Recently, the hybrid electric vehicle is more and more popular due to its advantage on the energy solution. Hence, the research of further optimizing the energy management of hybrid electric vehicle is crucial. Besides, the AI -assisted driving technologies are widely applied on such vehicles, which yields an important problem, that is how to manage the energy consumption when applying such technologies? As a typical AI-assisted driving situation, vehicle following needs more inves-tigation. Therefore, this paper proposes a method combining Multi-agent Deep Deterministic Policy Gradient (MADDPG) and sliding mode control (SMC) to deal with hybrid electric vehicle energy management and vehicle following problems. This method optimizes sliding mode control parameters through the exploration of MADDPG and takes into account the stability of SMC. The experimental results show that this method signifi-cantly achieves lower speed following error, more stable relative safety distance control and higher safety compared to traditional MADDPG algorithm, which implies less speed changes, higher passenger comfort under the constraints like safety and energy management. Meanwhile, less speed changes further contribute to less energy consumption.
Energy Management and Vehicle Following Strategy of Hybrid Electric Vehicle Based on Deep Reinforcement Learning and Sliding Mode Control
20.09.2024
2322566 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
SAE Technical Papers | 2023
|Fuel cell hybrid electric vehicle energy management method based on deep reinforcement learning
Europäisches Patentamt | 2024
|Europäisches Patentamt | 2025
|