An integrated geophysical system for automatic discrimination

Abstract

The application belongs to the field of geological exploration, and particularly relates to an automatic discrimination comprehensive geophysical prospecting system. A system named an automatic discrimination comprehensive geophysical prospecting system is disclosed, aiming at solving the problems of multi-solution, difficulty in multi-source data fusion and poor real-time performance caused by single geophysical prospecting method. The system comprises a multi-method integrated detection platform, a distributed acquisition and real-time transmission system, an automatic data processing system, an artificial intelligence automatic discrimination and interpretation system and an intelligent decision three-dimensional modeling system. By integrating multi-source detection modules such as gravity, magnetism, electricity and shock, using PTP clock synchronization, ELU self-encoder noise reduction and deep learning multi-model fusion determination technology, the feature extraction, automatic recognition and dynamic modeling of geological anomaly bodies are realized. Through the scheme, the sensitivity and real-time performance of geological body recognition are significantly improved, and the dependence on artificial interpretation is reduced.

Description

Technical Field

[0001] This invention belongs to the field of geological exploration, specifically relating to an automatic identification integrated geophysical exploration system. Background Technology

[0002] Geological exploration, as a core technological means to reveal the internal structure of the Earth's crust, explore mineral resources, and assess geological environmental risks, plays an irreplaceable role in national economic construction, energy security, and disaster reduction in large-scale engineering projects. Its basic principle lies in observing the spatial evolution and instantaneous response of geophysical fields (such as gravitational fields, magnetic fields, electromagnetic fields, and seismic wave fields) using precision instruments, and then inferring the lithology, geological structure, and distribution characteristics of anomalies based on differences in physical properties. As exploration targets shift from shallow to deep, and from simple structures to concealed and complex conditions, single physical detection methods, limited by specific physical response mechanisms, often only reflect one aspect of the geological body's properties, leading to significant "multiple interpretations" in data interpretation.

[0003] To alleviate this contradiction, the existing technological evolution path is gradually moving towards integrated geophysical exploration combining multiple methods. Within the current technological framework, gravity exploration is widely used to identify large-scale tectonic anomalies caused by density differences, magnetic exploration focuses on capturing changes in rock magnetism, electromagnetic exploration utilizes resistivity differences to detect aquifers or metallic ore bodies, and seismic exploration constructs high-resolution stratigraphic frameworks through elastic wave velocity fields. These methods, within their respective professional dimensions, can provide relatively reasonable physical inferences for specific geological targets. However, in practical applications, existing integrated geophysical systems often exhibit a loosely coupled state of physical integration rather than logical fusion.

[0004] This loose coupling has led to deep-seated technical contradictions, the core of which lies in the imbalance between the dimensionality curse caused by multi-source heterogeneous data and the bottleneck of human processing efficiency. Specifically, although integrated systems can simultaneously collect multiple physical field data, the fundamental differences in measurement principles, spatial resolution, temporal sampling frequency, and sensitivity to environmental interference among different geophysical methods result in extremely strong heterogeneity in the original form of massive amounts of data. In the data processing stage, traditional techniques typically rely on offline, step-by-step independent interpretation by experts from various disciplines, followed by a posteriori spatial alignment and feature synthesis based on human experience. This model not only leads to a long exploration cycle, but more seriously, the unavoidable subjective biases and experience limitations in the human interpretation process often obscure weak but crucial correlations between multiple physical fields. Especially when facing hidden geological hazards that are deep, small in scale, and have extremely low signal-to-noise ratios, this human-intervention-based discrimination model is prone to misjudgment or omission.

[0005] The lack of automation and intelligence in existing technologies prevents systems from achieving closed-loop feedback under complex and dynamic field conditions. For example, in borehole geophysical exploration or complex terrain exploration, the non-stationary nature of environmental noise requires algorithms with extremely high adaptive noise reduction capabilities. However, mainstream noise reduction operators often face a trade-off between signal fidelity and noise suppression when processing nonlinear and complex signals. Furthermore, the lack of a unified geological knowledge base and real-time deep learning inference mechanisms prevents the system from instantly identifying and warning of risks from collected abnormal signals. This hinders field personnel from optimizing exploration plans based on on-site feedback, resulting in wasted valuable engineering time and increased exploration costs.

