Modeling Li-ion batteries is extremely important for the simulation, design, and testing of EV’s. Unfortunately, modeling Li-ion batteries with granular detail from the ground-up is not the easiest task: complex electrochemical and physics-based diffusion equations that define the transport of Li-ion ions are computationally expensive. A popular method in industry is to use a simpler model: the equivalent circuit model (ECM). The ECM models a Li-ion battery as a circuit of a voltage source, resistors, and capacitors.
I used an Arbin BT-2000 to program current profiles of a Sony Li-ion battery, then collected corresponding voltage data. Then, I created a Simulink model that implemented the ECM in Matlab, and then created scripts that utilized various optimization algorithms to fit ECM parameters to laboratory data. For each algorithm, the fitted parameters were validated against real-life test runs. As it turns out, ECM is a great way to model the voltage response of a Li-ion battery, and genetic algorithms performed the best in our dataset.