While-drilling earthquake-electromagnetic fusion geological anomalous body identification system and method

Abstract

The invention belongs to the technical field of while-drilling geological anomalous body identification, and discloses a while-drilling seismic-electromagnetic fusion geological anomalous body identification system and method, and the system comprises a ground excitation system which is configured to excite seismic waves and electromagnetic waves into a rock mass at the same time; the while-drilling receiving system is configured to receive seismic wave and electromagnetic wave signals from the geological anomalous body in the drilling and digging process, perform noise reduction preprocessing on the signals and compress the signals into data packets; the underground transmission system is configured to transmit the preprocessed data packet to an early warning center on the ground in real time; the early warning center is configured to extract data packet parameters, calculate the data packet parameters and generate comprehensive early warning information including risk levels, positions and scales according to calculation results; a calculation result is obtained through a GAFF (geologic anomaly fusion factor) formula. According to the method, geological forecast is converted into a continuous and adjoint type working process from a discontinuous and staged working process, and the overall construction efficiency of tunnel and well drilling construction is greatly improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of geological anomaly body identification while drilling, and particularly relates to a seismic-electromagnetic fusion geological anomaly body identification system and method while drilling. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute the prior art.

[0003] In the construction of tunnels and drillings, unknown geological anomaly bodies (such as faults, fracture zones, karst caves, water-containing pockets, etc.) in front are the primary risk sources that restrict safety, efficiency and cost. In the construction process, a considerable proportion of disasters such as water inrush, mud gushing and collapse caused by sudden encounter with major geological anomaly bodies accounts for a considerable proportion of the total amount of tunnel and drilling engineering accidents.

[0004] At present, the prediction of the geological conditions in front mainly relies on single type parameters obtained from initial geological survey data, geophysical advanced prediction (TGP) or measurement while drilling (MWD). Due to the long line of the tunnel and drilling engineering, the density and accuracy of the survey points are limited, and the geological conditions revealed by the initial geological survey report often differ significantly from the actual tunneling conditions, which cannot accurately guide the construction. At the same time, due to the narrow working space, strong vibration of the tunneling equipment and serious electromagnetic interference in the tunnel and drilling engineering, the traditional advanced geological prediction method (such as the seismic wave method and the electromagnetic wave method) must be implemented in the working face in the stopped state, which is complicated and inefficient, and the obtained data are intermittent and static. In addition, the existing measurement while drilling technology mainly focuses on collecting and analyzing the mechanical parameters (such as the drilling pressure, torque and rotating speed) of the drilling machine, and indirectly inferring the lithology through the "rock-machine response" relationship. This method has poor anti-interference ability, strong multi-solution property and is extremely insensitive to key characteristics such as water content of the rock mass.

[0005] Geological anomaly bodies are complex geological phenomena with multiple attribute characteristics (such as wave impedance, electrical property and water content). The above methods rely on single type isolated indicators for judgment, ignore the cross verification and fusion complementation between various physical field information, and the accuracy and reliability of the prediction results are difficult to guarantee.

[0006] The existing technology discloses a tunnel prediction device for seismic wave and electromagnetic wave joint inversion. A borehole is drilled in the tunnel sidewall, and a shock rod, an electromagnetic emission rod and a composite detector are alternately arranged. The shock rod and the electromagnetic emission rod are connected with the controller through the concentrator, the composite detector synchronously receives the electromagnetic wave signal and the seismic wave signal, and the composite detector is connected with the controller through the concentrator. The excitation and reception of the seismic wave signal and the electromagnetic wave signal are realized, and the joint inversion analysis is performed to comprehensively judge the fracture development condition and the water content in front of the tunnel working face, which to some extent overcomes the multi-solution limitation of the single geophysical prospecting method.

[0007] The above scheme has the following defects: The above scheme clearly requires the excitation and reception of the seismic wave signal first, and then the equipment is replaced to excite and receive the electromagnetic wave signal. This time-sharing and alternating working mode causes the seismic wave and electromagnetic wave data to come from the rock mass state at different times. In a complex dynamic construction environment, the rock mass stress, underground water and other factors may change over time, causing the data of the two physical fields to not be strictly aligned in time, thereby introducing additional uncertainty and error for subsequent data fusion and joint inversion, affecting the accuracy of the prediction result. SUMMARY

[0008] Therefore, the purpose of the present application is to provide a drilling seismic-electromagnetic fusion geological anomaly body identification system and method, which can solve the technical problem that the seismic wave signal and electromagnetic wave signal in the prior art cannot be strictly aligned in time, affecting the accuracy of the prediction result.

[0009] To achieve the above purpose, the present application adopts the following technical scheme: In a first aspect, a drilling seismic-electromagnetic fusion geological anomaly body identification system is provided, comprising: A ground excitation system configured to excite seismic waves and electromagnetic waves into the rock mass at the same time; A drilling receiving system configured to receive seismic wave and electromagnetic wave signals from the geological anomaly body during drilling, and to perform noise reduction preprocessing and compression into data packets on the signals; A downhole transmission system configured to transmit the preprocessed data packets to the ground in real time to a warning center; The warning center is configured to extract data packet parameters and calculate, and generate comprehensive warning information containing risk level, location and scale according to the calculation result; The calculation result is obtained by the GAFF (geological anomaly fusion factor) formula, and the GAFF formula is: ; In the formula: GAFF is the geological anomaly fusion factor; is the P-wave velocity of the measurement point; is the reference P-wave velocity; is the apparent resistivity of the measurement point; is the reference apparent resistivity; is the seismic wave attenuation coefficient of the measurement point; is the reference attenuation coefficient; is the S-wave velocity of the measurement point; represents the wave velocity ratio; is a weight coefficient, ; is an exponential factor.