[0006] The limitations of existing geophysical exploration systems lie in their basic architecture's failure to address the deep semantic fusion of multi-source data and the lack of an intelligent logic platform capable of self-evolution and autonomous judgment. With the emergence of strategic needs such as transparent mines and geological support for smart cities, the requirements for exploration accuracy have evolved from qualitative description to quantitative analysis. This demands that the system not only record changes in the physical field but also automatically and in real-time extract features from disordered multidimensional data streams, construct models, and output deterministic geological conclusions.

[0007] Therefore, how to construct a highly integrated comprehensive geophysical exploration system with intelligent perception, fusion, and interpretation capabilities, and which can automate the entire process from data acquisition to 3D modeling, in order to solve the core technical challenges such as the difficulty of multi-source data fusion, low discrimination accuracy, poor real-time performance and high degree of manual dependence, as well as the difficulty of identifying hidden anomalies under complex working conditions, has become a major technical problem that the geological exploration field is currently facing and urgently needs to solve. Summary of the Invention

[0008] The purpose of this invention is to provide an automatic identification integrated geophysical exploration system to solve the technical defects of existing technologies, such as multiple solutions caused by the single geophysical exploration method, difficulty in fusion of multi-source data, high degree of manual intervention, poor real-time performance, and low accuracy in identifying hidden geological anomalies.

[0009] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows: The system comprises a multi-method integrated detection platform, a distributed data acquisition and real-time transmission system, an automated data processing system, an artificial intelligence automatic discrimination and interpretation system, and an intelligent decision support and 3D modeling system. The systems interact in a closed loop via a high-speed bus and industrial-grade wireless communication links, enabling control commands and data flow. The multi-method integrated detection platform integrates gravity exploration, magnetic exploration, electromagnetic exploration, seismic exploration, ground-penetrating radar, and integrated well logging modules to acquire raw observation data of the multi-dimensional physical field of the area to be explored. The distributed data acquisition and real-time transmission system includes a multi-channel synchronous acquisition unit, a wireless communication backbone network, and a real-time quality control unit. The multi-channel synchronous acquisition unit utilizes a clock synchronization mechanism based on the Precision Time Protocol (PTP) to synchronize data from the multi-method integrated detection platform. The system performs sub-microsecond synchronous triggering acquisition for each exploration module; the automated data processing system includes a multi-source data fusion center, an intelligent noise reduction module, a feature extraction and enhancement module, and a standardized preprocessing module, used to perform spatial registration, background noise suppression, and physical feature dimension unification on the acquired raw observation data; the artificial intelligence automatic discrimination and interpretation system consists of a deep learning inference engine, a geological knowledge base, a multi-model fusion judgment unit, and an uncertainty quantification module, which automatically discriminates the attributes of geological anomalies by extracting multi-physics joint feature vectors and combining them with geological prior constraints; the intelligent decision support and 3D modeling system includes a transparent geological support platform, an exploration scheme optimization module, and an automatic report generation unit, used to construct a 3D geological model covering the three-level scale of "mining area-mine shaft-working face", and dynamically update the 3D geological model based on the discrimination results and provide suggestions for intensified exploration.

[0010] Preferably, the gravity exploration module is equipped with a high-precision quartz spring gravimeter or a superconducting gravimeter with a resolution better than 0.001 milliger; the gravity exploration module adopts an orthogonal grid bidirectional observation structure, constructs a two-dimensional orthogonal measurement grid in the target detection area, and performs least-squares adjustment through redundant observation points in the longitudinal and lateral directions to eliminate system drift errors; the magnetic exploration module consists of an unmanned aerial vehicle-borne high-precision magnetometer, which adopts an optically pumped magnetometer or a superconducting quantum interference device with a sensitivity of 0.001 nanotesla.

