Reinforcement learning energy management control strategy of electric tractor based on condition identification

lithium-titanate battery(LTB) and supercapacitor(SC) hybrid power drive electric tractors (ET) are one of the necessary means to achieve truly green agriculture. However, low traction efficiency, short battery life, and high energy consumption are the main reasons hindering the industrialization of...

Full description

Saved in:
Bibliographic Details
Main Authors: Liqiao Li, Jiangchun Chen, Jing Nie, Zongyu Gao
Format: Article
Language:English
Published: Elsevier 2025-09-01
Series:International Journal of Electrical Power & Energy Systems
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S0142061525003941
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849768541286301696
author Liqiao Li
Jiangchun Chen
Jing Nie
Zongyu Gao
author_facet Liqiao Li
Jiangchun Chen
Jing Nie
Zongyu Gao
author_sort Liqiao Li
collection DOAJ
description lithium-titanate battery(LTB) and supercapacitor(SC) hybrid power drive electric tractors (ET) are one of the necessary means to achieve truly green agriculture. However, low traction efficiency, short battery life, and high energy consumption are the main reasons hindering the industrialization of ET. This paper proposed a Q-network reinforcement learning(RL) energy management control strategy(EMCS) of ET based on driving condition identification(CI). Firstly, the power demand of ET during driving is regarded as a Markov process. The historical driving data are used to construct the driving conditions of ET and obtain the Markov power state transfer probability matrix(MPSTPM) under different CI; Second, to minimize the energy consumption of lithium-titanate battery and supercapacitor hybrid power system(HPS), the power allocation strategy for ET under different CI is obtained by a Q-network RL algorithm; Finally, an learning vector quantization neural network(LVQNN) is used to identify the current ET driving CI through online and real-time, and the control system makes real-time power output decision through the current driving CI. Using the actual ET driving data, the simulation results show that the Q-network RL of based on CI optimized EMCS can reduce the energy loss of HPS. Compared with the state machine control strategy and dynamic programming (DP) control strategy through simulation platform where it is shown to Q-network RL control strategy based on CI provide better performance. Compared with the state machine control strategy, the Q-network RL reduction the totle energy consumption 13.28 % and achieved 94.3 % performance compared with DP. Finally, the 90 kW experiments platform was used for experimental verification, and the test results verified the feasibility of this Q-network RL based on CI energy management control strategy in practical engineering applications.
format Article
id doaj-art-c12ba2e40286458783ba7a0f2d3c60cb
institution DOAJ
issn 0142-0615
language English
publishDate 2025-09-01
publisher Elsevier
record_format Article
series International Journal of Electrical Power & Energy Systems
spelling doaj-art-c12ba2e40286458783ba7a0f2d3c60cb2025-08-20T03:03:45ZengElsevierInternational Journal of Electrical Power & Energy Systems0142-06152025-09-0117011084610.1016/j.ijepes.2025.110846Reinforcement learning energy management control strategy of electric tractor based on condition identificationLiqiao Li0Jiangchun Chen1Jing Nie2Zongyu Gao3College of Mechanical and Electrical Engineering, Shihezi University, Shihezi 832003, China; Xinjiang Production and Construction Corps Key Laboratory of Modern Agricultural Machinery, Shihezi 832003, China; Northwest Key Laboratory of Agricultural Equipment, Ministry of Agriculture and Rural Affairs Shihezi 832003 ChinaCollege of Mechanical and Electrical Engineering, Shihezi University, Shihezi 832003, China; Xinjiang Production and Construction Corps Key Laboratory of Modern Agricultural Machinery, Shihezi 832003, China; Northwest Key Laboratory of Agricultural Equipment, Ministry of Agriculture and Rural Affairs Shihezi 832003 ChinaCollege of Mechanical and Electrical Engineering, Shihezi University, Shihezi 832003, China; Xinjiang Production and Construction Corps Key Laboratory of Modern Agricultural Machinery, Shihezi 832003, China; Northwest Key Laboratory of Agricultural Equipment, Ministry