Research on Integrated Planning Techniques for Typical Low-Earth Orbit Satellite Platforms and Payload Missions

Given the exponential growth in the number of Low-Earth Orbit satellites and the diversification of payload types, the traditional mode of separate planning for platform and payload tasks is inadequate for meeting the demands of efficient and precise collaborative scheduling. A solution integrating platform resource modeling, mission resource modeling, intelligent payload task scheduling, and platform-payload collaborative optimization is proposed. By establishing a hierarchical planning framework, global characterization and dynamic scheduling of large-scale satellite constellation resources are achieved. The algorithm employs deep reinforcement learning to optimize routine task scheduling and introduces a rolling window optimization mechanism to address emergency tasks. The simulation results demonstrate that, in a typical hybrid constellation scenario, the proposed algorithm achieves an overall task completion rate of 78.9%, an emergency task response rate of 100%, and an average computation time of 0.028 s. This algorithm outperforms traditional methods in task completion rate and emergency response time.