Regression Analysis For Non-Linear Load Growth

(Load Forecasting)
 Muhammad Arslan Ahmad Technology, Lahore
Department of Electrical Lahore, Pakistan
Engineering [email protected]
University of Engineering and
 Abstract—Load growth of the geographical area served by energy prices increasing by a factor of ten or more during
a utility company is the most important factor influencing the peak situations, load forecasting is vitally important for
expansion of distribution system. For an energy-limited utilities. Over the last few decades a number of forecasting
economy system, it is crucial to forecast load demand methods have been introduced. Most of these methods use
accurately. Moreover, electric load forecasting is principal to statistical techniques sometimes combined with artificial
the economic and efficient provision of electric power to meet intelligence algorithms such as neural networks, fuzzy logic,
various load demands for a specified period of time. This paper and expert systems. Two of the methods, so-called end-use
presents the detailed aspects of regression techniques being and econometric approach are broadly used for medium and
used in forecasting the electricity load. Over the last few
long-term forecasting. A variety of methods, that include the
decades a number of forecasting methods have been
introduced. Most of these methods use statistical techniques
so-called similar day approach, various regression models,
some times combined with artificial intelligence algorithms time series, neural networks, statistical learning algorithms,
such as neural networks, fuzzy logic, and expert systems. A fuzzy logic, and expert systems, have been developed for
large variety of mathematical methods and ideas have been short-term forecasting.[1]
used for load forecasting. The development and improvements Traditional models for Load Forecast can be generally
of appropriate mathematical tools will lead to the development classified as time series models and Regression Models. In
of more accurate load forecasting techniques. In this paper, we
time series models, the previous load is extrapolated to
are concerned to talk mainly about the long-term forecast
using various regression techniques.
obtain future loads. These models are frequently augmented
with transfer functions in order to adapt the consumer
Keywords—Load growth, distribution system, forecast, response to changing weather patterns and intangible factors.
mathematical models, Regression techniques, These models require large amount of data and a complex
estimation procedure. Regression models constitute the
second major modeling technique, wherein the database is
I. INTRODUCTION frequently divided into smaller segments, whereby a
Electrical Energy is one of the most important sources for Regression models is built for each segment, such as a
social and economic development of all nations. The growth season or a day or a week.[2]
in energy consumption is essentially linked with the growth
in economy. Electricity demand increases due to the Regression analysis is a technique used for analyzing the
population growth, higher per capita consumption, rapid numerical data. The dependent variable yi is a linear
development of industrial & commercial growth, higher combination of the parameters, α, and the independent
Gross Domestic Product (GDP) growth and structural variables, xi , which could be linear or nonlinear. The simple
changes in the economy.[1] linear and multiple linear regressions are the two basic types
of linear regression. For instance, in simple regression of N
Load forecasting problem has been receiving great and data points’ modeling, there is one independent variable, xi ,
growing attention as being an important and primary tool in and two parameters, α0 and α1 , which yield a straight line,
power system planning and operation. It has been an called fitted regression line: [3]
attractive research topic in many countries all over the world,
especially in fast developing countries with higher load yi =α0 +α1xi +ei i =1,...,N (1)
growth rate in recent decades. Many techniques such as auto Regression analysis is an important tool for modeling and
regressive integrated moving average and regression analysis analyzing data. Here, we fit a curve / line to the data points,
(RA) have been investigated to solve the problem of Load in such a manner that the differences between the distances
Forecasting in the last few decades.[1] of data points from the curve or line is minimized.[4]
Load forecasts can be divided into three categories: 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.[1] 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 a next day load with
an accuracy of approximately 1-3%.[1] However, it is
impossible to predict the next year peak load with the similar
accuracy since accurate long-term 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
Figure 1. Regression Analysis[4]
load, which would take place for average annual peak
weather conditions or worse than average peak weather
conditions for a given area.[1]
Load forecasting has always been important for planning
and operational decision conducted by utility companies.
However, with the deregulation of the energy industries, load
forecasting is even more important. With supply and demand
--fluctuating with the changes of weather conditions an
 II. FORECASTING METHODOLGY
Forecasting is simply a systematic procedure for
quantitatively defining future loads. Depending on the time
periods of interest, a specific forecasting procedure may be
classified as a short term, intermediate or long term
technique.
Because system planning is our basic concern and
because planning for the flow generation, transmission and
distribution facilities must begin 4-10 year in advance of the
actual in service data.
Regression analysis is the most accurate method being
used for the forecasting. There are multiple benefits of using
regression analysis. They are as follows:[4]

