Abstarct: In recent years, functionality of energy infrastructures has been affected continuously by natural disasters, particularly earthquakes, due to the high dependence of power systems on energy. Due to reliability criteria have not a proper response to high impact rare (HR) events; power system resilience has become a concern for power grid planners to reduce the effects of the HR events. In this thesis, then, a new three-level planning frxamework is presented to optimally reinforce and expand the power distribution network with the consideration of multi-micro-grids (MMGs) in such a way that the devastating effect of seismic risks is mitigated. In the lower-level of newly proposed frxamework, after determining the worst destructive scenario, operation-baxsed corrective strategies such as resource scheduling and grid re-configuration are presented as short-term reaction of distribution network operator (DNO) to maximize benefit of micro-grids and minimize load shedding in the network. In the intermediate-level problem, destructive effects of seismic risks on the power distribution network components, such as substations, feeders, distributed energy resources (DERs) are modeled through a seismic scenario generation method (SSGM). In the SSGM, with a new point of view, maximum horizontal ground acceleration is modeled using a reduction procedure in terms of effective seismic parameters including soil type, seismic magnitude, depth of occurrence and surface distance. Additionally, and more importantly, the probability of complete destruction of the power distribution network components is extracted via predetermined fragility curves. Relying to maximum horizontal ground acceleration and probability of complete destruction, multiple seismic scenarios are generated by maximizing the technical-economic damage subject to the structural constraints; and then, the worst-case seismic scenario is selected. In the upper-level problem, however, the resilient optimal micro-grid expansion plans, as the long-term preventive actions after the seismic risks, are identified. The MMEP objectives, modeled through the third-level, are the minimization of the investment and operation costs and maximization of participation profits, while satisfying long- and short-term constraints over the planning horizon. To solve the proposed large-scale mixed-integer linear tri-level frxamework, a melody search algorithm (MSA) is widely employed. The proposed planning model is implemented on the 9-bus 33-kV test system to represent the feasibility and effectiveness of the newly developed model. The simulation results corroborate the effective performance of the proposed planning frxamework in improving the resilience of power distribution networks against seismic risks.