Short-term Load Forecasting Based on Chaotic BP Neural Network United with Improved Particle Swarm Optimization

Abstract

Short term load forecasting is the premise and guarantee for the safe and stable operation of power system. Based on the prediction method considering the similarity recognition of related factors, chaotic BP neural network (CBNN) united with improved particle swarm optimization (IPSO) model is proposed. The new hybrid algorithm combines the excellent learning ability of CBNN to capture relevant factors and the ability to obtain global optimal value in IPSO. In view of the shortcomings of CBNN, such as the difficulty of parameter determination and slow speed, the new hybrid algorithm is used to optimize the parameters of CBNN, so as to improve the global searching ability of the algorithm. Finally, the future load is predicted by an example; it verifies the feasibility and practicability of the new model, which is better than IPSO and IPSO respectively.