Power Demand Forecasting Based on BP Neural Network Optimized by Clone Selection Particle Swarm

摘要

Based on ordinary BP algorithm, firstly established the power demand forecasting model. Then the model's network structure was identified by using the power demand's influential factors as the input of the network, Repeated the optimization of the BP network's weight combination with the aid of clone selection particle swarm algorithm, and adopted the weight optimized as the initial value of the BP neural network, carried on the BP algorithm until the network met the training requirement. Finally recent years' annual data of relevant input and output variables were used to empirically forecast the power demand with the established model, mean absolute error is 156.8340, root mean square error is 160.9708, root mean square error rate is 0.0095. The results show that BP neural network based on clone selection particle swarm has both fast training speed and small error, the forecast precision also has been significantly improved.