Time Domain Impedance Parameters Identification Using Fractional Modeling Applied to Li-ion Batteries

Identification in time domain represents an accurate alternative technique for the Electrochemical Impedance Spectroscopy (EIS) measurements that requires long measurement time to characterize low frequency behavior. This paper focuses on the parameter identification of a Li-ion cell fractional model using measurements in time domain. The fractional model is based on an Equivalent Circuit Model (ECM) and is derived from the simplified Randles model. The parameters of the impedance fractional model are identified using Levenberg-Marquardt optimization algorithm. Results are validated in time domain using simulation data and experimental data obtained from chronopotentiometry experiments performed on a 3.4 Ah Li-ion cell. Experiments are carried out at different States Of Charge (SOC) and show the interest and the relevance of the employed methodology.

I. Introduction

The applications of batteries are numerous and have been developed first to meet the energy needs of embedded electronic equipment such as mobile phones, laptops and cameras [1]. Furthermore, they are used on automotive applications, particularly through the development of Hybrid Electric Vehicles (HEV) and Electric Drive Vehicles (EDV) requiring the design of batteries to satisfy their power needs [2]. Li-ion batteries have been chosen as storage systems for these kinds of applications due to their high energy and power densities [3]. Nowadays, the battery performance represents a major challenge which requires accurate modeling tools in order to evaluate and diagnosis its characteristics [4]. Models should ensure accurate description of the battery dynamics: electronic conduction, charge transfer and ionic diffusion. These phenomena behave according to very different dynamics. Therefore, a model based on the impedance physical characteristics is required.