TSP wavefield separation and noise suppression based on the LC-KL-DSW method

A complex seismic wavefield and strong random noise are the main factors that affect the detection distance and precision of tunnel seismic prediction (TSP). Considering that the Karhunen–Loève (KL) and directional seismic wave (DSW) methods have the ability to extract the target signals and suppress random noise, we attempt to apply these methods to solve the problem of weak effective signals being difficult to separate in a complex wavefield. The local correlation (LC) algorithm can not only improve the accuracy of delay time estimation but also reduce the intensity of coherent interference. In this study, a KL-DSW method based on LC (LC-KL-DSW) is proposed for wavefield separation and noise suppression. First, the time window of the target signal is selected, and the cross-correlation method is used to estimate the time delay parameters of the target signal. Then, the event of the target signal is levelled by the time delay correction, and the signal is stacked in phase by an N-element array. Finally, the KL transform is used to extract the target signal, which can be obtained by anti-delay correction. Numerical simulations show that the complex seismic wavefields are separated well and that the signal-to-noise ratio (SNR) of the target signal is increased by approximately k (20 < k 〈100) times via the N-element LC-KL-DSW technique. Under ideal conditions, only spherical diffusion attenuation is considered, and the detection distance increases by approximately k-fold. Hence, the LC-KL-DSW technique has great significance for high accuracy and long-distance advanced tunnel detection.