Research On Intelligent Advance Prediction Technology Based On Tunnel Face Using 3D Seismic Detection

When the tunnel constructing in complex geological settings,such as fault zones and karst areas,it is easy to cause disasters such as tunnel blockage,instability,or collapses.It can be seen that the advance forecast of the tunnel is vital to reduce the disaster.The existing advance prediction technology of tunnel based on seismic detection is mostly based on the full-space seismic wave field propagation model,and an geometry is arranged on the tunnel wall to obtain the information of the front anomaly.However,due to the interference of free interfaces such as the face and the tunnel wall,the propagation of the seismic wave field in the actual tunnel space does not fully comply with the law of seismic wave field propagation in the whole space.The detection ideas based on the full-space model are prone to misdetection and missed detection.The engineering practice also shows that detection reliability still needs to be improved.At the same time,there are few studies on the law of seismic wave field propagation and energy distribution in tunnels,and the lack of knowledge of wave field characteristics.Secondly,the geometry arranged on the tunnel wall is time-consuming,expensive,and has poor repeatability,which is not conducive to multiple detections.Thirdly,manual processing is often used to process data.When the data is large,the imaging time is long,the cost is high,the artificial influence is large,and the versatility is strong.Therefore,this article carry out the analysis of the propagation characteristics of the tunnel seismic wave field,and propose an approximate full-space model that more satisfies the real tunnel wave field propagation law.On this basis,we study the high-efficiency,low-cost advanced prediction technology that is more in line with the model.In this paper,we use finite difference numerical simulation technology to study the laws of tunnel seismic wave field propagation,seismic wave field energy distribution and wave field conversion under different excitation and reception positions.On this basis,we propose an approximate full-space model that is more in line with the true tunnel seismic wave field propagation law.Combining actual data and numerical simulation data,we analyzed the advantages and disadvantages of the advanced prediction technologybased on the detection of the tunnel wall,and proposed the advanced prediction technology of the tunnel based on the 3D seismic detection on the face.We verify that the method is more in line with the theory of seismic wave field propagation in approximately full space from the source excitation and signal reception.The wave field analysis results show that the detection distance of this method is farther,the detection accuracy is higher,the reliability of the 3D reflection information is higher,and it is less affected and interfered by the outside world.Secondly,this paper studied the method of data acquisition.We analyzed the particularity of the 3D geometry layout on the tunnel face and the similarities and differences with the design of the 3D geometry on the surface in terms of fold,azimuth and offset.Combining the seismic wave field propagation law and data processing methods,we studied the main factors that affect the layout of the tunnel 3D geometry.The quality control parameters of the tunnel 3D geometry are proposed.On this basis,combined with actual data collection requirements,we studied the layout of fast and low-cost geometry that better meets the needs of tunnel construction.Thirdly,this paper studies the processing and interpretation technology of 3D seismic data.We analyzed the characteristics of the3 D seismic prediction data,and proposed a processing and interpretation idea based on the combination of machine learning technology and seismic data processing and interpretation technology.The difficulties of machine learning technology in the specific application of data processing,label and sample data production methods,network model and parameter selection were studied,and the application method of machine learning technology in tunnel seismic data processing was proposed.At the same time,we studied the method of constructing the correlation model of seismic data collected in different tunnel construction stages based on machine learning to realize the intelligent interpretation of seismic data;Finally,we study the application of this method in actual data.The detection results show that the method can realize rapid,low-cost and efficient imaging of anomalies in a large distance.