摘要
在基本的思维进化算法(Mind Evolutionary Computation, 原名Mind-Evolution-Based Machine Learning)框架[1]的基础上,吸取遗传规划(GP)中的树形编码思想,提出了利用树结构进行信息抽取的方法,进而实现了用于常微分方程组演化建模的趋同、异化算子,并获取了优良的效果,使MEC在非数值优化领域中得到了进一步应用。最后的仿真实例的结果表明,同GP方法[2]比较,MEC方法具有计算速度快、建模结果优和全局搜索性能好等明显优点。
Based on the fundamental Mind Evolutionary Computation (MEC), an effective method of similartaxis and dissimilation for MEC is present to solve modeling problem of systems of the ordinary differential equations. In the paper, a tree construct called information tree is built to record the information produced during the process of similartaxis and dissimilation and guide the new generation抯 producing according to the tree coding method in Genetic Programming GP). Finally, it is proved by the results of some experiments that the method of MEC presented in this paper, compared with the GP, holds the advantages of higher computing speed and better capability for global searching.
出处
《系统仿真学报》
CAS
CSCD
2002年第5期539-543,550,共6页
Journal of System Simulation
基金
国家自然科学基金资助(编号:60174002)
863智能计算机系统主题资助(编号:863-306-ZT06-06-6)