GA-DFP分级优化技术的研究  

A New Step Optimum Approach Based on Genetic Algorithm and Davidon-Fletcher-Powell Method

作　　者：汪晓东 叶美盈 

出　　处：《浙江师大学报（自然科学版）》1997年第2期19-22,26,共5页Journal of Zhejiang Normal University(Natoral Sciences)

摘　　要：本文提出了一种把遗传算法（GA）和交度量法（DFP）相结合的分级代化方法．即先以遗传算法进行优化，将其计算结果作为第二级变度量法的初始值．再进行优化计算。结果表明，该方法是一种提高代化质量的有效方法，既具备良好的全局收敛特性，又有较快的收敛速度和较高的优化精度。本文还通过举例说明了该方法的应用。In this paper, a new step optimum approach based on genetic algorithm(GA) and Davidon -Fletcher- Powell(DFP) method is presented- The genetic algorithm is applied and then its optimum results are used as the beginning values of the variables in the next step optimum approach that is the Davidon -Fletcher- Powell method. The results show that, the step optimum approach is effective for improving optimum quality. This approach has a better convergence property and the times of global search decrease obviously and a fair degree of accuracy can be achieved- Finally, the application of this approach is illustrated with an example.

关 键 词：遗传算法 变度量法 全局寻优 函数优化 最优化 

分 类 号：O242.23[理学—计算数学] O224[理学—数学]