[0011] Preferably, the electromagnetic exploration module adopts a multi-parameter cableless integrated combined probe structure. The outer shell of the multi-parameter cableless integrated combined probe is made of high-strength non-metallic composite material, and the interior integrates a transient electromagnetic detection unit, a ground-penetrating radar high-frequency antenna unit, a natural gamma detection sensor, and a high-precision trajectory measurement unit. The multi-parameter cableless integrated combined probe is connected to the drill pipe through standard threaded interfaces at both ends, and performs joint detection of transient electromagnetic, ground-penetrating radar, natural gamma, and trajectory parameters simultaneously during a single drilling process.

[0012] Preferably, the distributed data acquisition and real-time transmission system has an adaptive identification technology for the multi-parameter cableless integrated combined probe. Specifically, a high-sensitivity accelerometer and a three-axis gyroscope are installed inside the multi-parameter cableless integrated combined probe to monitor the probe's motion vector in real time. When the multi-channel synchronous acquisition unit senses that the probe is stationary for a period of time exceeding a set time limit through the accelerometer, it automatically triggers the transient electromagnetic detection unit to start a high-redundancy superposition acquisition mode. When the probe is sensed to be in continuous motion, the system automatically switches to a high sampling rate mode and starts continuous scanning acquisition by the ground-penetrating radar high-frequency antenna unit and the natural gamma detection sensor.

[0013] Preferably, the seismic exploration module is equipped with a fully digital piezoelectric source and a distributed seismic data acquisition and recording system. The fully digital piezoelectric source has a frequency scanning function and can perform linear or nonlinear frequency sweep excitation in the range of 10Hz to 500Hz, with a sampling frequency of not less than 48kHz. The automated data processing system also includes a dynamic data calibration module, which is configured to use the high-resolution stratigraphic information obtained by the seismic exploration module as a geometric constraint boundary to perform real-time calibration of the inversion depth of the gravity exploration module and the electromagnetic exploration module.

[0014] Preferably, the intelligent noise reduction processing module adopts an autoencoder neural network based on the exponential linear unit ELU, which contains at least five hidden layers, and the number of neurons in each layer decreases and then increases proportionally to form a funnel-shaped feature compression structure; the feature extraction and enhancement module extracts multi-scale spatial features of geological anomalies based on a deep convolutional neural network.

[0015] Preferably, the deep learning inference engine in the artificial intelligence automatic discrimination and interpretation system integrates convolutional neural network (CNN), recurrent neural network (RNN), and Transformer architectures; among them, for the inversion of borehole transient electromagnetic data, a long short-term memory neural network (LSTM) algorithm is used to achieve a single-point inversion time of less than 1 second; the geological knowledge base is stored in a graph database structure, describing the topological relationships and sedimentary evolution logic between geological entities, and the discrimination results output by the deep learning inference engine are verified by the stratigraphic sequence law and tectonic mechanics principles through the knowledge base logic checking unit, and anomalies that do not conform to geological logic are eliminated.

[0016] Preferably, the multi-model fusion judgment unit is based on an integrated method of game theory and machine learning optimization, which assigns weights to the judgment results of different deep learning models and solves the overfitting problem under imbalanced sample conditions by calculating the Nash equilibrium point between the output features of each model; the uncertainty quantification module uses Monte Carlo sampling or variational inference methods to evaluate the confidence of the automatic judgment conclusion and outputs a probability distribution map of the location, size and nature of the geological anomaly.

[0017] Preferably, during the construction of the three-dimensional geological model, a mesh generation technique under topological constraints and a non-uniform rational B-spline NURBS technique are used to fit the complex and curved stratigraphic interfaces.

[0018] Preferably, the wireless communication backbone network in the distributed data acquisition and real-time transmission system supports traffic scheduling based on software-defined networking (SDN), and dynamically allocates bandwidth according to the real-time data priority of each exploration module.

