Analysis and key findings from real-world electric vehicle field data

Deploying battery state of health (SoH) estimation and forecasting algorithms are critical for ensuring the reliable performance of battery electric vehicles (EVs). SoH algorithms are designed and trained from data collected in the laboratory upon cycling cells under predefined loads and temperatures. Field battery pack data collected over 1 year of vehicle operation are used to define and extract performance/health indicators and correlate them to real driving characteristics (charging habits, acceleration, and braking) and season-dependent ambient temperature. Performance indicators (PIs) during driving and charging events are defined upon establishing a data pipeline to extract key battery management system (BMS) signals. This work shows the misalignment existing between laboratory testing and actual battery usage, and the opportunity that exists in enhancing battery experimental testing to deconvolute time and temperature to improve SoH estimation strategies.