TBM tunneling earthquake advanced detection observation system imaging method

Abstract

The application discloses a kind of based on TBM with the imaging method of seismic wave advanced detection observation system of excavation, by arranging detector in the excavation face, and roadway side help behind, constitute TMB with the seismic wave advanced detection observation system of excavation, make full use of the limited space of roadway, and using deep learning algorithm, construct wave field correction dataset, correct conventional reverse-time migration imaging process to form wave field data pair, then seismic interference processing is carried out to the data after correction, extract each seismic reflection wave common detector gather from the multichannel seismic wave received by detector, obtain the initial value wave travel time of each seismic reflection wave to reach detector, generate seismic interference record, carry out high-precision reverse-time migration imaging to seismic interference record, obtain the geological anomaly distribution in front of roadway.The application realizes the wave field correction of virtual observation system along the horizontal direction of excavation face, expands the migration aperture between shot-detection, improves the detection ability to small dip structure.

Description

Technical Field

[0001] This invention relates to the field of mine seismic data processing technology, and more specifically to an imaging method for a TBM (Tunnel Boring Machine) seismic advance detection and observation system. Background Technology

[0002] Due to the limited space in TBM construction tunnels, existing seismic observation systems are mainly deployed as linear observation systems behind the tunneling face. While these systems can effectively detect geological structures ahead of the tunneling face, their small migration aperture limits their effectiveness to detecting faults with large dip angles. The detection efficiency drops sharply when the fault dip angle is small. Furthermore, limitations in their migration imaging algorithms, which rely primarily on geometric migration, result in images formed by superimposing ellipsoids at the same spatial location. This leads to predominantly arc-shaped images, hindering precise spatial localization of anomalies.

[0003] Therefore, how to utilize TBM-based seismic advance detection data to achieve accurate and advanced prediction of geological conditions ahead of the tunneling face is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] In view of this, the present invention provides an imaging method for a TBM-based seismic advance detection and observation system, which can effectively solve the problem of the layout of the TBM-based seismic observation system. At the same time, the imaging method based on this observation system can effectively improve the accuracy of spatial positioning of anomalies in front of the tunnel excavation working face.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] An imaging method for a TBM (Tunnel Boring Machine) seismic advance detection and observation system includes the following steps:

[0007] Step 1: Deploy geophones in actual tunnels of different geological models to build a conventional observation system; In view of the characteristics of tunnels based on TBM construction, deploy high-precision three-component geophones on both sides of the TBM excavation face and behind the shield head.

[0008] Step 2: Construct a virtual observation system based on the conventional observation system, and perform acoustic wave field simulation on the conventional and virtual observation systems to obtain the actual seismic wave field records under the conventional observation system and the virtual seismic wave field records under the virtual observation system simulated by different geological models. Furthermore, construct the mapping relationship between the two based on a convolutional neural network to obtain the seismic wave field correction dataset during excavation.

[0009] Step 3: Collect measured seismic advance detection data during TBM excavation Based on the seismic wavefield correction dataset obtained in step 2, the corresponding seismic correction wavefield values ​​under the virtual observation system are determined using a convolutional neural network.

[0010] Step 4: Use seismic interferometry to correct the seismic correction wavefield values ​​in the virtual observation system. Cross-correlation processing is performed to obtain the processed seismic interferometric records.

[0011] Step 5: Process the seismic interferometry records Perform reverse time-shift imaging to obtain imaging values;

[0012] Step 6: Perform seismic advance migration imaging on the imaging value I calculated in Step 5 to obtain the advance prediction results of the geological structure ahead.

[0013] Preferably, for each geological structure's actual tunnel, acoustic wave field simulation is performed using both an actual tunnel observation system and a virtual tunnel observation system to obtain the actual seismic wave field record under the actual tunnel observation system and the virtual seismic wave field record under the virtual tunnel observation system. Then, a mapping relationship between the two is constructed based on a convolutional neural network. The specific steps are as follows:

[0014] Step 21: For each actual roadway, geophones are installed along the working face and the left and right sidewalls behind it to construct a conventional observation system. Acoustic wave field simulation is performed on the actual roadway, and Gaussian white noise is introduced to simulate the actual seismic wave field record M1 under the actual roadway observation system. obs The conventional observation system uses a geophone located behind the tunnel to map onto the left and right sides of the tunneling face after algorithm processing, i.e., perpendicular to the tunneling direction. The geophones are horizontally arranged to construct a virtual observation system. Acoustic wavefield simulation is performed, and Gaussian white noise is introduced to simulate the virtual seismic wavefield record M2 obtained from actual measurements. obs ;

