BatteryML:An Open-source platform for Machine Learning on Battery Degradation

TL;DR

Battery degradation modeling is hindered by data heterogeneity and fragmented tooling. The authors introduce BatteryML, an open-source, end-to-end platform that unifies data representation (BatteryData), feature engineering, automatic labeling, and a spectrum of models for $RUL$, $SOH$, and $SOC$ tasks, with a pipeline that supports transfer learning across diverse datasets. Through benchmarking on public datasets (CALCE, MATR, HUST, HNEI, RWTH, SNL, UL_PUR) and composite sets (CRUH, CRUSH, MIX), the work shows varying model strengths across contexts, underscoring the value of standardized data and modular modeling. BatteryML is positioned to accelerate battery research by enabling cross-disciplinary collaboration and reproducible experimentation, with future directions including one-click predictions and broader data integration from BMS and material properties.

Abstract

Battery degradation remains a pivotal concern in the energy storage domain, with machine learning emerging as a potent tool to drive forward insights and solutions. However, this intersection of electrochemical science and machine learning poses complex challenges. Machine learning experts often grapple with the intricacies of battery science, while battery researchers face hurdles in adapting intricate models tailored to specific datasets. Beyond this, a cohesive standard for battery degradation modeling, inclusive of data formats and evaluative benchmarks, is conspicuously absent. Recognizing these impediments, we present BatteryML - a one-step, all-encompass, and open-source platform designed to unify data preprocessing, feature extraction, and the implementation of both traditional and state-of-the-art models. This streamlined approach promises to enhance the practicality and efficiency of research applications. BatteryML seeks to fill this void, fostering an environment where experts from diverse specializations can collaboratively contribute, thus elevating the collective understanding and advancement of battery research.The code for our project is publicly available on GitHub at https://github.com/microsoft/BatteryML.

Figures (7)

• Figure 1: An overview of BatteryML.
• Figure 2: MATR1 degradation curve showcasing the train and test dataset discharge capacity
• Figure 3: Voltage curves of a battery in MATR1. Each curve corresponds to a cycle.
• Figure 4: A showcase of the correlation between actual and predicted values in the RUL task.
• Figure 5: Feature space ablations. The "variance", "discharge" and "full" features are designed by domain experts to capture the degradation pattern of LFP/graphite cells. The "Variance" feature refers to the Log Variance of $\Delta Q_{100-10}(V)$ during the discharge process.The "Discharge" feature encompasses multiple features extracted from the discharge process.The "Full" feature includes features extracted from both the charging and discharging processes. QdLinear feature is obtained by linear interpolation of discharge capacity with respect to voltage.