分布计算的遗传算法在无功优化中的应用

为了快速有效地求解电力系统无功优化问题,提出了一种分布式并行计算的遗传算法。它采用主从方式来组织局域网内的多台机器进行并行计算——由1台主机进行选择和遗传操作,并根据负荷均衡的原则调度多台从机计算潮流以给出个体适应值。根据无功优化的特点,为了增加算法并行度,就编码方案、基于多目标函数的适应度求解和遗传操作等方面对遗传算法进行了详尽的设计。文中还着重分析了并行处理效率的相关问题。算例表明该方法不仅取得了较好的优化效果,而且显著地提高了计算速度。

A DISTRIBUTED GENETIC ALGORITHM FOR REACTIVE POWER OPTIMIZATION

This paper presents a distributed genetic algorithm (GA) to solve optimal reactive power flow (OPF) problems inpower systems. A master-slave plan is adopted to process the parallel computation with several computers in a local areanetwork (LAN). In this plan, a 11 mastern controls genetic operations and balances the loads of n slaves" which solve theadaptability function simultaneously. Some aspects to improve the parallelism of this algorithm are discussed in detail, such asencoding methods, the solution of the adaptability function of multi--objectives, and genetic operations. Moreover. someproblems on parallel efficiency are analyzed. Based on the analysis of the numerical example, it can be concluded that thismethod achieves satisfactory optimization results as well as high computation speed.