Abstract: In this study, a typical geometric model for fault dislocation is established. The model uses total fitting length and superelevation change rate as dual objectives, and the lengths of transition curves, circular curves, and intermediate straight segments, along with the circular curve radius, as decision variables. The geometric closure constraints, based on Euler spirals and parametric equations for circular arcs, ensure engineering feasibility by iteratively aligning the transverse coordinates and azimuth at the end of the fitted line within preset tolerances. The nondominated sorting genetic algorithm Ⅱ (NSGA-Ⅱ) is used for multiobjective evolutionary computation to obtain a wellconverged Pareto solution set in the objective space, and a line offset area index I is introduced to quantitatively evaluate the scale of lateral overexcavation and the engineering volume of various schemes. Based on the resulting solutions, factors such as the rate of superelevation change, fitting length, seismic fortification impacts, and project costs are evaluated to screen representative schemes and recommend seismic fortification ranges and cross-sectional overexcavation. Using a railway tunnel as a case study, the  maximum sudden displacement is selected as the most unfavorable condition. The results indicate that the line-fitting method based on NSGA-Ⅱ multiobjective optimization provides operable tradeoff solutions among rapid restoration, engineering volume, and safety, offering a quantitative basis for the technical reserve in emergency plans and decision-making for rapid post-earthquake repairs.

Key words: active fault zone, railway tunnel, seismic displacement, line fitting, emergency restoration, nondominated sorting genetic-algorithm Ⅱ, multiobjective optimization