一种求解高维优化问题的多目标遗传算法及其收敛性分析

A Multi-Objective Genetic Algorithm for High Dimensionality Optimization Problem and Study of Its Convergence Properties

单纯Pareto遗传算法很难解决目标数目很多的高维多目标优化问题 ,在多个指标之间引入偏好信息 ,提出的多目标遗传算法使进化群体按协调模型进行偏好排序 ,改变了传统的基于Pareto优于关系来比较个体的优劣。另外讨论了算法在满足一定条件下具有全局收敛性 。

Pure Pareto genetic algorithms cannot be expected to perform well on problems that involve many competing objectives By introducing preference information among several goals, a multi objective genetic algorithm is proposed, whose character lies in that evolutionary population is preference ranked based on concordance model The algorithm transforms a normal method with which individuals are ranked by Pareto superior relationship Also, it is proven that the new algorithm can guarantee the convergence towards the global optimum under some condition Mathematics parses of typical computational samples and experiments show that it can achieve good convergent performance and speed