飞机总体设计中的一种多目标鲁棒优化方法

Multi-objective robust optimization method to aircraft conceptual design

提出一种基于灵敏度的多目标鲁棒优化方法.针对各维设计变量存在扰动的情况,在原约束多目标优化模型上,附加偏差目标函数,并采用最差估计法对约束条件进行鲁棒可行性调整.采用全局敏度方程方法来计算目标函数和约束函数对设计变量的敏度,进而采用Pareto遗传算法搜索约束多目标优化问题的非劣解集,设计者可以根据不同的设计准则从中选择合适的设计点.将上述方法用于飞机总体参数优化设计,并与采用常规优化方法所得的优化结果进行了分析和比较.

A multi-objective robust optimization method based on sensitivity was proposed. Aiming at the condition that there exists disturbance in design variables, by adding bias function to original multi-objective optimization model and adopting worst-case method, constraints were adjusted to ensure robust feasibility. Global sensitivity equation （GSE） method was utilized to compute the sensitivity of objective function and constraint function to design variables. Furthermore, Pareto Genetic Algorithm （PGA） was employed to search pareto-optimal set of multi-objective optimization problem with constraints. Designer can select suitable design points according to different design criteria. The above method is applied in aircraft conceptual parameter optimization design, and its result is compared with the ones obtained by traditional methods.