Drill hole joint detection method and system capable of reducing number of advance drill holes of tunnel roadway

Abstract

The invention provides a drill hole joint detection method and system capable of reducing the number of advance drill holes in a tunnel roadway, and relates to the field of roadway engineering detection, and the method comprises the steps: obtaining the depth of a measurement point and the radar data and transient electromagnetic data of a target region, and enabling the radar data and the transient electromagnetic data to be collected through an integrated detection device in a drill hole; performing feature extraction on the radar data and the transient electromagnetic data, and fusing the extracted features based on contribution weights of the radar and the transient electromagnetism to obtain a fused feature vector; based on the depth of the measuring point and the fused feature vector, utilizing the trained physical information neural network to invert the dielectric constant and the resistivity; according to the dielectric constant and the resistivity, the geological anomalous body of the target area is identified; according to the invention, the drilling radar and the drilling transient electromagnetism are combined, so that the number of advanced drilling holes is reduced, the detection cost is reduced, and the full-area coverage of the forepoling of the roadway is ensured.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of roadway engineering detection, in particular to a drilling combined detection method and system capable of reducing the number of tunnel and roadway advance boreholes. BACKGROUND

[0002] Before the construction of a mine, advance boreholes must be laid according to the requirements of the specifications; in the case of complex geology and hydrology, multiple advance boreholes need to be drilled to investigate the regional geological conditions; in general, at least two advance boreholes need to be drilled; however, advance boreholes are expensive, and multiple advance boreholes will significantly increase the cost of the mine; and only drilling will cause a "one-sided view" problem, and the understanding of the underground geological conditions is limited; therefore, the existing detection scheme is difficult to balance the detection cost and detection accuracy. SUMMARY

[0003] In order to solve the above problems, the present disclosure provides a drilling combined detection method and system capable of reducing the number of tunnel and roadway advance boreholes, which combines borehole radar and borehole transient electromagnetic method, reduces the number of advance boreholes, reduces the detection cost, ensures full-area coverage of the roadway advance exploration, and improves the detection accuracy.

[0004] According to some embodiments, the present disclosure adopts the following technical solutions: The drilling combined detection method capable of reducing the number of tunnel and roadway advance boreholes comprises: obtaining radar data and transient electromagnetic data of a target area at a measurement point depth, wherein the radar data and the transient electromagnetic data are collected by an integrated detection device in a borehole; performing feature extraction on the radar data and the transient electromagnetic data respectively, fusing the extracted features based on the contribution weights of the radar and the transient electromagnetic method, and obtaining a fused feature vector; based on the measurement point depth and the fused feature vector, using a trained physical information neural network to perform inversion on the dielectric constant and the resistivity; identifying a geological anomaly body in the target area according to the dielectric constant and the resistivity; The integrated detection device comprises a GPR probe module and a TEM probe module, the GPR probe module uses multiple sets of transmitting and receiving antennas, and the TEM probe module uses coplanar rectangular transmitting and receiving coils to realize zero magnetic flux by adjusting the transmitting and receiving distance.

[0005] According to some embodiments, the present disclosure adopts the following technical solutions: The drilling combined detection system capable of reducing the number of tunnel and roadway advance boreholes comprises: an acquisition module configured to obtain radar data and transient electromagnetic data of a target area at a measurement point depth, wherein the radar data and the transient electromagnetic data are collected by an integrated detection device in a borehole; The extraction module is configured to extract features from the radar data and the transient electromagnetic data respectively, fuse the extracted features based on the contribution weights of the radar and the transient electromagnetic, and obtain a fused feature vector; The inversion module is configured to invert the dielectric constant and the resistivity by using the trained physical information neural network based on the depth of the measuring point and the fused feature vector; The identification module is configured to identify the geological abnormal body of the target region according to the dielectric constant and the resistivity. The integrated detection device includes a GPR probe module and a TEM probe module, the GPR probe module adopts multiple sets of transmitting and receiving antennas, and the TEM probe module adopts coplanar rectangular transmitting and receiving coils to realize zero magnetic flux by adjusting the transmitting and receiving distance.

[0006] According to some embodiments, the present disclosure adopts the technical scheme as follows: A computer program product comprising a computer program which, when executed by a processor, implements the drilling combined detection method capable of reducing the number of advance drill holes in a tunnel and roadway.

[0007] According to some embodiments, the present disclosure adopts the technical scheme as follows: A non-transitory computer readable storage medium for storing computer instructions, which, when executed by a processor, implements the drilling combined detection method capable of reducing the number of advance drill holes in a tunnel and roadway.

