基于CBR-GA的建筑块重组  

Toward recombination of building blocks based on CBR-GA

作　　者：邹鹏[1] 闭应洲[1] 杨虎林[1] 刘阿宁[1] 元昌安[1] 

机构地区：[1]广西师范学院计算机与信息工程学院,南宁530004

出　　处：《计算机应用研究》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)

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.

关 键 词：案例推理 遗传算法 欺骗问题 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]