Short-Term Load Forecasting Based On BP Artificial Neural Network

The short-term load forecasting (STLF) of electric system is one of the important routines for power dispatch and utility departments. It is widely used in the dispatching and operation planning of power systems, and the accuracy of the load forecasting is helpful to the security, economy of power systems and quality of the power supply. So the study of short-term load forecasting has been paid enough attentions in the past decades.At aiming some problems in the short-term load forecasting of the power system, based on studying carefully and deeply the basic theory and method of the artificial neural network (ANN) and considering the analysis of many kinds of factors which impact load, the load forecasting model is set up. In its input features, the load characteristic of near days and every kind of weather factors that considered. Then the input variables are unified, and weather factors including temperature and weather condition are quantified.To improve convergence speed of ANN,various improved back propagation (BP) training algorithms are compared. By analyzing the characteristic of the model that has been set up, Levenberg-Marquardt algorithm is adopted. To avoid entering into local minimum point for improper selection of initial parameters value of forward-back neural network, particle swarm optimization (PSO) algorithm is introduced to determine initial parameters value of network,a short-term load forecasting method of power system based on PSO-BP is presented. In this method, PSO is used to optimize connection weights of forward-back neural network, then BP algorithm with optimized weights is used to finish short-term load forecasting process.Experimental results show that, compared with momentum backpropragation algorithm, Levenberg-Marquardt algorithm can speed convergence of ANN and improve predicting precision. By predicting the load of 28th in November, compared with the short-term load forecasting method of power system based on improved BP neural network, the short-term load forecasting method based on PSO-BP presented in this thesis can improve the forecast precision, and can be used in STLF system.