[0010] The parameters of the GAFF formula are preferably adaptively adjusted by a Bayesian optimization algorithm; the Bayesian optimization uses the parameters of the drilled sections 、 、 as a training set to minimize the loss function between the GAFF predicted value and the real geological label, and the output optimal parameter set is applied to the GAFF formula, and the parameter set The parameters are updated every preset distance drilled.

[0011] Preferably, the ground excitation system comprises a seismic emission source and an electromagnetic emission source, and the seismic emission source and the electromagnetic emission source are synchronously excited, and the excitation time stamp is recorded.

[0012] Preferably, the while-drilling receiving system comprises a sensor array, the sensor array is connected to a downhole preprocessing unit, the sensor array comprises a seismic sensor array and an electromagnetic sensor array, the seismic sensor array is embedded in the wall of the drill collar in a ring distribution and an axial distribution; the electromagnetic sensor array is installed on the composite insulation window of the outer wall of the drill collar.

[0013] Preferably, the downhole preprocessing unit is configured to perform matched filtering noise reduction on the signals of the sensor array, identify the first arrival time of the seismic wave, compress the preprocessed data into data packets, and output the time-stamped data to the downhole transmission system.

[0014] Preferably, the downhole transmission system comprises an adaptive routing protocol module and an ITM module, the ITM module is encapsulated in the two end connectors of each drill pipe section to form a network topology; the adaptive routing protocol module is configured to calculate the optimal transmission path based on the signal-to-noise ratio, the bit error rate, the signal strength, and the network topology.

[0015] Preferably, the early warning center comprises a decoding module, a calculation module and an early warning module; The decoding module is configured to extract the P-wave velocity , the S-wave velocity , the apparent resistivity and the seismic wave attenuation coefficient of the data packet, calculate the distance and azimuth of the abnormal body from the sensor, and assign an absolute time label to each data packet sample according to the extracted time stamp; The calculation module is configured to run the GAFF formula and output the GAFF value; The early warning module is configured to divide the GAFF range threshold by machine learning based on the existing engineering data, and combine the GAFF value output by the calculation module to perform early warning.

[0016] Preferably, the GAFF formula is or a plurality of or the average or median value.

[0017] Preferably, the early warning module is configured to: cluster the historical GAFF values using the K-Means clustering algorithm, set the early warning threshold values of the three risk levels of normal, slight abnormality and severe abnormality, and compare the GAFF value output by the calculation module with the early warning threshold value to obtain the risk level; extract the distance and direction data of the abnormal body distance sensor obtained by the decoding module; observe the continuous change of the GAFF value output by the calculation module, obtain the initial depth and the end depth of the abnormal body, and the difference between the two is the size of the abnormal body in the tunneling direction; fuse all the above information to generate comprehensive early warning information containing the risk level and the size and location information of the geological abnormal body.

[0018] In the second aspect, the identification method of the geological abnormal body identification system is provided, and the specific steps include: The ground excitation system is applied to synchronously emit seismic waves and electromagnetic wave signals into the rock mass, and record the excitation time stamp; The seismic wave and electromagnetic wave signals from the geological abnormal body are synchronously received by the while-drilling receiving system during the drilling process, and the seismic wave and electromagnetic wave signals are denoised and compressed into data packets; The preprocessed data packets are transmitted to the ground in real time through the downhole transmission system; The data packets are processed by the early warning center, and the geological abnormal fusion factor GAFF is calculated, and the comprehensive early warning information containing the risk level, the location and the size is generated according to the calculation result.

[0019] Compared with the prior art, the present application has the advantages and positive effects that: The ground excitation system of the application synchronously excites seismic waves and electromagnetic waves into the rock mass, the while-drilling receiving system synchronously receives seismic signals and electromagnetic signals from the geological anomaly body in the drilling process, and performs noise reduction preprocessing on the received seismic signals and electromagnetic signals, and compresses them into data packets for transmission to the early warning center through the downhole transmission system for processing and calculation to generate comprehensive early warning information containing risk level, location and scale; the while-drilling seismic-electromagnetic fusion geological anomaly body identification system of the application changes the geological prediction from an intermittent and phased process to a continuous and accompanying process, greatly improving the overall construction efficiency of tunnel and drilling construction. Moreover, the GAFF formula of the application adopts a nonlinear multiplication model of "weighted summation + exponential modulation", which goes beyond simple linear weighting and can more accurately depict complex geological responses. The GAFF formula nonlinearly fuses and jointly interprets seismic attributes and electromagnetic attributes, realizes complementary advantages, overcomes the multi-solution bottleneck of single physical methods, significantly improves the accuracy and reliability of detection, and provides a reliable basis for engineering construction site decision-making. BRIEF DESCRIPTION OF DRAWINGS

[0020] The drawings accompanying the specification of this application form a part thereof, serve to provide further understanding of the application, and together with the description, explain the application. The embodiments of the application and its description are used to explain the application without imposing undue limitation on the application.