[0015] Step 22: Data Preprocessing: Record the actual seismic wavefield M1 obs Virtual seismic wavefield record M2 obs The data is converted into actual and virtual seismic images in m×n two-dimensional image format;

[0016] Step 23: Construct a convolutional neural network;

[0017] Step 24: Determine the loss function: based on seismic wavefield records M1 obtained through two different observation methods. obs and M2 obs Calculate the loss function values ​​Loss1 and Loss2 for both respectively;

[0018]

[0019] In the formula, Loss1 and Loss2 represent the seismic wavefield loss functions during excavation under conventional and virtual observation systems, respectively; M 1N M 2N These represent actual forward propagation seismic records under conventional observation systems and virtual forward propagation seismic records under virtual observation systems, respectively. N represents the Nth geological model. Forward propagation seismic records refer to the records of the process of sound wave generation -> propagation -> detector reception during the acoustic wave field simulation.

[0020] Step 25: Training the network: Record the simulated actual seismic wavefield M1 obs Virtual seismic wavefield record M2 obs The loss function values ​​Loss1 and Loss2 are substituted into the convolutional neural network constructed in step 23 for training;

[0021] Step 26: Perform wavefield correction: Use a trained convolutional neural network to process the actual seismic wavefield records M1 simulated by different geological models. obs Perform corrections and obtain the corresponding seismic correction wavefield values ​​under the virtual observation system.

[0022] Step 27: Save the seismic correction wavefield values ​​from the virtual observation system in Step 26. And store it in the seismic wave field correction dataset during excavation.

[0023] Step 28: Seismic correction wavefield values ​​under the virtual observation system of all geological models Constructing a seismic wave field correction dataset during excavation Where N is a positive integer, representing the number of geological model types.

[0024] Preferably, the acoustic wave field simulation process includes: setting an initial velocity model; obtaining the actual wave equation forward modeling wave field record under a conventional observation system and the virtual wave equation forward modeling wave field record under a virtual observation system based on the initial velocity model; calculating the actual loss function under the conventional observation system based on the actual wave equation forward modeling wave field record, and calculating the virtual loss function under the virtual observation system based on the virtual wave equation forward modeling wave field record; obtaining the actual seismic wave field record based on the actual loss function, and obtaining the virtual seismic wave field record based on the virtual loss function.

[0025] Preferably, the convolutional neural network consists of an input layer, a convolutional layer, a fully connected layer, and an output layer, specifically:

[0026] Input layer: Input m×n two-dimensional image format;

[0027] Convolutional layer: According to the convolution calculation formula, a matrix of size l×l is applied to an m×n two-dimensional image to perform scanning operations;

[0028] Sampling layer: Removes unimportant samples from the m×n two-dimensional image to reduce the number of parameters;

[0029] Fully connected layer: Connects each current node to all nodes in the previous layer, such as the input layer and convolutional layer, to form a set of data mapping relationships.

[0030] Preferably, the specific process of step 4 is as follows:

[0031] Step 41: Select the geophone located at point A on the left side of the tunneling face as the reference signal path;

[0032] Step 42: Select point B from the remaining geophones on the sidewall of the tunneling face. The Green's functions of the signals received by the geophones at points A and B are as follows:

[0033]

[0034] Where G(A / x) represents the Green's function of the signal received at point A; G(B / x) represents the Green's function of the signal received at point B; d(x) is the distance the wave field propagates from each point on the cutterhead of the tunnel boring machine to point A or B when it acts as the seismic source; C is the cutterhead moving speed; V is the seismic wave propagation speed; x0 is the starting point of the tunnel boring machine's forward movement; x is the current distance the tunnel boring machine has moved; ω represents the angular frequency; i represents the imaginary part of the complex number; A(x) is the distance the wave field propagates from each point on the cutterhead of the tunnel boring machine to point A when it acts as the seismic source; B(x) is the distance the wave field propagates from each point on the cutterhead of the tunnel boring machine to point B when it acts as the seismic source.

[0035] Step 43: Perform cross-correlation on the Green's functions of points A and B to obtain the reference signal trace Green's function G*(A / x), which is used as the seismic interferometry record of point B. Represented as:

[0036] G(A / x)=G*(A / x)G(B / x)

[0037] Where G*(A / x) is the complex conjugate form of G(A / x);

[0038] Step 44: Repeat steps 42-43 to obtain the common detector gathers of the remaining cutterhead sources relative to the reference signal channel.

