Electric Power Systems Research 148 (2017) 3547

Contents lists available at ScienceDirect

Electric Power Systems Research

journal homepage: www.elsevier.com/locate/epsr

Online electricity demand forecasting based on an effective forecast

combination methodology
Abderrezak Laoua a, , Mourad Mordjaoui a , Salim Haddad b , Taqiy Eddine Boukelia c ,
Abderahmane Ganouche a
a
Department of Electrical Engineering, University of 20 August 1955-Skikda, Skikda, Algeria
b
Department of Mechanical Engineering, University of 20 August 1955-Skikda, Skikda, Algeria
c
Department of Mechanical Engineering, University of Constantine 1, Constantine, Algeria

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents a new forecast combination methodology for generating very short-term electricity
Received 8 May 2016 demand predictions under both normal and anomalous load conditions. The main contribution of the
Received in revised form 17 February 2017 work is to propose an online load forecasting system that has the ability to achieve good forecasting
Accepted 19 March 2017
accuracy, avoid large forecasting errors and ensure low computation time. The real-time load data from
the French power system and the Australian dataset for the state of New South Wales are used as an
Keywords:
illustrative example to evaluate the performance of the proposed methodology. The results reect that
Online load forecasting
the developed approach has better forecasting performance than other methods considered in this study.
Forecast combination
Hampel lter
For example, the results from the public holidays in France showed an average mean absolute percentage
Anomalous load error (MAPE) of 0.863%, and the accuracy improvements over a simple average combination method, the
best individual method, and a weighted combination are 15.887%, 13.353%, and 3.034%, respectively. For
the case of the Australian load dataset, our forecasting system achieved an average MAPE of 0.860% and
the improvement in comparison to a benchmark algorithm from the literature is equal to 8.316%.
2017 Elsevier B.V. All rights reserved.

1. Introduction of renewable energy sources intermittence on grid stability and

Electricity demand forecasting is the primary prerequisite for
achieving the goal of sustainable energy management and eco-
nomic and secure operation of modern power systems. The task
of knowing the electricity demand in advance is needed to sustain
supply/demand balance and to manage the process of electricity
production, distribution and consumption on a variety of temporal
scales: very short-term from few minutes to an hour, short-
term from an hour to one week [15], medium-term from one
week to one year [69], and long-term from one year to decades
[1012]. In particular, the accurate very short-term load fore-
casts (VSTLF) are needed for the real-time scheduling of electricity
generation as well as load-frequency control and economic dis-
patch functions. With the deregulation of electricity markets and
the growing penetration of renewable energy sources into the
energy matrix of todays power networks, such predictions are
also of importance to market participants to mitigate the effects
Corresponding author.
E-mail address: a.laoua@univ-skikda.dz (A. Laoua).
reliability.
Over the past few years, researchers have studied this prob-
lem of improving VSTLF accuracy and various classical and articial
intelligence techniques have been considered using different sim-
ulated datasets. Classical methods include BoxJenkins time series
models [13,14], exponential smoothing [15,16], and Kalman l-
tering [17,18]. Articial intelligence techniques include neural
network [1921], fuzzy logic [13], adaptive neuro-fuzzy inference
system [22,23], and support vector regression [24]. Hybrid methods
can be also used to combine the advantages of several algorithms.
A typical way to create a hybrid model is to use the empirical mode
decomposition as a tool for decomposing the electric load time
series into a limited number of intrinsic mode function compo-
nents, and to forecast the decomposed sequences with appropriate
computational intelligence models. However, the application of
empirical mode decomposition requires a signicant execution
time and the forecaster may face difculty to handle the mode-
mixing problem, since the components of new time series may
signicantly differ from those used while training the forecasting
models [25]. In the other hand, wavelet decomposition is an ef-
cient technique to unfold the inner features of the electric load
time series. In this manner, many recent papers discussed the

