An Ultrasonic Reflected Wave-Based Method for Estimating State of Charge of Hard-Shell Lithium-Ion Batteries

Efficiently detecting the state of charge (SOC) of lithium-ion battery is crucial for advancing new energy technologies, and using ultrasound for SOC characterization has been demonstrated as a viable method. However, there is relatively little work on ultrasonic characterization of SOC by combining the structure of lithium-ion batteries. In this study, we propose a multifeature indicators SOC estimation method for hard-shell lithium-ion battery using ultrasonic reflected waves. We analyze wave structure and X-ray computed tomography (CT) result to identify echo origins. On this basis, a feature extraction method combining "echo correlation features" and "bottom wave features" are proposed, with a total of 16-D acoustic features extracted. Then, a "classification-regression" machine learning model composed of support vector regression and extreme gradient boosting is designed. This method can accurately estimate the SOC of the battery, and the average root-mean-square error of the estimation results on the test set is about 1.62%. The estimation results under different experimental conditions and different models also prove that the method we proposed has strong accuracy and robustness, which is a new idea for the application of ultrasonic detection methods in battery management systems (BMSs).