See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267226399

Short-term load forecasting using ANN technique

Article in International Journal of Engineering Sciences and Emerging Technologies · February 2012
DOI: 10.7323/ijeset/v1_i2_12

CITATIONS READS

81 4,529

2 authors, including:

Samsher kadir sheikh

Padmashri Dr. Vithalrao Vikhe Patil College of Engineering
2 PUBLICATIONS 81 CITATIONS

SEE PROFILE

All content following this page was uploaded by Samsher kadir sheikh on 27 May 2015.

The user has requested enhancement of the downloaded file.

 International Journal of Engineering Sciences & Emerging Technologies, Feb 2012.
ISSN: 2231 – 6604 Volume 1, Issue 2, pp: 97-107 ©IJESET

SHORT-TERM LOAD FORECASTING USING ANN TECHNIQUE

Samsher Kadir Sheikh1, M. G. Unde2

1
Assistant Professor & 2Professor, Electrical Dept PDVVP COE, Ahmednagar, India

A BSTRACT
Load forecasting is the technique for prediction of electrical load. In a deregulated market it is much need for a
generating company to know about the market load demand for generating near to accurate power. If the
generation is not sufficient to fulfill the demand, there would be problem of irregular supply and in case of
excess generation the generating company will have to bear the loss. Neural network techniques have been
recently suggested for short-term load forecasting by a large number of researchers. This work studies the
applicability of this kind of models. The work is intended to be a basis for a real forecasting application .First, a
literature survey was conducted on the subject. Most of the reported models are based on the so-called Multi-
Layer Perceptron (MLP) network. There are numerous model suggestions, but the large variation and lack of
comparisons make it difficult to directly apply proposed methods. It was concluded that a comparative study of
different model types seems necessary. Several models were developed and tested on the real load data of a
Finnish electric utility. Most of them use a MLP network to identify the assumed relation. We carried out short-
term load forecasting for P.D.V.V.P.COE, Ahmednagar college campus using ANN (Artificial Neural Network)
technique ANN was implemented on MATLAB-10. MLP (Multi-layer Perceptions was made with input as days
and hourly load. Hourly load means the hourly power consumption in college. Error was calculated as MAPE
(Mean Absolute Percentage Error) and with error of about 0.956% this paper was successfully carried out. This
paper can be implemented by any intensive power consuming company/ college for predicting the future load
and would proved to be very useful tool while sanctioning the load

K EYW ORD
Load forecasting, Artificial Neural Network, MLP, MAPE.

I. INTRODUCTION
The most used thing in today’s world is energy. We use energy in various forms in our day to day life,
electricity, refined oils, LPG, solar energy, wind energy, chemical energies in form of batteries and
many other forms. Sometimes we are extravagant and sometimes we are careful. But to provide users
uninterrupted supply of electricity there must be proper evaluation of present day and future demand
of power. That’s why we need a technique to tell us about the demand of consumers and the exact
capability to generate the power and this need LOAD FORECASTING technique. It is used by power
companies to anticipate the amount of power needed to supply the demand. It tells about the scenario
of present and future load demand. It has many applications including energy purchasing and
generation, load switching, contract evaluation, and infrastructure development. Load forecasting has
become in recent years one of the major areas of research in electrical engineering. Load forecasting
is however a difficult task. First, because the load series is complex and exhibits several levels of
seasonality. Second, the load at a given hour is dependent not only on the load at the previous day, but
also on the load at the same hour on the previous day and previous week, and because there are many
important exogenous variables that must be considered [1]
1.1 Necessity of LOAD FORECASTING
Load forecasting is vitally important for the electric industry in the deregulated economy. It has many
applications including energy purchasing and generation, load switching, contract evaluation, and

97
 International Journal of Engineering Sciences & Emerging Technologies, Feb 2012.
ISSN: 2231 – 6604 Volume 1, Issue 2, pp: 97-107 ©IJESET
infrastructure development. A large variety of mathematical methods have been developed for load
forecasting. Accurate models for electric power load forecasting are essential to the operation and
planning of a utility company. Load forecasting helps an electric utility to make important decisions
including decisions on purchasing and generating electric power, load switching, and infrastructure
development. Load forecasts are extremely important for energy suppliers, ISOs, .national
institutions, and other participants in electric energy generation, transmission, distribution, and
markets.
1.2 Types of Load Forecasting Techniques Involved
Load forecasts can be divided into three categories:
[i] Short-term forecasts: This is usually from one hour to one week.

