Loading...

Predicting SOC Level of Electric Vehicles Battery based on Machine Algorithms in Cloud Environment

Vafaie Souraki, Roghaye | 2025

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58069 (46)
  4. University: Sharif University of Technology
  5. Department: Energy Engineering
  6. Advisor(s): Moeini Aghtaei, Moein
  7. Abstract:
  8. With the increasing adoption of electric vehicles and the growing need for improved energy management, accurate state-of-charge estimation for lithium-ion batteries has become a critical parameter in energy storage systems. Traditional methods, such as Coulomb counting and Kalman filters, lack sufficient accuracy when dealing with variable operational conditions and the nonlinear behavior of batteries. This study aims to present a comprehensive and precise approach for state-of-charge estimation. A fractional-order model was first developed to simulate the dynamic behavior of batteries, and initial state-of-charge and parameter estimation were achieved using a fractional-order unscented Kalman filter. Advanced machine learning and deep learning models, including LightGBM, LSTM, and Transformer, were subsequently employed to enhance predictive accuracy. To leverage the strengths of these methods, a mixture-of-experts architecture was implemented, utilizing operational data and realistic scenarios to achieve more accurate predictions. Results demonstrated that the proposed architecture successfully estimated the state of charge with an accuracy exceeding 98%. Furthermore, the method exhibited robust performance under complex and dynamic battery behaviors. The findings of this study not only contribute to improving battery management in electric vehicles but also hold significant potential for application in other sustainable energy storage systems
  9. Keywords:
  10. Cloud Computing ; Deep Learning ; Lithium Ion Batteries ; State of Charge ; Expert Mixture ; Charge Estimation

 Digital Object List

 Bookmark

No TOC