Cooperative Mission Planning of Multiple Spacecrafts Using a Multiobjective Optimization Algorithm Based on Reinforcement Learning

Mission planning of spacecrafts is critical to their successful execution. This article will focus on cooperative mission planning of multiple spacecrafts problem (CMP-MSP). To tackle this challenge, an efficient multiobjective snake optimizer (MOSO) based on reinforcement learning (MOSORL) is devel...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on aerospace and electronic systems Jg. 61; H. 2; S. 2703 - 2718
Hauptverfasser: Qi, Yuheng, Gu, Defeng, Liu, Yuan, Zhu, Jubo
Format: Journal Article
Sprache:Englisch
Veröffentlicht: New York IEEE 01.04.2025
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Schlagworte:
Adaptive algorithms
Aerospace and electronic systems
Earth
Heuristic algorithms
Iterative methods
Machine learning
Maintenance engineering
Mission planning
multiobjective snake optimizer (MOSO)
Multiple objective analysis
multispacecraft planning
Optimization
Planning
Reinforcement learning
reinforcement learning (RL)
Search problems
Simulation
Space vehicles
Strategy
ISSN:0018-9251, 1557-9603
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Mission planning of spacecrafts is critical to their successful execution. This article will focus on cooperative mission planning of multiple spacecrafts problem (CMP-MSP). To tackle this challenge, an efficient multiobjective snake optimizer (MOSO) based on reinforcement learning (MOSORL) is developed. The algorithm integrates an adaptive MOSO with a population classification and bidirectional search mechanism to improve the performance. In addition, a repair strategy is introduced to prevent the generation of suboptimal individuals. To improve the efficiency and accuracy of solutions, reinforcement learning (RL) is introduced to update the iterative strategy of MOSO. A dynamic learning rate approach is implemented to enhance the algorithm's learning speed. The effectiveness of the proposed algorithm is validated through numerical simulation experiments on 16 sets of instances with varying mission scales. The experimental results confirm that MOSORL outperforms competing methods significantly in solving CMP-MSP.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0018-9251
1557-9603
DOI:10.1109/TAES.2024.3476456