Design and Experiment of Nonlinear Observer with Adaptive Gains for Battery State of Charge Estimation

Abstract

State of charge (SOC) is an important evaluation index for lithium-ion batteries (LIBs) in electric vehicles (EVs). This paper proposes a nonlinear observer with a new adaptive gain structure for SOC estimation based on a second-order RC model. It is able to dynamically adjust the gains and obtain a better balance between convergence speed and estimation accuracy with less computational time. A sufficient condition is derived to guarantee the uniform asymptotic stability of the observer, and its robustness with respect to disturbances and uncertainties is analyzed with the help of input-to-state stability (ISS) theory. A selection guide of the observer gains in practical application is presented. The estimation accuracy and convergence rate of the observer are evaluated and compared with those of extended Kalman filter (EKF) based on multi-temperature datasets from two different types of LIB cells. The robustness against different disturbances and uncertainties that may appear in a real vehicle is validated and discussed in detail. The experimental results show that the proposed observer is capable of achieving better performance with less computational time in comparison to EKF for different types of LIB cells under various working conditions. The observer is also capable of estimating SOC accurately for real life conditions according to the validation results of datasets from a battery management system (BMS) in an EV battery pack. Furthermore, the observer is simple enough, and is suitable for implementation on embedded hardware for LIB cells of EVs.

A novel practical state of charge estimation method: an adaptive improved ampere‐hour method based on composite correction factor

“…Accurate SOC estimation can effectively enhance the safety and energy efficiency of batteries. 1 LIB is a complex nonlinear system, 2 complex working condition and sensor measurement process will produce inevitable noise, which makes accurate SOC estimation extremely difficult. 3 Plett defined SOC as a state quantity of physical meaning, 4 but SOC of battery cannot be measured by sensor directly, only by measuring other parameters such as voltage, current, resistance, and temperature.…”

“…At first, Li quantitatively analyzed the influence of initial SOC precision, 24 battery coulomb efficiency, charge-discharge efficiency, and battery capacity on the traditional Ah method. Then, OCV method 25 and load voltage method 26 are proposed to correct the initial error of SOC, the accuracy and auto disturbance rejection ability of improved algorithm are not improved substantially, 27 because the parameters in the integral formula(Formula [1]) are greatly affected by the ambient temperature, charge-discharge ratio, and state of health. 24 Feng proposed an improved Ah method, 28 which considered the influence of the change of battery available capacity on SOC estimation under different temperatures, but false to consider the influence of battery discharge rate and SOH.…”

An equivalent circuit model analysis for the lithium-ion battery pack in pure electric vehicles

“…At the same time, in order to reflect accurately the changing rule of battery, researchers have successively developed various resistance-capacitance (RC) equivalent circuit models. [14][15][16][17][18][19][20][21][22][23][24] More recent study in Wu et al 25 fully considers the difference of internal resistance between charging/ discharging and then improves the Thevenin model. An nRC equivalent circuit model is proposed in Lai et al 26 The high-order RC model is robust to the variations of model parameters and sensor errors.…”

Online State of Charge and State of Health Estimation for a Lithium-Ion Battery Based on a Data–Model Fusion Method

“…Amongst others, the ECMs have a better trade-off between accuracy and complexity and thus are favorable candidates for application in micro-controller units. Generally, ECMs are used to simulate the dynamics of an LIB, while the states of interest are estimated in real time with various observers, such as the Luenberger observer [20], the extended Kalman filter (EKF) [21][22][23], the square root cubature Kalman filter [24], the unscented Kalman filter (UKF) [25], the sliding mode observer (SMO) [26], the particle filter (PF) [27], and the nonlinear observer [28]. For these methods, the ECMs are calibrated offline and the model parameters are assumed to be fixed during operation.…”