振动信号的自适应非抽样提升小波降噪方法  

The adaptive non-sampling lifting wavelet noise reduction method of vibration signals

作　　者：胡振国[1] 张来斌[1] 段礼祥[1] 陈敬龙[1] 

机构地区：[1]中国石油大学(北京)机电学院

出　　处：《石油机械》2011年第5期60-63,97,共4页China Petroleum Machinery

基　　金：国家"863"计划项目资助课题"基于双扭环机制的输油管线泄漏诊断的新装置与方法研究"(2008AA06Z209);中国石油天然气集团公司中青年创新基金项目"往复压缩机剩余寿命的混沌关联预测方法研究"(07E1005)

摘　　要：振动信号是机械故障诊断中应用最广泛的信号,为了能够提取隐含在强噪声背景下振动信号的故障特征,设计了一种自适应非抽样提升混合小波。该方法利用LMS自适应法则构造了可与提升小波更好匹配的初始预测器和更新器,在计算预测差值平方和时考虑所有样本的预测信息,以锁定第2代小波信号的局部特征。试验和工程振动信号分析表明,使用该方法降噪的信号获得了较高的信噪比和较小的均方差,且能较理想地提取出故障特征。The vibration signal is the most extensively used signal in diagnosis of mechanical failure.To extract the fault characteristic of vibration signals hidden in the strong noise background,a kind of adaptive non-sampling lifting mixed wavelet was designed.The method adopted the LMS adaptive principle to construct the initial predictor and updater which could better match lifting wavelet.The prediction information of all samples was considered in calculating the prediction difference sum of squares so as to lock the local feature of the second-generation wavelet signals.The test and engineering vibration signals showed that the signals which used this method to reduce noise obtained relatively high signal-to-noise ratio and relatively small mean square deviation and the fault characteristic could be extracted desirably.

关 键 词：自适应 非抽样提升小波 振动信号 降噪 特征提取 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]