[0019] The present invention has the following significant technical effects: First, this invention achieves deep integration at the physical field level and intelligent fusion at the logical level. Through orthogonal network gravity observation, high-precision magnetic surveys by UAVs, and multi-parameter "one-trip" electromagnetic detection, the accuracy of the original observation data is significantly improved. By utilizing the consistency principle of multi-source data fusion and the principle of prioritizing high-precision data, the problem of multiple solutions caused by a single geophysical exploration method is fundamentally solved.

[0020] Secondly, this invention incorporates artificial intelligence technology in the data processing stage. The autoencoder neural network constructed using the ELU activation function significantly outperforms traditional filters and ReLU networks in suppressing random noise while preserving effective signal phase and amplitude information. Deep learning-based feature enhancement technology improves the system's sensitivity in identifying small, concealed geological bodies (such as micro-faults and deep cavities) by more than 30%.

[0021] Third, this invention improves the real-time performance and automation level of geophysical exploration operations. Through the collaborative work of a distributed acquisition architecture and edge computing nodes, a closed-loop process from data acquisition to inversion and interpretation is achieved. In particular, the application of the LSTM algorithm in borehole transient electromagnetic inversion reduces the traditional inversion time of several hours to seconds, making it possible to dynamically adjust exploration plans in real time in the field.

[0022] Fourth, this invention constructs a three-tiered transparent geological assurance system. Through highly expressive and high-resolution 3D modeling technology, it solves the problems of unclear representation and difficulty in updating complex fault morphologies in mining areas within the model. The system can automatically generate standardized exploration reports, reducing the tedious process of manual interpretation, significantly decreasing reliance on expert experience, and improving the objectivity and accuracy of geological outputs.

[0023] Fifth, this invention possesses strong environmental adaptability. The system employs a distributed architecture and adaptive recognition technology, enabling it to automatically switch working modes based on different geological conditions (such as surface, borehole, and mine tunnels). Through a game theory-optimized ensemble learning method, the system overcomes the model generalization challenge under small sample sizes and imbalanced geological data, ensuring robust performance across diverse geographical environments and exploration objectives.

[0024] Sixth, the engineering design of this invention ensures the reliability of the system under high-intensity operating environments. The discrete vibration damping design of the sensor platform, the high-strength non-metallic composite packaging of the probe, and the high-precision clock synchronization mechanism based on PTP jointly guarantee the physical authenticity and temporal consistency of data acquisition in harsh field environments.

[0025] In summary, this invention constructs a fully automated, high-precision comprehensive geophysical exploration technology platform through the collaborative integration of multiple methods, intelligent noise reduction, deep learning automatic discrimination, and real-time 3D modeling. This significantly improves the efficiency and accuracy of geological exploration and provides advanced technical means for geological exploration, engineering disaster early warning, and fine detection of mineral resources. Attached Figure Description

[0026] Figure 1 This is a block diagram of the integrated geophysical exploration system with automatic discrimination provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of the multi-parameter cableless integrated combined probe provided in an embodiment of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0028] like Figure 1 , 2 As shown, the automatic discrimination integrated geophysical exploration system of the present invention consists of a multi-method integrated detection platform, a distributed data acquisition and real-time transmission system, an automated data processing system, an artificial intelligence automatic discrimination and interpretation system, and an intelligent decision support and 3D modeling system. The systems are interconnected through a high-speed bus and an industrial-grade wireless communication link to achieve closed-loop interaction of control commands and data streams. The multi-method integrated detection platform integrates gravity exploration, magnetic exploration, electromagnetic exploration, seismic exploration, ground-penetrating radar, and integrated well logging modules to acquire multi-dimensional physical field raw observation data of the area to be explored. The distributed data acquisition and real-time transmission system includes a multi-channel synchronous acquisition unit, a wireless communication backbone network, and a real-time quality control unit. The multi-channel synchronous acquisition unit uses a clock synchronization mechanism based on the Precision Time Protocol (PTP) to perform sub-microsecond-level synchronous triggering acquisition of each exploration module in the multi-method integrated detection platform. The automated data processing system includes a multi-source data fusion center, an intelligent noise reduction module, a feature extraction and enhancement module, and a standardized preprocessing module, which are used to perform spatial registration, background noise suppression, and physical feature dimension unification on the collected raw observation data. The AI-powered automatic identification and interpretation system consists of a deep learning inference engine, a geological knowledge base, a multi-model fusion judgment unit, and an uncertainty quantification module. It automatically identifies the attributes of geological anomalies by extracting multi-physics joint feature vectors and combining them with geological prior constraints. The intelligent decision support and 3D modeling system includes a transparent geological assurance platform, an exploration scheme optimization module, and an automatic report generation unit. It is used to construct a 3D geological model covering three levels: "mining area-mine shaft-working face". Based on the discrimination results, the 3D geological model is dynamically updated and suggestions for intensified exploration are given.

