Toward Low Time Fragmentation of Equipment: A Double-DQN Based TT C Task Planning Approach

• Published in: Applied sciences Vol. 13; no. 10; p. 6326
• Main Authors: Hangkun Xu, Runzi Liu
• Format: Journal Article
• Language: English
• Published: MDPI AG 01-05-2023
• Subjects: deep reinforcement learning; low time fragmentation; satellite tracking, telemetry, and command; task planning
• ISSN: 2076-3417
• DOI: 10.3390/app13106326

With the increase of the number of satellites in space, satellite tracking, telemetry, and command (TT&C) is becoming more and more important for aerospace. This paper proposes a method for a low time fragmentation oriented. TT&C task planning method based on Double Deep Q-Network (DDQN). This method mainly solves the problem of poor responses to emergency tasks caused by the large amount of time fragments of equipment under traditional TT&C task-planning methods. Firstly, a multi-objective optimization model aiming at minimizing time fragments and maximizing task revenue is formulated. Then, according to the conflict characteristics of tasks, a task-planning conflict graph is proposed, based on which a TT&C task-planning problem is transferred into an independent set problem. Finally, DDQN is combined with graph structure embedding to solve the transferred independent set problem. The experimental results demonstrate that the proposed method reduces the time fragment of TT&C equipment by 32% and shortens the response time of emergency tasks by 36% compared to existing methods.