Escolar Documentos
Profissional Documentos
Cultura Documentos
a r t i c l e i n f o a b s t r a c t
Article history: This paper describes daily peak load forecasting using an adaptive hybrid two-stage methodology.
Received 14 July 2015 Because the time series of electricity consumption is mainly influenced by seasonal effects, the double
Received in revised form 3 October 2015 seasonal HoltWinters exponential smoothing method is firstly used for next-day peak electricity
Accepted 17 November 2015
demand forecasting. In the second stage, the secondary forecasting model is applied taking into account
the benefits of Fuzzy c-means clustering; K-nearest neighbors algorithm; Wavelet packet decomposition;
and Adaptive Neuro-Fuzzy Inference System, for further improvement in forecasting accuracy. The whole
Keywords:
architecture of the proposed model will be presented and the results will be compared with neural net-
Daily peak load forecasting
Exponential smoothing
works and stand-alone adaptive neuro-fuzzy inference system based approaches by using a gathered
Data classification data from the Algerian power system. The results show that: (1) the proposed methodology is the best
Wavelet packet among all the considered schemes, (2) the FKW-ANFIS has satisfactory performance in both normal
Adaptive neuro-fuzzy inference system and special daily conditions.
2015 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijepes.2015.11.046
0142-0615/ 2015 Elsevier Ltd. All rights reserved.
A. Laouafi et al. / Electrical Power and Energy Systems 77 (2016) 136144 137
[27]. Although they are accurate in normal days, univariate con- (workdays and weekends). Hence, it seems natural to try to include
ventional techniques such as the seasonal autoregressive inte- both seasonal cycles in a HWT forecasting model. This is the
grated moving average (SARIMA) model cannot give satisfactory approach taken by J.W. Taylor as part of their modeling of half-
results when subject to special daily conditions, because they hourly England and Wales electricity consumption for a lead time
ignore important weather effects and cannot be updated for facing up to a one day ahead [33]. This method compete well in normal
the change in load demand during holidays. Even though multi- days, but it may faces some difficulty when subject to special daily
variate classical methods such as multiple regression models conditions such the case of the rapid change in load demand during
express the peak load as a function of exogenous inputs, the linear holidays. The whole architecture of the proposed methodology for
modeling process faces difficulty to determine empirically the cor- forecasting the peak load in both normal and special daily condi-
rect complex relationship that exists between the peak load and tions will be presented in the next section.
the other explanatory inputs. In recent years, the advent of new To evaluate the performance of the proposed strategy, the elec-
intelligent techniques such as neural networks and adaptive tricity peak load data for four different months: January 2014;
neuro-fuzzy inference system (ANFIS) enabled to model automati- April 2014; July 2014; and November 2014, and for the public hol-
cally complex nonlinear inputoutput relationships through learn- idays of 2014 were used as an illustrative example.
ing process using a database of load and independent variables.
Current trends in research in the field of load forecasting are
Proposed daily peak load forecasting methodology
focused on the use of hybrid and combined methods [2831].
The main objective of this paper is to present a novel model for
The structure of the proposed DPLF method consists of two
performing the DPLF. The proposed methodology can be viewed as
principal stages. Through the use of morning and evening peak
a two-stage forecasting model. Since the time series of peak elec-
electricity demand records, HWT method is applied as a first part
tricity demand is mainly influenced by seasonal effects (daily and
at the day J L1 for ensuring a primary forecasted peak load for
weekly cycles), the double seasonal HoltWintersTaylor tech-
the day J L . At its turn, the second part of our model consists of five
nique (HWT) is firstly used to perform a primary forecasted load.
basic parts. First, the temperature observations are used as inputs
In the second part, the primary forecasted peak load is verified
to the Fuzzy c-means clustering method for classifying the peak
and improved by a secondary forecasting model; called FKW-
load data into two classes (load of a hot or cold period). Then,
ANFIS, which uses the benefits of Fuzzy c-means (FCM) clustering,
the morning peak loads; the evening peak electricity demand
k-Nearest neighbors (k-NN) algorithm, Wavelet packet decomposi-
information; and the primary forecast of HWT, are used as inputs
tion (WPD), and Adaptive neuro-fuzzy inference system. Our idea
for k-NN to distinguish whether the day J L is workday or weekend
for this structure is to take the output of HWT model as an entry
and so that to select closest samples that have the similar charac-
to the secondary forecasting model. FCM and k-NN are used for
teristic of the day J L . The third part of the secondary forecasting
classifying data into somewhat similar patterns, which lead to
method decomposes the clustered morning and evening peak load
noise reduction and to an improvement on the quality of the
data; via wavelet packet method, into approximation parts associ-
amount of data should be taken by the Wavelet packet-Adaptive
ated with low frequencies and detailed parts associated with high
neuro-fuzzy inference system models; and therefore, to a higher
frequencies. At the fourth part, the ANFIS is introduced to predict
accuracy. To illustrate the suitability of the proposed methodology,
the future patterns of each wavelet packet space. Finally, the eve-
the approach is implemented to a peak load data from the Algerian
ning peak load forecasting values of all the spaces are added up to
power system.