[0039] Preferably, the specific implementation process of step 5 is as follows:

[0040] Step 51: Extract the seismic interferometric records from the common detector gather obtained in Step 4. As the epicenter, the earthquake record R2 was obtained by back propagation of the wavefield using a virtual observation system.

[0041] Step 52: Using a reverse time migration imaging method based on cross-correlation conditions, the imaging values ​​are obtained from the seismic interferometric records and seismic records, expressed as:

[0042]

[0043] In the formula: R1 represents the seismic interferometry record, which is the forward propagation seismic record of the source wavefield reaching the detector; R2 represents the seismic interferometry record. The seismic record obtained by backpropagation of the wave field to the source; I is the imaging value under cross-correlation conditions; T represents time and is a positive integer.

[0044] As can be seen from the above technical solution, compared with the prior art, this invention discloses an imaging method based on a TBM-based seismic wave advance detection observation system. By arranging geophones on the TBM cutterhead, shield support shoes, and the roadway sidewalls behind the working face, a TBM-based seismic wave advance detection observation system is constructed, making full use of the limited space of the roadway. A deep learning algorithm is used to construct a wavefield correction dataset, namely, wavefield records from a conventional observation system arranged along the roadway and a virtual observation system in the horizontal direction of the working face. The conventional observation system wavefield records are actual seismic wavefield records, and the virtual observation system wavefield records are virtual seismic wavefield records. Furthermore, since the convolutional neural network used in the deep learning algorithm learns the mapping relationship from the actual seismic wavefield records to the virtual seismic wavefield records in the same iteration of reverse time migration imaging, this invention modifies the conventional reverse time migration imaging process. This allows the corrected wavefield to be used in the same iteration to calculate the wavefield records under the conventional observation system as TBM-based seismic advance detection data, and the virtual seismic wavefield records under the virtual observation system as seismic correction wavefield values. By forming wavefield data pairs, a virtual observation system with a large offset aperture along the horizontal direction of the tunnel face is used to replace the conventional small offset aperture observation system behind the roadway for calculating corrected wavefield values. This effectively solves the problem of small offset aperture caused by arranging linear observation systems along the rear sidewalls. Then, the seismic corrected wavefield values ​​are... Seismic interferometry is performed to generate seismic interferometric records from the common detector gathers of seismic reflections from multiple seismic waves received by the detector. These records can be directly used for high-precision reverse-time migration imaging (RTM) for advanced seismic detection during TBM tunneling. Finally, high-precision RTM is used to perform RTM imaging on the seismic interferometric records to obtain the distribution of geological anomalies ahead of the tunnel face, thereby achieving accurate and advanced prediction of geological conditions ahead of the tunneling face. This invention makes full use of the tunnel space and, through deep learning technology, achieves wavefield correction of the virtual observation system along the horizontal direction of the tunneling face, expanding the migration aperture between the shot and detector and improving the detection capability of small-dipping structures. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0046] Figure 1 The attached figure is a flowchart of the TBM-based seismic advance detection data processing method provided by the present invention.

[0047] Figure 2 The attached figure is a schematic diagram of the TBM-based seismic observation system provided by the present invention;

[0048] Figure 3 The attached figure is a geological structure model diagram provided by the present invention;

[0049] Figure 4 The attached figure is a flowchart of the wave field correction based on deep learning provided by the present invention;

[0050] Figure 5 The attached figure is a schematic diagram of the geophone layout at the TBM cutterhead tunneling face in the conventional observation system provided by the present invention;

[0051] Figure 6 The attached figure is a schematic diagram of the layout of the support shoe and the rear roadway sidewall detector in the conventional observation system provided by the present invention.

[0052] Figure 7 The attached figure is a seismic profile diagram under the conventional observation system provided by this invention;

[0053] Figure 8 The attached figure is a schematic diagram of the geophone layout at the TBM cutterhead tunneling face in the virtual observation system provided by this invention;

[0054] Figure 9 The attached figure is a seismic profile after wavefield correction under the virtual observation system provided by this invention;

[0055] Figure 10 The attached figure is a schematic diagram of the convolutional neural network structure provided by the present invention;

[0056] Figure 11 The attached figure is a schematic diagram of the seismic profile record after seismic interferometry processing provided by the present invention;

