Research On The Characteristic Response Of Fault Geological Radar In Mountain Tunnel

With the vigorous development of transportation construction in China in recent years,mountain tunnel construction is in full swing.The geological structure of mountain tunnels is complex.Various engineering accidents are prone to occur when tunnel construction encounters geological disasters.region.As a rapid development method in recent years,geological radar has been used in tunnel geological prediction due to its advantages of rapid detection,non-destructiveness,and strong anti-interference.However,the interpretation of radar images is relatively subjective today.There are few studies on the response to unfavorable geological features,which often leads to misreporting and underreporting.Aiming at these problems,this paper carries out the research on the characteristic response of the geological radar in the fractured area of the tunnel,and has a clear understanding of the fractured area from the theoretical analysis,forward experiment and engineering measurement to realize the automatic classification and recognition of radar signals.The main content of the paper is as follows:(1)Study the propagation law of geological radar electromagnetic waves in underground media,summarize the conventional processing methods of radar data;analyze and compare the conversion methods of signal frequency domain and time-frequency domain,and select mature wavelet transform(WT)as the signal conversion Methods;starting from Maxwell's curl equation and combining with the Yee-style network,the time-domain finite difference method is studied.(2)Carry out forward simulation based on the time-domain finite difference method,combine the relative dielectric parameters and conductivity that are common in underground media,explore the limit detection depth of different bottom radars;establish water,gas,water-gas mixture and interface The irregularly shaped fault model,through the joint analysis of time domain,frequency domain and time-frequency domain,summarizes the propagation law of electromagnetic waves in various faults and distinguishes different types of faults.(3)Build a network model based on XGBoost algorithm,use simulation signals as data sets,7 feature parameters in radar single-channel waves as classification features,and use it as input data;classify the external environment where the tunnel fault area is located,increase The prior conditions of classification,comparative analysis of the accuracy and recall of radar data classification under single conditions and combined conditions,both of which have reached a relatively high level,and the effect of the model under single conditions is better.(4)Based on the response characteristics of faults and the automatic classification and recognition model,combined with the engineering conditions and detection data of the Yangling tunnel in Yixing,the application of the above results in actual engineering is verified and evaluated.The radar detection data of the Yangling tunnel fault is basically consistent with the forward simulation data,which verifies the credibility of the forward experiment;the automatic recognition and classification accuracy of the signal and the recall rate have met the engineering requirements,and the faults of the Yangling tunnel fault Automatic recognition.In this paper,there are 48 figures,15 tables,and 90 references.