produce the final electricity demand forecasting result.
The rest of the paper is organized as follows. Section The Alge-
rian peak electricity demand time series presents the Algerian
peak electricity demand time series. Section Proposed daily peak HoltWintersTaylor method
load forecasting methodology describes the proposed estimation
method. Section Results and discussion provides and explains HoltWintersTaylor technique is an extension to the standard
forecasting results. Finally, Section Conclusion concludes the HoltWinters exponential smoothing, which allows accommodat-
paper. ing the intraday and intraweek seasonal cycles in the electricity
demand series. This is the approach taken by J.W. Taylor as part
of their modeling of half-hourly England and Wales electricity con-
The Algerian peak electricity demand time series sumption for a lead time up to a one day ahead [33]. However, the
present paper is the first in term of using HWT method for fore-
In this study, we consider the power peak load data that consist casting daily peak electricity demand. The formulation for HWT
of thirty-five months of observations for electricity demand in method is given in the following expressions:
Algeria from 01 January 2012 to 30 November 2014. These data
were collected from OSE (Operateur du Systeme Electrique) web- St ayt =Dts1 W ts2 1 aSt1 T t1 1
site [32]. The top subplot of Fig. 1 shows that the used data consist T t cSt St1 1 cT t1 2
of two time series. The first series records the morning peak elec- Dt dyt =St W ts2 1 dDts1 3
tricity demand, while the second series consists of the evening
W t xyt =St Dts1 1 xW ts2 4
peak load observations. The figure illustrates that the daily peak
load usually occurs during the evening period. Therefore, we lim- b k
y t k St kT t Dts1k W ts2k / yt St1 T t1 Dts1 W ts2
ited our interest in the present work to the evening peaks predic- 5
tion. The second subplot of Fig. 1 presents box-plots for the
evening electricity demand data. One can perceive an intraweek St and T t , are the smoothed level and trend; Dt and W t are respec-
seasonal cycle, where weekdays show similar patterns of demand, tively the seasonal indices for the intraday s1 (s1 2) and intra-
but Fridays are somewhat different and the load on which is rela- week s2 (s2 14) seasonal cycles; a; c; d and x are the smoothing
tively low compared to in workdays. parameters; yt is the actual value of the time series in period t;
By considering both two series cited above, the data will con- and by t k is the k step-ahead forecast made from forecast origin t.
tain then daily seasonal cycle of duration two periods (morning The term involving the parameter / in the forecast function expres-
and evening peak), and weekly cycle of duration fourteen periods sion Eq. (5) is a simple adjustment for first-order autocorrelation.
138 A. Laouafi et al. / Electrical Power and Energy Systems 77 (2016) 136144
In this paper, HWT method parameters are estimated by by Dunn in 1973 [35] and improved by Bezdek in 1981 [36] and it
minimizing the one-day ahead evening peak electricity demand is frequently used in pattern recognition. The FCM algorithm
forecasting error in three principal stages. First we generated attempts to partition a finite collection of n elements
2 103 vectors of parameters from a uniform random number gen- X x1 ; . . . ; xn into a collection of c fuzzy clusters with respect to
erator between 0 and 1. For each of the vectors, we then evaluated some given criterion. Given a finite set of temperature data, the
the absolute percentage error (APE) on the daily peak demand fore- algorithm returns a list of two cluster centers C fc1 ; c2 g and a
casting of 24 June 2013. The 200 vectors producing the lowest APE partition matrix, W wi;j 2 0; 1; i 1; . . . ; n; j 1; 2 where each
are evaluated; in their turns, over the period from 24 June 2013 to element wi;j indicates the degree to which element xi belongs to
30 June 2013. Among the 200 vectors, the 10 vectors that have the cluster cj . The FCM aims to minimize an objective function:
lowest mean absolute percentage error (MAPE) are evaluated at
X
n X
c 2
last on the one-day ahead evening peak load forecasting over the Jm
ij xi c j
wm 6
period from 01 June 2013 to 30 June 2013. The one producing i j
the lowest MAPE value was chosen as the final parameter
vector. The optimized values of HWT method parameters were: where, m is any real number greater than 1. The fuzzifier m deter-
a 0:189, c = 0.049, d 0:304; x 0:112, and / 0:660. mines the level of cluster fuzziness. wi;j is the degree of membership
of xi belonging to cluster j:
Fuzzy c-means clustering
1
wi;j 2 7
Weather conditions have a major impact on electricity con- Pc kxi cj k m1
sumption, especially for domestic and agricultural consumers. k1 kxi ck k
objects y to determine its k-nearest-neighbor list on the training ber 2014. This figure shows that data classifiers not only allow to
set of n elements: reduce the dimension of the problem, but also to improve the
quality of the amount of data should be taken by the Wavelet
Di maxi jx yi j; i 1; . . . ; n 8 packet-Adaptive neuro-fuzzy inference system models.
