Microseism wave arrival time point detection method and system

Abstract

The invention discloses a microseism wave arrival time point detection method and system. The method comprises steps of obtaining a microseism oscillogram, and constructing an arrival time point detection model based on a segmentation and integration multi-head self-attention mechanism; and finally, sequentially inputting the micro-seismic oscillogram into an extractor, an encoder and a generatorof the arrival time point detection model for detection to obtain an arrival time point. According to the method, the time point detection model is constructed based on the segmentation integration multi-head self-attention mechanism, i.e., the extractor encoder generator deep learning framework is constructed based on the segmentation integration multi-head self-attention mechanism and is used for the arrival time point detection of micro-seismic signals, so the processing speed is improved, arrival time point detection precision is also improved, and detection efficiency is improved; the precision requirement is met in the engineering environment, and the arrival time point is intelligently determined.

Description

technical field [0001] The present invention relates to the technical field of time point detection, in particular to a method and system for detecting the time point of arrival of microseismic waves. Background technique [0002] In order to meet the growing human demand for resources and underground infrastructure, my country will continue to increase the mining of mineral resources and the development and utilization of underground space. These major projects and infrastructure are closely related to rock mass engineering disasters. With the advancement of major national projects, catastrophe problems in rock mass engineering (rock bursts, landslides, water inrush, landslides, etc.) will become increasingly prominent. Therefore, analysis and early warning of rock mass catastrophe is an urgent and major demand. [0003] Microseismic activity monitoring caused by rock mass failure is widely used in resource mining and underground space development and utilization to evalua...

AI Technical Summary

Problems solved by technology

[0004] At present, the existing methods for determining the time point based on the waveform data of rock mass rupture mainly include two methods, the first is manual judgment; Noisy microseismic data can accurately predict the arrival point, but manual judgment has the following disadvantages: 1. The processing speed is slow and cannot handle a large amount of data

2. Most of them are handled by experience, without theoretical basis

3. Human beings will be affected by biological factors such as fatigue and distraction, and cannot work with high quality for a long time

In addition, the current automatic time point picking method can only accurately predict the time point for relatively pure microseismic signals without noise interference, but the signals monitored in actual engineering are often accompanied by a lot of noise, so it is impossible to accurately predict When the time comes

[0005] Based on the above problems, how to solve the problems of slow manual operation, inability to process a large amount of data, and unsatisfactory precision in engineering environments has become an urgent technical problem in this field.

Images