Battery Diagnosis: A Lifelong Learning Framework for Electric Vehicles

Expending manufacturing capacity and development of high-energy batteries greatly stimulate the growth and applications of electric vehicles (EVs). However, battery diagnostics and prognostics related to capacity degradation (referred as state of health, SOH) and safety issues (referred as state of safety, SOS) in real-world applications is still a big deal. Due to the uncertainties in materials and manufacturing, dynamic operation conditions as well as a lack of plentiful, high-quality on-road data, accurate diagnosis of battery performance for 'real EVs' is very challenging. Considering the difficulty in accurately predicting battery behaviors in real-world applications, brand-new control area networks (CAN) and cloud-based solution could have considerable benefits. An AI-powered cloud-based framework integrating longitudinal electronic health records with real-world data enables continuous battery performance evaluation for EVs. This offers opportunities for combining data generation with data-driven approaches to predict the behavior of complex, time-varying electrochemical systems. It is hoped that this paper will be of reference value to the EV and battery industries for ameliorating some of the hurdles for battery diagnostics and prognostics under realistic EV conditions.