[ii]Medium forecasts : This is usually from a week to a year.

[iii]Long-term forecasts: This is longer than a year.

The forecasts for different time horizons are important for different operations within a utility
company. The natures of these forecasts are different as well. For example, for a particular region, it is
possible to predict the next day load with an accuracy of approximately 1-3%. However, it is
impossible to predict the next year peak load with the similar accuracy since accurate long-term
weather forecasts are not available. For the next year peak forecast, it is possible to provide the
probability distribution of the load based on historical weather observations. It is also possible,
according to the industry practice, to predict the so-called weather normalized load, which would take
place for average annual peak weather conditions or worse than average peak weather conditions for a
given area. Weather normalized load is the load calculated for the so-called normal weather
conditions which are the average of the weather characteristics for the peak historical loads over a
certain period of time. The duration of this period varies from one utility to another. Most companies
take the last 25-30 years of data. Load forecasting has always been important for planning and
operational decision. However, with the deregulation of the energy industries, load forecasting is even
more important. With supply and demand fluctuating and the changes of weather conditions and
energy prices increasing by a factor of ten or more during peak situations, load forecasting is vitally
important for utilities. Short-term load forecasting can help to estimate load flows and to make
decisions that can prevent overloading. Timely implementations of such decisions lead to the
improvement of network reliability and to the reduced occurrences of equipment failures and
blackouts. Load forecasting is also important for contract evaluations and evaluations of various
sophisticated financial products on energy pricing offered by the market. In the deregulated economy,
decisions on capital expenditures based on long-term forecasting are also more important than in a
non-deregulated economy when rate increases could be justified by capital expenditure projects.
1.3 Important Factors for Forecasts
Consideration of various factors is the prerequisite for accurate forecasting of load. Time factor, past
weather data, class of consumers, load demanded by the region in past, growth of the region, amount
of load increased etc, these are the factors which play pivotal role in calculating the demand load.
For Short-term load forecasting several factors should be considered, such as
[i] Time factors,
[ii] Weather data, and
[iii] Possible customers’ classes.
The medium- and long-term forecasts take into account
[i] The historical load
[ii]Weather data,
[iii]The number of customers in different categories,
[iv]The appliances in the area and their characteristics including age,
[v] The economic and demographic data and their forecasts,
[vi]The appliance sales data and other factors.

98
 International Journal of Engineering Sciences & Emerging Technologies, Feb 2012.
ISSN: 2231 – 6604 Volume 1, Issue 2, pp: 97-107 ©IJESET
The time factors include the time of the year, the day of the week, and the hour of the day. There are
important differences in load between weekdays and weekends. The load on different weekdays also
can behave differently. For example, Mondays and Fridays being adjacent to weekends, may have
structurally different loads than Tuesday through Thursday. This is particularly true during the
summer time. Holidays are more difficult to forecast than non-holidays because of their relative
infrequent occurrence. Weather conditions influence the load. In fact, forecasted weather parameters
are the most important factors in short-term load forecasts.[2]

II. ARTIFICIAL NEURAL NETWORK

ANN is usually formed from many hundreds or thousands of simple processing units, connected in
parallel and feeding forward in several layers. Because of the fast and inexpensive personal computers
availability, the interest in ANN’s has blossomed in the current digital world. The basic motive of the
development of the ANN is to make the computers do what a human being cannot do
2.1 Benefits of ANN
1. They are extremely powerful computational devices.
2. Massive parallelism makes them very efficient
3. They can learn and generalize from training data – so there is no need for enormous feats of
programming.
4. They are particularly fault tolerant – this is equivalent to the “graceful degradation” found in
biological systems.
5. They are very noise tolerant – so they can cope with situations where normal symbolic systems
would have difficulty.
6. In principle, they can do anything a symbolic/logic system can do, and more.
2.2 Working of Brain Replica of Artificial Neural Network
ANN is relatively crude electronic models based on the neural structure of the human brain which
basically learns from experience .Learning in biological systems involves adjustments to the synaptic
connections that exist between the neurons .The basic building block of all biological brains is a
nerve cell, or a neuron. A basic component of a neuron shows the various inputs to the network. It is
represented by the mathematical symbol, x (n). Each of these inputs is multiplied by a connection
weight. These weights are represented by w (n)