 It indicates the significant relationships between

dependent variable and independent variable.
 It indicates the strength of impact of multiple Figure 2. Linear Regression[4]
independent variables on a dependent variable.
The difference between simple linear regression and
Regression in general is a relationship between the multiple linear regression is that, multiple linear regression
variable we want to forecast (dependent) and another has (>1) independent variables, whereas simple linear
variable (independent). regression has only 1 independent variable.
Or, we can say For multiple regression coefficient, complex linear
Y = f(x) regression is as;[1]
If the independent variable is time, then we call it simple
time-series regression, and simple refers to a single
independent variable.

A. Linear Regression
It is one of the most widely known modeling techniques.
Linear regression is usually among the first few topics
which people pick while learning predictive modeling. In
this technique, the dependent variable is continuous, B. Polynomial Regression
independent variable(s) can be continuous or discrete, and A regression equation is a polynomial regression equation
nature of regression line is linear.[4] if the power of independent variable is more than 1. The
Linear Regression establishes a relationship between equation below represents a polynomial equation:[4]
dependent variable (Y) and one or more independent
variables (X) using a best fit straight line (also known as y=a+bX2 (2)
regression line).[4] In this regression technique, the best fit line is not a straight
It is represented by an equation [4] line. It is rather a curve that fits into the data points.
Y=a+bX + e,
where a is intercept, b is slope of the line and e is error term.
This equation can be used to predict the value of target
variable based on given predictor variable(s).

Figure 3. Polynomial Regression[4]

III. PRACTICAL EXAMPLE OF FORECASTING

A Non-Linear Regression model was created for mid-

term load forecasting to improve the modeling of
seasonality. Regression analysis can also be combined with human error, some load values were either omitted or out of
Artificial Neural Networks to create a hybrid model. range. These values were corrected using interpolation or by
The case study is Covenant University, one of the first finding the average of the preceding and succeeding values
privately-owned universities in Nigeria. It is located in in the daily load log.
Canaan land, km 10 Idiroko expressway, Ota, Ogun state, The average load consumption for each day is then
Nigeria. Covenant University is a fast developing calculated and added to the other values of average load
community known for its in-depth research work.[5] consumption for the other days in a month to give the
This study will show the results carried out from a average load consumption for the particular month. The
medium-term load forecast that was carried out using values values for all the six months were collated and their
from August to October 2012 using the regression analysis respective average load curves plotted but to prepare the
method. The results derived from this forecast were also forecast models, the months of August, September and
compared with the actual values recorded within the period October 2012 are used. The values for these months are
that was forecasted (November to January 2013) and the arranged and placed in Table 1 in the column labelled ‘Y
accuracy of the forecasting results tested using the MAPE (load) [in MW]’. The months are in the column ‘X (month)’.
and RMSE tools.[6] In this table,
August 2012 =1, September 2012 = 2, October 2012 =3.
The Three models are used to calculate the medium term
A. Methods and implementation of the forecast
load forecast. They are:
Regression Analysis is a very important statistical tool 1) The Linear Regression Model:
which is very useful in determining the statistical This is the simplest model of regression analysis. It
relationship or dependence between a change in one variable comprises of an independent and dependent variable. It is
and the change in another when compared to it. To carry out denoted by the formula:
a medium term forecast of the electrical load consumed by
Covenant University, the following steps have to be taken Y = a + bx (3a)
using the collated load data:[7] Where
Y is the value of the load
B. Collection of data X is the number of months (i.e.) August = 1, September= 2
etc.
The data was collected on an hourly basis from the
The values a and b are derived by solving (3b) and (3c)
substation, the duration of one year January 2012-January
simultaneously.[8]
2013. The load data from August to October 2012 is will be
∑Y = n a + b∑X (3b)
used for the load forecast. It was then transferred and saved
∑XY = a∑X + b∑X2 (3c)
as a Microsoft Excel worksheet.[7]
To carry out the forecasts, the months to be forecasted are
given numbers with respect to X and then calculated.
C. Pre-processing of the Data Collected Substituting the values of Table 1 and solving the equations
The data collected is then analyzed and arranged into (3a) and (3b) simultaneously, the equation (3a)
days. The average load consumption per day is calculated becomes:
and then the total average monthly load consumption from Y = 20.44 + 19.26b (3d)
August to October 2012 is also derived. All the calculations To test the model, the values for the first three months (from
were carried out using Excel functions necessary for the b=1 till b=3) are substituted into equation (3d) then the
data collation. The data was also placed in tables for easier values for the months to be forecasted (from b= 4 till b=6)
display and comparison.[7] are also substituted into equation (3d).
The forecasted values are then compared with the actual
values gathered from the substation as shown in Table 2.
D. Building of the Forecast Models Figure 4 shows a comparison of the two values in a
To carry out medium term load forecasting, there are a graphical form.
lot of models available for use. These models incorporate
various data like time, temperature, load and the effect of 2) The Compound-Growth Model:
seasonality. Regardless of the complexity or simplicity of This is another model of regression analysis. It is denoted
any model, it is important to choose the model that best suits by the formula:
the data available for forecasting. Y = antilog(c + dX) (4a)
The data gathered is peculiar to Covenant University, Like the linear regression model, the values of c and d are
Nigeria. After representing the data in form of graphs, trends also derived by solving the equations (4b) and (4c)
were discovered and these trends were used for forecasting. simultaneously.[9]
The trends were as a result of observations made from the
gathered data.

