基于数据驱动的配电网光伏双层优化调控策略

Double-Layer Optimal Control Strategy for Distribution Network with Photovoltaic Power Driven by Data

高渗透分布式电源接入后,对配电网电压和网损的优化提出了更高的要求。而传统的集中式控制缺少大量的量测设备和通信设备,导致数据采集不完整,优化模型不精确,难以满足大规模光伏并网的运行要求。所以文章构建了一种双层优化模型来改善传统集中式控制的不足;在概率优化的电气距离矩阵的基础上,使用蚁群聚类进行有效分区和主导节点选择,以此分区将传统的配电网二级控制引入第1层模型,然后利用基于粒子群算法优化极限学习机(particle swarm optimization extreme learning machine,PSO-ELM)神经网络挖掘并拟合配电网参数数据之间的函数关系,对第1层控制模型进行反复迭代修正。最后,在IEEE-33节点上进行仿真计算,验证了该模型对于配电网电压和光伏出力调控的有效性。

With the access of high-proportion distributed power supply,higher requirement of the optimization of distribution network voltage and network loss is made.However,traditional centralized control lacks enough measurement and communication equipment,which leads to incomplete data acquisition and inaccurate optimization model,and makes it difficult to meet the operation requirements of large-scale photovoltaic power accessing to grid.In this paper,a double-layer optimization model is constructed to make up for the shortcomings of traditional centralized control.On the basis of the electrical distance matrix of probability optimization,ant colony clustering is used for effective partition and dominant node selection.Traditional secondary control of distribution network is introduced into the first layer model,and then the PSOELM neural network is used to mine and fit the functional relationship among the distribution network parameter data,and iteratively corrects the first-layer control model.Finally,the simulation results of IEEE 33-node system verify the effectiveness of the model for voltage and photovoltaic output control of distribution network.