Power restoration in distributed system is the problem to determine an operation for performing re-transfers to the section, which resulted due to power failure in some part of the power distribution system by feeder accidents and construction. Generally, the minimization of power distribution loss and load recovery is treated to optimize the problem, by switching the switches of the line segments to open or close state. Regarding to this problem, a number of studies about restoration methods such as distributed power grid interconnection and mitigation calculation time have been made. However, the evaluation method which used weighing factor so far, becomes a problem to be improved when trial and error is necessary for the coefficient determination. This paper compares the study of Genetic Algorithm (GA) which reduced the un-restored load recently with the method that applied multi-objective optimization called Non-dominating Sorting Algorithm GA-II(NSGA II) which able to solve the problem of weighting factor method. Furthermore, from the comparison to other references, the effectiveness of this proposed method is shown.