An adjacency representation for structural topology optimization using genetic algorithm

M3M Laboratory, University of Technology of Belfort-Montbeliard, 90010 Belfort Cedex, France

Corresponding author: matthieu.domaszewski@utbm.fr

Received: 9 July 2007
Accepted: 1 November 2007

Abstract

A new approach for continuum structural topology optimization using genetic algorithms is presented in this paper. The proposed approach is based on a representation by adjacency where the principle is founded on the concept of connectivity of finite elements, considered as cells. This principle is expressed by an adjacency matrix similar to that used in the graph theory. The encoding of the structure solutions uses this matrix by transforming it into a binary string. The research of optimal solution, i.e. the optimal material distribution, is interpreted in this approach by the determination of the connectivity of elements (cells). Using density variable, the approach has some common points with the homogenization techniques. The proposed approach is tested with simple benchmark applications.