Modeling of Lithium-ion Batteries via Tensor-Network-Based Volterra Model

Equivalent circuit models (ECMs) are a popular and important tool to characterize and safely use lithium-ion batteries, due to their parsimonious structure, fast computation, and physical interpretability. However, they are often limited in predictive accuracy and thus insufficient for some critical applications. To overcome the limitation, this paper proposes to integrate the linear double-capacitor model, an ECM, with a data-based Volterra model to build a physics-informed data-driven model for lithium-ion batteries, which is named as Volterra double-capacitor (VDC) model. The VDC model uses the ECM as a feature extractor to capture physical features of charging/discharging; taking the features, the Volterra model then approximates the complex nonlinear dynamics inherent to the battery and predicts the terminal voltage. In particular, the Volterra model exploits a tensor network representation to break the curse of dimensionality. Further, a parameter identification approach is constructed to extract the parameters of the model from data. The experimental validation demonstrates this new model’s high accuracy, suggesting its promise for various future applications.