Research On Remaining Life Prediction Method Of Lithium-ion Battery Based On Machine Learning

Lithium-ion battery has been widely used in all kinds of electronic equipment since its inception because of its excellent electrochemical performance.However,the performance of the battery gradually decreases during the specific use process,which brings the problems of safety and reliability.Therefore,knowing the remaining service life of the battery in advance is of great significance to the safe operation of the power system.The internal electrochemical reaction of the battery is relatively complex,and it is difficult to obtain the remaining service life of the battery through the traditional electrochemical model.Through the method of machine learning,combined with the analysis of the historical charge and discharge data of the battery,this thesis predicts the remaining useful life of the battery.The main research contents of this thesis include:In this thesis,a multi resolution decomposition method based on machine learning combined with capacity data is proposed to solve the problem of relatively few battery degradation data and capacity regeneration.The Improved Complementary Ensemble Empirical Mode Decomposition is used to process the battery capacity data,which increases the dimension of the data and reduces the influence of battery capacity regeneration.The sequence with similar randomness is reconstructed by using the permutation entropy theory,which reduces the number of subsequences and improves the training speed.Due to the lack of consideration of charging process and temperature in the battery service life prediction,this thesis extracts battery health factors from three aspects: battery current and voltage data,battery capacity increment analysis and battery temperature.Random Forest is used to analyze the importance of the obtained health factors.The thesis selects the important health factors to participate in Bi LSTM training,which reduces the dimension of training samples and improve the stability of prediction results.Combined with Bayesian optimization network parameters,the optimization time of hyperparameter is reduced.Finally,the heating aging experiment of lithium manganate battery pack is carried out,and the collected data are analyzed to verify the performance of the prediction method proposed in this thesis.