[0029] The gravity exploration module is equipped with a high-precision quartz spring gravimeter or superconducting gravimeter with a resolution better than 0.001 milliger. The gravity exploration module adopts an orthogonal grid bidirectional observation structure, constructs a two-dimensional orthogonal measurement grid in the target detection area, and performs least squares adjustment through redundant observation points in the longitudinal and transverse directions to eliminate system drift error. The magnetic exploration module consists of a high-precision magnetometer carried by an unmanned aerial vehicle (UAV). The magnetometer uses an optically pumped magnetometer or a superconducting quantum interference device, and its sensitivity reaches 0.001 nanotesla.

[0030] The electromagnetic exploration module adopts a multi-parameter cableless integrated combined probe structure. The outer shell of the multi-parameter cableless integrated combined probe is made of high-strength non-metallic composite material, and the interior integrates a transient electromagnetic detection unit, a ground-penetrating radar high-frequency antenna unit, a natural gamma detection sensor, and a high-precision trajectory measurement unit. The multi-parameter cableless integrated probe connects to the drill pipe through standard threaded interfaces at both ends, and simultaneously performs combined detection of transient electromagnetic, ground-penetrating radar, natural gamma, and trajectory parameters during a single drilling process.

[0031] The distributed data acquisition and real-time transmission system has adaptive identification technology for multi-parameter cableless integrated combined probes. Specifically, a high-sensitivity accelerometer and a three-axis gyroscope are installed inside the multi-parameter cableless integrated combined probe to monitor the motion vector of the probe in real time. When the multi-channel synchronous acquisition unit senses that the probe is stationary for a period of time exceeding the set time limit through the accelerometer, it automatically triggers the transient electromagnetic detection unit to start the high-redundancy superposition acquisition mode; when the probe is sensed to be in continuous motion, the system automatically switches to the high sampling rate mode and starts the continuous scanning acquisition of the ground-penetrating radar high-frequency antenna unit and the natural gamma detection sensor.

[0032] The seismic exploration module is equipped with a fully digital piezoelectric source and a distributed seismic data acquisition and recording system. The fully digital piezoelectric source has a frequency scanning function and can perform linear or nonlinear frequency sweep excitation in the range of 10Hz to 500Hz, with a sampling frequency of not less than 48kHz. The automated data processing system also includes a dynamic data calibration module, which is configured to use the high-resolution stratigraphic information obtained by the seismic exploration module as a geometric constraint boundary to calibrate the inversion depth of the gravity exploration module and the electromagnetic exploration module in real time.

[0033] The intelligent noise reduction processing module adopts an autoencoder neural network based on the exponential linear unit ELU. This autoencoder neural network contains at least five hidden layers, and the number of neurons in each layer decreases and then increases proportionally to form a funnel-shaped feature compression structure. The feature extraction and enhancement module extracts multi-scale spatial features of geological anomalies based on deep convolutional neural networks.

