After blackout, in order to improve the efficiency of system restoration for power grid containing distributed generation (DG), this paper partitions the grid into several subsystems and restores them in parallel. Firstly, black-start sources for each subsystem are selected by employing grey decision theory, and optimal restoration paths from black-start sources to candidate nodes are obtained by utilizing shortest path faster algorithm (SPFA). Secondly, multi-objective optimization model for optimal partitioning, which considers rapidity and security of restoration, is established. Then, the model is solved by multi-objective particle swarm optimization (MOPSO). Afterwards, for the purpose of reconfiguring primary skeleton network, load restoration inside each subsystem is optimized by adopting genetic algorithm (GA) based on the optimal partitioning scheme acquired previously. The effectiveness of the proposed scheme is validated by the simulation results of IEEE 30-bus system.