基于最大似然译码字的Petri网电网故障诊断方法

A Fault Diagnosis Approach of Power Networks Based on Maximum Likelihood Decoding Petri Net Models

考虑到畸变的动作信号对电网故障诊断准确度的影响,提出一种基于最大似然译码字的Petri网故障诊断方法。首先,根据保护和断路器的动作逻辑关系,建立故障元件的Petri网诊断模型,并运用CPN Tools软件验证其合理性;其次,推导相关矩阵,并给出确定监督码元数量的方法;最后,应用最大似然可信度和相关代价之间的关系确定最佳错误图样,以此对畸变信号进行纠正,并根据故障映射关系做出相应的故障诊断。算例分析和不同模型诊断结果的对比表明上述方法可提高故障诊断结果的容错性和准确性。

Considering that the actuating signals would affect the fault diagnosis accuracy of power grids, a fault diagnosis approach based on maximum likelihood decoding Petri net models is proposed. At first, the diagnostic models of Petri nets corresponding to the fault components are built according to the actuating logic relationship of protection and the circuit breaker. With the help of the computer-aided tool CPN Tools, the models are verified and rationalized. After that, the corresponding matrix is derived and a method to determine the number of supervision code is given and verified. Finally, a relationship between maximum likelihood credibility and correlation discrepancy is applied to determine the optimal error pattern so as to correct the distortion of the signals, and then the corresponding fault diagnosis can be made according to the mapping relationship of the fault. The example analysis and the comparison of diagnostic results of different models show that the suggested method can improve fault tolerance and accuracy of the fault diagnosis.