CN112305591B - Tunnel advanced geological prediction method and computer readable storage medium - Google Patents
Tunnel advanced geological prediction method and computer readable storage medium Download PDFInfo
- Publication number
- CN112305591B CN112305591B CN202011078912.2A CN202011078912A CN112305591B CN 112305591 B CN112305591 B CN 112305591B CN 202011078912 A CN202011078912 A CN 202011078912A CN 112305591 B CN112305591 B CN 112305591B
- Authority
- CN
- China
- Prior art keywords
- data
- tunnel
- seismic
- neural network
- network model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 55
- 238000003860 storage Methods 0.000 title claims abstract description 16
- 230000002159 abnormal effect Effects 0.000 claims abstract description 52
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 36
- 238000013135 deep learning Methods 0.000 claims abstract description 34
- 238000012545 processing Methods 0.000 claims abstract description 26
- 238000012937 correction Methods 0.000 claims abstract description 19
- 238000003062 neural network model Methods 0.000 claims abstract description 10
- 230000005284 excitation Effects 0.000 claims abstract description 8
- 238000001914 filtration Methods 0.000 claims abstract description 6
- 238000010276 construction Methods 0.000 claims description 111
- 238000012549 training Methods 0.000 claims description 32
- 239000011435 rock Substances 0.000 claims description 13
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 238000000513 principal component analysis Methods 0.000 claims description 5
- 238000011176 pooling Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 26
- 238000009826 distribution Methods 0.000 description 26
- 238000005516 engineering process Methods 0.000 description 21
- 238000010586 diagram Methods 0.000 description 20
- 230000006870 function Effects 0.000 description 20
- 230000008569 process Effects 0.000 description 18
- 230000000875 corresponding effect Effects 0.000 description 14
- 230000008030 elimination Effects 0.000 description 8
- 238000003379 elimination reaction Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 230000000694 effects Effects 0.000 description 6
- 239000010410 layer Substances 0.000 description 6
- 238000001228 spectrum Methods 0.000 description 6
- 210000003462 vein Anatomy 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000004088 simulation Methods 0.000 description 5
- 230000007423 decrease Effects 0.000 description 4
- 238000003384 imaging method Methods 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000002950 deficient Effects 0.000 description 2
- 239000010438 granite Substances 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 239000011044 quartzite Substances 0.000 description 2
- 230000005641 tunneling Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000007717 exclusion Effects 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000013277 forecasting method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000011295 pitch Substances 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 239000002344 surface layer Substances 0.000 description 1
- 238000003325 tomography Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/28—Processing seismic data, e.g. for interpretation or for event detection
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Acoustics & Sound (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- General Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Geophysics (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention discloses a tunnel advance geological prediction method and a computer readable storage medium, wherein the tunnel advance geological prediction method comprises the following steps: a three-dimensional observation system is arranged on the tunnel face to achieve three-dimensional seismic wave excitation and reception, and the acquired data are processed based on a deep learning algorithm to obtain the position of the abnormal body. The processing of the data based on the deep learning algorithm comprises: based on AlexNet convolutional neural network model algorithm, removing bad channel data from the collected data, and filtering; identifying the first-motion wave based on a U-Net convolution neural network model algorithm, and estimating the speed of the direct wave; performing linear dynamic correction on the original data from which the bad track data is removed; carrying out equal travel time section superposition on the data after linear dynamic correction; and carrying out reflective layer pickup on the superposed data. The tunnel advance geological prediction method and the computer readable storage medium can improve the detection efficiency and the detection accuracy of the abnormal body.
Description
Technical Field
The present invention relates to the field of tunnel engineering technology, and more particularly, to a method for advanced geological prediction of a tunnel and a computer-readable storage medium.
Background
The advanced tunnel prediction technology is a technical method for detecting the geological environment and the geological structure in front of a tunnel face before and during tunnel construction. The tunnel advance forecasting technology based on seismic detection is used for detecting surrounding rock information in front of a tunnel face according to an elastic wave field of the surrounding rock of the tunnel, is large in detection depth and high in detection precision, and is one of main advance forecasting technologies. According to different detection modes, the existing tunnel advanced prediction technology based on seismic detection can be roughly divided into a two-dimensional detection technology and a quasi-three-dimensional detection technology. The two-dimensional detection technology mostly adopts the mode of utilizing explosive, electronic detonator, hammering and the like to excite seismic waves on a tunnel wall or a tunnel face, and an observation system with three-component detectors arranged linearly on the tunnel wall collects the reflected seismic waves to carry out advanced prediction, such as HSP and TSP technologies and the like. The detection technology of the quasi-three-dimensional observation system mainly adopts a ring tunnel wall three-dimensional space arrangement wave detector, an observation system for exciting reflected waves on a tunnel wall or a tunnel face, and seismic data processing modes such as seismic tomography, scattered wave superposition, Hough depth migration and the like are adopted for imaging, such as TRT, TST, TSWD and the like.
The inventor discovers that the advanced forecasting technology based on the seismic exploration has the problems of false exploration or missed exploration, low exploration efficiency, long data processing period and strong ambiguity of interpretation results in the process of realizing the invention.
Specifically, the inventors have found that although observation systems using two-dimensional detection technologies such as HSP and TSP technologies have a certain spatial distribution, they are limited by the tunnel space, and therefore, they have a very small seismic wave excitation reception angle, poor imaging quality, and difficulty in obtaining accurate geological anomaly information. Compared with a two-dimensional observation technology, the wave velocity analysis accuracy of quasi-three-dimensional detection technologies such as TRT, TST, TSWD and the like and the imaging accuracy of poor geologic bodies are improved to a certain extent, however, the detection technology based on the ring tunnel wall three-dimensional space arrangement detectors is influenced by the tunnel wall and the tunnel face, shot points and wave detection points are difficult to be arranged in all directions, partial phase data are easy to be lost, and therefore real three-dimensional detection is difficult to realize; secondly, the abnormal body reflected signals received on the tunnel wall are easily interfered by strong non-reflected information such as tunnel surrounding rock surface waves, converted waves, horizontal strata parallel to the tunnel axis and the like, and the signal-to-noise ratio is low; meanwhile, the existing tunnel advance forecasting seismic data are processed on the basis of manual methods, the period is long, the cost is high, the processing result is greatly influenced by external factors and professional knowledge of technicians, the imaging multi-resolution is strong, and the tunnel advance forecasting method based on the manual method cannot intuitively compare the tunnel advance forecasting seismic data at different tunnel construction positions.
The information disclosed in this background section is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person skilled in the art.
Disclosure of Invention
The invention aims to provide a tunnel advance geological prediction method and a computer readable storage medium, which can improve the detection efficiency and the detection accuracy of an abnormal body.
