Suppression of periodic interference during tunnel seismic predictions via the Hankel-SVD-ICA method

Abstract

Periodic interference is one of the most common interference sources in tunnel seismic prediction (TSP). This noise can deleteriously affect signal quality, and the signal can be completely drowned out by the noise in some cases, which can seriously affect seismic wave velocity estimations and TSP detection distance and detection accuracy. To solve this problem, a single-channel independent component analysis (SICA) based on the Hankel matrix and a singular value decomposition (Hankel-SVD-ICA) is proposed to suppress the periodic interferences. The traditional band-pass filters and notch filters are ineffective when seismic data and the periodic interference bandwidths overlap. In this study, we used TSP data to evaluate the Hankel-SVD-ICA method in detail. The Hankel-SVD-ICA was tested on synthetic recordings and field data, and the results showed that the method effectively separates the seismic data from data mixed with strong periodic interferences when seismic data and periodic interference bandwidths overlap. Qualitative and quantitative comparisons show that the Hankel-SVD-ICA provides better recordings than both the Hankel matrix singular value decomposition (HMSVD) and notch filters. The comparisons also show higher quality data and a more reliable P-wave velocity assessment when the Hankel-SVD-ICA is applied to TSP data, which significantly increases the detection distance and improves the lithological interpretation. Moreover, we study the Hankel-SVD-ICA method for frequency time-varying periodic interference suppression in TSP, and the results provide a good reference for suppressing periodic interference in other fields.