Improved Genetic Algorithm-based Distribution Network Reconstruction

Distribution network reconfiguration is not only an important method for circulatingand controlling of distribution system, but also an important part of distributionmanagement system. In theory, there exists a best network structure for any distributionnetwork. Under the best network structure, the voltage at each load point, network lossesand load balance are all better than other feasible methods. The purpose of distributionnetwork reconfiguration is to adjust the network structures by changing the open/closestates of the tie and subsection switches under both normal and abnormal operatingconditions. It is helpful to enhance the security, economy and reliability of the system.In this paper, firstly the network topology is analyzed with the inverted sequencenumbering method adopting breadth first search plan for numbers. Original back/forwardsweep method based on branch current is improved, adding the method for PV nodeswhich make the model of power flow solution more practical. Secondly, by analyzing thesimple genetic algorithm in detail, its global convergence is rectified with elite geneticalgorithm and adaptive genetic algorithm. The adaptive formulas of crossover probabilityand mutation probability are further adjusted to eliminate the early maturing phenomenon.Thirdly, focusing on minimizing network losses, the mathematics model of distributionnetwork reconfiguration is constructed. The problem of infeasible solutions in distributionnetwork reconfiguration is discussed and improved by encode method. Infeasible solutionsafter crossing and mutating are corrected to feasible solutions by amended crossoveroperating and mutation operating. Finally, the application program is complied based onthe proposed algorithm in this paper. Two standard test computations of single powersource and multiple power sources are verified. The validity of the algorithm isdemonstrated by the results.