The invention discloses a fuel cell vehicle deep reinforcement learning energy management method based on experience pool migration, belongs to the field of fuel cell hybrid power energy management, and accelerates the development speed of a vehicle energy management strategy through a method of migrating training data among different types of fuel cell vehicles. And meanwhile, the automobile fuel economy can be improved. The method comprises the following steps: establishing two remarkably different fuel cell automobile models; establishing a depth deterministic strategy gradient agent model based on experience pool migration; the method comprises the following steps: training a DDPG proxy model for a vehicle type, and obtaining an experience pool of the trained DDPG proxy model; and migrating the experience pool to a new vehicle model, continuing to train the DDPG proxy model, and performing energy management on the new vehicle model by using the finally trained DDPG proxy. The empirical pool migration technology constructed by the invention can effectively utilize the training data of other vehicle types, and the training convergence speed of the migrated proxy model and the automobile fuel economy are improved.

    本发明公开了一种基于经验池迁移的燃料电池汽车深度强化学习能量管理方法,属于燃料电池混合动力能量管理领域,通过在不同类型的燃料电池汽车之间迁移训练数据的方法,加速车辆能量管理策略的开发速度,同时可以提升汽车燃油经济性。本发明方法包括:建立两种显著不同的燃料电池汽车模型;建立基于经验池迁移的深度确定性策略梯度代理模型;针对一种车型训练DDPG代理模型,获取训练后DDPG代理模型的经验池;迁移经验池至新的车型上,继续训练DDPG代理模型,利用最终训练完成的DDPG代理进行新车型的能量管理。本发明所构建的经验池迁移技术可以有效利用其他车型的训练数据,提高迁移后代理模型的训练收敛速度和汽车燃油经济性。


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Fuel cell vehicle deep reinforcement learning energy management method based on experience pool migration


    Weitere Titelangaben:

    一种基于经验池迁移的燃料电池汽车深度强化学习能量管理方法


    Beteiligte:
    ZHOU JIANHAO (Autor:in) / XU DA (Autor:in) / GUO AIJUN (Autor:in) / LI CHAOXIONG (Autor:in) / WU XUYANG (Autor:in)

    Erscheinungsdatum :

    2023-04-25


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Chinesisch


    Klassifikation :

    IPC:    B60L PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES , Antrieb von elektrisch angetriebenen Fahrzeugen / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen



    Fuel cell hybrid electric vehicle energy management method based on deep reinforcement learning

    MA YAN / LUAN YU | Europäisches Patentamt | 2024

    Freier Zugriff

    Hydrogen fuel unmanned aerial vehicle energy management method based on deep reinforcement learning

    WU YU / ZHANG MINGXUAN / WANG YUYANG et al. | Europäisches Patentamt | 2021

    Freier Zugriff

    Fuel cell vehicle energy management method combining working condition prediction and reinforcement learning

    LU YANHUI / ZHANG YUNRUI / WANG PENGYU et al. | Europäisches Patentamt | 2024

    Freier Zugriff

    Vehicle-mounted fuel cell system control method and system based on deep reinforcement learning

    LUAN WEILING / LI SONGYANG / WANG CHANG et al. | Europäisches Patentamt | 2022

    Freier Zugriff

    HEV energy management method based on deep reinforcement learning in vehicle following environment

    TANG XIAOLIN / CHEN JIAXIN / YANG KAI et al. | Europäisches Patentamt | 2020

    Freier Zugriff