基于谱熵及分形理论的压缩机气阀故障诊断

Fault diagnosis on compressor valves based upon the spectrum entropy and fractal theory

压缩机组振源较多,振动信号背景噪声强烈、干扰大,通过检测振动信号细微特征变化识别故障征兆异常困难。为此,从系统特性的角度出发,选择信息熵及分形维数作为特征参数,提取气阀无故障、轻微漏气与严重漏气3种典型故障信号特征,采用聚类分析方法来判断气阀故障。应用表明,较之传统的诊断方法,基于谱熵及分形理论的诊断方法具有故障特征提取工作量小、容错性强、准确率高的特点。

The compressor operating process is unstable and the operating condition is different,so the traditional fault diagnosis by detecting the faulty symptom is extremely difficult. The information entropy and fractal theory were used to analyze the system behavior of the signal under three representative different states,and then apply cluster analysis to diagnose the failure. The application demonstrates this method has more advantages than the conventional methods,such as simple operation,fault-tolerance,high accuracy.