Once the distance between the query-instance and all the train-
ing samples is calculated, the distances are then sorted to deter- Wavelet packet decomposition
mine nearest-neighbor list based on the k-th minimum distance.
In the k-NN method proposed in this paper, each individual is The wavelet transform (WT) is powerful and flexible tool that
defined by the morning and the evening peak load values corre- has been widely used in electric load forecasting for decomposing
sponding to a particular day J i . Thus, the k-NN classifier tries to find power load series into constitutive series that can be predicted
pairs which are similar to the actual morning peak and the primary more accurately than the original signal [3739]. Wavelet trans-
forecasted evening peak load of the day J L . Consequently, the algo- form is able to reveal breakdown points, discontinuities in higher
rithm allows the identification of the load patterns, and by the way derivatives and self-similarity aspects of data [37]. WT can be
to generally distinguish whether the day J L is a weekday or a week- divided in two categories: continuous wavelet transform (CWT),
end. An example of the k-Nearest neighbors algorithm results is and discrete wavelet transform (DWT).
presented in Fig. 3. The point to be classified, shown with X; is In continuous wavelet transform, each wavelet is created by
related to the peaks electricity demand information for 30 Novem- scaling and translating operations in a mother wavelet. The mother
140 A. Laouafi et al. / Electrical Power and Energy Systems 77 (2016) 136144
Fig. 5. WPD results for forecasting the peak load of 30 November 2014.
wavelet is an oscillate function with finite energy and zero average. and positions usually based on powers of two [40]. In multiresolu-
The CWT of a continuous time signal f t is defined as: tion analysis, the signal is passed through high-pass and low-pass
Z 1 filters. The high-pass filter gives the detailed resolution of the
CWT a; b f twa;b tdt 9 signal, while low-pass filter gives the approximate information of
1 the signal. An approximation (A) holds the general trend of a given
where w is the mother wavelet, a 2 R is scaling parameter, b 2 R is signal, whereas a detail (D) depicts high-frequency components of
translating parameter, and: it.
By setting a ac0 ; b db0 ac0 ; with a and b 2 Z, the DWT of the
1 tb sampled signal f t of length T is given by:
wa;b t p w 10
a a X
T1
However, the CWT is difficult to put into practice and the infor-
DWT c; d f twc;d t 11
t0
mation that has been picked up may overlap and result in redun-
dancy. It is therefore evident that a discretization of the where:
transform should be considered if it is desired to obtain a non-
1 t db0 ac0
redundant transformation. Therefore, instead of doing that, the wc;d t pc w 12
a0 ac0
mother wavelet can be scaled and translated using certain scales
A. Laouafi et al. / Electrical Power and Energy Systems 77 (2016) 136144 141
Table 1
The key elements, parameters, and inputs for the different models.
Wavelet packet analysis is an extension of wavelet analysis that the original signal S from its wavelet coefficients. In this paper, the
offers better frequency resolution and richer signal analysis. The following nodes from the three level wavelet packet tree are
process of WPD by filters is similar to the DWT. The difference selected: fAAA3 ; DAA3 ; ADA3 ; DDA3 ; AD2 ; DD2 g. The used mother
between wavelet packet decomposition and wavelet decomposi- wavelet function is the Daubechies of order 2 (abbreviated as
tion is that only low-frequency components are divided into two Db2) wavelet function. An example for wavelet packet decomposi-
parts continuously in multiresolution analysis, while both the tion results is shown in Fig. 5, where we have presented the
detail and approximation coefficients are decomposed in WPD. obtained wavelet sub-series from the forecasting process of 30
This attribute makes the WPD very suitable in our case studies November 2014.
