Hybrid Modeling of High-Voltage Battery Systems for Software Control

Abstract

Abstract In recent years, electric vehicles have taken up a greater part of the automotive industry. This thesis investigated the potential of using a hybrid approach to modeling the behavior of batteries used in electric vehicles on pack level. Using existing models, both empirical and physics-based, to create a model meant for software control. The aim was to find a way to model the aging of batteries and control the battery pack based on internal electrochemical dynamics but at a lower computational load than that of fully physics-based modeling. During the project, it was shown that the use of a hybrid model to simulate batteries on pack level is a working method that possesses the ability to control on the basis of physics-based dynamics of the batteries. Using a physics-based controller for fast charging of the battery pack was also shown to decrease fast charging time 18 % compared to using a fast charging protocol based on testing. This was possible without compromising the risks of aging. It was also shown that one can not only use the model to evaluate the performance of aged batteries, but also be able to see the aging effects that a given use of the battery has. The hybrid model was also found to operate on a substantially lower computational load. compared to a fully physical-based model, making it possible to run longer simulations.