ELECTRIC

LOAD
FORECASTIN
G

Purpose, Classification and

Methods.

Saqib Fayyaz
15-MS-EE-19

DEPARTMENT OF ELECTRICAL ENGINEERING

UNIVERSITY OF ENGINEERING AND TECHNOLOGY
TAXILA
Contents

1. INTRODUCTION................................................................................................ 2
1.1 Background:................................................................................................. 2
1.2 Application:................................................................................................... 3
1.3 Factors affecting load forecasting:...................................................................3
2. CLASSIFICATION OF LOAD FORECASTING........................................................4
3. FORECASTING METHODS................................................................................. 5
3.1. LONG AND MEDIUM TERM FORECASTING..................................................6
3.1.1 End use model:....................................................................................... 6
3.1.2 Econometric approach:...........................................................................7
3.2. SHORT TERM FORCASTING........................................................................7
3.2.1 Similar day approach:............................................................................. 8
3.2.2 Regression method:................................................................................ 8
3.2.3 Time series:............................................................................................ 8
3.2.4 Neural Networks:.................................................................................... 8
3.2.5 Fuzzy logic:............................................................................................. 9
3.2.6 Support vector machines:....................................................................10
4. Future of Load Forecasting:.............................................................................. 10
5. Conclusion:...................................................................................................... 10
6. References:.......................................................................................................... 11

1
1. INTRODUCTION

The unique characteristic of electric power industry that makes it

different from rest of the industries is that its product electricity cannot
be stored and therefore power system needs to be designed in such a way
that it meets maximum demand called peak load at all times. Because of
this characteristics it becomes essential for electric power utilities to
estimate there load in advance. By definition electric load forecasting is
the process used to predicting future electric load, based on historical
load, current and forecasted weather information.

1.1 Background:

The first central power plant Pearl street station in the US was
commissioned on September 4th, 1882. At that time power system was
designed by Thomas Edison for the sole purpose of promoting his
incandescent bulb sales. Because of limited lighting load it was easy to
determine peak demand simply by counting. Later with the expansion of
power system and introduction of devices like electric fan and electric iron
(1880s) power system became more diverse, but the limit transmission
capacity of DC at that time the stations were small and served a small
area.

The alternating current (AC) power system devised by Nikola Tesla,

initiated a major change to the industry. On November 16, 1896, electrical
power was generated and transmitted as alternating currents to industries
in Buffalo from the hydroelectric generators at the Edward Dean Adams
Station at Niagara Falls. The Niagara Falls generation project was the first
large-scale system to successfully supply electricity from one circuit for
multiple end uses, and opened the era of large-scale AC power systems.
Inventions in the 20th century like TV, radio, microwave oven etc. further
enriched the end uses and stimulated the growth of electricity demand.

Eventually the electricity demand began to be correlated with

consumer device activity and preferences. In the early 20th century,

2
economic factors, which indicated and impacted human activities
nationally or even worldwide, started to be used to forecast the long-term
trend (i.e. 10 years ahead) of the demand. It was observed that load
varied due to special events, such as radio broadcasts of presidents
speech or an important cricket match, which affected local human
activities. It was also observed that the electricity demand of certain
appliances was observed to be closely tied to meteorological conditions.
For instance, it was discovered in the 1940s that the increase of load
during the summer as the temperature increases is the result of
increasing operations of fans, refrigerators, air conditioning and other
cooling devices. While the sales of room air conditioners escalated from
74,000 in 1948 to 1,045,000 in 1953, a major effort in the field of electric
load forecasting was to investigate the impact of weather on electricity
demand.

These consumer trends and diverse range of loads have made

electric load forecasting complicated. For instance, with hundreds or even
thousands of messages, talk shows, and comedies broadcasted through
radio stations, TV stations, and the internet, it is no longer as easy as it
was in the 1940s to predict which ones could significantly affect the
electricity demand and the magnitude of the affect.

1.2 Application:
Load forecasting has many applications including energy purchasing
and generation, load switching, contract evaluation, and infrastructure
development. The purpose of load forecasting can be listed as below;

Planning of power system,

Planning of transmission and distribution facilities,
Proper power system operation
Financing,
Manpower development,
Grid formations and
Electrical sales.

