基于S变换谱核密度估计的齿轮故障诊断

发布时间：2018-09-09 15:38

【摘要】：针对齿轮在故障损伤状态下的振动信号,提出一种基于S变换谱二维核密度估计的冲击特征提取方法,以实现齿轮的故障诊断。该方法首先对包含冲击特征的振动信号进行S变换;然后将S变换谱乘以一个系数后圆整,得到一个整数矩阵;最后以S变换谱的时间和频率构成一个二维随机变量,以整数矩阵中的元素值作为二维随机变量各个采样样本的个数,对二维随机变量进行核密度估计,并最终得到一个二维核密度函数。该核密度函数相当于由S变换谱经过一次平滑去噪的过程获得,其中的噪声得到了有效的抑制,而冲击特征则得到了加强与突显。仿真振动信号和齿轮箱故障振动信号的分析结果表明,该方法能够有效地强化并提取出振动信号中周期性的冲击特征,从而实现齿轮箱相关故障的诊断。
[Abstract]:Aiming at the vibration signal of gear under the condition of fault damage, a method of extracting impulse feature based on S-transform spectrum two-dimensional kernel density estimation is proposed to realize the fault diagnosis of gear. The method firstly transforms the vibration signal containing shock characteristics, then multiplies the S transform spectrum by a coefficient to obtain an integer matrix. Finally, a two-dimensional random variable is constructed by the time and frequency of the S-transform spectrum. Taking the element value in the integer matrix as the number of sampling samples of two-dimensional random variables, the kernel density of two-dimensional random variables is estimated and a two-dimensional kernel density function is obtained. The kernel density function is equivalent to the S transform spectrum through a smooth de-noising process, in which the noise is effectively suppressed, while the impact feature is strengthened and highlighted. The analysis results of simulated vibration signals and gearbox fault vibration signals show that this method can effectively enhance and extract the periodic shock characteristics of vibration signals and thus realize the diagnosis of gearbox related faults.
【作者单位】：浙江工业大学之江学院;浙江大学浙江省先进制造技术重点研究实验室;浙江工业大学特种装备制造与先进加工技术教育部重点实验室;
【基金】：国家自然科学基金(51275453,51505424) 浙江省自然科学基金(LQ17E050006,LY15E050019)项目资助
【分类号】：TH132.41

【参考文献】