In order to achieve the aim, the invention provides a tunnel advance geological prediction method, which comprises the following steps: arranging a three-dimensional observation system on the tunnel face to realize the excitation and the reception of three-dimensional seismic waves; and processing the data acquired by the three-dimensional observation system based on a deep learning algorithm so as to obtain the position of the abnormal body in the tunnel. The method for realizing the excitation and the reception of the three-dimensional seismic waves based on the three-dimensional observation system arranged on the tunnel face comprises the following steps: arranging an observation system on the tunnel face; at the first tunnel construction position, the observation system excites seismic waves and receives reflected waves so as to acquire data. The method for processing the data acquired by the three-dimensional observation system based on the deep learning algorithm to obtain the position of the abnormal body in the tunnel comprises the following steps: based on AlexNet convolutional neural network model algorithm, removing bad channel data from the collected data, and filtering; identifying the first-motion wave based on a U-Net convolution neural network model algorithm, and estimating the speed of the direct wave; performing linear dynamic correction on the original data without the bad track data according to the first arrival wave identification result; carrying out equal travel time section superposition on the data after linear dynamic correction; and carrying out reflection layer pickup on the superposed data through a deep learning algorithm, thereby inferring the area of the abnormal body at the first tunnel construction position.
In an embodiment of the present invention, the observation system is centrally symmetrical as a whole.
In an embodiment of the present invention, the receivers of the observation system are arranged radially from the center to the periphery, and the seismic wave generating devices of the observation system are also arranged radially from the center to the periphery.
In an embodiment of the present invention, the removing bad track data from the collected data based on the AlexNet convolutional neural network model algorithm includes: training the AlexNet convolutional neural network model; and removing bad data from the acquired data by adopting the trained AlexNet convolutional neural network model.
In one embodiment of the invention, the AlexNet convolutional neural network model is based on a seismic wavelet field according to seismic exploration resolution, and takes a quarter of a wavelet wavelength as a convolution kernel length and an eighth of a wavelet wavelength as a pooling layer length.
In an embodiment of the present invention, the trained AlexNet convolutional neural network model is adoptedThe bad track removing comprises the following steps: judging the data with the data prediction value outside the range of the effective channel prediction threshold f (x) as noise data and removing the noise data,
wherein, N is label type, pre (i) represents the possibility predicted value of AlexNet belonging to a certain type label for each data, F (x) represents the sum of all label predicted values of each data, F (x)maxRepresents the sum of the maximum all-label predictors of all participating predictors.
In an embodiment of the present invention, the performing a reflector picking on the superimposed data through a deep learning algorithm includes: and identifying a similar region of the multi-shot records by adopting a principal component analysis algorithm, wherein the similar region is presumed to be the region of the geological abnormal body.
In an embodiment of the present invention, the method for advanced tunnel geological prediction further includes: after the estimation of the position of the abnormal body at the first tunnel construction position is completed, a seismic signal correlation model is built based on a U-Net convolution neural network model algorithm, and the estimation of the abnormal body is carried out at other construction positions of the same tunnel according to the seismic signal correlation model.
In an embodiment of the present invention, the building a seismic signal correlation model based on a U-Net convolutional neural network model algorithm, and performing anomaly prediction at other construction positions in the same tunnel according to the seismic signal correlation model includes: acquiring seismic single shot data of one or more tunnel construction positions; classifying the reflected wave homophase axes in the seismic single shot data of each tunnel construction position to obtain the label data of the reflected wave homophase axes of each tunnel construction position; training the label data of the reflection wave homophase axis of each tunnel construction position and the seismic single shot data of each tunnel construction position based on a U-Net convolution neural network algorithm, and establishing a seismic signal correlation model; acquiring seismic single shot data of a current tunnel construction position; identifying and positioning a reflected wave event in the seismic single shot data of the current tunnel construction position according to the seismic signal correlation model; and according to the earth surface elevation, the surrounding rock information of the construction area and the reflected wave homophase axis identification and positioning result in the seismic single-shot data of the current tunnel construction position, the abnormal body position in front of the current tunnel construction position is presumed.
Based on the same inventive concept, the present invention also provides a computer-readable storage medium for performing the following steps: based on AlexNet convolutional neural network model algorithm, removing bad channel data from data collected by an observation system, and filtering; identifying the first-motion wave based on a U-Net convolution neural network model algorithm, and estimating the speed of the direct wave; performing linear dynamic correction on the original data without the bad track data according to the first arrival wave identification result; carrying out equal travel time section superposition on the data after linear dynamic correction; and carrying out reflection layer pickup on the superposed data through a deep learning algorithm, thereby inferring the area of the abnormal body at the first tunnel construction position.
In an embodiment of the present invention, the computer readable storage medium is further configured to perform the following steps: acquiring seismic single shot data of one or more tunnel construction positions; classifying the reflected wave homophase axes in the seismic single shot data of each tunnel construction position to obtain the label data of the reflected wave homophase axes of each tunnel construction position; training the label data of the reflection wave homophase axis of each tunnel construction position and the seismic single shot data of each tunnel construction position based on a U-Net convolution neural network algorithm, and establishing a seismic signal correlation model; acquiring seismic single shot data of a current tunnel construction position; identifying and positioning a reflected wave event in the seismic single shot data of the current tunnel construction position according to the seismic signal correlation model; and according to the earth surface elevation, the surrounding rock information of the construction area and the reflected wave homophase axis identification and positioning result in the seismic single-shot data of the current tunnel construction position, the abnormal body position in front of the current tunnel construction position is presumed.
Compared with the prior art, the tunnel advance geological prediction method and the computer readable storage medium provided by the invention have the advantages that the seismic data of the single construction position of the tunnel are quickly identified based on the deep learning algorithm, the dependence on the professional knowledge of data processing personnel is low, the interference of the processing result by the external influence is small, and the intelligent, quick and low-cost processing of a large amount of data can be realized. Preferably, in an embodiment, a tunnel seismic wave field propagation rule is utilized, and a central symmetric observation system, particularly a radial observation system, is selected, so that a large amount of reflected wave information with higher reliability and stronger authenticity can be obtained, and meanwhile, the influence on data processing and interpretation is small. Preferably, in an embodiment, the characteristic that a reflected wave event formed by the same anomaly appears to be different in travel time on single shot data acquired at different detection positions is utilized, after a detection result of one or more previous construction positions is obtained, a correlation model can be established at subsequent construction positions by utilizing the characteristic similar to the waveform, specifically, the correlation model for acquiring data at different construction positions is established based on a deep learning algorithm, so that the problems of low efficiency and poor reliability of manual identification of special waveforms in a large amount of data can be avoided, the rapid and accurate identification of characteristic waveforms in the large amount of data can be realized, and the tunnel forecast seismic data at different tunnel construction positions can be visually contrasted and analyzed in advance, so that the method is very intelligent.
