摘要
配网重构的实质是求解大规模、多约束条件的非线性组合优化问题。除了经典方法以外,使用基于人工智能原理的遗传算法求解这类问题在近年来受到了广泛的关注。但因配电网辐射状运行条件的限制,简单遗传算法在求解该问题时容易陷入无效解过多、效率低下的困境,而采用基于Agent的改进遗传算法可以有效解决这一问题,但还有一些地方需要改进。针对这一问题,在已有算法基础上引入"自由度"概念,可以明显增加算法的效率和全局收敛性。通过采用IEEE14节点系统作为算例,具体分析了不同的"自由度"取值对基于Agent的改进遗传算法性能的影响并分析了其原因。相关结论表明所引入"自由度"概念的必要性和有效性。
Distribution network reconfiguration is in essence an issue of solving large-scale nonlinear combinatorial optimization with multi-constraints.Apart from applied classic methods,genetic algorithm based on artificial intelligence principle in solving this issue is widely concerned recently.But limited by the radialized mode of operation in distribution network,simple genetic algorithm becomes inefficient at solving this problem.Modified genetic algorithm based on Agent can solve this problem effectively,but some measures are needed to be improved.This paper introduces a concept named ＂Freeness＂ applying on the existing methods to improve their efficiency and global convergence.Taking IEEE-14 system as an example,this paper analyzed the effect of different values of Freeness on the performance of the modified genetic algorithm based on Agent.Conclusions prove that bringing the concept of Freeness is necessary and feasible.
出处
《华中电力》
2011年第3期35-39,共5页
Central China Electric Power
关键词
配网重构
基于Agent的遗传算法
自由度
distribution network reconfiguration
genetic algorithm based on Agent
freeness
