Hybrid artificial neural network system for short-term load forecasting

Abstract

This paper presents a novel hybrid method for Short-Term Load Forecasting (STLF). The system comprises of two Artificial Neural Networks (ANN), assembled in a hierarchical order. The first ANN is a simple Multi Layer Perceptron (MLP) which is used as integrated load predictor (ILP) of the forecasting day. The output of the ILP is then fed to another, more complex MLP, which acts as the hourly load predictor (HLP) for the forecasting day. By using a simple ANN that predicts the integral of the load (ILP), additional information is presented to the actual forecasting ANN (HLP), while keeping its input space relatively small. This property enables online training and adaptation, as new data becomes available, because of the short training time. Different sizes of the training sets have been tested, and the optimum of a 30 day sliding time-window has been determined. The system has been verified on recorded data from a Serbian electrical utility company. The results demonstrate better efficiency of the proposed method in comparison with non-hybrid methods, because it produces better forecasts, and yields a smaller mean average percentage error (MAPE).