Xu, Fuguo (Chiba University), Kuboyama, Tatsuya (Chiba University), Moriyoshi, Yasuo (Chiba University, Graduate School of Engineering)

Reinforcement Learning-Based Energy Efficiency Optimization of Hybrid Electric Vehicles Using Historical Traffic Data

Scheduled for presentation during the Regular session "Energy Management" (FrB2), Friday, November 1, 2024, 11:50−12:10, Room T2

7th IFAC Conference on Engine and Powertrain Control, Simulation and Modeling, Oct 30 - Nov 1, 2024, Dalian, China

This information is tentative and subject to change. Compiled on January 2, 2025

Abstract

In this paper, a reinforcement learning-based energy management strategy for power-split hybrid electric vehicle is proposed to improve the energy efficiency. Firstly, the modeling of HEV power-split powertrain is built. Then, an infinite horizon optimization problem is formulated to minimize the fuel consumption and to maintain the real time SoC within a pre-defined SoC. Then, the driving torque and vehicle speed is seen as the disturbance, and the probability transition is determined by the historical vehicle data collected on a fixed real-world traffic route. After that, the optimal solution is derived by the value iteration approach. Finally, the verification of the proposed algorithm is conducted in a high-performance simulator.