一种基于证据融合的故障诊断方法

A fault diagnosis method based on evidence fusion

结合故障诊断工作的实际场景,通过计算故障信息观测值与参考值之间的差异,生成对应的故障信度分配函数。然后基于证据分类的思想,将故障证据分为可信证据和冲突证据,并定义每个证据的可信度。最后,利用证据可信度修正最初的信度分配,再利用证据融合公式得到综合的故障诊断结果。电机转子故障诊断的实例表明本文提出的方法能够准确快速地识别产品故障,基于证据分类的修正方法能够加速故障信度分配的收敛。

In this paper we propose a method to generate the belief assignment of faults by calculating the difference between the observed and the reference values of fault information. Subsequently, based on the idea of evidence classification, the fault evidence is divided into two types： credible evidence and conflicting evidence, and the defi- nition of evidence credibility is given. Finally, we use the evidence credibility to modify the initial belief assign- ment and obtain the synthetic result of fault diagnosis by the evidence fusion formula. The example of motor rotor fault diagnosis shows that the method proposed in this paper can accurately and quickly identify the fault and the modified method based on evidence classification can accelerate the convergence of fault belief assignment.