Data‐Driven State of Health for Lithium‐Ion Batteries: Feature Engineering, Estimation Approaches, and Future Directions

Amidst the prevailing trends of electrification, intelligence, and connectivity, the convergence of new energy vehicles and big data herald transformative opportunities for China's automotive industry. In this context, the estimation of the state of health (SOH) of lithium‐ion batteries assumes critical importance for ensuring the safe and efficient operation of electric vehicles. This paper presents a comprehensive exposition of data‐driven methodologies for SOH estimation of lithium‐ion batteries. It begins with an overview of commonly accepted definitions and the key factors influencing battery SOH, followed by an in‐depth exploration of feature engineering—including the processes of feature extraction and selection. The discussion then delves into data‐driven approaches to battery SOH management, encompassing both nonprobabilistic and probabilistic models. While numerous methods exist for estimating SOH, each possesses distinct strengths and limitations. Looking ahead, data‐driven SOH estimation is poised to evolve toward the integration of multisource data fusion, enhancement through small‐sample and transfer learning techniques, incorporation with physical modeling, and expansion across domains. These advancements are anticipated to significantly enhance the precision and dependability of SOH estimation and catalyze the broader deployment of battery technologies across various sectors.