An Improved Arithmetic Optimization Algorithm With Multi-Strategy Variable Step Size for Optimizing Engineering Problems
Year: 2025
Type: article
Abstract: Arithmetic Optimization Algorithm (AOA) has attracted considerable attention owing to its efficient search mechanism; however, it still exhibits deficiencies in optimization accuracy and convergence speed. To overcome these limitations, this paper proposes an improved arithmetic optimization algorithm based on multi-strategy variable step size, named MSVSAOA. The algorithm integrates four key strategies: a Lévy flight-based variable step size for the exploration phase, an update direction guidance-based variable step size for the exploitation phase, a doubly randomized variable step size for the boundary handling phase, and an information fusion strategy that combines the information of the best individual with that of the updated individuals. These improvements effectively enhance global exploration capability, refine local exploitation precision, and accelerate convergence. To evaluate its performance, MSVSAOA was tested on 29 functions from the CEC2017 test suite and compared with nine mainstream optimization algorithms. A comprehensive evaluation was conducted through statistical analysis, convergence analysis, Wilcoxon rank-sum tests, and Friedman tests. According to the Friedman ranking results, the average ranks of MSVSAOA, AOA, RCLSMAOA, HSMAAOA, AOAAO, GWOCS, GAT, tGSSA, ESO, and ISO were 1.48, 9.79, 5.55, 2.28, 7.10, 6.17, 5.03, 8.62, 3.55, and 5.41, respectively. Furthermore, the algorithm was successfully applied to four real-world engineering optimization problems. In all cases, MSVSAOA demonstrated faster convergence and superior optimization performance compared with AOA, with performance improvements of 0.27%, 0.31%, 8.91%, and 99.9998%, respectively, highlighting its strong practicality and broad application prospects.
Source: IEEE Access
Author: Jun Liu
Institution: Shangluo University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,075
Award: G788636226
OpenAlex ID: W4413394089