Evaluation of the state of charge of lithium-ion batteries using ultrasonic guided waves and artificial neural network

In this paper, a method for detecting the state of charge (SOC) of lithium-ion (Li-ion) batteries based on ultrasonic guided waves and artificial neural network is proposed. Commercial Li-ion pouch batteries are taken as the experimental object, real-time ultrasonic guided wave detection is carried out during the operation of the battery, and the SOC is analyzed via signal processing. The guided wave parameters are taken as characteristic parameters, and the backpropagation (BP) neural network model is used to accurately estimate the battery SOC. It is found that the frequency band of the direct waves and the variation of their amplitude in the spectrum of the response signal have good correlations with the battery charge-discharge cycle. It is also found that the wave velocities of the two envelope peaks are the same as the change of the SOC, and the time of flight (TOF) decreases with the increase of the SOC. The research results can guide the development of a battery management system based on a guided wave framework that can be applied to the detection and monitoring of the SOC of Li-ion batteries.