Drawings
Fig. 1 is a block diagram of steps of a method for forecasting a tunnel ahead according to an embodiment of the present invention;
FIG. 2 is a diagram of a tunnel space seismic field energy distribution with tunnel face excitation of a seismic source according to an embodiment of the invention;
FIG. 3 shows a centrosymmetric observation system 3a, its three-dimensional seismic wavefield energy distribution map 3 b;
FIG. 4 shows a non-centrosymmetric observation system 4a, its three-dimensional seismic wavefield energy distribution map 4 b;
FIG. 5 shows a matrix orthogonal observation system 5a, an octagonal orthogonal observation system 5b, a radial observation system 5c, a non-orthogonal observation system 5d, and an observation system parameter table 5 e;
FIG. 6 shows shot and geophone link azimuthal profiles for four types of three-dimensional area observation systems;
FIG. 7 is a graph showing the distribution of the number of coverage for four types of three-dimensional area observation systems;
FIG. 8 is a graph of the effect of different numbers of lines or included angles of lines and spacing of shots on azimuthal distribution uniformity for a radial observation system in accordance with an embodiment of the present invention;
FIG. 9 is a graph of the effect of different shot line counts or shot angle and shot spacing on coverage distribution uniformity for a radial observation system in accordance with an embodiment of the present invention;
FIG. 10 shows three-dimensional reflection seismic signal raw data 10a, a single shot record spectrum curve 10b, a single-channel time spectrum 10c of a 4 th channel, and a single-channel time spectrum 10d of a 5 th channel acquired by a radial observation system;
FIG. 11 shows a graph 11a of the loss function as a function of the number of iterations for the AlexNet convolutional neural network model training process, a valid trace prediction threshold curve 11 b;
FIG. 12 shows the actual data 12a before the bad track culling and the actual data 12b after the bad track culling;
fig. 13 shows a loss function value variation curve 13a with iteration times during the training process of the U-Net convolutional neural network, the first arrival identification data and identification result 13b of a certain single shot record, the first arrival identification data and identification result 13c of another single shot record;
FIG. 14 shows data 14a before linear motion correction, and data 14b after linear motion correction;
FIG. 15 is a two-dimensional horizontal slice view of a stack of iso-travel time sections according to an embodiment of the present invention;
FIG. 16 is a schematic diagram of identifying similar regions of multi-shot records using a principal component analysis algorithm, according to an embodiment of the present invention;
fig. 17 shows an automatic recognition result 17a of a similar region of a multi-shot record based on deep learning and a reflected wave recognition result vertical section 17 b;
FIG. 18 is a block construction report diagram according to an embodiment of the present invention;
FIG. 19 is a numerical simulation model diagram of four different construction locations corresponding to four different times in accordance with an embodiment of the present invention;
FIG. 20 is a sample of deep learning training and label data for a construction location corresponding to time T1, in accordance with an embodiment of the present invention;
FIG. 21 is a sample of deep learning training and label data for a construction location corresponding to time T2, according to an embodiment of the present invention;
fig. 22 shows deep learning training samples and tag data of a construction site corresponding to time T3 according to an embodiment of the present invention.
FIG. 23 is a sample of deep learning training and label data for a construction location corresponding to time T4, according to an embodiment of the present invention;
FIG. 24 is a schematic illustration of a modification of a single acquired data processing result according to an embodiment of the invention;
FIG. 25 is a simulation diagram of data acquisition of four tunnels at different construction stages according to an embodiment of the present invention;
FIG. 26 is data collected experimentally according to an embodiment of the present invention;
FIG. 27 is a graph of loss function values versus iteration number according to one embodiment of the invention;
fig. 28 is a result of recognizing a step waveform according to an embodiment of the present invention.
Detailed Description
The following detailed description of the present invention is provided in conjunction with the accompanying drawings, but it should be understood that the scope of the present invention is not limited to the specific embodiments.
Throughout the specification and claims, unless explicitly stated otherwise, the word "comprise", or variations such as "comprises" or "comprising", will be understood to imply the inclusion of a stated element or component but not the exclusion of any other element or component.
In order to overcome the problems of the prior art, the tunnel advanced geological prediction method provided by the following implementation mode provides a rapid design and arrangement method of a tunnel face observation system by combining a tunnel seismic wave field propagation rule and an energy distribution rule, so that rapid arrangement of geophones is realized, and abnormal body reflection information in front of a tunnel face with higher signal-to-noise ratio and higher reliability is obtained under the condition of reducing influence on a tunnel construction process and tunnel face damage; and multiple data acquisition is carried out in the tunnel construction process, and more real abnormal body space position information is finally obtained on the basis of a deep learning technology, a seismic data rapid intelligent processing technology and an abnormal body rapid intelligent identification technology aiming at the obtained large amount of reflection seismic data.
Fig. 1 is a method for tunnel advanced prediction according to an embodiment of the present invention, the method comprising: exciting and receiving three-dimensional seismic wave signals based on an observation system on a tunnel face; and processing the data acquired by the observation system based on a deep learning algorithm so as to obtain the position of the abnormal body in the tunnel. Specifically, the method includes steps S1 to S8.
In step S1, a three-dimensional seismic observation system is placed on the tunnel face.
The inventors first conducted the following studies when setting up an observation system.
Fig. 2 is a diagram showing the energy distribution of the seismic wave field in the tunnel space when the tunnel face excites the seismic source, wherein the x axis is the direction in which the tunnel extends horizontally, the z axis is the direction perpendicular to the ground, and the tunnel represents the tunnel. It can be seen from fig. 2 that the energy distribution of the seismic wave field in the space on both sides of the tunnel is axisymmetric with respect to the excitation point when the tunnel face excites the seismic source. Therefore, in the actual arrangement process of the observation system, whether the observation system is reasonable or not is judged by evaluating the energy distribution uniformity of the observation system to the seismic wave field in the tunnel space.
The inventor finds that the observation system meeting the central symmetry can avoid the difference of the energy distribution of the seismic wave field in the tunnel space caused by the arrangement of the observation system.
Fig. 3 shows a centrosymmetric observation system 3a, a three-dimensional seismic field energy distribution map 3b of the observation system, a planar seismic field energy contour map 3c of the observation system at Z50, and a planar seismic field energy contour map 3d of the observation system at y 50, respectively. Wherein the y-axis is a direction parallel to the ground and perpendicular to the X-axis, and the observation system comprises a plurality of geophones (shown by inverted triangles in the figure) and a plurality of seismic wave generation devices (shown by dots in the figure). Comparing the boxes in the contour map 3c and the contour map 3d, it can be shown that the energy distribution of the tunnel space seismic wave field of the centrosymmetric observation system is very uniform.
