基于决策表约简的变压器故障诊断Petri网络模型及其应用研究

Study on Application of Petri Nets Model of Transformer Fault Diagnosis Based on Decision Table Reduction

通过对变压器故障诊断知识中大量的冗余特征进行压缩或约简 ,将粗糙集理论(RST)引入到变压器故障诊断中 ,以提高以往依赖先验知识进行诊断智能方法的效率。根据决策表约简实现故障特征的压缩与规则的简化 ,并利用该结果建立Petri网络模型 ,可获得最优的网络模型以充分发挥其快速的并行推理能力 ,实现高效的变压器故障诊断。对比分析结果表明 :约简后的特征具有与原来相同的分类能力 ,得出的主导特征也与实际相符 ;基于最小诊断规则所建立的故障诊断Petri网络模型 ,结构得到有效地优化 ,分类结果与原有的网络一致。

In order to improve the efficiency of intelligent approaches based on prior knowledge,the rough set theory(RST)is introduced to reduce the many redundant features in the transformer fault diagnosis rules.Through decision table reduction,the features are compressed and rules are simplified.And based on the reduced results,the optimal Petri nets(PN) are built to realize fast and parallel reasoning.Then more efficient fault diagnosis can be achieved.The results of comparison analysis show that after reduction fault classification is invariable,and main features are close to actual experiences.Although the structure of Petri nets constructed by minimal diagnosis rules is effectively simplified,diagnosis results are not changed.Finally,the analytical result of practical sample verifies the accuracy of the proposed idea.