Energy Management Strategy for Cross-Domain Vehicles Based on Multimodal Sensing under Large Disturbance Conditions

Shengzhao Pang, Siyu Zhao, Xiaoran Ren, Kaiyin Song, Yingxue Chen, Zhaoyong Mao

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

The Cross-Domain Vehicle (CDV) refers to a vehicle that can navigate underwater, on the water surface, and in the air. Also known as Unmanned Aerial-Underwater Vehicle (UAUV) by many researchers. It can enter and leave the water many times during the working process. This paper focuses on the intricate energy management issue of CDVs amidst large disturbances. Multimodal sensing technology is employed to collect comprehensive data from a variety of sensors, including motion, environmental, and energy-related sensors. A Improved Deep Q-Network algorithm is then utilized to optimize the energy distribution strategy. And attempt to address the lack of interpretability in deep learning models when dealing with practical problems. Simulation results demonstrate that this approach can effectively improve the energy utilization rate by 6.6% and reduce the SOC fluctuation of lithium battery 2.6%, thus offering an innovative solution for the energy management of cross-media vehicles.

Original languageEnglish
Title of host publication2025 IEEE Industry Applications Society Annual Meeting, IAS 2025
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781665457767
DOIs
StatePublished - 2025
Event2025 IEEE Industry Applications Society Annual Meeting, IAS 2025 - Taipei, Taiwan, Province of China
Duration: 15 Jun 202520 Jun 2025

Publication series

NameConference Record - IAS Annual Meeting (IEEE Industry Applications Society)
ISSN (Print)0197-2618

Conference

Conference2025 IEEE Industry Applications Society Annual Meeting, IAS 2025
Country/TerritoryTaiwan, Province of China
CityTaipei
Period15/06/2520/06/25

Keywords

  • Cross-domain vehicle
  • deep learning
  • energy management
  • hybrid electric vehicles
  • reinforcement learning
  • unmanned aerial-underwater vehicle

Fingerprint

Dive into the research topics of 'Energy Management Strategy for Cross-Domain Vehicles Based on Multimodal Sensing under Large Disturbance Conditions'. Together they form a unique fingerprint.

Cite this