摘要
为提高遗传算法的优化求解性能,从4个方面对遗传算法进行改进:即对障碍项及惩罚项因子的动态变化实现适应度函数的动态变化;对约束函数规格化处理以提高算法的搜索稳定性和算法的收敛速度;采用共享函数的方法使进化个体极易跳出局部最优而达到全局收敛;控制参数的动态变化以适应进化过程不同时期的需要.整个改进措施以面向对象的方法加以实现,并通过单级圆柱齿轮减速器设计实例验证,结果表明相对于常规优化算法,改进后的遗传算法使减速器体积减小25.8%,相比传统遗传优化算法使减速器体积减小5%,从而表明该改进遗传算法具有较高的优化求解效果.
In order to enhance solving ability of genetic algorithm,the genetic algorithm was improved from four aspects.Firstly,the dynamic change of fitness function was realized through the dynamic change of barrier item and penalty item.Secondly,the constrained functions were standardized in order to improve the searching stability and converging speed of the algorithm.Thirdly,sharing function was adopted to make individuals easy jump out of local solutions and reach globally optimal solutions.Last,parameters are dynamically controlled to meet the demand of the evolution process at different stages.The entire improved measures were realized by the object-oriented method and tested by the designed example of the single cylindrical gear velocity reducer.Experimental results show that the improved algorithm can save the volume of velocity reducer by 25.8 %.Compared with the 5 % save by conventional genetic algorithm,it is a remarkable improvement.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2008年第7期89-92,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
陕西省自然科学基金资助项目(2007E218)
河南省教育厅自然科学研究计划资助项目(2008B520030)
信阳师范学院青年骨干教师资助计划
信阳师范学院青年基金资助项目(20070204)
关键词
遗传算法
优化设计
面向对象
动态变化
genetic algorithm
optimization design
object-oriented
dynamic change
作者简介
刘道华（1974-），男，博士研究生，E—mail：ldhzzx@163．com．
