DRL-Based Dynamic Destroy Approaches for Agile-Satellite Mission Planning
Agile-satellite mission planning is a crucial issue in the construction of satellite constellations. The large scale of remote sensing missions and the high complexity of constraints in agile-satellite mission planning pose challenges in the search for an optimal solution. To tackle the issue, a dyn...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-09-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/15/18/4503 |
| _version_ | 1797577219373006848 |
|---|---|
| author | Wei Huang Zongwang Li Xiaohe He Junyan Xiang Xu Du Xuwen Liang |
| author_facet | Wei Huang Zongwang Li Xiaohe He Junyan Xiang Xu Du Xuwen Liang |
| author_sort | Wei Huang |
| collection | DOAJ |
| description | Agile-satellite mission planning is a crucial issue in the construction of satellite constellations. The large scale of remote sensing missions and the high complexity of constraints in agile-satellite mission planning pose challenges in the search for an optimal solution. To tackle the issue, a dynamic destroy deep-reinforcement learning (D3RL) model is designed to facilitate subsequent optimization operations via adaptive destruction to the existing solutions. Specifically, we first perform a clustering and embedding operation to reconstruct tasks into a clustering graph, thereby improving data utilization. Secondly, the D3RL model is established based on graph attention networks (GATs) to enhance the search efficiency for optimal solutions. Moreover, we present two applications of the D3RL model for intensive scenes: the deep-reinforcement learning (DRL) method and the D3RL-based large-neighborhood search method (DRL-LNS). Experimental simulation results illustrate that the D3RL-based approaches outperform the competition in terms of solutions’ quality and computational efficiency, particularly in more challenging large-scale scenarios. DRL-LNS outperforms ALNS with an average scheduling rate improvement of approximately 11% in Area instances. In contrast, the DRL approach performs better in World scenarios, with an average scheduling rate that is around 8% higher than that of ALNS. |
| first_indexed | 2024-03-10T22:05:00Z |
| format | Article |
| id | doaj.art-115abdd7771940a28013c66dce2659b6 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T22:05:00Z |
| publishDate | 2023-09-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-115abdd7771940a28013c66dce2659b62023-11-19T12:48:37ZengMDPI AGRemote Sensing2072-42922023-09-011518450310.3390/rs15184503DRL-Based Dynamic Destroy Approaches for Agile-Satellite Mission PlanningWei Huang0Zongwang Li1Xiaohe He2Junyan Xiang3Xu Du4Xuwen Liang5School of Information Science and Technology, ShanghaiTech University, Shanghai 201210, ChinaInnovation Academy for Microsatellites of Chinese Academy of Sciences, Shanghai 201306, ChinaSchool of Information Science and Technology, ShanghaiTech University, Shanghai 201210, ChinaSchool of Information Science and Technology, ShanghaiTech University, Shanghai 201210, ChinaInstitute of Mathematics HANS, Henan Academy of Science, Zhengzhou 450046, ChinaInnovation Academy for Microsatellites of Chinese Academy of Sciences, Shanghai 201306, ChinaAgile-satellite mission planning is a crucial issue in the construction of satellite constellations. The large scale of remote sensing missions and the high complexity of constraints in agile-satellite mission planning pose challenges in the search for an optimal solution. To tackle the issue, a dynamic destroy deep-reinforcement learning (D3RL) model is designed to facilitate subsequent optimization operations via adaptive destruction to the existing solutions. Specifically, we first perform a clustering and embedding operation to reconstruct tasks into a clustering graph, thereby improving data utilization. Secondly, the D3RL model is established based on graph attention networks (GATs) to enhance the search efficiency for optimal solutions. Moreover, we present two applications of the D3RL model for intensive scenes: the deep-reinforcement learning (DRL) method and the D3RL-based large-neighborhood search method (DRL-LNS). Experimental simulation results illustrate that the D3RL-based approaches outperform the competition in terms of solutions’ quality and computational efficiency, particularly in more challenging large-scale scenarios. DRL-LNS outperforms ALNS with an average scheduling rate improvement of approximately 11% in Area instances. In contrast, the DRL approach performs better in World scenarios, with an average scheduling rate that is around 8% higher than that of ALNS.https://www.mdpi.com/2072-4292/15/18/4503agile-satellite mission planninggraph attention networkdeep-reinforcement learninglarge-neighborhood searchdynamic destroy |
| spellingShingle | Wei Huang Zongwang Li Xiaohe He Junyan Xiang Xu Du Xuwen Liang DRL-Based Dynamic Destroy Approaches for Agile-Satellite Mission Planning Remote Sensing agile-satellite mission planning graph attention network deep-reinforcement learning large-neighborhood search dynamic destroy |
| title | DRL-Based Dynamic Destroy Approaches for Agile-Satellite Mission Planning |
| title_full | DRL-Based Dynamic Destroy Approaches for Agile-Satellite Mission Planning |
| title_fullStr | DRL-Based Dynamic Destroy Approaches for Agile-Satellite Mission Planning |
| title_full_unstemmed | DRL-Based Dynamic Destroy Approaches for Agile-Satellite Mission Planning |
| title_short | DRL-Based Dynamic Destroy Approaches for Agile-Satellite Mission Planning |
| title_sort | drl based dynamic destroy approaches for agile satellite mission planning |
| topic | agile-satellite mission planning graph attention network deep-reinforcement learning large-neighborhood search dynamic destroy |
| url | https://www.mdpi.com/2072-4292/15/18/4503 |
| work_keys_str_mv | AT weihuang drlbaseddynamicdestroyapproachesforagilesatellitemissionplanning AT zongwangli drlbaseddynamicdestroyapproachesforagilesatellitemissionplanning AT xiaohehe drlbaseddynamicdestroyapproachesforagilesatellitemissionplanning AT junyanxiang drlbaseddynamicdestroyapproachesforagilesatellitemissionplanning AT xudu drlbaseddynamicdestroyapproachesforagilesatellitemissionplanning AT xuwenliang drlbaseddynamicdestroyapproachesforagilesatellitemissionplanning |