Abstract
Microseismic (MS) signals recorded by sensors are often mixed with various noise, which produce some interference to the further analysis of the collected data. One problem of many existing noise suppression methods is to deal with noisy signals in a unified strategy, which results in low-frequency noise in the non-microseismic section remaining. Based on this, we have developed a novel MS denoising method combining variational mode decomposition (VMD) and Akaike information criterion (AIC). The method first applied VMD to decompose a signal into several limited-bandwidth intrinsic mode functions and adaptively determined the effective components by the difference of correlation coefficient. After reconstructing, the improved AIC method was used to determine the location of the valuable waveform, and the residual fluctuations in other positions were further removed. A synthetic wavelet signal and some synthetic MS signals with different signal-to-noise ratios (SNRs) were used to test its denoising effect with ensemble empirical mode decomposition (EEMD), complete ensemble empirical mode decomposition (CEEMD), and the VMD method. The experimental results depicted that the SNRs of the proposed method were obviously larger than that of other methods, and the waveform and spectrum became cleaner based on VMD. The processing results of the MS signal of Shuangjiangkou Hydropower Station also illustrated its good denoising ability and robust performance to signals with different characteristics.
Highlights
The purpose of this paper is to develop a new method combining variational mode decomposition (VMD) and Akaike information criterion (AIC) to suppress random noise in MS signals
It is a sign of insufficient decomposition that waveforms in different frequency intervals are not clearly separated, which has a negative effect on noise reduction
It can be seen from the table that no matter what the initial noise level is, the signal-to-noise ratios (SNRs) after VMD-AIC
Summary
Brittle failure of rock masses triggered by engineering activities and natural changes results in the energy release with the spreading of a seismic wave, which is called a microseismic (MS) event [1]. Battista et al [17] successfully applied it to remove cable interference from seismic data, and Bekara and Van der Baan [18] proposed a f-x EMD method to suppress the random and coherent noise in complex geologic sections. Dragomiretskiy and Zosso [35] developed a new recursive method, called variational mode decomposition (VMD), as an alternative of EMD-based methods. It can achieve adaptive separation between frequency bands of nonstationary time series through iterative optimization. The VMD algorithm, despite its outstanding performance, processes the noisy signal in the entire time window like most of the above algorithms It can only separate the high-frequency noise from the effective signal by choosing appropriate components. We denoise 40 noisy signals collected in Shuangjiangkou Hydropower Station by these methods
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
