Study On Parameter Identification Of The Equivalent Circuit Model In Lithium-ion Battery In Electric Car

The lithium-ion battery technology is one of the most important factors to developelectric vehicles and gain market profits, of which difficulty is the battery state estimation.An accurate battery modeling is the basic solution to the problem.Firstly, the author summarized the battery modeling including battery models,identification methods, and problems. The equivalent circuits were applied to model theelectrical performance of the lithium-ion batteries. In this thesis, the first order RC model-1RC and the second order RC model-2RC were focused on.Secondly, the experiment for the18650cells was set up. The electrical performance ofeach cell was analyzed with the definition of the energy efficiency, the capacity efficiency,the overcharge ratio, and the discharge undervoltage ratio.Thirdly, with the selected cells, the recursive least squares (RLS) was used to estimateparameters of both the1RC and2RC models. The simulation of the battery terminal voltageshowed that the2RC model had better accuracy than the1RC. Nevertheless, a novel onlinesequential algorithm called the multi-stage RLS-MSRLS was proposed to estimateparameters of the multi-order RC models for the lithium-ion batteries. Compared with theRLS, the MSRLS had the better real-time performance and similar accuracy by experimentand simulation data.Furthermore, the author adopted the Levenberg-Marquardt nonlinear least squares toidentify the parameters of the1RC and2RC models. Experiment and simulation showed thatit had better model accuracy than the other proposed LS with types of loads such as theconstant, the ladder, and the dynamic stress test-DST current profiles.Finally, the author summarized the study on the parameter estimation methods for themulti-order RC battery models, and prospected the further work in this field in the future.