基于概率有序树的预想故障集贪心筛选算法及其在可靠性评估中的应用

A Greedy Algorithm for Contingency Screening Based on Probability Ordered Tree and its Application in Reliability Assessment

预想故障集筛选是电力系统可靠性评估的重要组成功能,而现有筛选方法筛选效率较低且容易漏筛一些发生频率较高的高重故障,对系统评估精度造成不利影响。为此本文提出一种概率树贪心故障筛选法(P-tree greedy contingency screening technique,PGCST)以快速准确筛出系统中概率最大的前k个故障。基于概率距离及概率有序树定义,将原始问题转化为求前k个最短路径问题,并针对该问题设计了贪心求解策略。与快速排序法及改进快速排序法相比,PGCST算法复杂度与系统规模无关且算法效率有较大提升,可满足大系统实时在线筛选要求。将所提方法应用于电力系统可靠性评估,在RBTS和RTS79可靠性测试系统下的算例结果表明,基于PGCST的系统评估可达到较高的精度。

Contingency screening is an important part of power system reliability assessment. However, most existing screening methods do not have a high efficiency, and they may omit some high-frequency contingencies with high levels, which will cause an adverse impact on assessment accuracy. Therefore, a novel method, P-tree greedy contingency screening technique（PGCST）, was proposed to screen out the first k contingencies with maximum probability accurately and quickly. Based on the definition of probability distance and probability ordered tree, it turned into a problem seeking the first k shortest path. A greedy strategy was designed to figure it out. Compared with fast sorting technique and improving fast sorting technique, the algorithmic complexity of PGCST is independent of system size and it has higher algorithm efficiency. When applied into power system risk assessment, case study in RBTS and RTS79 illustrate that PGCST has high evaluation accuracy.