Research On Ekf Lithium Battery SOC Combined With Fuzzy Neural Network

As the world’s energy structure is undergoing a shift from relying on traditional fossil energy to pursuing clean,high-efficiency and low-carbon energy,clean energy has quickly become the backbone of my country’s acceleration of supply-side structural reforms in the energy sector.Therefore,new energy vehicles have successfully entered the mass market,and the development and application of power battery technology and battery management systems have gradually attracted the attention of enterprises and users.In fact,the key to solving the above problems lies in the reasonable estimation of the state of charge(SOC)of lithium batteries.Today’s battery SOC estimation model,through continuous innovation and development,has gradually formed an intelligent battery power calculation method incorporating information analysis technology,such as neural network algorithm,Kalman filter method,and fuzzy logic control algorithm.However,intelligent algorithms still have defects such as complex implementation,high requirements for model accuracy,poor robustness,and poor applicability,and may cause unpredictable estimation errors due to data uncertainty.Based on the above analysis,based on the establishment of the lithium battery model,this article uses fuzzy neural network combined with the extended Kalman filter algorithm to improve the accuracy and rapidity of battery power estimation.The main research contents and conclusions of this paper are as follows:1.First,according to the principle analysis and performance screening of the battery,the lithium battery is selected as the experimental research object,the battery experiment platform is built,the battery characteristic experiment is completed,and a solid data basis is provided for the verification of battery modeling and SOC estimation results.2.The establishment of the model is based on battery experimental data and characteristic analysis,and the Thevenin second-order RC equivalent circuit model is selected.Using HPPC experimental data to identify the circuit model offline parameters,and then further screen out the recursive least squares FFRLS with forgetting factor to realize the online parameter identification of the circuit model.It is hoped that the real-time update of battery model parameters can be realized by this.3.Taking the extended Kalman filter algorithm as the research object,the pros and cons of the algorithm and the specific calculation process are described in detail,and the accuracy of lithium battery SOC estimation is evaluated by FFRLS+EKF.The result shows that the error of the joint algorithm is still about 6%.4.In view of the problems exposed by the extended Kalman filter algorithm,the EKF and fuzzy neural network algorithm are combined to optimize and improve,and then the computer simulation software is used to analyze and study the SOC of the lithium battery.The results show that the improved EKF algorithm has strong noise Inhibition ability,its robustness is strong,the error result is small.