Two-Stage Retired Battery Sorting Strategy Based on Multi-Feature Fusion

Scale retirement of lithium-ion batteries has brought about the issue of their gradient utilization, and the consistency sorting technology is the key to it. However, the existing consistency sorting technology has a single sorting feature, low sorting effectiveness and other problems. This paper proposes a two-stage decommissioning lithium battery sorting method. The overall partial correlation coefficient analysis of the features is carried out to determine the sorting features. In the first stage, DBSCAN performs multiple clustering for different combinations of features to cross-validate the results and exclude abnormal batteries. In the second stage, the cohesive hierarchical clustering algorithm sorts retired batteries, and the number of clusters is determined by observing the dendrogram. The results show that the standard deviation of the capacity of the battery clusters is reduced by at least 48.19%; the standard deviation of the internal resistance is reduced by at least 36%; and the standard deviation of the OCV is reduced by at least 29% relative to that before the unsorted, proving the effectiveness of the proposed method.