A Sorting Method of Retired Lead-Acid Battery For Electric Power Based on Combined Density Clustering

The inconsistency among single-cell retired valve-regulated lead-acid (VRLA) battery for power systems is a primary challenge that restricts their echelon utilization in groups. This paper proposes a method for sorting retired battery based on a combination of density clustering techniques. Firstly, a feature vector is established by leveraging both the dynamic and static characteristics of battery performance, with criteria importance assessed through the intercriteria correlation (CRITIC) method to assign weights to the dimensions. Secondly, the density-based spatial clustering of applications with noise (DBSCAN) algorithm, optimized via the elbow method for its distance threshold, effectively filters out outliers. Finally, the density peaks clustering (DPC) algorithm is employed to generate clustering results. A sorting experiment involving 190 retired batteries utilized for electric power was conducted using this method, and comparative analysis was performed on the results to validate its effectiveness.