http://dx.doi.org/10.1016/j.epsr.2017.03.016
0378-7796/ 2017 Elsevier B.V. All rights reserved.
36 A. Laoua et al. / Electric Power Systems Research 148 (2017) 3547
use of wavelet transform to decompose the data into constitu-
tive series that can be predicted more accurately than the original
signal [26].
However, it is not an easy task to perform the accurate pre-
dictions of electricity consumption, since the electricity demand
behavior is complex and the load time series is non-stationary in
mean and variance, containing long run trend, multiple cycles of
seasonal uctuations as well as random noise [27,28]. In fact, there
is not any forecasting method which gives the best result for all
times: some approaches may perform very well under certain cir-
cumstances, and they fail to reect the variation laws and power
load information in others [29]. Hence, the methodology of fore-
cast combination, also called ensemble forecasting, is a suitable
tool which has been successfully applied in many disciplines for
improving forecasting accuracy [30,31].
New studies related to electric load forecasting research area
show that the strategy of combining different models is quite useful
to improve the forecasting performance. The simplest way to com-
bine forecasts is to assign equal weights to the individual forecasts.
This method often work reasonably well when we are uncertain
about which method is best. So practically, one can make important
performance improvements in electric load forecasting by sim-
ply averaging the forecasts. In the study given by Taylor [15], a
variety of forecasting methods have been evaluated using minute-
by-minute British electricity demand data, and the results showed
that the simple average of the forecasts from the weather-based
approach and the double seasonal Holt-Winters-Taylor (HWT)
method resulted in forecasts that outperformed all other methods
beyond about an hour ahead. Another popular method for combin-
ing forecasts is the unequal weighting based on the past forecast
performance of models. In the approach proposed in [32], the short-
term load is obtained by combining the forecasts generated by the
seasonal autoregressive integrated moving average (SARIMA) and
the neural networks models, and the weights of the combination
are determined using a variancecovariance approach. The results
of this study showed an important improvement in forecasting
accuracy compared to other methods. A combined short-term load
forecasting model based on ensemble empirical mode decomposi-
tion (EEMD) and sub-section particle swarm optimization (SS-PSO)
is proposed in [33]. Firstly, the load sequence is decomposed into a
limited number of intrinsic mode function (IMF) components and
one remainder by EEMD. Then, the decomposed sequences are fore-
casted with appropriate forecasting models. SS-PSO is proposed
and used to optimize the linear combination weights. The numer-
ical results indicate that the proposed method in this reference
paper gets a higher forecasting accuracy rate when compared with
other forecasting techniques. A number of recent research studies
have suggested also the use of forecasts combination as a promising
tool for improving the electricity demand forecasting performance
[3436].
However, the combined models cited above have the drawback
of ignoring the effect of large forecasting errors. Actually, most of
the existing methods in the literature of electric load forecasting
focus only to evaluate the performance in normal and typical con-
ditions and they often ignore the effect of the anomalous load that
occurs on special days (such as the case of public holidays). For
example, if a combined model constructed by a weighted combi-
nation is implemented to forecast the load of a particular public
holiday, a poor estimation could be obtained at the case where a
large forecasting error is obtained by the individual method that
had a higher weight. This poor estimation can be attributed to the
dissimilar load behaviors of holidays compared with those of nor-
mal days. In addition, for a given special day, the pattern of the
load can vary from year to year, depending on the day and sea-
son of the year when the abnormal load is occurred. So making an
accurate abnormal load forecasting model is a more complicated
task compared to the case of constructing a normal load forecasting
model.
There have been few research studies for developing accurate
load forecasting techniques for special days. Some of them give
more attention to the quality of the load data and the rigor of
time series analysis. Some other papers aim to provide the appro-
priate output from a combination of forecasts. While some other
works, like the study performed by Ghofrani et al. [37], focus on
the consideration of both the appropriate combination of different
forecasts and the capture of the most relevant information of the
time series as two key elements to perform the task of improv-
ing the forecast accuracy of special daily conditions. One of the
existing methods in the literature that focus on the quality of the
data to predict the atypical load is the triple seasonal exponen-
tial smoothing method developed by Arora and Taylor in [38]. The
methodological contribution of the authors work is to show how
the exponential smoothing method can be adapted to model the
load for special days, when used in conjunction with an approach
based on rules. The results of the study showed that the advan-
tage of this rule-based method is located on its ability to model
normal and abnormal load in a unied framework. However, this
model is too complicated, requires a signicant amount of load
data, and a forecaster intervention is almost essential to restruc-
ture the model. In another study, a research work to improve the
short-term load forecasting for anomalous load conditions has been
presented [39]. The method of this study uses articial neural net-
work (ANN) models to provide the forecast scaled load curve, and
fuzzy inference models to forecast maximum and minimum loads
of the special day. The 24 hourly forecast loads of the special day
are predicted by combining the results of the ANN and the fuzzy
inference method. The benet of this hybrid structure was to uti-
lize the advantages of both, i.e., the generalization capability of
ANN and the ability of fuzzy inference for handling and formal-
izing the experience and knowledge of the forecasters. The test
results showed that the hybrid forecasting method could provide
a considerable improvement of the forecasting accuracy for the
special days. In Ref. [40] Ghayekhloo et al. presented a hybrid
load forecasting framework with a data preprocessing algorithm to
enhance the forecasting accuracy. The data preprocessing method
is a combination of the input selection, wavelet decomposition and
a novel standardization to provide the most appropriate inputs for
the Bayesian neural network algorithms. Genetic algorithm is used
to optimize the weighting coefcients of different forecast com-
ponents and minimize the forecast error. The performance and
accuracy of this short-term load forecasting method have been
evaluated using New England load data. The results from this study
showed a signicant improvement in the forecast accuracy when
compared to other existing techniques.
However, we would point out that most of the forecasting
methods in the literature have been implemented off-line using
simulated datasets. In fact, the good use of any load forecasting
algorithm in practice not only requires the consideration of rea-
sonable accuracy, but requires also the consideration of many other
issues, such as the online adaptation ability, the simplicity of use
and the computational efciency. This is particularly important for
the case where the lead time is as short as few minutes ahead and
the adopted method is a forecast combination. To our knowledge,
there is a lack of works that discuss the use of combined methods for
a practical application of very short-term load forecasting. This lack
has motivated us to provide the present work to the development
of an appropriate online load forecasting method that can ensure
low computation time and achieve good forecasting performance
on both normal and special daily conditions.
The main objective of this paper is to present a new forecast
combination method, namely HFCM (Hampel lter-based Forecast
Combination Method), for performing the online predictions of the
 A. Laoua et al. / Electric Power Systems Research 148 (2017) 3547
electricity demand. The proposed methodology consists of three
principal stages. Firstly, different models are adopted to forecast
the electricity consumption of the next-hour. The second part of
our methodology tries to detect any possible large forecasting error
and replace it with a normalized one. This function is performed
by employing the Hampel lter in a data sequence formed by the
primary forecasts. However, the normal use of outlier detection
methods is to detect spikes in load data and correct them [41].
Hence, we think that the present work is the rst in terms of using
a data pre-ltering technique as a tool for detecting over- or under-
estimation. At the third part of our methodology, the nal load
estimation is obtained by considering the trimmed mean of the
normalized forecasts. The basic idea in our work is to propose an
online load forecasting scheme that has the ability to ensure a low
computation time, avoid large forecasting errors and achieve good
forecasting accuracy on both normal and special daily conditions.
The real-time load data from the French power system are used as
an illustrative example to evaluate the online application and the
computation time of the proposed methodology, while the Aus-
tralian data for the state of New South Wales are used for a purpose
of comparison with a benchmark algorithm from the literature.
The paper is organized as follows. Section 2 describes the char-
acteristics of the electric load data and presents our methodology
for analyzing the electricity demand behavior. Section 3 presents
the proposed forecasting method. Section 4 provides and explains
forecasting results. Finally, Section 5 concludes the paper.
2. Electricity demand analysis and processing
To evaluate the performance of our approach, we used two elec-
tricity load datasets from two different countries (Australia and
France), with different time interval (5 min and 15 min). The France
