Fast Charging Li-Ion Battery Capacity Fade Prognostic Modeling Using Correlated Parameters' Decomposition and Recurrent Wavelet Neural Network

Abstract

Continuous cycling of Lithium-Ion (Li-ion) batteries, as required by applications, degrades their resulting capacities over time. This degradation is generally negligible in the early charge/discharge cycles. An increase in charging/discharging rates (C-rate) applied on a continuous cycling battery reduces its charging time, thereby resulting in a fast charging battery, however, this also escalates the degradation. This degradation can be studied from the resultant decrease in charging/discharging capacities, also termed as capacity fade. To analyze capacity fade, an approach using reference C-rate based charging/discharging capacity analysis is proposed for a time-limited degradation analysis. Further, a step-ahead forecasting approach is proposed for all the charging/discharging capacities' correlated original, and corresponding deviation parameters, to present time-ahead modeling of all the impacted parameters. A combinatorial empirical mode decomposition (EMD)-recurrent wavelet neural network (RWNN) model is proposed as the step-ahead forecasting approach for the correlated parameters. Finally, a comparison of error values between the proposed EMD-RWNN model is performed with combinatorial EMD-wavelet neural network (WNN), standalone WNN and RWNN models to effectively analyze the resulting superior performance of the recurrent nature of the proposed model by forecasting every decomposition.