基于Kriging的多目标遗传算法

为了提高多目标优化问题的求解效率,提出了一种新的处理约束多目标优化问题的基于Kriging的多目标遗传算法(MOKGA)。MOKGA采用物理规划法将多目标优化转化为单目标优化,然后构建目标函数的考虑约束的EI(Expected Improvement)模型,并采用遗传算法进行求解。六峰值驼背函数和一个导弹多目标多学科设计优化问题用于MOKGA算法性能的测试。结果表明,与理论解相比,MOKGA算法有很好的优化结果;与NSGA II相比,MOKGA有很快的收敛性。

Multiobjective Genetic Algorithm Based on Kriging

In order to improve the efficiency for solving multiobjective optimization problem ,a new method dealing with constrained multiobjective optimization is proposed ,which is multiobjective kriging based ge-netic algorithm （ MOKGA） .Physical programming is used in MOKGA to convert the multiobjective to a single objective ,and then EI （ Expected Improvement ） is made for the single objective taking into consid-eration the various constraints .Genetic algorithm is used to optimize the EI .Six-hump camel back func-tion and a multiobjective multidisciplinary design optimization of missile are used to test the performance of MOKGA.The results show that MOKGA can find good results as compared to theory solution and has fast convergence as compared with NSGA II .