dataset is collected from the transmission system operator website
[42]. It covers the period from April 07, 2013 to January 31, 2015,
and is measured every 15 min. The Australian dataset is for the
state of New South Wales and is provided by the Australian Energy
Market Operator [43]. The data is for two years, 2006 and 2007,
and is measured every 5 min. However, since we have opted to use
the electricity demand information as the sole entry requirement
for developing an appropriate online load forecasting algorithm,
it is therefore of rst importance to examine the main character-
istics of the load series and to understand the electricity demand
behavior. We present in Fig. 1 the power consumption in France
for the period from January 01, 2014 to December 31, 2014. It is
clear from the plot that the electric load behavior is mainly inu-
enced by daily and annually seasonal patterns where the electricity
consumption typically begins with low values in early morning
and increases signicantly towards a daily peak load; whereas the
Fig. 1. Electricity demand data of the French power system: from January 01, 2014
to December 31, 2014.
37
annual cycle is characterized by an annual peak demand in win-
ter and a dip on summer. Fig. 2 presents the daily peak load in
France over four different weeks. This plot shows an intraweek
seasonal cycle including the weekdays with similar demand pat-
terns and the weekend, which is characterized by a decrease in the
consumption.
Univariate classical methods such as BoxJenkins time series
models and exponential smoothing methods are frequently con-
sidered to be sufcient for modeling and capturing the seasonal
features of the data. However, it seems that the inclusion of the
intra-year cycle in a forecasting approach will not lead to any
improvement in the accuracy of predictions as short as an hour
ahead. In this regard, our choice fell on the consideration of the sea-
sonal autoregressive integrated moving average (SARIMA) and the
double seasonal Holt-Winters-Taylor (HWT) exponential smooth-
ing as effective techniques for modeling intraday and intraweek
seasonal cycles in the electricity demand data.
The modeling process of HWT and SARIMA methods is based on
the sequential nature of the time series, where data are ordered
in time and future values are inuenced by past values. However,
as a consequence of ignoring the special days during the modeling
process, the drawback of HWT and SARIMA approaches is found on
its inability for given satisfactory results when subject to anoma-
lous load conditions such the case of the rapid change in load
demand during holidays. To incorporate this limitation, we pro-
pose a parallel processing approach for generating very short-term
load forecasts for both normal and special daily conditions. In this
approach, the electrical load time series is separated in autonomous
points. A set of independent points means that the load at each
quarter-hour (or ve-minutes for the Australian data) of the day
is independent from the load at any other quarter-hour (or ve-
minutes). The main idea behind this approach is to partition the
electric load data into a dened number of parallel series and use
a separate model for each particular sub-series, where each sub-
series is related to a particular time of the day [44]. Our parallel
approach considers two principal cases. For normal days, the paral-
lel load series are classied by calendar according to the type of day:
Saturdays, Sundays, and workdays. In previous works, this method
of arrangement showed satisfactory results for forecasting the very
short-term load of normal days [16,23]. However, the particular
contribution of the present work is the consideration of the anoma-
lous load conditions. The consideration of such load in previous
existing few works in the literature have been realized using com-
plex modeling process that requires a signicant amount of data,
and a forecaster intervention is almost essential to restructure the
model. In contract to those existing works, our special days model-
ing algorithm is very simple and requires only few recent samples to
perform predictions. Hence, each parallel series in our special days
modeling process contains either the electricity demand informa-
tion from previous Sundays plus the load from the day to be forecast
(if it is a weekday), or the electricity demand information from
previous Saturdays or Sundays (if it is a weekend). The forecast-
ing methods designed to be used in a parallel implementation are
the back-propagation neural network (BPNN), the adaptive neuro-
fuzzy inference system (ANFIS), and a recently developed adaptive
Holts exponential smoothing (AHES) technique. For these paral-
lel models, the four step-ahead forecasts (or 12 for the Australian
data) can be obtained by performing one-step ahead forecast on
each separated parallel series related to the evaluation period (1 h).
According to this procedure, four (or 12) models with a single out-
put are implemented separately to provide the hour-ahead load
forecasts. For the methods based on the use of a sequential imple-
mentation, we made simply k step-ahead forecast from forecast
origin t; where k can up to 5 (or 13 for the Australian data), and
t is a time related to the last observation of the electric load time
series. The forecast horizon k is a term of up to 5 (or 13), since for
38 A. Laoua et al. / Electric Power Systems Research 148 (2017) 3547
Fig. 2. Daily peak electricity demand in France for four different weeks.
an online implementation the electricity demand information of a
given actual time (denoted as t + 1) is not available at the forecasting
moment.
3. The proposed online load forecasting methodology
The structure of the proposed very short-term load forecasting
methodology consists of three principal stages. At the rst part, a set
of basic, yet efcient, methods are used to perform primary electric
load forecasts for an hour in advance. Since the time series of the
electricity demand knows multiple variations, it is of rst impor-
tant to take into account several hypotheses and methods in order
to treat these variations and capture the recent trend of the electric-
ity demand. For this reason, two hypotheses have been considered
in our method. The rst is based on the use of a sequential approach
and the second is based on the use of a parallel approach. Moreover,
a mixture between classical and computational intelligence meth-
ods has been arranged. Hence, two classical methods (HWT and
SARIMA) and two articial intelligence methods (BPNN and ANFIS)
have been used. The inclusion of these models is based on their rea-
sonable accuracy, simplicity of use, and acceptable execution time.
The Holt-Winters exponential smoothing and the seasonal ARIMA
approaches are, without doubt, the most popular classical meth-
ods that have been used in the literature of load forecasting. This
popularity is attributed to the ability of these methods to handle
the seasonal component of the electricity demand time series. In
the other hand, articial neural network and neuro-fuzzy systems
are among the computational intelligence methods that have great
ability to model the non-linearity and the complex relationship that
Fig. 3. General structure of the proposed online load forecasting method.
exist between the load demand and the other exogenous factors. A
fth individual model, named adaptive Holts exponential smooth-
ing, is also included. We opted to use this fth method because it is
a recently developed approach that showed promising results for
handling the rapid variation of the electricity demand and ensuring
a low very short-term load forecasting error [16]. A simple com-
bination of the forecasts from the ve individual models is also
included to be a part of the primary forecasting models. To avoid
large forecasting errors, the Hampel lter is introduced at the sec-
ond part of our approach to replace over- and under-estimations
by normalized points. Then, the nal load estimation is obtained
by considering the trimmed mean of the normalized forecasts. The
general structure of our online load forecasting methodology is rep-
resented in Fig. 3, and the key elements and parameters of all the
proposed methods are presented in Table 1. The blanks in Table 1
mean that the element or the parameter concerned is not included
for a specic forecasting model, while * designates the inclusion
of the element/parameter by the forecasting model.
3.1. Individual models
3.1.1. Seasonal autoregressive integrated moving average
Exponential smoothing models and BoxJenkins (ARIMA) mod-
eling are amongst the most widely used time series models in the
eld of electric load forecasting. The popularity of these methods
is attributed to their ability of handling the different components
that form the electricity demand time series. The modeling process
of exponential smoothing methods assumes that the time series
can be perfectly predictable using its past variations. It enables to
 A. Laoua et al. / Electric Power Systems Research 148 (2017) 3547
Table 1
The key elements and parameters for the different methods.
Methods Classical methods
Elements SARIMA HWT
Mathematical model Eq. (1) Eqs. (2)(6)
Over/Under estimation detection
Simple average
Weighted average
Trimmed mean
Sequential approach * *
Parallel approach
Special days modeling
capture and forecast the level of the electricity demand data along
with different types of trends and seasonal patterns. Hence, a
mathematical model is required to represents the load as the
sum (additive model) or the product (multiplicative model) of a
number of components (just like the HWT model in Eqs. (2)(6)).
BoxJenkins models are similar to exponential smoothing mod-
els in dealing with capturing patterns in the historical data and
extrapolating those patterns into the future. However, in contrast
to exponential smoothing where we suppose that the time series
is stationary, autoregressive integrated moving average models
are more benecial in cases where the time series show evi-
dence of non-stationarity, where an initial differencing step (the
integrated part of the model) can be applied to reduce the non-
stationarity. Hence, the seasonal autoregressive integrated moving
average method can be very useful in our load forecasting problem
for modeling trend, cyclical, and irregular components of the elec-
tricity demand time series. The general structure of this approach
