A Comparison of the Performance of Classification Methods and Artificial Neural Networks for Electricity Load Forecasting

Abstract

Electricity load forecasting plays a key role forutility companies. Short-term and medium-term electricity load forecastingprocesses allow the utility companies to retain reliable operation and highenergy efficiency. On the other hand, long-term electricity load forecastingallows the utility companies to minimize the risks. Long-term forecasting alsohelps the utility companies to plan and make feasible decisions in regard togeneration and transmission investments. Since there are commercial and technicalimplications of electricity load forecasting, the accuracy of the electricityforecasting is important not only to the utility companies but also to theconsumers. In this paper, we carry out a performance evaluation study toevaluate the accuracy of different classification approaches for electricityload forecasting. As shown with the results of the performance evaluationstudy, some of the investigated approaches can successfully achieve highaccuracy rates and therefore can be used for short-, mid-, or long-termelectricity load forecasting.