小样本条件下基于SGMM模型的滚动轴承故障诊断研究

Research on Rolling Bearing Fault Diagnosis Based on SGMM Model under Small Sample Condition

由于机械设备故障时间短,信号捕获难度高等原因的存在,会导致小样本故障集出现,但小样本往往是机械故障诊断的关键;针对小样本条件下传统滚动轴承故障诊断诊断算法存在检测率偏低等问题,提出一种基于SGMM模型的滚动轴承故障诊断算法;先确定与故障建模策略相关的提取任务,预估潜在的机械故障状态变化;对故障信号进行变分模态分解,得到最小熵解卷积结果,并满足端点效应的处理需求,实现对机械故障位置的精确定位与诊断;实验结论表明,SGMM模型更注重对故障脉冲成分的连续检测,在以峭度作为衡量标准的条件下,该方法增强故障冲击力的作用更强,能更早诊断出轴承类机械元件的早期故障状态,平均故障检测率能够达到99.4%。

Due to the short fault time of mechanical equipment and high difficulty of signal acquisition,it leads to small sample fault sets to occur,but small samples are often the key to mechanical fault diagnosis.A rolling bearing fault diagnosis algorithm based on subspace Gaussian mixture model(SGMM)model is proposed to address the problem of low detection rate in traditional rolling bearing fault diagnosis algorithms under small sample conditions.Firstly,identify the extraction tasks related to the fault modeling strategy and estimate potential changes in mechanical fault status;The variational modal on the fault signal is decomposed to obtain the minimum entropy deconvolution result,meet the processing requirements of endpoint effects,and realize the precise positioning and diagnosis of mechanical fault locations.The experimental conclusion shows that the SGMM model places more emphasis on continuous detection of fault pulse components.The kurtosis is taken as a measurement standard,this method has a stronger effect on enhancing fault impact and can diagnose the early fault status of bearing mechanical components,with an average fault detection rate of 99.4%.