Novel Fitting Algorithm for Parametrization of Equivalent Circuit Model of Li-Ion Battery From Broadband Impedance Measurements

Jussi Sihvo,Tomi Roinila,Daniel-Ioan Stroe

doi:10.1109/tie.2020.2988235

Jussi Sihvo, Tomi Roinila + Show 1 more

Open Access

https://doi.org/10.1109/tie.2020.2988235

Copy DOI

Abstract

The impedance of lithium-ion (Li-ion) batteries contains information about the dynamics and state parameters of the battery. This information can be utilized to improve the performance and safety of the battery application. The battery impedance is typically modeled by an equivalent-circuit-model (ECM), which provides the dynamic information of the battery. In addition, the variations in the model parameters can be used for the battery state estimation. A fitting algorithm is required to parametrize the ECM due to the nonlinearity of both the battery impedance and ECM. However, conventional fitting algorithms, such as the complex-nonlinear-least-squares (CNLS) algorithm, often have a high computational burden and require selection of initial conditions, which can be difficult to obtain adaptively. This article proposes a novel fitting algorithm for the parametrization of battery ECM based on the geometric shape of the battery impedance in the complex-plane. The algorithm is applied to practical and fast broadband pseudorandom sequence impedance measurements carried out at various state-of-charges (SOC) and temperatures for lithium-iron-phosphate cell. The performance of the method is compared to conventional CNLS algorithm with different initial conditions. The results show that the proposed method provides fast and accurate fit with low computational effort. Moreover, specific ECM parameters are found to be dependent on the battery SOC at various temperatures.

Highlights

T HE increasing usage of lithium-ion (Li-ion) batteries in electrical transportation and renewable energy storage applications is introducing strict demands for the safety and performance monitoring of Li-ion batteries
The monitoring of Li-ion batteries is based on the estimation of the battery state parameters, such as, state-of-charge (SOC) and state-of-health (SOH), which are indirectly obtained from the voltage, current, and temperature measurements of the battery system
This is most likely caused by the fact that the CNLS-init-3 found different local minimum than the other CNLSs due to the poor selection of the initial conditions [9]

Summary

Introduction

T HE increasing usage of lithium-ion (Li-ion) batteries in electrical transportation and renewable energy storage applications is introducing strict demands for the safety and performance monitoring of Li-ion batteries. Due to the nonlinearity of the ECM, an optimization algorithm, such as particle-swarm-optimization [7] or complex-nonlinear-least-squares (CNLS) algorithm, is required to fit the ECM accurately to the impedance data [3], [9]. Adaptively obtained accurate initial conditions can improve the performance of the fitting algorithm in changing operating conditions of the battery, but can reduce the complexity needed for the fitting algorithm [15]. This presents an opportunity to design the algorithm in a lighter manner than the conventional algorithms, making it more suitable for practical applications

Methods

Results

Conclusion