基于听觉模型和极值点概率密度的断齿故障特征提取方法研究

摘要

齿轮断齿故障的重要特征是啮合过程中在断齿处产生碰撞与冲击。考虑到人耳听觉系统对于突发的瞬态声信号具有本能的反应，为提取断齿故障诱发的瞬态冲击响应成分，提出一种基于听觉模型和信号极值点概率密度的特征提取方法。该方法首先对信号进行 GT 带通滤波、相位调整及极值点提取，然后计算各极值点的幅值概率密度，通过对其求导判断各滤波通道中是否存在瞬态冲击成分，继而提取与之相关的极值点。同时，由于系统振动时会产生与断齿冲击无关的极值点，为准确提取断齿冲击，根据瞬态信号频带连续性和多频段分布特点，设计了相应的提取方法。经实测信号验证表明，所提方法能准确刻画及提取断齿故障特征，可以在含有多种类型的瞬态冲击响应成分中提取出只由断齿故障所诱发的冲击成分，且提取结果精确度较高。%The collision and impact at gear teeth fractures are the most important features in meshing process of gear pairs.Considering a human auditory system has an instinctive response to sudden transient acoustic signals,in order to extract transient impulse response components induced by gear teeth fracture faults,a method extracting fault features of gear teeth fractures based on an auditory model and signal probability density of extreme points was proposed.Firstly, band-pass filtering with Gammatone filters ,phase adjustment and extreme points extraction for signals were conducted, and then the amplitude probability densities of extreme points were calculated,their derivatives were used to judge if there are transient impact components in all filtered signals.Those extreme points with transient impact components were extracted.Meanwhile,the whole system vibration might produce extreme points being irrelevant to gear teeth fracture impacts.In order to accurately extract impacts of gear teeth fractures,according to transient signals'frequency band continuity and multi-band distribution characteristics,an appropriate extraction method was designed.The real measured signals showed that the proposed method can accurately extract fault features of gear teeth fractures,and can extract impact components only induced by gear teeth fracture faults in a variety of transient impulse response components,and the extraction accuracy is higher.