An Online Model-Based Diagnosis Method for Micro-Short Circuits in Series-Connected Lithium-Ion Battery Packs

Short circuit faults are a primary cause of safety incidents in lithium-ion batteries. During the micro-short circuit (MSC) stage, the large short circuit resistance and small short circuit current make effective detection and diagnosis difficult for traditional battery management systems. This article proposes an online fault diagnosis framework for series-connected battery packs. First, leveraging the slow time-varying characteristics of the open-circuit voltage (OCV), an online tracking algorithm for battery OCV is constructed based on a second-order equivalent circuit model. This effectively addresses the challenge of the battery OCV relying on offline measurements. Second, the online estimation of battery state of charge (SOC) and maximum available capacity is achieved by combining Gaussian filtering, particle filtering, and the total recursive least squares algorithm with a forgetting factor. This method enhances computational efficiency while fully considering the bias noise of SOC estimation and the error of cumulative Ampere-hour metering. Furthermore, this article designs a set of general fault diagnosis current profiles, considering the influence of charge and discharge capacity and dynamic excitation. Finally, based on the capacity characteristics of normal, MSC, and low-capacity batteries, corresponding fault diagnosis indicators are constructed to identify different fault types effectively. The effectiveness and applicability of the proposed algorithm under varying degrees of micro-short-circuit resistance are validated through real-world experiments.