基于混合遗传算法的多目标动态优化设计

针对多目标动态优化设计中，解空间复杂、目标函数高度非线性、部分目标函数灵敏度容易求得等特点，综合利用多目标优化技术、遗传算法和梯度优化技术，提出了一种新的多目标混合遗传算法。这种算法在“并列选择遗传法”的基础上尽可能多地利用目标函数的灵敏度信息以加速收敛速度，同时通过各目标的重要程度和满意程度来决定相应的子种群大小以达到各目标的协同优化。在探讨了该方法的思想之后，着重研究了其实现步骤和关键技术，最后用一个浸水旋转壳的多目标动态优化设计为例，证明了本文方法的有效性。

Hybrid genetic algorithm based multiobjective dynamic optimization

A hybrid genetic algorithm,HMGA,for solving multiobjective,nonlinear,and bad robustiness dynamic optimization problem is proposed.Compared to existed methods,the algorithm of this paper takes into account virtues of genetic algorithm,multiobjective optimization and sensitivity analysis.By introducing the concepts of weight and satisfaction degree,HMGA harmonize the optimization progress of all objective functions.At the same time,the use of sensitivities accelerates the efficiency of HMGA.A complex example of fluid-shell interaction system is optimized by HMGA.It is shown that the presented method of this paper has good efficiency and powerful optimization ability.