An improved goose optimization algorithm (I-GOOSE) based on self-adapting attenuation factors and lévy flight

To address the limitations of premature convergence, insufficient local search capability, and limited convergence accuracy in the Goose Optimization Algorithm (GOA) when applied to complex optimization problems, this paper proposes an Improved Goose Optimization Algorithm (I-GOOSE) based on dynamic exploration and exploitation ratio, self-adapting attenuation factors and Lévy flight. The proposed algorithm introduces three key enhancements: First, a dynamic exploration–exploitation ratio adjustment mechanism prioritizes global exploration in the early iterations and shifts to local exploitation in the later stages, thereby improving search efficiency. Second, an adaptive decay coefficient and step size parameter (coe) are introduced to optimize the dynamic adjustment of perturbation factors and displacement amplitudes, balancing global exploration and local exploitation capabilities. Third, the Lévy flight strategy is incorporated to leverage its long-jump characteristics, enhancing the algorithm's ability to escape local optima. Extensive experiments using the CEC2017 benchmark test suite demonstrate that I-GOOSE improves convergence accuracy by 3–10 orders of magnitude compared to the original algorithm, while exhibiting superior stability and faster convergence speed. Under identical experimental conditions, the average runtime of the improved algorithm is on the same order of magnitude as the original algorithm and mainstream comparison algorithms, indicating that performance gains do not come at the cost of significantly increased computational overhead. The improved algorithm demonstrates significant performance in escaping local optima and adapting to complex high-dimensional problems, providing a more competitive solution for engineering optimization challenges.