Research On Modeling And SOC Algorithm Of LIFePO₄ Battery

In recent years, the energy crisis and the environmental problems have become more prominent, LiFePO4 batteries to its own security, no pollution and many other performance characteristics are gradually to be concerned, and have a wide range of applications in the car and backup power supply etc .to be the further study of the battery, setting up accurate model of characteristic for batteries battery management system in engineering development has a very important significance, as the electric car in the power battery, accurate state-of-charge (SOC) estimates to improve battery life and vehicle performance is very important.In order to establish the accurate battery model, this paper made a lot of experiment of charging and discharging LiFePO4 batteries to test the battery performance characteristics, based on the existing first and second order model, the model was improved through the HPPC composite pulse condition acquire the parameters of different temperature,different discharge rate and different state of charge. compared the two models in the three conditions with dynamic response value of the voltage and gained a specific test precision, in the complex conditions, the first-order correction in the model has the average error of 0.025V, the average second-order error correction model is 0.0103V, and later provides accurate models for extended kalman filter algorithm to estimate SOC.The state of charge (SOC) estimation as power battery is the one of core technology of the electric car development, it's the key of utility and industrialization. In this paper, basis for setting up the model, battery remaining power are calculated by extended kalman filter. Through compared in varying conditions based on the setting model, First-order model kalman filter algorithm accuracy is better than the second-order in the constant conditions. But in the changing conditions, the second-order models have higher accuracy. Complex conditions through the convergence of second-order model validation found that the initial error of 20%, the battery after a nine-minute run, the algorithm converges to the estimated accuracy of 5% around; the initial error of 50%, After 40 minutes of adjustment algorithm, the algorithm converges to the estimated accuracy of 10% range.The improved model of this paper can better reflect the dynamic performance of battery, and can meet the practical needs of the simulation, the state of charge estimation of battery also have the high accuracy, good dynamic adaptability, and has a faster convergence speed.