[0008] According to some embodiments, the present disclosure adopts the technical scheme as follows: An electronic device comprising a processor, a memory and a computer program, wherein the processor is connected to the memory, and the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the drilling combined detection method capable of reducing the number of advance drill holes in a tunnel and roadway.

[0009] Compared with the prior art, the present disclosure has the following beneficial effects: (1) The present application adopts a modular multi-parameter integrated probe, which contains a zero-magnetic-flux transient electromagnetic coil and a double-frequency borehole radar antenna, and is suitable for Φ75-160mm boreholes; through the design of a ceramic wear-resistant ring, an electromagnetic telescopic arm and a backup hydraulic retraction system, the probe realizes dynamic adjustment of the radial size (the contraction ratio reaches 1:2.5) and reliable detection in extreme environments (pressure resistance of 10MPa, waterproof IP68), and solves the problem of jamming caused by the size deviation of the traditional equipment.

[0010] (2) The application adopts a low-frequency butterfly antenna and a high-frequency microstrip antenna combination suitable for borehole GPR, can measure the surrounding rock crack within 0-5m, and can measure the abnormal body section within 0-15m; the standardized microstrip antenna design (such as Rogers RO4350B substrate) is combined with FPGA hardware acceleration, and the manufacturing cost is reduced.

[0011] (3) The application adopts a zero-flux transient electromagnetic transmitting and receiving coil, and proposes a transmitting and receiving distance calculation method, the zero-flux is formed by eccentric placement, and the range of the transient electromagnetic blind area is greatly reduced.

[0012] (4) The application designs a borehole positioning system based on laser ranging (accuracy ±1mm) and inertial navigation (heading angle error <0.5°), and realizes real-time feedback of the probe inclination and depth; a double-redundancy anti-stuck mechanism is developed, including an electromagnetic trigger type retractable arm (response time <0.5s) and a hydraulic emergency retracting system (pressure 21MPa), combined with vibration spectrum analysis and wear-resistant ceramic ring wear monitoring, to realize early warning and millisecond-level emergency response of the stuck risk, and ensure the safety of deep hole detection.

[0013] (5) The application proposes a dynamic weight fusion technology based on borehole ground penetrating radar (GPR) and transient electromagnetic (TEM), the contribution weights of GPR reflection coefficient and TEM apparent resistivity are adaptively allocated through signal-to-noise ratio (SNR), and a joint feature matrix (including reflection gradient and resistivity time-varying derivative) is constructed. This method overcomes the limitations of single geophysical method in complex strata (such as insufficient penetration of GPR in high-conductivity strata and low near-field resolution of TEM), improves the identification accuracy of geological abnormal bodies, and reduces the number of verification boreholes by more than 50%.

[0014] (6) The application proposes a joint inversion framework combining physical priori and data-driven, constructs an objective function containing data fitting, total variation (TV) and sparsity constraint, and embeds a physical constraint neural network (PINN) to force to meet Maxwell equation; combined with genetic algorithm (GA) global search and L-BFGS local optimization, the efficient inversion of complex geological model is realized, and the physical rationality of the inversion result is significantly enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0015] The accompanying drawings, which form a part of this disclosure, are used to provide further understanding of the disclosure, and the illustrative embodiments of the disclosure and their descriptions are used to explain the disclosure, and do not constitute improper limitations on the disclosure.

[0016] Figure 1 It is the method flowchart of example 1. Figure 2 It is the structure diagram of the integrated detection device of example 1. DETAILED DESCRIPTION The present disclosure will be further described below in conjunction with the accompanying drawings and examples.

[0017] It should be noted that the following detailed description is exemplary in nature and is intended to provide further description of the present disclosure. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs.

[0018] It should be noted that the terms used herein are only intended to describe specific embodiments and are not intended to limit exemplary embodiments according to the present disclosure. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise, and it should also be understood that when the terms "comprise" and / or "comprises" are used in the specification, they indicate the presence of the features, steps, operations, devices, components and / or combinations thereof.

[0019] The borehole exploration method is a method of placing a detection instrument in a borehole for detection, which mainly includes borehole radar and borehole transient electromagnetic method; the radar (GPR) is a detection technology based on high-frequency electromagnetic wave reflection, which can realize decimeter-level detection, but the detection distance is generally within 10m; the transient electromagnetic method (TEM) is a low-frequency electromagnetic method, which inputs a step current to a transmitting source and obtains underground electromagnetic induction signals during current shutdown using a receiving source, and has the characteristics of sensitivity to low-resistance bodies and large detection depth, but it has a shallow blind area problem.

