K-SVD小波降噪在齿轮故障诊断中应用

针对早期齿轮故障诊断中噪声干扰大,故障特征难以提取的问题提出基于K-SVD稀疏表示小波降噪算法。该算法克服传统小波阈值降噪算法只对小波系数进行逐点处理,而忽略小波系数整体架构的缺点,充分考虑小波系数结构特点,在强噪声下仍具有很好稳健性。通过对模拟信号和实测发动机减速器齿轮毂信号分析,证明小波降噪算法正确性和在实际工程应用中的价值。

Application of K-SVD Wavelet Denoising to Gear Fault Diagnosis

As for the problem of strong noise which made it difficult to extract feature at the early stage of gear fault, this paper proposed a new method of wavelet denoising based on K-SVD sparse representation. The proposed method has improved the efficiency of conventional threshold method which dealt with the wavelet coefficients one by one but ignored the whole structure of coefficients. Through the analysis of simulated signal and measured signal of aero-engine gear hub, the validity and value in engineering application of the proposed method has been testified.