Abstract: To address the aerodynamic shape optimization of supersonic split line （SSSL） nozzle under multiple operation conditions, an efficient surrogate-assisted Bayesian optimization method was established in combination with three-dimensional numerical simulations, and the optimal solution was obtained. The variance-based Sobol sensitivity analysis method was employed to quantify the sensitivity of characteristic variables. The results demonstrate that the performance of the optimized SSSL nozzle is improved by 15.6%, indicating significant overall optimization benefits and validating the effectiveness of the optimization design method for the SSSL nozzle under multiple operating conditions. Partial dependence plot analysis indicates that the initial expansion-section half-angle α, the outlet expansion half-angle β and the downstream arc radius R exhibit weak overall effects on the response. The distance from the swing center to the throat diameter L_r dominates the response, whereas the interaction between the half-angle of the initial expansion section ( α ) and the expansion ratio at the split line ( ε₂ ) is strong.