The Improved Load Forecasting Model of BP Neural Network

The traditional BP Neural Network forecasting model influenced clearly by the complexity of network frame and sample, which lead to the over-learning or low-extensive ability. The paper puts forward a new model to improve the BP Neural Network load forecasting model, which use attribution reduction algorithm of rough sets to reduce the various historical data related to the load and eliminate the property not related to the decision-making information. It was tested that this method reduces the input variables of the BP Neural Network, so shortens the training time of the BP. Neural Network load forecasting model. At last the forecasting capability is improved.