[0021] Figure 1 is a schematic diagram of a while-drilling seismic-electromagnetic fusion geological anomaly body identification system according to the embodiment 1 or the embodiment 2 of the application. DETAILED DESCRIPTION It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the application belongs.

[0022] The application will be described in detail below with reference to the accompanying drawings.

[0023] Embodiment 1 The embodiment discloses a while-drilling seismic-electromagnetic fusion geological anomaly body identification system, as shown in Figure 1 The system includes a ground excitation system, a while-drilling receiving system, a downhole transmission system and an early warning center. The ground excitation system is configured to simultaneously excite seismic waves and electromagnetic waves into the rock mass for the while-drilling receiving system to receive signals. The while-drilling receiving system is configured to synchronously receive seismic signals and electromagnetic signals from the geological anomaly body in the drilling process, and perform noise reduction preprocessing on the received seismic signals and electromagnetic signals, and compress them into data packets for the downhole transmission system to transmit. The downhole transmission system is configured to transmit the preprocessed data packets to the early warning center on the ground in real time; The early warning center is configured to process the transmitted data packets, extract data packet parameters and calculate, generate comprehensive early warning information containing risk level, location and scale according to the calculation result; The calculation result is obtained by GAFF (geological anomaly fusion factor) formula, and the GAFF formula is: ; Among them: GAFF is the geological anomaly fusion factor; is the P-wave velocity of the measurement point; is the reference P-wave velocity; is the apparent resistivity of the measurement point; is the reference apparent resistivity; is the seismic wave attenuation coefficient of the measurement point; is the reference attenuation coefficient; is the S-wave velocity of the measurement point; represents the wave velocity ratio; is the weight coefficient, ; is the exponential factor.

[0024] In this embodiment, the output value of the GAFF calculation represents the comprehensive risk level of the geological anomaly body, and the higher the GAFF value, the greater the risk.

[0025] In this embodiment, the parameter set of the GAFF formula is given to the GAFF formula, which has flexible adjustability and strong adaptability, so that the GAFF formula can learn from historical data, dynamically adjust the contribution of each parameter according to the geological characteristics of the specific project through machine learning algorithm (such as Bayesian optimization), find the optimal configuration that best fits the current environment, and realize lifelong learning and dynamic self-adaptation of the formula.

[0026] The feature X of Bayesian optimization is the parameter , , the label Y is the artificial interpretation result from the field cutting logging (for example, a geological engineer has marked a certain section as "complete limestone-0, water-bearing fracture zone-1"), and a loss function is defined, the goal is to minimize the difference between the GAFF prediction value and the true label, for example, using mean square error loss. Through Bayesian optimization optimization, the optimal parameter set found in the history is output and applied to the GAFF formula, while setting rolling update, for example, the system automatically triggers a new round of Bayesian optimization every 20 meters of drilling, updates the high-dimensional adjustable parameter space, so that the dynamic self-adaptive learning of the GAFF formula runs through the entire engineering construction.

[0027] In the embodiment, the Bayesian optimization algorithm is trained by the actual geological data of the drilled section to obtain the parameter set The optimal parameter combination is automatically found, so that the GAFF formula has self-learning and adaptability and can be self-optimized according to different geological environments.

[0028] In the embodiment, the ground excitation system includes a seismic emission source and an electromagnetic emission source, the ground excitation system sends a trigger signal to the seismic emission source and the electromagnetic emission source according to the built-in expert knowledge base, and the while-drilling receiving system starts to work and receive and process signals after the excitation signal is sent. The built-in expert knowledge base recommends optimal excitation parameters (seismic scanning frequency band, scanning length, electromagnetic emission frequency and emission power) according to different lithology, different burial depth and different detection targets, to ensure the accuracy of geological anomaly body identification.

[0029] In the embodiment, the seismic emission source is a controllable seismic source, a servo system of the seismic emission source drives a vibrator to vibrate according to the trigger signal, so that the tunnel floor or wall rock coupled with the seismic emission source radiates specific seismic waves into the stratum; the electromagnetic emission source is composed of a high-power transmitter and a transmission coil arranged in the tunnel, and is configured to emit electromagnetic wave signals with known characteristics into the rock mass. During the entire excitation process, the seismic emission source and the electromagnetic emission source ensure completely synchronous excitation, and record the excitation time stamp for absolute time alignment of data, so as to receive subsequent signals and calculate after decoding.

[0030] It should be noted that the ground excitation system is the starting end of the entire detection and identification process, and the core purpose is to generate high-quality and high-synchronization source signals. Before the tunnel and drilling construction starts or when a new geological unit is entered, the field construction personnel set the geological anomaly body target of this detection cycle (for example, a large fault or a water-bearing cave), the ground excitation system calls the built-in expert knowledge base after receiving the instruction, the expert knowledge base is composed of pre-trained software modules, stores a large amount of simulation and historical project data, and establishes an optimal mapping relationship between the geological target and the excitation parameters. For example, when the detection target is to detect a small cave, the knowledge base will recommend high-frequency and short scanning length seismic waves, because high-frequency signals have high resolution and are suitable for detecting small-scale anomalies, and electromagnetic sources with high frequency and moderate power, because high-frequency electromagnetic fields are more sensitive to small conductors.

