摘要
针对现有风速预测方法在解决预测问题时存在的不足,提出了采用收缩因子对粒子速度更新的方式进行改进,保证粒子群算法前期的全局搜索能力和后期的局部寻优能力,提高算法的收敛性能。利用改进PSO对LSSVM参数进行寻优,建立基于改进PSO优化LS-SVM的短期风速预测模型。采用实际风速数据进行仿真分析,所建模型的平均相对误差和均方根误差分别为3.72%和0.21,均好于其他几种风速预测方法,验证了所提出方法的正确性和实用性。
Aiming at the problem of wind speed prediction,the shortcomings of the existing wind speed prediction methods are pointed out.In order to ensure the global searching ability and local searching ability of PSO in the early stage and the local searching ability in the later stage,the shrinkage factor is used to improve the way of particle velocity updating,and the convergence performance of PSO is improved.The improved PSO is used to optimize LS-SVM parameters,and a short-term wind speed prediction model based on the improved PSO is established.The actual wind speed data is used for simulation analysis,and the PSO is improved to optimize LS-SVM.The average relative error and root mean square error of the wind speed prediction model are 3.72%and 0.21 respectively,which are better than other wind speed prediction methods,and verify the correctness and practicability of the proposed method.
作者
范曼萍
周冬
FAN Man-ping;ZHOU Dong(College of Electrical and New Energy,Three Gorges University,Yichang 443000,China)
出处
《电力学报》
2020年第2期123-128,142,共7页
Journal of Electric Power
关键词
风速预测
改进粒子群
最小二乘支持向量机
收缩因子
wind speed prediction
improved particle swarm
least squares support vector machine
contractile factor
