Dynamic gold rush optimizer: fusing worker adaptation and salp navigation mechanism for enhanced search
Abstract
The Dynamic Gold Rush Optimizer (DGRO) is presented as an advanced variant of the original Gold Rush Optimizer (GRO), addressing its inherent limitations in exploration and exploitation. While GRO has demonstrated efficacy in solving optimization problems, its susceptibility to premature convergence and suboptimal solutions remains a critical challenge. To overcome these limitations, DGRO introduces two novel mechanisms: the Salp Navigation Mechanism (SNM) and the Worker Adaptation Mechanism (WAM). The SNM enhances both exploration and exploitation by dynamically guiding the population through a stochastic strategy that ensures effective navigation of the solution space. This mechanism also facilitates a smooth transition between exploration and exploitation, enabling the algorithm to maintain diversity during early iterations and refine solutions in later stages. Complementing this, the WAM strengthens the exploration phase by promoting localized interactions among individuals within the population, fostering adaptive learning of promising search regions. Together, these mechanisms significantly improve DGRO's ability to converge toward global optima. A comprehensive experimental evaluation was conducted using benchmark functions from the Congress on Evolutionary Computation (CEC) CEC2013 and CEC2020 test suites across 30 and 50-dimensional spaces, alongside seven complex engineering optimization problems. Statistical analyses, including the Wilcoxon Rank-Sum Test (WRST) and Friedman Rank Test (FRT), validate DGRO's superior performance, demonstrating significant advancements in optimization capability and stability. These findings underscore the effectiveness of DGRO as a competitive and robust optimization algorithm.