A Fast SNR-based Vibration Events Detection Algorithm for AETA Geo-acoustic Data

Abundant information can be mined from AETA geo-acoustic data. In this paper, a fast SNR-based vibration events detection algorithm is proposed by improving traditional STA/LTA algorithm for these data. Background noise sampling and different update strategies during different states are applied to improve detection performance. Furthermore, we studied relative method to update the eigenvalues in O(1) time when an element in the sliding window is operated so data stream can be detected in O(N) time. Finally, in the experience part, white Gaussian noise with SNR range from 3 to 10dB is added to the geo-acoustic data of 72 AETA stations around the Longmenshan fault zone from October to November 2019 and result is measured by start point relative error, end point relative error and detected energy ratio. Result shows that the proposed method achieves considerable detection effect for vibration events in AETA geo-acoustic data with low computation complexity.