Abstract: Accurate monitoring of energy storage battery degradation anomalies is the key to ensure the safe operation of battery energy storage systems. The adoption of reconfigurable battery topology is a major trend for future battery energy storage systems, and existing data-driven battery health assessment methods primarily emphasize improvements at the algorithmic level, making it difficult to leverage the advantages of this framework. Aiming at this problem, this paper proposes a two-level diagnosis method for abnormal batteries; the primary diagnosis adopts a least-squares support vector machine classification model trained by full-condition full-life cycle simulation dataset to screen out the suspected abnormal battery modules; the secondary diagnosis adopts a health state estimation model based on residual connection and gated cyclic unit to realize the accurate estimation of the health state of the storage batteries and to verify the results of the primary diagnosis. The experimental results show that the diagnosis method proposed in this paper has a high accuracy rate in both diagnostic links, and realizes the accurate monitoring of energy storage battery attenuation anomalies on the basis of reconfigurable battery topology.

Key words: lithium-ion battery, fire, fire extinguishing agent, hydrogel, flame retardant