[0020] The present application combines borehole radar and borehole transient electromagnetic method, proposes a borehole GPR-TEM integrated detection method, uses GPR probe module and TEM probe module, reduces the number of advance boreholes, reduces the detection cost, and ensures full-area coverage of roadway advance exploration, based on the full-coverage GPR-TEM data, proposes a borehole GPR-TEM data fusion interpretation method.

[0021] Example 1 In an embodiment of the present disclosure, a borehole combined detection method is provided, which can reduce the number of advance boreholes in a tunnel and roadway, comprising: Step S1: obtaining radar data and transient electromagnetic data of a measurement point depth and a target area, the radar data and transient electromagnetic data being collected by an integrated detection device in a borehole; Step S2: respectively extracting features of the radar data and the transient electromagnetic data, fusing the extracted features based on the contribution weights of the radar and the transient electromagnetic method, and obtaining a fused feature vector; Step S3: based on the measurement point depth and the fused feature vector, using a trained physical information neural network to perform inversion on the dielectric constant and the resistivity; Step S4: identifying a geological anomaly body in the target area according to the dielectric constant and the resistivity. The integrated detection device includes a GPR probe module and a TEM probe module. The GPR probe module uses multiple sets of transceiver antennas, and the TEM probe module uses coplanar rectangular transceiver coils to achieve zero magnetic flux by adjusting the transceiver distance.

[0022] As an embodiment, the disclosed combined drilling detection method can reduce the number of advance drilling holes in tunnels and lanes. It combines drilling radar with drilling transient electromagnetic to reduce the number of advance drilling holes and reduce the detection cost while ensuring full coverage of the lane advance exploration. The specific implementation process is described below from the two parts of data acquisition and data processing.

[0023] 1. Data Collection This embodiment proposes a borehole GPR-TEM integrated detection method, which uses an integrated detection device in the borehole to collect radar data and transient electromagnetic data covering the entire area of ​​the tunnel advance.

[0024] Integrated detection devices, such as Figure 2 As shown, it includes a GPR probe module, a TEM probe module, a control module and an auxiliary module.

[0025] (1) GPR probe module The GPR probe module uses two antenna structures: one is a low-frequency butterfly antenna and the other is a high-frequency microstrip antenna. Each antenna is divided into two sections, the front section is the transmitting section and the rear section is the receiving section.

[0026] The low-frequency butterfly antenna measures 200mm×150mm, operates at a frequency of 800-150MHz, has a gain of 5dBi, and a center-to-center distance of 300mm between the transmitting antenna (TX) and the receiving antenna (RX). TX is horizontally polarized, while RX is vertically polarized, with cross-polarization isolation further enhancing power. The receiving coil is housed in an aluminum cylindrical shielding cabin with a 90° opening and lined with absorbing material. It is primarily used for deep detection, with a detection range of 15m.

[0027] The high-frequency microstrip antenna uses a rectangular microstrip patch antenna, operating at a frequency of 400 ± 10% MHz and measuring 30 × 40 mm. The substrates for the transmitting antenna (TX) and receiving coil (RX) are made of high-frequency board material (such as Rogers RO4350B), both measuring 50 × 60 × 1.6 mm. The TX is fed by coaxial backfeed, while the RX is fed by electromagnetic coupling. The TX is vertically polarized, while the RX is horizontally polarized, with cross-polarization isolation to further enhance power. The axial spacing between the TX and RX antennas is 40 cm, and a metal spacer coated with absorbing material is installed between the two antennas. The detection range is 10 meters.

[0028] (2) TEM probe module TEM probe module adopts coplanar rectangular transmitting-receiving coil, and zero magnetic flux is realized by adjusting transmitting-receiving distance, so as to eliminate the covering influence of primary field of transmitting-receiving coil on secondary field of shallow geological body, and greatly reduce the blind area range.

[0029] Specifically, the TEM probe module adopts a zero-flux coplanar transmitting-receiving coil, which is placed eccentrically to make the magnetic flux received by the receiving coil inside and outside the transmitting coil compensate each other, so as to ensure that the mutual inductance coefficient of the transmitting-receiving coil is 0, thereby eliminating the covering of the primary field and realizing the collection of the secondary field of the shallow geological body.

