多目标进化算法求解无功优化问题的对比分析  

Comparison and Analysis of Multiobjective Evolutionary Algorithm for Reactive Power Optimization

作　　者：李智欢[1] 段献忠[1] 

机构地区：[1]电力安全与高效湖北省重点实验室(华中科技大学),湖北省武汉市430074

出　　处：《中国电机工程学报》2010年第10期57-65,共9页Proceedings of the CSEE

基　　金：国家自然科学基金项目(50595414);中国留学基金委建设高水平大学公派研究生项目([2008]3019)~~

摘　　要：对经典的多目标进化算法(multi-objective evolutionary algorithms,MOEAs)在电力系统无功优化中的应用展开比较研究。与传统设定偏好参数、将多目标问题转化为单目标问题的方法不同,直接采用计及系统网损与电压偏移的多目标模型。提出无功优化多目标进化算法统一框架,采用一致的编码策略、约束处理方法。以IEEE30节点标准系统的多目标无功优化为算例,从帕累托前沿、外部解及C指标等方面,比较各种算法的性能特点,并按照其优劣将算法分为5个性能等级。参考算法的性能等级,详细分析每种算法的优缺点。相关结论对MOEAs在无功优化及电力系统其他优化问题中的应用和改进,都具有一定的参考价值。Reactive power optimization based on multi-objective evolutionary algorithms （MOEAs） was studied. Different with the traditional approach that combines multiple objective functions into a single one by setting preference parameters, the approach adopts and optimizes the multi- objective models directly. The uniform framework of MOEAs- based reactive power optimization was proposed and the detailed procedures were discussed. Based on the test case of IEEE 30 bus system, from the view point of Pareto fronts, outer solutions and C metric, the computing performances of five MOEAs were compared, which were classified into five performance levels. According to the levels, the advantages as well as disadvantages of all the MOEAs were outlined. These conclusions may be a great reference for further application and improvements of MOEAs in reactive power optimization and other optimization problems of electric power systems.

关 键 词：无功优化 多目标进化算法 帕累托前沿 非支配解 多目标优化 

分 类 号：TM71[电气工程—电力系统及自动化]