基于蚁群算法的多维有约束函数优化研究  

Research on multi-dimension and constrained function optimization based on ACA

作　　者：原思聪[1] 刘道华[1] 江祥奎[1] 曹体进[1] 

机构地区：[1]西安建筑科技大学机电工程学院,西安710055

出　　处：《计算机应用研究》2008年第6期1682-1684,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(50275113);陕西省自然科学基金资助项目(2007E218)

摘　　要：利用蚁群算法的基本原理,将多维有约束的优化问题通过罚惩因子方式转换为统一的多变量目标函数形式,并将所有独立变量分成不同的等份区域,以蚂蚁走过每一变量的一个区域并访问完所有变量所构成的构造图作为优化问题的可行解,获得这一可行解的过程即为蚁群算法的粗搜索;再将粗搜索所获得的解执行遗传交叉及变异操作,从而构建另一种精搜索蚁群算法以获得更精确的全局优化解。给出了基于蚁群算法的多维有约束函数优化的具体算法。通过其他三种优化方法及本文方法对行星轮系优化设计的对比求解,验证了该优化方法的高效性及准确性。This paper made use of basic principles of ant colony algorithm,optimization problem which was multi-dimensional and constrained were transformed into unified multi-variable target function by the way of punished factors,and all of the independent variables were divided into different equivalent area,and obtained constructional-graph which the ant visited an area of each variable acts as feasible solutions of optimization problem.So the process for accomplishing feasible solutions was called rough searching of ant colony algorithm.The solutions that rough searching obtained would be implemented the operation of crossover and mutation in order to construct the subtle searching ant colony algorithm and acquire more accurate global solution. The specific algorithm was given.Compared the others three kinds optimization methods with the method of this paper for solving the planetary gear system＇s optimization problem,the results indicate that this optimization method has higher accuracy and efficiency.

关 键 词：蚁群算法 多维函数优化 粗搜索过程 精搜索过程 具体算法 

分 类 号：TP202.7[自动化与计算机技术—检测技术与自动化装置]