[0031] In the embodiment, the seismic signals and electromagnetic signals from the geological anomaly body are received synchronously during the drilling process through the while-drilling receiving system, and the received signals are preprocessed by the downhole preprocessing unit in the while-drilling receiving system.

[0032] The while-drilling receiving system comprises a sensor array and a downhole preprocessing unit, receives seismic wave signals and electromagnetic wave signals through the sensor array, and performs noise reduction and compression through the downhole preprocessing unit, and subsequently transmits data to a warning center to perform GAFF calculation to identify a geological anomaly body.

[0033] In the embodiment, the sensor array comprises a seismic sensor array and an electromagnetic sensor array, and the seismic sensor array and the electromagnetic sensor array are electrically connected with the downhole preprocessing unit; the seismic sensor array adopts three-component MEMS accelerometers (hereinafter referred to as "accelerometers"), and a plurality of accelerometers are embedded in the wall of the drill collar in a ring-shaped distribution (to capture different azimuth seismic wave field information) and an axial distribution (to form spatial sampling, which is helpful for seismic wave field separation and velocity analysis), when the seismic waves excited by the ground and the waves reflected from the front anomaly body propagate to the positions of the accelerometers, each accelerometer simultaneously measures the vibration acceleration in the X, Y and Z directions, so as to completely record the vector information of the seismic wave length.

[0034] In the embodiment, the electromagnetic sensor array is installed on the composite insulating window of the outer wall of the drill collar to overcome the electromagnetic shielding effect of the metal drill collar, detect the change of the total magnetic field, measure the change rate of the magnetic field by the coil, measure the potential difference in the formation by the electrode, and synchronously measure the electromagnetic field vector. By installing the sensor array on the drill collar, the while-drilling receiving system can receive seismic wave and electromagnetic wave signals and obtain real-time while-drilling signals.

[0035] It can be understood that the prior art requires a series of cumbersome procedures such as drilling, installation and coupling of physical rods (shaking rods, electromagnetic emission rods and composite detectors) on the tunnel sidewall, which not only requires the tunneling equipment to stop working, but also takes a long time in the entire detection process, and the obtained data is static and discrete, which cannot be synchronized with the tunneling process, forming an intermittent operation mode of "tunneling - stopping - detection - tunneling again", which affects the overall construction efficiency. The while-drilling receiving system of the embodiment can obtain real-time while-drilling signals, which can improve the overall construction efficiency.

[0036] In the embodiment, the signals received by the sensor array are weak and rich in noise, and the downhole preprocessing unit is configured to perform matched filtering noise reduction on the signals of the sensor array, extract weak effective signals from strong background noise, perform first arrival picking and time marking, identify the first arrival time of the seismic wave, finally compress the preprocessed data into data packets through wavelet transform, and mark the time stamp, and output the processed and compressed data packets to the wireless transmission module in the downhole transmission system.

[0037] In the embodiment, the data packet preprocessed by the downhole preprocessing unit is transmitted to the early warning center on the ground in real time through a downhole transmission system. The downhole transmission system comprises an adaptive routing protocol module and a wireless transmission module (hereinafter referred to as an “ITM module”), the ITM module can perform a store-and-forward function, the ITM module is encapsulated in the two end connectors of each drill pipe to form a dynamic network node, and the downhole transmission system adopts a wireless Mesh self-organizing network technology to disperse the transmission task to each network node of the drill pipe and replace single long-distance transmission by a multi-hop relay mode. In the embodiment, when the ITM module of each drill pipe is powered on and started after being lowered into the well, the ITM module automatically searches for the adjacent ITM module nodes above and below to establish a connection and form a network topology by sending a “beacon” signal.

[0038] The adaptive routing protocol module is configured to calculate an optimal transmission path with the highest comprehensive quality to the ground (maximum bandwidth, lowest error rate and most stable) based on the signal-to-noise ratio, error rate, signal strength and network topology, and the data packet is transmitted to the early warning center by the optimal transmission path. The data packet transmission mainly adopts millimeter waves to transmit through the natural waveguide formed by the internal metal cavity of the drill pipe, and each intermediate node of the ITM module plays a role in store-and-forward until the data packet is transmitted to the early warning center.

[0039] In the embodiment, the core algorithm of the adaptive routing protocol module is to dynamically calculate an optimal transmission path with the highest comprehensive quality to the ground by a multi-index weighted fusion mechanism, which organically combines the four key parameters of the signal-to-noise ratio, error rate, signal strength and network topology. The original data of different dimensions are converted into normalized scores between 0 and 1 (the higher the signal-to-noise ratio, the higher the score, the lower the error rate, the higher the score, and the stronger the signal strength, the higher the score), and the network topology is calculated by considering the number of paths (the fewer the better) to calculate the topology score. Then appropriate weights are assigned to the four indexes according to the actual working conditions on site, and then the comprehensive quality score of each path is calculated by a weighted summation formula, and the correlation between the indexes is corrected, and a load balancing factor is introduced to avoid network congestion. Finally, the adaptive routing protocol module selects the path with the highest comprehensive score, but sets a hysteresis threshold to prevent frequent switching of the path, and ensures that the selected path meets the minimum quality requirement for transmission (set on site).

