中文摘要：

针对电力系统继电保护故障,利用概率Petri网对电力系统进行建模,获取有效的故障信息,并运用DS(Dempster-Shafer)证据理论对信息进行融合,得出诊断结果.针对传统D-S证据理论在处理冲突证据时会存在结果与源证据相悖的问题,在加权平均法的基础上提出一种改进的融合方法,根据各个证据到平均证据的距离与证据权重大小成反比的关系,计算每个证据的权重,再进行加权平均,最后利用D-S组合规则进行迭代计算.与传统方法相比,本文方法拥有更好的融合效果和更高的效率.仿真结果验证了改进方法的有效性.

英文摘要：

The power system fault diagnosis problem was investigated in this paper .First ,the power system was modeled by a probability Petri net .Then based on this model ,the diagnosis information was obtained and fused with D-S (Dempster-Shafer) evidence theory .Finally ,the diagnosis result was concluded with the fusion performance .However ,the diagnosis result based on the D-S evidence the-ory may be contrary to the source evidence when dealing with the evidence ,of which some conflicts with each other .To solve the problem ,an improved fusion approach was proposed based on method of weighted average .With the method ,first the weight for each piece of evidence was calculated by its distance away from the average evidence ,which was inversely proportion to the weight .Then accord-ing to the D-S combination rules ,the final result was obtained by the iterative computation of the weighted average evidence .With the proposed approach ,better fusion performances were achieved and higher efficiencies were guaranteed .And the simulation results proved the effectiveness of this ap-proach .