基于假设检验和支持向量机的旋转机械故障诊断方法

摘要

A fault diagnosis method was proposed based on adaptive statistic test filter (STF) and SVM for condition diagnosis of rotating machinery to extract weak fault features and identify fault types.STF was based on the statistic of the hypothesis testing in the frequency domain to evaluate similarity among reference signals (noise signal) and original signals,and remove the components of high similarity.The optimal level of significance a was obtained by using particle swarm optimization(PSO).To evaluate the performances of the STF,evaluation factor IpqWaS also defined.Finally,a sequential diagnosis method,using sequential inference and SVM was also proposed,by which the conditions of rolling bearings might be identified sequentially.Practical examples of fault diagnosis for structural faults often occurring in the shafts,such as unbalance,misalignment states were shown to verify that the method is effective.%针对旋转机械故障诊断中存在的早期非平稳微弱故障信号特征提取困难、故障诊断不准确等问题,提出了一种基于自适应假设检验滤波和支持向量机(SVM)的故障诊断方法.该方法采用统计学假设检验原理来评估参考信号(噪声信号)和原始信号(故障信号)在频域上的相似性,删除具有高相似性的频域成分;通过粒子群优化算法获得最佳的显著性水平α;定义评估因子Ipq来评价假设检验滤波的效果.最后通过SVM来逐次诊断轴系构造异常.验证结果表明该方法能够有效地诊断出传动轴不对中和不平衡的故障类型.