use a combination of past observations and past errors as well
as using a differencing scheme between an observation and the
corresponding observation from a previous period at the end of
translating the time series into smooth and a stationary process. A
general expression of a SARIMA (p, d, q) (P, D, Q )s model can be
given by:
   D  
p (B) P Bs (1 B)d 1 Bs yt = q (B) Q Bs t
where yt and t are respectively the actual value and random error
at time t. B is a backward shift operator satisfying: Bt yt = yt1 . P (Bs ),
Q (Bs ) and (1 Bs )D are related to autoregressive, moving aver-
age and differencing for seasonal components. While p (B), q (B)
and (1 B)d are related to autoregressive, moving average and dif-
ference for the non-seasonal component. S is the seasonal order.
The adopted SARIMA model for the France and the Australian load
data can be described as SARIMA (1, 1, 1) (2, 1, 1)96 and SARIMA
(3, 1, 3) (2, 1, 1)288 , respectively.
3.1.2. Holt-Winters-Taylor method
Exponential smoothing method has established itself as one
of the robust and accurate strategies for forecasting time series
of different characteristics. In particle, the Holt-Winters exponen-
tial smoothing is a very successful method for dealing with time
series which contain both trend and seasonal variations. However,
the standard Hold-Winters is suitable for accommodating only one
seasonal pattern on the data. In this work we apply an extended
version of the standard Holt-Winters method, originally developed
by Taylor [45], for accommodating both the intraday and intraweek
seasonal cycles in the electricity demand series. The formulation for
the double seasonal Holt-Winters-Taylor method is given in the
following expressions:
St = (yt /Dts1 Wts2 ) + (1 )(St1 + Tt1 )
Tt = (St St1 ) + (1 )(Tt1 )
39
Computational intelligence methodsCombined methods
AHES BPNN ANFIS SFCM HFCM- HFCM-TM
WAM
Eq. (7)(9) Not required Not required Required in Required in Required in
some parts some parts some parts
* *
* *
*
*
* * *
* * * * * *
* * * * * *
Dt = (yt /St Wts2 ) + (1 )Dts1 (4)
Wt = (yt /St Dts1 ) + (1 )Wts2 (5)
yt (k) = (St + kT t )Dts1+k Wts2+k
+ k (yt ((St1 + Tt1 )Dts1 Wts2 )) (6)
St and Tt , are the smoothed level and trend; Dt and Wt is the seasonal
indices for the intraday s1 (s1 = 96 for the French data, s1 = 288 for
the Australian data) and intraweek s2 (s2 = 672 for the French data,
s2 = 2016 for the Australian data) seasonal cycles, respectively; ,
, and are the smoothing parameters; yt is the actual value of
the time series in period t; and yt (k) is the k step-ahead forecast
made from forecast origin t. The term involving the parameter  in
the forecast function expression Eq. (6) is a simple adjustment for
rst-order autocorrelation.
In this paper, HWT method parameters are estimated by tak-
ing into account the minimum forecasting error for three principal
stages. For the French load series, we rst generated 2 103 vectors
of parameters from a uniform random number generator between
0 and 1. For each of the vectors, we then evaluated the absolute
percentage error (APE) on the daily peak demand forecasting for
August 02, 2014. The 200 vectors producing the lowest APE are
evaluated; in their turns, for predicting the load curve of August
(1)
02, 2014 (96 load values). Among the 200 vectors, the 10 vec-
tors that have the lowest mean absolute percentage error (MAPE)
are evaluated at last for next-hour load prediction over the period
from August 02, 2014 to August 08, 2014. The one producing
the lowest MAPE value was chosen as the nal parameter vector.
The optimized values of HWT method parameters were: = 0.987;
 = 0.057; = 0.006; = 0.272; and  = 0.008. For the Australian load
series, the parameters are estimated by a similar manner and the
optimized values were: = 0.996;  = 0.191; = 0.273; = 0.545;
and  = 0.206.
3.1.3. Back-propagation neural network
Due to their inherent capability for solving any complex rela-
tionships between inputs and outputs through a learning process,
the use of neural networks models is nowadays an alternative
method for dealing with the electric load forecasting problem
[46,47]. In this regard, a univariate back-propagation neural net-
work model is used in this work. The conguration of the proposed
neural network is very similar to the one used in [48]. The proposed
BPNN model contains one hidden layer with four sigmoidal neurons
followed by an output layer of linear neuron. The load information
from a previous parallel series is used as input, and the load data
for the quarter-hour (or the ve-minutes) to be forecast is adopted
as the output of the network. Such univariate modeling procedure
(2) is frequently considered to be sufcient for very short-term load
forecasting because the weather variables tend to change in a
(3) smooth manner over short time scales, and the effects of exogenous
40 A. Laoua et al. / Electric Power Systems Research 148 (2017) 3547
factors can be captured in the electricity demand time series itself.
During the building process of our neural network model, the load
data of one year are divided into three subsets: 70% for training,
15% for validation and 15% for test. The LevenbergMarquardt algo-
rithm is considered to train the network and the mean square error
(MSE) criteria is used as the performance function criteria. The
number of epochs for training the network is limited to a maximum
of 20 epochs.
3.1.4. Adaptive neuro-fuzzy inference system
The adaptive neuro-fuzzy inference system is one of the suc-
cessful schemes that combine the merits of human reasoning from
fuzzy logic and the adaptive learning ability of articial neural
networks. Such hybrid system is a fuzzy TakagiSugeno model
where the parameters and the optimal topology are obtained by
applying a learning process of an adaptive articial neural net-
work [49]. It can be viewed as a multi-layer feed-forward network
in which each node (neuron) performs a particular function on
incoming signals. In this architecture, the rst layer executes the
fuzzication process and gives the degree of fuzzy membership of
the input. The second computes the ring strengths of the fuzzy
rules. The third layer normalizes the ring strengths from the pre-
vious layer. The fourth layer gives the consequent part of the fuzzy
rules. Finally, the fth layer presents a summation of all incoming
signals from the fourth layer [23]. The proposed neuro-fuzzy infer-
ence structure uses generalized bell-shaped membership function,
two fuzzy rules, and 20 epochs for identifying the fuzzy inference
system (FIS) parameters and the optimal topology of the ANFIS. The
input/output pairs are similar to that of BPNN model.
3.1.5. Adaptive Holts exponential smoothing
AHES is an adaptive load forecasting approach which has been
introduced in [16] to produce one step-ahead prediction. The
method is adaptive at the manner that their smoothing parame-
ters are updated every time when a new data point is recorded.
It is found that the adaptive scheme of this methodology has the
advantage to perceive the variation that could be effected in the
electric load time series, and so that the ability of providing a slight
error when forecasting the electricity consumption of a particular
time. The formulation for the Holts exponential smoothing could
be set up as:
St = yt + (1 )(St1 + Tt1 ) (7)
Tt = (St St1 ) + (1 )(Tt1 ) (8)
yt (k) = St + kT t (9)
St and Tt are the smoothed level and trend; and  are the smooth-
ing parameters; yt is the actual value of the time series in period t;
and yt (k) is the k step-ahead forecast made from forecast origin t.
For the purpose of predicting the electricity consumption at a
particular time m; we create at a primary step two parallel load
series: a rst series for a previous time m 1, and a second series
for the time m. In the second stage, we look to nd a pair of smooth-
ing parameters that give the smallest forecasting error when going
from the penultimate to the last data point of the rst electricity
demand time series. Since the exponential smoothing methods are
mostly inuenced by newest data points, and especially the abil-
ity of the Holts method to identify the tendency in non-seasonal
time series, it seems that the fact of searching an optimal model
assuring the minimum error at the time m 1 may signicantly
help to provide a small error for predicting the load of the specied
time m.
3.1.6. Simple average-based forecast combination method
In this method, a simple average of the primary forecasts from
the individual models is performed with the aim of improving the
forecasting accuracy. This method is based on assigning an arith-
metic mean to the individual forecasts for a time m and often
generates reliable forecasts for normal days. Its mathematical rep-
resentation is given below:
1
n
c
Ym = Ym,i (10)
n
i=1
c is the combined forecast; Y
where Ym m,i is the individual forecast
obtained from the ith model; n is equal to 5, the total number of
proposed individual models.
3.2. Data cleaning lter to avoid large forecasting errors
Sometimes, even the more advanced combination method may
fails to improve the forecasting accuracy, especially for the anoma-
lous loads where the accurate prediction is almost considered as a
difcult task. For such cases, we can nd frequently that the fore-
cast of one of the individual models is signicantly outside the range
of the majority of methods. To avoid large forecasting errors, the
Hampel lter is introduced at the second part of our approach to
replace over- and under-estimations by normalized points, which
can help to improve the forecasting accuracy when combining the
normalized forecasts.
The Hampel lter is a moving window nonlinear data clean-
ing lter that searches outliers or abnormal local data in a time
sequence and replaces them with a more reasonable value when it
nds them [50]. For a data sequence (Y1 , Y2 , . . ., Yn ), the lter forms
a collection of samples from which it computes the median and the
median absolute deviation (MAD) and performs the outlier test.
The median of a data sequence is obtained by rst rank-ordering
it from smallest to largest, i.e. Y1 Y2 Y3 Yn1 Yn and then
taking either the middle value (if it is odd) or the average of the
middle two values (if it is even) [51]. The Hampel lter denes out-
liers as those data points Yj whose absolute difference from the
median of the short series in the moving window is greater than a
pre-determined threshold:
 
