A BATTERY CELL PERFORMANCE GRADING APPROACH

摘要

The performance grading for battery cells is an important research issue to relevant fabricators and downstream purchasers. This research aims to develop a systematic performance grading approach for effectively and efficiently evaluating the performance of lithium-ion battery cells. After identifying the essential quality characteristics, this approach employs some common techniques for the evaluation: clustering analysis, Multiple Criteria Decision-Making (MCDM), and correlation analysis. For clustering analysis, an Ant-Based Self-Organizing Map (ABSOM) algorithm embeds the exploitation and exploration transition rules of ant colony optimization inside as well as U-matrix block diagram to make the algorithm more sound and identifiable. Additionally, this algorithm is combined with the K-Means into a two-stage method to improve the clustering result. After compared with SOM+K-Means and K-Means alone, the ABSOM shows better visualization and clustering results. Subsequently, the TOPSIS method is applied for grading analysis to provide battery fabricators a basis to assign a proper level for each cell and the following applications.