An Unbiased Fuzzy Double Q-Learning based Energy Management for Fuel Cell Hybrid Electric Vehicles Publisher: IEEE PDF
Résumé
In the paper, a fuzzy double Q-learning (FDQL) based energy management strategy is proposed for fuel cell hybrid electric vehicles (FCHEV). Model-free characteristic of the proposed novel reinforcement learning enable the agent to enhance performance through environment interactions without relying on specific models. To enable the continuous space application for the studied energy system, fuzzy logic is involved to approximate the state-action value function of conventional Q-Learning. Moreover, the introduction of dual estimators solves its inherent overestimation problem. With python-based environment, low computation and fast convergence of the proposed FDQL are reflected in the training process. Also, adaptability to the changes in driving conditions and initial states are verified in the tests. Finally, the goal of reducing hydrogen fuel consumption and maintaining battery operation of FCHEV are both achieved.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |