Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Interpretable Battery Aging without Extra Tests via Neural-Assisted Physics-based Modelling

Yuan Qiu, Wei Li, Wei Zhang, Yi Zhou, Fang Liu, Jianbiao Wang, Zhi Wei Seh

Abstract

State of health (SoH) is widely used for battery management, but it is a single scalar and offers limited interpretability. Two batteries with similar SoH can exhibit very different degradation behaviors and the lack of interpretability hinders optimal battery operation. In this paper, we propose IBAM for interpretable battery aging modelling with a neural-assisted physics-based framework. IBAM outputs a 2-D aging fingerprint without extra diagnostic tests and uses only routine logs from the battery management system. The fingerprint offers great interpretability by capturing a battery's curve-wide polarization voltage loss and the tail loss near the end-of-discharge. IBAM first creates a physics-based battery model based on a fractional-order equivalent circuit model, and then extracts per-cycle fingerprints from the model using a two-stage least-squares method. IBAM further anchors fingerprints on the SoH axis with physics-guided regression, where the per-cycle SoH is estimated via a bidirectional gated recurrent unit with customized multi-channel voltage features. Across batteries with short-, medium-, and long-lifespans, IBAM consistently yields the best physics model fidelity at different aging stages, and provides clear interpretations of degradation mechanisms and fingerprint patterns about batteries of different lifespans. The resulting fingerprints support interpretable battery health assessment and can inform battery control choices.

Interpretable Battery Aging without Extra Tests via Neural-Assisted Physics-based Modelling

Abstract

State of health (SoH) is widely used for battery management, but it is a single scalar and offers limited interpretability. Two batteries with similar SoH can exhibit very different degradation behaviors and the lack of interpretability hinders optimal battery operation. In this paper, we propose IBAM for interpretable battery aging modelling with a neural-assisted physics-based framework. IBAM outputs a 2-D aging fingerprint without extra diagnostic tests and uses only routine logs from the battery management system. The fingerprint offers great interpretability by capturing a battery's curve-wide polarization voltage loss and the tail loss near the end-of-discharge. IBAM first creates a physics-based battery model based on a fractional-order equivalent circuit model, and then extracts per-cycle fingerprints from the model using a two-stage least-squares method. IBAM further anchors fingerprints on the SoH axis with physics-guided regression, where the per-cycle SoH is estimated via a bidirectional gated recurrent unit with customized multi-channel voltage features. Across batteries with short-, medium-, and long-lifespans, IBAM consistently yields the best physics model fidelity at different aging stages, and provides clear interpretations of degradation mechanisms and fingerprint patterns about batteries of different lifespans. The resulting fingerprints support interpretable battery health assessment and can inform battery control choices.

Paper Structure

This paper contains 14 sections, 14 equations, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. IBAM Methodology
  3. Problem Formulation and Interpretation Targets
  4. Physics-based Model for Fingerprint Extraction
  5. Per-cycle Fingerprint Identification with Two-Stage LSM
  6. Physics-guided SoH-domain Fingerprint Mapping
  7. BiGRU-based SoH Modelling
  8. Physics-guided Mapping
  9. Experimental Study
  10. Experiment Setup
  11. Interpretable Fingerprint Insights
  12. Physics-Model Fitting Performance
  13. Impact of Multi-Channel Inputs on BiGRU SoH Modelling
  14. Conclusion

Figures (4)

  • Figure 1: An illustration of IBAM architecture.
  • Figure 2: Illustration of the polarization loss and tail loss for a representative battery cell across different discharge cycles.
  • Figure 3: Illustration of the physics-based models for ECM and FOECM.
  • Figure 4: IBAM learned degradation fingerprints for representative batteries in the short-life (Bat039 and Bat025), medium-life (Bat116), and long-life (Bat100 and Bat124) categories; each battery’s lifespan (in cycles) is indicated in the legend. Overall, short-life batteries exhibit stronger tail loss.