Research On SOC Estimation Algorithm And Equalization Strategy For Battery Management System Of Electric Vehicles

The development of electric vehicle has been paid more and more attention by academia and industry,but the safety accidents of electric vehicles are frequent.According to statistics,battery management system(BMS)cause the fire cases of electric car,which is accounted for more than 60%of all accidents,so the BMS security greatly restricts the rapid development of the industry.At the same time,the cost of battery box and battery management system is for a third and more of the total cost of an electric vehicle,and the replacement and repairing cost is also huge,so a stable and reliable and long life BMS system for an electric car is very important.The two main functions of BMS are accurate estimation of SOC and effective balance between batteries.They play an irreplaceable role in protecting battery safety,extending battery life and maximizing battery capacity.For new energy vehicles for the demand of the battery SOC estimation and equalization,this paper established a kind of second-order RC equivalent circuit model considering bidirectional internal resistance,used the variable forgetting factor least square method to realize the model parameters online identification,put forward the modified covariance extended kalman filtering algorithm for SOC estimation,design a any cell to any cell balance topology structure with the consideration of zero current switch and put forward the dynamic equalization variable threshold algorithm,developed a multivariable fuzzy fusion equalization control strategy.Finally,the above work is verified through the experiment and simulation.The battery model is the premise of SOC estimation and equilibrium control.In order to realize these functions,a model describing the nonlinear characteristics of the battery must be established.The equivalent circuit model use ideal circuit elements(resistor,capacitor,controllable voltage source,etc.)to simulate the dynamic operating characteristics of lithium battery.Based on these characteristics of the battery and the charging and discharging rules of the battery,a second-order RC equivalent circuit model considering the bidirectional internal resistance is established.In order to accurately identify battery parameters,this paper achieves online identification of model parameters by using variable forgetting factor recursive least squares(MFF-RLS),and then these parameters are used in the next step of SOC estimation.Finally,through constant current pulse experiments,the output voltage error of the established model is controlled within 0.3v,and it is found that the maximum error only occurs in the high and low states of the battery,i.e.,SOC > 0.9 and SOC < 0.1,and at other moments,such as 0.1 < SOC < 0.9,the estimated error value of the model is very small.Since the effective operating range of battery is 0.1-0.9,the established model and identified parameters can meet the requirements of electric vehicles.Accurate SOC can help drivers choose the right driving behavior.When the traditional extended kalman filter(EKF)is applied to the battery SOC estimation,the covariance cannot gradually reduce to zero in the recursive process,which affects the accuracy of the SOC estimation.In view of this situation,a modified covariance extended kalman filter(MVEKF)algorithm is proposed in this paper,which uses the modified state estimation to update the process gain,recalculates the covariance in the iterative process,and applies the new process gain value to the next state estimation to ensure the stability of the filter.At last,the method is verified by the experiments of constant current discharge,constant pulse discharge and dynamic stress test(DST).The results show that MVEKF filter algorithm is better than EKF algorithm,especially in the complex charging and discharging conditions,MVEKF filter algorithm has more obvious advantages.Under the condition of DST,EKF has a large deviation and is unstable especially in the charging process,while MVEKF algorithm can steadily estimate SOC with high accuracy and strong robustness.This proposed algorthm is suitable for the complex and changeable working conditions of electric vehicles.The equalization topology is the platform for the equalization control of the equalization management system(EMS).The commonly used balanced circuit topology often requires a large number of switches and complex control algorithms,and it is difficult to meet the needs of efficient and safe balanced power batteries of electric vehicles due to its low efficiency,large volume,high cost and low reliability.In this paper,a kind of any cell to any cell equalization topology structure considering zero current is proposed.At the same time,setting a fixed value of variable threshold leads to false equalization problems in the whole cycle.This paper puts forward a kind of dynamic threshold algorithm.Finally through constant current pulse simulation,the proposed balanced structure and algorithm is verified.The result shows that the proposed equalization structure realizes zero current switch function,reduces the stress in the circuit switch and prolongs the service life of circuit components.The proposed algorithm can reduce the number of open and close switch,avoiding the repetition of equalization.Excellent equalization control algorithm is an important guarantee to improve the equalization speed.In the actual work,the energy demand of electric vehicle power lithium battery is changeable.Considering a single equalization variable(SOC,voltage)to carry out the equalization control and deside the equilibrium current,it cannot meet the real demand and may lead to misbalance and overbalance.In this paper,a multivariable fusion fuzzy equalization strategy considering voltage and SOC is developed.The fuzzy controller can determine the equalization current by selecting the ratio of voltage and SOC according to the level of current battery box SOC and the charge and discharge current.Finally the algorithm is verified by comparison with simulink simulation.The results show that based on the actual state,the proposed algorithm adjusts the equalization current,make the circuit work near the optimal working point,and reduce the time of equilibrium.At the same time,the equalization current is stable,no big ups and downs in the process,which protects circuit components and extends the service life of the equalization circuit.