Managing the Quantity of Battery Data Required to Deliver High Accuracy Ageing Models and Support the Battery Passport

Today’s world is marked by a significant increase in the amount of data generated, and Electric Vehicles batteries are no exception. Although the importance of analysing and storing battery data is clear, methods to characterize dynamic profiles, distinctive of real-driving, are not standardized. This study evaluates different ways of characterizing Electric Vehicle battery data, balancing data volume and battery health assessment accuracy. Different features are extracted and used to build degradation models. Voltage and State of Charge show the most relevance for degradation prediction and boxplot-based characterization yields the best model performance. Data aggregation, meaning storing only relevant features per driving cycle, is proposed as the best method to reduce data quantity and generate accurate results, especiallywhen employing histogram-based models. The results of this study can be used to guide feature selection for degradation models, reduce the dimensionality of laboratory datasets and as information for the Battery Passport.