An analytical examination of the performance assessment of CNN-LSTM architectures for state-of-health evaluation of lithium-ion batteries

Abstract

Precisely determining the State of Health of Lithium-Ion Batteries is essential for maintaining their dependability, security, and durability in diverse applications such as electric vehicles and energy storage systems. This study examines recent advancements in health estimation techniques with a focus on deep learning methodologies. A bibliometric analysis was conducted to identify the current trends in the state of health estimation. This research specifically examines the potential of the convolutional neural network–long short-term memory algorithm to improve the precision of State of Health forecasts for the battery model. A one-dimensional COMSOL battery model was created and examined to facilitate a thorough assessment of battery characteristics. Further validation was achieved by determining the battery parameters through a cost-effective hardware prototype. Various learning models were employed to analyse the data obtained from this hardware, with the support vector machine-based model delivering the most accurate results, showing an error rate of .48 percent. In this study, an assessment was conducted on various Convolutional Neural Network-Long Short-Term Memory architectures to explore their complete potential, with the generic architecture yielding the most favourable outcomes. Finally, this work explores the prospects and direction of the State of Health estimation, offering perspectives on enhancing deep learning frameworks for practical applications. The results of this study contribute to the advancement of more dependable and effective battery health monitoring systems, advancing the domain of battery management and predictive upkeep

Topics & Concepts