A multi-strategy fusion improved Kepler optimization algorithm: Comprehensive performance evaluation and three-dimensional UAV trajectory path planning under multiple constraints
Year: 2026
Type: article
Abstract: The Kepler Optimization Algorithm (KOA), as a recently developed metaheuristic algorithm, has demonstrated promising performance in various optimization tasks. However, its performance is still constrained by slow convergence speed, limited convergence accuracy, and an imbalance between exploration and exploitation capabilities. To overcome these limitations, this paper proposes a multi-strategy improved Kepler Optimization Algorithm, termed the Sophisticated Policies-based Kepler Optimization Algorithm (SPKOA). First, Latin Hypercube Sampling (LHS) is introduced for population initialization to enhance population diversity. Then, a dynamic orbital perturbation-improved Lévy flight mechanism is developed to expand the global search capability and achieve an adaptive balance between exploration and exploitation. Furthermore, a multi-elite gravitational mechanism is proposed to guide solutions toward promising regions by utilizing multiple high-quality individuals. A local fine-tuning mechanism is incorporated to improve exploitation capability, while an improved elite crossover recombination strategy is designed to preserve elite solutions and accelerate convergence. Extensive comparative experiments on the CEC2017, CEC2020, and CEC2022 benchmark test suites demonstrate that SPKOA achieves superior optimization performance in terms of convergence accuracy and stability compared with several advanced algorithms. Moreover, three-dimensional UAV path planning experiments are conducted in complex environments involving terrain obstacles, no-fly zones, and dynamic obstacles. The results show that SPKOA obtains the shortest path length among the seven compared algorithms while maintaining zero constraint violations and zero infeasible runs in all 30 independent trials, verifying its effectiveness and applicability for complex constrained optimization problems.
DOI: 10.1093/jcde/qwag060
Institution: Shenyang University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,650
OpenAlex ID: W7165961398