Abstract
Public battery aging datasets are a critical asset for advanced health management, but their practical use is often limited by inconsistent formats, unclear schemas, and metadata scattered across repositories and publications. Current curation remains largely manual and hard to reproduce, while general-purpose data integration tools miss the domain-specific semantics of electrochemical time-series data.
We present BatteryLake, a governed data lakehouse that turns raw public battery data into benchmark-ready assets through an agentic, physics-grounded curation framework, with three contributions:
-
Metadata Extraction and Dataset-Specific Converters: LLM agents extract metadata and synthesize dataset-specific converters, grounding every output in verbatim evidence and abstaining when none supports a value.
-
Human-in-the-Loop Verification Mechanism: Framing verification as selective prediction, gating admitted data through 26 schema, statistical, and physical-plausibility rules.
-
Open Benchmark Release: We release an open benchmark of 41 datasets from over 25 institutions, with standardized State of Health (SOH) and Remaining Useful Life (RUL) tasks, three split protocols, and eight baseline model families.
The platform, benchmark, and curation protocol are publicly available at BatteryLake Website.
Blogger's Review: The innovation of the BatteryLake project lies in its integration of physical principles with intelligent tools, addressing multiple pain points in battery data management. It provides robust support for battery health management, particularly in standardizing benchmark testing, showcasing broad application potential. The platform's openness also offers rich resources for researchers.