Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Cylindrical Battery Fault Detection under Extreme Fast Charging: A Physics-based Learning Approach

Roya Firoozi, Sara Sattarzadeh, Satadru Dey

TL;DR

This work constructs the detection observers based on an experimentally identified electrochemical-thermal model, and subsequently design the observer tuning parameters following Lyapunov’s stability theory, and utilizes Gaussian Process Regression technique to learn the model and measurement uncertainties which in turn aid the Detection observers in distinguishing faults and uncertainties.

Abstract

High power operation in extreme fast charging significantly increases the risk of internal faults in Electric Vehicle batteries which can lead to accelerated battery failure. Early detection of these faults is crucial for battery safety and widespread deployment of fast charging. In this setting, we propose a real-time {detection} framework for battery voltage and thermal faults. A major challenge in battery fault detection arises from the effect of uncertainties originating from sensor inaccuracies, nominal aging, or unmodelled dynamics. Inspired by physics-based learning, we explore a detection paradigm that combines physics-based models, model-based detection observers, and data-driven learning techniques to address this challenge. Specifically, we construct the {detection} observers based on an experimentally identified electrochemical-thermal model, and subsequently design the observer tuning parameters following Lyapunov's stability theory. Furthermore, we utilize Gaussian Process Regression technique to learn the model and measurement uncertainties which in turn aid the {detection} observers in distinguishing faults and uncertainties. Such uncertainty learning essentially helps suppressing their effects, potentially enabling early detection of faults. We perform simulation and experimental case studies on the proposed fault {detection} scheme verifying the potential of physics-based learning in early detection of battery faults.

Cylindrical Battery Fault Detection under Extreme Fast Charging: A Physics-based Learning Approach

TL;DR

This work constructs the detection observers based on an experimentally identified electrochemical-thermal model, and subsequently design the observer tuning parameters following Lyapunov’s stability theory, and utilizes Gaussian Process Regression technique to learn the model and measurement uncertainties which in turn aid the Detection observers in distinguishing faults and uncertainties.

Abstract

High power operation in extreme fast charging significantly increases the risk of internal faults in Electric Vehicle batteries which can lead to accelerated battery failure. Early detection of these faults is crucial for battery safety and widespread deployment of fast charging. In this setting, we propose a real-time {detection} framework for battery voltage and thermal faults. A major challenge in battery fault detection arises from the effect of uncertainties originating from sensor inaccuracies, nominal aging, or unmodelled dynamics. Inspired by physics-based learning, we explore a detection paradigm that combines physics-based models, model-based detection observers, and data-driven learning techniques to address this challenge. Specifically, we construct the {detection} observers based on an experimentally identified electrochemical-thermal model, and subsequently design the observer tuning parameters following Lyapunov's stability theory. Furthermore, we utilize Gaussian Process Regression technique to learn the model and measurement uncertainties which in turn aid the {detection} observers in distinguishing faults and uncertainties. Such uncertainty learning essentially helps suppressing their effects, potentially enabling early detection of faults. We perform simulation and experimental case studies on the proposed fault {detection} scheme verifying the potential of physics-based learning in early detection of battery faults.

Paper Structure

This paper contains 15 sections, 1 theorem, 19 equations, 7 figures, 2 algorithms.

Table of Contents

  1. Introduction
  2. Battery Electrochemical-Thermal Model
  3. Failure Modes
  4. State-Space Model with Faults and Uncertainties
  5. Fault Detection Framework
  6. Gaussian Process Regression based Learning
  7. Design of Detection Observers
  8. Design of Decision Maker
  9. Results and Discussion
  10. Voltage Fault Detection
  11. Thermal Fault Detection
  12. Minimum Detectable Thermal Fault
  13. Performance Comparison of Model-only and Model-and-Learning Approaches
  14. Performance under Uncertainties
  15. Conclusion

Key Result

Proposition 1

Considering the estimation error dynamics ss-echem-e-ss-therm-e-2, the following are true: if there exist symmetric positive definite matrices $P_1$ and $P_2$ such that the following conditions are satisfied: where $\Gamma$ and $\vartheta$ are arbitrary positive numbers, $x_1 = \left\|(A_1-L_VC_1)^TP_1\right\|$, $x_2 = \left\|(A_2-L_TC_2)^TP_2\right\|$, $\underline{\lambda}_Q$ and $\underline{\l

Figures (7)

  • Figure 1: Fault detection scheme.
  • Figure 2: Comparison of model output and experimental data under constant current constant voltage (CCCV) charging scenario. For model only case, the RMS errors are $27.4$$mV$ and $0.3$$^oC$ whereas model along with learning have RMS errors of $0.56$$mV$ and $0.012$$^oC$.
  • Figure 3: Residual responses under voltage faults.
  • Figure 4: Residual response under thermal fault.
  • Figure 5: Thermal fault detection performance.
  • ...and 2 more figures

Theorems & Definitions (4)

  • Remark 1
  • Remark 2
  • Proposition 1
  • proof