The State-of-charge Estimation For Lithium-ion Battery Using Adaptive Approach

The exacerbation of energy crisis and environmental pollution have made the microgrid energy storage and electric vehicle to attract extensive attention. The common point of these fields is that they need a battery management system to manage and control the battery effectively. The state estimation of the battery is not only a great important task but also a difficult one which have a great influence on energy management, cycle life, safety and stable performance of the battery, and costing. It is challenging to estimate the state of the battery rapidly and accurately due to the fact that the operation of the battery is a complicated electrochemical process, which is influenced by the environment and ageing process. Therefore, the study of battery state estimation has significance in both theory and application. In this dissertation, the specific research work is studied on the lithium iron phosphate battery(LiFePO4).The mathematical model from zero-order to n-order RC network structure is constructed in this dissertation, which aims at modeling and parameter identification of Lithium-ion battery. In order to weigh the trade-offs of precision and complexity of model prediction, the method of four-step parameter identification is built, including battery parameter identification, model optimization and model evaluation. In model parameter identification, since the online identification of model parameters needs to consume the amount of computation continuously and has strong data dependence, an offline identification method of model parameters based on measuring current and voltage driving data is put out in this dissertation. The estimation method based on data and model fusion is proposed to estimate SOC. This method reduces the computational complexity using a piecewise linear approximation of the model parameters, and describes the attenuation of battery using Arrhenius theory, which achieves the precise state under the influence of aging of the battery. On this basis, the available capacity and internal resistance are two indexes representing the health condition of battery. The joint estimation of SOH and SOC is achieved by the estimator based on dual Kalman filter.In this dissertation, the method of modeling and parameter identification is given, which had solved the issue of weighing the trade-offs of precision and complexity of modeling. The piecewise linear approximation of the model parameters had reduced the computational burden caused by nonlinear between model parameter and SOC. In order to solve the influence of battery aging, a method based on Arrhenius theory is proposed to describe the attenuation of the battery. On this basis, the proposed estimation algorithm based on dual Kalman filter achieved joint estimation of health condition of battery and SOC.