A Comprehensive Algorithm for Estimating Lithium-Ion Battery Parameters From Measurements

Abstract

The use of equivalent circuit models for simulating the operating behavior of lithium-ion batteries is well established in the automotive and the renewable energy sector. However, finding the correct parameter set for these models is still a challenging task. This manuscript proposes a comprehensive methodology for estimating the required, temperature dependent simulation parameters from battery measurements. Based on a specific load current and prior system knowledge, an algorithm first analyses the correlation between current steps and the measured terminal voltage. Second, a combination of particle swarm optimization and Gauss-Newton algorithm fits the initially estimated parameters from the first algorithm to the measurement data. Finally, the dependency of each simulation parameter on both the state of charge and the battery temperature is determined. As this contribution aims at modeling reversible effects of lithium-ion batteries, ageing effects are neglected. The validation against measurement data proves that the generated parameter set enables the user to accurately reproduce and investigate the operating behavior of the chosen battery. Applied to a lithium-iron-phosphate cell, the comparison between measurements and simulations in standardized real-life automotive driving cycles (Artemis, FTP75 and WLTC) shows a terminal voltage error of less than 1.09% within the typical operational window between state of charge 0.15 and 0.95.