一种滚动轴承异音特征的综合识别方法  

A hybrid analysis method of abnormal sound signal of rolling bearings

作　　者：张根源[1] 陈芳华[2] 

机构地区：[1]浙江传媒学院电子信息学院,浙江杭州310018 [2]杭州轴承试验研究中心有限公司,浙江杭州310022

出　　处：《振动工程学报》2009年第4期425-429,共5页Journal of Vibration Engineering

摘　　要：介绍了家用电器类滚动轴承音质异音特征的识别方法。通过小波变换原理,对轴承异音信号进行时频分析,采用一种无频带错位的小波包算法进行轴承的异音特征信号提取,清晰地刻画出轴承异音信号的特征函数,提出了基于子频带的K-L信息量的滚动轴承异音分析检测方法,然后以无异音状态为参考状态,利用K-L信息量建模方法进行分析判断,将分析判断的结果用于RBF神经网络的训练,不断地调整神经网络的输入参数,达到逐步提高异音信号判断精度和准确度。该方法同样适用于具有脉冲性质的异常音检测。实验结果表明,该方法在轴承异音探测以及分析方面具有较高的准确度、有效性和工程实用性,较之以往的时域、频域信号处理技术,该方法对异音信号分解更趋合理,是一种可靠和有效的滚动轴承异音特征的综合识别方法。The detect and analysis method of acoustic fidelity of abnormal sound from the rolling bearings in household appliances is proposed. The character of abnormal sound is analyzed in frequency and time domain by wavelet transformation algorithm, which is based on wavelet pack algorithm without frequency band derangement and precisely indicates the character function of abnormal sound. Then the method to analyze and detect abnormal sound of rolling bearings based on K- L information with sub-frequency band is presented, which is used to analyze the abnormal sound by comparing normal one. The result of analysis is used to train RBF network and adjusts RBF network input parameters in order to improve the accuracy of analysis. The method can he used to detect the abnormal sound with pulse. The result of experiments shows that the approach is more accurate, valid and practicable in engineering than other methods in detecting the presence of the abnormal sound. It＇s a reliable and efficient hybrid method in analyzing the abnormal sound from rolling bearing.

关 键 词：滚动轴承 异音 小波分析 RBF网络 

分 类 号：TH133.33[机械工程—机械制造及自动化]