解全局优化问题的仿生类算法(I)—模拟进化算法

The Simulated-Biology-Like Algorithms for Global Optimization,Part Ⅰ:Simulated Evolutionary Aslgorithms

模拟进化计算(simulated evolutionary computation)与人工神经网络是近年来信息科学、人工智能与计算机科学的两大"热点"研究领域,由此所派生的求解优化问题的仿生类算法(遗传算法、演化策略、进化程序、神经优化、免疫系统等),由于其鲜明的生物背景、新颖的设计原理、独特的分析方法和成功的应用实践,正日益形成最优化理论与方法的一个崭新分支。本文扼要介绍这一新分支的形成,发展与现状,提出仿生类算法当前研究的热点与待解决问题,以引起数学工作者的广泛注意与兴趣。

The simulated evolutionary computation and artificial neural network research are two 'hor-point'areas in information science,computational intelligence and computer science in recent yeas.These investigations have lead to a rapid development of what we will call simulated-Biology -Like algorithms for solving optimization problems,among which,for instance,are genetic algorithm,evolution strategies,evolutionary programming,neural-network-based algorithms, self-organization map and simulated annealing techniques.All of these are forming a new branch of optimization theory,due to their solid biological fundation,novel idea of algorithm derivation,fresh approach for analysis as well as very successful practices in application s to side-range difficulty optimization problems.As an introduction to this new branch,we present a brief account of each approach of simulated-Biology-Like algorithms in this series of papers.The development of simulated evolutionary computation is described in the present pater.Some recent efforts and open problems in the areas are reviewed.