[0040] Specifically, the allocation of weights is not static but dynamically adjusted based on actual field conditions through an adaptive learning algorithm to ensure that the path selection is always optimal. The initial weight setting usually relies on expert experience and historical network data. To adapt to the dynamically changing network environment, the weights can be optimized in real time through online machine learning algorithms, such as through a deep reinforcement learning model, in which the adaptive routing protocol module acts as an agent that selects paths based on the current network state (such as node energy, packet loss rate) and updates the weight parameters through a reward function (such as data transmission success rate, delay penalty). The reward function is designed as a linear combination of weighted indicators, and the long-term loss is minimized through gradient descent method, so as to automatically adjust the weights to maximize the comprehensive quality score.

[0041] In addition, the weight allocation also needs to consider the correlation between indicators, such as the positive correlation between signal-to-noise ratio and signal strength. If both are given high weights, it may cause overfitting, so through principal component analysis (PCA) or covariance matrix correction to decorrelate, ensure the independence and fairness of weight allocation. The load balancing factor also indirectly affects the weight, for example, when the load of a certain path is high, the system will temporarily reduce its topology weight to avoid congestion (this part is realized by introducing a dynamic weight adjustment term in the weighted summation formula, such as weight = base weight x (1-current load / max load)), so that when calculating the comprehensive quality score, the path with light load is automatically selected preferentially.

[0042] The appropriate value of the maximum weight is also fine-tuned through field testing and simulation verification. Before deployment, network simulator (NS-3) is used to simulate transmission scenarios under different geological conditions, and grid search or Bayesian optimization is used to find the weight combination that minimizes the average delay and maximizes the throughput, and recalibration is performed periodically during runtime to adapt to complex environmental changes. This dynamic weight mechanism ensures that the adaptive routing protocol module not only responds quickly to network fluctuations, but also optimizes itself through data-driven methods in the long run, thereby maintaining efficient and reliable data transmission.

[0043] In this embodiment, ITM modules are built into each section of drill pipe, and the data transmission line is optimized by the adaptive routing protocol module, ensuring the timeliness and integrity of massive data transmission.

[0044] In this embodiment, the warning center receives and calculates the data packets transmitted by the downhole transmission system, and generates comprehensive warning information containing risk level, location and scale according to the calculation results. The warning center includes a decoding module, a calculation module and a warning module.

[0045] The seismic attribute and electromagnetic attribute of the data packet are extracted by the decoding module to obtain parameters required in a GAFF (geological anomaly fusion factor) calculation formula with a multi-parameter nonlinear fusion and a reinforcement learning mechanism created in a subsequent calculation module; the GAFF formula is run by the calculation module to calculate and output a GAFF value; and the warning module divides a GAFF range threshold based on existing engineering data through machine learning, and compares the GAFF value output by the calculation module to give a warning.

[0046] In the embodiment, the decoding module is configured to extract seismic attributes and electromagnetic attributes of the data packet to obtain P-wave velocity , S-wave velocity , apparent resistivity and seismic wave attenuation coefficient , calculate the distance and direction of the anomaly body distance sensor; and the decoding module assigns an absolute time label to each data packet sample according to the extracted time stamp.

[0047] The decoding module demodulates the uploaded data packet, parses the packet header of the data packet according to a predefined communication protocol, obtains information such as packet number, source depth and data length, performs cyclic redundancy check, and performs wavelet inverse transform to reconstruct the data from the frequency domain back to the time domain to obtain complete and denoised seismic waveform channels and electromagnetic waveform channels; the decoding module assigns a unified and high-precision absolute time label to each data packet sample according to the extracted time stamp, so that subsequent calculations are based on a unified time reference.

[0048] In the embodiment, the decoding module automatically and accurately identifies the first arrival time of P-wave and S-wave by using the Akaike information criterion for the decompressed seismic waveform channel, and calculates the wave velocity to obtain the wave velocity ratio; the frequency spectrum ratio method is used to select a direct wave time window and a background noise time window, and Fourier transform is performed to obtain the frequency spectrum, and the attenuation coefficient can be obtained by linear fitting combined with the calculation formula of the frequency spectrum. The decoding module processes the decompressed electromagnetic waveform channel to obtain the amplitude ratio and phase difference of the received signal and the transmitted signal, and converts these observation values into apparent resistivity and phase data by solving the coupled equations according to the frequency domain electromagnetic method theory. The decoding module finally outputs a series of parameter physical groups with clear meaning, including P-wave velocity , S-wave velocity , apparent resistivity and seismic wave attenuation coefficient , and arranged by depth.

[0049] In the embodiment, the decoding module calculates the distance of the anomaly body distance sensor according to the principle of seismic reflection wave by using the reflection wave travel time and the formation velocity; and estimates the direction of the incoming reflection wave by using polarization analysis or beam forming technology based on the seismic wave field information received by the seismic sensor array to determine the direction of the anomaly body.

