Investigation of microseismic signal denoising using an improved wavelet adaptive thresholding method.

There are high- and low-frequency noise signals in a microseismic signal that can lead to the distortion and submersion of an effective waveform. At present, effectively removing high- and low-frequency noise without losing the effective signal of local waveform spikes remains a challenge. This work addresses this issue with an improved wavelet adaptive thresholding method. Because a denoised signal conceptually approximates the minimum error, a dynamic selection model is established for the optimal threshold. On this basis, an adaptive correction factor a is proposed to reflect the noise intensity, which uses the 1/2 power of the ratio of the median absolute value to the amplitude of the monitoring data to reflect the noise intensity of the wavelet detail signal and corrects the size of the denoising scale. Finally, the performance of the improved method is quantitatively evaluated in terms of the denoising quality and efficiency using the signal-to-noise ratio, root-mean-square error, sample entropy and running time.