Yj Y0  > T S0 (11)
where Y0 represents the median value of the rank-ordered samples
of the data sequence of length n, T is a scalar threshold, and S0 is
the median absolute deviation, as dened by:
 
S0 = median Yj Y0  (12)
If the Hampel lter declares a data point to be an outlier, its value
is then replaced by the median of its neighbors. For the Hampel lter
used in our load forecasting model, the time sequence is dened
by the forecasts from the individual models and from the simple
average-based forecast combination method.
3.3. Final electric load estimation
Once the Hampel lter is performed, the nal load forecasting is
obtained by taking into account the trimmed mean of the normal-
ized forecasts (the mean of the normalized forecasts, excluding the
highest and lowest value). For the purpose of comparison, we pro-
pose also a second variation of HFCM. In this method, the weighted
average of the normalized forecasts is employed to combine fore-
casts from primary models. This combined forecast can be written
as follows:
F (f1 (Y ) , f2 (Y ) , . . ., fn (Y )) = w1 f1 (Y ) + w2 f2 (Y ) + + wn fn (Y ) (13)
where fi is related to a primary forecasting method transforming
the real electricity consumption Y of a previous quarter-hour (or
ve-minutes) to its prediction Y . In the present work, the inverse
 A. Laoua et al. / Electric Power Systems Research 148 (2017) 3547 41

