Abstract:A key challenge for Lithium-ion batteries (LiB) is the capability to manage battery performance and predict the battery's lifetime, which are crucial information for optimum operation of electric-vehicles (EVs). In that regard, early detection of battery-aging phenomena and their implications on performance is necessary for maintaining warranty and avoiding safety-related liabilities. However, rapid diagnosis of the root causes of battery degradation has not yet been well-developed nor studied upon a broad mixture of cell designs and use cases. Here, a machine-learning (ML)- based framework is presented that aims at identifying aging modes or mechanisms during battery cycling. Cycle-by-cycle electrochemical signatures (e.g., end-of-rest voltages, columbic efficiency, and capacity fade) were analyzed then correlated with aging modes, such as loss of active materials in the cathode (LAM PE ) or loss of Li inventory (LLI). Multiple battery designs and use conditions were analyzed to develop this ML framework, including 44 single layer pouch cells representing two cathode chemistries (NMC532 and NMC811), two electrode loadings (low and moderate), and five charging rates (1, 4, 4.5, 6, and 9C CC-CV profiles). Predominant aging behaviors depend on both the cell design and use. Aging behaviors can be classified into a combination of LAM PE and LLI. A classification accuracy of 90% was achieved using only the first 200aging cycles, and the percentage of LAM PE was predicted with 4.3% of error after 600 cycles. As seen here, the overall aging behavior of the battery was found to be dependent on cathode chemistry, electrode build, and usage conditions. This ML framework is promising for battery developers to save the time and effort that is typically associated with prolonged testing. The ML framework will also advise on innovative design strategies to counter specific degradation modes, thus significantly accelerating the development cycle for building longer lasting batteries. From the end user's perspective, this aging-mode diagnosis framework enables the ability to fine-tune battery service scenarios to minimize degradations thereby improving both lifetime and safety in EVs or stationary storage systems. Figure 1

Degradation Modes Identification of Lithium-Ion Batteries Based on Flexible Discharge Data

A Physics-Informed Hybrid Data-Driven Approach with Generative Electrode-Level Features for Lithium-Ion Battery Health Prognostics

A Physics-Informed Hybrid Multitask Learning for Lithium-Ion Battery Full-Life Aging Estimation at Early Lifetime

Reliability Study of Battery Lives: A Functional Degradation Analysis Approach

Diagnosing and Decoupling the Degradation Mechanisms in Lithium Ion Cells: An Estimation Approach

A data-driven approach for diagnosing degradation in lithium-ion batteries using data transformation techniques and a novel deep neural network

A facile physics-based model for non-destructive diagnosis of battery degradation

Lithium-ion battery degradation: using degradation mode analysis to validate lifetime prediction modelling

Invariant learning based multi-stage identification for Lithium-ion battery performance degradation

Degradation mode estimation using reconstructed open circuit voltage curves from multi-year home storage field data

Battery degradation diagnosis with field data, impedance-based modeling and artificial intelligence

Degradation Diagnosis of Lithium-Ion Batteries Considering Internal Gas Evolution

Untangling Degradation Chemistries of Lithium‐Sulfur Batteries Through Interpretable Hybrid Machine Learning

Data-driven State-of-health Estimation for Lithium-Ion Battery Based on Aging Features

Non-destructive Degradation Pattern Decoupling for Ultra-early Battery Prototype Verification Using Physics-informed Machine Learning

Analysis of Performance Degradation in Lithium-Ion Batteries Based on a Lumped Particle Diffusion Model

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

Lifetime prediction and maintenance assessment of Lithium-ion batteries based on combined information of discharge voltage curves and capacity fade

Data-Driven Battery Aging Mechanism Analysis and Degradation Pathway Prediction

Non-invasive Characteristic Curve Analysis of Lithium-ion Batteries Enabling Degradation Analysis and Data-Driven Model Construction: A Review

A generalizable, data-driven online approach to forecast capacity degradation trajectory of lithium batteries