As one of core subjects in the modern power system restoration research, the power grid reconfiguration is discussed, and a novel optimal strategy involving the corresponding model and approach for power grid reconfiguration is presented in this paper. The goal of the proposed model is to find the shortest weighted path for generation unit start-up or load recovery in restoration duration whilst considering all kinds of constraints. The proposed model is considered as a typical partial minimum spanning tree problem from the mathematical point of view. The genetic algorithm method with characteristics of global optimization and handling the discrete variables easily and effectively is employed to solve this problem. Furthermore, the performance of genetic algorithm is optimized in order to improve calculation speed, stability and search efficiency further. To some extent, the proposed method can make the trade off between the simulation precision and the computational efforts much better. Finally, the IEEE 30-bus test system is applied as benchmark to demonstrate the effectiveness and validity of the proposed model and method.