• Energy Research

Abstract We propose a multiple parameter identification method for vanadium redox flow batteries (VRFBs) to estimate the model parameter in a VRFB model. The proposed method consists of an evaluation of identifiability based on the Fisher Information Matrix (FIM) to determine the best subset of model parameters to be identified, a numerical modeling of semi-two-dimensional steady-state VRFB model, and a genetic algorithm to estimate optimal model parameters. In the optimization, we introduce a fitness function involving the mean square errors of the voltage between available experimental data and results of the VRFB model. We validate the proposed method by calculating confidence intervals of identifying parameters in the subset based on the FIM from the state of charge-voltage data obtained from a small VRFB cell experiment; we compare the curves of the identified-parameter model with those obtained experimentally. Further, we demonstrate the robustness of the proposed method through its application to a kW-scale VRFB stack utilizing advanced mixed electrolytes. The capacity-voltage curves predicted by the identified-parameter model show good agreement with those obtained experimentally under various operating conditions, with mean relative errors of less than 1.9%.

Abstract Accurate forecasting of state-of-health and remaining useful life of Li-ion batteries ensure their safe and reliable operation. Most previous data-driven prediction methods assume the same distributions between training and testing batteries. Because of different operating conditions and electrochemical properties of batteries, however, distribution discrepancy exists in real-world applications. To address this issue, we present a deep-learning-based health forecasting method for Li-ion batteries, including transfer learning to predict states of different types of batteries. The proposed method simultaneously predicts the end of life of batteries and forecasts degradation patterns with predictive uncertainty estimation using variational inference. Three types of batteries are used to evaluate the proposed model; one for source and the others for target datasets. Simulation results reveal that the proposed model reduces efforts required to collect data cycles of new battery types. Further, we demonstrate the generality and robustness of the proposed method in accurately forecasting the state-of-health of Li-ion batteries without past information, which applies to cases involving used batteries.

AbstractParameter identification (PI) is a cost‐effective approach for estimating the parameters of an electrochemical model for lithium‐ion batteries (LIBs). However, it requires identifiability analysis (IA) of model parameters because identifiable parameters vary with reference data and electrochemical models. Therefore, we propose a PI and IA (PIIA) framework for a robust PI that can adapt to discharge data. The IA results show that the best subset with 15 parameters is determined by the Fisher information matrix and the sample‐averaged RDE criterion under various operating conditions. The identification process based on a genetic algorithm determines the optimal parameters. The identified‐parameter model predicts voltage curves with uncertainty bounds, considering the confidence intervals of identified parameters. Further, we demonstrate that the proposed PIIA framework robustly identifies the parameters of the electrochemical model from experimental data.

This paper proposes a fully unsupervised methodology for the reliable extraction of latent variables representing the characteristics of lithium-ion batteries (LIBs) from electrochemical impedance spectroscopy (EIS) data using information maximizing generative adversarial networks. Meaningful representations can be obtained from EIS data even when measured with direct current and without relaxation, which are difficult to express when using circuit models. The extracted latent variables were investigated as capacity degradation progressed and were used to estimate the discharge capacity of the batteries by employing Gaussian process regression. The proposed method was validated under various conditions of EIS data during charging and discharging. The results indicate that the proposed model provides more robust capacity estimations than the direct capacity estimations obtained from EIS. We demonstrate that the latent variables extracted from the EIS data measured with direct current and without relaxation reliably represent the degradation characteristics of LIBs.