Fig :- 1Block diagram of Human Nervous system

Fig :-2 working principle of ANN system

Above two figures shows similarity in the working principle of brain and ANN system.

99
 International Journal of Engineering Sciences & Emerging Technologies, Feb 2012.
ISSN: 2231 – 6604 Volume 1, Issue 2, pp: 97-107 ©IJESET
2.3 Multi-Layer Perception (MLP)
Multi-Layer Perceptron (MLP) network is the most popular neural network type and most of the
reported neural network STLF models are based on it .The MLP network consists of several layers of
neurons. Each neuron in a certain layer is connected to each neuron of the next layer. There are no
feedback connections.
2.4 Identification of Input Data
This technique uses 2 variables:
[i] The historical load demand and
[ii] Time for 24 hours.
The network learned through historical data to give the network input signals and desired outputs. To
each input signal the network produced an output signal, and the learning aimed at minimizing the
sum of squares of the differences between desired and actual outputs. The learning was carried out by
repeatedly feeding the input-output patterns to the network. One complete presentation of the entire
training set was called an epoch . The learning process was usually performed on an epoch-by-epoch
basis until the weights stabilized and the sum of squared errors converged to some minimum value or
goal .The other parameters were the learning rate and momentum factor . Both parameters
were system dependent. The momentum factor caused the weight changes to be dependent on more
than one input pattern. The useful range for this parameter was between 0 and 1. [3]

III. ACTUAL IDENTIFICATION OF INPUT DATA

This paper used two variables:
[i] The historical load demand, and
[ii] Time for 24 hours.
Historical load demand data of our college was used as a reference to get the profile of weekends.
Data was taken from January 17th 2011 until January 24th 2011. (1WEEK)
The selections of data were carried out according to the following steps in order to get good
predictions:-
i) Separate the data by day:
_ Data Friday
_ Data Saturday
_ Data Sunday
ii) Select the data without holiday
iii) Choose the best range of data to train
MATLAB can be trained by typing out the right profile or pattern for each day.
MATLAB 10 has been used by us carry out our work.
Based on the data we have plotted the graph shown below using Microsoft Excel 2007

Fig:-3 This graph shows the actual variation of load for 24 hrs of Friday, Saturday and Sunday

100
 International Journal of Engineering Sciences & Emerging Technologies, Feb 2012.
ISSN: 2231 – 6604 Volume 1, Issue 2, pp: 97-107 ©IJESET
Observation can be made that the load variation is mainly during 10am to 5pm i.e. working hour of
college. Friday being the full day has the max load throughout the day while Saturday is half and
Sunday is holiday.

3.1 Implementation of ANN using MATLAB.10

Fig 3.1 Open nftool in MATLAB.10 Fig;-3.2 Prepare input data and output data in
Workspace. Import the prepare data from workspace

Fig:-3,3 Training validation and testing percentage Fig :- 3.4 error Retrain
can be set according to the need. In this paper Training
was set to 70%, Validation was 15% and Testing was 15% [4]

If error is high then to minimize the error Retrain is done

Then we can get plots of performance, Training state, Fit and Regression

101
 International Journal of Engineering Sciences & Emerging Technologies, Feb 2012.
ISSN: 2231 – 6604 Volume 1, Issue 2, pp: 97-107 ©IJESET

Fig :- 3.5 Training state, Fit and Regression. Fig:-3.6 This shows the performance plot for Sunday
with 5 Epochs. This plot is Mean squared error vs
Epochs.

Best Validation performance is 24.8816 at epoch 1.

