An interpretable aging prediction model for Ni-rich all-solid-state batteries informed by mechanistic modeling

Ni-rich all-solid-state batteries (ASSBs) are promising candidates for next-generation energy storage technology, while they suffer from much more severe capacity fading than lithium-ion batteries with liquid electrolyte, severely compromising their practical application and commercialization. Accurate degradation prediction is required to improve the operational stability of Ni-rich ASSBs. A novel degradation prediction framework for NCA (LiNi 0.8 Co 0.16 Al 0.03 , Nickel content: 80%) cathode ASSBs is proposed based on a mechanism model combined with deep neural network in this study. This study first constructs a three-dimensional physical model of ASSBs with NCA cathodes. On the basis of experimental observations, three cathode capacity degradation processes, namely interfacial side reactions, interfacial delamination and particle cracking, are parameterized. Subsequently, the CNN-LSTM hybrid framework is constructed for model training, and a multi-level prediction module is further embedded into the architecture to leverage the advantage of physical model in capturing battery internal characteristics, e.g., the Li concentration as an additional feature. Validated by simulation data, the proposed model possesses the high prediction accuracy in estimating NCA-based ASSBs degradation. Comparative analyses are carried out against the prediction model without concentration correction to further verify the rationality of the proposed prediction framework, indicating that concentration correction substantially enhances the prediction accuracy under small-sample scenarios. The three-dimensional mechanism model-neural network fusion framework proposed in this work verifies that neural networks embedded with physical prior information such as lithium-ion concentration exhibit stronger learning capacity. Even under limited data conditions, the framework can accurately predict capacity degradation of ASSBs with NCA cathodes. This study provides theoretical references for performance prediction and battery management of NCA-based ASSBs.