SOS-ELM based Prediction Model for Electrical Load Forecasting

Abstract

Load forecasting is an interesting issue in which the electrical load demands are dynamic and non-linear in nature. This paper develops a forecasting model in which, the next hour load is predicted. The most popular machine learning algorithm called as Extreme Learning Machine (ELM) is implemented for the load prediction. The performance of ELM mainly depends on the output weights which are determined from the input weights and biases. In this paper both machine learning algorithm and optimization technique have been applied for optimal design of input weights and biases for getting a better result. Symbiotic Organism Search (SOS) algorithm has been implemented for determination of the optimal weights and biases.