Comparison of forecasting methods for power system short-term load forecasting based on neural networks

Abstract

In this paper, the periodicity and variation of power system load data has been analysed with bad data removed when correlation process was conducted, and proper parameter has been applied to be the restraint weight of neuron. Then back propagation (BP) neural network and radial basis function (RBF) neural network has been established by means of MATLAB. The load is predicted by the use of model and meanwhile the effectiveness and veracity of the neural network was verified via the comparison with the actual load. On this basis, we introduce the wavelet analysis which was used with the combination of the neural network to establish incompact wavelet analysis neural network of which the effectiveness has been testified. Finally, comparison has been made among the three forecasting methods.