关键词: 齿轮, 二进小波变换, 新阈值函数, 动态阈值, 信号重构

Abstract: In the process of gear laser detection, due to the influence of various noise sources, the phenomenon of the original signal weakening of feature information will appear, which impacts the accurateness of gear detection. To solve this problem, a new de-noising method of combining a new threshold function and dyadic wavelet transform is proposed based on the analysis of the traditional soft and hard threshold function. In this method, adjustable threshold function is built. The threshold adjusts dynamically in the form of index, with the change of binary wavelet coefficients, overcoming the discontinuity of the hard threshold and the offset of the soft threshold. By simulation experiments, the signal to noise ratio and applicability of the method have significantly improved.

Key words: gear, dyadic wavelet transform, new threshold function, dynamic threshold, signal reconstruction