[0057] Figure 12 The attached figure is a schematic diagram of the reverse time migration imaging results based on the modified wave field provided by the present invention. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] This invention discloses an imaging method based on a TBM-based seismic wave advance detection and observation system, the process of which is as follows: Figure 1 As shown, the specific steps are as follows:

[0060] S1: Considering the characteristics of TBM-based tunnel construction, high-precision three-component geophones are deployed on both sides and behind the TBM excavation face and shield head. A schematic diagram of the geophone deployment in the actual tunnel observation system is shown below. Figure 2 As shown in the figure, R1, R2, ..., Rn-1, Rn, as well as R111, R112, R11, and R12 all represent detectors. R1, R2, ..., Rn-1, and Rn are installed on the sidewalls of the roadway behind the tunneling face. R111 and R112 are installed on the cutterhead of the tunnel boring machine. R11 and R12 are installed on the shield support shoes on both sides of the shield head. S1 and S2 represent the cutters on the cutterhead of the tunnel boring machine, and each cutter is equipped with a detector.

[0061] S2: For the same geological model, such as Figure 3 As shown, acoustic wave field simulations were performed according to both the actual tunnel observation system layout and the virtual observation system, obtaining seismic wave field records under both the conventional and virtual observation systems. Then, a mapping relationship between the two was constructed based on a convolutional neural network. The constructed convolutional neural network was trained using the seismic wave field records from both the conventional and virtual observation systems, as follows: Figure 4 As shown, the specific steps are as follows:

[0062] a) For the same geological model, linear observation systems are deployed along the left and right sides of the rear of the tunneling face. The geophones at the TBM cutterhead tunneling face are arranged as follows: Figure 5 As shown, the arrangement of the support boot and the rear tunnel sidewall detectors is as follows: Figure 6 As shown, wavefield simulation was performed and Gaussian white noise was introduced to simulate the seismic record M1 under the measured conventional observation system. obs ,like Figure 7 As shown; a horizontal virtual linear observation system is arranged along the tunneling face, and the detectors on the TBM cutterhead tunneling face are arranged as follows. Figure 8 As shown, a wave field simulation was performed and Gaussian white noise was introduced to simulate the measured wave field record M2. obs ,like Figure 9 As shown; where the longitudinal section of the tunneling face is as follows: Figure 5As shown, the cross-section of the detector arrangement in a conventional observation system is as follows: Figure 6 As shown, geophones are arranged along the left and right sides of the tunnel face and behind the roadway, as indicated by the blocks in the figure; the cross-section of the geophone arrangement in the virtual observation system is shown in the figure. Figure 8 As shown, the detectors on the left and right sides behind the tunnel are mapped to the left and right sides of the tunneling face through algorithm processing. As shown in the figure, the dashed squares are transformed into squares on the dashed line horizontal to the tunneling face. The dashed box represents the original detector arrangement, and the green box represents the virtual detector arrangement.

[0063] b) Data preprocessing: Seismic record M1 obs M2 obs Convert the data to an m×n two-dimensional image format;

[0064] c) Construct and train the neural network: The structure of a convolutional neural network is as follows: Figure 10 As shown, it is generated by an input layer, a convolutional layer, a sampling layer, a fully connected layer, and an output layer, with the convolutional layer, sampling layer, and fully connected layer serving as hidden layers. The specific steps are as follows:

[0065] 1) Input layer: Input m×n two-dimensional image data;

[0066] 2) Convolutional layer: According to the convolution calculation formula, a matrix of size l×l is applied to a two-dimensional m×n feature matrix to perform scanning operations;

[0067] 3) Sampling layer: Remove unimportant samples from the two-dimensional m×n image to reduce the number of parameters;

[0068] 4) Fully connected layer: Connects each current node to all nodes in the previous layer, such as the input layer and convolutional layer;

[0069] d) Determine the loss function: Based on the seismic wavefield data obtained by two different observation methods, calculate the loss function values ​​Loss1 and Loss2 for each.

[0070]

[0071] In the formula, Loss1 and Loss2 represent the seismic wavefield loss functions during excavation under conventional and virtual observation systems, respectively; M 1N M 2N These represent the forward propagation seismic records of the wavefield under conventional and virtual observation systems, respectively. The subscript N is a positive integer. When N=1, it represents the forward propagation seismic record of the initial velocity model wavefield corresponding to the first conventional observation system. When N=2, it represents the forward propagation seismic record of the second model wavefield, and so on.

