基于多维特征量融合的配电网拓扑异常溯源与应用模型研究

An Anomaly Tracing and Application Model of Distribution Network Topology Based on Multi-dimensional Feature Fusion

线路拓扑异常是线损治理中影响较大的因素之一,目前中压配电网基础数据量大、异常诊断分析复杂等原因造成线损治理排查效率低。结合多年中压线损管理经验,对中压线路-变压器档案关系构建配电网拓扑异常溯源与应用模型,将电网GIS(地理信息系统)、 PMS2.0系统、用电采集系统、营销业务系统、调度系统等多源融合,利用大数据挖掘技术,充分挖掘数据价值,以机器学习算法为基础,规则引擎为补充,从线变关系档案异常、双电源配置错误和线路转供运行三大问题入手,多维度全方位实时分析10(20) kV线路拓扑异常,精准定位线路拓扑异常点。该模型将有效指导拓扑异常基础数据治理,进一步提升线损管理水平,为维护企业经济利益和可持续精益化管理提供有力支撑。

Abnormal circuit topology is one of the influential factors in line loss management, and the large basic data and complicated abnormity diagnosis and analysis lead to inefficiency of line loss management and investigation. Based on medium-voltage line loss management experience over the years, an anomaly tracing and application model is established for the relationship between medium-voltage lines and transformer files to integrate grid GIS(geographical information system), PMS2.0 system, electricity information acquisition system, marketing service system and dispatching system and to employ big data mining technology to fully exert data value;based on machine learning algorithm and supplemented by rule engine, real-time analysis of 10(20) kV line topology anomaly is conducted multidimensionally and ultimately from the perspectives of linetransformer relation file anomaly, duplicate supply misconfiguration and line transferred operation to precisely locate line topology anomaly point. The model will help guide basic data management due to topology anomaly, furthering line loss management level and providing robust support for maintaining economic interest of enterprises and sustainable lean management.