基于知识与模糊神经网络的故障诊断专家系统

KNOWLEDGE AND FUZZY NEURAL NETWORK BASED FAULT DIAGNOSTIC EXPERT SYSTEM

提出了一种基于知识与模糊神经网络专家系统故障诊断方法 ,设计构造了诊断专家系统的整体框架 ,在框架中体现了在任务调度机控制下 ,规则符号推理和模糊神经网络推理综合诊断的思想。知识库是诊断专家系统的核心 ,在本系统中定义了广义三层规则库结构 ,即元规则、故障诊断规则、结论合并规则。 Rule型的模糊联想记忆器实现专家系统中的分类和综合功能 ,并讨论了模糊神经网络输入和输出模糊化的问题。本文为旋转机械故障诊断专家系统提供了一个易于实现的框架结构。以汽轮机故障诊断为例进行了实验分析 。

A fault diagnosis expert system based on the knowledge and fuzzy neural network is developed. The frame of fault diagnosis system is designed and constructed. The comprehensive diagnosis idea of expert system under the control of task scheduler is showed in this frame. Knowledge base is the kernel of fault diagnosis expert system. It is defined as generalized three layer rulebase structure in this system which consists of basic rules, fault diagnosis rules, result merging rules. The function to classify faults and combine diagnosis results in expert system is realized by FNN. The issue of input and output parameter fuzzification of FNN is also discussed in this paper. Finally, as an example, a turbine fault diagnosis is analyzed. It shows that the expert system has the high reasoning efficiency and diagnosis accuracy.