Research On High-precision Modeling And Multi-state Estimation Of Lithium-ion Power Batteries

The development of electric vehicles is the key way to solve the energy crisis and environmental pollution problems,and it is also the only way to realize the transformation of role from a big country to a powerful country in automobile industry for China.According to the latest data issued by China Association of Automobile Manufacturers,the annual sales volume of electric vehicles in China will be 1,367 million in 2020,which ranks the first in the world.Due to the unique advantages in power density,energy density,cycle life and self-discharge rate,lithium-ion power batteries are widely used in electric vehicles.The management of power batteries plays an important role in maximizing the utilization of the available capacity and energy,and improving the overall vehicle performance.Improper battery management will not only cause over-charge and over-discharge abuses,but also lead to premature battery failures and dangerous accidents such as fire and explosion,which directly threaten the safety of lives and property,and cause irreparable losses.Therefore,the researches of theory and technology that related to battery management have become the focus of concern in both academia and industry.Scholars at home and abroad have carried out a lot of fruitful work in power battery modeling and state estimation.However,until now,some key problems have not been solved substantially:1)The important effect of the identification data on the accuracy of parameter identification is ignored,what is particularly important is that the existing excitation signal design methods cannot effectively deal with the nonlinear characteristics of the lithium-ion battery;2)Due to unaware of the fact that the lithium-ion battery is essentially a stiff system,the traditional method is adopted to estimate the model parameter,which results in that low accuracy of parameter identification and poor modeling performance occur;3)The factors that affect the accuracy of co-estimation of model parameters and state of charge(SOC)are not taken into consideration,two of which are the sensor measurement error and the electromagnetic interference that generated by power electronic converter,motor controller,motor and other high-power devices;4)The multi-state estimation method is highly coupled among the estimators,which results in that the parameter tuning is complex and difficult,and the stability is also difficult to guarantee.In view of the above problems,the main work and innovation of this thesis are as follows:1)Lithium-ion battery has a strong nonlinear time-varying characteristic.The traditional excitation signal design method is easy to stimulate the nonlinear response of the battery,which leads to the low modeling accuracy of the commonly used equivalent circuit model.Therefore,a novel excitation signal design method called inverse repeat binary sequence is proposed.The antisymmetric characteristic of the signal eliminates the response of the even order nonlinear part of the battery system effectively,and the modeling performance of the proposed method is compared to that of the other three excitation signal design methods under Urban Dynamometer Driving Schedule(UDDS)at different temperatures.The results show that the mean absolute error(MAE)and root mean square error(RMSE)of the proposed method are within 6.86 mV and 8.61 mV respectively,and the modeling accuracy is significantly higher than other methods,which validates the effectiveness of the proposed method.2)Aiming at the problem that low accuracy of parameter estimation and poor modeling performance can occur when the conventional recursive least squares(RLS)method is used to estimate model parameters of the battery stiff system,the decentralized least squares method is proposed to estimate the model parameters of dual polarization model.According to the information that different time scales of the battery can be separated,the battery model is divided into two sub-models for identification,and model parameters can be estimated accurately because of the elimination of the mutual interference between the parameters.Experimental results show that compared with RLS,the proposed method can reduce MAE and RMSE by 50.0%and 46.43%under UDDS test respectively,which verifies the effectiveness of the proposed method.3)Aiming at the low accuracy of SOC estimation caused by the complex and harsh electromagnetic environment of electric vehicles and the measurement error of sensors,co-estimation of model parameters and SOC of lithium-ion batteries with noise immunity is proposed,which is based on the analysis of the bias of the traditional method.The proposed method is mainly composed of two parts:parameter identification based on restricted total least squares and SOC estimation based on unscented Kalman filter.The performance of the proposed method is evaluated by simulation and a series of experiments at different temperatures and working conditions.The results show that the MAE of SOC estimation is less than 1.19%,which verifies the effectiveness of the proposed method.4)Aiming at the problem of high-coupling between estimators in traditional multi-state estimation methods,a reduced-coupling co-estimation method of SOC and state of health(SOH)is proposed.The proposed method owns strong self-correction abilities for inaccurate initial values,and is mainly implemented by using battery model and two migration factors.It is composed of two parts:SOC estimation based on adaptive coevolutionary particle swarm optimization algorithm(ACPSO)and SOH estimation based on restricted total least squares.The performance of the proposed method is evaluated by simulation and experiment.The results show that MAEs of SOC and capacity of the proposed method under UDDS test are within 1.58%and 0.136 Ah respectively,which verifies the effectiveness of the proposed method.To sum up,four research contents for the lithium-ion power batteries are carried out in this thesis,which include excitation signal design,parameter identification method,SOC estimation and multi-state estimation.Everything in good order and well arranged,and the innovative research results have been achieved that lays a solid theoretical foundation to ensure the safe,reliable and efficient operation of power batteries.