E. Implementation of the Models

It is important to note that the electric load is the most
important input needed when carrying out an electrical load
forecast. Due to wrong readings of the values and general
 Table 1[9]

Σ logY = nc + dΣX (4b)

ΣX (log) Y = cΣX + dΣX2 (4c)
Substituting the values of Table 1 and solving the equations
(4b) and (4c) simultaneously, the equation (4a)
becomes:
Y = antilog (1.4685 + 0.1432x) (4d)
To test the model, the values for the first three months (from
x=1 till x=3) are substituted into equation (4d) then the
values for the months to be forecasted (from x= 4 till x=6)
are also substituted into equation (4d). The forecasted values
are then compared with the actual values gathered from the
substation as shown in Table 3. Figure 5 shows a
comparison of the two values in a graphical form. Figure 4[9]

3) The Cubic Regression Model: Figure 4 shows the graph that compares the actual and
This model is denoted by the formula: forecasted average load consumption of Covenant
Y = a0 + a1X + a2X2 (5a) University from August 2012 – January 2013 using the
The values of a0, a1 and a2 can be derived simultaneously linear method.
using these formulas:[9]
aon + a1∑xi + a2∑xi2= ∑yi (5b)
ao∑xi+ a1∑xi2 + a2∑xi3 = ∑ xiyi (5c)
ao∑xi2+ a1∑xi3 + a2∑xi4 = ∑ xi2yi (5d)
The values to be used in this calculation are all derived
from Table 1. Substituting the values of Table 1 and solving
the equations (5b), (5c) and (5d) simultaneously, the
equation (5a) becomes:
Y = 36.34 + 0.18X + 4.77X2 (5e) Table 3[9]
To test the model, the values for the first three months (from
x=1 to x=3) are substituted into equation (5e) then the
values for the months to be forecasted (from x= 4 to x=6)
are also substituted into equation (5e).
The forecasted values are then compared with the actual
values gathered from the substation as shown in Table 1.
The MAPE and RMSE of the different models are
calculated by substituting values into equations (1) and (2).
The values are shown in Table 5.

Figure 5[9]

Figure 5 shows the graph that compares the actual and

forecasted average load consumption of Covenant
University from August 2012 – January 2013 using the
compound growth method.
Table 2[9] Table 4 shows the actual and forecasted values using the
cubic regression method.

F. Presentation of forecast results

Table 1 above shows the computation of the respective
models. The values in Y are the values of the average load
consumed in each month where the first month is August
2012, second month is September 2012 and third month is
October 2012.

