基于竞争ISPO双胞支持向量回归短期负荷预测  

Short-term Load Forecasting Based on Competitive ISPO and Twin Support Vector Regression

作　　者：彭显刚[1] 王洪森[1] 

机构地区：[1]广东工业大学自动化学院,广州510006

出　　处：《电力系统及其自动化学报》2014年第10期46-50,68,共6页Proceedings of the CSU-EPSA

基　　金：广东省自然科学基金项目(10151009001000045)

摘　　要：双胞支持向量回归TSVR(twin support vector regression)参数的合理选择严重影响回归结果的准确性。该文采用竞争型智能单粒子算法CISPO(competitive intelligent single particle optimizer)优化参数。CISPO针对智能单粒子算法中各因子值难以确定的问题,在每次迭代中根据待优化参数的变化情况自动选择最佳的因子值,同时引入迭代竞争因子,避免算法前期陷入混乱,而后期又能更好地找到全局最优值。将基于CISPO优化的TSVR模型应用到电力系统短期负荷预测中,结果表明,该方法能有效提高负荷预测的速度和精度。The reasonable choice of twin support vector regression （TSVR） parameters can significantly affect the accuracy of the regression results.The competitive intelligent single particle optimizer （CISPO） is utilized to optimize the parameters in this paper.CISPO aims at the problem that each factor value in intelligent single particle optimizer is intricate to determine and automatically chooses the best factor values in each iteration according to the variation of the parameters to optimize.In the meantime,the iterative competition factors are introduced,which can avoid the algorithm in chaos in the early stage and find the global optimal value better in the following stage.TSVR based on CISPO optimization model is applied to the short-term load forecasting of power system.The results indicate that the proposed method can effectively improve the velocity and precision of load forecasting.

关 键 词：双胞支持向量回归 竞争型智能单粒子算法 短期负荷预测 参数优化 智能单粒子算法 

分 类 号：TM714[电气工程—电力系统及自动化]