An Improved Archimedes Optimization Algorithm for Solving Optimization Problems

The Archimedes Optimization Algorithm (AOA) algorithm, a multi-agent-based metaheuristic, has garnered attention for its remarkable accuracy in real-world optimization. This research addresses solutions for the inherent limitation of original AOA, notably its susceptibility to uneven exploration and exploitation phases and its propensity to become ensnared in local optima. To overcome these limitations, we employ two strategies: the modification of the density decreasing factor and the introduction of a safe updating mechanism inspired by game theory. These enhancements are subjected to rigorous evaluation using 23 benchmark functions, and their performance is compared against that of the original AOA and other prominent algorithms, including the Multiverse Optimization (MVO), Grasshopper Optimization Algorithm (GOA), Sine Cosine Algorithm (SCA), and Ant Lion Optimizer (ALO). The test results reveal significant improvements achieved by the newly proposed improved AOA (IAOA), surpassing the performance of the original AOA in 69% of the optimization cases among the 23 test functions. It is noteworthy that it also outperformed the other mentioned algorithms. The potential of the proposed algorithm as an effective tool for addressing real-world optimization challenges is underscored by these encouraging findings, adhering to research conventions.