Combination between adaptive SMO and DWT-based an adjusted EDCV signal for robust SOC estimation in battery pack applications

This research elaborately investigates an innovative work for high-accuracy SOC estimation using an adjusted experimental discharging/charging voltage (EDCV) signal of series/parallel-cell configured battery pack through combination between adaptive sliding-mode observer (SMO) and discrete wavelet transform (DWT). The steps for robust SOC estimation in the proposed approach as follows. First, after discharging/charging of the battery pack, an obtained EDCV signal is decomposed into different frequency sub-bands (low-and high-frequency components, A_n and D_n) using the DWT-based multi-resolution analysis (MRA) method. Second, a low frequency component A_n is specifically served as a terminal voltage and sent to the equivalent circuit model (ECM)-based SMO for obtaining SOC information. The ECM of the battery pack previously constructed by discrimination process that selects unit cells with similar electrochemical characteristics is well considered. Finally, the SOC performance is compared with that of Ampere-hour counting for validation of this proposed work. Experimental results clearly show that this approach sufficiently enables us to provide a reliable SOC estimation whole discharging/charging period using only fundamental ECM without additional consideration such as internal state variation and noise model for compensating the ECM errors. This approach used two experimental packs of 6S1P and 8S2P that respectively connected in series and in series/parallel using 2.2Ah unit cells discriminated in advance.