Medium Term Load Forecasting Using Statistical Feature Self Organizing Maps (SOM)

Abstract

Load forecasting is an essential tool for power system activity and planning. With the increase in development and the expansion of power system, it is important for the electrical utility to make a decision in ensuring that there would be enough supply of electricity to deal with the increasing demand. This research presents the Medium Term Load Forecasting using the artificial neural networks: Kohonen’s Self-organizing Maps. The main purpose of this paper was to understand the ability of Self-Organizing Maps in forecasting the load demand and to train and test via SelfOrganizing Maps method using the selected features. Using data provided by the Global Energy Forecasting Competition (GEFCom2012), this paper focused on the missing data from the year 2005 and 2006 for the load forecasting. The loaded data were trained, tested, and forecasted using SOM Toolbox in MATLAB software. The accuracy of the forecasted data was determined by calculating the error of each forecasted data by comparing them with the actual data. Then, the Mean Absolute Percentage Error was computed to determine the accuracy of the results.