基于多传感器信息融合的滚动轴承故障诊断研究

Rolling Bearing Fault Diagnosis Research Based on Multi-sensor Information Fusion

由于单一传感器所包含的故障信息不能全面地反映滚动轴承的故障状态,提出了一种基于多传感器信息融合的滚动轴承故障诊断方法。首先,利用不同位置的加速度传感器采集滚动轴承故障振动信号,经集成经验模态分解(EEMD)后,前8个固有模态分量(IMF)的能量值作为分类器支持向量机(SVM)的输入故障特征参量;其次,利用故障特征参量训练分类器SVM,并对测试样本进行分类,实现故障的初步分离;然后,根据混淆矩阵获得各分类器的全局可信度和局部可信度,并与各测试样本的后验概率输出结合实现DS证据理论中基本概率分配函数的赋值;最后,利用DS证据理论实现融合以获得最终诊断结果。试验结果表明:提出的方法可有效融合不同传感器的故障信息,最大限度地避免误诊现象。

Since the fault information obtained by the singular sensor could not reflect the fault condition of rolling element bearing completely,a new fault diagnosis method of rolling element bearing based on the multi-sensor information fusion was proposed in this paper. Firstly,three acceleration sensors of different locations were utilized to collect the vibration signals of rolling element bearing. The first eight energy value of IMF components extracted by ensemble empirical mode decomposition（ EEMD）were utilized as the input features of support vector machine（ SVM） respectively. Secondly,the fault features of train samples were utilized to train SVM,which were utilized to classify the test samples following. Then the primary fault diagnosis was realized. Thirdly,the global reliability and local reliability of each classifier were obtained from the confusion matrix. The basic probability assignment of each evidence was realized by combining the reliability with posterior probability estimation. Finally,The final fault diagnosis was obtained by using DS evidence theory to fuse the information of each evidence. The result of fault diagnosis exercise shows the high reliability of this proposed method. The method can fuse the fault signal of each sensor effectively and avoid the phenomenon of misdiagnosis.