The rational generator start-up sequence and reliable restoration paths are very important for fast restoration of a power system. The unit restoration process of network reconstruction is analyzed

and a new multi-objective optimization model of unit restoration is proposed. The main optimization goal is to maximize the total power generation capability

and the restoration reliability of transmission lines and the impact on successive restoration process are also considered. The solution domain is set to slacken the constraint on primary and secondary optimization goals

while the differential evolution-estimation of distribution algorithm (DE-EDA) is introduced to improve the solution efficiency. Meanwhile

the Pareto-optimal generator start-up sequence

target network reconstruction plan as well as line restoration sequence can be obtained. Simulation results show that the computation burden can be reduced significantly by setting the solution domain and that the Pareto solutions obtained by DE-EDA have better convergence and distribution.