基于自适应非抽样提升小波的故障特征提取方法研究  

Study on Adaptive Non-decimated Lifting Wavelet Packet Method for Failure Feature Extraction

作　　者：孙如田[1] 

机构地区：[1]克拉玛依职业技术学院,新疆克拉玛依834000

出　　处：《自动化与仪器仪表》2016年第7期147-149,共3页Automation & Instrumentation

基　　金：国家自然科学基金(61304211)

摘　　要：针对传统小波在处理非线性往复式机械故障时存在的能量损耗和虚假频率的缺点,本文结合自适应策略和非抽样算法对现有提升小波算法进行改进,并利用仿真和实验验证了新算法在处理虚拟数据集以及实际信号的有效性。室内实验及现场应用表明:新方法可显著消除虚假频率,可在噪音污染信号中提取关键特征,改进后的算法能有效应对往复和旋转机械故障的特征提取。This paper aims at removing the drawbacks of traditional wavelet transform in analyzing nonlinear reciprocating machinery failures,like energy losses and false frequencies. We combined the self- adapting strategy and non- decimating algorithm as an improvement in existing lifting scheme wavelet packet transform. Utilizing the Matrix Laboratory software,we tested and verified the performance of this new method in processing virtual datasets and actually collected signal. The spectrograms and tables in example and application indicate that this new method can obviously eliminate the false frequencies and extract the crucial characteristics in contaminated signals. And this property will enable it to be used in case of reciprocating and rotating machinery failure diagnosis.

关 键 词：提升小波 自适应算法 特征提取 

分 类 号：TH17[机械工程—机械制造及自动化]