[0030] The coplanar coil usually adopts a square coil, and the mutual inductance coefficient calculation formula is as follows:

[0031] Among them, is the number of turns of the receiving coil; is the number of turns of the transmitting coil; is the vacuum permeability; is the path of the transmitting coil, represents the first side of the square transmitting coil; represents the first side of the square receiving coil; is the path of the receiving coil, represents the first side of the square receiving coil; represents the first side of the square receiving coil; is the microelement of the transmitting coil; is the microelement of the receiving coil; r is the distance between a source point Q1 on the transmitting coil and an arbitrary point Q2 on the receiving coil. By calculating the transmitting-receiving distance when the mutual inductance coefficient is 0, zero magnetic flux can be realized.

[0032] In this embodiment, the transmitting antenna (TX) and the receiving coil (RX) of the TEM probe module adopt a rectangular coil with a size of 30mm*200mm, the TX line diameter is 1mm, the number of turns is 20, the receiving coil line diameter is 0.1mm, and the number of turns is 200; a cylindrical shield is arranged outside the coil, a 120° opening is arranged, and the material is selected from foamed nickel; and the detection distance is between 10-30m.

[0033] (3) Control module The control module controls the transmission and reception of the borehole GPR and TEM, and has a built-in real-time calculation module to realize intelligent switching of low-frequency and high-frequency channels according to real-time lithology conditions.

[0034] Specifically, the internal circuit of the control module includes a transmitting circuit and a receiving circuit, an output continuous voltage adjustable power supply circuit, a bipolar pulse transmitting circuit, and a logic control circuit, etc.

[0035] As some possible implementations, the power supply circuit can support connection with an external power supply or internally set a battery power supply; the receiving circuit is connected with each receiving coil through an insulating wire, and the collected data is stored in the receiving control system, which is internally provided with a slot for placing electronic components such as a collection card, an amplifier, a data transmission unit, and an embedded computer.

[0036] The transmitting circuit adopts an ARM Cortex-A processor, integrates an FPGA (Field Programmable Gate Array), processes signals in real time, and performs time synchronization; a GPR pulse signal generator is set, which is based on an avalanche transistor circuit, generates a pulse with a adjustable pulse width of 0.1-5 ns and a peak voltage of 1 kV, and has a adjustable repetition frequency of 1 kHz-1 MHz; a TEM transient electromagnetic current source is set, which adopts an SIC MOSFET H-bridge topology, and has a maximum output current of 50 A and supports bipolar square wave output.

[0037] The receiving circuit sets a GPR receiving channel, which adopts a 12-bit ADC, has a sampling rate of 5 GS / s, a bandwidth of 2 GHz, and an input impedance of 50 Ω; a TEM receiving channel is set, which adopts a 24-bit Σ-Δ ADC, has a sampling rate of 100 kS / s, a dynamic range of 120 dB, and integrates a programmable gain amplifier (PGA) with a adjustable gain of 1-1000 times.

[0038] The power supply is designed with double inputs, which can be compatible with AC 220 V or DC 24 V, and is provided with multiple outputs, including ±12 V (analog circuit), 5 V (digital circuit), and 3.3 V (FPGA), and has an efficiency of >90%. The battery pack can adopt a lithium iron phosphate battery pack (48 V / 20 Ah), supports hot plugging and fast charging, and has a reserved protection interval (50 μs) to start TEM current off measurement after GPR pulse emission; the GPR / TEM circuits are separately arranged in independent shielding cabins, and are provided with three-level isolation of power supply ground-signal ground-chassis ground.

[0039] (4) Auxiliary module The auxiliary module includes a probe positioning system and an anti-stuck mechanism. The probe positioning system includes a borehole laser ranging unit and an inertial navigation unit, has a distance accuracy of 1 mm and a heading angle error of <0.5°, and can transmit data back to the borehole to display the probe inclination and depth in real time. The anti-stuck mechanism includes a retractable arm based on an electromagnet, an emergency retraction device of a backup hydraulic retraction system, and a wear detection device using a shell inlaid with a wear-resistant ceramic ring and an embedded vibration sensor.

[0040] The borehole laser ranging unit adopts a 1550 nm infrared laser emitter and sets an APD avalanche photodiode at the tail end of the probe. The laser is vertically emitted from the borehole, returns through the tail mirror of the probe, and the distance, i.e. the depth of the current measuring point of the probe, is calculated by measuring the round-trip time .

