摘要： 传统交互式遗传算法在优化隐式性能指标时会使用户产生疲劳,影响优化质量与优化效率。为此,提出一 种改进的交互式遗传算法。采用二元排序确定适应值评价的不确定度,根据评价序列的最大信息差异计算种群的 收敛率,通过收敛率衡量种群进化状态,基于适应值不确定度和种群收敛率设计自适应交叉算子和变异算子,给出 交叉概率和变异概率的计算公式,利用包含用户偏好信息的遗传策略引导进化,从而使进化结果更加客观。将该 算法应用于服装进化设计系统,结果表明,与传统交互式遗传算法(T-IGA)相比,该算法可获取更多的满意解,提高了优化效率。

关键词: 遗传算法, 自适应交叉, 变异概率, 适应值, 交互环境, 不确定性

Abstract: The traditional interactive Genetic Algorithm ( GA ) can make user fatigue on optimizing the implicit performance index which affects the optimization quality and optimization efficiency. It is necessary to enhance the performance of interactive GA in order to apply it to complicated optimization problems successfully. The uncertainty of individual fitness is calculated based on the evaluation difference between the adjacent individuals;the convergence rate is abstracted according to the biggest information differences in evaluation sequence which reflecting the convergence of evolutionary population. Based on these,the probabilities of crossover and mutation operation of evolutionary individuals are presented. It makes the results more objective by guiding the evolutionary strategy through user preference information,and it allows a better exploration of the searching space and gives better findings compared with the traditional interactive GA(T-IGA).

Key words: Genetic Algorithm ( GA ), adaptive crossover, mutation probability, fitness value, interactive environment, uncertainty

中图分类号: 

• TP399