Suggested Citation:
Patnaik, Ayush, Jackson Fogelquist, Adam Zufall, Stephen K. Robinson, Xinfan Lin (2025)

Machine Learning Detection of Lithium Plating in Lithium-ion Cells: AGaussian Process Approach

. Institute of Transportation Studies, University of California, Davis, Journal Article UCD-ITS-RP-25-94

Abstract:

 Lithium plating during fast charging is a critical degradation mechanism that accelerates capacity fade and can trigger catastrophic safety failures. Recent work has identified a distinctive dQ/dV peak above 4.0 V as a reliable signature of plating onset; however, conventional methods for computing dQ/dV rely on finite differencing with filtering, which amplifies sensor noise and introduces bias in peak location. In this paper, we propose a Gaussian Process (GP) framework for lithium plating detection by directly modeling the charge–voltage relationship Q(V) as a stochastic process with calibrated uncertainty. Leveraging the property that derivatives of GPs remain GPs, we infer dQ/dV analytically and probabilistically from the posterior, enabling robust detection without ad hoc smoothing. The framework provides three key benefits: (i) noise-aware inference with hyperparameters learned from data, (ii) closedform derivatives with credible intervals for uncertainty quantification, and (iii) scalability to online variants suitable for
embedded BMS. Experimental validation on Li-ion coin cells across a range of C-rates (0.2C–1C) and temperatures (0–40°C) demonstrates that the GP-based method reliably detects plating peaks under low-temperature, high-rate charging, while correctly reporting no peaks in baseline cases. The concurrence of GP-identified differential peaks, reduced charge throughput, and capacity fade measured via reference performance tests confirms the method’s accuracy and robustness, establishing a practical pathway for real-time lithium plating detection.