基于改进自组织映射聚类算法的连锁故障预测模型

为提高电力系统连锁故障预测的快速性和全面性,提出一种基于改进自组织映射聚类算法的连锁故障预测模型.针对事故链预测模型的故障指标模糊、事故链模型过于庞大等问题,该模型首先设计了初始故障集评价指标、上下级支路间关联性指标.其次改进自组织映射聚类算法,对关联性指标进行聚类分类,并在上下级支路关联性确定中,设定适当阈值.最后,利用IEEE36节点系统进行仿真验证.结果 显示:本文方法在3.245 s时可预测到6级总计37条故障线路,故障预测范围达到97％.可见,本文方法与传统方法相比能有效缩短预测时间,扩大预测范围.

A Chain Failure Prediction Model Based onImproved Self-organizing Maps Clustering Algorithm

In order to solve the problems existing in power system chain fault prediction, such as fuzzy fault index and too large fault chain prediction model, an improved fault chain prediction model was proposed.The model designs the evaluation index of the initial failure set and the correlation index between the upper and lower branches.Improved the ambiguity of indicators;Improved the accuracy of prediction.The optimal SOM clustering algorithm is introduced to cluster the correlation indexes, and the appropriate threshold value is set in the determination of the correlation between upper and lower branches to ensure that the predicted subordinate branches are comprehensive and not redundant.An adaptive method is proposed to solve the problem that the learning rate is difficult to determine in the clustering algorithm, which makes the clustering results more accurate.The simulation results of IEEE36 node system show that this method can effectively shorten the time of fault prediction, improve the accuracy of fault prediction, and give attention to both the rapidity and comprehensiveness of the prediction results.