of Agriculture and Rural Affairs Shihezi 832003 ChinaCollege of Mechanical and Electrical Engineering, Shihezi University, Shihezi 832003, China; Xinjiang Production and Construction Corps Key Laboratory of Modern Agricultural Machinery, Shihezi 832003, China; Northwest Key Laboratory of Agricultural Equipment, Ministry of Agriculture and Rural Affairs Shihezi 832003 China; Corresponding author at: College of Mechanical and Electrical Engineering, Shihezi University, Shihezi 832003, China.lithium-titanate battery(LTB) and supercapacitor(SC) hybrid power drive electric tractors (ET) are one of the necessary means to achieve truly green agriculture. However, low traction efficiency, short battery life, and high energy consumption are the main reasons hindering the industrialization of ET. This paper proposed a Q-network reinforcement learning(RL) energy management control strategy(EMCS) of ET based on driving condition identification(CI). Firstly, the power demand of ET during driving is regarded as a Markov process. The historical driving data are used to construct the driving conditions of ET and obtain the Markov power state transfer probability matrix(MPSTPM) under different CI; Second, to minimize the energy consumption of lithium-titanate battery and supercapacitor hybrid power system(HPS), the power allocation strategy for ET under different CI is obtained by a Q-network RL algorithm; Finally, an learning vector quantization neural network(LVQNN) is used to identify the current ET driving CI through online and real-time, and the control system makes real-time power output decision through the current driving CI. Using the actual ET driving data, the simulation results show that the Q-network RL of based on CI optimized EMCS can reduce the energy loss of HPS. Compared with the state machine control strategy and dynamic programming (DP) control strategy through simulation platform where it is shown to Q-network RL control strategy based on CI provide better performance. Compared with the state machine control strategy, the Q-network RL reduction the totle energy consumption 13.28 % and achieved 94.3 % performance compared with DP. Finally, the 90 kW experiments platform was used for experimental verification, and the test results verified the feasibility of this Q-network RL based on CI energy management control strategy in practical engineering applications.http://www.sciencedirect.com/science/article/pii/S0142061525003941electric tractor(ET)condition identification(CI)Q-networkEnergy management control strategy(EMCS)Hybrid power system(HPS)
spellingShingle Liqiao Li
Jiangchun Chen
Jing Nie
Zongyu Gao
Reinforcement learning energy management control strategy of electric tractor based on condition identification
International Journal of Electrical Power & Energy Systems
electric tractor(ET)
condition identification(CI)
Q-network
Energy management control strategy(EMCS)
Hybrid power system(HPS)
title Reinforcement learning energy management control strategy of electric tractor based on condition identification
title_full Reinforcement learning energy management control strategy of electric tractor based on condition identification
title_fullStr Reinforcement learning energy management control strategy of electric tractor based on condition identification
title_full_unstemmed Reinforcement learning energy management control strategy of electric tractor based on condition identification
title_short Reinforcement learning energy management control strategy of electric tractor based on condition identification
title_sort reinforcement learning energy management control strategy of electric tractor based on condition identification
topic electric tractor(ET)
condition identification(CI)
Q-network
Energy management control strategy(EMCS)
Hybrid power system(HPS)
url http://www.sciencedirect.com/science/article/pii/S0142061525003941
work_keys_str_mv AT liqiaoli reinforcementlearningenergymanagementcontrolstrategyofelectrictractorbasedonconditionidentification
AT jiangchunchen reinforcementlearningenergymanagementcontrolstrategyofelectrictractorbasedonconditionidentification
AT jingnie reinforcementlearningenergymanagementcontrolstrategyofelectrictractorbasedonconditionidentification
AT zongyugao reinforcementlearningenergymanagementcontrolstrategyofelectrictractorbasedonconditionidentification