An improved wavelet adaptive logarithmic threshold denoising method for analysing pressure signals in a transonic compressor

Abstract

This paper proposes an improved wavelet threshold denoising algorithm based on orthogonal wavelet transform. The algorithm, called adaptive logarithmic threshold denoising algorithm based on wavelet (ALTDAW), processes the dynamic pressure signals generated by a transonic axial compressor. Combined with an adaptive logarithmic threshold function, it sets the optimal threshold for each decomposition level. In this way, noise is effectively identified and eliminated. Since ALTDAW is adaptable, it reduces the maximum decomposition level, thereby decreasing the processing time and improving the computational efficiency. In numerical experiments, the performance of ALTDAW was compared with that of the classical soft and hard threshold algorithms. Relative to the classical algorithms, ALTDAW increased the signal-to-noise ratio (SNR) by 27.9% and 44.2%, and reduced the processing time by 38.5% and 37.6%, respectively. The practicality of the algorithm was validated on the complex dynamic pressure signals of a transonic axial compressor. When processing these signals, the algorithm depicted that disturbance passes through the tips and the spike three revolutions before the compressor stalls, consistent with the physical properties of transonic axial compressors.