New energy battery cabinet single voltage detection

As electric vehicles advance in electrification and intelligence, the diagnostic approach for battery faults is transitioning from individual battery cell analysis to comprehensive assessment of the entire battery system. This shift involves integrating multidimensional data to effectively identify and predict faults.

The detection method of battery parameters in battery management system is simple and the accuracy is limited [, , ], but the accuracy of parameters is the direct factor affecting the fault diagnosis results. Wang et al. proposed a model-based insulation fault diagnosis method based on signal injection topology.

At present, the analysis and prediction methods for battery failure are mainly divided into three categories: data-driven, model-based, and threshold-based. The three methods have different characteristics and limitations due to their different mechanisms. This paper first introduces the types and principles of battery faults.

The battery overvoltage or undervoltage fault can be diagnosed using the threshold-based method. The voltage information collected by the voltage sensor is compared with the preset threshold. When the battery voltage exceeds the threshold, the fault occurrence state and fault occurrence time are defined .

Sun et al. proposed a two-layer fault detection strategy like Gan et al., with the difference that they monitored voltage and temperature and other parameters simultaneously in the first layer strategy, which improved the reliability of battery thermal fault detection.

Battery degradation is inevitable, and it will also affect various battery parameters, and the existing sensor fault detection and isolation (FDI) methods ignore this important factor [, , ]. Tran et al. took battery degradation into account and proposed a sensor FDI scheme based on a first-order RC-equivalent circuit model.