粗糙集与神经网络在发动机故障诊断中的融合应用

电喷发动机运行的状态信息众多而复杂,故障与状态信息的关系模糊而不确定,如何从复杂的多元信息中获取有用部分并加以利用是电喷发动机故障诊断的关键.本文应用粗糙集理论对冗余信息进行约简,得到更为简明的诊断规则,将约简结果与神经网络相结合,建立了故障诊断系统.网络的训练对比结果表明,粗糙集理论的约简处理简化了神经网络结构,提高了网络的训练效率;通过实例验证了粗糙集理论与神经网络相结合进行电喷发动机故障诊断的可行性.

Blending application of rough sets and neural network in EFI engine fault diagnosis

The information of reflecting the EFI engine running state is complicated,the relationship between fault and numerous stating information is uncertain and fuzzy.The key of EFI engine fault diagnosis process is how to acquiring and using valuable information from those complicated multi-source information.In this paper,rough sets theory was applied to reduction of incomplete information to find necessary conditions for diagnosis,and according to the results of reduction the fault diagnosis system of based on neural networks was founded.The comparison results of network training indicated that the structure of neural network was simplified by reduction process of based on rough sets theory and efficiency of network training was enhanced;the feasibility of application of rough sets integrated with neural networks to fault diagnosis of EFI engine was verified by the practical example.