[0041] The inertial navigation unit adopts a three-axis MEMS gyroscope, a three-axis accelerometer, and a three-axis magnetometer. The three-axis magnetometer is used for heading angle correction; a real-time three-axis direction angle of the probe is generated and input to the control module for storage.

[0042] The trigger type retractable arm is installed at the tail end of the entire device, adopts a solenoid electromagnet (rated voltage 24V, suction force 200N, response time <10ms) and a titanium alloy connecting rod retractable arm (length 150mm, retractable stroke 50mm), and when the contact force sensor detects that the resistance is >50N, the electromagnet is instantaneously powered on, and the retractable arm is retracted to prevent the probe from being stuck in front.

[0043] The emergency retraction device, when the electromagnetic trigger type retractable arm fails, starts the hydraulic circuit of the pressure sensor, adopts a hydraulic circuit composed of a miniature gear pump, a double-acting hydraulic cylinder and a bladder type nitrogen energy accumulator, the hydraulic stroke is 60mm, and the retraction speed is set to 100mm / s.

[0044] The wear detection device adopts a wear-resistant ceramic ring, for example, a zirconia toughened ceramic (Zr02), the ring is 8mm thick and 15mm wide, a MEMS accelerometer is used, and when the ring thickness decreases by 0.5mm, an early warning is given; and time domain and frequency domain vibration analysis is performed: the time domain RMS vibration value >5g for 1s triggers an early warning; the frequency domain characteristic frequency matches the stuck mode (such as 200-400Hz resonance peak).

[0045] The process in which each part of the auxiliary module works cooperatively is as follows: 1) Normal detection: The ceramic ring of the wear detection device rubs against the hole wall during detection, and the vibration sensor monitors the baseline vibration.

[0046] 2) Stuck pre-judgment: When the vibration sensor detects a sudden increase in vibration energy or the contact force sensor detects that the contact force exceeds the limit, the trigger type retractable arm performs electromagnetic retraction.

[0047] 3) Emergency treatment: When electromagnetic retraction fails, the hydraulic system of the emergency retraction device takes over within 0.5s and continues to retract.

[0048] 4) Wear maintenance: When the wear detection device detects that the cumulative wear reaches the threshold value, the device is stopped and the ceramic ring is prompted to be replaced.

[0049] The GPR probe module and the TEM probe module are fixedly connected, and then connected to a motor drill. An explosion-proof stepper motor (such as the 42BYGH34) is used, with a programmable speed range of 0.1 to 10 rpm. Step scan mode is selected (1° step size for GPR and 15° step size for TEM). The acquisition process is as follows: 1) Push the probe into the borehole to the target depth (e.g. 20m); 2) Start the motor and set the scanning mode; 3) The motor drives the antenna to rotate to the initial angle ; 4) If rotated to angle +k·1° (k is any non-zero positive integer, k=1,2,3...), the GPR probe module transmits a low-frequency 80-150MHz signal and synchronously collects and stores it; otherwise, no transmission or reception is performed; 5) After the low-frequency signal ends, a high-frequency 400MHz signal is transmitted and synchronously collected and stored; 6) If rotated to angle +k·15° (k is any non-zero positive integer, k=1,2,3...), the TEM probe module transmits a transient electromagnetic field, and collects and stores data during the shutdown period; otherwise, no transmission or reception is performed; 7) Rotate to the next angle +1°, repeat steps 3-6; 8) After the 360° scan is completed, GPR and TEM data are obtained at different positions in the hole; 2. Data Processing This example proposes a borehole GPR-TEM data fusion interpretation method. It fuses the characteristics of ground penetrating radar (GPR) and transient electromagnetic (TEM) data through dynamic weight allocation. It also constructs an objective function for the joint inversion of permittivity and resistivity. By integrating the constraints of the electromagnetic field equations with a physical information neural network (PINN), a hybrid optimization strategy of genetic algorithm (GA) and L-BFGS is employed to achieve global search and local fine convergence.

[0050] The interpretation method includes feature-level fusion and joint inversion, combined with intelligent algorithms to achieve high-precision identification of geological anomalies.

[0051] (1) Feature-level fusion Feature extraction is performed on GPR and TEM data, specifically: The GPR time domain signal is mapped to the depth domain through time-to-depth conversion, the reflection coefficient R(z) at the depth z of each measuring point is extracted, and the spatial derivative is calculated using the central difference:

[0052] in, is the difference of the depth of the survey point between adjacent survey points.

