Gradient-based methods are utilized in traditional topology optimization studies. This approach is based on a point-by-point technique, which depends on the gradient information of the objective functions regarded as independent variables. Despite the fast response, this approach can lead them to find local optimal solutions, but it faces problems in defining the global optimal solutions in large-scale problems or high-degree functions. For this reason, non-gradient methods that give better solutions have been developed to solve these struggles. This study examines the nature-inspired methods developed over the last 25 years, which are called Genetic Algorithms, Ant Colony Optimization, Particle Swarm Optimization, and Artificial Neural Networks, and presents flowcharts to illustrate their principles. These methods are clarified via a cantilever beam. According to the results, methods are compared, and Particle Swarm Optimization can provide a reasonable solution to determine the optimal solution. These comparisons are tabulated and may be a guide for researchers.
Şen et al. (Thu,) studied this question.