[0050] In the present embodiment, the calculation module is configured to run a GAFF formula with multi-parameter nonlinear fusion of reinforcement learning mechanism, the GAFF formula being: ; wherein: GAFF is a geological anomaly fusion factor, as a final calculation output value, the value of which represents the comprehensive risk level of the geological anomaly body, the higher the GAFF value, the greater the risk; Vp represents the P-wave velocity of the current measurement point, reflecting the strength and integrity of the rock; Vp represents the P-wave velocity of the current measurement point, reflecting the strength and integrity of the rock; Rt represents the apparent resistivity, the apparent resistivity of the current measurement point, reflecting the water content or conductivity of the formation; Rt represents the apparent resistivity, the apparent resistivity of the current measurement point, reflecting the water content or conductivity of the formation; D represents the attenuation coefficient, the seismic wave attenuation coefficient of the current measurement point; D represents the attenuation coefficient, the seismic wave attenuation coefficient of the current measurement point; Vs represents the S-wave velocity, the shear wave velocity of the current measurement point; Vp / Vs represents the wave velocity ratio, used to identify the pore fluid properties and lithology; constitute an adjustable parameter space wherein W represents a weight coefficient, respectively giving the weight of the velocity, resistivity and attenuation anomalies, ; Ex represents an exponential factor, representing a reinforcement factor and a modulation factor, used for nonlinear amplification or modulation of the corresponding item's abnormal response.

[0051] It should be noted that each background value in the GAFF formula: or is obtained through historical data, for example, the field construction personnel can select the average value or median value of each parameter , , within a set distance (such as the last 50m drilled) as the background reference value , , This ensures that the GAFF formula adapts to changes in the surrounding rock background. For each depth, the calculation module substitutes the physical quantities extracted by the decoding module into the GAFF formula, and simultaneously substitutes the corresponding background reference values ​​into the GAFF formula, calculating the GAFF value in real time.

[0052] It needs to be explained that the GAFF formula is essentially a nonlinear multi-parameter fusion model with a deep foundation in rock physics. It adopts a multi-physics nonlinear deep fusion architecture, specifically a nonlinear multiplication model of "weighted summation + exponential modulation", which surpasses simple linear weighting and is more in line with the response law of geological anomalies. The core input of the GAFF formula is not the absolute value of the physical quantity, but the relative rate of change of the physical quantity with respect to the background value (logarithmic ratio). This enables consistent evaluation across the entire domain. The logarithmic value measures the percentage change of the physical quantity relative to the background value, which is a dimensionless quantity. This frees the algorithm from the constraints of absolute values ​​and greatly improves the adaptability and consistency of the formula in different geological regions. The physical significance of this design lies in the fact that it measures the percentage change of a physical quantity rather than the difference. Taking the logarithm makes the increase and decrease of the same proportion symmetrical, which is more in line with the characteristics of geophysical response. It minimizes the differences caused by different surrounding rock backgrounds, in order to achieve a consistent evaluation across the entire domain and make the GAFF formula adaptable to complex geological environments.

[0053] In this embodiment, the addition term part of the GAFF formula An exponential enhancement factor was introduced. This gives the GAFF formula the intelligent characteristics of ignoring weak anomalies and being sensitive to significant anomalies, effectively suppressing background noise interference, greatly highlighting strong anomaly signals, and significantly improving the signal-to-noise ratio and the ability to identify major risk entities.

[0054] When the exponential enhancement factor When the value is greater than 1, the GAFF formula will produce a nonlinear response to changes in physical quantities. For small changes, its contribution increases slowly; but for significant changes, its contribution will be amplified exponentially, making the GAFF formula selectively changeable. It can effectively suppress background noise and interference from small fluctuations, while becoming more sensitive to truly noteworthy and strong signals, greatly improving the signal-to-noise ratio and the ability to distinguish major risk entities.

[0055] The key innovation of the GAFF formula lies in the fluid effect modulator. The GAFF formula does not simply... Instead of adding an addition term, it is treated as an independent multiplication modulation term. When the preceding addition term calculates a high outlier, this value will be combined with... Multiply, if at this time The values ​​are also high and If the value is greater than 0, the outlier calculated by the addition term will be amplified a second time. If the value is not high, the outlier calculated by the additive term will not be amplified. This conditional amplification mechanism allows the GAFF formula to identify various geological anomalies. The GAFF formula introduces a fluid effect modulator section— This solves the problem of multiple solutions to anomalies such as "low speed and low resistance" and enables quantitative identification of the water content of anomalies.

[0056] In this embodiment, the parameter set of the GAFF formula Bayesian optimization dynamically adjusts the contributions of various parameters based on the specific geological characteristics of the project to find the optimal configuration that best suits the current environment. The characteristic X of Bayesian optimization is the parameter... , The label Y is the result of manual interpretation from the cuttings logging data (e.g., a geological engineer has labeled a section as "intact limestone-0, water-bearing fractured zone-1"). A loss function is defined to minimize the difference between the GAFF predicted value and the true label. Bayesian optimization is used to find the optimal parameter set found in history and apply it to the GAFF formula. A rolling update is also set; for example, the system automatically triggers a new round of Bayesian optimization every 20 meters of drilling, updating the high-dimensional adjustable parameter space. This allows the dynamic adaptive learning of the GAFF formula to continue throughout the entire engineering construction process.

[0057] It's important to explain that Bayesian optimization is an advanced sequence optimization strategy used to optimize "black-box functions." In the GAFF formula, the "black-box function" to be optimized is the function defined by the parameter set... The function is taken as input and outputs a loss function L(P) (such as the mean square error between GAFF predictions and true geological labels).

