无重访单亲遗传算法在配电规划中的应用  

Application of Non-revisit Partheno-genetic Algorithm on Distribution Network Planning

作　　者：胡宇行[1] 卫志农[1] 孙国强[1] 陈婷[1] 马骏毅 

机构地区：[1]河海大学能源与电气学院,南京210098 [2]江苏镇江供电公司,镇江212000

出　　处：《电力系统及其自动化学报》2014年第11期47-52,共6页Proceedings of the CSU-EPSA

摘　　要：在求解配电网优化问题时,普通遗传算法GA(genetic algorithm)不能保证解的可行性,而单亲遗传算法PGA(partheno-genetic algorithm)虽然避免了不可行解,但会产生与已知解重复的新解,所以这两种算法都要增加额外的工作量。为减少类似的计算,提高算法搜索效率,提出无重访单亲遗传算法NRPGA(non-revisit parthenogenetic algorithm),将无重访功能与单亲遗传算法相结合,使用单亲遗传算法确保所有新解符合配电网的相关约束;然后对每一个新解使用无重访算法进行校验,以保证新解的唯一性;最后以配电网网架建设投资、运行费用、停电损失之和为目标函数,通过算例分析验证了NRPGA在空间搜索效率、快速收敛性、最优解命中率等方面的优势。Using genetic algorithm to solve the problem of optimizing distribution planning,GA cannot insure that all new distribution topologies are radial and connected.Although partheno-genetic algorithm （PGA） avoids generating the infeasible solutions,it can cause many same topologies.These two algorithms will increase the calculating burden.In this paper,non-revisit partheno-genetic algorithm is proposed to reduce unnecessary calculating burden and to improve search efficiency.As PGA can insure that all new topologies must meet the constrains and non-revisit approach makes sure that all topologies are unique.Based on minimizing the sum of construction investment,operating cost and power cut cost,it can be illustrated from simulation results that the proposed algorithm is better on space search efficiency,convergence and higher hit rate.

关 键 词：配电网规划 单亲遗传算法 无重访算法 多叉分割树存储 

分 类 号：TM71[电气工程—电力系统及自动化]