The Particle Swarm Optimization (PSO) algorithm is a well-known nature-inspired technique used to tackle complex optimization problems, widely used by researchers and practitioners due to its simplicity and effectiveness. This paper introduces an improved version of PSO, called Particle Swarm Optimization-based Variables Decomposition Method (PSO-VDM), which utilizes a de-composition technique and a semi-random initialization strategy to divide the problem into subproblems, enhancing exploration and exploitation of the search space. To evaluate the proposed algorithm, a comparison with seven other well-known algorithms is conducted across 13 benchmark problems. The search performance of the algorithms is analyzed using both the test of Wilcoxon signed-rank and Friedman rank. The results of the comparisons and statistical analyses demonstrate that the strategies employed in the PSO-VDM algorithm make a significant contribution to the search process. These comparisons indicate that the PSO-VDM algorithm outperforms other state-of-the-art optimization algorithms in terms of solution quality, highlighting its potential to effectively tackle challenging optimization problems.