[0072] e) Training the network: Input the forward modeling data of the TBM tunneling seismic wavefield under the two observation systems into the convolutional neural network constructed in step c) for training;

[0073] The forward modeling data M1 of the TBM tunneling seismic wavefield from the conventional observation system was used. obs As the input vector of the neural network, the TBM forward modeling data M2 of the tunneling seismic wavefield of the virtual observation system obs As the target output, calculate the output of each unit in the hidden and output layers; calculate the difference between the neural network's output vector and the target output as the bias; if the bias satisfies the stopping condition, e≤10. -6 If the training ends, the weights and threshold are fixed; otherwise, the error of the neurons in the neural network is calculated, the error gradient is solved, and the weight parameters are updated according to the error gradient. A convolutional neural network is composed of different multi-layer networks. The weights are determined by the stride and number of kernels used in each convolution calculation. Here, the stride is 2 and the number of kernels is 2. The threshold is the value that satisfies the bias termination condition. The weights are passed to the "given input vector and target output," and then the convolutional network calculates to make the bias approach the termination condition, ultimately obtaining the target vector.

[0074] f) Perform wavefield correction: Using the trained model, based on steps d) and e), obtain the TBM-based seismically corrected wavefield record under the virtual observation system.

[0075] g) Save the corrected wavefield from step f) and store it in the dataset.

[0076] h) Repeat steps a), b), c), d), e), f), and g) to obtain seismic wavefield correction datasets {M} for different geological structures during excavation. 21 M 22 ,…,M 2N}, where N is a positive integer representing the number of geological model types;

[0077] S3: Measured TBM seismic advance detection data during excavation Substituting into S2, the corresponding corrected wavefield value under the virtual observation system is found using a deep learning algorithm.

[0078] S4: Using seismic interferometry, the corrected seismic records were processed. Cross-correlation processing is performed to obtain the processed seismic interferometric records. like Figure 11 As shown, the specific process is as follows:

[0079] 1) First, select a detector located at point A on the left side behind the tunneling face as the reference signal channel G(A / x). Then, its Green's function with respect to the receiving point A is:

[0080]

[0081] Where d(x) is the distance the wave field propagates to point A when each point on the cutterhead is the source of the earthquake, C is the speed of the cutterhead, V is the speed of the seismic wave propagation; x0 is the starting point of the tunnel boring machine's forward movement, and x is the current distance the tunnel boring machine has moved.

[0082] 2) Based on step 1), the Green's functions at points A and B are as follows:

[0083]

[0084] The cross-correlation between two points is shown in equation (4):

[0085] G(A / x)=G*(A / x)G(B / x) (4)

[0086] 3) Repeat steps 1)-2) to obtain the common shot gather results of the cutterhead source for each channel received by the other detectors, such as... Figure 5 As shown;

[0087] S5: Based on the virtual observation system in S2, perform reverse time migration imaging, i.e., perform seismic interferometric imaging on the seismic interferometry records acquired in S4. Seismic records R2 were obtained through wavefield backpropagation using a virtual observation system, and the imaging values ​​were calculated using a reverse time migration imaging method based on cross-correlation conditions, as follows:

[0088]

[0089] In the formula: R1 represents the source wavefield after wavefield correction, arriving at the detector and transmitting the seismic record; R2 represents the data from the probe. The wavefield is backpropagated from the earthquake source; I is the imaging value under cross-correlation conditions.

[0090] S6: Perform seismic forward migration imaging on the imaging value I calculated in S5, such as... Figure 12 As shown, advanced predictions of the geological structures ahead are obtained.

[0091] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0092] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An imaging method for a TBM-based seismic advance detection and observation system, characterized in that, Includes the following steps: Step 1: Deploy geophones in actual tunnels of different geological models to construct a conventional observation system; Step 2: Construct a virtual observation system based on the conventional observation system, and perform acoustic wave field simulation on the conventional and virtual observation systems to obtain the actual seismic wave field records under the conventional observation system and the virtual seismic wave field records under the virtual observation system simulated by different geological models. Furthermore, construct the mapping relationship between the two based on a convolutional neural network to obtain the seismic wave field correction dataset during excavation. Step 3: Collect measured TBM-based seismic advance detection data during tunneling, and determine the corresponding seismic correction wavefield value under the virtual observation system based on the seismic wavefield correction dataset during tunneling in Step 2 using a convolutional neural network; Step 4: Using seismic interferometry, cross-correlation processing is performed on the seismic correction wavefield values ​​under the virtual observation system to obtain the processed seismic interferometric record; Step 5: Perform reverse time migration imaging on the processed seismic interferometric records to obtain imaging values; Step 6: Perform seismic advance migration imaging on the imaging values ​​to obtain advance prediction results of the geological structures ahead.

