基于SLMGRBF的配电网分布式光伏集群智能划分策略

Intelligent Partition Strategy of Distributed Photovoltaic Cluster inDistribution Network Based on SLM-RBF

分布式电源大规模分散接入给配电网的优化调度带来计算上的维数灾难,需要对分布式电源进行集群以降低调控难度,因此合理的集群划分十分重要.同时,配电网实时量测数据不全造成分布式电源进行实时集群划分难度大、时间效率低,因此提出一种智能局部移动(SLM)算法与径向基神经网络相结合的分布式电源集群智能划分策略.首先,选取有功和无功功率调节范围以及有功和无功功率-电压的灵敏度作为集群划分的指标,构造相似度矩阵并基于SLM形成分布式电源的集群划分方案库.然后,离线建立电压拟合模型,拟合可实时观测节点的功率与电压之间的关系;同时,离线建立电压-划分结果模型,在线通过电压得到实时划分结果,创新性地解决了潮流模型缺失时无法进行集群划分的问题,提高了集群划分的实时性.最后,在MATLAB平台通过仿真计算验证了算法的合理性和优越性.

Access of large-scale distributed power supply to the distribution network brings dimensionality disaster to the optimal dispatching of the distribution network.Therefore,it is necessary to cluster the distributed power supply to reduce the difficulty of regulation and control,and a reasonable division of distributed power supply cluster is very important.However,the incomplete real-time measurement data of the distribution network has caused difficulty and low time efficiency in real-time cluster division of the distribution network.Therefore,this paper proposes a distributed power cluster division strategy based on the smart local moving(SLM)algorithm and the radial basis function(RBF)neural network.First,the range of active power and reactive power regulation and the sensitivity of active power and reactive power to voltage are selected as the indexes of cluster division.By constructing a similarity matrix,the SLM algorithm is used to form the historical strategy library of cluster division of distributed power sources.Then,a voltage fitting model is established offline,which can observe the relationship between the power and voltage of buses in real time.Meanwhile,a voltage-division result model is established offline,and the real-time division result is obtained through the voltage online,which solves the problem that cluster division cannot be performed when the power flow model is missing,and improves the real-time performance of cluster division.Finally,the rationality and superiority of the algorithm are verified by simulation on MATLAB platform.