基于支持向量机的风电偏航回转支承故障诊断  

Fault diagnosis of wind power yawing slewing bearing based on support vector machine

作　　者：钮满志 陈捷[1] 封杨[1] 王华[1] 

机构地区：[1]南京工业大学机械与动力工程学院,江苏南京210009

出　　处：《南京工业大学学报（自然科学版）》2014年第1期117-122,共6页Journal of Nanjing Tech University(Natural Science Edition)

基　　金：国家十二五科技支撑计划(2011BAF09B02);国家自然科学基金(51105191)

摘　　要：针对风电回转支承故障样本少、信号微弱且不易提取的特点,提出一种基于小波能谱和支持向量机相结合的故障诊断方法。采用加速度信号的小波能谱与温度、扭矩信号组合构成特征向量,用支持向量机对正常、单个螺栓断裂、多个螺栓断裂3种状态进行分类识别,结果分类准确率都达到100%。样本不变,采用BP神经网络方法分类的准确率分别为84%、92%和80%。结果表明,支持向量机方法比BP神经网络更适用于风电回转支承的故障诊断。A fault diagnosis method based on wavelet energy spectrum and support vector machine （SVM） was proposed in terms of less fault samples, weakness of signals and difficulty to be extracted of wind power slewing bearing.Feature vectors were constructed by combining wavelet energy spectrum of acceleration signal with temperature and torque signal.And three normal states： single bolt fracture and multiple bolt fracture were classified by using SVM with classification accuracy of 100%,while the classification accu- racies for the same samples reached only 84%, 92%, and 80%, respectively by the BP neural network method.Results show that SVM method is more suitable than BP neural network method for wind power slewing bearing fault diagnosis.

关 键 词：支持向量机 回转支承 故障诊断 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]