Fig. 4. HFCM-TM results for the French load data: from August 09, 2014 to August 12, 2014.

error weights are used [1], so that a model with a smaller error
obtains more weight:
e1
wi = n i , i = 1, 2, . . ., n
e1
i=1 i
where ei designates the absolute percentage error realized by the
ith prediction method. By this approach, the forecasting technique
with the lowest value of the absolute percentage error receives a
higher weight in composing the combined prediction. For the rest
of this paper, the method based on the weighted arithmetic mean
algorithm will be dened by HFCM-WAM, while the proposed
method based on the trimmed mean will be noted by HFCM-TM.
4. Results and discussion
In this section, we describe an evaluation of the Hampel
lter-based forecast combination method for hour-ahead load fore-
casting on both normal and special daily conditions. We will also
(14) compare the results with a number of benchmark algorithms and
forecast combination methods. In addition, a discussion about the
computation time of our provided tool for realizing the online pre-
diction of the electricity demand will be presented.
4.1. Evaluation of forecasts for the French load data
Figs. 47 show the results of HFCM-TM for next-hour elec-
tricity demand prediction in France along the week from 09th to
15th August 2014, and the week from 10th to 16th January 2015,

Fig. 5. HFCM-TM results for the French load data: from August 13, 2014 to August 15, 2014.
42 A. Laoua et al. / Electric Power Systems Research 148 (2017) 3547

Fig. 6. HFCM-TM results for the French load data: from January 10, 2015 to January 13, 2015.

Fig. 7. HFCM-TM results for the French load data: from January 14, 2015 to January 16, 2015.

respectively. It is obvious from this gure that satisfactory results where Y m is the predicted load for the time m, Ym is the actual load
were obtained by using the Hampel lter-based forecast combi- at the time m, n is the total number of forecasting cases.
nation approach, since the shape of predicted load curves is very All detailed forecasting performances are listed in Table 2. These
similar to that of actual load curves. However, we need to compare results clearly demonstrate the effectiveness of the proposed fore-
the performance of our HFCM-TM against the performance all other casting strategy and show that HFCM-TM outperformed all other
considered methods. For the evaluation of the methods considered methods. It can be seen from the relative performance that the
in this study, we used APE, and MAPE criteria: average MAPE value of HFCM-TM in normal days is equal to
0.478%, which is 5.720% and 22.023% more accurate than the simple
|Y m Ym | average-based forecast combination method and the best individ-
APE = 100 (15)
Ym ual forecasting approach (AHES), respectively. The results show
also that the method based on a weighted mean (HFCM-WAM) has
1  |Ym Ym |
n been slightly outperformed by HFCM-TM. Among the individual
MAPE = 100 (16) forecasting methods, the recently developed AHES has showed a
n Ym
m=1 very good forecasting accuracy. The obtained performance of this
 A. Laoua et al. / Electric Power Systems Research 148 (2017) 3547 43

Table 2
The relative error (MAPE) for forecasting the electricity consumption in France over two weeks.

Time SARIMA HWT BPNN ANFIS AHES SFCM HFCM-WAM HFCM-TM

09/08/2014 1.006 0.566 0.979 0.843 0.721 0.583 0.556 0.570

10/08/2014 1.337 0.758 1.080 0.906 0.739 0.661 0.641 0.642
11/08/2014 0.666 0.569 1.420 0.840 0.583 0.582 0.565 0.552
12/08/2014 0.533 0.488 0.685 0.740 0.393 0.406 0.400 0.404
13/08/2014 0.430 0.634 0.703 0.788 0.479 0.404 0.399 0.376
14/08/2014 0.437 1.511 0.877 0.777 0.334 0.499 0.416 0.421
15/08/2014 1.088 1.710 0.982 0.737 0.743 0.649 0.740 0.642
10/01/2015 1.027 0.483 1.095 0.513 0.703 0.472 0.460 0.438
11/01/2015 1.204 0.777 1.058 0.649 1.040 0.690 0.663 0.648
12/01/2015 1.004 0.608 0.524 0.446 0.685 0.439 0.437 0.430
13/01/2015 0.806 0.758 0.519 0.490 0.617 0.470 0.431 0.430
14/01/2015 0.684 0.728 0.560 0.500 0.589 0.493 0.486 0.486
15/01/2015 0.574 0.728 0.454 0.521 0.641 0.444 0.425 0.415
16/01/2015 0.452 0.829 0.520 0.530 0.445 0.447 0.393 0.407
Normal days 0.781 0.726 0.806 0.657 0.613 0.507 0.483 0.478
All days 0.803 0.796 0.818 0.663 0.622 0.517 0.501 0.490

Table 3
The relative error (MAPE) for the public holidays of 2014 in France.

Time AHES SFCM HFCM-WAM HFCM-TM

01/01/2014 1.433 1.517 1.382 1.400

21/04/2014 0.603 0.992 0.722 0.747
01/05/2014 1.128 1.111 0.901 0.837
08/05/2014 1.115 0.900 0.789 0.783
29/05/2014 0.946 1.190 0.917 0.898
09/06/2014 1.218 0.883 0.797 0.730
14/07/2014 0.649 0.803 0.628 0.636
15/08/2014 0.743 0.644 0.740 0.642
01/11/2014 1.132 0.812 0.835 0.799
11/11/2014 0.572 0.991 0.670 0.639
25/12/2014 1.563 1.586 1.547 1.525
All days 0.996 1.026 0.890 0.863

