摘要
本文提出了一种用机器学习来改进遗传算法搜索效率的新方法———基于范例学习的遗传算法(简称CLGA),并将该方法成功应用于连续体结构拓扑优化。CLGA的基本思想是利用已搜索过的点的信息来指导后续的搜索,避免重复计算,从而提高了GA搜索的效率。本文提出了目标向量的概念,可以在不同尺度下量化链码产生的不同结构个体之间的相似性。算例的计算结果表明,该方法是一种高效的连续体拓扑优化方法。
A new method was described to improve the genetic algorithms based optimization by using the case based learning, which is applied successfully to the topology optimization of continuum structures. The basic idea is to utilize the sequence of points explored during a search to guide future exploration, which makes genetic algorithms more efficient. A new concept, Mass Vector, was proposed to make it possible to measure similar degree of different structures constructed by chain-code.
出处
《力学季刊》
CSCD
北大核心
2006年第3期489-494,共6页
Chinese Quarterly of Mechanics
基金
西南交通大学校基金(2002B08)
关键词
遗传算法
连续体结构
拓扑优化
范例学习
genetic algorithms
continuum structures
topologgy optimization
case based learning