基于证据理论的多传感器融合故障诊断方法∗

摘要

In multisensor fault diagnosis system, a relationship between the parameter space of feature and the type space for engine fault needs to be established. Also the problem that the diagnostic results from dif-ferent sensors for the same engine may conflict needs to be addressed. The method of fault diagnosis based on D-S evidence theory using multisensor data fusion can solve the problems above to a large extent. The paper first discussed D-S theory in conjunction with fault diagnosis system;then an instance of fusion diag-nosis using two sensors was given;finally we compared the simulation results in single sensor with fusion di-agnosis. Analysis shows that the method based on D-S theory can solve the problem of evidence conflict in multisensor fault diagnosis effectively, and improve the accuracy of fault diagnosis decision to some extent.%在多传感器故障诊断系统中，需要建立故障特征参数空间到故障类型空间的对应关系，同时还需要解决不同传感器对同一机器的诊断结果可能存在冲突的问题。基于证据理论的多传感器融合故障诊断方法可以较好的解决上面的问题。论文首先结合机械故障诊断系统讨论了D-S证据理论，然后给出两传感器融合诊断的具体实例，最后对比分析了单传感器诊断与融合诊断的仿真结果。实验表明，基于D-S证据理论的融合诊断方法可以有效的解决多传感器之间的证据冲突问题，并在一定程度上提高了故障诊断决策的准确性。