基于知识图谱的主动配电网电压控制策略

随着新能源大规模接入以及负荷的随机波动性，对配电网的电能质量提出了更高的挑战及要求。主动配电网控制无功调压设备抑制电压波动通常转化为混合整数规划问题，难以做到实时控制且需频繁进行复杂计算。本文从历史数据中提取源荷状态，生成基于二阶锥最优潮流模型的电压控制策略，构建以调压装置状态、系统数据与控制策略为核心实体的配电网电压控制知识图谱；在实时电压控制时，基于控制策略子图匹配及数据检索算法，以当前网络状态匹配知识图谱中相似状态，进行安全校验和优化求解，并更新知识图谱中的状态策略。同时，在无功设备调节过程中增加人机交互环节，对于时间尺度、电压及设备动作及关键点电压实现精准控制。基于改进IEEE系统算例的仿真结果表明，所提出的基于电压控制策略知识图谱的检索方法及交互策略能够有效提升配电网无功电压控制策略生成效率，并具有不同场景适用性。

Active Distribution Network Voltage Control Strategy Based on Knowledge Graph

With the large-scale access of new energy and the random fluctuation of load, the power quality of the distribution network has been put forward higher challenges and requirements. The control of reactive voltage regulator based active distribution network to suppress voltage fluctuation is usually transformed into a mixed integer programming problem, which is difficult to achieve real-time control and requires frequent complex calculations.In this paper, source and charge states were extracted from historical data, and voltage control strategies based on the second-order cone optimal power flow model were generated in batches. The voltage regulator states of all generating strategies, and the knowledge graph of system data and strategies were recorded. In the real-time voltage control, based on the time series similarity retrieval algorithm, the current network state is matched with the similar state in the knowledge graph, the security verification and optimization solution are carried out, and the state strategy in the knowledge graph is updated. Meanwhile, human-computer interaction is added in the reactive equipment regulation process to achieve accurate control of time scale, voltage, equipment action and voltage of key points. The simulation results based on the improved IEEE system examples show that the proposed retrieval method and interaction strategy based on the knowledge graph of voltage control strategy can effectively improve the generation efficiency of reactive voltage control strategy in distribution network, and can be applied to different scenarios.