Improving the Puma Optimization Algorithm Using a Hybridization of the Gradient-Dependent Religion Algorithm with a New Beta Coefficient

Abstract

Metaheuristic optimization algorithms often face difficulties in achieving an effective balance between exploration and exploitation, leading to slow convergence or entrapment in local optima. The Puma Optimization (PO) algorithm is a modern metaheuristic known for its strong exploratory capabilities; however, it suffers from deficiencies in refining solutions during the final stages of the optimization process. In this paper, a hybrid model is proposed that combines the PO algorithm with the hybrid Divine Religion Algorithm-Conjugate Gradient (DRA-CG) method, aiming to enhance overall performance. Specifically, PO is utilized for global exploration, while DRA-CG is employed to bolster local exploitation. Furthermore, a novel parameter, β, has been developed within the conjugate gradient component to improve convergence speed and numerical stability. Experimental results demonstrate that the proposed method achieves superior performance in terms of both accuracy and convergence speed, thereby validating its effectiveness in addressing complex optimization problems across diverse benchmark functions and real-world scenarios.