摘要
提出了常微分方程组的演化建模的一种新算法 ,新算法在 3个方面改进了作者原有的算法 :(1)采用新的适应值评估方式 ;(2 )采用一种基于子空间搜索的遗传算法来优化模型的参数 ;(3)将传统的遗传程序设计方法与局部搜索技术相结合来优化模型的结构 .将新算法分别应用于人口增长与化学反应模型的自动建模 ,并比较两种算法的实验结果 ,表明新算法发现的模型更稳定、精确度更高 .
A new algorithm for the evolutionary modeling of system of ordinary differential equations ( ODEs) is presented in this paper.It has improved the old algorithm proposed by the authors in three aspects.First,it adopts a new way of fitness evaluation function.Secondly,it uses a genetic algorithm based on subspace search to optimize the parameters of a model.Thirdly itcombines the traditional genetic programming with local search technique to optimize the structure of a model.By applying the new algorithm to the automatic modeling of population growth and chemical reaction,it shows that the new algorithm can discover superior ODEs models which are more stable and accurate than the old algorithm does.
出处
《武汉大学学报（自然科学版）》
CSCD
2000年第5期549-553,共5页
Journal of Wuhan University(Natural Science Edition)
基金
国家 8 63计划!( 863 -3 0 6-ZT0 6-0 6-3 )
高等学校骨干教师资助计划
并行与分布处理国家重点实验室基金资助项目
关键词
演化建模
遗传算法
子空间搜索
常微分方程
evolutionary modeling
genetic algorithm
genetic programming
subspace search
