基于SFCE与规则推理的发动机集成故障诊断方法

An Engine Integrated Fault Diagnosis Method Based on SFCE and Rule Inference

针对发动机磨损过程的复杂性以及故障症状(磨粒信息)与故障原因之间对应关系的不确定性,首先对磨粒数量特征信息进行了模糊化,建立了某发动机典型磨损故障的模糊规则库。针对单一故障诊断方法的局限性,引入了对称模糊交互熵的故障推理方法,在此基础上建立了基于对称模糊交互熵和规则推理相结合的发动机故障诊断模型。本模型首先采用对称模糊交互熵对故障原因进行"过滤",滤除出现可能性较小的原因,然后利用规则推理进一步确认最后的故障原因。实例表明,该方法在发动机多症状多原因的复杂情况下能够取得较好的诊断结果。

Owing to the complexity of engine wear process and the uncertainty of corresponding relation between fault symptom and fault reason, the fuzzy set was employed to measure the wear particle information and the fuzzy rule library was also created. For the limitation of single fault diagnosis method, the symmetric fuzzy cross entropy （SFCE） was introduced. Then the engine fault diagnosis model based on the combination of SFCE and rule inference was built. By the model, the fault reasons that appeared rarely were eliminated with SFCE and then the further fault reasons were determined with rule inference. The ex- ample proves that the model shows an evident advantage in dealing with the complicated relations of multi-symptom and multireason.