计及信息畸变影响的电网故障诊断分级优化方法

Power System Fault Diagnosis Based on Gradual Optimization in Consideration of Alarm Information Aberrance

基于优化算法的电网故障诊断方法可归纳为相似性诊断方法和覆盖性诊断方法。在故障信息发生畸变时,两类方法分别存在漏诊断和误诊断的问题。为了提高电网故障诊断的准确性,结合两类诊断方法的特点提出了电网故障诊断的分级优化方法。该方法依次进行所有警报信息的相似性诊断,保护信息的相似性诊断,以及覆盖性诊断,通过分级诊断分离出故障发生可能性不同的几类可疑故障元件。算例测试表明该方法避免了传统优化方法的弊端,可在警报信息发生畸变且存在多次保护和断路器不正确动作的情况下对多重故障进行准确地诊断。针对已有方法中特征警报状态矢量造成寻优计算量大的问题,提出了构造自适应特征警报状态矢量的方法;并基于故障后的停电区域研究了限定及建立诊断模型中各类变量的简单方案;通过采用这些方法提高了诊断过程的快速性。

Recent optimization-based fault section estimation methods for power systems can be categorized as consistency diagnostic methods or parsimonious covering diagnostic methods. The two kinds of methods may reach wrong diagnosis results in complex scenerios in which maloperation of protections and alarm aberrance exist. To overcome their drawbacks, a multi-objective gradual optimization method was built by combining the merits of the two methods. The proposed method took turns to make the consistency diagnosis based on all the alarm and the protection alarm repectively, and then the parsimonious covering diagnosis. By simple calculation of the diagnosis results, several kinds of fault section candidates are listed according to their fault probability. Case studies show that the method presented can accurately locate the faulty element in the aforementioned complex scenerios but also give a detailed evaluation of the operation of protections. To reduce the heavy calculation load caused by multiplicity of characteristic alarms, a method to construct adaptive characteristic alarms is proposed. Specefic approaches to confine and list various state vectors are provided based on the outage area, which make the diagnosis process much faster.