A radial basis function neural network approach for multi-hour short term load-price forecasting with type of day parameter

In 1990s, after deregulation of Australian electricity market, electricity became a commodity that can be bought and sold. This led power industry to change their planning strategies. In this planning Short Term Load Forecasting (STLF) plays a vital role to provide unit commitment, economic generation scheduling etc. In this paper, RBF neural network (RBFNN) is applied as short term load as well as price forecaster. While modeling process, day-type (Sunday, Monday, etc.) is considered as an extra input to the neural network. The prediction performance of proposed RBFNN architecture is evaluated using Mean of Mean Absolute Percentage Error (MMAPE) between actual data and forecasted data of New South Wales (Australia). The results obtained are compared with the results gained from classical moving average (MA), Holt-Winters and Feed Forward Neural Network (FFNN) methods. It is, in general, observed that the RBFNN is more accurate and works better with inclusion of day type input parameters.