基于AR模型的螺纹连接故障诊断  

Fault Diagnosis of Bolted Joints Based on AR Model

作　　者：张言[1] 曾国英[2] 赵登峰[2] 

机构地区：[1]成都工业学院电加工实验室,四川成都611730 [2]西南科技大学制造科学与工程学院,四川绵阳621010

出　　处：《仪表技术与传感器》2013年第11期78-80,83,共4页Instrument Technique and Sensor

基　　金：国家自然科学基金项目(10876034);中国工程物理研究院联合资助基金项目(10876034)

摘　　要：应用AR(自回归)模型分析法兰螺纹连接结构上下连接法兰间的差值信号,进行螺纹连接故障诊断。首先以正常工况下的差值信号建立系统AR模型,并将AR模型作为滤波器,处理待检状态的差值信号;之后将待检信号输入滤波器,得到滤波后输出的残差序列;在此基础上,利用各种统计量对残差序列进行分析(如Itakura信息距离、偏度、峰度和方差)。结果表明:Itakura信息距离能作为诊断螺纹连接故障的有效值,具有实际应用价值。The AR model was used to analyze the difference signal which between upper and lower shells of flange bolted joints structure for fault diagnose. Initially,AR model was built by difference signals in healthy state. This AR model can be considered as a filter whose output is residual series to deal with difference signals in test state. Then the statistic measurements, such as Itakura distance, skewness, kurtosis and variance, are used to handle the residual series. The result of experiment shows that ltakura dis- tance is a useful eigenvalue of bolted joints fault diagnose.

关 键 词：AR模型 故障诊断 残差序列 Itakura信息距离 

分 类 号：TH131[机械工程—机械制造及自动化]