Degradation Mode Knowledge Transfer Method for LFP Batteries

Lithium-ion (Li-ion) batteries are widely utilized as energy storage units owing to their high energy density and safety. However, when battery degradation occurs, Li-ion batteries deteriorate and become untrustworthy. Accurate diagnosis and identification of the degradation modes (DMs) constitute a critical task for systems employing Li-ion batteries. Current diagnosis methods are usually postanalysis and cannot be directly employed for diagnosing the batteries that are in operation. This study proposes a ResNet-50-based diagnosis model for DMs, which can quantify the contribution of three DMs for the synthetic datasets. Because the real and synthetic datasets are independent and identically distributed, it is difficult to apply this model to the real datasets. To bridge the gap, this article proposes a deep domain adaptation method to minimize the classification loss and domain adaptation loss between the source domain (synthetic) and the target domain (real), such that the degradation knowledge learned from the synthetic batteries can be transferred to the real batteries. The model's input, structure, and parameters are optimized through simulation tests to improve the diagnosis accuracy. A validation session is designed to verify the classification accuracy of unlabeled DMs of the lithium iron phosphate (LFP) battery. The results show that the proposed method can effectively transfer the knowledge of degradations from synthetic batteries to real-world LFP batteries to diagnose and identify DMs of LFP batteries with relatively high classification accuracy.