Fig :-3.7 This is the plot fit for output and targets Fig 3.8 regression plot for Sunday similarly, various
Vs input. Errors are shown in the graph. This error is plots can be found for Saturday and Friday
difference between actual data and predicted data.

The regression plot for Sunday is shown below.

102
 International Journal of Engineering Sciences & Emerging Technologies, Feb 2012.
ISSN: 2231 – 6604 Volume 1, Issue 2, pp: 97-107 ©IJESET

Fig 3.9:- Saturday Fig3.10:- For Saturday

Fig3.11 for Saturday The regression plot shown. Fig3.12 for friday

Fig 3.13for Friday Fig3.14 for Friday The regression plot shown

103
 International Journal of Engineering Sciences & Emerging Technologies, Feb 2012.
ISSN: 2231 – 6604 Volume 1, Issue 2, pp: 97-107 ©IJESET

IV. PROGRAM
Generated m-filefunction net = create_fit_net(inputs,targets)

%CREATE_FIT_NET Creates and trains a fitting neural network.

%
% NET = CREATE_FIT_NET(INPUTS,TARGETS) takes these arguments:
% INPUTS - RxQ matrix of Q R-element input samples
% TARGETS - SxQ matrix of Q S-element associated target samples
% arranged as columns, and returns these results:
% NET - The trained neural network
%
% For example, to solve the Simple Fit dataset problem with this function:
%
% load simplefit_dataset
% net = create_fit_net(simplefitInputs,simplefitTargets);
% simplefitOutputs = sim(net,simplefitInputs);
%
% To reproduce the results you obtained in NFTOOL:
%
% net = create_fit_net(input1,unnamed);

% Create Network
numHiddenNeurons = 20; % Adjust as desired
net = newfit(inputs,targets,numHiddenNeurons);
net.divideParam.trainRatio = 70/100; % Adjust as desired
net.divideParam.valRatio = 15/100; % Adjust as desired
net.divideParam.testRatio = 15/100; % Adjust as desired % Train and Apply Network
[net,tr] = train(net,inputs,targets);
outputs = sim(net,inputs);
% Plot
plotperf(tr)
plotfit(net,inputs,targets)
plotregression(targets,outputs [5]

4.1) COMPARISONS OF ACTUAL AND PREDICTED RESULTS [6] [7]

Fig4.1 Friday Fig4.2 Saturday

104
 International Journal of Engineering Sciences & Emerging Technologies, Feb 2012.
ISSN: 2231 – 6604 Volume 1, Issue 2, pp: 97-107 ©IJESET

Fig4.3 Sunday

TABULATION OF RESULTS & CALCULATION OF ERROR

Calculation of Error as
Hours Actual load(kWh) Predicted load(kWh)
1 10.4 14.3
2 10.7 10.7
3 10.26 3.7
4 12.3 10.7
5 13.6 10.3
6 14.1 11.8
7 18.2 10.2
8 17.5 10.6
9 22.5 10.4
10 25.3 10.6
11 24.1 20
12 25.3 25.2
13 26.9 23.3
14 28.4 16.1
15 28.4 17.2
16 28.5 28.2
17 29.6 19.1
18 29.3 18.9
19 28.3 17
20 29.9 25
21 31.3 29
22 33.7 24.9
23 31.6 25.7
24 32.5 20.1

V. CONCLUSION
We have carried out short- term load forecasting of our college and predicted the future load. Actual
data was the actual energy consumed .We have used MATLAB.10 for prediction of load. The result
of MLP network model used for one day ahead short term load forecast for the P.D.V.V.P. college of
engineering, Ahmednagar shows that MLP network has a good performance and reasonable
prediction accuracy was achieved for this model. Its forecasting reliabilities were evaluated by
computing the mean absolute percentage error between the actual and predicted values.
We were able to obtain a Mean Absolute Percentage Error (MAPE) of 1.066 % which represents the
high degree of accuracy. The results suggest that ANN model with the developed structure can
perform good prediction with least error and finally this neural network could be an important tool for
short term load forecasting.

