State of Health Estimation for Lithium Battery via Indirect Health Factor with Gaussian Process Regression

To improve battery management systems's (BMS) safety and dependability, state-of-health (SOH) estimation must be done accurately. In the paper, we offer an estimating approach using differential thermal voltammetry (DTV), incremental capacity (IC), and differential thermal capacity (DTC) curves, which utilize the Gaussian Process Regression (GPR) algorithm. First, the IC, DTV, and DTC curves are smoothed using a Savitzky-Golay (SG) filter technique. Subsequently, person correlation analysis was employed to select relevant health features, which were then utilized in GPR to develop and establish the battery aging model. Finally, NASA and Oxford datasets are used to validate the estimation accuracy, robustness, and reliability of the proposed model. The results indicate that the proposed model is capable of delivering precise forecasts regarding battery health status.