基于分层遗传算法的电力变压器优化设计  

Design Optimization of Power Transformer Based on Multilevel Genetic Algorithm

作　　者：刘彦琴[1] 孟向军[1] 王曙鸿[1] 邱捷[1] 冯斌 刘方华 魏彩霞 刘凤琴 

机构地区：[1]西安交通大学电气工程学院,西安710049 [2]青岛变压器集团有限公司,山东青岛266109

出　　处：《西安交通大学学报》2009年第6期113-117,共5页Journal of Xi'an Jiaotong University

摘　　要：在研究了分层遗传算法的基础上,分析了该方法动态编码及动态调整遗传算子的优点,提出了一种改进的交叉变异机制.该机制在交叉操作时,不能在关联变量内部选择交叉点,在变异操作时,如果其中一个变量发生了变异,则与其关联的变量也要同时进行变异操作,并且可以将所有个体限制在有效的搜索空间内.文中将分层遗传算法应用于电力变压器优化的问题中,与单层遗传算法相比,在相同计算量的情况下,多层遗传算法成本节约1.83%,负载损耗下降1.01%.An optimization method, multilevel genetic algorithm （MLGA）, is presented. The advantages of dynamic encoding and adjustment of genetic operators in MLGA are analyzed. And improved crossover and mutation techniques in MLGA are proposed. The correlated variables, described as points, in individuals can be operated with crossover and mutation simultaneously. These techniques ensure that all the individuals are limited in effective search space. This MLGA is applied in optimization design of power transformer successfully. A comparison with single level ordinary GA shows that MLGA reduces the cost of transformer by 1.83% and the no-load loss by 1.01% with similar computational consumption.

关 键 词：分层遗传算法 电力变压器 优化设计 

分 类 号：TM402[电气工程—电器]