Short-term load forecasting using H∞ filter and Elman neural network

In view of the dynamic nonlinear characteristics of short-term load forecasting (STLF) in power systems, a short-term load forecasting model is proposed based on Wiener model, and Elman recursive neural network is used to fit into its nonlinear part in this paper. H^∞ filter is introduced to overcome the unknown disturbance and noise in the linear part of the systems during forecasting, and then, Elman dynamic neural network is applied to implement the nonlinear loads prediction. Compared with normal Kalman filter and BP neural networks, the proposed method possesses high learning efficiency, strong adaptability, high forecasting accuracy and good forecasting behavior, and is very suitable to short-term load forecasting for power system. In the end, simulation results indicate the availability of the proposed method.