基于模糊聚类与学习自动机的多目标无功优化.pdfVIP

第 36 卷 第 7 期 电 网 技 术 Vol. 36 No. 7 2012 年 7 月 Power System Technology Jul. 2012 文章编号：1000-3673 （2012 ）07-0224-07 中图分类号：TM 72 文献标志码：A 学科代码：470·40 基于模糊聚类和学习自动机的多目标无功优化 王玉荣，万秋兰，陈昊 （东南大学电气工程学院，江苏省 南京市 210096 ） Multi-Objective Reactive Power Optimization by Fuzzy Cluster and Learning Automata WANG Yurong, WAN Qiulan, CHEN Hao (School of Electrical Engineering, Southeast University, Nanjing 210096, Jiangsu Province, China) ABSTRACT: To perform reactive power compensation, it is 层面的问题[3-5] 。首先，无功补偿设备安装地点的正 necessary to select the site where the compensator is configured 确选择，能够保证无功补偿的有效性。现有的选择 and the needed compensation capacity. The weak nodes in power 无功补偿设备安装地点的方法主要有 2 大类：一是 system are searched by fuzzy cluster and the information of 通过混合整数规划(mixed integer programming，MIP) candidate nodes is obtained; during the dynamic fuzzy cluster, 法，同时提供补偿设备的地点和容量的优化结果； the indices of U/ U , Γ and voltage deviation are applied. 0 Synthetically considering the minimum generation cost and 二是采用弱节点指标、模态分析、电压稳定裕度、 reactive compensation cost, the minimum voltage deviation and 灵敏度法等[6-9]，对无功容量优化之前先进行选点优 minimum active network loss, a multi-objective optimization 化。对于大规模系统而言，MIP 需要在大范围内搜 model for candidate nodes, where reactive compensator may be 索无功补偿的最优选点及补偿容量，计算时间较 configured, is built and the optimal trade-off solution of the 长。实际中无功补偿的选点一般选用第 2 种方法。 optimization problem is obtained by learning automata. Applying 但对于原则、指标的选用策略及如何确定无功补偿 fuzzy cluster and learning automata to IEEE 57-bus s