Volume 03: Proceedings of 11th International Conference on Applied Energy, Part 2, Sweden, 2019

A Deep Neuroevolution Based Energy Management Strategy for Plug-in Hybrid Electric Vehilce Yuankai Wu, Huachun Tan, Jiankun Peng, Yuecheng Li, Hongwen He

Abstract

Energy management strategy is important for improving fuel economic of hybrid electric vehicles. We present a deep neuroevolution based energy management strategy for hybrid electric vehicles, which learns optimal energy split strategies through evolution of its deep neural networks structure. We define the optimization objective of the deep neural networks by the fuel consumption and properties of target HEV. The deep neural networks controller is learnt through a parallel and evolution way. The simulation results on a standard driving cycles show that the proposed deep neuroevolution method outperforms the DRL based model, and achieves comparative performance to global–‐optimal method–‐dynamic programming.

Keywords deep neuroevolution, plug--‐in hybrid electric vehicle, energy management strategies

Copyright ©
Energy Proceedings