为解决静态近似模型所需样本量大、优化效率低的问题，基于粒子群算法（PSO）的最小二乘支持向量回归（LSSVR）自适应近似模型构建优化算法，并通过构建全局和局部自适应近似模型以减小优化算法陷入局部最优解的可能，加速收敛过程。文中将Branin函数作为测试函数，证明构建的自适应PSO-LSSVR近似模型用于单目标优化问题的有效性；将自适应PSO-LSSVR近似模型用于GTS模型低风阻尾板的快速优化上，以上尾板倾角、下尾板倾角、侧尾板倾角和尾板长度为设计变量，仅通过31组数据集样本便收敛至最优解，且近似模型预测气动阻力系数误差仅为0.18%。相比初始尾板，优化后的尾板使得GTS模型气动阻力下降9.38%，证明了自适应PSO-LSSVR近似模型优化算法对小样本快速寻优问题具有较好的可行性。

To solve the problems of large sample size and low optimization efficiency of static approximation model, the least squares support vector regression (LSSVR) based adaptive approximation model with particle swarm optimization (PSO) algorithm was introduced to construct the optimization algorithm. The global and local adaptive approximation models were constructed to reduce the possibility of the optimization algorithm falling into the local optimal solution and to accelerate the convergence process. The Branin function was used as test function to prove the effectiveness of the proposed adaptive PSO-LSSVR approximation model for single-objective optimization problems. The adaptive PSO-LSSVR approximation model was applied to the rapid optimization of boat-tail of GTS model. The upper boat-tail angle, the lower boat-tail angle, the side boat-tail angle and the tail plate length were taken as design variables, and the optimal solution could be obtained only with 31 sample data sets.  And the error of aerodynamic drag coefficient predicted by the approximation model is only 0.18%. The aerodynamic drag of GTS model with optimized boat-tail is reduced by 9.38% after optimization, which proves that the adaptive PSO-LSSVR approximation model optimization method is feasible for fast optimization problem with small samples.