Study On Modeling And SOC Estimation Of Ni-Mh Battery For Hybrid Electric Vehicles

With the increasing scarcity of oil resources in the world,the market scale of new energy vehicles is increasing.However,the development of pure electric vehicles has been limited by the performance of power batteries.Hybrid electric vehicles occupy a position in the market because of the characteristics of both fuel vehicles and pure electric vehicles.Power Ni-MH battery is then used in hybrid vehicles with its outstanding specific power and excellent advantages of high current charge and discharge performance.In this paper,a Ni-MH battery module with six single batteries in series is taken as the research object,and its modeling and SOC estimation are studied.The main work is as follows.Firstly,the working principle and main performance parameters of Ni-MH battery are introduced,and some important working characteristics are tested.The factors affecting the performance of Ni-MH battery were analyzed.The experimental data provided data for the identification of battery model parameters in the following chapters.Secondly,the second-order RC network model is selected by comparing several equivalent models of battery.Considering that the open-circuit voltage hysteresis characteristic of Ni-MH battery is very obvious in the experiment,the hysteresis characteristic factor is added into the model.The unknown parameters in the model are identified one by one according to the experimental data of battery.Compared with the model without considering hysteresis characteristics,the simulation results in Simulink show that the model has better adaptability to the actual working conditions and the addition of hysteresis voltage module significantly improves the accuracy of the model.Last the recursive process of EKF(Extended Kalman Filter)algorithm and UKF(Unscented Kalman Filter)algorithm is introduced.On the basis of describing the state equation and the observation equation of the battery model system,the two algorithms are applied to estimate the SOC value.The simulation results show that both algorithms have the ability to correct the initial SOC errors,but the UKF algorithm is more accurate and stable,and the root mean square error of the estimation error is less than 5%.The state equation and observation equation of the algorithm are based on the equivalent model mentioned above,so the accuracy of the model is also verified.