改进的遗传算法在连续体结构多目标拓扑优化中的应用

连续体结掏多目标拓扑优化是结构优化领域中的一个较难的研究课题.本文提出了一种改进的SPEA2多目标优化算法.该算法中采用数学形态学中的四方向链码的编码方式进行结构拓扑表达,使产生的拓扑结构清晰且无异议,完全消除了模糊的拓扑边界和棋盘格现象.提出了"质量向量"的概念,用以量化两个结构拓扑之间的相似性,从而使得不同结构拓扑之间的相似性比较成为可能.同时还将机器学习中的范例学习的思想融入到新算法中,利用前面的有限元分析结果对后来的结构分析进行指导,剔除了很多不必要的重复的计算,从而使算法的总计算量大幅度地减少.将新算法应用于悬臂和简支两类深梁的多目标拓扑优化,获得了高质量的Pareto最优解,且具有很好的分布.

Multi-objective Topology Optimization of Continuum Structure using Improved Genetic Algorithm

Multi-objective topology optimization of continuum structure is a hard problem in structure opti- mization field.An improved SPEA2 algorithm was proposed.A coding method called four direction chain- code was used to present structure topology,which can make clear toplogy and remove fuzzy boundary and chessboard.A new concept"quality vector"was presented,which makes it possible to compare two dif- ferent topologies.And the machine learning was integrated with the new method,guiding upcoming cal- culation to ...