, K.K., 2005. Crude oil price forecasting with TEI@I methodology.
J. Syst. Sci. Complex. 18 (2), 145–166.
Wang, S.Y., 2004. TEI@I: A New Methodology for Studying Complex Systems. The
International Workshop on Complexity Science, Tsukuba, Japan.
Wu, Z., Huang, N.E., 2009. Ensemble empirical mode decomposition: a noise-assisted
data analysis method. Adv. Adapt. Data Anal. 1 (01), 1–41.
Yu, L., Dai, W., Tang, L., 2016. A novel decomposition ensemble model with extended
extreme learning machine for crude oil price forecasting. Eng. Appl. Artif. Intel. 47,
110–121.
Yu, L., Wang, S.Y., Lai, K.K., 2008. Forecasting crude oil price with an EMD-based neural
network ensemble learning paradigm. Energ. Econ. 30 (5), 2623–2635.
Yu, L., Wang, S.Y., Lai, K.K., 2009. A neural-network-based nonlinear metamodeling
approach to financial time series forecasting. Appl. Soft Comput. 9 (2), 563–574.
Yu, L., Wang, Z., Tang, L., 2015. A decomposition–ensemble model with data-char-
acteristic-driven reconstruction for crude oil price forecasting. Appl. Energ. 156,
251–267.
Yu, L., Zhao, Y., Tang, L., 2014. A compressed sensing based AI learning paradigm for
crude oil price forecasting. Energ. Econ. 46, 236–245.
Zhang, X., Lai, K.K., Wang, S.Y., 2008. A new approach for crude oil price analysis based
on empirical mode decomposition. Energ. Econ. 30 (3), 905–918.