The seismic disaster survey shows that substation has higher vulnerability under the strong seismic action

which significantly affects post-seismic disaster recovery and rescue efforts. In this paper

a 220 kV substation is taken as a case study to perform the sampling simulation of the system disaster

and the mixed integer linear programming (MILP) is used to generate optimal repair decision sequence. In this study

the MILP decision result is used to train an artificial neural network (ANN) model. The repair strategy based on the model is proposed and is suit-able for scenarios which is lack of high performance calculation condition. Then

the repair priorities for various components of substation after the seismic is analyzed

which can achieve the repair strategy in case of unknown fault position after the disaster. Finally

the seismic resilience retrofitting of substation is performed by using three distinct retrofitting strategies. The result shows that compared to MILP strategy

the ANN-based repair strategy can significantly reduce computational costs while maintaining the efficiency of repair decisions. In addition

The optimal retrofit-ting strategy with different budget is given. The effectiveness of the repair strategies and retrofitting strategies proposed in this paper are verified by those results.