基于多尺度样本熵的往复式压缩机故障诊断方法研究

Study on Fault Diagnosis Method of Reciprocating Compressors Based on Multi-scale Sample Entropy

往复式压缩机工作时的振动信号非常复杂,对采样数据经过替代数据法和最大Lyapunov指数分析后发现具有一定的非线性和弱混沌性,这样的数据特征在其故障诊断时较适合采用样本熵(Sample Entropy,SampEn)进行处理。为实现往复式压缩机振动故障诊断,研究了振动信号符号化处理方法,提出了符号序列的SampEn算法,分析了不同时延参数下SampEn值曲线的几何特征,并利用该几何特征构建了故障的标准模板,通过该模板实现了故障诊断。实验证明,利用SampEn可以进行压缩机典型故障的诊断,该方法具有工程实用价值。

The vibration signals are very complex and nonlinear in the reciprocating compressor working process . Based on surrogated date method and max Lyapunov exponent method , it shows weak nonlinearity and chaos . Such data character in its fault diagnosis is more suitable for the sample entropy models .For the realization of the reciprocating compressor vibration fault diagnosis , it presents the method to transfer dynamic signals into symbol-ic sequence , proposes SampEn algorithm based on symbolic sequence and the geometry characters of SampEn value.Experiments show that the proposed method is a valid tool to distinguish different fault diagnosis of recip -rocating compressors .