higher performance on the validation period. Therefore, if an over

Fig. 8. The exponential smoothing parameters of AHES over a week in summer and or under-estimation of the load is obtained by this model, we will
a week in winter. get then an erroneous forecast if the weighted combination method
is not able to detect and to replace this over or under-estimation.
method is mainly attributed to its adaptive character where the To incorporate this limitation it will be more appropriate to assign
smoothing parameters, as shown in Fig. 8, are updated every time equal weights to each individual model, or better; to exclude the
when a new observation is recorded. A satisfactory performance highest and lowest data value from the normalized forecasts and
of a MAPE value equal to 0.743% has been also achieved by this use then the arithmetic mean of the four remaining values. Thus,
method for the public holiday of August 15th 2014. For this spe- this is the case with our HFCM-TM.
cic day, a further improvement in forecasting accuracy has been In order to clarify the key idea of our HFCM-TM, we present
obtained using HFCM-TM. in Fig. 9 and in Table 4 the forecasting results for December 26,
However, we cannot be denitively sure about the superiority
of our combined method for forecasting atypical loads if we base
only on the results of just one comparing day. So an extended eval-
uation period is extremely necessary to make a clear idea about the
performance of each method during the atypical load conditions.
For that reason, a thorough assessment of AHES, SFCM, HFCM-WAM
and HFCM-TM is performed off-line on the length of the public hol-
idays in France for the year of 2014 and the results of the MAPEs
are listed in Table 3. It can be seen from the results that the average
MAPE of HFCM-TM is less than all the average MAPEs of the other
methods. Compared to the case of the normal days where HFCM-
TM showed only 5.720% higher accuracy over SFCM, we can notice
for the case of anomalous days that a very signicant improvement
is achieved by HFCM-TM; with an average MAPE value lower by
15.887%. By comparing the results obtained by HFCM-TM with the
results obtained by HFCM-WAM, we can clearly observe that this
last method has a lower performance. We think that the main rea-
son for this is attributed to the weighted algorithm of HFCM-WAM.
In this method of combination, the forecasting model that has a
lower forecasting error in a training period gets a higher weight in
composing the combined forecast. However, a model with higher Fig. 9. Results of HFCM-TM and HFCM-WAM for the forecasting process of
performance in a training period does not mean a model with December 26th 2014 at 00:00.
44 A. Laoua et al. / Electric Power Systems Research 148 (2017) 3547
Table 4
Forecasting results for December 26, 2014 at 00:00 AM.
Results Actual load (MW) Predicted load (MW)
Method
SARIMA 60363
HWT 60005
BPNN 59439
ANFIS 59597 59560
AHES 61300
SFCM 60133
HFCM-WAM 60027
HFCM-TM 59919
2014 at 00:00. We can perceive three important points from these
results: (1) an over-estimation of the load has been occurred by the
best individual forecasting method, (2) HFCM-TM is outperformed
by BPNN and ANFIS models, (3) Although that ANFIS model has
received the second lower weight in composing the combined
forecast of HFCM-WAM, the obtained results indicated that this
individual approach has achieved the best forecasting performance
among all the considered methods. This is consistent with what
we have reported above about the non-existence of a forecasting
method which gives the best results for all times. However, the
advantage of our proposed HFCM-TM is located on its ability to
reduce the uncertainty and provide electricity demand predictions
which are very close to the actual values of the load. This can be
clearly observed by Figs. 10 and 11, where we compared the APE
distribution for AHES, SFCM, HFCM-WAM and HFCM-TM on nor-
mal and anomalous load conditions, respectively. Although that
the simple averaging used in SFCM is among the most commonly
used approaches to produce the combined forecast, we can perceive
that this method of combination has not allowed to any improve-
ment in the case of the anomalous load conditions. In contrast,
we can observe that HFCM-WAM and HFCM-TM give almost the
best results and show excellent forecasting performances, since an
absolute percentage error lower than one percent has been
Fig. 10. APE distribution of AHES, SFCM, HFCM-WAM, and HFCM-TM for normal loads.
Fig. 11. APE distribution of AHES, SFCM, HFCM-WAM, and HFCM-TM for anomalous loads.
occurred for more than 85% and 65% of the evaluation period of
normal and special days, respectively. However, the results from
APE (%) these gures show that HFCM-TM is more effective since it avoided
the most signicant number of large forecasting errors.
1.285
0.685 4.2. Evaluation of forecasts for the Australian load data
0.265
0.062
2.857 In this subsection, we carry out a further evaluation of our
0.895 method through performing a simulation of the Australian load
0.721 data for the state of New South Wales. The main reason for this
0.540
application is to compare our model with the one proposed in
[26]. The model proposed in the cited reference, called Advanced
Wavelet Neural Network (AWNN), combines an advanced wavelet
algorithm for load decomposition, mutual information for feature
selection and neural networks as prediction algorithm. AWNN was
tested on the Australian load dataset of 2007 and achieved a MAPE
equal to 0.938% for a forecast horizon of an hour-ahead. For a fair
comparison, we used the same evaluation period (2007) to build
hour-ahead load forecasting with the prediction algorithm pro-
posed in the current study. Fig. 12 presents the predicted load with
HFCM-TM as a function of an actual load and plots also the actual
and predicted load values for a typical working day (December
31, 2007). This gure indicates the high correlation between the
predicted and actual load values. Table 5 compares the predictive
accuracy results of HFCM-TM with those of AWNN. The results in
this table show that HFCM-TM achieved overall prediction errors as
follows: MAPE = 1.048% for the public holidays and MAPE = 0.860%
for all the days of 2007. Furthermore, HFCM-TM showed better
forecasting results and the MAPE improvement by using HFCM-TM
compared to AWNN is equal to 8.316%.
4.3. Presentation of the proposed online load forecasting tool
In addition to an effective model that ensures a very high
accuracy, a friendly forecasting tool takes also an important role
 A. Laoua et al. / Electric Power Systems Research 148 (2017) 3547 45

Fig. 12. Predicted load by HFCM-TM plotted against actual load from the Australian dataset.

Table 5 led us; through an extensive study, to the development of the

The relative error (MAPE) for the load data of the New South Wales.
Day type All the public holidays of 2007 All the days of 2007
Forecasting method
AWNN 0.938
HFCM-TM 1.048 0.860
at the real implementation of any electric load forecasting tech-
nique. As a natural extension of the work in [16], the AHES is
the method that has been implemented for performing the online
electricity demand prediction in France over the summer week
and winter week of the evaluation period. We opted to use the
French power data for the simple reason that we found the RTE
(transmission system operator of France) website very useful and
accurate in providing the electric load data in real-time. How-
ever, the aim of achieving forecasting error not higher than one
percent under both normal and anomalous load conditions has
Hampel lter-based forecast combination method. For the online
implementation of this method, many considerations, algorithms
and functions have been taken into account for developing suit-
able software using MATLAB R2014b environment. As shown in
Fig. 13, our method is currently accompanied with a forecasting
tool and working very well for the online electricity demand predic-
tion of the French power system. This forecasting tool can perform
an hour ahead forecasting from any moment selected from the
user. This is important, because in practice, the power compa-
nies need to perform the load predictions in an iterative manner,
not once every an hour. However, for an evaluation purpose, the
same quarter-hours of a given day are considered to form the
rst step in an hour ahead load forecasting over all the evalua-
tion period of our study. Hence, the rst steps in the 1-h ahead
forecasting are 00:15, 01:15, 02:15, . . ., 23:15. Thus, in this manner
we get only one forecasted value for each specic moment of the
day.

Fig. 13. Developed tool for online electricity demand prediction.

