Noise suppression of the DWT-based MRA on mother wavelet and decomposition level optimization for a robust adaptive SOC estimator in multi-cell battery string

Abstract

This approach newly presents an innovative method that implements the noise suppression considering the discrete wavelet transform (DWT)-based multi-resolution analysis (MRA). Specifically, this approach elaborately carries out two comparative analyses of the noise suppression depending on information on mother wavelet and decomposition level in the MRA. The first is to select the decomposition level having a high performance of the noise suppression under the same condition of mother wavelet. On the contrary, with an identical condition of the decomposition level, the second is to compare the noise suppression of different mother wavelets under the same condition of the decomposition level. All comparative analyses are evaluated in terms of signal-to-noise ratio (SNR). This approach entirely makes a comparison between two noise suppression techniques considering hard- and soft-thresholding. From these results, our analytic results suggest the clear comparison by showing the SNR difference with various mother wavelets and decomposition levels. Consequently, it can be surely expected to optimize the mother wavelet and the decomposition level for noise suppression of the noise-riding EDCV signal in multi-cell battery string. Our experimental multi-cell battery string in series/parallel (2S3P) using 2.2Ah unit cells previously discriminated was used.