基于CEEMD能量熵特征提取和VNWOA-LSSVM的风力机轴承故障诊断方法研究  

Fault diagnosis for wind turbine bearings based on CEEMD energy entropy and VNWOA-LSSVM

作　　者：万晓静[1] 孙文磊[1] 陈坤 WAN Xiao-jing;SUN Wen-lei;CHEN Kun(School of Mechanical Engineering,Xinjiang University,Urumqi 830047,China)

机构地区：[1]新疆大学机械工程学院,新疆乌鲁木齐830047

出　　处：《机电工程》2020年第10期1186-1191,共6页Journal of Mechanical & Electrical Engineering

基　　金：国家自然科学基金资助项目(51565055,51765062);新疆大学博士生科技创新项目(XJUBSCX-2014019)。

摘　　要：针对极端复杂工况下风力发电机组轴承故障诊断问题,对风力机运行状态监测中常用的故障诊断方法进行了研究,提出了一种基于互补总体经验模式分解能量熵的故障特征提取和改进的鲸鱼算法来优化最小二乘支持向量机的风力机轴承故障诊断方法;通过互补总体经验模式分解,降低了噪声对微弱故障信号的干扰,提取了各分量的能量熵构建故障特征集合,作为诊断模型的输入;利用冯诺依曼拓扑结构的特性,克服了鲸鱼算法中收敛慢、寻优精度低的问题,构建了改进的鲸鱼算法优化最小二乘支持向量机的诊断模型分类器,实现了对不同故障类型特征参数的准确分类;最后利用试验数据集进行了测试。研究结果表明:所提出的方法计算速度快、泛化能力强、分类正确率高,其诊断结果优于基于鲸鱼算法优化的最小二乘支持向量机,远优于传统的最小二乘支持向量机算法。Aiming at the problem of fault diagnosis of wind turbine bearing under extremely complicated working conditions,the fault diagnosis methods commonly used in the condition monitoring of wind turbine operation were studied,and the fault diagnosis method for wind turbine bearings based on CEEMD energy entropy and VNWOA-LSSVM was proposed.CEEMD method was used to reduce the interference of noise on weak fault signal,energy entropy of each IMF was extracted to construct fault feature set and serve as the input of diagnosis model.Von Neumann structure was used to overcome the problems of slow convergence and low optimization accuracy in WOA algorithm,and the VNWOA-LSSVM diagnostic model classifier was constructed to realize the accurate classification of characteristic parameters of different fault types.The results indicate that the fault diagnosis method proposed has fast computing speed,strong generalization ability and high classification accuracy,and its diagnosis result is better than WOA-LSSVM,far better than the traditional LSSVM method.

关 键 词：风力机轴承 互补总体经验模式分解 能量熵 冯诺依曼拓扑结构优化鲸鱼算法 最小二乘支持向量机 

分 类 号：TH133.3[机械工程—机械制造及自动化] TP301.6[自动化与计算机技术—计算机系统结构]