2. The imaging method of a TBM-based seismic advance detection and observation system according to claim 1, characterized in that, The specific implementation process of step 2 is as follows: Step 21: For each actual roadway, construct a conventional observation system, simulate the acoustic wave field of the actual roadway, and introduce Gaussian white noise to simulate the actual seismic wave field record under the actual roadway. A virtual observation system was constructed based on the mapping of conventional observation systems. Acoustic wavefield simulation was performed, and Gaussian white noise was introduced to simulate the virtual seismic wavefield records obtained from actual measurements. ; Step 22: Record the actual seismic wavefield Virtual seismic wavefield records Convert to Two-dimensional image formats for actual and virtual seismic images; Step 23: Construct a convolutional neural network; Step 24: Based on the actual seismic wavefield record Virtual seismic wavefield records Calculate the loss function values ​​for both. Loss function value ; ； In the formula , These represent the seismic wavefield loss functions during excavation under conventional and virtual observation systems, respectively. M 1N , M 2N These represent the actual forward propagation seismic records under conventional observation systems and the virtual forward propagation seismic records under virtual observation systems, respectively, with N representing the Nth geological model. Step 25: Record using simulated actual seismic wavefield Virtual seismic wavefield records Loss function value Loss function value Train the constructed convolutional neural network; Step 26: Use the trained convolutional neural network to record the actual seismic wavefields simulated by different geological models. Perform corrections and obtain the corresponding seismic correction wavefield values ​​under the virtual observation system. ; Step 27: Seismic correction wavefield values ​​under the virtual observation system of all geological models Constructing a seismic wave field correction dataset during excavation , where N is a positive integer representing the number of geological model types.

3. The imaging method of a TBM-based seismic advance detection and observation system according to claim 1, characterized in that, The specific implementation process of step 4 is as follows: Step 41: Select the geophone located at point A on the left side of the tunneling face as the reference signal path; Step 42: Select point B from the remaining geophones on the sidewall of the tunneling face. The Green's functions of the signals received by the geophones at points A and B are as follows: ； Where G(A / x) represents the Green's function of the signal received at point A; G(B / x) represents the Green's function of the signal received at point B; d(x) is the distance the wave field propagates from each point on the cutterhead of the tunnel boring machine to point A or B when it acts as the seismic source; C is the cutterhead moving speed; V is the seismic wave propagation speed; x0 is the starting point of the tunnel boring machine's forward movement; x is the current distance the tunnel boring machine has moved; ω represents the angular frequency; i represents the imaginary part of the complex number; A(x) is the distance the wave field propagates from each point on the cutterhead of the tunnel boring machine to point A when it acts as the seismic source; B(x) is the distance the wave field propagates from each point on the cutterhead of the tunnel boring machine to point B when it acts as the seismic source. Step 43: Perform cross-correlation on the Green's functions at points A and B to obtain the reference signal Green's function. The seismic interferometry record at point B is represented as follows: ； in, for The complex conjugate form; Step 44: Repeat steps 42-43 to obtain the common detector gathers of the remaining cutterhead sources relative to the reference signal channel.

4. The imaging method of a TBM-based seismic advance detection and observation system according to claim 1, characterized in that, The specific implementation process of step 5 is as follows: Step 51: Using the seismic interferometric record obtained in Step 4 as the source, the seismic record is obtained by backpropagation of the wavefield using a virtual observation system; Step 52: Using a reverse time migration imaging method based on cross-correlation conditions, the imaging values ​​are obtained from the seismic interferometric records and seismic records, expressed as: ； In the formula: Represents seismic interferometry records; For seismic interferometry records Seismic records obtained by backpropagation of the wave field to the earthquake source; The image value represents the cross-correlation condition; T represents time.

5. The imaging method of a TBM-based seismic advance detection and observation system according to claim 1, characterized in that, Based on the actual tunnel characteristics of TBM construction, several high-precision three-component geophones are arranged on the left and right sides behind the TBM excavation face and shield head.

6. The imaging method of a TBM-based seismic advance detection and observation system according to claim 5, characterized in that, Based on the geological model of the actual tunnel, the geophones behind the tunnel in the conventional observation system are mapped to the left and right sides of the tunneling face after being processed by an algorithm, thus constructing a virtual observation system.

Images