改进的双向联想记忆网络在凝汽器故障诊断中的应用  

Application of Improved Bidirectional Associative Memerory Network on Condenser Fault Diagnosis

作　　者：任燕燕[1] 戴毅姜[1] 翟永杰[1] 董泽[1] 

机构地区：[1]华北电力大学控制科学与工程学院,河北保定071003

出　　处：《电力科学与工程》2009年第10期23-27,共5页Electric Power Science and Engineering

摘　　要：在确立凝汽器典型故障知识库的基础上,应用双向联想记忆(BAM)网络对凝汽器进行故障诊断。网络学习算法采用强化系数的多重训练算法。在该算法的作用下,BAM网络将被强化矢量对存储在以此矢量对为中心的Hemming距离为1的邻域里的能量最小点,从而保证矢量对的正确联想。设计了诊断模型,实现了对凝汽器典型故障的诊断,并分析了该模型在实际应用中可能出现的问题。BAM neural network is applied on condenser fault diagnosis in this paper,based on the typical failures knowledge of condenser.The multiple training algorithm is adopted to intensify the coefficient of BAM.With this algorithm,the BAM network can store the corresponding training vector pair at the minimum energy point within the area of 1 Hemming distance from the vector pair.It ensures that the vector pair can be recalled correctly.The diagnosis model is designed in detail in the paper.Typical fault diagnosi...

关 键 词：双向联想记忆(BAM) 故障诊断 多重训练算法 Hemming距离 BAM能量函数 

分 类 号：TK268[动力工程及工程热物理—动力机械及工程]