Data-Driven Characterization of Batteries Using Neural Networks

This thesis presents a data-driven approach to characterizing the thermal behavior of lithium-ion batteries using neural networks. The model leverages a lab test dataset containing impedance, temperature, and State of Charge measurements to accurately estimate the battery temperature under various operating conditions. By employing advanced machine learning techniques, the study demonstrates how neural networks can enhance the predictive accuracy of battery temperature, a critical factor in battery performance and safety. The results underscore the potential of neural networks in advancing battery temperature monitoring and predictive analytics.