基于混合优化算法的小子样复杂规律建模方法

针对在高速飞行器建模及动力学特性分析过程中,用多项式拟合方法求解飞行器小子样复杂规律建模问题的不足,提出了基于混合优化算法的高速飞行器数据拟合的新思路,即将建模中气动、动力小子样参数拟合问题转化为最短路径的优化问题,要求拟合出来的函数经过各数据节点的同时,走最短的路径,从而将数据拟合问题转化为非线性优化问题。该方法突破了传统多项式拟合阶次小于节点数的限制,获得了能够准确反映模型特性的平滑曲线。鉴于混合优化算法（仿生优化算法与序列二次规划的局部优化算法相结合）收敛半径大、优化精度高等优点,以某型高速飞行器小子样气动参数点为例,采用混合优化算法进行优化,在飞行器建模及动力学特性分析中用拟合得到的模型计算气动参数,可大大提高计算精度,仿真试验亦取得了满意的效果。

The Modeling Method for Small Sample Complicate Situation Based on the Hybrid Optimization Method

To overcome the runger phenomenal of the high order polynomial fitting, a new aerodynamic parameter fitting method for small sample complicate situation is proposed, which can be applied to high speed vehicle modeling and characteristic analyzing. Through imposing proper constrains, the parameter fitting problem can be transformed into nonlinear optimization problem. The selected fitting function at- tempts to cross the existing points and travels on the shortest path, thus determine the optimal objective. The present method breaks the traditional rule, that is, the order of polynomial fitting function must be smaller than grid points. The obtained accurate smooth curve can reflect the internal characters of aerody- narnic data, thus provides great improve on data analysis. Owing to large convergence district and high efficiency of hybrid optimization method （Genetic Algorithm combined Sequential Quadratic Program- ming）, the proposed method highly improves computational accuracy. The simulation results show that the performance of the proposed method is satisfactory.