Short-Term Electric Load Forecasting Using Support Vector Machines

Abstract

This research paper proposes a short-term electrical load forecasting model based on support vector machines and the ensemble machine learning method. Historical data is reviewed, and acceptable features for developing the model are discovered. Load demands from the previous twenty-four hours are utilized as the goal value, with the day of the week and the hour of the day as the input values. The effect of temperature has been also analyzed as an input. It has been found that temperature is not always a very good feature for designing the short-term load forecast model. In order to train the model efficient data and then prediction of short-term electrical load achieved. The comparative result between SVM and ensemble method have been carried out, which shows that SVM performed better in STLF model.