for peak electricity demand forecasting, since relevant information
may have located in higher frequency components of the load data. Adaptive neuro-fuzzy inference system
For a better understanding of wavelet packet decomposition
process, a third-order WPD is shown in Fig. 4. In wavelet packet The adaptive neuro-fuzzy inference system is one of the most
decomposition, the original signal S is firstly decomposed into successful schemes which combine the benefits of artificial neural
detail (D1 ) and approximation (A1 ) coefficients. In the second level, network and fuzzy inference system (FIS) paradigms into a single
not only the approximation part is decomposed into new detail capsule [41]. An ANFIS works by applying neural learning rules
(DA2 ) and approximation (AA2 ), but detail coefficient vector (D1 ) to identify and tune the parameters and structure of a Fuzzy Tak-
is also decomposed into two parts (DD2 ; AD2 ) using the same agiSugeno Inference System. The network is a multi-layer feed-
approach as in approximation vector splitting. This process is forward that consists of a collection of artificial neurons in which
repeated in a similar manner at the third level of the wavelet each node (neuron) performs a particular function on incoming
packet tree. However, there is not a unique manner to reconstruct signals. The first layer executes the fuzzification process. The
Fig. 7. Results of FKW-ANFIS for four different months: January, April, July, and November 2014.
Table 2
The relative error (MAPE) of all proposed methods.
Table 3
Obtained results when HWT and FKW-ANFIS are used to forecast the peak load in 2014 public holidays.
temperature, and morning peak load of the day J L are used as ANFIS for the public holidays of 2014. The Table shows that MAPE
inputs for S-BPNN, S-NARX, and S-ANFIS models. The architecture value of HWT and FKW-ANFIS is respectively 3.283% and 2.796%,
of FKW-BPNN and FKW-NARX is very similar to that of meaning that FKW-ANFIS model has a 14.834% higher accuracy
FKW-ANFIS, where only the morning peak of the day on which over HWT. Therefore, the proposed FKW-ANFIS model in this paper
we would like to predict the load; is used as input to the neural is very suitable tool for forecasting the daily peak electricity
networks. The key elements and parameters of these approaches demand in both normal and special daily conditions.
are indicated on Table 1, and schematized in Fig. 6. The blanks in
Table 1 mean that the input or the parameter concerned is not
Conclusion
included for a specific forecasting model, while designates
the inclusion of the input/parameter by the forecasting model.
This paper has addressed the problem of daily peak load fore-
For the evaluation of the methods considered in this study, we
casting using an adaptive hybrid two-stage structure, which has
used APE, and MAPE criteria:
obtained better forecast results in comparison with double sea-
jb
y i yi j sonal HoltWinters, neural networks, and stand-alone adaptive
APE :100 13
yi neuro-fuzzy inference system based approaches. The obtained
1X n
jby i yi j results and forecasting performance showed that: (1) the average
MAPE :100 14 MAPE value of HWT primary forecaster and FKW-ANFIS secondary
n i1 yi
forecaster in normal days is 1.879% and 1.641%, respectively, and
where b y i is the forecasted peak demand of the day i; yi is the real (2) the average MAPE value of HWT and FKW-ANFIS in public hol-
load at the day i. n is the total number of forecasting cases. idays is 3.283% is 2.796%, respectively. Therefore, the proposed
Fig. 7 shows the results of FKW-ANFIS method for evening peak two-stage methodology in this paper is an efficient tool for improv-
demand prediction along four different months: January 2014, ing the peak load forecasting accuracy in both normal and special
April 2014, July 2014, and November 2014. This figure illustrates days.
that the proposed FKW-ANFIS model has successfully predict the
daily peak demand, since the shape of the predicted peak curves References
follows almost the shape of the actual peak curves. Hence, detailed
forecasting results; as indicated on Table 2; clearly demonstrate [1] Abdel-Aal RE. Modeling and forecasting electric daily peak loads using
abductive networks. Int J Electr Power Energy Syst 2006;28(2):13341.
the effectiveness of the proposed forecasting strategy and show [2] Hagan MT, Behr SM. The Time Series Approach to Short Term Load Forecasting.
that FKW-ANFIS outperformed all others methods. The results IEEE Trans Power Syst 1987;2(3).
from MAPE illustrate that HWT method is the more accurate [3] Papalexopoulos AD, Hesterberg TC. A regression-based approach to short-term
system load forecasting. IEEE Trans Power Syst 1990;5(4).
among the primary forecasting models, while the best technique
[4] Dillon TS, Sestito S, Leung S. Short term load forecasting using an adaptive
among the secondary forecasting models is the proposed FKW- neural network. Int J Electr Power Energy Syst 1991;13(4):18692.
