The battery information format (BIF): A memory-efficient standard for communicating cell usage and status for battery digital twins (BDTs)

The massive expansion in lithium-ion battery (LIB) powered applications demands attention to reliability and increased lifetime, for a truly sustainable value chain. Data-driven lifetime modelling offers a route to ensure optimal usage patterns, but the lack of widely accessible LIB fleet data for training hampers benchmarking and optimization. Here we approach the lack of available training data by outlining an implementation of a battery information format (BIF). We suggest a feature-first approach, simultaneously respecting memory constraints of embedded systems and heralding a mechanistic understanding of LIB ageing. First, we examine existing feature designs. Second, we define qualitative requirements for collected battery data. From this combined, we arrive at a BIF where trade-offs can be matched to the constraints of the application. The BIF follows a histogram-based strategy, which allows to share battery state-of-health (SOH) and usage history in an interpretable, accessible, and reusable way. This BIF strategy furthermore allows training of data-driven models for predicting SOH and a vast reduction of the data footprint – by 2-4 orders of magnitude as compared to time-series data collection schemes. • Limited datasets curb quantitative design of field data for ageing prognostics. • A battery information format can be used to mitigate the data shortage. • A qualitative selection framework should form the basis for feature selection. • Histograms are found to fulfill selection criteria. • Harmonised format design, e.g. binning schemes, ensures data transferrability