Massive battery pack data compression and reconstruction using a frequency division model in battery management systems

Abstract

Abstract Battery data storage is one of the essential functions in battery management systems (BMS). For a limited memory space in BMS, reducing the recording frequency obviously increases the total recording time. However, it also increases the trend of the recording signal distortion. In this study, a novel battery data storage method which uses the frequency division model of the battery pack is presented. The higher frequency dynamic voltage signal data of single cell is obtained by the reconstruction of the voltage recorded with the lower recording frequency. The real cell voltage signal data of the battery pack is obtained by the dynamic current condition experiment. By comparing the root mean square error (RMSE) and required storage space as two indicators for accuracy and efficiency respectively between real recording cell voltage signal and the reconstruction cell voltage under the New European Driving Cycle (NEDC), the results show that there is a trade-off between the required storage space and the mean cell voltage RMSE. With the flexible recording frequency reconstruction approach, the required storage space is shortened and the reconstructed voltage distortion is increased for lower recording frequency in BMS. The cell voltage RMSE and required storage space results confirm the proposed method that can effectively save the required storage space on the premise of ensuring the accuracy for BMS at different recording frequency in practical application.