Time-Frequency Distribution Analysis of the Stress Relaxation of Sandstones Affected by Dynamic Disturbance

Low frequency noisy signals are often generated during studies of uniaxial compression and stress relaxation. Denoising describes signal processing methods and devices that extract signal from a mixture of signal and noise to improve the signal-to-noise ratio of signals. These tools include the use of appropriate grounding connections, filters, the use of shielded and twisted wires, signal averaging methods, and the incorporation of differential input voltage amplifiers. This study investigated the denoising process for the study of rock under the combined conditions of stress relaxation and dynamic disturbance using a wavelet denoising scheme. The signal screening process in the wavelet domain identifies and eliminates noise-corrupted portions that may lead to inaccurate prognosis of laboratory environmental noise conditions. The wavelet transform time-frequency analysis of the experimental data evaluates denoising effects throughout the rock time-dependent deformation process. The frequency typically ranges from about 0 to 50 Hz for the original signal of dynamic disturbance. For the denoised signal of one-stage stress relaxation under dynamic disturbance, the frequency ranges from 0 to 10 Hz. This study demonstrates the feasibility of wavelet-based denoising, and shows experimentally that the use of the upgraded stress relaxation-disturbance testing machine is an effective and reliable tool to monitor the time-dependent behavior of rocks.