解决区间混合性能指标优化问题的进化算法

Evolutionary Algorithms for Optimization Problems with Interval and Hybrid Indices

区间混合性能指标优化问题普遍存在,但是已有的解决方法比较少.为此,提出一种有效解决该问题的区间偏好大种群进化优化方法.利用大种群进化,提高算法的搜索能力;采用基于相似度的策略,估计用户没有评价的进化个体的隐式性能指标值,以减轻用户疲劳;采用Pareto占优比较不同进化个体时,通过计算序值相同的进化个体的用户偏好值,进一步细分进化个体的优劣;此外,通过求解优化问题,将采用区间表示的用户对不同性能指标的偏好量化.将该方法应用于室内布局这一典型的区间混合性能指标优化问题,并与其他3种方法比较,结果验证了该方法的优越性.

Optimization problems with interval and hybrid indices are common in real-world applications,but there are few methods to solve these problems.An evolutionary algorithm with a large population and interval preferences is presented to effectively solve the problems above in this paper.A large population is adopted to improve the performance of the algorithm in exploration.A similarity-based strategy is employed to estimate the values of implicit indices of the individuals not having been evaluated by the user to alleviate the user＇s fatigue.When Pareto domination is utilized to compare different individuals,the user＇s preference values of the individuals with the same rank are calculated to further clarify their performance.In addition,the user＇s preferences to different indices are expressed as intervals,and are quantized by solving an optimization problem.The proposed algorithm is applied to the interior layout problem,a typical optimization problem with both interval parameters in the explicit index and interval value of the implicit index.By comparing with other three optimization algorithms,the results validate its superiority.