Investigation on Simulation and Parameter Estimation for Lithium-ion Batteries
|School||Harbin Institute of Technology|
|Course||Power Electronics and Power Drives|
|Keywords||Lithium-ion batteries P2D model SP model GA parameter estimation|
With the widespread applications of lithium-ion batteries, a series of problemsabout batteries have appeared, which mainly reflect in its safety and life. The life andsafety problems have already been the most important factor that restricts thedevelopment of batteries. Therefore, the research on the mechanism modeling andparameter estimation of batteries has significant influence on monitor of batteryhealth state and prediction of battery life.Single particle model and pseudo-two-dimensional model are elaborated in thispaper. The discretization of PDEs in P2D model is implemented by FDM in bothelectrode and radial direction, which reduces the system of PDEs to a system ofdifferential algebraic equations of index1with time as the only independent variable.In order to reduce the number of DAEs in simulation, the two-parameter and three-parameter parabolic approximation equations are studied to simplify the particle solidphase diffusion process. And under different rate constant current and FUDSdischarge operating conditions, the simulation and theoretical curve of particle solidphase surface concentrate are compared, which validates the effectiveness of thesetwo kinds of approximate equations.Based on the backward difference operator, the simulation of two mechanismmodels is realized in MATLAB. Under different rates constant current and FUDSdischarge operating conditions, the terminal voltage curve of batteries obtained by SPmodel and P2D model and the theoretical curve got by COMSOL software arecompared. And it can be conclude that the simulated program of these two models iseffective. At the same time, comparison and analysis about these two equivalentmodels are conducted according to the simulated results. After that, we gained thesuitable operating condition of SP model.Finally, variance between the fitting curve about models and theoretical terminalvoltage curve is regard as the objective function of genetic algorithm which can helpus to identify parameters in battery. A large number of experiments are conducted toanalysis the sensitivity of parameters, and suitable operating conditions are designedto achieve the parameter estimation step by step. On the one hand, via computing theerror between the eight parameters values obtained by GA and truth values, on theother hand, comparing the fitting degree between the fitting curve and theoreticalcurve about the battery terminal voltage, we can get a conclusion that the result ofparameter estimation and identification method are effective.