Research On State Estimation Of Power Batteries Based On Filtering Algorithms And Incremental Capacity Analysis

In recent years,China is vigorously building a clean,low-carbon,safe and efficient modern energy system.Electric vehicles?EVs?,as the main direction of new energy vehicles in our country,are developing rapidly.As the energy density of lithium-ion batteries for EVs continues to increase,the safety and economy of EVs become more prominent.The effectiveness of battery energy management,cycle life and service safety of batteries largely depend on the accuracy of battery state information.However,the strong time-variable and nonlinear electrochemical reactions,the immeasurability of the battery state variables,and the random factors such as operating conditions,environment and measurement errors of electrical signals in practical application of EVs,present a significant challenge to online accurate estimation of battery state.Aiming at the related problems of battery modeling and state estimation,the research work in this paper has been carried out on the LiNiCoAlO₂ battery which has the highest energy density in commercial batteries.The main work can be summarized as follows:Firstly,for the modeling of lithium-ion battery,an improved model equation based on Gaussian function trinomial is proposed.Different from the traditional modeling method based on power function polynomial,the proposed method does not increase the order of polynomials to obtain higher precision.Owing to a simpler form based on Gaussian function,the proposed model function can bring higher accuracy with a smaller amount of computation.By applying this model function to different filtering algorithms including algorithms for state estimation proposed in this paper,its superiority in precision and computational efficiency as well as the universal applicability are proved.Secondly,for the online estimation of the battery state of charge?SOC?,an SOC estimation method based on an adaptive high-degree cubature Kalman filter is proposed.This approach adopts the fifth-degree spherical-radial cubature rule to select the sampling points to approximate the probability density function of the state,and introduces the adaptive law to update the state noise covariance and the measurement noise covariance at each step,which improves the SOC estimation accuracy and the ability to deal with various random factors.The SOC estimation results in two dynamic driving cycles show that the estimation error of the proposed method is less than 0.8%even in the presence of measurement errors.Thirdly,for the online estimation of the battery state of energy?SOE?,an SOE estimation method based on a spherical simplex-radial cubature Kalman filter is proposed.Compared with other methods that approximate the probability density function of the state,the spherical simplex-radial rule uses the n-simplex to calculate the spherical area,which is more efficient and accurate when dealing with nonlinear problems,thus improving the accuracy and robustness of SOE estimation.With the tests by two highly dynamic driving cycles,the proposed method demonstrates better performance than the existing methods in SOE estimation accuracy and initial error correction.Fourthly,for the online estimation of the battery state of health?SOH?,an SOH estimation method based on incremental capacity partitioned analysis with dual filter is proposed.Based on the long-term battery aging experiments for six groups,the fading characteristics of lithium-ion battery has been studied.By adopting dual filters to process the incremental capacity data,both the number of data for smoothing the incremental capacity curve and the voltage threshold for obtaining the effective value of the incremental capacity have been greatly reduces.By applying the partitioned analysis approach,the application scope of this method is extended,making online SOH estimation faster and more flexible.Various types of verification experiments show that the proposed method is effective for batteries with different aging modes,different discharge currents and different depths of charge/discharge.The experimental results show that the SOH estimation error is less than 2.5%under various conditions in the whole life cycle.On the basis of the above work,an effective improvement in the accuracy,stability and reliability of battery state estimation for EVs is achieved.The research results are easy to implement and can be extended to other fields such as portable sets,energy storage power stations and so on.