[0053] The obtained reflection coefficient R(z) and spatial derivative are taken as GPR features.

[0054] TEM is based on the late-time field formula to calculate the apparent resistivity, which is converted to the depth domain by depth migration imaging , and the central difference is used to calculate the time domain rate of change:

[0055] wherein, is the time step used when calculating the time derivative of the apparent resistivity (or electromagnetic field attenuation signal), i.e. the interval between adjacent two sampling time points The obtained depth domain and time domain rate of change are taken as TEM features.

[0056] Stack GPR and TEM features into a multi-dimensional matrix:

[0057] wherein, N z is the number of depth points.

[0058] Based on the signal-to-noise ratio SNR, the contribution weight of GPR features and the contribution weight of TEM features are automatically adjusted:

[0059]

[0060]

[0061] wherein, represents the logarithm with base 10, respectively represent the signal-to-noise ratios of the borehole radar and the borehole transient electromagnetic signal, and the signal-to-noise ratio SNR is calculated from the signal power and the noise power SNR directly determines the credibility and availability of the features (GPR features and TEM features) extracted from the data, and in the following, the contribution weights and calculated using SNR are used to obtain by fusion for subsequent inversion.

[0062] For GPR: In the time domain waveform, the target reflection wave window (such as the reflection pulse interval after the direct wave) is selected, and the signal power where is the mean value of the signal window, is the mean value of the signal window, N is the number of sampling points in the window. The noise power is calculated from a non-reflecting background segment (e.g. the direct wave front) and needs to be detrended to eliminate baseline drift, where is the mean value of the noise window, is the mean value of the noise window, is the number of sampling points in the noise window. M

[0063] For TEM: The signal power is calculated by integrating the square of the voltage over the target segment (e.g. the early high SNR segment) of the decay curve, in the early time window , ] of the signal, the signal power where is the time-decaying induced voltage. The noise power is taken from a segment before the transmitter current is switched off or in the late signal decay to the instrument noise level , ], the noise power where is the time-decaying induced voltage, and frequency domain filtering can be applied to suppress power line interference.

[0064] The features are fused with the contribution weights, and the measured fusion feature is expressed by the formula:

[0065] (2) Joint inversion The objective function of joint inversion contains the data fitting term , the total variation regularization term , and the sparse regularization term , which is expressed by the formula:

[0066] The final prediction model parameter m is calculated by solving the minimum value of the objective function . Wherein, is the model parameter; The dielectric constant and resistivity are fused from the time-depth conversion and apparent resistivity imaging of the data forward from the model parameter m, and the processing method is consistent with , and the SNR is consistent with ; λ is the data weight, generally 1; η, γ are regularization weights; is the boundary preservation term, which performs TV regularization (total variation) to suppress noise and maintain sharp boundaries, ; The sparse regularization term is used to enhance the sparsity of the physical property mutations,​ ; The initial model population for subsequent inversion is generated by the physical information neural network (PINN). The input is the measured point depth z and the fused feature vector , and the output is the model parameter , which contains the dielectric constant and the resistivity . Four layers are set in the hidden layer, and the input measured point depth z is converted into intermediate features in hidden layer 1, using the swish activation function to balance the non-linear expression ability and gradient stability, i.e.

[0067] where is the weight matrix obtained by training learning, and b1 is the bias term.

[0068] In hidden layer 2, z = hidden layer 1 output is used to further extract high-order features; in hidden layer 3, repeated processing is performed, z = hidden layer 2 output, and the target function is gradually approximated. Finally, in hidden layer 4, the high-order features are converted into physical parameters:

[0069]

[0070] PINN requires that the output results meet the physical law, one of which is based on the output and , combined with the Maxwell equation approximation: , is the vacuum dielectric constant, is the electric field vector, which is calculated by automatic differentiation technology , and compared on the right side; the second is to calculate the difference between the predicted electric field divergence and the charge density: physical error = .

[0071] The loss function guides the adjustment of the model parameters. First, compare the difference between the network predicted and and the measured data, here only the data fitting term is used to calculate the data loss:

[0072] Then the physical constraint loss is performed to force the network output to meet the above electromagnetic field equation, and finally the total loss is calculated: total loss = data loss + 0.1·physical loss, the gradient is calculated according to the total loss, and the network weight is updated by the Adam optimizer. Through the PINN neural network, 50 preliminary prediction model parameters m=[ , ], replacing the random initialization of the genetic algorithm, as the initial individual population calculated by the hybrid optimization algorithm in the following text.

