A Method Of Real-time Monitoring The State Of Charge Of Lithium-ion Battery By The Ultrasonic Wave

As the most promising power and energy storage device,lithium-ion battery has attracted a lot of attention.It is still necessary to further improve its key performance indexes,such as energy density,safety performance and environmental adaptability.Therefore,the demand for lithium-ion battery research increases day by day.Based on the consideration of battery safety,overcharge and overdischarge induced problems in lithium-ion batteries have gradually attracted attention.Accurately monitoring of State of Charge can effectively prevent these problems.However,in conventional battery management system,the estimation of SoC is completely dependent on the battery’s current,voltage and other parameters,which lead to poor estimation accuracy and thus cannot guarantee the consistency of the battery pack.This thesis mainly focuses on the study of NCM and LiFePO₄ batteries which are prevalent in the market.We proposes a new SoC monitoring method based on acoustic principles.This method measures the ultrasonic signal under different charge and discharge current,and then analyze the ultrasonic signal characterizations and the SOC.The relationship between ultrasonic signal and the SOC has been successfully established.Then,the ultrasonic signal of LiFePO₄ system battery was further processed by wavelet transformation to obtain the time-frequency information of the signal.Based on the obtained ultrasonic signal,a neural network deep learning algorithm was constructed,which uses to classified and labelled ultrasonic signal data to achieve estimation of SoC.At present,the estimation of SOC of battery by this method can achieve the self identification analysis with error better than 3%.