Early-Cycle Internal Impedance Enables ML-Based Battery Cycle Life Predictions Across Manufacturers

Tyler Sours,Shivang Agarwal,Marc Cormier,Jordan Crivelli-Decker,Steffen Ridderbusch,Stephen L. Glazier,Connor P. Aiken,Aayush R. Singh,Ang Xiao,Omar Allam

2024-10-06

Abstract:Predicting the end-of-life (EOL) of lithium-ion batteries across different manufacturers presents significant challenges due to variations in electrode materials, manufacturing processes, cell formats, and a lack of generally available data. Methods that construct features solely on voltage-capacity profile data typically fail to generalize across cell chemistries. This study introduces a methodology that combines traditional voltage-capacity features with Direct Current Internal Resistance (DCIR) measurements, enabling more accurate and generalizable EOL predictions. The use of early-cycle DCIR data captures critical degradation mechanisms related to internal resistance growth, enhancing model robustness. Models are shown to successfully predict the number of cycles to EOL for unseen manufacturers of varied electrode composition with a mean absolute error (MAE) of 150 cycles. This cross-manufacturer generalizability reduces the need for extensive new data collection and retraining, enabling manufacturers to optimize new battery designs using existing datasets. Additionally, a novel DCIR-compatible dataset is released as part of ongoing efforts to enrich the growing ecosystem of cycling data and accelerate battery materials development.

Machine Learning,Materials Science

What problem does this paper attempt to address?

### What problem does this paper attempt to solve? This paper aims to address the challenges in predicting the end-of-life (EOL) of lithium-ion batteries across different manufacturers. Specifically, due to differences in electrode materials, manufacturing processes, battery formats, and the lack of universal data, existing methods based on capacity curve features often fail to generalize across manufacturers and chemistries. These issues lead to a significant decline in the predictive capability of models when applied to different chemistries or operating conditions. To overcome these challenges, the study introduces a new approach that combines traditional voltage-capacity features with direct current internal resistance (DCIR) measurements to improve the accuracy and generalization ability of the models. By using DCIR data from early cycles, this method can capture key degradation mechanisms related to internal resistance growth, thereby enhancing the robustness of the models. The study shows that this combined approach of DCIR and traditional features can successfully predict the lifespan of batteries from different manufacturers and chemistries, with a mean absolute error (MAE) of only 150 cycles. This not only reduces the need for new data collection and retraining but also allows manufacturers to leverage existing datasets to optimize new battery designs. Additionally, the research team has released a new DCIR-compatible dataset to enrich the growing ecosystem of cycling data, accelerating the development of battery materials.

Early-Cycle Internal Impedance Enables ML-Based Battery Cycle Life Predictions Across Manufacturers

Data-driven prediction of battery cycle life before capacity degradation

Data driven analysis of lithium-ion battery internal resistance towards reliable state of health prediction

Prediction of Battery Cycle Life Using Early-Cycle Data, Machine Learning and Data Management

Predicting Battery Lifetime Under Varying Usage Conditions from Early Aging Data

A Battery Aging Mode Identification Framework: Encompassing Multiple Cell Chemistries, Electrode Designs, and Use Conditions

Predicting the Cycle Life of Lithium-Ion Batteries Using Data-Driven Machine Learning Based on Discharge Voltage Curves

A Hybrid Machine Learning Model for Battery Cycle Life Prediction with Early Cycle Data

A Data-Driven Approach for Lithium-ion Battery Lifetime Classification Based on Early Cycles

Performance Classification and Remaining Useful Life Prediction of Lithium Batteries Using Machine Learning and Early Cycle Electrochemical Impedance Spectroscopy Measurements

A Data-Driven Modeling Framework for Lithium Metal Battery Health Estimation

Data Driven Prediction of Battery Cycle Life Before Capacity Degradation

Optimizing Cycle Life Prediction of Lithium-ion Batteries via a Physics-Informed Model

Cycle Life Prediction for Lithium-ion Batteries: Machine Learning and More

Feature engineering for machine learning enabled early prediction of battery lifetime

An open access tool for exploring machine learning model choice for battery life cycle prediction

Data-Driven Cycle Life Prediction of Lithium Metal-Based Rechargeable Battery Based on Discharge/Charge Capacity and Relaxation Features

Early prediction of Li-ion cell failure from EIS derived from current-voltage time series

Early prediction of cycle life for lithium-ion batteries based on evolutionary computation and machine learning

Integrating Physics-Based Modeling and Machine Learning for Degradation Diagnostics of Lithium-Ion Batteries

Deep Learning to Predict Battery Voltage Behavior after Uncertain Cycling-Induced Degradation