Fig. 4 shows a non-centrosymmetric observation system 4a, its three-dimensional seismic field energy distribution 4b, its planar seismic field energy contour diagram 4c at Z50, and its planar seismic field energy contour diagram 4d at y 50. Comparing the boxes in the contour map 4c and the contour map 4d can show that the energy distribution of the tunnel space seismic wave field of the centrosymmetric observation system is different.
As can be seen from FIGS. 3 and 4, the energy distribution of the tunnel space seismic wave field of the centrosymmetric observation system is more uniform. Therefore, the observation system is preferably arranged to be a centrosymmetric observation system in the present embodiment, so that the propagation law of the tunnel seismic wave field can be utilized to acquire data with stronger reliability and higher signal-to-noise ratio.
In addition, the inventor also finds that when the covering times of the observation system are more uniform and the distribution of the azimuth angle of the connecting line of the shot point and the wave detection point is more uniform, the signals received by the observation system are more consistent with the propagation rule of the tunnel seismic wave field, so that the signal-to-noise ratio and the reliability of the original data acquired by the observation system can be further improved.
Specifically, the inventors analyzed four different types of three-dimensional area observation systems. As shown in fig. 5, four different types of three-dimensional area observation systems are a matrix orthogonal observation system 5a, an octagonal orthogonal observation system 5b, a radial observation system 5c, and a non-orthogonal observation system 5 d. Fig. 5 also shows a layout parameter table 5e of the four observation systems. Fig. 6 is a view showing azimuthal angle distribution of connecting lines of shot points and geophone points of the four types of three-dimensional area observation systems. Fig. 7 is a distribution diagram of the number of times of coverage of the four types of three-dimensional area observation systems.
From fig. 6 and 7, it can be seen that the uniformity of the coverage times, shot point and geophone point link azimuth distribution of the radial observation system is the strongest.
It should be further noted that, in the actual exploration process, on the premise of meeting the exploration requirement, for the radial observation system, in order to reduce the arrangement cost and the arrangement duration of the observation system, the distribution uniformity of the azimuth angles of the shot point and the geophone point connecting lines and the distribution uniformity of the coverage times are both considered, a proper shot line (a shot point connecting line passing through the center) and a proper signal receiving line (a geophone point connecting line passing through the center) are selected, and the number of a proper seismic wave generating device (a shot point) and a proper number of geophones (geophone points) are selected. FIG. 8 is a graph showing the effect of different numbers of lines or included angles of lines and pitches on azimuthal distribution uniformity for a radial observation system. FIG. 9 is the effect of different shot line numbers or shot angle and shot spacing of the radial observation system on the uniformity of coverage distribution.
After the observation system is set, in step S2, the observation system acquires three-dimensional seismic data: at the time of T1 at a certain construction position of the tunnel face, the observation system performs seismic wave excitation, and the geophone receives signals, so that three-dimensional seismic data acquisition is realized.
Fig. 10 shows three-dimensional reflection seismic signal raw data 10a, a single shot spectral curve 10b, a single-channel time spectrum 10c of the 4 th channel, and a single-channel time spectrum 10d of the 5 th channel acquired by the radial observation system according to the present embodiment. The single-channel time frequency spectrum is marked with a region of the first arrival wave and a region where the effective reflected wave possibly exists.
In step S3, based on the AlexNet convolutional neural network algorithm, bad track elimination is performed on the collected original data. Specifically, an AlexNet convolutional neural network model is trained, wherein the AlexNet convolutional neural network model takes a seismic wavelet field as a reference according to seismic exploration resolution, a quarter of a wavelet wavelength as a convolution kernel length, and an eighth of the wavelet wavelength as a pooling layer length. And then bad track elimination is carried out based on the trained model.
In order to better distinguish between valid tracks and bad tracks (empty track data and noisy data) and thereby improve the effect of bad track culling,preferably, in the embodiment, in the bad track elimination, an effective track prediction threshold f (x) is further set, wherein
Wherein, N is a label type (i.e. valid data type), pre (i) represents the probability prediction value of AlexNet belonging to a certain type label (valid data) for each data, F (x) represents the sum of all the label prediction values of each data, F (x)maxRepresents the sum of the maximum all-label predictors of all participating predictors.
Fig. 11 shows a curve 11a of a loss function of the AlexNet convolutional neural network model training process according to the present embodiment as a function of the number of iterations, and a curve 11b of an effective trace prediction threshold.
As can be seen from the curve 11a, the loss function value gradually decreases as the number of iterations increases, and after the number of iterations is greater than 400, the loss function value tends to 0, indicating that the AlexNet convolutional neural network model tends to be stable here, thereby ending the training.
According to the training sample data, the upper limit and the lower limit of the predicted value of the valid track record corresponding to different sample labels can be obtained, as shown in a curve 11b, the upper limit and the lower limit are used as valid track prediction threshold values, if the predicted value of the data is in the range, the predicted value of the data is judged to be a valid value, and if the predicted value is out of the range, the predicted value of the data is judged to be noise. Since the original seismic data which is not processed is adopted for constructing the bad track identification network, partial noise influence exists in the predicted value, therefore, preferably, in the embodiment, the effective track prediction threshold is expanded to a certain extent according to the signal-to-noise ratio, the data processing in the preferred embodiment expands the range shown by 11b by 50%, and finally the upper limit of the effective data is determined to be 63.2, and the lower limit is-94.5. And leading the 48 common receiving point shot sets into the trained convolutional neural network for bad channel identification.
Fig. 12 shows the actual data 12a before the defective track elimination and the actual data 12b after the defective track elimination, respectively.
The actual data 12a before bad track elimination is 48 original single-track recorded data, the upper curve is a predicted value curve of different single-track records, the horizontal axis represents different common receiving point shot set records, the vertical axis represents a bad track identification predicted value based on an AlexNet convolutional neural network algorithm, the smaller the predicted value is, the greater the possibility that the predicted value is a bad track is, and the smaller the possibility that the predicted value is a bad track is. Comparing the actual data 12a after the bad track elimination with the actual data 12b before the bad track elimination shows that the high-frequency noise, the sine noise, the low-frequency noise and the strong-amplitude noise are effectively eliminated.
In step S4, the data after the bad track rejection is subjected to filter processing.
In step S5, first arrival recognition is performed based on the U-Net convolutional neural network algorithm, and the direct wave velocity is estimated. Firstly, training a U-Net convolutional neural network model, then identifying a first arrival wave based on the trained convolutional neural network model, and then estimating the speed of the direct wave.
Fig. 13 shows a variation curve 13a of the loss function value with the number of iterations in the U-Net convolutional neural network training process, first arrival identification data and identification result 13b of a certain single shot record, first arrival identification data and identification result 13c of another single shot record, respectively.
