Charge State Detection Of Lithium-ion Square Power Battery Based On Ultrasonic Detection

When lithium-ion batteries are actually being charged and discharged,accurate monitoring of the charge state(SOC)is useful in preventing overcharge and overdischarge issues.There are several different SOC estimating techniques available right now.The SOC of a lithium-ion battery is estimated using an ultrasonic nondestructive testing approach in this paper.The SOC of a large-capacity square battery is evaluated under high current conditions utilizing two continuous sinusoidal signal waves.The research contents and conclusions are as follows:The battery and ultrasonic probe were fixed using a specific fixture once the ultrasonic test platform was constructed.First,the same battery was used for the ultrasonic test trials,which were conducted under one-cycle charging and discharging and three-cycle charging and discharging,respectively,with current densities of 0.25 C,0.5C,0.75 C,and 1C.The ultrasonic signal as well as the battery voltage and current were recorded.Then,ultrasonic signals were gathered while three batteries with the same specs through 100 cycles of charging and discharging at a 1C ratio.The link between SOC and ultrasonic sounds is then investigated using additional analysis of the experimental data.The Fourier rapid shift technique is used in Matlab to convert the recorded signal’s amplitude to time frequency.Two continuous ultrasonic experimental sounds are used to compute the phase difference using the cross-correlation function.Following a thorough analysis of the processed data,it is determined that the phase difference between the two signals gradually increases with an increase in SOC,demonstrating a linear relationship with SOC,and the battery signal amplitude gradually increases with an increase in lithium battery capacity during charging.As the capacity of a lithium battery is discharged,the battery signal’s amplitude diminishes.The phase disparity between the two signals increases as SOC decreases.In the process of charging,the anode material will be more inclined to a rigid sphere in the process of ultrasonic propagation due to the embedding process of lithium ions,and the sound wave will lose part of its energy due to the scattering on the surface of the rigid sphere.Therefore,the amplitude of low current density will decrease in the case of high current density.In the discharge process,the negative material will have a fast delithium rate,high internal lithium ion density and high internal stress,resulting in a slow outward migration rate of lithium ions.At this time,the negative material tends to be elastic.With the removal of lithium ions,the surface of the negative material tends to be an elastic sphere,and the energy attenuation will decrease when the ultrasonic wave passes by.So at low current density the amplitude goes up relative to high current density.The Support Vector Machine(SVM)algorithm of machine learning was then used to create a prediction model from the ultrasonic test data of three sets of 100 cycles of charge and discharge at 1C rate.The test results were categorized.Amplitude,voltage,and phase difference were inputs to the training set,and SOC was the outcome.Charge and discharge SOC forecast errors are 7.4% and 8%,respectively.With a SOC prediction error of 4.2% for charging and2.3% for discharging,the model was retrained using the Particle Swarm Optimization(PSO)method after being optimized to find the best parameters.The prediction effect of PSO-SVM is better when compared to the prediction error before and after optimization.