Table 4[9]
 this period, and major maintenance operation can take place
as well. Meanwhile, this drop in load consumption accounts
for the huge deviation in the forecast values for the three
regression models. Work is under way to incorporate
artificial intelligence into the study.

REFERENCES

[1] Chapter 2: Electrical Load Forcasting

https://www.scribd.com/document/23712060/Chapter-2-Load-
Forecasting Retrieved 25 October 2018

Figure 6[9] [2] JieWu (2013), “Short term load forecasting technique based on the
seasonal exponential adjustment method and the regression model”
Figure 6 shows the graph that compares the actual and
forecasted average load consumption of Covenant [3] J. Kumaran, “A Simple Regression Model for Electrical Energy
Forecasting” International Journal of Advanced Research in
University from August 2012 – January 2013 using the Electrical, Electronics and Instrumentation Engineering
cubic method growth method.
[4] Analytics vidya
https://www.analyticsvidhya.com/blog/2015/08/comprehensive-
guide-regression/ Retrieved on 25 October 2018
[5] Isaac A. Samuel, Felly-Njoku Chihurumanya F., Adewale Adeyinka
A. and Ayokunle A. Awelewa (2014), “Medium-Term Load
Forecasting Of Covenant University Using the Regression Analysis
Methods”, Journal of Energy Technologies (Online)Vol.4, No.4, 2014

Table 5 [9] [6] MAPE & RMSE

https://www.researchgate.net/post/what_is_the_difference_between_
Mean_Absolute_Error_MAE_the_Mean_Absolute_Percentage_Error
_MAPE_and_the_Root_Mean_Square_Error_RMSE Retrieved on 25
October 2018
IV. CONCLUSION
[7] Isaac A. Samuel, Felly-Njoku Chihurumanya F., Adewale Adeyinka
As electricity markets have deregulated over the last decade, A. and Ayokunle A. Awelewa (2014), “Medium-Term Load
accurate load forecasts have become a vital part of a utility’s Forecasting Of Covenant University Using the Regression Analysis
Methods”, Journal of Energy Technologies (Online)Vol.4, No.4, 2014
long-, medium-, and short-term generation and procurement Page 11
planning. An inaccurate load forecast can have severe
consequences for customers in the form of higher rates.[10] [8] Isaac A. Samuel, Felly-Njoku Chihurumanya F., Adewale Adeyinka
The objective of this paper was to forecast electricity load A. and Ayokunle A. Awelewa (2014), “Medium-Term Load
using different regression techniques. Short-term load Forecasting Of Covenant University Using the Regression Analysis
forecasting can help to estimate load demands and to make Methods”, Journal of Energy Technologies (Online)Vol.4, No.4, 2014
Page 12
decisions that can prevent overloading. Timely
[9]
implementations of such decisions lead to the improvement
[10] Isaac A. Samuel, Felly-Njoku Chihurumanya F., Adewale Adeyinka
of network reliability and to the reduction occurrences of A. and Ayokunle A. Awelewa (2014), “Medium-Term Load
equipment failures and blackouts. Load forecasting is also Forecasting Of Covenant University Using the Regression Analysis
important for contract evaluations and evaluations of Methods”, Journal of Energy Technologies (Online)Vol.4, No.4, 2014
Page 13
various sophisticated financial products on energy pricing
[11]
offered by the market.
[12] Jennifer Hinman and Emily Hickey (2019), “Modeling and forcasting
Within multiple types of regression models, it is important Short-term electricity load using regression analysis”, Illinois State
to choose the best suited technique based on type of University.
independent and dependent variables, dimensionality in the [13]
data and other essential characteristics of the data. One of [14] Carakostas MC1, Green JW, ‘Using the correct regression analysis
the best trick to find out which technique to use, is by technique in method comparison studies.’ Vet Clin Pathol.
checking the family of variables i.e. discrete or continuous. 1991;20(4):91-94
[11]
A medium-term load forecast of an educational institution is
the thrust of this paper. Regression analysis of sample load
data (collected between January 2012 and January 2013) has
been detailed. Results of the models used reveal a drastic
decrease in load consumption in the month of December.
This means that fuel supply can be at the minimum during