Study On State Estimation Method Of Lithium-ion Battery Pack For Electric Vehicle

With the rapid development of the global automobile industry,the shortage of energy and the environmental pollution caused by a large number of fuel vehicles exhausts are becoming increasingly serious.Electric vehicles use vehicle-mounted battery systems as power sources and motor-driven vehicles.They have the advantages of energy conservation,waste reduction,and low noise,effectively solving the two major problems facing the world,and have a very broad prospect for development.Vehicle to electric development,vehicle-mounted power batteries are the main core technology of electric vehicles,providing all the necessary power and energy for the vehicle.In order to maximize the battery life and ensure the safe driving of the vehicle,it is very important for the effective management of the vehicle.Among them,the state-of-charge(SOC)of the battery is one of the important parameters in the battery management system(BMS).How to quickly and accurately estimate the SOC is the focus and difficulty of current research.In this paper,the estimation algorithm of SOC in electric vehicle lithium batteryis studied,and the hardware and software of lithium battery pack are designed.(1)Firstly,the working principle of the battery is introduced,and the characteristics of the battery are analyzed.At present,the lithium ion battery has the advantages of small self-discharge,good cycle characteristics,rapid charge and discharge,and high energy efficiency.This paper selects the lithium battery as the research object.The common battery equivalent circuit models are analyzed and compared,and the PNGV equivalent circuit model is used to estimate the SOC of the battery.(2)Secondly,the traditional SOC estimation method based on Extended Kalman Filter(EKF)has the following drawbacks: it needs to establish an accurate battery model,and the system noise must obey Gaussian white noise.This paper improves the extended Kalman filter algorithm and proposes a SOC joint estimation method based on the model error EKF-HIF algorithm.The Back Propagation Neural Network(BPNN)was used to predict the battery model error,and the Extended Kalman Filter(EKF)and H Infinity Filter(HIF)algorithms were deduced,and the relationship between the two algorithms was analyzed.According to the model error,different algorithms are selected.When the model error is small,the EKF algorithm is used to estimate the state of charge.When the model error is large,the HIF algorithm is used to estimate thestate of charge.(3)In this paper,HPPC testing experiments were performed on single lithium iron phosphate battery.The parameters of the battery model were identified based on the test data,and the functional relationship between each parameter and SOC was obtained by the method of least squares fitting.Then the hardware and software of the lithium battery pack were designed.The main design modules include DSP2812 main control circuit,current and other parameters sampling circuit,equalization control circuit,protection mechanism and alarm circuit,data communication circuit and so on.(4)Finally,a Simulink simulation platform is built to verify the accuracy of the established battery model and the validity of the proposed battery state of charge joint estimation method.The battery is discharged under constant current,pulsed and variable current conditions.The experimental results show that the PNGV model can reflect the dynamic state of the lithium battery;The SOC joint estimation method based on EKF-HIF is compared with the traditional SOC estimation method based on EKF.It is verified by simulation that the joint algorithm can effectively eliminate the SOC estimation error introduced by the large model error and measurement noise,indicating that the proposed method has high accuracy and good convergence.