The loss function value is shown in a curve 13a of variation of the loss function value along with the iteration number, the loss function value of the convolutional neural network gradually decreases and finally tends to be unchanged along with the increase of the iteration number, wherein the loss function value decreases at a high speed when the iteration number is 0-30, the loss function value decreases at a low speed within a range of 30-150, and the loss function value tends to 0 after more than 170 times, which indicates that the U-Net network tends to be stable at the moment.
From the first-arrival wave identification data and identification result 13b of a single shot record and the first-arrival wave identification data and identification result 13c of another single shot record, the deep learning technique based on the U-Net algorithm effectively identifies the first-arrival wave position. In addition, as shown in FIG. 13b, the direct wave velocity was 4000 m/s.
In step S6, linear motion correction is performed: linear motion correction is performed on the data after the bad track rejection according to the first arrival wave position identification result, and fig. 14 shows data 14a before linear motion correction and data 14b after linear motion correction, so that the linear motion correction result is good.
The travel-time cross-section superimposition is performed in step S7. Specifically, the velocity scan is performed using the first arrival apparent velocity as the initial velocity, and when the velocity on the grid node is correct, a plurality of travel-time profiles are superimposed at the point and appear as a strong energy value, whereas the velocity is a weak energy value.
Fig. 15 is a two-dimensional horizontal slice view in which the equal travel time sections of the present embodiment are superimposed. Observing the two-dimensional horizontal slice, the x can be divided into three parts of 1, 2 and 3 according to the energy change of the reflection in-phase axis, wherein the x is 50-300.
In step S8, the reflection layers are picked up based on a deep learning algorithm, and specifically, similar regions of the multi-shot records are identified using a principal component analysis algorithm. The similar region is presumed to be the region of the geological anomaly. As shown in fig. 16, the region framed by the rectangle is estimated as the region where the abnormal body is located.
Fig. 17 shows the automatic recognition result 17a and the reflected wave recognition result vertical section 17b of the similar region of the multi-shot record based on the deep learning. Similar region positions (regions shown as black boxes in the figure) are represented by the phase inversion of the in-phase axis. The horizontal section at y 100m has stronger energy, and the vertical section at z 100m has weaker energy, which is mainly because the azimuth angle of the connecting line of the shot point and the geophone point is mainly distributed in the z direction, and is less distributed in the y direction, so that the energy distribution in the direction is stronger. Reflected wave pickup was performed simultaneously in conjunction with three regions in the two-dimensional horizontal slice of fig. 16, and the pickup result is shown in fig. 17. The in-phase axis position and shape are divided into three parts, x 100 to 150, x 200 to 250, and x 300, and the presence of an abnormal body structure in these three regions is estimated.
In order to verify the estimation result, fig. 18 is a construction report diagram of the 2+966-3+311 section of the present embodiment, and the triangle position in the diagram is the position detected this time. As can be seen from the construction report, a sausage-shaped quartzite vein and a granite pegmatite vein are penetrated in the rock at a position 60 m (3 +074 in fig. 18) in front of the tunnel face, and the positions of the sausage-shaped quartzite vein and the granite pegmatite vein are coincided with the reflection wave abnormal body at a position 55 m away from the tunnel face in the reflection wave recognition result vertical section 17b (as shown by a rectangular frame 1 in fig. 17 b); a Huanggang rock vein interpenetration with the width of 0.1-0.5m and the length of 2-10m and intruding along the fissure exists in the surrounding rock 260 meters away from the face, and in the spatial position, the Huanggang rock vein interpenetration is matched with the same phase axis of the place 250 meters in front of the face in fig. 17b (as shown by a rectangular frame 3 in fig. 17 b); meanwhile, a strong energy reflected wave exists between 60 meters and 250 meters from the tunnel face, and compared with the same phase axis at 60 meters from the tunnel face (as shown in a rectangular frame 2 in fig. 17 b), the phase change rule is similar, and meanwhile, as can be seen from subsequent construction reports, in the pile number 2+966-3+240, the integrity and stability of surrounding rock are poor, and the fold deformation is serious, so that the reflected wave is inferred to be a multiple wave formed by the same phase axis in the rectangular frame 1 in fig. 17 b.
In addition, the tunnel is a dynamic construction process, the position of the detection target body is not changed in the whole construction process, and the detection position changes along with the construction progress. The inventor finds that the characteristic enables the reflection wave in-phase axis formed by the same abnormal body to show that the traveling time is different on the single shot data collected at different detection positions, and the wave forms are similar. Therefore, in order to improve the efficiency of estimating the abnormal body in the subsequent construction process, in a preferred embodiment, a seismic signal correlation model acquired at different construction positions of the tunnel is constructed based on a U-Net convolution neural network algorithm, and a computer is used for intelligently learning characteristic waves of a large amount of data to realize intelligent identification and extraction of the same abnormal body reflection information in different seismic signals, so that the influence of external factors such as ground surface coverings, ground surface layers, tunnel walls and the like on the identification of the abnormal body in front of the tunnel face is eliminated.
Specifically, after the estimation of the abnormal body position at the first time is completed, the estimation of the abnormal body at each subsequent construction position is performed as follows: firstly, acquiring seismic single shot data of one or more tunnel construction positions; classifying the reflected wave homophase axes in the seismic single shot data of each tunnel construction position to obtain the label data of the reflected wave homophase axes of each tunnel construction position; training the label data of the reflection wave homophase axis of each tunnel construction position and the seismic single shot data of each tunnel construction position based on a U-Net convolution neural network algorithm, and establishing a seismic signal correlation model; acquiring seismic single shot data of a current tunnel construction position; identifying and positioning a reflected wave event in the seismic single shot data of the current tunnel construction position according to the seismic signal correlation model; and according to the earth surface elevation, the surrounding rock information of the construction area and the reflected wave homophase axis identification and positioning result in the seismic single-shot data of the current tunnel construction position, the abnormal body position in front of the current tunnel construction position is presumed.
It should be noted that the more sample data is trained, the more accurate the result of the estimation of the abnormal body is. Therefore, after the abnormal position of the first construction position is estimated, the subsequent abnormal position of the second construction position is obtained according to the process; then, when the position of the abnormal body at the third construction position is estimated, preferably, the single shot data and the label data of the first construction position and the second construction position are used as training samples to train; by analogy, when the position of the abnormal body at the fourth construction position is estimated later, the single shot data and the tag data at the first construction position, the second construction position and the third construction position should preferably be trained as training samples. Thus, a more accurate estimation result of the abnormal body can be obtained.
