Impact Of Colored Noise On SOC Estimation Of Lithium-ion Batteries

Now, the power lithium-ion battery is the main onboard energy storage device for electric vehicles, whose state of charge(SOC) has a serious influence on their drive, economy, and safety. Hence, the battery SOC estimation has been a research hotspot. Although the Kalman filtering theory has been widely applied on the battery SOC estimation, the colored noise is hardly investigated which can contaminate the battery current in practice. However, the Kalman filtering theory is built on white noises. In this thesis, the first order resistance-capaciatance equivalent circuit model is built for power lithium-ion batteries. The battery SOC estimation is studied on the Kalman filtering techniques impacted by the battery current cololed noise.Firstly, the battery SOC estimation techniques and the colored noise Kalman filtering are reviewed. Both the extended Kalman filtering(EKF) and the unscent Kalman filtering(UKF) are used to estimate the battery SOC with Gaussian noises. Results show that both EKF and UKF have good accuracy of the battery SOC, and the latter is a little better.Secondly, the three colored noises are introduced as the random constants, the random walk, and the first order Markov process, which hybrids the battery current. The battery SOC estimation by EKF and UKF is investigated with the battery current colored noises, which is studied on the Matlab software. Results show that the battery SOC estimation by EKF and UKF has large errors when the battery current is contaminated by the colored noises, and both of Kalman filters may be divergent.Finally, the state augument Kalman filtering(SAKF) is used to estimate the battery SOC with the colored noise current. The SAKF can consider the colored noise as a part of the state variables, which gets the colored noise white to improve the battery SOC estimation. Results show that the SAKF is effective to filter the colored noise on the battery SOC estimation, its maximum error is less than 3%.