A Comparative Study of the Dwt and Wpt for Electrochemical Information Extraction of a LiCoO2 Cell

In general, an experimental discharging/charging terminal voltage is highly dependent to the pulse current. According to the pulse currents including different magnitude, time interval, and type (discharging/charging), it is possible to get various terminal voltages with non-stationary and transient phenomena. In this work, for elaborate information extraction of electrochemical characteristics from a LiCoO₂ cell, the wavelet transform (WT) can be properly considered as a significant role. The WT is a state-of-the-art mathematical approach that decomposes a time domain signal into different frequency groups and provides an effective solution for analyzing the non-stationary signals. Specifically, this work introduces a comparative study of the discrete wavelet transform (DWT) and wavelet packet transform (WPT) for electrochemical feature extraction of a LiCoO₂ cell. Both of DWT and WPT have the framework of multi-resolution analysis (MRA) with a vigorous function of both time and frequency localization. Through MRA requiring filtering and down-sampling, the information on the electrochemical characteristics of a LiCoO₂ cell can be extracted from the discharging/charging voltage signal (DCVS) over a wide frequency range. Namely, the structure of the WPT is similar with that of the DWT. The main difference in the two techniques is the decomposition ability of high frequency band. In the DWT-based MRA, only the low frequency band is decomposed giving a right recursive binary tree structure, whereas in the WPT-based MRA, the low, as well as the high frequency bands are decomposed giving a balanced binary tree structure. Thus, the WPT can be defined as extension of the DWT. According to the DCVS with and without frequent discharging/charging, one of the two techniques will be finally selected as the optimum solution of electrochemical characteristics analysis. Extracted information can be usefully applied to cell’s discrimination for stable pack configuration, model-based state-of-charge (SOC) and state-of-health (SOH) estimations, etc.