基于CIM的配电网重构的并行遗传计算  

Parallel genetic computation for CIM-based distribution network reconfiguration

作　　者：欧阳帆[1] 刘东[1] 廖怀庆[1] 陆融[2] 

机构地区：[1]上海交通大学电气工程系,上海200240 [2]上海市电力公司市东供电公司,上海200122

出　　处：《华东电力》2008年第1期71-75,共5页East China Electric Power

基　　金：国家自然科学基金重点项目(90612018)

摘　　要：针对基于遗传算法的配网重构计算搜索时间过长、过早收敛的问题,引入发布式并行计算方法,从函数级主从式并行计算模型和群体级对等式计算模型两个方面组织局域网内的多台机器进行网络重构的并行计算,在配电网的建模中采用基于公用信息模型(C IM)的配网模型以及基于此模型的广度优先遍历方法有效地提高配网拓扑分析的效率,同时扩展了配网重构算法的开放性,算例结果表明,两种并行计算模型取得了较好的优化结果,显著地提高了计算速度。To solve the problem of long search time needed and early convergence of the genetic algorithm for distribution network reconfiguration computation, the distributed parallel computation method is introduced. The parallel computation of network reconfiguration for multi-machine in the local network was carried out based on two kinds of models, i.e. the function level master to slave parallel model and the community level peer to peer parallel model, and the CIM-based distribution network model and its breadth-first traverse method were applied to distribution network modeling, which greatly improves the efficiency of the net topology analysis and expands the openness of the distribution network configuration algorithm. Example calculation shows that optimization results are achieved based on the two parallel models and the computation speed is substantially increased.

关 键 词：并行计算 配网重构 遗传算法 公用信息模型(CIM) 

分 类 号：TM727[电气工程—电力系统及自动化]