Lithium Battery Health State Estimation Based on PSO-GPR Algorithm

Abstract

The State of Health (SOH) estimation of lithium-ion batteries is a critical factor in battery management systems, directly influencing performance evaluation and lifespan prediction. Traditional SOH estimation methods suffer from limitations such as insufficient accuracy and poor robustness, making it difficult to meet the demands of complex operating conditions. This paper proposes a novel SOH estimation method for lithium-ion batteries based on Particle Swarm Optimization (PSO) and Gaussian Process Regression (GPR). By analyzing the temperature variation curve during the constant-current charging process of lithium-ion batteries, three geometric features are extracted as health indicators. These features are combined with the peak information from the incremental capacity curve to construct a PSO-GPR model for SOH estimation. Experimental results demonstrate that the proposed method outperforms traditional GPR and Backpropagation (BP) neural network methods in terms of prediction accuracy and robustness. The SOH estimation error for most batteries is within 2%, effectively improving the estimation accuracy of lithium-ion battery SOH.