【摘要】 省级全景电网一张图可以通过各种形式的数据源进行构建，但是由于电网运行环境中遥测遥信数据及设备间拓扑连接关系的动态变化特征，省级全景电网一张图也会面临过时的问题。为解决以上问题，提出了基于全动态连接的图化模块的图数据动态更新机制，基于不影响省级全景电网图数据库生产环境高压状态使用的前提，在满足高可用、可靠性等特性的同时，基于Neo4j图数据库，对Neo4j的更新机制进行改进，结合机器学习的方法，引入动态更新预测器集成最新的数据，如：开关变位信息、设备故障信息、新增电力设备信息、设备操作与检修信息等等，避免因为滞后信息影响图计算结果。实验结果证明了所提出的机器学习驱动的图数据动态更新机制在预测电力系统实体更新状态方面具有优秀的准确率、泛化能力和可靠性。更多还原

【Abstract】 A provincial-level panoramic power grid map can be constructed through various forms of data sources, but due to the dynamic changes in telemetry and signaling data and topological connections between devices in the power grid operating environment, a provincial-level panoramic power grid map may also face the problem of obsolescence. To solve the above problems, this article proposes a graph data dynamic update mechanism based on a fully dynamic connection graph module. Based on the premise of not affecting the use of high-voltage states in the production environment of the provincial panoramic power grid graph database, while meeting the characteristics of high availability and reliability, the update mechanism of Neo4j is improved based on the Neo4j graph database. A new solution is adopted to integrate the latest data, such as switch displacement information, equipment fault information, newly added power equipment information, equipment operation and maintenance information, etc., to avoid the influence of lag information on graph calculation results. The experimental results demonstrate that the proposed machine learning driven graph data dynamic update mechanism has excellent accuracy, generalization ability, and reliability in predicting the update status of power system entities.更多还原