Lithium-ion Batteries’ Charge-discharge Method Improvement And State Of Charge Estimation

Lithium-ion batteries will be popularized and applied in in many fields such as energy storage and power.in response to the national " carbon peaking and carbon neutrality " strategy to alleviate the energy and environmental crisis.Due to the harsh requirements of water and oxygen environment and the special active properties of battery materials,the consistency of battery status is an important guarantee for the safe operation of batteries after packing.The estimation of battery online status information and retired batteries offline status information have become an important defense line for battery safety control.Among them,the information of battery state of charge is the basis of all battery state research,and its accurate estimation has become a hot research issue.Based on the preliminary preparation work of battery state estimation,this paper mainly completed the following aspects of research:Samsung INR-18650-25 R cylindrical battery was taken as the research object in this paper.Firstly,the buckle battery was assembled by Lithium nickel-cobalt aluminate commercial battery material corresponding to the research object.The influence of the liquid-phase diffusion process on the relaxation voltage variation of the battery after discharge was investigated by adjusting the concentration of lithium ions in the buckle battery electrolyte and the thickness of the diaphragm,and the reasons for relaxation of the battery were analyzed with the research of diffusion polarization mechanism of lithium-ion battery.According to the idea of reducing the polarization of charge and discharge and speeding up the equilibrium process,the charge and discharge method before the battery characteristics test is improved based on standard method.The advantages of the improved method in the characteristic testing were verified by the open-circuit voltage and electrochemical impedance spectroscopy test.The improved charge-discharge method provides conditions and ideas for shortening the testing period of battery characteristics for battery state estimating and rapid separation of retired batteries.The characteristics,advantages and disadvantages of five common battery equivalent circuit models are compared secondarily.According to the working principle of the battery,the second-order resistance-capacitance battery model was selected as the state estimation model in this paper with the mind of moderate computational load and accuracy suitable for the current requirement of state of charge estimating,and the mechanism of battery chemical process corresponding to each component in the model was explained.Aiming at the problems of poor robustness of traditional recursive least squares algorithm in battery parameter identification,weak adaptability of identification ability in different working conditions,slow convergence of identification algorithm and so on,the forgetting factor in recursive least squares algorithm was improved.By comparing the parameter identification effects of recursive least square method with different forgetting factor terms under different working conditions,the robustness and accuracy of the algorithm under large current stress are greatly improved.The advantages of the improved charge-discharge method in parameter identification are verified by comparing the influence of open circuit voltage acquisition method on the accuracy of parameter identification.Finally,the parameter identification algorithm is joined into the basic extended Kalman filter algorithm for the battery online state of charge estimation.The improved charge-discharge method greatly optimizes the estimation accuracy of the state of charge,especially in the case of low terminal electric quantity.the improved charge-discharge method greatly improved the estimation.Finally,the parameter identification algorithm is joined into the basic extended Kalman filter algorithm for the battery charging state online estimation.The improved chargedischarge method greatly improves the estimation accuracy of the state of charge was verified,especially when the terminal electric quantity is low.The differences in convergence speed,accuracy and iteration speed of the forgetting factor term on the estimation of the state of charge by the joint algorithm were compared,and the ability of the improved algorithm to estimate the state of charge under the actual complex working conditions was verified by the experimental data of different working conditions.The results show that the adaptive forgetting factor recursive least squareextended Kalman filter algorithm can estimate the battery state of charge in different working conditions and initial value of state of charge robust and accurately.