基于数据特性分析的风电机组叶片结冰辨识  

Identification of wind turbine blade icing based on data characteristic analysis

作　　者：董兴辉[1] 张劲草 李佳 高迪 杨志凌[1] Dong Xinghui;Zhang Jingcao;Li Jia;Gao Di;Yang Zhiling(l.School of Energy Power and Mechanical Engineering,North China Electric Power University,Beijing 102206,China;Beijing Municipal Lighting Management Center,Beijing 100078,China;China Electric Power Research Institute,Beijing 100192,China)

机构地区：[1]华北电力大学能源动力与机械工程学院,北京102206 [2]北京市城市照明管理中心,北京100078 [3]中国电力科学研究院有限公司,北京100192

出　　处：《可再生能源》2023年第1期53-59,共7页Renewable Energy Resources

基　　金：国家重点研发计划资助项目(2017YFE0109000)。

摘　　要：风电机组叶片结冰影响叶片的气动特性,降低机组的输出功率并形成安全隐患。及时、准确判定叶片结冰,有利于提高风电机组运行可靠性和做出运维应对决策。文章通过分析风电能效指标变化特性、部件振动特性、测风仪测值波动特性,选取与叶片结冰相关的SCADA中耦合变量,利用SVR回归算法挖掘耦合参数的历史样本特征值,构建叶片结冰辨识模型。现场数据验证表明,基于SCADA实时数据分析与设备监测结果完全一致,判定准确。模型只通过对实时数据分析判断,无需额外增设硬件,具有较好的实用价值。Icing affects the aerodynamic characteristics of wind turbine(WT)blades,reducing the output power of wind turbine and causing potential safety hazards.Timely and accurate determination of blade icing is conducive to improving the operation reliability of WTs and making scientific operation and maintenance decisions.By analyzing the variation characteristics of wind energy efficiency indicators,vibration characteristics of components,and fluctuation characteristics of measured values of wind gauges,Supervisory Control And Data Acquisition(SCADA)coupling variables reflecting blade icing were selected,and the historical sample characteristic values of coupled parameters were excavated by Support Vector Regression(SVR)regression algorithm to construct the identification model of blade icing.The field data verification shows that the real-time data analysis based on SCADA is consistent with the equipment monitoring results,and the identify icing is accurate.The model in this paper only analyzes and judges the real-time data without additional hardware,so it has a good practical value.

关 键 词：叶片结冰 辨识模型 输出功率 风能利用系数 塔架振幅 

分 类 号：TK81[动力工程及工程热物理—流体机械及工程]