Application of an improved BP neural network in the forecasting of urban power supply

The electricity is closely related to the residents´ living. The satisfaction of living and industrial electricity consumption is directly related to economic development and social stability. Accurately predicting urban electricity consumption for the foreseeable future there will help decision makers make the adjustment and specific work. In recent years, in order to solve the problem of the forecasting of urban power supply, many models have been proposed, such as multiple regression analysis, gray prediction algorithm, and so on... It is the basis of the forecasting of urban power supply that artificial neural networks have strong holographic associative learning ability and fault tolerance. Because the data collection for forecasting electricity consumption is often inaccurate, and may even have larger errors, but only when the selection of forecasting model parameters is changed in different environments, a better prediction will come out. The traditional forecasting methods are difficult to resolve these problems. The improved BP neural network was put forward to resolve these problems, after analysis, this method got more realistic results than other models.