基于递归图分析的压缩机故障诊断方法研究

摘要

Recurrence plot of the vibration signal is researched by studying the internal regularity and characteristic based on recurrence quantification analysis effectively for extracting the nonlinear feature of vibration system of reciprocating natural gas compressor under the fault working condition.The classification and recognition method based on the extracted feature is applied in diagnosis of compressor fault successfully.Simulations and experiments were worked based on some samples of 3 types of typical faults.Results show that the feature parameter which is extracted by this method has nice classification performance,and shows nice performance in the application of fault diagnosis.%通过有效提取往复式天然气压缩机在故障工作状态下机械振动系统的非线性特征量,对振动信号的递归图进行定量分析,研究递归图内部规律性和特征,对提取特征量的分类识别成功应用于对某型压缩机的故障诊断.仿真实验以三类典型故障类别一定样本数量的振动数据为研究对象进行,仿真结果表明定量递归分析提取的特征量具有很好的聚类性,能很好地把三类故障区分,达到对故障诊断的目的.