Lithium ion battery dual Kalman filtering capacity estimation method

Abstract

The invention provides a lithium ion battery dual Kalman filtering capacity estimation method based on the combination of partial charge curves and a semi-empirical life model. The method effectivelycombines the respective advantages of a data driving method and an empirical model method for battery capacity estimation, for the situation that complete charging is rare in the actual use process ofan electric automobile, a method based on the capacity online estimation of the partial charge curves is studied, by monitoring constant current charging data between two static voltage characteristic points, the battery capacity is estimated online, it is considered that an actual charging process may not be in accordance with the conditions required by the estimation method, the method only canintermittently obtain capacity estimation results, the Arrhenius life model is further adopted for real-time capacity estimation, while parameters of the Arrhenius life model may have a mismatching problem, by designing a DEKF algorithm, the real-time update of the model parameters is achieved, and the battery capacity is also estimated online.

Description

technical field [0001] The invention relates to the field of lithium ion battery capacity estimation, in particular to a lithium ion battery dual Kalman filter capacity estimation method. Background technique [0002] As the number of cycles and storage time increases, the battery will age, and the capacity will gradually decay. The cell with the fastest capacity decay in the entire battery pack will determine the overall life of the battery pack, which will then restrict the driving range of the electric vehicle. Due to the complex and changeable cycle conditions of the actual vehicle, the inconsistency of the manufacturing and use environments among the battery cells makes the capacity attenuation quite different from the laboratory calibration results, and the battery cells on the real vehicle cannot be fully charged and fully charged. Put the test capacity. Therefore, accurate online estimation of battery cell capacity in a battery pack and prediction of battery life b...

AI Technical Summary

Problems solved by technology

However, this method is a life model established according to the specific cycle working conditions in the laboratory. The model is open-loop, and it is difficult to adapt to variable working conditions and the inconsistency of different cells in the battery pack. If it is directly applied in the vehicle BMS, the estimated accuracy Usually poor; the data-driven method mainly makes full use of the charge and discharge data, one is to estimate based on the change of SOC, and the accuracy of capacity estimation is highly dependent on the accuracy of SOC estimation; the other is to use the peak value of the incremental capacity (Incremental Capacity, IC) curve , establish a relationship model between peak characteristics and capacity, and use the model to estimate the capacity online. One of the main difficulties of this method is that the voltage in the measurement is sensitive to noise. All peaks on the IC curve are located in the flat area of ​​the OCV-SOC curve. Calculations with the obtained data may lead to inaccurate results

Therefore, the traditional empirical model method or data-driven method will have a large capacity estimation error

Images