Estimaton Of State-of-charge For Lithium-ion Batteries Based On Fractional-order Model

Lithium-ion batteries are widely used in electric vehicle power systems because of their high energy density and good safety performance.The SOC characterizes the current remaining battery power,and is one of the most important performance indicators of batteries.Therefore,SOC estimation is a hot issue of lithium-ion battery research.This article focuses on the core of SOC estimation method battery model and estimation algorithm.First,different from the traditional Integer-order equivalent circuit model,this paper studies the electrochemical impedance spectroscopy characteristics of Li-ion batteries,and then based on fractional calculus theory,puts forward a fractional-order equivalent circuit model which is a better description of the electrochemical characteristics of lithium-ion batteries.The verification results show that this model have a higher model accuracy.Then,because the model parameters of the battery in use will be affected by ambient temperature,current conditions and aging of the battery,and then affect the SOC estimation accuracy,so the accuracy sensitivity of the fractional model to its parameters is analyzed.The internal resistance and SOC joint estimation model is established.The fractional-order extended Kalman estimator is designed and through several simulation under different working conditions and temperatures,the results show that the algorithm can adapt to the changes of temperature and current,accurately identify the model parameter,as well as SOC.It also has strong antijamming capability for measurement noise,and has higher SOC estimation accuracy than the integer-order extended Kalman algorithm under the same conditions.Finally,for the problem that fractional-expanded Kalman estimation has large amount of computation and low algorithm efficiency due to its iterative updating of variables such as covariance matrix,this paper proposes a relatively simple sliding mode state observers based on fractional-order model.Finding the best sliding mode gain matrix and performming simulation experiments under diffenent current conditions,temperatures,and initial state values,the simulation results show that the observer can adapt to change of the current,temperature and the initial state value,and has strong compensability to modeling errors and immunity to measurement noise.Compared with the extended Kalman estimator algorithm,the sliding mode observer algorithm obtains higher computational efficiency only at the expense of a verry small SOC estimation accuracy,and then has stronger practical applicability.