ANFIS. They also show that there is a notable improvement in fore- [5] Yao SJ, Song YH, Zhang LZ, Cheng XY. Wavelet transform and neural networks
cast accuracy when using ANFIS secondary forecasting model for short-term electrical load forecasting. Energy Convers Management
2000;41(18):197588.
rather than the double seasonal HWT technique, and that there is [6] Mandal P, Senjyu T, Urasaki N, Funabashi T. A neural network based several-
a further noticeable improvement achieved by using the FKW- hour-ahead electric load forecasting using similar days approach. Int J Electr
ANFIS model. From these results, we can notice that average MAPE Power Energy Syst 2006;28(6):36773.
[7] Taylor JW. Short-Term Load Forecasting With Exponentially Weighted
of FKW-ANFIS has 08.838%, and 12.678%, higher accuracy over Methods. IEEE Trans Power Syst 2012;27:45864.
S-ANFIS, and HWT, respectively. The most important improvement [8] Li P, Li Y, Xiong Q, Chai Y, Zhang Y. Application of a hybrid quantized Elman
was in summer, with a MAPE value lower respectively by 12.958%, neural network in short-term load forecasting. Int J Electr Power Energy Syst
2014;55:74959.
and 21.418%. This is a very important point, since we know that the [9] Barakati SM, Gharaveisi AA, Rafiei SMR. Short-term load forecasting using
electricity consumption reaches its highest levels in summer as a mixed lazy learning method. Turkish J Electr Eng Comput Sci 2015;23(1):
result of huge utilization of air conditioning appliances. In the rest 20111.
[10] Zhai MY. A new method for short-term load forecasting based on fractal
of this section, we are interested in further comparison between interpretation and wavelet analysis. Int J Electr Power Energy Syst 2015;69:
HWT and FKW-ANFIS models. 2415.
We present in Fig. 8 the MAPE results of the four months dis- [11] Charytoniuk W, Chen MS. Very Short-Term Load Forecasting Using Artificial
Neural Networks. IEEE Trans Power Syst 2000;15:2638.
cussed above; consolidated according to the type of day. This figure
[12] Laouafi A, Mordjaoui M, Dib D. Very short-term electricity demand forecasting
shows that HWT gives the minimum MAPE values for Fridays and using adaptive exponential smoothing methods. In: 15th International
Saturdays, and FKW-ANFIS gives the minimum MAPE values for conference on sciences and techniques of automatic control & computer
workdays. For Mondays, the MAPE value obtained by FKW-ANFIS engineering STA2014, Hammamet, Tunisia; 2123 December 2014. p. 553
7.
is somewhat around the half of the obtained value by HWT. We [13] Laouafi A, Mordjaoui M, Dib D. One-hour ahead electric load forecasting using
present in Table 3; the forecasting results of HWT and FKW- neuro-fuzzy system in a parallel approach. Computational intelligence
144 A. Laouafi et al. / Electrical Power and Energy Systems 77 (2016) 136144
applications in modeling and control. Studies in computational intelligence, [27] Son SY, Lee SH, Chung K, Lim JS. Feature selection for daily peak load
vol. 575; 2015. p. 95121. forecasting using a neuro-fuzzy system. Multimedia Tools Appl; March 2014.
[14] Gonzalez-Romera E, Jaramillo-Moran MA, Carmona-Fernandez D. Monthly [28] Saini LM, Soni MK. Artificial neural network-based peak load forecasting using
electric energy demand forecasting based on trend extraction. IEEE Trans conjugate gradient methods. IEEE Trans Power Syst 2002;17(3):90712.
Power Syst 2006;21(4):194653. [29] Fan S, Mao C, Chen L. Peak load forecasting using the self-organizing map. In:
[15] Jaramillo-Morn MA, Gonzalez-Romera E, Carmona-Fernndez D. Monthly Advances in neural networks. Lecture notes in computer science, vol. 3498;
electric demand forecasting with neural filters. Int J Electr Power Energy 2005. p. 6407.