The intelligent interpretation technology of the tunnel advance forecast data is explained by numerical simulation data. Fig. 19 is a numerical simulation model of four different construction positions at different times. Fig. 20 shows the deep learning training samples and the label data at the construction position corresponding to time T1. Fig. 21 shows deep learning training samples and tag data of the construction location corresponding to time T2. Fig. 22 shows deep learning training samples and tag data of the construction site corresponding to time T3. Fig. 23 shows deep learning training samples and tag data of the construction location corresponding to time T4. In order to increase the randomness of the sampling data, the distances between the four working face surfaces at the time of construction are not equal, the working face corresponding to the time T1 is located at the position x being 25m, the working face corresponding to the time T2 is located at the position x being 47m, the working face corresponding to the time T3 is located at the position x being 53m, and the working face corresponding to the time T4 is located at the position x being 73 m. Each single shot recorded 20 tracks of data, with a track spacing of 0.5 m. The initial multi-time single-shot data homophase axis relevance numerical simulation is established by using single-shot data at the time T1. As can be seen from the T1 time data, two types of reflected wave event axes, namely a negative apparent velocity event (75 ms) and a small apparent dip event (50 ms), exist in the reflected wave region, and accordingly, the event axes on the T1 single shot record are divided into two types, the 50ms and small apparent dip event are regarded as a first type of event axis, and the 75ms negative apparent velocity event axis is regarded as a second type of event axis, so as to produce T1 time tag data (as shown in fig. 20). The shot gather records show that the travel time of the same-phase axis of the reflected wave formed by the abnormal body gradually becomes smaller along with the approach of the tunnel to the position of the abnormal body, and the depth change of the horizontal stratum is small in the range of x-25 and x-75, so that the travel time of the same-phase axis of the reflected wave formed by the influence of the abnormal body is almost unchanged. And then automatically identifying shot gather record homophase axes of different tunnel construction positions based on a deep learning technology, as shown in figures 21-23. It can be seen that the multi-time seismic data correlation network model established based on the deep learning algorithm can effectively identify and reflect the reflection wave event change rule formed by the same abnormal body. Secondly, because the waveform similarity of the multiples and the primaries is high, when a multi-time shot gather record relevance model is constructed by only using T1 data as training data, the prediction result is poor, the primaries and the multiples are difficult to separate, and a large amount of multiple information (shown in FIG. 21) is remained in the prediction result; however, when training sample data is increased, the multi-time shot gather record relevance model constructed based on multiple data can effectively distinguish primary waves and multiple waves (as shown in fig. 22 and 23), the anti-interference capability of waveform identification is also enhanced, for example, in the shot gather record at the time T4, reflected waves of effective abnormal bodies are mixed with direct waves, when data at 3 times are adopted as training data, the identification capability of a same-phase axis is obviously improved, which shows that the prediction result is positively correlated with the training data set quantity, and the larger the data quantity participating in model training, the stronger the representativeness, and the more accurate the obtained prediction result. On the basis, the data processing result of single acquisition is corrected, and as shown in fig. 24, the interferences of the earth surface, the stratum parallel to the axis of the tunnel and the like are effectively filtered.
In order to verify the effect of identifying the characteristic waves by establishing the correlation model, a description will be given using a physical experiment. Firstly, two measuring lines are arranged by utilizing a smartsolo 5Hz node type seismograph, each measuring line has 20 receiving points, and the track interval is 0.5 m. The walking process of the pedestrian along the measuring line is similar to the forward tunneling process of the tunnel, and when the pedestrian walks to a certain receiving point, the pedestrian can consider that the tunnel is constructed to a certain position, so that the tunneling process of the tunnel is simulated through a physical experiment. In the data acquisition process, the tunnel is walked back and forth for 2 times along the measuring line, so that the data acquisition of the tunnel in different construction stages is simulated for four times, and the data acquisition is shown in figure 25. The step recorded in the first walk of the pedestrian is taken as the reflected wave received by the tunnel at the time T1 (indicated by the m-box in fig. 26), and the step recorded in the subsequent three walks is taken as the reflected wave signal acquired at the subsequent different time (indicated by the n-box in fig. 26). To increase the data difference, the walker holds one 5kg weight shot in the first walk and two 5kg weight shots in the subsequent three walks. The data in the m frames are used as training data, the track number is used as a sample label for training, the data in the n frames are identified on the basis, and a curve of the loss function value along with the change of the iteration times is shown in fig. 27. The recognition result shown in fig. 28 indicates that the neural network model constructed by using the data at the time T1 effectively recognizes the position of the subsequent step waveform, and this experiment indicates that the method can be used for establishing a single shot data relevance model acquired at different times in a tunnel, so as to realize characteristic wave recognition.
Based on the same inventive concept, the present embodiment also provides a computer-readable storage medium for performing the following steps: based on AlexNet convolutional neural network model algorithm, removing bad channel data from data collected by an observation system, and filtering; identifying the first-motion wave based on a U-Net convolution neural network model algorithm, and estimating the speed of the direct wave; performing linear dynamic correction on the original data without the bad track data according to the first arrival wave identification result; carrying out equal travel time section superposition on the data after linear dynamic correction; and carrying out reflection layer pickup on the superposed data through a deep learning algorithm, thereby inferring the area of the abnormal body at the first tunnel construction position.
Preferably, the computer-readable storage medium of this embodiment is further configured to perform the following steps: acquiring seismic single shot data of one or more tunnel construction positions; classifying the reflected wave homophase axes in the seismic single shot data of each tunnel construction position to obtain the label data of the reflected wave homophase axes of each tunnel construction position; training the label data of the reflection wave homophase axis of each tunnel construction position and the seismic single shot data of each tunnel construction position based on a U-Net convolution neural network algorithm, and establishing a seismic signal correlation model; acquiring seismic single shot data of a current tunnel construction position; identifying and positioning a reflected wave event in the seismic single shot data of the current tunnel construction position according to the seismic signal correlation model; and according to the earth surface elevation, the surrounding rock information of the construction area and the reflected wave homophase axis identification and positioning result in the seismic single-shot data of the current tunnel construction position, the abnormal body position in front of the current tunnel construction position is presumed.
