徐可君，任　帅，秦海勤，李志农

(1.海军航空工程学院青岛校区航空机械系,青岛　266041； 2.南昌航空大学测试与光电工程学院,南昌　330063)

第一作者徐可君男，教授，博士，1963年生

基于QGA和SK的滚动轴承故障诊断新方法

徐可君1，任帅2，秦海勤1，李志农2

(1.海军航空工程学院青岛校区航空机械系,青岛266041； 2.南昌航空大学测试与光电工程学院,南昌330063)

摘要：基于量子遗传算法收敛速度快、全局寻优能力强的特点，将谱峭度法和量子遗传算法相融合，提出一种滚动轴承故障诊断的新方法。该方法以经初始带通滤波后的原信号的谱峭度作为量子遗传算法的适应度函数，重新优化设计最优滤波器，以设计的最优带通滤波器对原始信号滤波并进行包络分析，从而实现轴承故障的诊断。实测数据分析表明，方法所得峭度谱值明显增大，包络谱线更加干净，故障特征频率更加明显。同时研究发现，方法诊断精度几乎不受传感器位置的影响。

关键词：滚动轴承；量子遗传算法；谱峭度；包络分析；故障诊断

基金项目：国家自然科学基金资助项目(50775208，51075372)

收稿日期：2014-03-10修改稿收到日期：2014-05-16

中图分类号:TH133

文献标志码：A

DOI:10.13465/j.cnki.jvs.2015.11.019

Abstract:Based on rapid convergence and good global search capacity, a new method for rolling bearing fault diagnosis was presented by combining quantum genetic algorithm(QGA) and spectral kurtosis(SK). The original signal’s SK gained by the primary band-pass filtering was taken as the fitness function of QGA to redesign the optimal band-pass filter. Then the original signal was filtered with the optimal band-pass filter redesigned. The characteristic frequencies of rolling bearing faults could be nicely obtained with the envelope analysis. The method was used to analyze actual testing signals. The results showed that the SK values obtained with the new method increase obviously, the envelope spectral lines of the filtered signals and the characteristic frequencies of faults are much clearer than those of the traditional SK results; the accuracy of the new method is not dependent upon the positions of sensors.

Rolling bearing fault diagnosis based on quantum genetic algorithm and spectral kurtosis

XUKe-jun1,RENShuai2,QINHai-qin1,LIZhi-nong2(1. Department of Aviation Mechanism, Qingdao Branch of Naval Aviation Engineering Institute, Qingdao 266041, China;2. School of Measuring and Optical Engineering, Nanchang HangKong University, Nanchang 330063, China)

Key words:rolling bearing; quantum genetic algorithm (QGA); spectral kurtosis (SK); envelope analysis; fault diagnosis

包络分析是滚动轴承故障诊断中最常用的方法之一[1]，该方法不仅能给出轴承的损伤程度，而且能进一步指出轴承的损伤部位[2]。在包络分析中带通滤波参数的选择至关重要，严重依赖操作人员的专业知识和以往的经验积累，得出诊断结论不仅费时，且可靠性不高。峭度作为一种统计量，在噪声较小时可以基于对奇异信号的敏感性进行系统的异常检测，但峭度作为一个全局指标无法反映信号某一分量的变化。……