Deep learning enabled state of charge, state of health and remaining useful life estimation for smart battery management system: Methods, implementations, issues and prospects

State of Charge (SOC), state of health (SOH), and remaining useful life (RUL) are the crucial indexes used in the assessment of electric vehicle (EV) battery management systems (BMS). The performance and efficiency of EVs are subject to the precise estimation of SOC, SOH, and RUL in BMS which enhances the battery reliability, safety, and longevity. However, the estimation of SOC, SOH, and RUL is challenging due to the battery capacity degradation and varying environmental conditions. Recently, deep learning (DL) has received wide attention for battery SOC, SOH, and RUL estimation due to the accessibility of a vast amount of data, large storage volume, and powerful computing processors. Nevertheless, the application of DL in SOC, SOH, and RUL estimation for EVs is still limited. Therefore, the novelty of this paper is to deliver a comprehensive review of DL-enabled SOC, SOH, and RUL estimation for BMS, focusing on methods, implementations, strengths, weaknesses, issues, accuracy, and contributions. Moreover, this study explores the numerous important implementation factors of DL methods concerning data type, features, size, preprocessing, algorithm operation, functions, hyperparameter adjustments, and performance evaluation. Additionally, the review explores various limitations and challenges of DL in BMS related to battery, algorithm, and operational issues. Finally, future opportunities and prospects are delivered that would support the EV engineers and automotive industries to establish an accurate and robust DLbased SOC, SOH, and RUL estimation technique towards smart BMS in future sustainable EV applications.

Automated summary

This paper provides a comprehensive review of deep learning-enabled estimation of state of charge (SOC), state of health (SOH), and remaining useful life (RUL) in electric vehicle battery management systems. It explores methods, implementations, strengths, weaknesses, issues, accuracy, and contributions, as well as important implementation factors and limitations of deep learning in battery management systems. The paper also discusses future opportunities and prospects for accurate and robust deep learning-based techniques in sustainable EV applications.