从往复发动机点火系统的工作机理出发,分析了点火系统的常见故障。构建了模糊隶属度函数对故障征兆信号进行模糊化处理,得到了多元故障敏感特征,并建立了三层动态神经网络进行基于多元信息融合的点火系统故障诊断。实例分析表明,基于模糊动态神经网络进行的往复发动机点火系统故障诊断高效可靠,能够为往复机械的故障诊断提供新的方法。
Starting from the working mechanism of reciprocating engine ignition system, the common faults of the system are analyzed. The fuzzy membership functions are built to blur the fault symptom in order to get multi-dimension fault sensitive characteristics. The three layers of dynamic neural network are constructed to make fauh diagnosis based on the multi-dimensional information infusion. Example analysis shows that the proposed method has efficient and reliable performances, and can provide a new method for fault diagnosis of reciprocating machinery.