基于信息融合技术的磁轴承转子故障诊断  

Fault Diagnosis of Magnetic Rotor Based on Information Fusion Technology

作　　者：范文[1] 孙冬梅[1] 熊鑫[1] 徐海鹏[1] 

机构地区：[1]南京工业大学自动化与电气工程学院,江苏南京211816

出　　处：《仪表技术与传感器》2015年第12期136-140,共5页Instrument Technique and Sensor

基　　金：国家自然科学基金(51277092);江苏省人事厅江苏省博士后资助计划(1201012C);江苏省科技支撑计划项目(BE2011188)

摘　　要：故障诊断信息融合过程可表述为检测层、特征层和决策层的信息融合。文中根据磁轴承转子振动分析的特点,提出了信息融合的故障诊断方案:检测层的融合创新性采用了基于小波分析的加权算法,特征层以希尔伯特-黄变换(HHT)分析法为基础,对边际谱进行特征频段能量的计算,采用BP神经网络对磁轴承转子故障类型进行特征层的识别诊断。决策层采用经典的D-S证据理论,对特征层获得的多个诊断结果做决策融合处理,最终确定磁轴承转子的故障类型。实验结果表明该方法有效地提高了故障诊断结果的可靠性,充分显示了该系统方案的有效性。The process of fault diagnosis based on information fusion can be described as information fusion of detection layer,feature level and the decision-making layer. According to characteristics of the vibration analysis of magnetic bearing rotor,the paper proposed fault diagnosis system solutions based on information fusion technology. Detection layer using the weighted algorithm was based on wavelet analysis. The feature layer is using the HHT algorithm to calculate the characteristic power of marginal spectrum. BP neural network is used as the recognition in feature layer to make a preliminary diagnosis of rotor fault. The decision-making layer uses the classical D-S theory to obtain the final diagnosis result. The feature layer is using the HHT algorithm,to calculate the characteristic power of marginal spectrum. BP neural network is used as the recognition in feature layer,to make a preliminary diagnosis of rotor fault. The decision- making layer uses the classical D-S theory to obtain the final diagnosis result.

关 键 词：磁轴承转子 信息融合 小波加权分析 D-S证据理论 故障诊断 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]