分布式机械故障诊断系统的协调机制及其分析

Cooperation strategies for distributed mechanical exception diagnosis systems

基于主体(Agent)的分布式机械故障专家系统可以有效地提高诊断效率,但Agent间诊断结果的不一致性会影响整个系统的诊断结果,需要采取适当的协调机制调节Agent之间的矛盾.为此给出了可信度优先、权威性优先、相邻影响、彼此影响等4种Agent协调机制以及相应的协调公式,并对它们进行了量化比较,分析了各自的优越性和适应性情形.最后给出了基于协调的Agent分组诊断的一致性处理机制.

In distributed mechanical exception diagnosis system, the inconsistency of the multi-Agent decisions can influence the results of the whole system. Four different inconsistency processing strategies for the multi-Agent diagnosis were introduced, which are reliability-first, authority-first, border-influence, and all-influence strategies. The decision formula of these four strategies were presented, and their diagnosis results were compared by listing the histogram, with which their advantages and adaptability were shown. The consistency diagnosis strategy of the (Agent) group was given finally.