| description abstract | Abstract. With the widespread adoption of electric vehicles, early detection of micro-short circuits and aging in lithium-ion batteries has become a critical issue in safety management. To address the problem of insufficient capacity increment (IC) curve reconstruction accuracy due to the low sampling rate (0.1 Hz), this paper proposes a method that optimizes the Lorentzian function fitting (LFF) using the Markov Chain Monte Carlo (MCMC) algorithm. In order to resolve the issue of overlapping voltage/current signal responses under different faults, the Particle Swarm Optimization (PSO) algorithm is employed to optimize the ɛ and MinPts parameters in the DBSCAN clustering algorithm, effectively distinguishing between normal, aging, and internal short-circuit states. Validation through short-circuit experiments at various levels demonstrates that the proposed method improves the accuracy of IC curve reconstruction to some extent, simplifies the diagnostic process, and offers significant advantages over traditional methods. Additionally, the Lorentzian function allows for direct analysis of peak area and peak position, avoiding the dependency on complex feature extraction algorithms required by traditional differential methods. Furthermore, an improved method for short-circuit resistance estimation based on the difference in peak area of the IC curve is proposed, providing a feasible quantitative approach for assessing internal short-circuit severity. The diagnostic framework built through these methods enables both qualitative and quantitative analysis of internal short-circuit faults. | |
