Abstract: With the increasing complexity of the power grid operation situation, the traditional operation rules obtained through a small number of typical operation modes can no longer meet the actual operation requirements. In view of the shortcomings in the existing technology, this study provides a multi-objective optimization mathematical model for power grid reconstruction based on two-stage fault analysis. Data mining and K-means clustering were used to generate fault scenario sets, and a multi-objective optimization reconstruction model was constructed, which was solved by differential evolution algorithm. The maximum loss load under different operation modes is calculated, and finally the optimal solution is selected through the evaluation model. The experimental results show that the reliability of the power grid is significantly improved, the stability and continuous power supply of the power system are guaranteed, and the loss and operation cost of the system are effectively reduced.