Service Restoration Of Distribution Systems Based On Genetic Algorithm

Service restoration of distribution systems is an important part of smart grid in the future and a major controlling method of realizing self-healing for power grid. After the occurrence of permanent fault, swiching of the loading in the unfalut area through closing tie switch and opening section switch is beneficial for the economy, safty and reliability of power system.This paper analizes the structure of distribution systems and the value and realizing process of service restoration; it concludes the domestic and overseas research status and raises new questions faced by optimizing process under the background of smart grid.This paper describes and simplifies the structure of network, builds three different models including restoration models of level0feeder, level1feeder and cut loading, and proposes objective functions and constraint conditions for every model. Gain fetures of every strategy as fetures of every strategy acquired through the back/forward sweep methord is used as the basis of evaluation.This paper analizes the theory of genetic algrorithm, shortbacks in multi-objective service restoration and advantages of adptive GA; it proposes that the applying of NSGA-II in service restoration can avoid the influence of the traditional weight. On the basis of NSGA-II, this paper offers a way of controlling selecting and viaration through Pareto optimizing path(POP); this way can adaptively change selecting size and viaration probility basing on evolution, reduce the probility of premature through enlarging searching area if there is no progress in continuous generation,and stop evolution duly to reduce redundant caulation;this paper increases efficiency and avoid to inefficient evolutions through similality cross operator, and method of deal with infeasible solution.through fuzzy set theory to ensure final strategy, this way can be applied to service restoration of distribution systems.In this paper, every model is testified by AGA, NSGA-II and POP NSGA-II.The results demonstrate feasibility and better robustness and global convergence of proposed algorithm, compared to AGA and NSGA-II.