Research Of Battery Modelused In Evs During Whole Life Cycle

As a response to environmental and energy issues, the electric carindustry has developed rapidly in recent years. Battery managementtechnology is one of the core technologies. Technologies of state of charge(SOC) and state of health (SOH) estimation are the most important.A mainstream solution to SOC estimation is based on battery modelingand Kalman filter. Most studies focus on the cell modeling and ignore toimprove Kalman filter algorithm itself, and thus the accuracy of SOCestimation falls. The subsequent battery management and electric vehicles(EVs) will not perform well. The current solution to SOH estimation is basedon off-line test and it is impossible to build an effective management forbatteries during life cycle. It is not conducive to extend battery life. Batterycharging technology is also an important part of the battery managementtechnology. The core is a design of DC-DC converter or rectifier. Controlmethod should be optimized, in addition to improving main circuit topology.In recent years, scholars put forward, the charger control methods should becombined with the identification of the battery parameters (i.e. SOC andSOH).Based on the in-depth analysis of the Kalman filter, this paper presents adeformation algorithm, which is essentially the same to Kalman filter,without the initial value and mixed with enumeration algorithm. Byintroducing maximum likelihood estimation and genetic algorithm, thecomputing efficiency of the algorithm is improved. The improved algorithmeliminates errors caused by improper initial value, linearization of nonlinearsystems, noise whose statistical properties are unknown, systematic errors ofmodeling and other factors. As a result, it achieves a more precise battery SOC and SOH estimation. Also, the improved algorithm can be used inbattery parameter identification. The identification result can be used in theaccurate linearization control for the boost circuit used as battery charger. Forboost charger requirements, this paper presents a new accurate linearizationalgorithm to eliminate fluctuations caused by switching between constantvoltage (CV) control mode and constant current (CC) control mode.The main research results are as follows:1) To solve the slow convergence caused by improper initial value andrecursive format of Kalman filter algorithm, an enumeration algorithmwithout recursive is proposed based on the principle and the derivation of thetraditional algorithm. Theoretical analysis shows that the result of theenumeration algorithm and that of traditional algorithm after convergence isexactly the same, but the new algorithm needs a lot of computing space, andthe computing efficiency is low.2) To solve the steady-state error in traditional Kalman algorithm causedby ignoring high-order terms during linearization, noise whose statisticalproperties are unknown, systematic errors of modeling and other factors, anew algorithm is proposed by the introduction of maximum likelihoodestimation and genetic algorithm, based on the enumeration algorithm. Thesimulation results show that the new algorithm meets the demand of batterymanagement during life cycle, and there is almost no error in the result of thenew algorithm, while there is about10%error in the result of the traditionalalgorithm.3) To solve the load parameter estimation problem in the exactlinearization control of boost charger, the improved algorithm is used andgets better battery parameter identification. Simulation results show that theimproved algorithm achieves the identification of the battery parameters witha very small error, when there is a10%random noise in the system and10%systematic errors in part of the model parameters.4) To solve the fluctuations caused by switching between CV controlmode and CC control mode in boost charger, an improved exact linearizationalgorithm is proposed, where CV mode and CC mode share the same coordinate transformation. In order to improve the response speed of thecharger, pole-zero configurations are used. The simulation results show thatthere are no fluctuations. The response speed is fast and meets therequirement of10Hz pulse charging.5) Battery module, the traditional Kalman filter algorithm module usedin SOC and SOH estimation, the improved traditional Kalman filteralgorithm module, main circuit module, exact linearization control algorithmmodule and other modules are built in MATLAB/SIMULINK environment.The results showed that the new estimation algorithm is more accurate andthe new control algorithm performs better.