Engineering, 2013, 5, 108-114
doi:10.4236/eng.2013.51b020 Published Online January 2013 (http://www.SciRP.org/journal/eng)
Copyright © 2013 SciRes. ENG
Seasonal Regression Models for Electricity Consumption
Characteristics Analysis
Yusri Syam Akil
1,2
, Hajime Miyauchi
1
1
Department of Frontier Technology for Energy and Devices, Kumamoto University, Kumamoto, Japan
2
Department of Electrical Engineering, Hasanuddin University, Makassar, Indonesia
Email: yusuri@st.cs.kumamoto-u.ac.jp, miyauchi@cs.kumamoto-u.ac.jp
Received 2013
Abstract
This paper presents seasonal regression models of demand to investigate electricity consumption characteristics. Elec-
tricity consumption in commercial areas in Japan is analyzed by using meteorological variables, namely temperature
and relative humidity. A dummy variable for holidays is also considered. We have developed models for two levels of
period to analyze demand characteristics, that is, half year models and seasonal models. Some options for each model
are calculated and validated by statistical tests to obtain better models. As results, half year and seasonal models present
explicit information about how the variables affect the demand differently for each period. These specific information
help in analyzing characteristics of studied commercial demand.
Keywords: Commercial Area; Demand Characteristics; Regression Model; Seasons; Relative Humidity; Temperature
1. Introduction
In general, electricity consumption analyses such as cha-
racteristics investigation and forecasting can provide
much information related to the time variance of demand.
The result of demand analysis is useful for electric utili-
ties in many aspects, for instance, to manage and control
their systems more effective. Therefore, it is valuable to
analyze an electricity demand in detail through develop-
ment of demand models. A number of methods can be
used for a demand analysis, and one of them is regres-
sion analysis. As a tool, a regression model needs a
number of data (dependent and explanation variables).
The implementation of proper explanation variables is
required to get a good model. As an explanation variable,
meteorological parameters such as temperature, humidity,
wind speed, and so on, are commonly used and con-
firmed electricity demand effectively. Prior studies
which employed meteorological parameters and regres-
sion models for electricity demand can be found in ref-
erences such as [1-4]. Reference [1] develops a demand
model using a stepwise procedure to forecast Spanish
daily electricity demand. Reference [2] develops regres-
sion equations to analyze electricity consumption for
residential area in Hong Kong by using climatic and
economic variables. Reference [3] develops the model of
electricity consumption for residential area in Bangkok
Metropolis and analyzes effect of climatic and economic
factors for demand. Meanwhile, in [4], authors have de-
veloped two statistical models for demand in Greece,
namely daily and monthly models to forecast demand up
to 12 months ahead (mid-term demand).
As electricity demand may differ to time [2,3] and
place generally, we have an interest to analyze demand
characteristics for commercial area in a typical city in
Japan by developing demand models. This study also
aims to find the electricity consumption characteristics in
Japan. We analyze demand characteristics based on
seasonal periods for commercial area in Japan. To
achieve the aim, two demand model of two period levels
based on different time length, namely half year models
(CTEChy1, and CTEChy2) and seasonal models
(CTECSM, CTECA, CTECW, and CTECS) are
proposed to reveal further demand characteristics by
regression analysis. They are developed from an initial
model (CTEC) that is derived from all data (whole
period) with the same explanation variables and
statistical validation processes. In the context of this
study, the application of regression approach is effective
enough. Beside simple in composing the models, the
obtained regression coefficients and statistics contain
specific information about the direct relationship
between variables and demand. It may be useful to draw
a seasonal strategy and to meet demands in maintaining
power system performance.