Research on operation optimization of heavy-haul combined trains in long and steep downhill sections based on reinforcement learning
To mitigate longitudinal impulse and address challenge posed by continuous air braking operations in long and steep downhill sections for 20 000-ton heavy haul combined trains, this paper proposes an approach for operation optimization of such trains featuring a long formation in such sections based...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
Editorial Department of Electric Drive for Locomotives
2023-11-01
|
| Series: | 机车电传动 |
| Subjects: | |
| Online Access: | http://edl.csrzic.com/thesisDetails#10.13890/j.issn.1000-128X.2023.06.017 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850274409050275840 |
|---|---|
| author | WANG Jianhua WANG Chunyi ZENG Zhou WANG Cong WANG Qingyuan YANG Hang |
| author_facet | WANG Jianhua WANG Chunyi ZENG Zhou WANG Cong WANG Qingyuan YANG Hang |
| author_sort | WANG Jianhua |
| collection | DOAJ |
| description | To mitigate longitudinal impulse and address challenge posed by continuous air braking operations in long and steep downhill sections for 20 000-ton heavy haul combined trains, this paper proposes an approach for operation optimization of such trains featuring a long formation in such sections based on a data-driven algorithm. An air braking force prediction model was developed based on neural network learning focusing on the variation rules of air braking performance across different operating states, to incorporate differences in air braking characteristics across different trains and varied braking system states on same trains. Then, an operation optimization algorithm was designed, establishing a reward function that prioritized speed following features, based on reinforcement learning. This algorithm incorporated constraints including train traction/electric braking characteristics, braking application and release times of train pipes, speed limits, and operational smoothness, to optimize the operation strategy for heavy-haul combined trains in long and steep downhill sections by reinforcement learning. The proposed air braking simulation model and operation optimization algorithm were verified using data collected from operation of real trains, confirming their feasibility and rationality. The results show the effectiveness of the proposed air braking force prediction model in predicting performance of air braking systems on running trains. Compared to manual driving, the optimized operation strategy plays an effective role in reducing longitudinal impulse and maximum coupler force to ensure the safety during train operation. |
| format | Article |
| id | doaj-art-4f67a244ee534e08be5e072a18009c40 |
| institution | OA Journals |
| issn | 1000-128X |
| language | zho |
| publishDate | 2023-11-01 |
| publisher | Editorial Department of Electric Drive for Locomotives |
| record_format | Article |
| series | 机车电传动 |
| spelling | doaj-art-4f67a244ee534e08be5e072a18009c402025-08-20T01:51:09ZzhoEditorial Department of Electric Drive for Locomotives机车电传动1000-128X2023-11-0113914647324154Research on operation optimization of heavy-haul combined trains in long and steep downhill sections based on reinforcement learningWANG JianhuaWANG ChunyiZENG ZhouWANG CongWANG QingyuanYANG HangTo mitigate longitudinal impulse and address challenge posed by continuous air braking operations in long and steep downhill sections for 20 000-ton heavy haul combined trains, this paper proposes an approach for operation optimization of such trains featuring a long formation in such sections based on a data-driven algorithm. An air braking force prediction model was developed based on neural network learning focusing on the variation rules of air braking performance across different operating states, to incorporate differences in air braking characteristics across different trains and varied braking system states on same trains. Then, an operation optimization algorithm was designed, establishing a reward function that prioritized speed following features, based on reinforcement learning. This algorithm incorporated constraints including train traction/electric braking characteristics, braking application and release times of train pipes, speed limits, and operational smoothness, to optimize the operation strategy for heavy-haul combined trains in long and steep downhill sections by reinforcement learning. The proposed air braking simulation model and operation optimization algorithm were verified using data collected from operation of real trains, confirming their feasibility and rationality. The results show the effectiveness of the proposed air braking force prediction model in predicting performance of air braking systems on running trains. Compared to manual driving, the optimized operation strategy plays an effective role in reducing longitudinal impulse and maximum coupler force to ensure the safety during train operation.http://edl.csrzic.com/thesisDetails#10.13890/j.issn.1000-128X.2023.06.017heavy-haul combined trainair brakinglong and steep downhillneural networkreinforcement learningheavy-haul rallway |
| spellingShingle | WANG Jianhua WANG Chunyi ZENG Zhou WANG Cong WANG Qingyuan YANG Hang Research on operation optimization of heavy-haul combined trains in long and steep downhill sections based on reinforcement learning 机车电传动 heavy-haul combined train air braking long and steep downhill neural network reinforcement learning heavy-haul rallway |
| title | Research on operation optimization of heavy-haul combined trains in long and steep downhill sections based on reinforcement learning |
| title_full | Research on operation optimization of heavy-haul combined trains in long and steep downhill sections based on reinforcement learning |
| title_fullStr | Research on operation optimization of heavy-haul combined trains in long and steep downhill sections based on reinforcement learning |
| title_full_unstemmed | Research on operation optimization of heavy-haul combined trains in long and steep downhill sections based on reinforcement learning |
| title_short | Research on operation optimization of heavy-haul combined trains in long and steep downhill sections based on reinforcement learning |
| title_sort | research on operation optimization of heavy haul combined trains in long and steep downhill sections based on reinforcement learning |
| topic | heavy-haul combined train air braking long and steep downhill neural network reinforcement learning heavy-haul rallway |
| url | http://edl.csrzic.com/thesisDetails#10.13890/j.issn.1000-128X.2023.06.017 |
| work_keys_str_mv | AT wangjianhua researchonoperationoptimizationofheavyhaulcombinedtrainsinlongandsteepdownhillsectionsbasedonreinforcementlearning AT wangchunyi researchonoperationoptimizationofheavyhaulcombinedtrainsinlongandsteepdownhillsectionsbasedonreinforcementlearning AT zengzhou researchonoperationoptimizationofheavyhaulcombinedtrainsinlongandsteepdownhillsectionsbasedonreinforcementlearning AT wangcong researchonoperationoptimizationofheavyhaulcombinedtrainsinlongandsteepdownhillsectionsbasedonreinforcementlearning AT wangqingyuan researchonoperationoptimizationofheavyhaulcombinedtrainsinlongandsteepdownhillsectionsbasedonreinforcementlearning AT yanghang researchonoperationoptimizationofheavyhaulcombinedtrainsinlongandsteepdownhillsectionsbasedonreinforcementlearning |