Research On Performance Degradation And Remaining Useful Life Prediction Method Of Lithium-ion Battery Under Working Conditions

Lithium-ion batteries are the power source of new energy vehicles,and their reliability and safety are crucial.In order to accurately grasp the health state of lithium-ion batteries,it is necessary to carry out research on the performance degradation law and remaining service life of lithium-ion batteries.In actual working conditions,due to different user habits,the discharge rate,discharge depth and discharge sequence of the battery are random,the battery discharge current fluctuates greatly,and the battery degradation behavior shows nonlinear,difficult to evaluate and other complex characteristics.To this end,the paper carried out battery charging and discharging tests according to the New European Driving Cycle(NEDC driving cycle),approximated the complex operating conditions of the battery,and based on sensing information such as voltage and current,the performance degradation analysis,capacity evaluation and the prediction method of Remaining useful life(RUL)of lithium-ion battery were studied.Specific research contents include:(1)Sorting out the working principle of lithium-ion batteries and influencing factors of performance degradation,building a lithium-ion battery test system,taking18650 terpolymer lithium batteries as test objects,designing and carrying out discharge capacity performance test,mixed power pulse characteristic test,NEDC discharge test and cycle life test,and getting data of battery charging and discharging voltage,current and available capacity.The capacity degradation trend and performance degradation law of lithium-ion battery were analyzed.(2)Based on the external state parameters obtained from the lithium-ion battery test,the particle diffusion lumped model was constructed to reflect the internal mechanism variation characteristics of the battery and further reveal the performance degradation rule under the battery working condition.Under NEDC discharge conditions,ohmic loss,activation loss and concentration loss all show an increasing trend as the aging state of lithium-ion batteries becomes more and more serious.From the influence degree of the three voltage losses,ohmic loss has a greater contribution,which is the main factor affecting the voltage loss during the battery use.(3)Based on the charging data collected from the battery life cycle experiment,the traditional artificial method and the One-Dimensional Convolutional Neural Network(1DCNN)are used to extract and evaluate the health features that characterize the battery performance degradation.On this basis,the characteristic parameter space was constructed,and the available capacity of lithium-ion battery was evaluated by using Support Vector Machine(SVM),Long short-term Memory(LSTM)network and SVM-LSTM fusion models respectively.The results show that the evaluation results based on SVM-LSTM have higher accuracy,which provides a good data basis for the prediction of the remaining service life of lithium-ion batteries.(4)Based on the capacity evaluation results of the SVM-LSTM fusion model,the LSTM network prediction model was used to predict the remaining service life of lithium-ion batteries.With the change of the starting point of prediction and the increase of training set data,the consistency between the capacity degradation prediction trend and the actual capacity change trend of lithium-ion batteries is gradually enhanced,and the absolute error of the prediction of the remaining service life of lithium-ion batteries is gradually reduced,and the predicted value of the remaining service life of lithium-ion batteries is gradually close to the real value.