基于ACO优化参数的模糊Petri网故障诊断技术研究

摘要

Aiming at the traditional fuzzy Petri net relied on the parameters from the experience of specialist to diagnose the fault, let the precision of fault diagnosis take the limit of the knowledge level of specialist, a new method using weight fuzzy neural net and the respecting constructing method is advanced, the method using fuzzy Petri net to diagnose the fault, the parameters of the models is obtained by the training of the BP network. In order to improve the diagnosis precision, the ACO (Ant Colony Optimization) algorism is used to optimize the parameters in the models. Finally, through the Dynamo fault diagnosis experiment, the ACO fuzzy neural Petri net is proved to have higher diagnosis precision and efficiency compared with the ordinary fuzzy neural Petri net, and can diagnose the fault correctly. The method in this paper is applicable and feasible.%针对传统模糊Petri网在进行故障诊断推理时,需要依靠专家经验给出所有产生式规则的参数,使得故障诊断的精确度受限于专家知识水平的问题,提出了一种加权模糊神经Petri网模型以及相应的构造方法,此方法使用模糊Petri网进行故障诊断,网模型中各参数由BP神经网络训练而得,为了进一步提高诊断精确度,定义了使用ACO (Ant Colony Optimization)对网模型的各参数进行优化的算法；最后通过发电机故障诊断实例对比试验,验证了文中ACO优化的模糊神经Petri网,能够对各种故障进行正确的诊断,且在诊断精度和效率上较常规的模糊神经Petri网有了很大的提高,具有很强的实用性和可行性.