动态环境中的进化算法

Evolutionary algorithms in dynamic environments

目前关于进化算法(EA)的研究主要局限于静态优化问题,然而很多现实世界中的问题是动态的,对于这类时变的优化问题通常并不是要求EA发现极值点,而是需要EA能够尽可能紧密地跟踪极值点在搜索空间内的运行轨迹.为此,综述了使EA适用于动态优化问题的各种方法,如增加种群多样性、保持种群多样性、引入某种记忆策略和采用多种群策略等.

Evolutionary algorithms（EAs） are widely and often used for solving stationary optimization problems where the fitness landscape or objective function does not change during the course of computation.However,the environments of real world optimization problems may fluctuate or change sharply.If the optimization problem is dynamic,the goal is no longer to find the extrema,but to track their progression through the search space as closely as possible.All kinds of approaches that have been proposed to make EAs suitable for the dynamic environments are surveyed,such as increasing diversity,maintaining diversity,memory-based approaches,multi-population approaches and so on.