Research On Consistency Screening And Diagnosis Strategy Of Li-ion Power Battery For Vehicle

Power battery is the energy source of new energy vehicles,and the performance of the power battery system largely determines the performance of the vehicle.The consistency of the power battery is mainly due to the degree of difference in the characteristics of the individual monomers in the battery pack.The consistency of the power battery has a great impact on the performance,stability and battery life of the power battery system.This article concludes from the experiment and theoretical analysis of different battery grouping methods that the difference between the series charging and discharging battery cells is obvious in the voltage curve.Use 30 cylindrical batteries to design and carry out series charging and discharging experiments;combine the variation law of the voltage variation coefficient in the whole charging and discharging process,analyze the curve section by section,extract the consistency and screen the primary characteristic variables;analyze the characteristic variables based on the principal component analysis method Perform further extraction and dimensionality reduction to obtain secondary feature variables;perform cluster analysis on secondary feature variables based on K-means clustering,and divide battery cells into different categories;select the category with the largest number of samples that is closest to the cluster center The 18 batteries of the battery are connected in series again for charging and discharging tests.Through the analysis and comparison of the test results,it is verified that the method can effectively complete the battery consistency screening work.The characteristics are more comprehensive,and the process is simple and easy to implement.In view of the obvious noise interference of the voltage data during the actual vehicle operation,and the uneven performance of the voltage difference time,the data of 108 cells were transformed based on the DWT;it was found that the local difference of the voltage curve can be used to highlight the overall difference,but with wavelet The decomposition level deepens and the voltage curve information loss increases;this paper divides the transformed voltage curve into normal curve group and difference curve;introduces the LOF algorithm to diagnose the outlier degree of the difference curve,and finds that the LOF score of the curve is affected by its sparseness;But they all show the characteristics that the LOF score of the normal curve group is relatively uniform,and the LOF score of the outlier voltage curve is much larger than the average;it can effectively detect the difference of the cells.