EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Establishment of a Neural Network-Based Prediction Model for Wheel–Sand Dynamics

Zhang Ni, Weihong Wang, Chenyu Hu, Zhi Li, Bo Li

World Electric Vehicle Journal · 01.04.2026 · 17 (4) · 186

DOI: 10.3390/wevj17040186 ↗

Kurz gefasst

Fachartikel in der Zeitschrift World Electric Vehicle Journal, online erschienen am 01.04.2026. Verfasst von Zhang Ni und 4 weiteren Autor:innen. Die Arbeit nennt 21 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Open Fund Project of State Key Laboratory of Intelligent Green Vehicle and Mobility, Open Fund Project of the National Key Laboratory of Automotive Chassis Integration and Bionics, Hubei Provincial Natural Science Foundation Innovation and Development Joint Fund Project in China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

With the expansion of electric vehicle (EV) applications into unstructured sandy terrains such as deserts, accurately characterizing tire–sand dynamic interactions is essential for enhancing off-road performance. However, traditional terramechanics models, the discrete element method (DEM), and purely data-driven neural networks all have inherent limitations, failing to balance physical interpretability and computational efficiency. This study proposes a wheel–sand dynamics prediction model that integrates DEM simulation, semi-physical modeling, and deep learning. A DEM tire–sand contact platform is built to acquire longitudinal slip and cornering properties, and a dimensionless semi-physical tire model is derived using frictional constitutive relations and tire theory. A 3-DOF vehicle dynamics model is then established to generate high-fidelity physics-based datasets, and a residual neural network is adopted to avoid performance degradation in deep networks. The model is validated and optimized via real-vehicle sandy terrain tests, with its performance compared against other network structures. The proposed model achieves high prediction accuracy, with engineering-acceptable errors, and outperforms conventional neural networks. The dimensionless framework improves generality, overcoming the weaknesses of traditional and purely data-driven models. This work provides theoretical and statistical support for EV traction control design and tire structure optimization, promoting driving stability and terrain passability in unstructured sandy environments.

Förderung

  • Open Fund Project of State Key Laboratory of Intelligent Green Vehicle and Mobility
  • Open Fund Project of the National Key Laboratory of Automotive Chassis Integration and Bionics
  • Hubei Provincial Natural Science Foundation Innovation and Development Joint Fund Project in China
  • Graduate Practice and Innovation Program Project of Jiangsu University of Technology

Zitieren

APA

Ni, Z., Wang, W., Hu, C., Li, Z., & Li, B. (2026). Establishment of a Neural Network-Based Prediction Model for Wheel–Sand Dynamics. World Electric Vehicle Journal, 17(4), 186. https://doi.org/10.3390/wevj17040186

Vancouver

Ni Z, Wang W, Hu C, Li Z, Li B. Establishment of a Neural Network-Based Prediction Model for Wheel–Sand Dynamics. World Electric Vehicle Journal. 2026;17(4):186. doi:10.3390/wevj17040186

BibTeX
@article{Ni2026,
  author = {Ni, Zhang and Wang, Weihong and Hu, Chenyu and Li, Zhi and Li, Bo},
  title = {Establishment of a Neural Network-Based Prediction Model for Wheel–Sand Dynamics},
  journal = {World Electric Vehicle Journal},
  year = {2026},
  volume = {17},
  number = {4},
  pages = {186},
  doi = {10.3390/wevj17040186}
}
Kostenlos starten →