基因遗传算法的理解与在城市肌理设计中的应用试验

Study on Genetic Algorithm and Experiments on Urban Tissue Design

在建筑与城市设计的过程中,经常会遇到多种相互冲突的设计目标,基因算法的理论为解决这类设计问题提供了一种新的方法。本研究通过三次相互关联、由简单到复杂的试验,讨论了基因遗传算法在设计中应用的可能。试验以三维模型为操作对象,引入了Octopus基因算法工具,按照生物进化概念设定模型的"身体结构""基因""变异率""自然选择条件"等试验参数。不同于传统设计方法一般只生成一个设计方案,基因算法可以生成大批量方案,并且测算出众多方案中的最优化方案。

In the design and architecture fields,the application of genetic algorithm（GA） as a design tool, offers the possibility to find trade-off solutions for problems that present multiple fitness criteria, which are usual y conflicting with one another. This research aims to understand how Evolution and Embryological Development theories can be applied to design through three experiments. The developed experiments are based on three consecutive sequences, which start from applying standard operations to a simple geometry and progressively become more complex. The Octopus, one of GA softwares, with its various parameters are introduced in the experiments, such as ＂Body plan＂,＂Gene＂,＂Mutation＂, and ＂Fitness criteria＂.Unlike traditional design method, GA could generate numerous solutions extensively, and the optimized solution could come out through data analysis.