DescriptionThe imbalance of electrical potential, temperature, and the State of Charge (SOC) within battery cells may cause safety issues in applications. Models that can predict battery cells’ thermal and electrical behaviors become necessary for real-time battery management systems to regulate the imbalance.
This dissertation introduces a Gaussian Process Regression (GPR) based data-driven framework that succeeds the Multi-Scale Multi-Dimensional (MSMD) modeling structure developed by (Kim, et al. 2011). The framework predictions can reach high accuracies as the full-order full-distribution simulations based on MSMD. It relies on the pseudo-2D model developed by (Doyle, Fuller and Newman 1993) to generate training data, which shifts computation burdens from real-time battery management systems to lab data preparation.
This dissertation introduces deterministic and uncertain input GPR models in the battery cell’s particle and electrode domains. These models work in different phases of the battery cell potentiostatic discharge simulation, where different types of probabilistic finite element analysis (FEA) procedures are applied correspondingly. Dimensional model representation approximations combined with Gauss-Hermite quadrature methods are employed to give solution distributions of temperatures and plate electrical potential differences in the cell domain.
Testing results illustrate the reliability of the GPR based data-driven framework in accuracy and stability under various circumstances; this was accomplished by comparing our results from other researchers’ methods.