Revamped dynamic opposite learning the hiking optimization algorithm for global optimization and engineering applications

In recent decades, numerous notable meta-heuristic algorithms have been developed to tackle complex optimization problems. Nevertheless, these existing meta-heuristics need to be enhanced with various evolutionary techniques to tackle new challenges in engineering applications. This analysis aims to improve the efficiency of the recently introduced Hiking Optimization Algorithm (HOA), a human-inspired method that simulates the complexities of a hiking journey aiming to reach a peak. The hiking environment, featuring rugged terrain with multiple local peaks and a global peak, mirrors an optimization search landscape. Like other algorithms, HOA has limitations, such as becoming trapped in local optima and lacking sufficient exploration capability, leading to premature convergence in challenging optimization problems. This study offers an upgraded version of HOA incorporating a revamped dynamic opposite learning (RDOL) strategy, called the revamped dynamic opposite learning HOA (RDOLHOA). RDOLHOA dynamically adjusts the weighting factor using a nonlinear function to balance exploration and exploitation. This innovation aims to reduce the randomness of the initial population and decrease the search effort required to find the optimal solution. Experiments on twenty-three diverse functions from the CEC 2005 test suites and various real world engineering problems show that RDOLHOA outperforms HOA with a faster convergence rate. Additionally statistical tests, particularly the Wilcoxon rank-sum test, demonstrate that RDOLHOA exceeds the performance of other competing algorithms in real world engineering design problems.