Abstract: In response to the diverse types of faults and the scattered resources in the emergency restoration work of distribution systems within a smart grid environment, this paper proposes an optimization method for the emergency restoration process based on an improved combination of the XGBoost algorithm and ant colony algorithm. This method utilizes historical data from smart grids to establish models for fault classification and restoration time prediction, and employs an enhanced ant colony algorithm for intelligent scheduling of emergency resources. The research results indicate that the proposed method can accurately predict fault types and restoration times, optimize emergency response paths, and significantly improve restoration efficiency.