Reconstructing long-time voltage relaxation from short pulse tests for faster battery testing and modeling

The galvanostatic intermittent titration technique (GITT) is a widely used method for obtaining open-circuit voltage (OCV) profiles and kinetic information in rechargeable batteries. However, its practical use is limited by the long relaxation period required after each current pulse. This study proposes a physics-consistent short-window extrapolation method to reconstruct long-time relaxation behavior from only minutes of post-pulse data. Based on this framework, two pulse-wise OCV extraction strategies, namely threshold-based convergence and time-based extrapolation, are evaluated. The utility of reconstructed relaxation trajectories for downstream equivalent-circuit-model (ECM)-based dynamic battery modeling is also assessed. The method was evaluated using more than 5.7 k pulses across 8 different rest-duration protocols. Using 10 min of measured relaxation data, the threshold-based convergence yielded RMSE values of 4.35 mV during charge and 8.98 mV during discharge relative to the measured 3 h relaxation, reducing the required relaxation time from 180 min to 10 min. For downstream modified 2-RC ECM analysis, the time-based extrapolation-assisted framework reduced the mean absolute error by 78% relative to direct identification from 10 min data, and achieved accuracy comparable to direct identification from 1 h relaxation data while using only 1/6 of the measured relaxation time. Independent validation on five additional commercial cells confirmed that reconstruction-assisted ECM consistently reduces the error relative to direct ECM identification. The reconstruction was especially consistent over the mid-SOC range (20–80% SOC), with relatively larger deviations observed near the voltage cut-offs.