中文摘要：

针对滚动轴承早期微弱冲击性故障信号特征难以提取的问题,提出了共振解调结合小波包系数熵阈值降噪的综合算法,用于准确确定并提取早期微弱冲击性故障引起的共振调制边频带。该算法应用时延相关和小波包系数熵阈值算法实现信号的双重降噪,并依据共振带能量比确定小波包分解的最佳分解尺度和选取熵阈值的最佳阈值,寻求共振带的最优解,然后进行共振解调提取故障信号特征。实验数据分析结果表明了该算法对滚动轴承早期冲击性故障提取的可行性和有效性。

英文摘要：

Aiming at the difficulty of early weak impactive fault feature extraction of roller bearings, an integrated algorithm based on resonance demodulation method and entropy threshold denoising of wavelet packet coefficients is proposed to accurately determine the resonance modulation bands resulted by early weak impactive fault signals. Firstly, the algorithm uses delayed correlation pretreatment and entropy threshold arithmetic of wavelet packet coefficients to realize double noise reduction of the signals. In the algorithm, the optimal decomposition scale of wavelet packet decomposition and optimal entropy threshold are chosen based on resonance band energy ratio to locate the best solution of resonance bands. Then resonance demodulation is applied to the best resonance bands to extract the fault signal feature. Experiment data analysis results prove the validity and practicability of the proposed method for early weak impactive fault feature extraction of roller bearings.