Study On Double Bayesian Collaborative Estimation Algorithm Of State And Parameters Of Ternary Li-ion Battery

The development of lithium-ion battery provides more possibilities and challenges for the automotive power field.With its high power density,excellent cycle performance and temperature performance,the ternary lithium-ion battery has become one of the important choices of the new energy vehicle power battery.The accurate estimation of the internal state of battery is closely related to the safe,reliable and efficient use of battery.As one of the most important battery States,State-of-Charge（SoC）can not only provide the driver with battery endurance information,but also provide important basis for the safe operation of battery.In this paper,for the collaborative estimation of state and parameters of ternary lithium-ion batteries,based on the Bayesian filtering theory[including extended Kalman filter（EKF）,unscented Kalman filter（UKF）and particle filter,PF)]proposed several battery SoC and model parameter collaborative estimation algorithms,which improved the estimation accuracy of SoC by real-time estimating model parameters.This paper also proposed an improved collaborative estimation algorithm to improve the accuracy of SoC estimation and the convergence performance.The main work includes:（1）Based on the introduction of the working principle of the ternary lithium-ion battery,a charge-discharge experiment is designed.The discharge tests of two kinds of ternary lithium-ion batteries with different ternary materials are carried out respectively.The results tell:（ⅰ）the discharge energy efficiencyη_E of the two kinds of batteries is lower than the discharge capacity efficiencyη_Q under the same discharge rate;（ⅱ）the discharge capacity Q_d or energy E_d of the two kinds of batteries under different discharge rate are lower than that of the standard charge mode,the Q_d and E_d decrease with the increase of discharge rate,so do theη_Q andη_E.（ⅲ）The decrease speed ofη_Eandη_Q of Ni-Co-Mn Li-ion battery is slower than that of Ni-Co-Al Li-ion battery.（2）Two new polarization parameters,attenuation coefficientα_i and steady-state coefficientβ_i,are proposed,which replace the polarization resistance and polarization capacitance in the original model,simplify the equation,and laid foundation for estimating the parameters by using linear Kalman filter.Hybrid Pulse Power Pharacterization（HPPC）was used to identify the polarization parameters of Ni-Co-Mn ternary lithium-ion battery in different SoC.The results show that the output voltage accuracy of the second-order model is higher than that of the first-order model under different current conditions.In addition,a recursive least square method with forgetting factor is designed to identify the polarization parameters on-line under different conditions.The results show that,compared with the off-line parameter identification method,the results of the on-line method can improve the accuracy of the model output voltage.However,the algorithm takes the open circuit voltage(u_oc)as one of the identified parameters and ignores the coupling relationship between u_oc and current,which makes the estimation results out of the actual range,so the method can not be used to improve the estimation accuracy of SoC.（3）In this study,three different Bayesian filter combinations of collaborative estimation algorithms are designed,and the SoC and parameter estimation of the algorithm are simulated and analyzed under different conditions.The results show that:（ⅰ）the collaborative estimation algorithm has higher accuracy of SoC and parameter estimation than the single filter algorithm.（ⅱ）AEKF-KF and AUKF-KF algorithms are proposed,the results show approximate estimation accuracy due to low nonlinear degree of u_oc-SoC relationship and small noise covariance,the estimation accuracy of both are lower than that of PF-KF.（ⅲ）The error analysis shows that it is not reliable to use the predicted voltage error of the algorithm as the standard for evaluating the accuracy of parameter estimation,however the output voltage error of the open-loop model can reflect the real situation.An improved IAEKF-KF algorithm is also proposed in this paper.First,SoC is treated as the only state variable,which eliminates the strong correlation between state variable and polarization parameter,so as to improve the estimation accuracy;Second,the problem of poor convergence of the original algorithm is overcome by adding the trigger mechanism of parameter filter.The simulation results of IAEKF-KF and AEKF-KF under different conditions show that the improved collaborative estimation algorithm has higher SoC and parameter estimation accuracy and better convergence performance.（4）Based on AD7606 sampling chip and TMS320F28335 digital signal processor,a collaborative estimation platform of SoC and parameters for ternary Li-ion battery is designed.The system can estimate the state and polarization parameters in real time and draw the curve in PC.Two kinds of working conditions are designed to discharge Ni-Co-Al ternary batteries.The performance and data results of the collaborative estimation platform equipped with AEKF and IAEKF-KF algorithms are compared and analyzed.The experimental results show that the platform works stably and reliably,and the estimation results meet the simulation expectations.