An improved method of state of health prediction for lithium batteries considering different temperature

Due to the inaccurate extraction of battery aging characteristics, low accuracy of available capacity, and a large amount of training data, a data-driven algorithm is used to predict the functional capacity of Li-ion batteries. We propose a method to predict the health status of Li-ion batteries under different temperature conditions using health factor (HF) and weighted particle swarm optimization Gaussian process regression (WPSO-GPR) models. The health factor (HF) is extracted adaptively in the voltage and time profiles during the discharge stages for the HF extraction problem in different temperature predictions. A Gaussian process regression (GPR) model for SOH prediction is created for the capacity regeneration problem, with rational quadratic coefficients as the kernel function. An adaptive weight particle swarm algorithm is used to optimize the GPR model. The proposed framework for designing single-cell experiments on the cell data set is validated using two evaluation metrics: root mean square error (RMSE) and means absolute percentage error (MAPE). The results demonstrate that the proposed method has a short sample size, high prediction accuracy, and broad applicability.