基于分类决策树——关联规则的远程停复电故障溯源

Remote Power Outage and Restoration Fault Tracing Based on Classification Decision Tree-Association Rules

远程费控系统已成为智能电网的重要环节,但费控系统停复电故障率达到10.64%,目前只有1/3故障靠人工凭历史经验逐步排查而解决,其他故障原因仍不得而知。为快速实现故障诊断,推进智能电网建设,该文提出一种基于决策树—关联规则的停复电故障溯源方法。采用决策树算法将不同类型的数据进行分类,依此建立属性特征表,计算故障特征的信息增益率,进行决策划分,得到故障原因数据集;利用关联规则计算故障→原因的提升度,找出各个故障与原因的强关联性,建立故障溯源规则,实现故障溯源。该方法在故障发生时可迅速找到原因,提高工作效率,并通过实例验证了所提方法的有效性和高效性。

The electricity cost remote control system has become an important link in smart grids,but the power outrage and restoration fault rate of the cost control system has reached 10.64%.At present,only 1/3 of the faults are gradually checked by manual labor based on historical experience,and the rest of the faults cannot be found.In order to realize the fault diagnosis quickly,and promote the construction of smart grids,based on decision tree association rules,this paper proposes a method for tracing power outrage and recovery failures.The decision tree algorithm is used to classify different types of data,and the attribute feature table is established accordingly.The information gain rate of fault features is calculated,and the decision is divided to obtain the data set of fault causes.The association rules are used to calculate the degree of improvement of fault→cause,find out the strong correlation between each fault and cause,establish fault traceability rules,and realize fault traceability.This method can quickly find the cause when the fault occurs and improve the work efficiency,and the effectiveness and efficiency of the proposed method are verified through examples.