基于改进KNN算法的风电功率实时预测研究  

Wind power real- time prediction research based on the improved KNN algorithm

作　　者：杨茂[1] 贾云彭[1] 穆钢[1] 严干贵[1] 刘佳 

机构地区：[1]东北电力大学电气工程学院,吉林吉林132012 [2]泰安东平供电公司,山东泰安271500

出　　处：《电测与仪表》2014年第24期38-43,共6页Electrical Measurement & Instrumentation

基　　金：国家重点基础研究发展计划项目(973计划)(2013CB228201);国家自然科学基金资助项目(51307017);吉林省科技发展计划项目(20140520129JH);吉林省教育厅"十二五"科学技术研究项目(吉教科合字[2014]第474号);吉林市科技发展计划资助项目(2013625004)

摘　　要：大规模风电并入电网将对电网的规划建设、分析控制以及电能质量等方面产生显著的影响,高精度的超短期风电功率预测可以对含大规模风电电力系统的安全调度和稳定运行提供可靠的依据。文章对风电功率的超短期预测方法进行了研究,以混沌理论为基础,对相空间重构参数进行了计算,提出了基于改进KNN(KNearest Neighbor)算法的风电功率实时预测方法,并且应用多个评价指标来对预测结果进行评价,以吉林西部某风电场实测数据为例,验证了模型的有效性。Integration of large-scale wind power into the power grid will greatly influence grid planning and construc-tion, analysis and control, and energy quality.Accurate short-term wind power forecasting can provide a reliable basis for safety dispatching and stable operation of the power system containing large-scale wind power generating units. This paper studied wind power short-term prediction methods.With the chaos theory as the basis, the parameters for phase space reconstruction were calculated, and a wind power real-time prediction method based on the improved KNN（ K-Nearest Neighbor） algorithm was proposed.Multiple evaluation indexes were applied to evaluate the forecast results, and the effectiveness of the model was verified with the measured data of a wind farm in the west of Jilin as the sample.

关 键 词：风力发电 功率预测 混沌时间序列 相空间重构 C-C方法 KNN算法 

分 类 号：TM71[电气工程—电力系统及自动化]