[0034] The deep learning inference engine in the AI ​​automatic discrimination and interpretation system integrates convolutional neural network (CNN), recurrent neural network (RNN), and Transformer architectures. Among them, for the inversion of transient electromagnetic data from boreholes, the Long Short Time Memory (LSTM) neural network algorithm is used to achieve a single-point inversion time of less than 1 second. The geological knowledge base uses a graph database structure for storage, describing the topological relationships and sedimentary evolution logic between geological entities. The knowledge base logic checking unit verifies the judgment results output by the deep learning inference engine based on stratigraphic sequence law and tectonic mechanics principles, and removes anomalies that do not conform to geological logic.

[0035] The multi-model fusion judgment unit is based on an integrated method of game theory and machine learning optimization. It assigns weights to the judgment results of different deep learning models and solves the overfitting problem under imbalanced sample conditions by calculating the Nash equilibrium point between the output features of each model. The uncertainty quantification module uses Monte Carlo sampling or variational inference methods to assess the confidence level of the automatically determined conclusions and outputs probability distribution maps of the location, size, and nature of geological anomalies.

[0036] In the process of constructing the three-dimensional geological model, the mesh generation technique under topological constraints and the non-uniform rational B-spline NURBS technique are used to fit the complex and curved stratigraphic interfaces.

[0037] The wireless communication backbone in the distributed data acquisition and real-time transmission system supports traffic scheduling based on software-defined networking (SDN), and dynamically allocates bandwidth according to the real-time data priority of each exploration module.

[0038] To further demonstrate the technical superiority of the present invention, a specific embodiment based on a deep polymetallic mining area is given below, and a comparative example is provided with traditional geophysical exploration methods.

[0039] In this embodiment, the detection target is a hidden fault zone and its associated water-bearing cavities at a depth of approximately 1200 meters.

[0040] The configuration parameters for this embodiment of the invention are as follows: Gravity exploration: A quartz spring gravimeter with an accuracy of 0.001 millidigrams was used, with a 20m×20m orthogonal grid.

[0041] Magnetic exploration: The UAV is equipped with an optically pumped magnetometer, flies at an altitude of 50m, and has a flight path spacing of 10m.

[0042] Electromagnetic exploration: One drilling pipe, transient electromagnetic frequency 25Hz, ground-penetrating radar antenna center frequency 100MHz.

[0043] Seismic exploration: Fully digital piezoelectric source, sweep frequency range 20-250Hz, sampling rate 48kHz.

[0044] Data processing: Noise reduction using a five-layer ELU autoencoder and LSTM inversion.

[0045] The configuration parameters for the comparative scale (traditional method) are as follows: Gravity exploration: traditional unidirectional survey line, point spacing 50m, without automatic leveling.

[0046] Magnetic exploration: Manual measurement using a ground magnetometer, with a point spacing of 20m.

[0047] Electromagnetic exploration: Traditional transient electromagnetic method on the surface, without drilling linkage.

[0048] Seismic exploration: conventional explosive source, simulated detector.

[0049] Data processing: Traditional manual filtering and manual inversion.

[0050] The performance comparison data of the two are shown in Table 1 after actual exploration and subsequent drilling verification.

[0051] Table 1 Comparison of performance data between embodiments of the present invention and comparative examples.

[0052] As can be clearly observed from the data in Table 1, this invention, through multi-method integration and artificial intelligence discrimination, demonstrates overwhelming advantages in signal-to-noise ratio improvement, boundary positioning accuracy, and operational efficiency. Particularly in terms of inversion computation time, the introduction of a high-performance deep learning inference engine elevates traditional manual inversion to a real-time automated inversion level.

[0053] As a crucial consideration in the implementation of this invention, all sensor modules have undergone rigorous anti-magnetic, moisture-proof, and high-pressure resistant encapsulation treatment. For example, the multi-parameter cableless integrated probe is equipped with a highly reliable pressure-bearing sealing joint at the connection point, with a pressure resistance tested to be no less than 100MPa, fully adaptable to the extreme environment of ultra-deep wells. Simultaneously, to prevent electromagnetic compatibility issues between modules, the high-power transient electromagnetic emission unit employs a multi-layer conductive polymer shielding layer, while the high-sensitivity gravimeter base is equipped with an active magnetic field compensation coil, eliminating interference from external environmental electromagnetic field fluctuations on the high-precision sensors.

