基于自适应神经网络加权模糊故障诊断研究  

Adaptive Weighted Fuzzy neural network Petri net for Fault Diagnosis

作　　者：潘海兰[1] 蒋文蓉[1] 吴嘉琪[1] 

机构地区：[1]上海第二工业大学计算机与信息学院,上海201209

出　　处：《计算机测量与控制》2011年第7期1599-1601,1632,共4页Computer Measurement &Control

摘　　要：研究故障诊断问题;针对传统Petri网难以精确地描述故障现象和故障原因之间的复杂关系,基于模糊逻辑BP神经网络和传统Petri网模型结合,提出了一种新的自适应的加权模糊神经网络Petri网模型故障检测方法;该方法首先采用改进的BP神经网络算法对模型的权值进行训练,然后采用构造的自适应模糊Petri网模型对故障进行诊断;在柔性制造系统实例中进行了故障诊断,实验结果表明,该方法具有很强的故障推理能力以及自适应能力,能有效地对故障进行诊断,具有一定的实际应用价值。A method for using neural network and weighted fuzzy Petri net to diagnose fault is proposed.Aiming at the traditional Petri net can not predict precisely the complex relation of the default phenomenon and the cause,neural network、fuzzy logic and the traditional Petri net were combined,and a constructing method for adaptive weighted fuzzy Petri net model is proposed,and a improved BP algorism is introduced to train the weight of the model,then the specific process for using the model to diagnose the fault is given.Finally,the model was applied to the instance of FMS,and the model was proved to have the advantages of Petri net and neural network and have reasoning and adaptive ability.

关 键 词：故障诊断 神经网络 柔性制造 权值 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]