Topology optimization using an adaptive genetic algorithm and a new geometric representation

A new method for structural topology optimization using a genetic algorithm is proposed in this paper. This method uses a topology representation by Bézier curves with varying thickness material distribution in a finite element model. Two groups of variables are considered: control points of Bézier curves and thickness values for each curve. This new technique avoids the formation of disconnected elements and checkerboard patterns in optimal topology design. New adaptive strategies for crossover and mutation operators are also proposed. The numerical program has been developed in MATLAB and tested on a benchmark 2-D problem of topology optimization. The results obtained are compared with those obtained by the homogenization method and ANSYS code.