Analysis of Wavelet Decomposition on Acoustic Spectrums for Gas Composition Sensing

Abstract

Wavelet decomposition on acoustic spectrums is a novel method in the field of gas composition sensing. However, the existing method doesn’t explain the relationship between the characteristics of acoustic gas spectrums and the results of wavelet decomposition, and it has high computational complexity. In this paper, we firstly find it can basically keep the recognition rate if the detail coefficients from wavelet decomposition are used to construct samples instead of characteristic coefficients extracted from the detail ones. Nevertheless, the detail coefficients can represent the properties of acoustic gas spectrums, which can’t be represented by characteristic coefficients. Then, we find points of detail coefficients with relatively high recognition rate are mainly distributed in the first three layers. Therefore, it can reduce the computational complexity when three-level decomposition is directly utilized in the method of gas composition sensing.