Discrete Structural Optimization with Set-Theoretical Jaya Algorithm

Discrete optimization of structures is known as a complex optimization problem with many local optima. Since metaheuristic algorithms do not require gradient information of the objective function and constraints, they are suitable for discrete optimization problems. A recently developed version of the Jaya algorithm (JA), called set-theoretical Jaya algorithm (ST-JA), has proven its effectiveness and robustness in solving structural optimization problems with continuous search spaces. In this paper, the ST-JA is applied to the discrete optimization of truss structures under stress and displacement constraints. The main idea of ST-JA is based on the division of the population of solutions into smaller well-arranged subpopulations of the same size. It follows that different subpopulations have different best and worst solutions. In this way, the ST-JA aims to strengthen both the exploration and exploitation capabilities of the classical JA and strike a balance between them. The performance of the ST-JA is demonstrated through four well-known truss optimization problems with discrete design variables, and its results are compared with those of the classical JA as well as other metaheuristic algorithms in the literature. To the best of our knowledge, this is the first time to apply ST-JA to discrete structural optimization. Numerical results reveal that ST-JA significantly outperforms the classical JA, especially in terms of convergence speed and accuracy, and provides results superior to other state-of-the-art metaheuristics.