[0058] Its implementation process is an iterative "hypothesis-evaluation-update" loop, as follows: First, we need to check the parameter set. Randomly select a small number (e.g., 5-10) different parameter combinations P1, P2, ..., P from the parameter space. n For each set of parameters, an evaluation must be performed: that is, the GAFF values ​​of all points in the entire drilled section are calculated using these parameters, and compared with the actual geological labels obtained from cuttings logging, to calculate a loss function value L(P1), L(P2), ..., L(P n） These initial points provide a basic understanding of the objective function.

[0059] The next step is the core loop of Bayesian optimization. The first step is to construct a surrogate model. A Gaussian process (GP) is chosen as the surrogate model because it can not only predict the function values ​​at unknown points but also provide the uncertainty (variance) of the prediction. GP will evaluate the points {P}... i , L(P i As observed data, the covariance function (kernel function) is used to characterize the similarity between any two points in the parameter space, thereby learning the probability distribution of the entire parameter space. Ultimately, GP forms a global probability estimate of the true loss function: for any set of untried parameters P*, GP will give a predicted value μ(P*) (i.e., the predicted average loss) and σ(P*) (i.e., the predicted uncertainty) that follow a normal distribution.

[0060] The second step is to optimize the acquisition function to determine the next evaluation point. The role of the acquisition function is to use the predictions (μ and σ) provided by the generalized system (GP) to weigh "exploration" against "exploitation". Expected Improvement (EI) is a commonly used and efficient acquisition function, EI(P) = E[max(0, Lmin-L(P))], where Lmin is the minimum loss value found so far.

[0061] The EI value is high in two places: first, where the GP prediction loss is low (μ(P) is small, which can be utilized); and second, where the GP prediction uncertainty is high (σ(P) is large, which can be explored). The parameter point Pnext that maximizes the EI(P) value is found using optimization algorithms (such as L-BFGS-B or differential evolution).

[0062] The third step is to evaluate the new points and update the model. A complete GAFF calculation and loss evaluation are performed using the found Pnext parameter set to obtain the true L(Pnext). This new data point {Pnext, L(Pnext)} is then added to the initial observation dataset. The GP surrogate model uses this augmented dataset to retrain and predict, updating its understanding of the entire parameter space and making it more accurate.

[0063] Finally, repeat steps two and three, iterating this process continuously. Each iteration either discovers a better point (development) or explores an unknown region (exploration), thus gradually narrowing the search range for optimal parameters. The loop continues until a preset number of iterations is reached (e.g., 100), or the loss function no longer shows significant improvement in consecutive iterations (convergence). Ultimately, the algorithm outputs the best-performing set of parameters P found throughout the entire optimization history. optimal This process efficiently finds the global optimal solution for complex black-box functions with as few expensive "evaluations" as possible, perfectly solving the problem of adaptive tuning of GAFF parameters.

[0064] In summary, the GAFF formula is an intelligent framework that deeply integrates rock physics principles, nonlinear mathematical enhancement, fluid intelligent identification, and adaptive machine learning. In this embodiment, the calculation process of the computation module is performed in parallel on the server, outputting the GAFF curve that varies with depth in real time.

[0065] In this embodiment, the early warning module of the early warning center sets a risk threshold using unsupervised learning, such as the K-Means clustering algorithm. The early warning module collects all GAFF values ​​calculated from data within a certain period or historical data, sets the number of clusters to 3, and the K-Means clustering algorithm automatically clusters these GAFF values ​​into three clusters, using the boundaries between clusters as the threshold. Assuming the centroids of the three clusters are ordered as follows... This automatically determines "normal" (GAFF< "Minor abnormalities" <GAFF< 2) "Severe abnormality" (GAFF>) 2) A three-level early warning system is implemented. Based on the principle of seismic reflection waves, the distance between the anomaly and the sensor is calculated using the travel time of the reflected wave and the formation velocity. This distance calculation is completed by the decoding module, and the early warning module directly calls this result. Using the seismic wavefield information received by the seismic sensor array, polarization analysis or beamforming technology is used to estimate the direction of the reflected wave arrival, thereby determining the location of the anomaly. By observing the continuous changes in the GAFF curve during drilling, the starting and ending depths of the anomaly can be clearly seen; the difference between the two is the extent (scale) of the anomaly in the drilling direction. Finally, the early warning module integrates all the above information to generate a final comprehensive early warning message containing the risk level, size, and location information of the geological anomaly, providing a basis for decision-making by on-site construction personnel.

[0066] Example 2 This embodiment discloses a geological anomaly identification method based on seismic-electromagnetic fusion during drilling, which utilizes a geological anomaly identification system based on seismic-electromagnetic fusion during drilling disclosed in Embodiment 1. The specific steps are as follows: A ground-based excitation system was used to simultaneously transmit seismic and electromagnetic wave signals into the rock mass and record the excitation timestamp. During the drilling process, the system simultaneously receives seismic and electromagnetic wave signals from the response of geological anomalies through a drilling receiving system, and then denoises and compresses the seismic and electromagnetic wave signals into data packets. The pre-processed data packets are transmitted to the surface in real time via an underground transmission system. The early warning center processes data packets and calculates the geological anomaly fusion factor (GAFF). It then classifies historical GAFF values, determines the threshold boundaries for normal, minor anomalies, and severe anomalies, compares real-time GAFF values ​​with these thresholds, and triggers a level-three early warning.