Above discussion clearly indicates that Load forecasts are extremely

important for energy suppliers, ISOs, financial institutions, and other
3
participants in electric energy generation, transmission, distribution, and
markets [1].

1.3 Factors affecting load forecasting:

A good forecasting model has to capture all the salient features

dictating the behavior of power system demand. A competent forecasting
model must take into account:

1. Economy

2. Climate/ Weather

3. Time factors

4. Consumer trends

5. Customer categories

Most short-term load forecasting models factors consider, time

factors, weather data, and possible customers classes. Whereas medium-
and long-term forecasts take into account the historical load and weather
data, the number of customers in different categories, the appliances in
the area and their characteristics including age, the economic and
demographic data and their forecasts, the appliance sales data, and other
factors.

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.

Weather conditions greatly influence the load and are in fact the
most important factors in short-term load forecasts. Temperature and
humidity are the most commonly used load predictors. THI (temperature-
humidity index) and WCI (wind chill index), are broadly used by utility
companies for estimating weather impact. THI is a measure of summer
heat discomfort and similarly WCI is cold stress in winter.

4
 Most electric utilities serve customers of different types such as
residential, commercial, and industrial. The electric usage pattern is
different for customers that belong to different classes but is somewhat
alike for customers within each class. Therefore, most utilities distinguish
load behavior on a class-by-class basis

2. CLASSIFICATION OF LOAD FORECASTING

Load forecasting can be classified into three categories based on

time:

1. Short term forecast

2. Medium forecast
3. Long term forecasting

Short-term forecasts which are usually from one hour to one week,
medium forecasts which are usually from a week to a year, and long-term
forecasts which are longer than a year. The forecasts for different time
horizons are important for different operations within a utility company.

Load forecasting can also be classified broadly into:

1. Demand forecast
2. Energy forecast

Demand forecast is used to determine generation, transmission and

distribution system additions which give the expected rate of growth of
load. Whereas energy forecast is used to determine types of facilities
required i.e., future fuel requirements [2].

3. FORECASTING METHODS

There is a growing tendency towards unbundling the electricity

system. This is continually confronting the different sectors of the
industry (generation, transmission, and distribution) with increasing
demand on planning management and operations of the network. The
operation and planning of a power utility company requires an
adequate model for electric power load forecasting. Load forecasting plays
a key role in helping an electric utility to make important decisions on
power, load switching, voltage control, network reconfiguration, and

5
infrastructure development. Owing to its importance, choosing a better
approach for prediction is also vital.

Over the resent years a lot of methods have emerged to carry out
efficient load forecasting. For short term load forecasting we have
methods like similar day approach, regression models, time series, neural
networks, expert systems, fuzzy logic, and statistical learning algorithms
whereas for medium and long term forecasting we have two common
methods end use method and econometric approach.

Generally statistical approaches require a mathematical model that

represents load as function of different factors such as time, weather, and
customer class. The two important categories of such mathematical
models are: additive models and multiplicative models. They differ in
whether the forecast load is the sum (additive) of a number of
components or the product (multiplicative) of a number of factors.

For example, H. Chen presented an additive model that takes the

form of predicting load as the function of four components [3]:

L=ln+ Lw + Ls+ Lr (1)

Where L is the total load, Ln represents the normal part of the

load, which is a set of standardized load shapes for each type of day
that has been identified as occurring throughout the year, Lw represents
the weather sensitive part of the load, Ls is a special event component
that create a substantial deviation from the usual load pattern, and Lris a
completely random term, the noise.

A multiplicative model may be of the form;

L=ln Fw Fs Fr (2)

Where Ln is the normal (base) load and the correction factors Fw,
Fs, and Fr are positive numbers that can increase or decrease the overall
load. These corrections are based on current weather (Fw), special events
(Fs), and random fluctuation (Fr). Factors such as electricity pricing (Fp)
and load growth (Fg) can also be included.

6
 Various methods used for forecasting are briefly discussed in
following section.

3.1. LONG AND MEDIUM TERM FORECASTING

Following methods are included in this category [4]:

1. End use model

2. Econometric approach

3.1.1 End use model:

The information about customers along with dynamics of change are
the main characteristics of this forecasting method. This method is
capable of estimating energy demand by collecting extensive information
on consumers, their age, size of house, device preferences, age, behaviors
etc.

