面向复杂设备的分布式故障诊断研究

基于复杂系统的设备故障诊断的要求 ,按故障信息流动的方向将诊断过程分为实时监测层、故障诊断层及综合决策层 ,并建造一种分布式多级黑板诊断结构模型。根据不同诊断层次的特点 ,赋予其不同的诊断方法。在实时监测层中各实时监测单元采用神经网络和知识基模块的混合结构完成该层任务 ;故障诊断层则基于广义知识表达模型 ,建立三库结构 ,即静态实例库、动态实例库和规则库 ,整个推理过程基于三级推理机制 ;综合决策层为一智能型的辅助决策支持系统 ,提供对整个系统生产与设备运行状态的综合诊断与决策。

Study on Distributed Fault Diagnosis of Complex Equipment

Based on the demands of fault diagnosis of equipment in complex systems, the diagnostic process is divided into three levels, i.e, the realtime monitoring level, the fault diagnostic level, and the synthetic decisionmaking level. A distributed multilevel blackboard model is built. Different methods are applied to different levels according to their particular characters. At the realtime monitoring level, the hybrid structure integrating neural networks and knowledgebased modules is used to accomplish diagnostic tasks in each monitoring unit. The fault diagnostic level is designed on the basis of the generalized knowledge demonstration model to build a triple basis structure, including the static case basis, the dynamic case basis, and the rule basis The whole diagnostic course abides by the threegrade inference mechanism. The synthetic decisionmaking level is an intelligent decision support system used to provide the synthetic diagnosis and decisionmaking to the conditions of the whole system and equipment.