In summary, according to the tunnel advance geological prediction method and the computer-readable storage medium of the embodiment, the seismic data at the single construction position of the tunnel is quickly identified based on the deep learning algorithm, the dependence on the professional knowledge of data processing personnel is low, the interference of the processing result by the external influence is small, and the intelligent, quick and low-cost processing of a large amount of data can be realized. Preferably, a tunnel seismic wave field propagation rule is utilized, a seismic observation system on a tunnel face is firstly provided, a centrosymmetric observation system is selected, particularly a radial observation system is selected, the observation system can obtain a large amount of reflected wave information with higher reliability and stronger authenticity through theoretical and practical verification, and meanwhile, the influence on data processing and interpretation is small. Preferably, in an embodiment, the characteristic that a reflected wave event formed by the same anomaly appears as different traveling time on single shot data acquired at different detection positions is utilized, and after a detection result of one position is obtained, a correlation model can be established at subsequent construction positions by utilizing the characteristic that the waveform is similar, specifically, the correlation model of the data acquired at different construction positions is established based on a deep learning algorithm, so that the problems of low efficiency and poor reliability of manual identification of special waveforms in a large amount of data can be avoided, and the rapid and accurate identification of characteristic waveforms in the large amount of data can be realized.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing descriptions of specific exemplary embodiments of the present invention have been presented for purposes of illustration and description. It is not intended to limit the invention to the precise form disclosed, and obviously many modifications and variations are possible in light of the above teaching. The exemplary embodiments were chosen and described in order to explain certain principles of the invention and its practical application to enable one skilled in the art to make and use various exemplary embodiments of the invention and various alternatives and modifications as are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the claims and their equivalents.
Claims (6)
1. A method for advance geological prediction of a tunnel is characterized by comprising the following steps:
exciting and receiving three-dimensional seismic wave signals based on an observation system on a tunnel face; and
processing the data acquired by the observation system based on a deep learning algorithm to obtain the position of the abnormal body in the tunnel;
the three-dimensional seismic wave signal excitation and reception based on the observation system on the tunnel face comprises the following steps:
arranging a three-dimensional earthquake observation system on the tunnel face;
at a first tunnel construction position, the three-dimensional earthquake observation system excites earthquake waves and receives reflected waves so as to acquire data;
wherein the processing the data collected by the observation system based on the deep learning algorithm to obtain the position of the abnormal body in the tunnel comprises:
based on AlexNet convolutional neural network model algorithm, removing bad channel data from the collected data, and filtering;
identifying the first-motion wave based on a U-Net convolution neural network model algorithm, and estimating the speed of the direct wave;
performing linear dynamic correction on the original data without the bad track data according to the first arrival wave identification result;
carrying out equal travel time section superposition on the data after linear dynamic correction; and
carrying out reflection layer pickup on the superposed data through a deep learning algorithm, thereby inferring the area of the abnormal body at the first tunnel construction position;
the method for removing the bad track data from the collected data based on the AlexNet convolutional neural network model algorithm comprises the following steps:
training the AlexNet convolutional neural network model;
removing bad track data from the acquired data by adopting the trained AlexNet convolutional neural network model;
wherein, the step of adopting the trained AlexNet convolutional neural network model to remove the bad channel comprises the following steps:
judging the data with the data prediction value outside the range of the effective channel prediction threshold f (x) as noise data and removing the noise data,
，
，
wherein, N is label type, pre (i) represents the possibility predicted value of AlexNet belonging to a certain type label for each data, F (x) represents the sum of all label predicted values of each data, F (x)maxRepresenting the sum of the maximum all-label predictors of all participating predictors;
wherein the step of performing reflector picking on the superimposed data through a deep learning algorithm comprises:
identifying a similar region of the multi-shot records by adopting a principal component analysis algorithm, wherein the similar region is presumed to be a region where the geological abnormal body is located;
the method for forecasting the advance geology of the tunnel further comprises the following steps:
after the estimation of the position of the abnormal body at the first tunnel construction position is completed, a seismic signal correlation model is built based on a U-Net convolution neural network model algorithm, and the estimation of the abnormal body is carried out at other construction positions of the same tunnel according to the seismic signal correlation model.
2. The method for advanced geological forecasting of tunnels as claimed in claim 1, wherein the observation system is centrally symmetric as a whole.
3. A method for advanced geological prediction of a tunnel as claimed in claim 2, wherein the geophones of said observation system are arranged radially from the center to the periphery, and the seismic generators of said observation system are also arranged radially from the center to the periphery.
4. The method for look-ahead geological prediction of a tunnel of claim 1, wherein the AlexNet convolutional neural network model is based on a seismic wavelet field according to seismic exploration resolution, with one-quarter of a wavelet wavelength as the length of the convolutional kernel and one-eighth of a wavelet wavelength as the length of the pooling layer.
5. The method for forecasting tunnel geology according to claim 1, wherein the building of the seismic signal correlation model based on the U-Net convolutional neural network model algorithm, and the estimation of the abnormal body at other construction positions of the same tunnel according to the seismic signal correlation model comprises the following steps:
acquiring seismic single shot data of one or more tunnel construction positions;
classifying the reflected wave homophase axes in the seismic single shot data of each tunnel construction position to obtain the label data of the reflected wave homophase axes of each tunnel construction position;
training the label data of the reflection wave homophase axis of each tunnel construction position and the seismic single shot data of each tunnel construction position based on a U-Net convolution neural network algorithm, and establishing a seismic signal correlation model;
acquiring seismic single shot data of a current tunnel construction position;
identifying and positioning a reflected wave event in the seismic single shot data of the current tunnel construction position according to the seismic signal correlation model;
and according to the earth surface elevation, the surrounding rock information of the construction area and the reflected wave homophase axis identification and positioning result in the seismic single-shot data of the current tunnel construction position, the abnormal body position in front of the current tunnel construction position is presumed.