105
 International Journal of Engineering Sciences & Emerging Technologies, Feb 2012.
ISSN: 2231 – 6604 Volume 1, Issue 2, pp: 97-107 ©IJESET
Error Calculation for Sunday,

MAPE= {[564.66-435] /564.66}*100/24 %

=0.956%
Similarly calculation of error for Friday and Saturday can be done.
Days MAPE(%)
Friday 1.279
Saturday 0.963
Sunday 0.956

VI. FUTURE SCOPE

Future studies on this work can incorporate additional information (such as customer class and season
of the year) into the network so as to obtain a more representative forecast of future load. Network
specialization (i.e. the use of one neural network for the peak periods of the day and another network
for the hours of day) can also be experimented upon.

REFERENCES
[1] “Load Forecasting” Chapter 12, E.A Feinberg and Dora Genethlio, Page 269 – 285, from links:
www.ams.sunysb.edu nd www.usda.gov
[2] Moghram, S. Rahman, “Analysis and Evaluation of Five Short-Term Load Forecasting Techniques,”
Proceedings of the IEEE Transaction on Power systems, pp. 1484-1491, Vol. 4,
[3] Application of Neural Network to Load Forecasting in Nigerian Electrical Power System by G.A. Adepoju,
M.Sc.1, S.O.A Ogunjuyigbe, M.Sc.2, and K.O. Alawode, B.Tech.1
[4] One Hour Ahead Load Forecasting Using Artificial Neural Network for the Western Area of Saudi Arabia
by A. J. Al-Shareef, E. A. ohamed, and E. Al-Judaibi
[5] Neural network toolbox for use with malab by Howard Demuth, Mark Beale, Martin Hogan
[6] Artificial Neural Network Approach short Term Load Forecasting for Illam Region by Mohsen Hayati, and
Yazdan Shirvany
[7] Ho, K.-L., Y.-Y. Hsu, C-C. Yang, 1992, "short-term load forecasting using a multilayer neural network
with an adaptive learning algorithm", IEEE Transactions on Power Systems, Vol. 7, No. 1, February 1992, pp.
141-148.
[8] Gross, G., F. D. Galiana, 1987, "Short-term load forecasting", Proceedings of the IEEE, Vol. 75, No. 12,
December 1987, pp. 1558-1573.
[9] Gupta, P. C. and K. Yamada, 1972, "Adaptive short-term load forecasting of hourly loads using weather
information", IEEE Transactions on Power Apparatus and Systems, Vol. PAS-91, No. 5, Sept./Oct. 1972, pp.
2085-2094.
[10] Hagan, M. T. and S. M. Behr, 1987, "The time series approach to short term load forecasting", IEEE
Transaction on Power Systems, Vol. PWRS-2, No. 3, August 1987, pp, 785-791.
[11] Papalexopoulos, A. D., T. C. Hesterberg, 1990, "A regression-based approach to short-term system load
forecasting", IEEE Transactions on Power Systems, Vol. 5,No. 4, November 1990, pp. 1535-1547.

Authors

Samsher Kadir Sheikh was born in Shegaon (India) on July15th 1968 & working as Assistant
Professor (Electrical Engineering Department) in P.D.V.V.P College of Engineering,
Ahmednagar (M.S) India.He has completed graduation from Amravati University (M.S) India
in 1995. He also completed his post graduation in Electrical Power System from Pune
University (M.S) India. Presently he is pursuing PhD in Power system from NIMS University,
Rajasthan .Now he is currently working on sugar cogeneration, power deregulation in power
system & neural network etc.

106
 International Journal of Engineering Sciences & Emerging Technologies, Feb 2012.
ISSN: 2231 – 6604 Volume 1, Issue 2, pp: 97-107 ©IJESET

M.G. Unde is working as Professor in Electrical Engineering Deptt..at Pad. Dr. Vithalrao
Vikhe Patil College of Engineering, Ahmednagar from last 24 years. He has completed his
Graduate degree B.E. Elect. In 1986 from Government College of engineering Pune and
completed Masters Degree M.E. Power System in 1998 from the same college. Presently ,he
is a Ph.D. research scholar of Pune university. His research interest is power system
analysis and power system grounding design and analysis. Electrical machines, matlab
programming, neural network etc.

107

View publication stats