Game theory and kernel boost efficient clustering and echelon utilization for decommissioned Li-ion batteries of electric vehicles

With the rapid expansion of electric vehicles, the echelon utilization of decommissioned lithium-ion batteries has become an urgent research priority, yet current studies remain preliminary and the technology is still immature. This study proposes a two-stage framework, Combination Weighting Method for Game Theory (CWMGT)-VIse Kriterijumska Optimizacija I Kompromisno Resenje (VIKOR)- Kernel Self-Organizing Map (KSOM), for comprehensive sorting and clustering of decommissioned lithium-ion batteries of Electric Vehicles (EVs) to enhance echelon utilization. CWMGT balances subjective and objective weights, VIKOR provides multi-criteria ranking, and KSOM ensures high-accuracy clustering. By replacing Euclidean distance with Gaussian kernel functions, KSOM improves clustering quality by 12% (Dataset 1) and 17.1% (Dataset 2) compared with traditional SOM, while eliminating misclassifications. The two-stage design also improves efficiency, reducing runtime by 41% on Dataset 2 relative to single-stage KSOM. Validation on laboratory and MIT-Stanford-Toyota datasets confirms that the framework effectively reduces inconsistency, improves classification reliability, and supports optimized reuse in energy storage. Key innovations include: (1) game-theoretic weight optimization, (2) VIKOR-based preliminary sorting, and (3) kernel-enhanced clustering. Overall, the proposed approach advances decommissioned battery management by combining precision and efficiency, thereby supporting large-scale echelon utilization and the circular economy of the EVs industry.