Abstract: For the problem of muti-objective of gas-ejection interior ballistic leading to difficult optimization by method of exhaustion， in the condition where the length of initial chamber was shortened to improve vehicle maneuverability， the flow mechanisms of smoothing missile-bottom pressure shock by arranging annular baffle and reducing canister-bottom pressure by increasing diversion-cone radius were analyzed， and the structure scheme for combining change of the fifth annular baffle and diversion-cone radius was put forward. The approximate mathematical model based on Optimal Latin Hypercube Design and the sixth order response surface and the combination optimization strategy platform based on Multi-Island Genetic Algorithm and Sequential Quadratic Programming were established. The numerical results show that the coupling relationship among four optimal targets can be balanced by proposed structural strategies， and the limitation of large amount of calculation for optimization by method of exhaustion can be overcome and the combination strategy can complement advantages of two algorithms in order to capture optimal solution quickly. After optimization， the acceleration peak is reduced by 6.6%， and the acceleration curve smoothness increased by 79.5%， and the canister-bottom pressure peak increased by 16.2%， and the thermal environment quality of missile-bottom increased by 6.1%.

Key words: Gas-ejection；Annular baffle；Diversion-cone；Experimental design；Approximate mathematical model；Combination optimization design

摘要： 针对燃气弹射内弹道目标多导致传统穷举法寻优难的问题，在缩短初容段以提高车载机动性能的工况下，分析了环形隔板平滑弹底压力冲击、增大导流锥半径、减小筒底压力的流场机理，提出了第5块环形隔板和导流锥半径共同变化的结构方案；建立了基于优化拉丁超立方和六阶响应面的近似数学模型，利用多岛遗传算法和序列二次规划算法搭建了组合优化策略平台。数值结果表明，提出的结构策略能平衡4个优化目标间耦合关系，搭建的优化平台能克服穷举寻优计算量大的局限，算法组合策略能互补优点而迅速捕捉最优解；优化后，加速度峰值减小了6.6%，加速度曲线平稳性增加了79.5%，筒底压力峰值增加了16.2%，弹底热环境品质增加了6.1%。

关键词: 燃气弹射；环形隔板；导流锥；试验设计；近似数学模型；组合优化设计