| With the rapid consumption of petroleum energy in the world and the increasingly serious environmental pollution,electric vehicles,due to their low energy consumption,zero emissions,no noise and many other advantages,have become the best solution to the problems of noise,exhaust gas,smoke,oil pollution and other problems caused by urban traffic pollution,and also an important development direction of the automobile industry in recent years.Lithium battery has become an important energy storage device for mobile equipment such as electric vehicles due to its high energy density,recyclability and other advantages.It is also the important energy source for electric equipment such as pure electric vehicles.Therefore,accurate evaluation and management of the operating state of lithium battery(battery Pack)is the key foundation and important basis for safe and reliable operation of electric vehicles.However,the complexity of lithium battery usage scenarios,the sensitivity to temperature,and the nonlinearity determined by the battery principle pose certain challenges to the battery status performance evaluation,prediction,and balance management.Therefore,the relevant research on lithium battery usage scenarios has important application value and significance for improving the performance of electric vehicles and promoting the development of new energy vehicles in China.In this paper,several issues such as state of charge(SOC)estimation,battery equalization,and remain useful life(RUL)evaluation of lithium batteries are studied.The specific research contents are as follows:(1)In order to solve the inaccurate estimation of the state of charge(SOC)of lithium batteries,the working mechanism of lithium batteries was studied,and a second-order RC equivalent circuit model was established based on the physical characteristics of the batteries;The discrete mathematical model equation of the battery is derived and solved;The model parameters of the battery were identified;The SOCOCV curve based on multipoint symmetric sampling method and fusion method is proposed,and the SOC-OCV curve is fitted by polynomial.These works will serve as the basis for subsequent SOC estimation and battery equalization.(2)In order to solve the problem of decreasing the accuracy of state of charge estimation caused by the time-varying noise of lithium battery in operation,a method of state of charge estimation based on variable Bayesian unscented Kalman(VBUKF)and variable Bayesian square root volume Kalman filter(VBSRCKF)was proposed.In order to solve problem of outliers in state of charge estimation,a SOC estimation method based on extended student’s-T(ESTF),extended robust student’s-T(ERSTF)and KLD extended student’s-T Filter(KLDESTF)are proposed.The state of charge estimation methods based on VBUKF,VBSRCKF are compared with UKFand SRCKF,and these four methods are used to verify the effect of the constructed SOC-OCV mapping curve;the state of charge estimation methods of lithium battery based on ESTF,ERSTF,KLDESTF are compared with EKF and AEKF.Both simulation and experimental results show that the proposed state of charge estimation method can effectively improve the estimation effect.(3)As a result of the difficulty of controlling the equalizing current and equalizing speed during battery equalization,a 4-string battery pack active equalization scheme is proposed that uses SOC as the equalizing variable and two controllable flyback converters as the transfer circuit.Based on SIMULINK,a second-order equivalent circuit model of 2000 m Ah battery is constructed,and a circuit equalization strategy and equalization circuit are designed for equalization simulation.According to the experimental results,the active equalization scheme has a good equalization effect.In addition,a passive equalization experiment is conducted,which also meets the equalization requirements.(4)A battery health indicator based on charge/discharge time difference sequence is proposed,and the remain useful life is predicted and evaluated by using particle filter and double exponential empirical model.The prediction results show that the health indicator can well represent the health of lithium batteries.On the basis of the constructed battery health indicator,a fusion prediction method is proposed.This method decomposes the feature sequence data into trend items and non trend items based on CEEMDAN algorithm,and then uses ARIMA and LSSVM to predict the trend items and non-trend items respectively,and combines the prediction results of both to comprehensively evaluate the remaining life of the battery.The prediction results show that when the battery life curve reversely increases due to energy regeneration and other reasons,or lack of full life-cycle reference model,the scheme can also predict and evaluate the remaining useful life of batteries. |
