Grey Wolf Optimizer Based on Variable Population and Strategy for Moving Target Search Using UAVs

Unmanned aerial vehicles (UAVs) are increasingly favored for emergency search and rescue operations due to their high adaptability to harsh environments and low operational costs. However, the dynamic nature of search path endpoints, influenced by target movement, limits the applicability of shortest path models between fixed points in moving target search problems. Consequently, the moving target search problem using UAVs in complex environments presents considerable challenges, constituting an NP-hard problem. The Grey Wolf Optimizer (GWO) is known for addressing such problems. However, it suffers from limitations, including premature convergence and instability. To resolve these constraints, a Grey Wolf Optimizer with variable population and strategy (GWO-VPS) is developed in this work. GWO-VPS implements a variable encoding scheme for UAV movement patterns, combining motion-based encoding with path-based encoding. The algorithm iteratively alternates between global optimization and local smoothing phases. The global optimization phase incorporates: (1) a Q-learning-based strategy selection; (2) position updates with obstacle avoidance and energy consumption reduction; and (3) adaptive exploration factor. The local smoothing phase employs four path smoothing operators and Q-learning-based strategy selection. Experimental results demonstrate that GWO-VPS outperforms both enhanced GWO variants and standard algorithms, confirming the algorithm’s effectiveness in UAV-based moving target search simulations.