混合粒子群径向基神经网络的短期负荷预测的应用

电力短期负荷预测的结果对电力系统的经济效益具有重要影响.为了克服基本粒子群(Particle Swarm Optimization,PSO)算法收敛精度不高、易陷入局部最优的缺点,提出一种将自然选择和变异结合的混合粒子群(Hybrid Particle Swarm Optimization,HPSO)算法,可以保持种群的多样性,有效地避免粒子早熟,并利用混合粒子群算法优化径向基神经网络的权值,最后将优化好的径向基神经网络进行广西某市的短期电力负荷预测.计算结果表明,该算法收敛速度快,并达到了提高预测精度和改善网络性能的要求.

Short-Term Load Forecasting Based on Hybrid Particle Swarm Optimization and Radial Basis Function Neural Network Model

The result of short-term load forecasting is important to economic benefit.In order to improve the convergence precision of the basic PSO algorithm and overcome the defect of easily falling into the local optimum,a hybrid particle swarm optimization algorithm combined with natural selection and the mutation is presented,which can avoid the partilce＇s premature effectively.Then the weights of the radial basis neural network are optimized by the improved PSO algorithm.Finally,the algorithm is employed to the short-term load forecasting in a city of Guangxi.The results of simulation show that the convergence speed is accelerated,and it meets the requirements of enhancing the forecasting accuracies and improving the performance of the network.