故障分析中的小波软阈值改进除噪方法

软阈值消噪是信号消噪中的标准算法。故障检测中的信号去噪,要求在降低噪声水平的同时,保留信号中用于故障检测的奇异特征。通过分析信号和噪声的小波系数在小波空间的不同特性,在"噪声强度估计"、"阈值选择策略"和"小波系数调整策略"三个方面对软阈值除噪技术进行了改进。仿真研究表明改进方法较好地解决了保护信号局部特征和抑制噪声之间的矛盾,并对各种类型的故障信号均能进行有效的分析。

Denoising by wavelet soft-thresholding in fault analysis

Denoising by soft-thresholding has been a standard algorithm in signal denoising. In fault measurement, it not only requires to reduce the noise, but also to reserve the singularity of signal. By analysing the different characteristics of signal and noise in wavelet analysis,this paper proposes an improved algorithm to enhance the denoising technique of wavelet soft-thresholding in estimating noise intensity, selecting threshold and adjusting wavelet coefficients. Simulations show that the new method solves the conflict commendably between protection of the local signals characteristics and restriction of noise, and can make proper analysis for different fault signals.