Comparative Research On Battery Inconsistency Diagnosis Algorithms

Power battery is the key component of electric vehicle energy storage,and it is very important to the safety of electric vehicle,so battery fault diagnosis has received extensive attention.The inconsistency of battery cells is one of the important factors of battery failure.Although domestic and foreign scholars have conducted in-depth research on the inconsistency diagnosis algorithm of battery cells,they are affected by the lack of selection criteria such as algorithm parameter selection principles and fault benchmark true values,the differences in the performance of these algorithms are still vague.Based on this,this paper selects 8 typical individual inconsistency diagnosis algorithms,and conducts a comparative study on their performance on the same data set.Valuable reference.Firstly,this paper extensively investigates the research on battery cell inconsistency fault diagnosis,selects 8 typical cell inconsistency diagnosis algorithms,and classifies them into two categories: outlier detection and information entropy analysis.The principle of 8 algorithms is briefly introduced.Secondly,the specific problems of the practical application and implementation of different diagnostic algorithms are studied.The original data required for diagnosis is selected;the original data is preprocessed,and the process and results of the preprocessing are analyzed and discussed considering the physical rationality and application background;the specific algorithm implementation process for the application of the 8algorithms is analyzed and discussed.The research and practice are carried out,focusing on the analysis of the selection principles of different algorithm parameters,and the method of selecting reasonable parameters for different algorithms is determined.Finally,the diagnostic performance of different algorithms is compared and analyzed.The fault matrix method is selected as the basis for algorithm comparison,and the diagnostic performance of different algorithms under static and dynamic characteristic data is compared based on experimental samples.