End use model states that electricity demand is derived from

customers demand for light, cooling, heating, refrigeration, etc. Therefore
end-use models represent energy demand as a function of the number of
appliances in the market. Mathematically this accounting approach can be
expressed as:
I
Aj= A ijUECij( 3)
t =1

Where Aj is the quantity of electricity demanded in the jth sector

(residential,, commerical, etc) and Aij is the number, or saturation, of
appliance i in demand class j. UEC represents the unitary consumption of
each appliance i for demand sector j. Summary overall demand classes
yields total electricity, peak or energy, demand, Q.
j
Q= Qj (4 )
j=1

The number of existing appliances can usually be found through

surveys. Future values of saturation are forecasted using economic and
demographic variables such as income, future appliance cost, and
population. Future appliance saturation is theorized to grow along an S-

7
shaped curve. Past surveys help identify where one is at on the curve and
future saturations are extrapolated from there. End use models, then, can
be used to forecast either type of electricity demand.

Ideally this approach is very accurate. However, it is sensitive to the

amount and quality of end-use data. For example, in this method the
distribution of equipment age is important for particular types of
appliances. End-use forecast requires less historical data but more
information about customers and their equipment.

3.1.2 Econometric approach:

This approach combines economic theory and statistical techniques
for forecasting the energy demands. This model estimates the relationship
between demand and factors influencing demand through the use of least
square method or time series methods. These factors are called driving
variables. Mathematically in load forecasting econometric approach, if the
quantity of electricity demanded is Qt then;

Qt=f ( At , Rt ) (5)

Where At is the stock of appliances and Rt is the utilization rate of

the appliances; and

At =g ( Pt , yt , Xt )(6)

Rt=h ( Pt , yt , Zt ) (7)

Where Pt is a vector of fuel prices including electricity, Yt is an

income measure, and Xt and Zt are vectors of other relevant variables
possibly relating to weather and demographics. Substituting for At and Rt:

Qt=k ( Pt Yt Xt Zt ) (8)

The above equation represents a basic econometric model. These

models are used to analyze two different energy parameters namely
energy demand and peak demand.

8
 3.2. SHORT TERM FORCASTING
A large variety of statistical and artificial intelligence techniques
have been developed for short-term load forecasting. Some of these
techniques are:

1. Similar day approach

2. Regression method
3. Time series
4. Neural networks
5. Fuzzy logic
6. Support vector machines

Their brief description is provided below.

3.2.1 Similar day approach:

This approach is based on searching historical data for days within
one, two, or three years with similar characteristics to the forecast day.
Similar characteristics include weather, day of the week, and the date.
The load of a similar day is considered as a forecast. Instead of a single
similar day load, the forecast can be a linear combination or regression
procedure that can include several similar days. The trend coefficients can
be used for similar days in the previous years.

3.2.2 Regression method:

Regression is the one of most widely used statistical techniques. For
electric load forecasting regression methods are usually used to model the
relationship of load consumption and other factors such as weather, day
type, and customer class. The model is similar to the additive and
multiplicative models discussed earlier.

3.2.3 Time series:

Time series methods are based on the assumption that the data
have an internal structure, such as autocorrelation, trend, or seasonal
variation. Time series forecasting methods detect and explore such a
structure. Time series have been used for decades in such fields as
economics, digital signal processing, as well as electric load forecasting. In
particular, ARMA (autoregressive moving average), ARIMA (auto
regressive integrated moving average), ARMAX (autoregressive moving

9
average with exogenous variables), and ARIMAX (autoregressive
integrated moving average with exogenous variables) are the most often
used classical time series methods. ARMA models are usually used for
stationary processes while ARIMA is an extension of ARMA to non-
stationary processes. ARMA and ARIMA use the time and load as the only
input parameters. Since load generally depends on the weather and time
of the day, ARIMAX is the most natural tool for load forecasting among the
classical time series models.

3.2.4 Neural Networks:

The use of artificial neural networks (ANN or simply NN) has been a
widely studied electric load forecasting technique since1990 [5]. Neural
networks are essentially non-linear circuits that have the demonstrated
capability to do non-linear curve fitting.