[0054] In the operational logic of the data dynamic calibration module, the system utilizes the high-resolution geometric information of the stratigraphic interface obtained by the seismic exploration module as a "hard boundary" to constrain the inversion process of gravity and electromagnetic methods in real time. This cross-physics spatial response function compensation can effectively eliminate depth calculation distortion caused by uneven density or abrupt changes in electrical properties of the surrounding rock, ensuring the physical authenticity of the inversion results.

[0055] Furthermore, the 3D modeling system of this invention also features dynamic slice display functionality. For any cross-section defined by the user in the 3D visualization interface, the system can calculate and overlay slices of gravity anomalies, magnetic scalar fields, resistivity contour lines, and seismic wave fields in real time. This multi-dimensional visual overlay greatly assists geologists in their intuitive understanding of complex structures.

[0056] The exploration risk assessment submodule in the intelligent decision support system can automatically calculate the evaluation index of potential water inrush risk during tunnel excavation by combining automatically judged geological conclusions (such as the distribution of water-bearing faults) and engineering rock mechanics parameters. The risk level is marked in real time in the form of red warning labels in the three-dimensional model, realizing the direct conversion of geophysical exploration results into safe production instructions.

[0057] In summary, the automated and intelligent integrated geophysical exploration system constructed in this invention completely solves the core pain points of traditional geophysical exploration work, such as strong ambiguity, poor real-time performance, and high dependence on experts, through high-level physical integration of hardware, high-precision synchronization of communication, intelligent noise reduction in processing, and deep learning discrimination in interpretation. This system has been field-verified in multiple large-scale mining and geological exploration projects, and its stability and accuracy have reached internationally leading engineering application levels, providing solid technical support for precise perception of underground space.

[0058] For those skilled in the art, the technical solution provided by this invention can be fine-tuned according to the specific exploration target depth and geological environment during implementation, with adjustments made to the sensor parameters of each module. However, its core multi-physics coupling discrimination logic and closed-loop automated process still fall within the protection scope of this invention. The collaborative optimization among the components of this invention not only improves the detection depth of a single physical quantity but also achieves a "see-through" understanding of complex underground geological entities through logical-level information complementarity.

[0059] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this invention, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention.

[0060] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features, and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An automatic identification integrated geophysical exploration system, characterized in that, The system consists of a multi-method integrated detection platform, a distributed data acquisition and real-time transmission system, an automated data processing system, an artificial intelligence automatic discrimination and interpretation system, and an intelligent decision support and 3D modeling system. The systems interact in a closed loop through a high-speed bus and an industrial-grade wireless communication link to achieve control commands and data flow. The multi-method integrated detection platform integrates gravity exploration module, magnetic exploration module, electromagnetic exploration module, seismic exploration module, ground-penetrating radar module and comprehensive well logging module, which are used to acquire multi-dimensional physical field raw observation data of the area to be explored; The distributed data acquisition and real-time transmission system includes a multi-channel synchronous acquisition unit, a wireless communication backbone network, and a real-time quality control unit. The multi-channel synchronous acquisition unit uses a clock synchronization mechanism based on the Precision Time Protocol (PTP) to perform sub-microsecond-level synchronous triggering acquisition of each exploration module in the multi-method integrated detection platform. The automated data processing system includes a multi-source data fusion center, an intelligent noise reduction module, a feature extraction and enhancement module, and a standardized preprocessing module, which are used to perform spatial registration, background noise suppression, and physical feature dimension unification on the collected raw observation data. The AI ​​automatic identification and interpretation system consists of a deep learning inference engine, a geological knowledge base, a multi-model fusion judgment unit, and an uncertainty quantification module. It automatically identifies the attributes of geological anomalies by extracting multi-physics joint feature vectors and combining them with geological prior constraints. The intelligent decision support and 3D modeling system includes a transparent geological support platform, an exploration scheme optimization module, and an automatic report generation unit. It is used to construct a 3D geological model covering three levels: "mining area-mine shaft-working face", and dynamically update the 3D geological model based on the discrimination results and provide suggestions for intensified exploration.

