Application Of Fuzzy Neural Network To Power System Short-Term Load Forecast

Short-term load forecast is very important for reasonable and economic arrangement of the electrical generator operation, decisions of the fuel purchase plan and the power exchange. So the study of short-term load forecast has been paid enough attentions in the past decades. At aiming some problems in the short-term load forecast of the power system, this thesis studies carefully and deeply the basic theory and method of the Artificial Neural Network (ANN) and fuzzy system, and puts forward two load prediction models based on the Artificial Neural Network (ANN) and Fuzzy Neural Network (FNN). In the load forecast method based on the ANN, all load data should be normalized firstly, and then accessional momentum method are compared with L-M Back Propagation algorithm in efficiency and precision, at last L-M Back Propagation algorithm is winner due to its better forecast effect. Based on theory of fuzzy and artificial neural network, this thesis presents a forecast model which based on neural network combined with fuzzy inference principles. A kind of rule-extracted algorithm of fuzzy inference is put forward in this thesis, and the definition of fuzzy rule tree is given also. By this algorithm, the problem of dimension explosion in modeling process can be avoided, as a result the node number is decreased greatly and the complexity of load forecasting model is reduced. Also it improves the efficiency of load forecast model. In the end, by comparing the load forecast result of 30th in September in Wuhan with actual load data, the load forecast model presented in this thesis is proved that it has better forecast precision and shorter training time.