An Improved Electric Model with Online Parameters Correction for Large Li-Ion Battery Packs

Abstract

Lithium-ion battery packs based battery energy storage system (BESS) has proved its high potential in energy storage. In studying the BESS, A precise battery model is critical for implementing the system simulation, estimating the SOC, SOH of the battery, as well as optimizing the design. An electric model which can reflect the hysteresis effect and dynamic characteristic is adopted in this paper, and based on that, we proposed a novel method for battery model identification and on-line correction of model parameters by extended Kalman filter (EKF). Thus the parameters of battery model can be regulated in the progress of battery charge and discharge. The method is proved to be efficient and accurate by comparing the simulation result with the experiment data. receive. However, the parameter identification algorithm is also more complicated, resulted in complex calculation. With the aging of the battery and the changing of temperature, the battery performance will change constantly, so do the model parameters. The model parameters got in a certain test, which are consistent with the actual characteristics of the battery, but may not work well in other conditions. Therefore, a battery model which could be able to change over test environment is needed. The battery model that proposed in this paper is improved on the basis of the Thevenin circuit, in order to capture the nonlinear characteristics and dynamic characteristics. We took EKF method, which is an intelligent means for estimating state value of a dynamic system, as the model parameter identification algorithm (4-6). In many research and applications, the model parameters are fixed. In fact, the circuit model cannot fully reflect performance of the battery, of which the parameters are needed to change with different battery states and ambient temperature, in order to adapt to the actual battery better. Especially, taking the Kalman filter or EKF algorithm to estimate the battery state, such as SOC, requires an accurate model. In order to get model parameters under different conditions, a large number of specific charge-discharge experiments have to be conducted, wherein the workload is quite large. Therefore, we studied an online method for parameters correction, which is able to fix the parameters of the original model when the energy storage system is running. For large-scale battery energy storage system, the method can achieve a goal that the system runs with model identification and parameters correction.