基于超椭球贝叶斯网络的配电系统可靠性评价-电测与仪表.DOC

基于超椭球贝叶斯网络的配电系统可靠性评估 葛轶，（四川大学 电气信息学院，成都 ） 摘要：IEEE-RBTS Bus 6 母线系统进行算例分析，证明了该方法适用于实际配电系统的可靠性评估。 关键词：配电系统；可靠性；贝叶斯网络；超椭球模型；区间概率；证据理论 中图分类号：TM933 文献标识码：B 文章编号：1001-1390（201）00-0000-00 Reliability assessment method for distribution system based on hyper-ellipsoidal Bayesian network Ge Yi, Zhou Buxiang (School of Electrical and Information, Sichuan University, Chengdu 610065, China) Abstract: Aiming at the problem that the failure probability of components in actual operation of the distribution system can not be described by exact value, a reliability assessment method for distribution system based on hyper-ellipsoidal Bayesian network is proposed in this paper. Evidence theory is applied to obtain belief function and plausibility function of components in order to get the initial fault interval probability. Hyper-ellipsoidal model is applied to restrict the initial fault interval probability in order to get the root nodes interval probability of Bayesian network. The forward reasoning Bayesian network is applied to get the interval probability of leaf nodes in order to get the reliability indices of distribution system. The backward reasoning of Bayesian network is applied to get the posterior interval probability of root nodes in order to find the weakest part of the system. By analyzing the IEEE-RBTS 6 bus system, the method is proved to be applicable to the reliability assessment of the actual distribution system. Keywords: distribution system, reliability, Bayesian network, hyper-ellipsoidal model, interval probability, evidence theory 0　引 言 [3]应用手动块层级结构和自动块层级结构以得到负荷点故障持续时间和故障率，进而对配电系统进行可靠性评估；文献[]应用改进威布尔分布和对数正态分布以得到元件寿命周期和修复时间，进而对配电系统进行可靠性评估。上述文献从不同角度对配电系统可靠性Bayesian Network, BN）是一种对有限离散随机变量间概率关系的图解描述，在评估系统可靠性和辨识系统薄弱环节等方面得到了广泛应用19]。由于实际系统元件，有学者将区间模型与贝叶斯网络相结合，产生了区间贝叶斯网络法。文献[]应用区间数以得到逻辑门区间算子，进而对系统进行可靠性评估。区间模型对于统计数据样本较少的系统可靠性评估具有较强的适应性，但该模型所结论过于保守，精确度不高。 Hyper-ellipsoidal Model, HM）是一种凸集模型[20]，具有参数变化连续、模型结构简单和易于进行相关性分析的优点，可以较好地弥补区间模型结论过于保守