[0067] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A geological anomaly identification system integrating seismic-electromagnetic fusion during drilling, characterized in that, include: The ground-based excitation system is configured to simultaneously excite seismic waves and electromagnetic waves into the rock mass; The drilling receiving system is configured to receive seismic and electromagnetic wave signals from geological anomalies during the drilling process, and to perform noise reduction preprocessing on the signals and compress them into data packets. The underground transmission system is configured to transmit pre-processed data packets to the surface early warning center in real time; The early warning center is configured to extract data packet parameters and calculate them, and generate comprehensive early warning information including risk level, location and scale based on the calculation results; The calculation results were obtained using the GAFF formula, which is: ; In the formula: GAFF is the geological anomaly fusion factor; The longitudinal wave velocity at the measurement point; For reference P-wave velocity; The apparent resistivity at the measurement point; For reference apparent resistivity; The seismic wave attenuation coefficient at the measurement point; Used as a reference attenuation coefficient; The S-wave velocity at the measurement point; Indicates wave speed ratio; These are the weighting coefficients. ; It is an exponential factor.

2. The geological anomaly identification system based on seismic-electromagnetic fusion during drilling as described in claim 1, characterized in that, The parameter set of the GAFF formula Adaptive adjustments are made using a Bayesian optimization algorithm; Bayesian optimization uses parameters from already drilled sections. , As a training set, the optimal parameter set is output to minimize the loss function between the GAFF predictions and the true geological labels and then applied to the GAFF formula. The parameters are updated after each preset drilling distance.

3. The geological anomaly identification system based on seismic-electromagnetic fusion during drilling as described in claim 1, characterized in that, The ground-based excitation system includes a seismic emission source and an electromagnetic emission source. The seismic emission source and the electromagnetic emission source are excited synchronously, and the excitation timestamp is recorded.

4. The geological anomaly identification system based on seismic-electromagnetic fusion during drilling as described in claim 1, characterized in that, The drilling receiving system includes a sensor array connected to the downhole preprocessing unit. The sensor array includes a seismic sensor array and an electromagnetic sensor array. The seismic sensor array is embedded in the drill collar wall in a ring-shaped and axially distributed manner. The electromagnetic sensor array is installed on the composite material insulating window on the outer wall of the drill collar.

5. The geological anomaly identification system based on seismic-electromagnetic fusion during drilling as described in claim 4, characterized in that, The downhole preprocessing unit is configured to perform matched filtering noise reduction on the signals from the sensor array, identify the first arrival time of the seismic waves, compress the preprocessed data into data packets, and output them to the downhole transmission system with a timestamp.

6. The geological anomaly identification system based on seismic-electromagnetic fusion during drilling as described in claim 1, characterized in that, The downhole transmission system includes an adaptive routing protocol module and an ITM module. The ITM module is encapsulated in the connectors at both ends of each drill pipe section, forming a network topology. The adaptive routing protocol module is configured to calculate the optimal transmission path based on signal-to-noise ratio, bit error rate, signal strength, and network topology.

7. The geological anomaly identification system based on seismic-electromagnetic fusion during drilling as described in claim 1, characterized in that, The early warning center includes a decoding module, a calculation module, and an early warning module; The decoding module is configured to extract the longitudinal wave velocity of the data packets. transverse wave velocity Apparent resistivity and seismic wave attenuation coefficient Calculate the distance and orientation of the abnormal object from the sensor; And based on the extracted timestamps, each data packet sample is assigned an absolute time label; The calculation module is configured to run the GAFF formula and output the calculated GAFF value; The early warning module is configured to use machine learning to divide the GAFF range threshold based on existing engineering data, and combine the GAFF value output by the calculation module to issue an early warning.

8. The geological anomaly identification system based on seismic-electromagnetic fusion during drilling as described in claim 7, characterized in that, The or To acquire multiple within a set distance or The average or median.

9. The geological anomaly identification system based on seismic-electromagnetic fusion during drilling as described in claim 7, characterized in that, The early warning module is configured as follows: The K-Means clustering algorithm is used to cluster historical GAFF values. Warning thresholds for three risk levels—normal, slightly abnormal, and severely abnormal—are set. The risk level is obtained by comparing the GAFF values ​​output by the calculation module with the warning thresholds. Extract the distance and orientation data of the abnormal object from the distance sensor obtained by the decoding module; Observe the continuous change of the GAFF value output by the calculation module to obtain the starting depth and ending depth of the anomaly. The difference between the two is the scale of the anomaly in the tunneling direction. By integrating all the above information, a comprehensive early warning information is generated, which includes the risk level and the size and location information of geological anomalies.

10. The identification method of a geological anomaly identification system based on seismic-electromagnetic fusion during drilling as described in any one of claims 1-9, characterized in that, The specific steps include: The ground excitation system is used to simultaneously transmit seismic and electromagnetic wave signals into the rock mass and record the excitation timestamp; The drilling receiving system synchronously receives seismic and electromagnetic wave signals from geological anomalies during the drilling process, and denoises and compresses the seismic and electromagnetic wave signals into data packets. The pre-processed data packets are transmitted to the surface in real time via the downhole transmission system. The early warning center processes the data packets and calculates the geological anomaly fusion factor GAFF. Based on the calculation results, it generates comprehensive early warning information that includes risk level, location, and scale.

Images