| Microseismic/Acoustic Emission(MS/AE)monitoring technology is a new monitoring method that is widely used in geotechnical engineering.By accurately acquiring and processing the monitoring signals,this technology can reflect the occurrence time,location,strength,and fracture mechanism of crack fractures,and can characterize the evolution process of rock mass damage.However,due to the heterogeneity and dynamic change of rock velocity model and instability of monitoring data processing effects,it is difficult to conduct accurate MS/AE event locations in practical applications.Whether accurate source locations can be achieved has become a technical bottleneck restricting the application of MS/AE monitoring technique,and it is urgent to make a technical breakthrough.In this paper,MS/AE event location is taken as the research objective and corresponding studies in laboratory scales have been carried out deeply.The high efficient automatic processing of MS/AE monitoring signals and the development of high precision source location methods are realized,and they are applied to the quantitative analysis of rock crack behaviors.This research has achieved the following innovative results:(1)A deep-learning-based automatic acoustic first arrival picking method is established to picking the first arrivals of massive AE waveforms effectively and accurately.In this method,a convolutional neural network model that is driven by acoustic amplitude and corresponding higher-order statistics is designed to classify the sampling points in acoustic waveform as signals or noises.Then,a picking algorithm that combines nonlinear curve fitting method and unsupervised density-based clustering method is established to identify the boundary between signal section and noise section as the acoustic first arrival.Compared with the traditional automatic picking method,the accuracy is significantly improved.At the same time,the influence of high-order statistics on first arrival picking is quantitatively analyzed.(2)A 3D fast sweeping method is proposed to accurately calculate the travel time of stress waves in complex velocity models.Based on the fast sweeping method and database matching technology,the MS/AE source inversion location method for the complex velocity model is developed.Compared with traditional source location method,the localization error of AE events is greatly reduced by the method developed in this study.Because that the algorithm format of 3D fast sweeping method can be used for parallel computation,our location method can significantly improve the efficiency of source inversion location.(3)A dynamic velocity model acquisition method using time-shift tomography algorithm is proposed.Based on the fact that the rock velocity model changes in the process of deformation and fracture,a time-shift acoustic tomography method is proposed.This method is applied to active/passive AE event data of the fractured sandstone uniaxial compression test to obtain the dynamic velocity model of the rock sample.The analysis shows that the influence of velocity change on the source location error can not be ignored.Compared with static velocity model,the travel time difference calculated based on the time-lapse velocity model is more consistent with the observed data,and the location errors of active AE events are smaller.The results show that this method can effectively eliminate the negative effect of velocity variation that is induced by the rock structure change on AE events location accuracy.(4)The evolution law of AE signal characteristics in rock fracture behavior is studied,which provides a theoretical basis for the application of MS/AE monitoring to dynamic disaster monitoring and early warning.Based on research 2 and research 3,accurate AE event location results of fractures sandstone uniaxial compression test are obtained.The spatial and temporal distribution of AE events and the variation of AE signal statistical characteristics during the formation of macroscopic cracks were compared and analyzed.The b value,which represents the magnitude-frequency distribution of AE events,and the D value,which represents the fractal dimension of AE events spatial distribution,are used to analyze the crack evolution law of compressed rock quantitatively.The results show that the statistical characteristics of AE monitoring data can be used as a short-term precursor for deep rock dynamic disasters monitoring and early warning.(5)The evolution law of stress wave propagation attributes during the process of rock micro-crack initiation is studied.The location and time of microcrack initiation were determined by analyzing vertical stress-strain data and surface strain field change of the rock sample.A time-shifted spectral ratio method is developed to obtain the attenuation coefficient variation by processing the waveform data of active ultrasonic transmission waveforms.By analyzing velocity and attenuation coefficient variation in the region where microcrack initiates,this study proves that compared with the stress and strain monitoring data,attenuation coefficient change-velocity difference data of transmission ultrasonic waveforms is more sensitive to small scale fracture of rock cracking behavior.The propagation attributes can be used to reflect the early activities of the key structure plane,to realize the early warning of deep rock dynamic disasters. |
PDF Full Text Request