利用模型诊断降维的电网故障诊断完全解析方法

A Complete Analytic Method for Fault Diagnosis of Power Systems Based on Model-based Diagnosis for Dimensionality Reduction

现有基于解析模型的电网故障诊断方法为了提高诊断的准确性,考虑了保护和断路器的动作及其告警信息存在的不确定性,增加诊断的难度,并且存在多解或误诊的可能。针对此问题,首先提出一种电网故障诊断的改进完全解析模型,根据各类保护和断路器不确定事情发生的概率,赋予保护和断路器不同权值,使模型更加合理。通过模型诊断的方法获取候选诊断,由此得到的故障假说可作为解析模型的已知变量,降低模型中待求变量的维度,同时,为了提高模型求解方法的通用性,通过分析保护和断路器的动作状态与告警信息之间的因果关系,提出一种基于关联规则的模型求解方法。最后,通过含有多重不确定事情的电网故障算例验证所提方法的有效性。

The existing fault diagnosis based on analytic models in power system considered the uncertainties about action of protective relays（PRs）, trip of circuit breakers（CBs） and alarm information, which improved the accuracy of fault diagnosis but also led to the increase of fault diagnosis difficulty and had the multiple solution and misdiagnosis problems. Aimed at these problems, firstly, an improved complete analytic model for fault diagnosis of power systems was proposed. Different weight factors were assigned to each class of PRs and CBs according to the probability of the uncertain things, which improved the reasonability of analytic model. Then, model-based diagnosis method was adopted for gaining candidate diagnosis which was defined as fault hypotheses as the known variables of model. The variable dimension of model was effectively reduced by fault hypothesis. Meanwhile, in order to improve the generalization of the model solution, a commonality solving method based on association rules was developed by analyzing causal relationship between action states and alarms of PRs and CBs. Finally, the validity of the proposed method is verified by examples of complex power system fault containing multiple uncertain things.