数据驱动的风电齿轮箱油液监测磨损度量化及其参数权重研究  

Data-driven Study on Oil Monitoring Wear Degree Quantification and Parameters Weight of Wind Power Gear Box

作　　者：徐启圣[1] 王俊[1] 白琨[1] 徐厚昌[1] 张春鹏[1] XU Qisheng;WANG Jun;BAI Kun;XU Houchang;ZHANG Chunpeng(College of Advanced Manufacturing Engineering,Hefei University,Hefei Anhui 230601,China)

机构地区：[1]合肥学院先进制造工程学院,安徽合肥230601

出　　处：《润滑与密封》2020年第11期125-129,共5页Lubrication Engineering

基　　金：安徽省高校优秀青年人才支持计划重点项目(gxyqZD2016276).

摘　　要：机械设备磨损状态的评价参数具有多元化、阶段性及动态性的特点,决策往往取决于技术人员的经验。通过油液监测判断机械设备健康状态时,参数多难以凭主观选择,以致磨损状态的准确界定一直是瓶颈。为此,以风电齿轮箱为对象,研究其颗粒信息、污染度及油品信息对磨损状态的影响,在根据磨粒种类的危害程度初步量化磨损度的基础上,对监测数据降噪后,通过支持向量机获取指标的显著性系数,进而算得参数权重,确定的关键参数为磨损颗粒数量、铁磁性颗粒质量分数、平均颗粒尺寸,简化了磨损度的指标,能提高风电齿轮箱的故障诊断效率。When judging the health state of mechanical equipment through oil monitoring,it is difficult to select the parameters subjectively,so that the accurate demarcation of wear state is always the bottleneck.This is because the evaluation parameters of wear state are diversified and dynamic in different stages,and the final decision depends on the experience of technicians to a large extent.Therefore,the influence of particle information,pollution degree and lubrication oil product information on the wear state of wind power gearbox was studied,and the wear degree was quantified initially.The significance coefficient of the index is obtained based on denoised data by support vector machine,and then the weight of the parameter is calculated.The key factors is chosen as number of wear particles,mass fraction of ferromagnetic particles,average particle size so that the index of wear degree is simplified to improve the fault diagnosis efficiency of wind power gearbox.

关 键 词：风力发电机 齿轮箱 油液监测 磨损度 支持向量机 

分 类 号：TH117.1[机械工程—机械设计及理论]