The outputs of an artificial neural network are some linear or

nonlinear mathematical function of its inputs. The inputs may be the
outputs of other network elements as well as actual network inputs. In
practice network elements are arranged in a relatively small number of
connected layers of elements between network inputs and outputs.
Feedback paths are sometimes used. In applying a neural network to
electric load forecasting, one must select one of a number of architectures
(e.g. Hopfield, back propagation, Boltzmann machine), the number and
connectivity of layers and elements, use of bi-directional or unidirectional
links, and the number format (e.g. binary or continuous) to be used by
inputs and outputs, and internally.

The most popular artificial neural network architecture for electric

load forecasting is back propagation. Back propagation neural networks
use continuously valued functions and supervised learning. That is, under
supervised learning, the actual numerical weights assigned to element
inputs are determined by matching historical data (such as time and
weather) to desired outputs (such as historical electric loads) in a pre-
operational training session. Artificial neural networks with unsupervised
learning do not require pre-operational training.

10
3.2.5 Fuzzy logic:
Fuzzy logic is a generalization of the usual Boolean logic used for
digital circuit design. An input under Boolean logic takes on a truth value
of 0 or 1. Under fuzzy logic an input has associated with it a certain
qualitative ranges. For instance a transformer load may be low,
medium and high. Fuzzy logic allows one to (logically) deduce outputs
from fuzzy inputs. In this sense fuzzy logic is one of a number of
techniques for mapping inputs to outputs (i.e. curve fitting). Among the
advantages of fuzzy logic are the absence of a need for a mathematical
model mapping inputs to outputs and the absence of a need for precise
(or even noise free) inputs. With such generic conditioning rules, properly
designed fuzzy logic systems can be very robust when used for
forecasting. Of course in many situations an exact output (e.g. the precise
12PM load) is needed. After the logical processing of fuzzy inputs, a
defuzzification process can be used to produce such precise outputs.

3.2.6 Support vector machines:

Support Vector Machines (SVMs) are a more recent powerful
technique for solving classification and regression problems. This
approach was originated from Vapniks statistical learning theory. Unlike
neural networks, which try to define complex functions of the input
feature space, support vector machines perform a nonlinear mapping (by
using so-called kernel functions) of the data into a high dimensional
(feature) space. Then support vector machines use simple linear functions
to create linear decision boundaries in the new space. The problem of
choosing architecture for a neural network is replaced here by the
problem of choosing a suitable kernel for the support vector machine

4. Future of Load Forecasting:

As the future power industry is moving towards deregulation and

smart grid the significance of load forecasting will greatly increase. In
order to make proper decisions for energy purchases, utilities must have a
good idea of their future electricity demand and price. Failing to do so

11
would result in excess costs in the market. Demand response initiatives
are a dominant feature of deregulated smart grid which help utilities
manage and reduce peak demands thus eliminating outages but accurate
demand response steps cannot be implemented without top notch load
forecasting. These load forecast can carry critical information about
location and magnitude of demand peaks that serves as critical parameter
for demand response. With increasing integration of renewable sources
like wind and solar which are highly dependent on weather conditions the
load forecasting problem has become a challenge.

5. Conclusion:
We can conclude by stating that load forecasting is need of the
moment of power industry especially in a deregulated market. The
increasing integration of renewable sources has opened new research
avenues in the domain of load forecasting. New and improved techniques
are bound to surface because of continual advancement in statistical
theories, stochastic approaches and mathematical techniques in general.

12
6. References:
[1].Feinberg, Eugene A., and Dora Genethliou. "Load forecasting." In Applied
mathematics for restructured electric power systems, pp. 269-285. Springer US, 2005.

[2].Sivanagaraju, S. Power system operation and control. Pearson Education India,

2009.

[3].Chen, Hong, Claudio A. Canizares, and Ajit Singh. "ANN-based short-term load
forecasting in electricity markets." In Power Engineering Society Winter Meeting,
2001. IEEE, vol. 2, pp. 411-415. IEEE, 2001.

[4].A Review of Load Forecasting Methodologies, George E. Oamek et al, Iowa State
University

[5].Peng, T. M., N. F. Hubele, and G. G. Karady. "Advancement in the application of

neural networks for short-term load forecasting." IEEE Transactions on Power
Systems 7, no. 1, pp. 250-257. 1992.

13