Adaptive Parameter Identification of Battery Pack in Electric Vehicles with Real-Driving Signals

EasyChair Preprint 15882

Abstract

This paper presents an adaptive identification method for battery parameters in automotive applications. A simple yet accurate electrical equivalent model (ECM) with varying parameters is used to represent the whole battery pack. The modeling process requires the current, voltage, and SOC signals of the battery. Detailed physical knowledge of the battery pack and inside cells are not necessary. The ECM parameter identification approach is developed by employing the NLMS (normalized least mean square) algorithm, which is an advanced adaptive algorithm having fast convergence rate and easier to be implemented. This approach is verified on a 51.2 V, 95AH LiFePO4 battery pack operated in three-wheeler electric bikes. Battery signals during vehicle daily real-world driving were collected over a period of time and used for the ECM parameter identification. The identified internal resistance R0, R1 and capacitance C1 changes obviously over the period of time and the battery degradation is well reflected through the identified parameters of the ECM.

Keyphrases: EV batteries, adaptive parameter identification, battery capacitance, battery degradation, battery internal resistance, battery modeling

@booklet{EasyChair:15882,
  author    = {Chunling Du and Tomi Wijaya and Choon Lim Ho},
  title     = {Adaptive Parameter Identification of Battery Pack in Electric Vehicles with Real-Driving Signals},
  howpublished = {EasyChair Preprint 15882},
  year      = {EasyChair, 2025}}