Lithium Battery State Of Charge And State Of Health Prediction Based On Fuzzy Kalman Filtering

With the rapid increase of energy consumption around the world and China,the air pollution problem and ecological friendly development are increasingly attracting the attention of many people and governments.Electric vehicles have the large potential in reducing environment pollution and prevention of energy crisis.Although the electric vehicles industry is supported by many governments,there are still some problems impending the electric vehicles development.Presently the battery as a key component of electric vehicles also is its bottleneck.Power battery in electric vehicles controlled by battery management system(BMS).In order to ensure the safe and efficient operation of electric vehicle,prevent deep discharge or overcharge of the battery,accurate estimate residual mileage,extend the lifetime,prevent progressive permanent damage to the battery and maximize battery performance,the BMS must have an accurate battery state of charge value.In addition,to improve the reliability of operation,and to warn the driver about the future replacement of the battery,the BMS needs the battery state of heath value.In the actual application,the battery working conditions,its temperature,aging and other factors take nonlinearities into state prediction task and make the accurate state prediction difficult.First,the lithium batteries working principles,structure and main characteristic are analyzed.Then,taking into account the lithium battery chemical peculiarity,the battery second-order network equivalent circuit model is established based on the comparative analysis of existing battery models.The equivalent circuit model parameters are estimated by Levenberg-Marquardt least square error optimization algorithm under the different ambient temperatures.The model accuracy is validated in the discharge pulse test.The shortcomings of the existing state of charge(SOC)estimation methods are identified.Aiming on overcome these shortcomings,this dissertation presents the adaptive extended Kalman filter(AEKF)algorithm to predict the battery SOC.This method helps to overcome disadvantages of the Coulomb counting,Kalman filter and the neural network estimation method Application of the extended Kalman filter(EKF)removes the need of prior known initial SOC.Although EKF can provide good estimation results,this method is not suitable for non-Gaussian noise and highly nonlinear systems because of large cumulative estimation error.Therefore,application of the fuzzy logic adaptive approach helps to eliminate the measurement noise estimation.Applying the AEKF algorithm the value of the measurement noise covariance is adaptively adjusted in the estimation process.In the end,the prediction accuracy of Coulomb counting,EKF and AEKF methods are compared in the Urban Dynamometer Driving Schedule(UDDS)test.The simulation results show,that the AEKF method reaches the best performance.After analyze the battery state of health(SOH)definition and influencing factors,the battery capacity is chosen as a reflecting battery SOH measure.The battery capacity is estimated by designed offline event-based Kalman filter through the discharge-charge cycle current integration.The representing city driving condition UDDS test is used for continuously simulation of degrading battery.The results show,the event-based Kalman filter capacity estimation reach the accuracy requirements and can be used in SOH prediction.