Sorting Strategy for Retired Batteries Based on Voltage Segments

With the application of lithium-ion batteries (LIBs) becoming more and more widespread, how to efficiently and accurately classify and grade retired batteries is the biggest challenge facing the stepwise utilization of LIBs. To shorten the feature extraction time and maximize the battery information, this paper extracts the voltage segment with the largest amount of effective information, which is in the high-rate constant current charge/discharge curve, by using the mutual information index, and extracts 12 characteristic parameters (CPs) on this voltage segment. In addition, since there may be some correlation between different HIs, this paper further eliminates the redundant data by dimensionality reduction through principal component analysis (PCA). Finally, the advantages of the processed data are verified by the fuzzy mean (FCM) algorithm. Experimentally, it is proved that this method can improve the accuracy and speed of clustering well.