基于模糊k近邻的变电站主接线类型自动识别方法

摘要

针对目前变电站接线图(静态手工绘制)维护耗时久、出错概率高等问题,提出一种基于模糊 k 近邻的变电站主接线类型自动识别方法——通过有向无环图(directed acyclic graph,DAG)构建不同电压等级电气设备拓扑岛以及构建相应的AOV网((activity on vertex network)进行拓扑排序,获取关键设备之间有向连通路径来表现设备之间的连接约束关系;对于不同类型的主接线,选取典型、非典型有向连通路径样本赋予隶属度进行训练,提高样本置信度,最后利用模糊k近邻分类方法对未知电气岛进行分类判别,获取变电站主接线类型.该方法基于设备模型动态获取变电站主接线类型,为后续主接线类型可视化成图奠定坚实的基础.%In allusion to problems of current statistic manual substation wiring diagrams such as long maintenance time and high error probability,this paper presents a kind of automatic identification method for substation main wiring modes based on fuzzy k-nearest neighbor algorithm which is to construct topological islands of electrical equipment of different voltage levels by means of directed acyclic graph and build corresponding activity on vertex networks for topological sorting so as to obtain directed connection paths among key devices to represent connection constraint relationships between devices. For different main wiring modes,this method selects typical and atypical directed connection paths for sampling training and im-proving confidence coefficients of samples.Finally,it uses fuzzy k-nearest neighbor classification method for classification of unknown electrical islands and obtain main wiring models of substations.This method can dynamically obtain the main wiring mode of substation based on the equipment model and lay a solid foundation of visualization of main wiring modes.