Modeling and forecasting of Turkey's long term electricity consumption with least square support vector machines

Managing electrical energy supply is a complex task. The most important part of energy resource planning is forecasting of the future electricity consumption in the regional or national service area. Accurate consumption models help government to make important decisions including decisions on purchasing, generating electric power and infrastructure development. This study deals with estimation of the net electricity consumption of Turkey until the year 2018 based on multiple lineer regression analysis(MLR), artificial neural network(ANN) and least square-support vector machines(LS-SVM) methods. Installed capacity, gross electricity generation, population and total subscribership are selected as independent variables. The results obtained by LS-SVM are compared to those obtained by MLR and ANN technique. It is shown that, LS-SVM is a good forecasting tool for forecasting of electric energy consumption.Key Words : Electricity consumption, Least-Square Support Vector Machine, LS-SVM, ANN