Research On SOH Estimation Method Of Lithium Battery Pack Based On Electrochemical Mechanism And Data-driven

With the change of power structure and energy market under the goals of “Carbon Neutral”and “Carbon Peak”,the demand for new electrochemical energy storage equipment,represented by lithium-ion batteries,has soared in the global energy industry and is widely used in energy storage power stations and other new power systems.For energy storage lithium batteries,the state of health(SOH)is a key indicator used by the battery management system(BMS)to optimize energy management strategies and provide economic reference.In this paper,the long-term SOH estimation of energy storage lithium battery pack is studied as follows.(1)In this thesis,the motion mechanism of lithium ion in energy storage lithium battery is analyzed,which provides a theoretical basis for electrochemical modeling.By exploring the law of liquid-phase potential in the reaction process of energy storage lithium battery,and considering the influence of solid electrolyte interphase(SEI)film on the single particle model,an extended single particle electrochemical model is built.Based on the series characteristics of energy storage lithium battery pack,a multi-cell model of energy storage lithium battery pack is established.(2)This thesis designed two kinds of extraction methods of different health indicators to provide input basis for SOH estimation model.Aiming at the problem that the results of particle swarm optimization algorithm are easy to fall into local optimum in practical application,the improved particle swarm optimization algorithm is constructed according to the cooperative strategy and competitive strategy,and the cooperation competition particle swarm optimization(CCPSO)algorithm is proposed to achieve accurate identification of electrochemical parameters.Through the quantitative analysis of the characteristic points of the incremental capacity(IC)curve and differential voltage(DV)curve,the IC-DV method is developed to realize the quantitative study of degradation mode.(3)Aiming at the problem of insufficient accuracy of back propagation(BP)neural network estimation results in experimental tests,the initial parameters of the traditional BP neural network are optimized through the nonlinear temperature decreasing step size simulated annealing strategy,and a nonlinear temperature decreasing step size simulated annealing-back propagation(NSA-BP)model is constructed.Combined with multi-dimensional health indicators and NSA-BP data-driven model,the SOH of lithium battery pack can be effectively estimated.(4)To study the accuracy of the SOH estimation method of lithium battery pack proposed in this thesis,the battery model and SOH estimation results are verified by clustering different typical energy storage conditions.The maximum voltage error and mean absolute voltage error of the multi-cell model under two different energy storage conditions are 0.0187 V and0.0364 V,respectively.The root mean square error(RMSE)and mean absolute error(MAE)of the SOH estimation method for energy storage lithium battery pack based on multi-dimensional health indicators and NSA-BP model under low-rate constant current energy storage conditions are 0.0059 and 0.0048,respectively.And the corresponding values under variable-rate energy storage condition are 0.0073 and 0.0064,respectively.Through the analysis of the experimental verification results,the long-term SOH estimation method of the energy storage lithium battery pack developed in this thesis can effectively meet the SOH estimation requirements of the energy storage lithium battery in the new energy storage environment.And the results can provide a strong basis of the operation monitoring and health management of the energy storage lithium battery.