Advancements in global optimization with an empowered capuchin search algorithm

Capuchin search algorithm (CSA) is a newly matured meta-heuristic algorithm based on the natural roaming habits of capuchin monkeys while foraging. The key flaws of this meta-heuristic include convergence to local optimums as well as early convergence. To get around these issues, this study presents a new variant of CSA, referred to as heterogeneous comprehensive opposite learning-based CSA (HCOLCSA), which combines heterogeneous comprehensive learning (HCL) and opposite-based learning (OBL) strategies. The HCL strategy is presented to modify the velocity term of all capuchins for the purpose of improving the exploration and exploitation behaviors in HCOLCSA, while the OBL strategy is embedded into HCOLCSA to boost its exploration and exploitation capabilities of the search space and prevent the trapping of local optimums. To verify the performance of the developed HCOLCSA algorithm, it was evaluated on a set of 29 benchmark functions of the IEEE CEC-2017 test group for dimensions 10, 30, 50, and 100, and then applied to the benchmark functions of the IEEE CEC-2011 evolutionary algorithm competition to demonstrate its reliability and suitability to solve real-world problems. Friedman’s tests, Holm’s tests, and convergence analysis were performed to assess the strength of HCOLCSA compared to others. The numerical findings demonstrate that in more than 92 percent of cases, HCOLCSA produced better results and earned the highest ranking. The results show that HCOLCSA outperforms competing algorithms, demonstrating that it can solve real-world problems represented by CEC-2011 standard functions.