Short Term Load Forecasting Using Improved Particle Swarm Fuzzy Neural Network

The development of electricity market requires more accurate short-term load forecasting (STLF). A novel bionic algorithm is proposed to improve the STLF accuracy and speed. In this paper, improved particle swarm algorithm (IPSO), fuzzy theory (Fuzzy) and BP neural network (BPNN) are combined to form a new STLF method which is called IPSO-F-BPNN. When we build this model, the impact of climate and temperature is processed with fuzzy technique and considered as input data of the network. The improved particle swarm algorithm is used to train network parameters until the learning error tends to be stable. Then weights optimized BP algorithm is adopted to accomplish load forecasting. The case analysis shows that, compared with the traditional BP neural network method, the new model has better predictive ability on power system short-term load due to the higher predict precision and faster convergence.