基于支持向量机的风电机组故障预测  

Fault Prediction of Wind Turbine Based on Support Vector Machine

作　　者：李森娟 张萍 岳大为 王秋富 ZHANG Ping;LI Sen-juan;YUE Da-wei;WANG Qiu-fu(School of Artificial Intelligence,Hebei University of Technology,Tianjin 300130,China)

机构地区：[1]河北工业大学人工智能与数据科学学院,天津300130

出　　处：《计算机仿真》2022年第5期84-88,180,共6页Computer Simulation

基　　金：国家自然科学基金资助项目(51207042)。

摘　　要：针对风电机组部件故障导致的计划外或事后维护会导致长时间的风电机组停运,从而导致运营成本上升,因此,提出基于支持向量机的风电机组故障预测。首先通过对风电场数据采集与监视控制(Supervisory Control And Data Acquisition,SCADA)系统的数据进行预处理,筛选出合适的SCADA数据,结合功率曲线过滤和分析故障和报警数据;其次,从参数阈值以及维修记录等角度对数据集进行标注;然后应用支持向量机(Support Vector Machine,SVM)分类技术,通过分析其它数据,如温度、桨距角和风速数据,识别故障和无故障运行;最后将其扩展到在特定故障发生之前进行预测和诊断。结果表明在预测某些特定类型的故障方面预测精度高达92%,可以检测早期故障并根据需要安排维护,从而避免不必要的定期检查。In view of the unplanned or post-maintenance caused by the failure of wind turbine components,it will lead to long-term wind turbine shutdown,which will lead to serious economic losses.Therefore,a support vector machine-based wind turbine failure prediction is proposed.First,by preprocessing the data of the wind farm Supervisory Control And Data Acquisition(SCADA) system,the appropriate SCADA data were filtered out,and the fault and alarm data were filtered and analyzed in combination with the power curve.Second,the data were labeled from the perspective of parameter thresholds and maintenance records.Then SVM classification technology was applied to identify faults and fault-free operation by analyzing other data,such as temperature,pitch angle and wind speed data.Finally,it was extended to predict and diagnose before specific faults occur.The results show that the evaluation results in predicting certain specific types of failures reach 92%.Early failures can be detected and maintenance can be scheduled as needed,thereby avoiding unnecessary periodic inspections.

关 键 词：风电机组 数据采集与监视控制系统 故障预测 支持向量机 

分 类 号：TM743[电气工程—电力系统及自动化]