[0073] A hybrid optimization algorithm (genetic algorithm (GA) + memory-limited BFGS quasi-Newton method (L-BFGS)) is used to optimize the model parameters. , ] is converted into real-number coded gene sequences, and 50 individuals (50 different groups) generated by PINN are used. and Distribution). GA is used to perform global search and calculate the objective function value for each individual. , the fitness is defined as 1 / ( +1), the larger the value, the better the individual. Parent individuals are selected according to the fitness ratio (the higher the fitness, the greater the probability of being selected), and the top 5 best individuals are directly retained to enter the next generation. On this basis, the two parent individuals are mixed according to the weight to generate offspring (for example, parent A with high fitness accounts for 70%, and parent B with low fitness accounts for 30%). The termination conditions are set to the maximum number of iterations (50 times) and the optimal fitness has no significant improvement for multiple generations (the change in 10 generations is <1%), and the optimal individual is output. L-BFGS is then used for local refinement, and the optimal individual is used as the initial point m0 ​​to calculate the gradient of the objective function. and the quasi-Hessian matrix H k , and update the parameters:

[0074] in, is the kth model parameter individual, k is any positive integer; is the step size, determined by line search to ensure that the objective function decreases, and the initial learning rate , decays to 0.5 times of the original value every 10 iterations; based on Calculate the objective function value.

[0075] Set the termination conditions as follows: gradient norm < 1e-6, or relative residual change < 1e-4, or number of iterations > 100. And dynamically adjust the parameters of the objective function:

[0076] in, The measured data is obtained by borehole radar. The measured data are obtained by borehole transient electromagnetic.

[0077] Example 2 In one embodiment of the present disclosure, a combined drilling detection system is provided that can reduce the number of advance drillings in tunnels and roadways, including: An acquisition module is configured to acquire radar data and transient electromagnetic data of a measuring point depth and a target region, the radar data and the transient electromagnetic data being collected by an integrated detection device in a borehole; An extraction module is configured to perform feature extraction on the radar data and the transient electromagnetic data respectively, and fuse the extracted features based on contribution weights of the radar and the transient electromagnetic to obtain a fused feature vector; An inversion module is configured to perform inversion on permittivity and resistivity based on the measuring point depth and the fused feature vector by using a trained physical information neural network. An identification module is configured to identify a geological abnormal body of the target region according to the permittivity and the resistivity. The integrated detection device includes a GPR probe module and a TEM probe module, the GPR probe module adopts multiple sets of transmitting and receiving antennas, and the TEM probe module adopts coplanar rectangular transmitting and receiving coils to realize zero magnetic flux by adjusting the transmitting and receiving distance.

[0078] Embodiment 3 In an embodiment of the present disclosure, a computer program product is provided, which includes a computer program, and the computer program is executed by a processor to implement the drilling combined detection method capable of reducing the number of advanced boreholes in a tunnel and roadway.

[0079] Embodiment 4 In an embodiment of the present disclosure, a non-transitory computer readable storage medium is provided, which is used to store computer instructions, and the computer instructions are executed by a processor to implement the drilling combined detection method capable of reducing the number of advanced boreholes in a tunnel and roadway.

[0080] Embodiment 5 In an embodiment of the present disclosure, an electronic device is provided, which includes a processor, a memory and a computer program, wherein the processor is connected with the memory, and the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the drilling combined detection method capable of reducing the number of advanced boreholes in a tunnel and roadway.

[0081] The computer program instructions can 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 or flows and / or blocks specified in the block diagrams. Figure 1 one or more flows and / or blocks. Figure 1 one or more flows and / or blocks.

[0082] These computer program instructions can also be loaded into 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 or flows and / or blocks specified in the block diagrams. Figure 1 one or more flows and / or blocks. Figure 1 one or more flows and / or blocks.

[0083] Although the present disclosure has been described with reference to specific implementations, it will be apparent to those skilled in the art that various modifications and changes can be made to the implementations without departing from the scope of the present disclosure as set forth in the claims below. Accordingly, the disclosure is not to be limited to the above described surely.

