Xiaokai Chen , Zhiming Wu , Hamid Reza Karimi , Qianhui Li , Zhengyu Li
{"title":"基于元学习和硬样本挖掘的多动力总成汽车统一深度强化学习能量管理策略","authors":"Xiaokai Chen , Zhiming Wu , Hamid Reza Karimi , Qianhui Li , Zhengyu Li","doi":"10.1016/j.conengprac.2025.106396","DOIUrl":null,"url":null,"abstract":"<div><div>Hybrid electric vehicles (HEVs) encompass diverse powertrain configurations and serve varied purposes. Commonly, energy management strategies (EMSs) have been developed separately for individual vehicle types and powertrain configurations under specific operating scenarios, often lacking generalizability across vehicle models and operating scenarios. To fill this gap, we propose a unified deep reinforcement learning (DRL) EMS based on meta-learning and online hard sample mining. This strategy enables adaptation to diverse vehicle types and powertrain configurations with minimal sample training through online fine-tuning. Firstly, meta-reinforcement learning is employed to simultaneously learn EMS for multiple vehicle types across various operating scenarios, establishing a base-learner capable of achieving satisfactory performance with minor adjustments when confronted with new configurations and operating scenarios. Furthermore, to mitigate the slow convergence associated with training multiple vehicle types and operating scenarios concurrently, hard sample mining method is used to optimize the presentation of random operating scenarios during training. This entails recording poorly performing conditions during training and prioritizing the training of simpler conditions before advancing to more challenging ones, thereby enhancing training efficiency through a scientifically informed approach. Additionally, we validate the proposed EMS on a simulated vehicle emulator. Results demonstrate a significant improvement in convergence efficiency, with respective enhancements of 40% in convergence efficiency while achieving comparable final performance metrics.</div></div>","PeriodicalId":50615,"journal":{"name":"Control Engineering Practice","volume":"163 ","pages":"Article 106396"},"PeriodicalIF":5.4000,"publicationDate":"2025-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"A unified deep reinforcement learning energy management strategy for multi-powertrain vehicles based on meta learning and hard sample mining\",\"authors\":\"Xiaokai Chen , Zhiming Wu , Hamid Reza Karimi , Qianhui Li , Zhengyu Li\",\"doi\":\"10.1016/j.conengprac.2025.106396\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>Hybrid electric vehicles (HEVs) encompass diverse powertrain configurations and serve varied purposes. Commonly, energy management strategies (EMSs) have been developed separately for individual vehicle types and powertrain configurations under specific operating scenarios, often lacking generalizability across vehicle models and operating scenarios. To fill this gap, we propose a unified deep reinforcement learning (DRL) EMS based on meta-learning and online hard sample mining. This strategy enables adaptation to diverse vehicle types and powertrain configurations with minimal sample training through online fine-tuning. Firstly, meta-reinforcement learning is employed to simultaneously learn EMS for multiple vehicle types across various operating scenarios, establishing a base-learner capable of achieving satisfactory performance with minor adjustments when confronted with new configurations and operating scenarios. Furthermore, to mitigate the slow convergence associated with training multiple vehicle types and operating scenarios concurrently, hard sample mining method is used to optimize the presentation of random operating scenarios during training. This entails recording poorly performing conditions during training and prioritizing the training of simpler conditions before advancing to more challenging ones, thereby enhancing training efficiency through a scientifically informed approach. Additionally, we validate the proposed EMS on a simulated vehicle emulator. Results demonstrate a significant improvement in convergence efficiency, with respective enhancements of 40% in convergence efficiency while achieving comparable final performance metrics.</div></div>\",\"PeriodicalId\":50615,\"journal\":{\"name\":\"Control Engineering Practice\",\"volume\":\"163 \",\"pages\":\"Article 106396\"},\"PeriodicalIF\":5.4000,\"publicationDate\":\"2025-05-20\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Control Engineering Practice\",\"FirstCategoryId\":\"94\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S0967066125001595\",\"RegionNum\":2,\"RegionCategory\":\"计算机科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"AUTOMATION & CONTROL SYSTEMS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Control Engineering Practice","FirstCategoryId":"94","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0967066125001595","RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"AUTOMATION & CONTROL SYSTEMS","Score":null,"Total":0}
A unified deep reinforcement learning energy management strategy for multi-powertrain vehicles based on meta learning and hard sample mining
Hybrid electric vehicles (HEVs) encompass diverse powertrain configurations and serve varied purposes. Commonly, energy management strategies (EMSs) have been developed separately for individual vehicle types and powertrain configurations under specific operating scenarios, often lacking generalizability across vehicle models and operating scenarios. To fill this gap, we propose a unified deep reinforcement learning (DRL) EMS based on meta-learning and online hard sample mining. This strategy enables adaptation to diverse vehicle types and powertrain configurations with minimal sample training through online fine-tuning. Firstly, meta-reinforcement learning is employed to simultaneously learn EMS for multiple vehicle types across various operating scenarios, establishing a base-learner capable of achieving satisfactory performance with minor adjustments when confronted with new configurations and operating scenarios. Furthermore, to mitigate the slow convergence associated with training multiple vehicle types and operating scenarios concurrently, hard sample mining method is used to optimize the presentation of random operating scenarios during training. This entails recording poorly performing conditions during training and prioritizing the training of simpler conditions before advancing to more challenging ones, thereby enhancing training efficiency through a scientifically informed approach. Additionally, we validate the proposed EMS on a simulated vehicle emulator. Results demonstrate a significant improvement in convergence efficiency, with respective enhancements of 40% in convergence efficiency while achieving comparable final performance metrics.
期刊介绍:
Control Engineering Practice strives to meet the needs of industrial practitioners and industrially related academics and researchers. It publishes papers which illustrate the direct application of control theory and its supporting tools in all possible areas of automation. As a result, the journal only contains papers which can be considered to have made significant contributions to the application of advanced control techniques. It is normally expected that practical results should be included, but where simulation only studies are available, it is necessary to demonstrate that the simulation model is representative of a genuine application. Strictly theoretical papers will find a more appropriate home in Control Engineering Practice''s sister publication, Automatica. It is also expected that papers are innovative with respect to the state of the art and are sufficiently detailed for a reader to be able to duplicate the main results of the paper (supplementary material, including datasets, tables, code and any relevant interactive material can be made available and downloaded from the website). The benefits of the presented methods must be made very clear and the new techniques must be compared and contrasted with results obtained using existing methods. Moreover, a thorough analysis of failures that may happen in the design process and implementation can also be part of the paper.
The scope of Control Engineering Practice matches the activities of IFAC.
Papers demonstrating the contribution of automation and control in improving the performance, quality, productivity, sustainability, resource and energy efficiency, and the manageability of systems and processes for the benefit of mankind and are relevant to industrial practitioners are most welcome.