Abstract: An improved genetic algorithm combined with second-order cone programing is proposed for distribution network reconfiguration to minimize power losses and eliminate voltage violations. The chromosome length is shortened by using the decimal genetic coding rule based on the loop. For eliminating the infeasible solutions generated during the optimization, the infeasible solution judgment method based on the switch loop-node matrix and the node layer strategy is adopted in this paper. Through the elite reserve and dynamic control mutation rate, the problem about the premature convergence of the genetic algorithm is effectively solved. Then, some equivalent conversion and relaxation are presented to cast the initial nonlinear power flow equation into a second-order cone model, which can reliably and efficiently solve the global optimality by using the available commercial software. Finally, case studies on IEEE 33-node test feeder are conducted and the results demonstrate the validity as well as effectiveness of the proposed method.

Key words: Distribution network reconfiguration, second-order cone programing, genetic algorithm, node layer strategy, infeasible solution

摘要： 提出一种结合二阶锥规划的遗传算法应用于配网重构，降低了网损，消除了过电压。采用基于环路的十进制遗传编码规则，缩短了染色体长度，并利用开关环路矩阵和节点分层策略进行可行性判定。通过精英保留和动态控制变异率，避免了早熟，提高了收敛速度。通过等效变换和有效松弛，把原始的潮流方程转化为可以利用现有解算器高效求得全局最优解的二阶锥规划模型。应用所提算法对IEEE33节点配电网进行算例分析，仿真结果验证了所提出算法的可行性和计算高效性。

关键词: 配网重构, 二阶锥规划, 遗传算法, 节点分层策略, 不可行解