复合无量纲指标在旋转机械故障分类中的应用  

The application of a compound dimensionless parameter for fault classifying of rotating machinery

作　　者：覃爱淞[1,2] 张清华[2] 李铁鹰[1] 胡勤[2,3] 

机构地区：[1]太原理工大学信息工程学院,太原030024 [2]广东石油化工学院计算机与电子信息学院,茂名525000 [3]广东工业大学自动化学院,广州510006

出　　处：《现代制造工程》2013年第4期10-14,121,共6页Modern Manufacturing Engineering

基　　金：国家自然科学基金项目(61174113);广东省自然科学基金项目(8152500002000011)

摘　　要：针对传统的无量纲指标在旋转机械故障分类中的不足,利用遗传编程方法对传统的无量纲指标进行重新组合和优化来构建新的无量纲指标,以分类效果作为判断新无量纲指标优劣程度的准则,通过该方法最终获得了一个具有最佳分类能力的指标,并用于旋转机械轴系故障诊断。通过试验验证,与传统的无量纲指标相比,经过遗传编程构建的新无量纲指标对轴系故障具有更好的识别能力,能够对轴系故障进行准确分类,可以应用于实际诊断中。In order to overcome the disadvantages of traditional dimensionless parameters in rotating machinery fault classification,a new method based on genetic programming is proposed to construct the new dimensionless parameters. According to this method, simple dimensionless parameters are combined and compound dimensionless parameters are formed, then fitness function is adopted to measure the performance of new generated indexes. With this proposed method, the dimensionless parameter having best classification ahility could be obtained for shaft fault diagnosis of rotating machinery. Experimental result shows that the recognition ability of optimum dimensionless parameter is better than that of existing ones, and accurate classification of shaft fault can be achieved, and the optimum dimensionless parameter can be applied in the actual diagnosis.

关 键 词：旋转机械 遗传编程 故障诊断 无量纲指标 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]