基于时延相关及小波包系数熵阈值的增强型共振解调方法

Enhanced Resonance Demodulation Based on the Delayed Correlation and Entropy Threshold of Wavelet Packet Coefficients

针对滚动轴承等旋转机械早期微弱损伤性故障,提出一种基于时延相关及小波包系数熵阈值的增强型共振解调方法,以解决传统共振解调技术中共振带难以确定的问题。该方法对原始故障信号做时延相关预处理,然后进行小波包分解,求出各叶结点系数香农熵,依据熵值谱图,设定有效阈值来取舍叶结点系数,重构信号并做频谱分析,有效确定并提取故障信号共振带,对共振带进行包络解调分析,有效提取出机械设备早期微弱损伤性故障的冲击特征。滚动轴承试验数据分析结果表明该方法的可行性和有效性。

To the early weak damage of the roller bearing and other rotating machinery,an enhanced resonance demodulation method based on delayed correlation pretreatment and entropy threshold of wavelet packet coefficients is proposed to solve the difficulty in determining the resonance bands in the traditional resonance demodulation technique.In this method,the weak damage signal should be preprocessed by using delayed correlation analysis method firstly.Then wavelet packet transform is carried out to find the Shannon entropy of all leaf nodes coefficients,an effective threshold is set on the basis of the entropy spectrum to choose the leaf nodes coefficients,reconstruct the signal,then do spectrum analysis,and extract the resonance modulation bands effectively.Then use the envelope demodulation method,and successfully extract the early impact characteristics of a weak damage signal of the machinery.The analysis results of roller bearing experimental data confirm the feasibility and validity of this method.