46 A. Laoua et al. / Electric Power Systems Research 148 (2017) 3547
Table 6
The computation time of all considered methods in four different forecasting cases.
Method
SARIMA HWT BPNN
Results
Evaluation period Sat, 09 Aug 2014, 12:45-13:30
Computation time (s) 4.38 5.16 6.08
Evaluation period Thu, 25 Dec 2014, 23:15-00:00
Computation time (s) 6.57 7.39 9.20
Evaluation period Thu, 27 Oct 2016, 21:15-22:00
Computation time (s) 6.77 7.28 7.84
Evaluation period Tue, 01 Nov 2016, 21:30-22:15
Computation time (s) 6.27 6.93 8.35
However, there are three principal tasks that can be performed
by using our provided forecasting tool. Obviously, the main task of
our tool is to enable the online prediction of electricity demand.
All what the user is required to do for achieving this function is to
select the forecasting method and the type of day in which he would
like to perform the predictions and the software complete then the
task by passing through four stages: (1) getting the current time in
France, (2) connecting to RTE website and downloading the avail-
able load data for the forecasting day, (3) updating the database by
adding up the downloaded data to the main le the include histor-
ical load data, (4) performing the required algorithms and showing
the load forecasting results for the next-hour. In addition to the
online implementation, the proposed tool enables also to perform
the off-line predictions. This is important if the user would like to
see the previous forecasting performance by taking into account
different error measurement criteria. The third task that can be
achieved by our provided software is the ability to update the his-
torical load data in case where the user neglected the use of the
tool for several days. In that case, an intervention from the user is
required to select a desired day from Update The Historical Data
panel in order include the missed data to the main historical data
le.
We present in Table 6 the computation time of our software on a
laptop of 4 GB of RAM, i3 380 M processor, a hard drive of 5200-rpm,
and operating in Windows 10 platform. The table shows the com-
putation time of each considered method in four different cases:
the rst case (09 August 2014) is related to an off-line execution
for performing hour ahead load prediction during a weekend; the
second case (25 December 2014) is related to an off-line implemen-
tation for performing hour ahead load prediction during a special
day; the third case (27 October 2016) is related to an online execu-
tion for performing hour ahead load prediction during a weekday;
and the fourth case (01 November 2016) is related to an online
execution for performing hour ahead load prediction during a pub-
lic holiday. One can clearly perceive from these results that the
individual models require less computation time compared to the
combined methods. However, the difference in computation time
is very acceptable if we take into account that the combined meth-
ods require the simultaneous use of many algorithms and methods
to perform predictions. Moreover, the execution time of our HFCM-
TM method is lower than 15 s. This time includes the time to update
the historical load data and to write the forecasting results on an
XLS le. If we compare the execution time of HFCM-TM for the pub-
lic holiday of 2014 with that of the public holiday of 2016, we can
clearly perceive that both cases require practically the same exe-
cution time of about 11 s. This is can be easily justied if we take
into account that we have limited the data to be used by each indi-
vidual method to a predened amount of samples. For example,
even if the main historical load data is consisted by twenty years
of electricity demand observations, the HWT method is supposed
to use only the last year of observations as a sufcient amount of
data for handling the seasonality of the time series and performing
the predictions. Hence, with an execution time lower than 15 s, our
ANFIS AHES SFCM HFCM-WAM HFCM-TM
4.29 4.71 7.96 11.85 8.39
5.28 7.66 11.24 15.83 11.61
5.97 8.51 11.39 16.61 12.35
5.84 8.25 10.84 13.24 11.06
combined method can be easily implemented at any power system
for performing the online predictions of the electricity demand.
5. Conclusion
In this paper, a new technique was developed to perform the
online predictions of the electricity demand for a lead time of an
hour ahead. Such very-short term forecasts are important for the
real-time scheduling of electricity generation as well as the load-
frequency control and the economic dispatch functions. However,
since the load demand behavior is complex and mainly affected
by several factors, there is no forecasting method which gives the
best result for all times. It is therefore necessary to integrate differ-
ent models for reducing the uncertainty of the predictions. In this
regard, the proposed method is categorized in combined forecast-
ing methods. To deal with the seasonal features and the non-linear
character of the electric load time series, two classical methods and
two computational intelligence approaches, have been considered.
A recently developed adaptive exponential smoothing method and
a simple combination of the forecasts from the ve individual mod-
els were also used as candidate models. To improve the forecasting
accuracy when combining forecasts from the primary models, a
data cleaning lter and the trimmed mean were employed as two
key elements at this function of improving the performance and
avoiding the large forecasting errors. The obtained results showed
that the proposed forecasting scheme has satisfactory performance
in both normal and special daily conditions, with an average mean
absolute percentage error not higher than one percent in both two
cases. Moreover, our method is also accompanied with a user-
friendly graphical interface and the execution time of this software
on a laptop of 4 GB of RAM; i3 380 M processor; and a hard drive of
5200-rpm, is lower than 15 s. Therefore, the proposed methodology
can be easily implemented at any power system for performing the
online predictions of the electricity demand.
References
[1] J.X. Che, J.Z. Wang, Short-term load forecasting using a kernel-based support
vector regression combination model, Appl. Energy 132 (1) (2014) 602609.
[2] S. Li, P. Wang, L. Goel, Short-term load forecasting by wavelet transform and
evolutionary extreme learning machine, Electr. Power Syst. Res. 122 (2015)
96103.
[3] A. Laoua, M. Mordjaoui, F. Laoua, T.E. Boukelia, Daily peak electricity
demand forecasting based on an adaptive hybrid two-stage methodology, Int.
J. Electr. Power Energy Syst. 77 (2016) 136144.
[4] G. Dudek, Pattern-based local linear regression models for short-term load
forecasting, Electr. Power Syst. Res. 130 (2016) 139147.
[5] L. Zjavka, V. Snsel, Short-term power load forecasting with ordinary
differential equation substitutions of polynomial networks, Electr. Power
Syst. Res. 137 (2016) 113123.
[6] E. Gonzalez-Romera, M.A. Jaramillo-Moran, D. Carmona-Fernandez, Monthly
electric energy demand forecasting based on trend extraction, IEEE Trans.
Power Syst. 21 (4) (2006) 19461953.
[7] N. Abu-Shikhah, F. Elkarmi, Medium-term electric load forecasting using
singular value decomposition, Energy 36 (7) (2011) 42594271.
 A. Laoua et al. / Electric Power Systems Research 148 (2017) 3547
[8] Z. Hu, Y. Bao, R. Chiong, T. Xiong, Mid-term interval load forecasting using
multi-output support vector regression with a memetic algorithm for feature
selection, Energy 84 (2015) 419431.
[9] O. Ahmia, N. Farah, Multi-model approach for electrical load forecasting, in:
2015 SAI Intelligent Systems Conference (IntelliSys), London, 1011
November 2015, 2015, pp. 8792.
[10] H.M. Al-Hamadi, S.A. Soliman, Long-term/mid-term electric load forecasting
based on short-term correlation and annual growth, Electr. Power Syst. Res.
74 (3) (2005) 353361.
[11] U. Perwez, A. Sohail, S.F. Hassan, U. Zia, The long-term forecast of Pakistans
electricity supply and demand: an application of long range energy
alternatives planning, Energy 93 (2) (2015) 24232435.
[12] A. Laoua, M. Mordjaoui, F. Laoua, An evaluation of conventional and
computational intelligence methods for medium and long-term load
forecasting in Algeria, in: 3rd International Conference on Control,
Engineering & Information Technology CEIT2015, Tlemcen, Algeria, 2527
May, 2015, 2015.
[13] K. Liu, S. Subbarayan, R.R. Shoults, M.T. Manry, C. Kwan, F.L. Lewis, J.
Naccarino, Comparison of very short-term load forecasting techniques, IEEE
Trans. Power Syst. 11 (2) (1996).
[14] A. Kotillov, Very short-term load forecasting using exponential smoothing
and ARIMA models, J. Inf. Control Manag. Syst. 9 (2) (2011).
[15] J.W. Taylor, An evaluation of methods for very short-term load forecasting
using minute-by-minute British data, Int. J. Forecast. 24 (2008) 645658.
[16] A. Laoua, M. Mordjaoui, D. Dib, Very short-term electricity demand
forecasting using adaptive exponential smoothing methods, in: 15th
International Conference on Sciences and Techniques of Automatic Control &
Computer Engineering STA2014, Hammamet, Tunisia, 2123 December
2014, 2014, pp. 553557.
[17] D.J. Trudnowski, W.L. McReynolds, J.M. Johnson, Real-time very short-term
load prediction for power-system automatic generation control, IEEE Trans.
Control Syst. Technol. 9 (2) (2001).
[18] C. Guan, P.B. Luh, L.D. Michel, Z. Chi, Hybrid Kalman lters for very short-term
load forecasting and prediction interval estimation, IEEE Trans. Power Syst. 28
(4) (2013) 38063817.
[19] W. Charytoniuk, M.S. Chen, Very short-term load forecasting using articial
neural networks, IEEE Trans. Power Syst. 15 (1) (2000) 263268.
[20] P. Qingle, M. Zhang, Very short-term load forecasting based on neural network
and rough set, in: 2010 International Conference on Intelligent Computation
Technology and Automation, vol. 3, 1112 May 2010, 2010, pp. 11321135.
[21] D. Chen, M. York, Neural network based very short term load prediction, IEEE
PES Transm. Distrib. Conf. Expos. 2124 (2008) 19.
[22] L.C.M. de Andrade, I.N. da Silva, Very short-term load forecasting using a
hybrid neuro-fuzzy approach, in: 2010 Eleventh Brazilian Symposium on
Neural Networks (SBRN), 2328 October 2010, 2010, pp. 115120.
[23] A. Laoua, M. Mordjaoui, D. Dib, One-hour ahead electric load forecasting
using neuro-fuzzy system in a parallel approach, in: Computational
Intelligence Applications in Modeling and Control, Studies in Computational
Intelligence, vol. 575, 2015, pp. 95121.
[24] A. Setiawan, I. Koprinska, V.G. Agelidis, Very short-term electricity load
demand forecasting using support vector regression, in: Proceedings of
International Joint Conference on Neural Networks, Atlanta, Georgia, USA,
2009.
[25] Y. Wang, L. Wu, On practical challenges of decomposition-based hybrid
forecasting algorithms for wind speed and solar irradiation, Energy 112
(2016) 208220.
[26] M. Rana, I. Koprinska, Forecasting electricity load with advanced wavelet
neural networks, Neurocomputing 182 (2016) 118132.
[27] G. Dudek, Pattern similarity-based methods for short-term load forecasting.
Part 1: Principles, Appl. Soft Comput. 37 (2015) 277287.
47
[28] G. Dudek, Pattern similarity-based methods for short-term load forecasting.
Part 2: Models, Appl. Soft Comput. 36 (2015) 422441.
[29] C.E. Borges, Y.K. Penya, I. Fernandez, Evaluating combined load forecasting in
large power systems and smart grids, IEEE Trans. Ind. Inf. 9 (3) (2013)
15701577.
[30] L.M. de Menezes, D.W. Bunn, J.W. Taylor, Review of guidelines for the use of
combined forecasts, Eur. J. Oper. Res. 120 (1) (2000) 190204.
[31] R. Clemen, Combining forecasts: a review and annotated bibliography, Int. J.
Forecast. 5 (1989) 559583.
[32] J. Wang, J. Wang, Y. Li, S. Zhu, J. Zhao, Techniques of applying wavelet
de-noising into a combined model for short-term load forecasting, Int. J.
Electr. Power Energy Syst. 62 (2014) 816824.
[33] Z. Liu, W. Sun, J. Zeng, A new short-term load forecasting method of power
system based on EEMD and SS-PSO, Neural Comput. Appl. 24 (3) (2014)
973983.
[34] L. Xiao, J. Wang, R. Hou, J. Wu, A combined model based on data pre-analysis
and weight coefcients optimization for electrical load forecasting, Energy 82
(2015) 524549.
[35] J. Nowotarski, B. Liu, R. Weron, T. Hong, Improving short term load forecast
accuracy via combining sister forecasts, Energy 98 (2016) 4049.
[36] L. Xiao, W. Shao, T. Liang, C. Wang, A combined model based on multiple
seasonal patterns and modied rey algorithm for electrical load
forecasting, Appl. Energy 167 (2016) 135153.
[37] M. Ghofrani, M. Ghayekhloo, A. Arabali, A. Ghayekhloo, Hybrid short-term
load forecasting with a new input selection framework, Energy 81 (2015)
777786.
[38] S. Arora, J.W. Taylor, Short-term forecasting of anomalous load using
rule-based triple seasonal methods, IEEE Trans. Power Syst. 28 (3) (2013)
32353242.
[39] K.H. Kim, H.S. Youn, Y.C. Kang, Short-term load forecasting for special days in
anomalous load conditions using neural networks and fuzzy inference
method, IEEE Trans. Power Syst. 15 (2) (2000) 559565.
[40] M. Ghayekhloo, M.B. Menhaj, M. Ghofrani, A hybrid short-term load
forecasting with a new data preprocessing framework, Electr. Power Syst. Res.
119 (2015) 138148.
[41] C. Guan, P.B. Luh, L.D. Michel, Y. Wang, P.B. Friedland, Very short-term load
forecasting: wavelet neural networks with data pre-ltering, IEEE Trans.
Power Syst. 28 (1) (2013) 3041.
[42] RTE, Transmission System Operator of France, Website: www.rte-france.com.
[43] AEMO, Australian Energy Market Operator, Website: www.aemo.com.au.
[44] D. Fay, J.V. Ringwood, M. Condon, M. Kelly, 24-h electrical load data sequential
or partitioned time series? Neurocomputing 55 (34) (2003) 469498.
[45] J.W. Taylor, Short-term electricity demand forecasting using double seasonal
exponential smoothing, J. Oper. Res. Soc. 54 (2003) 799805.
[46] D. Niu, H. Shi, D.D. Wu, Short-term load forecasting using Bayesian neural
networks learned by Hybrid Monte Carlo algorithm, Appl. Soft Comput. 12 (6)
(2012) 18221827.
[47] X. Lu, J. Wang, Y. Cai, J. Zhao, Distributed HS-ARTMAP and its forecasting
model for electricity load, Appl. Soft Comput. 32 (2015) 1322.
[48] A. Laoua, M. Mordjaoui, D. Dib, One-hour ahead electric load and wind-solar
power generation forecasting using articial neural network, in: 2015 6th
International Renewable Energy Congress (IREC), Sousse, Tunisia, 2426
March 2015, 2015, pp. 16.
[49] J.S.R. Jang, ANFIS: adaptive network based fuzzy inference system, IEEE Trans.
Syst. Man Cybern. 23 (3) (1993) 665685.
[50] F.R. Hampel, The inuence curve and its role in robust estimation, J. Am. Stat.
Assoc. 69 (346) (1974) 383393.
[51] R.K. Pearson, Outliers in process modeling and identication, IEEE Trans.
Control Syst. Technol. 10 (1) (2002).