遺伝アルゴリズムによる進化的形状設計法

概要

論文の詳細を見る
In the CAD system, shape design or shape modeling is one of the bottle-necks because the shape design is a trial and error process, and it depends on designer's intuition and experience. Therefore, it is expected to develop an automatic shape design system or automatic shape design methodology. This paper presents methodology to generate the suitable shape automatically by using Genetic Algorithm (GA) and Free-Form Deformation (FFD). GA is one of the optimization techniques based on concepts from population genetics. To apply GA for shape design problem, arbitrary shapes must be coded as chromosomes and genetic operators(reproduction, crossover, mutation, etc.) should be defined for those chromosomes. In this paper, the lattice of the FFD is introduced as the chromosome of the shape. As the FFD technique, the shape is deformed by moving the positions of the control points of the FFD lattice. By this technique, the chromosome is expressed as a set of a few numbers of control points (floating point representations). Using GA and FFD, the genetic shape design system which permits user to define shape evaluation functions (fitness functions) is constructed. As the fitness functions, any technique can be used, such as FEM analysis, by the form of subroutine programs.
社団法人精密工学会の論文
1993-09-05