Compressing and reconstructing the voltage measurement of lithium-ion battery using compressed sensing

Accurate state estimation is critical for the stable operation of lithium-ion (Li-ion) battery energy storage systems. The high-precision single-cell voltage measurement is the prerequisite of accurate state estimation, yet it requires the battery management system (BMS) sampling with required frequency to capture the dynamic changes of the voltage. However, the transmission and storage of large-scale battery voltage data is a significant challenge due to the limited memory and transmission bandwidth of BMS. In response to this issue, this paper proposes a compressed sensing driven net (CSDNet) for voltage data compression and reconstruction, which consists of a sampling module and a reconstruction module. First, the sampling module is designed to learn the prior knowledge of the signal during the model training and achieve data compression . Then, a reconstruction module is employed to learn the mapping relationship between the compressed signal and the original signal to achieve the signal reconstruction. The feasibility of the proposed CSDNet is validated using the Oxford and NASA battery datasets. The results indicate that even for a compression ratio as low as 1 %, the proposed CSDNet can achieve a mean absolute error (MAE) of 14.63 mV within 0.0937 ms for the voltage reconstruction. • A novel CSDNet is proposed to fast compress and reconstruct the battery voltage data. • A sampling module is designed to learn the prior knowledge of the signal and acquire the CS measurements. • A reconstruction module is designed to map the relationship between the CS measurements and the original voltage signal.