摘要
为了解决遗传算法(GA)中好的建筑块被破坏的问题,提出基于CBR-GA的建筑块重组方法。首先反复运行简单的遗传算法找到多个局部最优解,并选择多个不同的局部最优解构建案例库;然后应用差异化学习方法产生新案例,有效解决了GA中随机交叉对建筑块破坏的问题;最后通过4阶欺骗问题验证了新算法的可行性。
In order to solve the problem of the destruction of good building blocks in GA,this paper presented a novel approach to recombine the building blocks based on CBR-GA.Firstly,ran simple GA repeatedly to find multiple local optimal solutions,and chose the different local optimal solutions to build the case-base.Secondly,generated the new case by differentiation learning which effectively solved the problem of the destruction of building block in GA.Finally,verified the new algorithm with 4-order deceptive problem.
出处
《计算机应用研究》
CSCD
北大核心
2011年第5期1658-1660,1673,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(60763012
40761027)
广西自然科学基金资助项目(0991104)
关键词
案例推理
遗传算法
欺骗问题
case-based reasoning
genetic algorithm
deceptive problem
