Energy management strategy for hybrid electric vehicles based on deep reinforcement learning with consideration of electric drive system thermal characteristics
•An optimized energy management strategy based on deep deterministic policy gradient is proposed.•The strategy considers the thermal characteristics of the battery and the motor.•The temperature rise curves of both the battery and the motor are optimized by the strategy.•The strategy reduces fuel co...
Uložené v:
| Vydané v: | Energy conversion and management Ročník 332; s. 119697 |
|---|---|
| Hlavní autori: | , , , |
| Médium: | Journal Article |
| Jazyk: | English |
| Vydavateľské údaje: |
Elsevier Ltd
15.05.2025
|
| Predmet: | |
| ISSN: | 0196-8904 |
| On-line prístup: | Získať plný text |
| Tagy: |
Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
|
| Shrnutí: | •An optimized energy management strategy based on deep deterministic policy gradient is proposed.•The strategy considers the thermal characteristics of the battery and the motor.•The temperature rise curves of both the battery and the motor are optimized by the strategy.•The strategy reduces fuel consumption by approximately 8.46 % during the actual driving cycle.•Energy consumption reaches up to 95.82 % of the dynamic planning.
Deep reinforcement learning has emerged as a promising candidate for online optimised energy management in hybrid vehicles. However, previous studies have not considered the impact of the overall thermal characteristics of key components in a hybrid electric system on the system performance. In this paper, an energy management strategy based on deep deterministic policy gradient algorithm considering the thermal characteristics of the electric drive system is proposed for plug-in hybrid electric vehicles, aiming at controlling the battery and motor temperatures within a safe range and improving the vehicle’s overall performance of the vehicle. Firstly, the temperature models of battery and motor are constructed and introduced into the energy management strategy framework. Secondly, the weight coefficients are adjusted using an intelligent algorithm based on deep deterministic policy gradient to achieve the trade-off between multiple objectives. Simulation experiments are carried out based on a variety of typical cycling conditions, and the results show that the proposed strategy maintains the powertrain in the optimal operating temperature range by dynamically adjusting the operating states of the battery and motor. Compared with the original strategy, the final battery temperature is reduced by 2.557 °C, the motor temperature is reduced by 1.806 °C, and the fuel consumption is reduced by about 8.46 %. Moreover, the energy consumption can reach 95.82 % of the dynamic planning. These results not only verify the effectiveness of the proposed strategy in energy optimization, but also fully demonstrates its robustness by maintaining a stable output in stress tests with different boundary conditions. |
|---|---|
| ISSN: | 0196-8904 |
| DOI: | 10.1016/j.enconman.2025.119697 |