Abstract:

In view of the problem of outliers in sensor measurement in lithium battery, a chi-square detector and corresponding filtering algorithm are designed to eliminate the influence of outliers on lithium battery state of charge estimation in this study. The second-order RC equivalent circuit is selected to describe the battery dynamic model, and the parameters of the battery model are identified by Kalman filter in an off-line way. Considering the existence of outliers in sensor data, the chi-square detector is used to detect the outliers in real-time. When outliers are detected, an improved SOC estimation algorithm that only depends on the model is proposed according to the idea of zero-order preservation, which can resist the measured outliers well. Under FUDS condition, the experimental simulation shows that the designed outlier detector and the improved SOC estimation algorithm can accurately detect the occurrence of outliers, and the estimation error of SOC is guaranteed within 2%, reflecting good estimation performance.

Key words: lithium-ion battery, equivalent circuit model, Kalman filter, parameter identification, sensor measurement outlier, chi-square detection, SOC estimation, FUDS