一种基于贝叶斯Petri网的故障诊断方法

Power System Fault Diagnosis Method Based on Bayesian Petri Nets

针对Petri网在分析复杂电力系统时的容错性差且难以适应网络拓扑变化的问题,提出一种贝叶斯Petri网模型(BPN),并基于该模型提出一种电网故障诊断方法.该方法通过电力系统网络拓扑分析确定停电区域,随后按照故障蔓延方向对停电区域内的元件分别建立BPN模型,应用Petri网推理和贝叶斯概率计算确定故障元件,最后采用均值方法对各方向上的分析结果进行融合.诊断分析表明,该方法在信息不完备的情况下具有较好的容错性,并且在网络拓扑结构发生变化后仍具有较好的适应性.由于BPN推理时根据基于统计的先验概率求取元件故障的发生概率,避免了直接对计算参数进行设定的主观性.

For Petri net＇s fault tolerance was poor and it was difficult to adapt to the topological change of network in the analysis of complex power system, this paper proposed a kind of Bayesian Petri net model, and gave a fault diagnosis method for power grid based on this model. According to the fault spread directions, this method established the BPN models for each component in power cut area separately, and determined the failed component by the application of Petri net inference and Bayesian probability calculation, finally, fused the results by averaging method. Diagnosis analysis showed that this method had better fault tolerance under the condition of incomplete inforalation, and had better adaptability after the change of network topology structure. In the BPN inference, calculating the fault probability of component based on the prior probability from statistics, avoided the subjectivity of setting related parameters directly.