双自寻优小波去噪方法及其在滚动轴承故障诊断中的应用

Wavelet based de-noising double self-optimizing method and its application to ball bearing fault diagnosis

阈值和阈值函数的选取是小波阈值去噪方法中影响去噪效果的最重要的两个因素.经典方法与后续诸多改进方法都存在两者之间相互选取无法解决耦合性的问题,难以在噪声环境中提取有用的滚动轴承故障信号.在论述了已有方法中存在的耦合性和数与数组合的问题后,提出了一种双自寻优的方法解决这种问题.双自寻优方法中把所有阈值和阈值函数的参数可能性相互组合,在这些组合中找出最优去噪组合.阈值的寻优范围由信号自身的高频死去值决定,文中给出了解决方案并予以证明,阈值函数的选取本文以带参数的软硬阈值折衷法为基准,其参数寻优范围为0到1.通过模拟轴承信号和真实轴承故障信号的实验结果表明,此算法得到了更高的信噪比和更好的图像外观.

Threshold and threshold function are the key factor in threshold based wavelet de- noising. There exists coupled one-sidedness between classical and subsequent many improve- ment methods, and it difficult to extract useful rolling bearing fault signal. This paper put forward a double self-optimizing algorithm to solve the problem which existed methods exist coupled and the problem of number combination. The selection of threshold and threshold function influence each other in double self-optimizing algorithm, which can produce a varie- ty of combination and find the optimal de-noising parameters. Threshold optimal range is de- cided by the dead value of high frequency. This paper gives the solvable methods and proves it. Threshold function set benchmark for the compromise of the hard and soft threshold method and the optimal de-noising parameters is 0 to 1. The experimental results indicate that the method is highly effective in noise reduction and fault feature extraction.