故障诊断神经网络系统的专家知识表达方法

EXPERT KNOWLEDGE REPRESENTATION METHODS FOR NEURAL NETWORK FAULT DIAGNOSIS SYSTEMS

通过介绍其它故障诊断技术存在的问题，研究了故障诊断神经网络系统，提出了两种故障诊断神经网络系统的专家知识表达方法，并将其应用于飞行器液压动力系统的故障诊断．仿真结果表明：提出的方法是正确可行的，并且专家知识表达方法对诊断结果有很大影响．该方法可推广应用于其它动力系统的故障诊断．

Some drawbacks in general fault diagnostic techniques can be avoided by using neural networks. The Neural Network Systems in Fault Diagnosis(NNSFD) and proposed two kinds of expert knowledge representation methods for NNSFD, which are applied to fault diagnosis for aerospace vehicle hydraulic dynamic systems. The results of simulation experiments show the methods presented are correct and feasible, the expert knowledge representation methods have great influence upon the performances of NNSFD.The methods can also be used for fault diagnosis of other dynamic systems.