Multi-strategy enhanced Archimedes algorithm for global optimization and engineering problems

Abstract

Abstract The Archimedes optimization algorithm (AOA) is a novel metaheuristic algorithm based on Archimedes’ principle. Despite the competitive performance of AOA, it is still subject to drawbacks like local–global search imbalance, low convergence efficiency, and local stagnation. To overcome these limitations, this study proposes a multi-strategy enhanced Archimedes optimization algorithm (MEAOA) that incorporates three strategies into the AOA. The adaptive evolution strategy helps objects select an appropriate formula to update their positions in agreement with their evolutionary status, achieving a balance between holistic and specific search. Furthermore, the hybrid update mechanism of density and volume raises convergence efficiency and accuracy by integrating valid information into the optimization process. Finally, dual opposition-based learning is introduced to mitigate the inferior offspring in each iteration, which avoids optimization stagnation. To evaluate the performance of the MEAOA, experiments were conducted on the classical, CEC2019 and CEC2021 test suites, which comprise 43 benchmark functions with varying complexity levels. Its performance was compared with that of the AOA, its three enhanced versions, two famous optimizers, three CEC winners, and four latest algorithms. Statistical, convergence, and complexity analyses, as well as an ablation test, are conducted to validate the efficacy of MEAOA. Furthermore, an evaluation was performed on five engineering design problems in real-world applications to verify the performance of the algorithm. MEAOA outperforms the competing algorithms in over 88% of the test cases and engineering issues in terms of precision, efficiency, and stability.