基于模糊灰色关联神经网络的综合故障诊断方法研究

Research on Synthesized Fault Diagnosis Technique Based on Fuzzy Gray Relational Neural Network

考虑到灰色关联诊断中模式向量各参数量纲不同,数量级存在差异,且各种特征因子有各自的应用缺陷,通过将模糊理论和灰色关联度相结合,形成所提取故障特征值的模糊关联度,将模糊关联度作为神经网络的输入量进行故障种类的识别,提出了模糊灰色关联神经网络故障识别方法,实现了针对在不同工作状态下的故障识别.对舰船主动力系统故障进行实例诊断研究,诊断结果表明,该方法准确、有效.

Considering mode vectors＇ parameters with different dimensions and orders of magnitude, and various characteristic parameters with their own application defects in gray relational fault diagnosis, fuzzy gray relational neural network fault diagnosis technique is presented. Through integrating fuzzy theory and grayrelational grade, fuzzy relational grade of fault eigenvalue is formed, and identifying of faults is done through making fuzzy relational grade to neural network inputting measures, then fault identifying in different working states is realized. Finally an example is applied for fault diagnosis of ship main power system, and diagnostic results indicate that the method is precise and available, which puts forward a synthesized fault diagnosis means for intelligent diagnosis.