The accurate and rapid estimation of the state of health (SOH) for batteries and capacitors is a crucial determinant of performance prognostics and health management. However, data-driven methods often require exhaustive data curation under random SOH and state of charge (SOC) conditions, leading to increased computational demands on battery management systems (BMS). In this study, we introduce a deep neural network (DNN) model that utilizes a Gaussian probability distribution (GPD) as the loss function to probabilistically estimate the SOH of both lithium-ion capacitors (LICs) and supercapacitors (SCs). By leveraging data from a single charge–discharge cycle, our model accurately predicts the degradation trajectories of LICs and SCs under various operational conditions, including different time intervals and charging protocols. The proposed DNN achieves SOH estimations with a root-mean-squared error (RMSE) of ≤1% and degradation trajectory predictions with an RMSE of ≤7%, reflecting high accuracy and robustness. This approach highlights exploiting limited data for accurate SOH estimation and degradation trajectory prediction by deep learning models, which will enhance the development of universal battery management algorithms for next-generation energy storage devices.
Min et al. (Fri,) studied this question.