解多目标优化的均匀正交遗传算法.pdfVIP

18 6 Vol. 18 No. 6 2003 12 JOURNAL OF SYSTEMS ENGINEERING Dec. , 2003 ¹ 1 2 2 王宇平, 焦永昌 , 张福顺 ( 11 , 710071; 21 , 710071) :在多目标优化中, 各目标通常相互冲突且不可公度, 其最优解( 常称为Pareto 最优解) 往往有无穷多, 如 何在最优解集 中求出一组分布均匀且数量充足的代表解供决策者选择十分重要. 论文将均匀设计正交设 计与遗传算法相结 给出了解多目标优化的一种新方法, 并证明了其全局收敛性. 新方法用少的计算量便可 求出一组在最优解集 中分布均匀且数量充足的最优解. 计算机仿真也表明这种方法对不同的试验函数均 可用少的计算量求出一组在最优解集 中分布均匀且数量充足的最优解. : 遗传算法; 多目标优化; 均匀设计; 正交设计 : O224; TN911.7; TP18 :A : 1000- 5781(2003) 06- 0481- 06 Uniform and orthogonal genetic algorithm for multiobjective optimization 1 2 2 WANG Yu-ping , JIAO Yong-chang , HANG Fu-shun ( 11Dept. of Mat . Science, Faculty of Science, Xidian University, Xi. an 710071, C ina; 2. Institute of Antennas and Electro_Magnetic Scattering, Xidian University, Xi. an 710071, C ina) Abstract: Multiobjective optimization problems often involve incommensurable and competing objectives, and t e number of t eir Pareto optimal solutions is usually infinite, t us ow to find a sufficient number of uniformly distributed and representative Pareto optimal solutions for t e decision maker isvery important. A novel algorit m is proposed for multiobjective optimization by combining t e uniform design and ort ogonal design wit genetic algorit m in t is paper, and its global convergence is proved. T e new algorit m can find a sufficient number of uniformly distributed and representative Pareto optimal solutionswit less com- putation