Research On SOH Estimation Method Of Power Lithium Battery Based On Feature Processing And Data Drivin

As the global demand for passenger vehicles increases year by year,there is a huge demand for fossil fuels.In order to alleviate the global energy crisis and environmental pollution problems,and to respond to the national policy of peak carbon emissions and carbon neutrality,electric vehicles have been vigorously promoted.Lithium-ion batteries are widely used as energy storage systems for electric vehicles due to their unique performance advantages,and it is crucial to achieve accurate,reliable,and robust estimation of battery SOH to help prevent battery overuse,Extend the service life and ensure the safe and stable operation of electric vehicles.Aiming at the problems of complex principles and insufficient estimation accuracy in traditional SOH estimation methods,this paper proposes a feature processing and data-driven SOH estimation method for lithiumion batteries.Firstly,the working principle and main performance parameters of lithium-ion batteries are analyzed,the internal and external factors that cause the SOH decline of the battery are discussed,and the capacity is used as the evaluation index of SOH.Analyze the parameters that characterize the battery capacity decay,and choose to extract the robust constant current charging time,constant voltage charging time,constant current charging time proportion,equal voltage rise time interval,equal current from the aging data of the battery charging stage.Five health features are used to indirectly represent the SOH in the falling time interval.Hampel filter and principal component analysis are used to remove abnormal data and reduce data dimensionality on the feature vector respectively.While improving the validity of feature data and reduce computational complexity.Secondly,a SOH estimation model based on the combination of the improved ant lion optimization algorithm and the SVR is constructed.Aiming at the problem that the hyperparameters of SVR are difficult to select,an improved ant lion optimization algorithm is proposed to optimize the parameters of SVR globally,which improves the estimation accuracy and generalization ability of SVR.Simulation experiments are carried out based on two battery data sets of NASA and CALCE,and the principal components after Hampel filtering and PCA processing are used as the input of the estimation model to realize the estimation of SOH.The results show that compared with the SVR and BP methods,IALOSVR has higher estimation accuracy and stability,the estimation error is within 2%,the mean absolute error and the root mean square error are both less than 1%,and it is used to estimate the same type of battery.The SOH also maintains high precision and has good applicability.Finally,the SOH estimation function is realized on the self-built experimental platform.The aging data of the battery is collected from the test equipment,which provides the data basis for training the estimation model.Then,the aging data of the battery is analyzed,and the health features are extracted from the charging curve and processed.The processed feature samples are used to train the IALO-SVR estimation model,and the program and parameters of the estimation model are written into the BMS,and the estimation function of SOH is realized in the BMS.The SOH estimation of the batteries Cell01 and Cell02 is carried out.The experimental results show that the method proposed in this paper has a good estimation accuracy for the SOH of the complete life cycle of batteries,with the overall error within 3% and the maximum error within 4%.It has good estimation accuracy and high practical value.