Abstract: To address the insufficient dynamic adaptability in collaborative observation mission planning for low-earth orbit mega-constellations, an adaptive teaching-learning-based optimization algorithm was proposed. Within the teaching-learning framework, adaptive mechanisms and hybrid learning strategies were incorporated. The teaching phase was enhanced through time-varying teaching factors and an elite-guided mechanism, while the learning phase was improved using hybrid learning strategies to dynamically balance global exploration and local exploitation capabilities. Simulations demonstrated that the proposed algorithm outperformed both the improved genetic algorithm and the improved differential teaching-learning-based optimization algorithm in terms of task completion rate and computational time. In large-scale, high-complexity multi-satellite collaborative mission scenarios, it achieved 6% and 16% higher task completion rates compared to baseline algorithms, proving suitable for high-dimensional discrete optimization problems. The algorithm exhibits advantages in task completion rate, operational efficiency, and robustness, making it applicable to collaborative observation missions in low-earth orbit constellations.

Key words: agile satellite, mission planning, multi-point targets, earth observation, on-orbit planning, teaching-learning-based optimization algorithm