Syst 2013;49:253263, 201. [30] Khan GM, Khan S, Ullah F. Short-term daily peak load forecasting using fast
[16] Lee WJ, Hong J. A hybrid dynamic and fuzzy time series model for mid-term learning neural network. In: 11th International conference on intelligent
power load forecasting. Int J Electr Power Energy Syst 2015;64:105762. systems design and applications (ISDA), Cordoba; 2224 November 2011.
[17] Laouafi A, Mordjaoui M, Laouafi F. An evaluation of conventional and [31] Moazzami M, Khodabakhshian A, Hooshmand R. A new hybrid day-ahead peak
computational intelligence methods for medium and long-term load load forecasting method for Irans National Grid. Appl Energy 2013;101:
forecasting in Algeria. In: 3rd International conference on control, 489501.
engineering & information technology CEIT2015, Tlemcen, Algeria; 2527 [32] Oprateur du systme electrique, subsidiary of Sonelgaz Group. <http://www.
May 2015. ose.dz/courbes.php>.
[18] Zhang Z, Ye S. Long term load forecasting and recommendations for China [33] Taylor JW. Short-term electricity demand forecasting using double seasonal
based on support vector regression. In: 2011 International conference on exponential smoothing. J Oper Res Soc 2003;54:799805.
information management, innovation management and industrial engineering [34] http://www.wunderground.com/history/airport/DAAG.
(ICIII); 2627 November 2011. p. 597602. [35] Dunn JC. A fuzzy relative of the ISODATA process and its use in detecting
[19] Kandil MS, El-Debeiky SM, Hasanien NE. Long-term load forecasting for fast compact well-separated clusters. J Cybernet 1973;3:3257.
developing utility using a knowledge-based expert system. IEEE Trans Power [36] Bezdek JC. Pattern recognition with fuzzy objective function algoritms. New
Syst 2002;17(2):4916. York: Plenum Press; 1981.
[20] Chikobvu D, Sigauke C. Regression-SARIMA modelling of daily peak electricity [37] Mordjaoui M, Boudjema B, Bouabaz M. A hybrid wavelet transform and ANFIS
demand in South Africa. J Energy Southern Africa 2012;23(3):2330. model for short term electric load prediction. In: 2012 2nd International
[21] Amjady N. Short-term hourly load forecasting using time-series modeling with conference on advances in computational tools for engineering applications
peak load estimation capability. IEEE Trans Power Syst 2001;16(4):798805. (ACTEA), Beirut, Liban; 1215 December 2012. p. 2925.
[22] Sigauke C, Chikobvu D. Daily peak electricity load forecasting in South Africa [38] Guan C, Luh PB, Michel LD, Wang Y, Friedland PB. Very short-term load
using a multivariate non-parametric regression approach. ORiON: J ORSSA forecasting: wavelet neural networks with data pre-filtering. IEEE Trans Power
2010;26(2):97111. Syst 2013;28(1):3041.
[23] Jin L, Lai YJ, Long TX. Peak load forecasting based on robust regression model. [39] Sudheer G, Suseelatha A. Short term load forecasting using wavelet transform
In: 2004 International conference on probabilistic methods applied to power combined with HoltWinters and weighted nearest neighbor models. Int J
systems; 2004. p. 1238. Electr Power Energy Syst 2015;64:3406.
[24] Amin-Naseri MR, Soroush AR. Combined use of unsupervised and supervised [40] Catalao JPS, Pousinho HMI, Mendes VMF. Hybrid intelligent approach for
learning for daily peak load forecasting. Energy Convers Management 2008;49 short-term wind power forecasting in Portugal. IET Renew Power Gener
(6):13028. 2011;5(3):2517.
[25] Iizaka T, Matsui T, Fukuyama Y. A novel daily peak load forecasting method [41] Jang JSR. ANFIS: adaptive network based fuzzy inference system. IEEE Trans
using analyzable structured neural network, 2002: Asia Pacific. In: IEEE/PES Syst Man Cybernet 1993;23(3):66585.
transmission and distribution conference and exhibition 1; October 2002. p. [42] Laouafi A, Mordjaoui M, Dib D. One-hour ahead electric load and wind-solar
3949. power generation forecasting using artificial neural network. In: 2015 6th
[26] Kiartzis SJ, Bakirtzis AG, Theocharis JB, Tsagas G. A fuzzy expert system for International renewable energy congress (IREC), Sousse, Tunisia; 2426 March
peak load forecasting. Application to the Greek power system. In: MELECON 2015. p. 16.
2000: 10th mediterranean electrotechnical conference 3; 2931 May 2000. p.
1097100.