Onboard Health Estimation using Distribution of Relaxation Times for Lithium-ion Batteries

Abstract

Real-life batteries tend to experience a range of operating conditions, and undergo degradation due to a combination of both calendar and cycling aging. Onboard health estimation models typically use cycling aging data only, and account for at most one operating condition e.g., temperature, which can limit the accuracy of the models for state-of-health (SOH) estimation. In this paper, we utilize electrochemical impedance spectroscopy (EIS) data from 5 calendar-aged and 17 cycling-aged cells to perform SOH estimation under various operating conditions. The EIS curves are deconvoluted using the distribution of relaxation times (DRT) technique to map them onto a function $\textbf{g}$ which consists of distinct timescales representing different resistances inside the cell. These DRT curves, $\textbf{g}$, are then used as inputs to a long short-term memory (LSTM)-based neural network model for SOH estimation. We validate the model performance by testing it on ten different test sets, and achieve an average RMSPE of 1.69% across these sets.

Figures (7)

• Figure 1: Discharge capacity of 22 cells in the dataset as a function of days for calendar-aged cells (top-left), and Ah-throughput for cycling-aged cells (top-right and bottom). Cell S4, S12, S17, and S22 show maximum degradation in each panel.
• Figure 2: Example EIS curves for Cell S22 at five different SOCs, and 0, 10, 20, 40 and 90 days of aging. Magnitude of impedance is large at 0% SOC, but relatively small at other SOCs. Black arrow indicates the direction of increasing frequency for all EIS curves.
• Figure 3: Example DRT curves for Cell S22 at five SOCs and 0, 10, 20, 40 and 90 days of aging. With aging, SEI/CEI resistance peak ($\tau \approx10^{-2}$s) shows an increase in height and shifts to the left. Charge transfer resistance ($10^{-2} < \tau < 10^0$s) shows two peaks for 0% SOC which increase in height while for other SOCs, it shows a small increase in height and shifts to the left. Diffusion resistance peak ($\tau > 10^0$s) increases in height for 0%, 25% and 50% SOC, but it shows a small decrease in height at 75% and 100% SOC.
• Figure 4: SOH pipeline representing flow of impedance data and architecture of the data-driven model. Data from EIS at 25% SOC is transformed into DRT curves, which are used as input to the LSTM-based model. The model has three LSTM layers with SELU activation followed by three fully-connected layers. On the output, the model provides SOH estimation at different days.
• Figure 5: Estimation results of test set 1 consisting of cells S8, S13, S19, and S23. LSTM has good performance in comparison to the experimental data with an RMSPE of 0.7129%. Linear regression fails to capture the SOH behavior, and underestimates the SOH for majority of the test points.