基于油液监测的风机主齿轮箱磨损预测  

Wear Prediction of Wind Turbine Main Gearbox Based on Oil Monitoring

作　　者：许少凡 李秋秋 覃楚东 杨智宏 何伟楚 XU Shaofan;LI Qiuqiu;QIN Chudong;YANG Zhihong;HE Weichu(Equipment Lubrication and Testing Research Institute,Guangzhou Mechanical Engineering Research Institute Co.,Ltd.,Guangzhou Guangdong 510000,China)

机构地区：[1]广州机械科学研究院有限公司设备润滑与检测研究所,广东广州510000

出　　处：《润滑与密封》2022年第4期183-188,共6页Lubrication Engineering

基　　金：国家重点研发计划项目(2018YFB2001604);广东省科技计划项目(2020B1212070022)。

摘　　要：针对风电机组主齿轮箱运行过程中因润滑不良导致异常磨损等故障的问题,提出偏最小二乘回归分析(PLS)方法来处理油液监测数据,并采用灰色等维递补模型进行故障预测。通过对油液监测数据中的运动黏度、酸值和铜元素、硅元素、锌元素、磷元素以及铁元素含量进行数据初值化与主成分提取,构建PLS模型并进行数据拟合分析。结果表明:运用偏最小二乘回归分析方法所拟合出的线性回归方程可以较为准确地预测齿轮箱中的异常磨损元素含量。使用灰色等维递补模型对数据进行预测,PLS模型进行诊断分析,可得出未来一段时间Fe元素含量变化趋势,为制定齿轮箱维护策略提供依据。Aimed at the problems of abnormal wear caused by poor lubrication during the operation of the main gearbox of wind turbines, partial least square regression analysis(PLS)was proposed to process oil monitoring data, and the grey equidimensional grading model was used for fault prediction and diagnosis.Through the data initialization and principal component extraction of kinematic viscosity, acid value and the content of copper, silicon, zinc, phosphorus and iron elements in the oil monitoring data, the PLS model was constructed and the data fitting analysis was carried out.The results indicate that the linear regression equation fitted by the PLS method can accurately predict the abnormal wear element content in the gear box.Using the grey equidimensional grading model to predict the data and PLS model for diagnostic analysis, the change trend of Fe content in the future can be obtained, which can provide the basis for formulating the gearbox maintenance strategy.

关 键 词：风电机组 主齿轮箱 油液监测 偏最小二乘回归分析 灰色等维递补模型 故障诊断 

分 类 号：TH17[机械工程—机械制造及自动化]