Abstract
Structural topology optimization is a recognized technique for designing structures. Genetic algorithm (GA) provides a reliable approach to finding the optimal structure; however, it has been criticized for its high computational cost. Although many studies aim to design the representation to reduce the number of variables and thereby increase search efficiency, the actual computational time in the structure’s evaluation by the finite element method (FEM) remains high. This study proposes two methods, i.e., GA with fractal iteration in the loop (GAFI-ITL) and GA with fractal iteration out of the loop (GAFI-OOTL). These two methods combine GA and fractals to address high computational cost. GAFI-ITL leverages the fractals to increase the structural complexity without requiring additional parameters. GAFI-OOTL utilizes the self-similarity information between fractals and enables non-iterative state structures during the GA process, thus saving the time for running the FEM. Once the best non-iterative structure is determined, GAFI-OOTL builds the internal structure with fractal rules. This study uses two test problems to validate the effects of fractal iteration on the structures obtained and the efficiency of algorithms. The results indicate that fractals can reduce the compliance of the original structure while imposing limitations on the distribution of materials during optimization.