6. A computer-readable storage medium based on the method for advanced geological prediction of tunnels according to any of claims 1 to 5, characterized in that it is adapted to perform the following steps:
based on AlexNet convolutional neural network model algorithm, removing bad channel data from data collected by an observation system, and filtering;
identifying the first-motion wave based on a U-Net convolution neural network model algorithm, and estimating the speed of the direct wave;
performing linear dynamic correction on the original data without the bad track data according to the first arrival wave identification result;
carrying out equal travel time section superposition on the data after linear dynamic correction; and
carrying out reflection layer pickup on the superposed data through a deep learning algorithm, thereby inferring the area of the abnormal body at the first tunnel construction position;
the method for removing the bad track data from the collected data based on the AlexNet convolutional neural network model algorithm comprises the following steps:
training the AlexNet convolutional neural network model;
removing bad track data from the acquired data by adopting the trained AlexNet convolutional neural network model;
wherein, the step of adopting the trained AlexNet convolutional neural network model to remove the bad channel comprises the following steps:
judging the data with the data prediction value outside the range of the effective channel prediction threshold f (x) as noise data and removing the noise data,
，
，
wherein, N is label type, pre (i) represents the possibility predicted value of AlexNet belonging to a certain type label for each data, F (x) represents the sum of all label predicted values of each data, F (x)maxRepresenting the sum of the maximum all-label predictors of all participating predictors;
wherein the step of performing reflector picking on the superimposed data through a deep learning algorithm comprises:
identifying a similar region of the multi-shot records by adopting a principal component analysis algorithm, wherein the similar region is presumed to be a region where the geological abnormal body is located;
the method for forecasting the advance geology of the tunnel further comprises the following steps:
after the estimation of the position of the abnormal body at the first tunnel construction position is completed, a seismic signal correlation model is built based on a U-Net convolution neural network model algorithm, and the estimation of the abnormal body is carried out at other construction positions of the same tunnel according to the seismic signal correlation model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011078912.2A CN112305591B (en) | 2020-10-10 | 2020-10-10 | Tunnel advanced geological prediction method and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011078912.2A CN112305591B (en) | 2020-10-10 | 2020-10-10 | Tunnel advanced geological prediction method and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112305591A CN112305591A (en) | 2021-02-02 |
| CN112305591B true CN112305591B (en) | 2022-04-29 |
Family
ID=74489584
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011078912.2A Active CN112305591B (en) | 2020-10-10 | 2020-10-10 | Tunnel advanced geological prediction method and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112305591B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113202481A (en) * | 2021-05-10 | 2021-08-03 | 中铁第四勘察设计院集团有限公司 | Method and device for acquiring geological information, electronic equipment and storage medium |
| CN114609668B (en) * | 2022-03-11 | 2023-09-19 | 西安交通大学 | High-quality reservoir identification method, device and equipment based on scattering transformation and neural network and storage medium |
| CN115201901A (en) * | 2022-06-30 | 2022-10-18 | 中铁第四勘察设计院集团有限公司 | Method, device and equipment for determining tunnel wave front travel time and readable storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105676279A (en) * | 2016-01-18 | 2016-06-15 | 长江地球物理探测（武汉）有限公司 | Earthquake reflection data collection method with concentric-circle equivalent shot-geophone distance |
| CN111929728A (en) * | 2020-08-13 | 2020-11-13 | 高军 | Three-dimensional three-component advanced refined geological prediction method |
Family Cites Families (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105259572B (en) * | 2015-07-23 | 2018-06-19 | 成都理工大学 | The seismic facies computational methods classified automatically based on seismic multi-attribute parametrical nonlinearity |
| CN106650951B (en) * | 2016-12-29 | 2019-06-21 | 中国科学技术大学 | Multi-channel measurement data automatic screening method based on time domain overall situation similarity |
| CN109557583B (en) * | 2017-09-26 | 2020-12-01 | 中国石油化工股份有限公司 | Seismic attribute extraction method and system |
| CN107807387B (en) * | 2017-10-31 | 2019-08-27 | 中国科学技术大学 | Acquisition methods when seismic first break neural network based is walked |
| KR102027252B1 (en) * | 2017-12-28 | 2019-10-01 | (주)대우건설 | Methods for differentiation of earthquake signal and prediction of earthquake intensity using randomly generated artificial seismic training data for an arbitrary zone |
| CN108337153B (en) * | 2018-01-19 | 2020-10-23 | 论客科技（广州）有限公司 | Method, system and device for monitoring mails |
| CN111478783B (en) * | 2019-01-23 | 2023-01-13 | 中国移动通信有限公司研究院 | Method and equipment for configuring wireless transmission parameters |
| CN111123351B (en) * | 2019-11-29 | 2022-03-15 | 中铁工程服务有限公司 | Advanced forecasting system and method for shield construction |
| CN110988981B (en) * | 2019-12-23 | 2021-09-14 | 山东大学 | Phased array sound wave advanced prediction system and method suitable for drilling and blasting method tunnel |
| CN111007566B (en) * | 2019-12-27 | 2020-12-18 | 西南石油大学 | Curvature-driven diffusion full-convolution network seismic data bad channel reconstruction and denoising method |
| CN111178320B (en) * | 2020-01-07 | 2020-11-17 | 中国矿业大学(北京) | Geological abnormal body recognition method and model training method and device thereof |
| CN111289250A (en) * | 2020-02-24 | 2020-06-16 | 湖南大学 | Method for predicting residual service life of rolling bearing of servo motor |
| CN111626355A (en) * | 2020-05-27 | 2020-09-04 | 中油奥博(成都)科技有限公司 | Unet + + convolutional neural network-based seismic data first arrival pickup method |
-
2020
- 2020-10-10 CN CN202011078912.2A patent/CN112305591B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105676279A (en) * | 2016-01-18 | 2016-06-15 | 长江地球物理探测（武汉）有限公司 | Earthquake reflection data collection method with concentric-circle equivalent shot-geophone distance |
| CN111929728A (en) * | 2020-08-13 | 2020-11-13 | 高军 | Three-dimensional three-component advanced refined geological prediction method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112305591A (en) | 2021-02-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112305591B (en) | Tunnel advanced geological prediction method and computer readable storage medium | |
| CN105607121B (en) | A kind of coal karst collapse col umn recognition methods and device | |
| CA2580355C (en) | Microseismic event detection and location by continuous map migration | |
| CN106646598B (en) | A kind of FAST-AIC methods microseism picking up signal method | |
| CN105785439B (en) | The Forecasting Methodology and device of small scale heterogeneous geologic body spatial distribution position | |
| CN105510880A (en) | Microseism focus positioning method based on double-difference method | |
| CN107290722B (en) | The localization method and device of microquake sources | |
| CN110529087B (en) | Method and device for evaluating hydraulic fracturing effect of stratum | |
| US10345463B2 (en) | Methods and systems for using known source events in seismic data processing | |
| CN103091709B (en) | Method and device for acquiring longitudinal wave and converted wave seismic data time matching relation | |
| CN104765064A (en) | Microseism interference imaging method | |
| CN103758511A (en) | Method and device for identifying hidden reservoir through underground reverse time migration imaging | |
| CN109856677A (en) | A kind of seismoelectric joint obtains the localization method of crack information | |
| CN104570078B (en) | A kind of solution cavity detection method of the similitude cross directional variations rate based on frequency domain inclination angle | |
| CN105425299A (en) | Method and apparatus for determining formation fracture distribution | |
| CN104199088B (en) | Incident angle gather extraction method and system | |
| CN112114358B (en) | Underground volcanic channel identification method based on three-dimensional seismic data representation | |
| CN108693560A (en) | A kind of scattering wave imaging method and system based on cross-correlation road | |
| CN110579799A (en) | Seismic acquisition observation method and system with equal travel time intervals | |
| CN108375794B (en) | VSP (vertical seismic profiling) slit-hole diffraction imaging technical method based on symmetrical observation | |
| CN107664771B (en) | A kind of microseism Full wave shape localization method based on likeness coefficient | |
| CN105785435A (en) | Seismic physical model experiment system and analog data collection method and device thereof | |
| CN104820240A (en) | Seismic tomography method based on WSNs | |
| CN108897046B (en) | Far detection sound wave reverse time migration imaging method and device | |
| Goertz-Allmann et al. | Integrating active with passive seismic data to best constrain CO2 injection monitoring |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |