State of Health of Lithium-ion Batteries by Data-Driven Technique with Optimized Gaussian Process Regression

Abstract

Lithium ion batteries are a promising energy source for electric vehicles due to their high specific energy and power output. Overall system reliability and stability can be improved by effectively planning battery replacement intervals and monitoring their condition. To guarantee the battery system operates safely, steadily, and effectively, it is necessary to accurately assess the state of health (SOH) of the lithium-ion battery. Capacity might be used to anticipate it directly. To improve the accuracy of the SOH estimate, hyperparameter-optimized Gaussian process regression (GPR) is used. Gaussian process models have the advantage of being flexible, stochastic, nonparametric models with uncertainty forecasts, and may have variance around the mean forecast to account for the associated uncertainties in evaluation and forecasting. The lithium-ion battery data set made available by NASA is examined in this article. The outcomes demonstrate its efficacy and demonstrate that the algorithm may be successfully used for battery monitoring and prognostics. Additionally, the prediction for battery health has been improved through the comparison of predictions with various quantities of training data.