组合优化问题中GA性能分析及多样性评价

Performance Analysis and Diversity Evaluation of Genetic Algorithms in Combinatorial Optimization

随着生产调度、机器学习、最优规划等组合优化问题的大规模化,复杂化,传统的基于运筹学的搜索算法已显得无能为力.具有广域搜索能力的遗传算法(GA)也因“完备性”与“健全性”的不充分不能有效地对应上述问题.为此,本文提出了保证GA上述两个性质地方法,使其能有效地解决复杂组合优化问题.

With the large-scale and complexion of those combined optimization such as production scheduling, machine learning and optimization layout, the traditional optimization algorithm based on operational research had not be adapt for resolving those problems. While GA, which has wide scope searching capacity, also could not resolve those problems because of its insufficiency of completeness and soundness. This paper proposed a method that can guarantee the GAs two characters. It can resolve complex combined optimization problems effectively.