Claims

1. A combined drilling detection method capable of reducing the number of advance drilling holes in tunnels and lanes, characterized in that: include: Acquiring radar data and transient electromagnetic data at all measurement point depths and target areas, the radar data and transient electromagnetic data being collected by an integrated detection device in the borehole; Feature extraction is performed on radar data and transient electromagnetic data respectively. Based on the contribution weights of radar and transient electromagnetic, the extracted features are fused to obtain a fused feature vector. Based on the depth of the measuring point and the fused feature vector, the dielectric constant and resistivity are inverted using the trained physical information neural network. Identify geological anomalies in the target area based on dielectric constant and resistivity; The integrated detection device includes a GPR probe module and a TEM probe module. The GPR probe module uses multiple sets of transceiver antennas, and the TEM probe module uses coplanar rectangular transceiver coils to achieve zero magnetic flux by adjusting the transceiver distance.

2. The combined drilling detection method for reducing the number of advance drilling holes in tunnels and lanes according to claim 1, characterized in that: The GPR probe module uses multiple sets of transmitting and receiving antennas, including low-frequency butterfly antennas and high-frequency microstrip antennas, to measure surrounding rock cracks within 0-5m and abnormal body cross-sections within 0-15m.

3. The combined drilling detection method for reducing the number of advance drilling holes in tunnels and lanes according to claim 1, characterized in that: The zero magnetic flux is achieved by adjusting the transmitting and receiving distance. The zero magnetic flux is achieved by calculating the transmitting and receiving distance when the mutual inductance coefficient is 0. The formula of the mutual inductance coefficient is: in, is the vacuum permeability; is the transmitting coil path; is the receiving coil path; is the transmitting coil element; is the receiving coil element; r is the distance between a source point Q1 on the transmitting coil and an arbitrary point Q2 on the receiving coil; is the number of turns of the receiving coil; is the number of turns of the transmitting coil.

4. The combined drilling detection method for reducing the number of advance drilling holes in tunnels and lanes according to claim 1, characterized in that: The specific process of the collection is as follows: (1) Push the integrated detection device into the borehole to the target depth; (2) Start the motor of the integrated detection device and set the scanning mode; (3) The motor drives the antenna to rotate to the initial angle; (4) If the GPR is rotated to the preset angle, the GPR probe module transmits a low-frequency signal and collects and stores it synchronously; otherwise, no transmission or reception is performed; (5) After the low-frequency signal ends, the high-frequency signal is emitted and synchronously collected and stored; (6) If the TEM is rotated to the preset angle, the TEM probe module transmits a transient electromagnetic field and collects and stores the data during the shutdown period; otherwise, no transmission or reception is performed; (7) Rotate to the next angle and repeat steps (3)-(6); (8) After the 360° scan is completed, the acquisition process ends.

5. The combined drilling detection method for reducing the number of advance drilling holes in tunnels and lanes according to claim 1, characterized in that: The feature extraction is to calculate the reflection coefficient and spatial derivative of radar data, and the depth domain and time domain change rate of transient electromagnetic data respectively; The contribution weight is a contribution weight based on the automatic adjustment feature of the signal-to-noise ratio (SNR).

6. The combined drilling detection method for reducing the number of advance drilling holes in tunnels and lanes according to claim 1, characterized in that: The physical information neural network uses data loss and physical loss as loss functions and adopts a hybrid optimization algorithm based on genetic algorithm and L-BFGS for training optimization.

7. A drilling joint detection system that can reduce the number of advance drilling holes in tunnels and lanes, characterized by: include: an acquisition module configured to: acquire radar data and transient electromagnetic data of a measuring point depth and a target area, wherein the radar data and transient electromagnetic data are acquired by an integrated detection device in the borehole; The extraction module is configured to: extract features from the radar data and the transient electromagnetic data respectively, fuse the extracted features based on the contribution weights of the radar and the transient electromagnetic data, and obtain a fused feature vector; The inversion module is configured to: invert the dielectric constant and resistivity based on the measured point depth and the fused feature vector using the trained physical information neural network; The identification module is configured to: identify geological anomalies in the target area based on dielectric constant and resistivity; The integrated detection device includes a GPR probe module and a TEM probe module. The GPR probe module uses multiple sets of transceiver antennas, and the TEM probe module uses coplanar rectangular transceiver coils to achieve zero magnetic flux by adjusting the transceiver distance.

8. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the drilling joint detection method capable of reducing the number of advance drilling holes in tunnels and lanes as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium, characterized in that The non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by the processor, the drilling joint detection method capable of reducing the number of advance drilling holes in tunnels and lanes as described in any one of claims 1 to 6 is implemented.

10. An electronic device, characterized in that: include: A processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the drilling joint detection method that can reduce the number of advance drilling in tunnels and lanes as described in any one of claims 1 to 6.

Images