چکیده:
This study focuses on the forecasting of energy demands of residential and commercial sectors using linear and exponential functions. The coefficients were obtained from genetic and particle swarm optimization (PSO) algorithms. Totally, 72 different scenarios with various inputs were investigated. Consumption data in respect of residential and commercial sectors in Iran were collected from the annual reports of the central bank, Ministry of Energy and the Petroleum Ministry of Iran (2010). The data from 1967 to 2010 were considered for the case of this study. The available data were used partly to obtain the optimal, or near optimal values of the coefficient parameters (1967–2006) and for testing the models (2007–2010). Results show that the PSO energy demand estimation exponential model with inputs, including value addition of all economic sectors, value of constructed buildings, population, and price indices of electrical and fuel appliances using the mean absolute percentage error on tests data were 1.97%, was considered the most suitable model. Finally, basing on the best scenario, the energy demand of residential and commercial sectors is estimated at 1718 mega barrels of oil equivalent up to the year 2032.
خلاصه ماشینی:
Results show that the PSO energy demand estimation exponential model with inputs, including value added of all economic sectors, value of made buildings, population, and price indices of electrical and fuel appliances using the mean absolute percentage error on tests data were 1.
Summary of energy demand estimation empirical studies Independent variables (period Method used Author(s) Forecasting for of data for model development) Ordinary Least Squares (OLS) (Leticia, Boogen, & Filippini, 2012) Empirical analysis on the residential demand for electricity-Spain influence of price, income, weather conditions (2000-2008) Continue Table 1.
Summary of energy demand estimation theoretical studies Independent variables Method used Author(s) Forecasting for (Ünler, 2008) electricity demand Turkey (period of data for model development) Population, GDP, import and export (1979-2005) Particle Swarm Optimization (PSO ) Genetic algorithm (GA) grey hybrid model and genetic algorithm colony algorithm (CA) Artificial Neural Networks (ANN) (AlRashidi & El-Naggar, 2010) (Kıran, Özceylan, Gündüz, & Paksoy, 2012) (Ardakani & Ardehali, 2014) (Canyurt & Ozturk, 2008) (Lee & Tong, 2011) (Kıran, Özceylan, Gündüz, & Paksoy, 2012) (Ardakani & Ardehali, 2014) maximum annual electricity load Kuwait Energy Demand Turkey electricity load Demand Iran and U.
According to the results, the best scenario for predicting the energy demand of the residential and commercial sectors of Iran is derived from the exponential model simulated by the PSO algorithm.