Ultrasonic estimation of lithium-ion battery state parameters using hybrid sparrow search algorithm and relevance vector machine

Abstract

State of Charge and State of Health are critical indicators for evaluating the operational status of lithium-ion batteries. Accurate estimation of these parameters is essential for ensuring safe operation and optimizing charging-discharging performance. This study presents a novel method for estimating SOC and SOH based on ultrasonic characteristics. A pouch battery is selected as the experimental object, with ultrasonic waves extracted in real-time during battery operation. Acoustic parameters are obtained via signal processing to establish a training dataset for SOC estimation. Additionally, the impact of varying discharge rates on battery capacity degradation is analyzed to support battery aging experiments. During the aging experiments, real-time acoustic parameters are collected to develop a training dataset for SOH estimation. The results indicate a strong correlation between the acoustic parameters and the SOH. Moreover, the acoustic parameters serve as the characteristic parameters, and a hybrid-model combining the sparrow search algorithm and relevance vector machine is employed to accurately estimate the state parameters. The maximum relative error for the SOC estimation is 1.51 %, while the maximum absolute error for the SOH estimation is 0.79 %. The research results provide a new technical support for the non-destructive quantitative characterization of lithium-ion battery state parameters.