Assessment Of The State Of Health Of Retired Lithium-ion Batteries And Screening Into Modules Based On Internal Resistance

After long-term use of power batteries in new-energy electric vehicles,there will be inconsistencies among cells.Therefore,for retired batteries,the first thing to do is to estimate the state of health to determine the echelon utilization locations of different aging cells.Next,to ensure the working capacity and voltage,echelon utilization batteries always exist in the form of modules,so a set of screening rules are required for retired batteries to select suitable cells to reduce inconsistencies.Finally,appropriate welding parameters need to be adjusted during the welding process to ensure the quality of the modules.For the above series problems,in this paper,the cylindrical cell battery made of lithium iron phosphate and ternary material were studied,and provided technical and data support for the echelon utilization of two different materials of retired batteries:(1)Capacity was estimated based on internal resistance curve.Resistance-state of charge(SOC)curves of retired lithium batteries were obtained by using Hybrid Pulse Power Characteristic(HPPC)test methods,then it was found that the trend of total internal resistance was same compared with polarization internal resistance,so it was believed that the change of polarization internal resistance mainly contributes to the change of total internal resistance.And due to the influence of the cathode material on the lithium-ion diffusion coefficient,the internal resistance of the ternary battery was generally smaller than that of lithium iron phosphate.The characteristic values such as the minimum internal resistance and the average internal resistance in the quadratic curve were extracted as the input of the model,and the capacity was estimated by using multiple linear regression(MLR)and back propagation(BP)neural network.The experimental results showed that for the retired batteries of same material,the prediction accuracy of the two models was relatively close,its average errors were less than 10%.For mixed cells,the maximum error of the MLR model reached 22.34%,while for the BP neural network model it was 13.46 % and the average error was only 4.95%,thus the capacity can be accurately estimated even when material of cells is unknown.(2)Capacity was estimated based on low frequency amplitude The suitable EIS measurement steps were determined,including the input current type,excitation amplitude,etc.Impedance spectrum of different aging degrees of lithium iron phosphate and ternary material cells were tested,it was found that with the decrease of state of health,the phase-frequency characteristics fluctuated left and right,its relationship was not obvious,while the amplitude of low-frequency impedance had a significant increase trend,and it could be seen that the curve resembled a quadratic exponential function.Therefore,the impedance amplitude under 0.01 Hz was chosen to estimate SOH,then the SOH evaluation function was obtained by fitting and its effect was verified.According to the results,the goodness of curve fitting of the two materials was 0.96 and 0.99,and the effect was very good.The root mean square error of the estimated capacity and the measured capacity were 1.78% and 1.5%,indicating that the method was effective and the test time was short,thus it had certain engineering application value.(3)The estimated capacity and internal resistance was used to screen the batteries,which were welded into groups after the parameters were optimized.By observing the charging and discharging voltage difference curve of the series module,it was found that the maximum voltage difference between the cells always appears in the full SOC at the end of charging,and because the upper cut-off voltage was sensitive to the capacity and cannot be reduced,therefore,a retired battery screening rule based on internal resistance was proposed,that was,the capacity and internal resistance at full SOC were used as the consistency screening index,and similar cells were selected in series to form a module.Compared with the comparative experimental module that did not use this rule,it was found that the maximum voltage difference could be reduced by 96.3%,and the maximum capacity retention rate could be increased by 3.27%.Afterwards,the orthogonal test was used to optimize the spot-welding process parameters,and the optimal process parameter combination was obtained by the range analysis as welding current 13 A,pulse time 2ms and welding pressure 3.5N.Then the analysis of variance was used to perform F test on the significance level of the parameters,and the results obtained were consistent with the range analysis,indicating that the data was true and valid.The test results showed that the welding current has the greatest influence on the quality of the joint,followed by the pulse time,and finally the welding pressure.