2. The integrated geophysical exploration system with automatic discrimination according to claim 1, characterized in that, The gravity exploration module is equipped with a high-precision quartz spring gravimeter or superconducting gravimeter with a resolution better than 0.001 millidimeters. The gravity exploration module adopts an orthogonal grid bidirectional observation structure, constructs a two-dimensional orthogonal measurement grid in the target detection area, and performs least squares adjustment through redundant observation points in the longitudinal and transverse directions to eliminate system drift error. The magnetic exploration module consists of an unmanned aerial vehicle-borne high-precision magnetometer, which uses an optically pumped magnetometer or a superconducting quantum interference device, with a sensitivity of 0.001 nanotesla.

3. The integrated geophysical exploration system with automatic discrimination according to claim 1, characterized in that, The electromagnetic exploration module adopts a multi-parameter cableless integrated combined probe structure. The outer shell of the multi-parameter cableless integrated combined probe is made of high-strength non-metallic composite material, and the internal components include a transient electromagnetic detection unit, a ground-penetrating radar high-frequency antenna unit, a natural gamma detection sensor, and a high-precision trajectory measurement unit. The multi-parameter cableless integrated probe is connected to the drill pipe through standard threaded interfaces at both ends, and performs simultaneous detection of transient electromagnetic, ground-penetrating radar, natural gamma and trajectory parameters during a single drilling process.

4. The integrated geophysical exploration system with automatic discrimination according to claim 3, characterized in that, The distributed data acquisition and real-time transmission system has an adaptive identification technology for the multi-parameter cableless integrated combined probe. Specifically, a high-sensitivity accelerometer and a three-axis gyroscope are installed inside the multi-parameter cableless integrated combined probe to monitor the motion vector of the probe in real time. When the multi-channel synchronous acquisition unit senses that the probe is stationary for a period of time exceeding a set time limit via the accelerometer, it automatically triggers the transient electromagnetic detection unit to start the high-redundancy superposition acquisition mode; when the probe is sensed to be in continuous motion, the system automatically switches to the high sampling rate mode and starts the continuous scanning acquisition of the ground-penetrating radar high-frequency antenna unit and the natural gamma detection sensor.

5. The integrated geophysical exploration system with automatic discrimination according to claim 1, characterized in that, The seismic exploration module is equipped with a fully digital piezoelectric source and a distributed seismic data acquisition and recording device. The fully digital piezoelectric source has a frequency scanning function and can perform linear or nonlinear frequency sweep excitation in the range of 10Hz to 500Hz, with a sampling frequency of not less than 48kHz. The automated data processing system also includes a dynamic data calibration module, which is configured to use the high-resolution stratigraphic information obtained by the seismic exploration module as a geometric constraint boundary to perform real-time calibration of the inversion depth of the gravity exploration module and the electromagnetic exploration module.

6. The integrated geophysical exploration system with automatic discrimination according to claim 1, characterized in that, The intelligent noise reduction processing module adopts an autoencoder neural network based on the exponential linear unit ELU. The autoencoder neural network contains at least five hidden layers, and the number of neurons in each layer decreases and then increases proportionally to form a funnel-shaped feature compression structure. The feature extraction and enhancement module extracts multi-scale spatial features of geological anomalies based on deep convolutional neural networks.

7. The integrated geophysical exploration system with automatic discrimination according to claim 1, characterized in that, The deep learning inference engine in the AI ​​automatic discrimination and interpretation system integrates convolutional neural network (CNN), recurrent neural network (RNN), and Transformer architecture; among them, for the inversion of borehole transient electromagnetic data, the long short-term memory neural network (LSTM) algorithm is used to achieve single-point inversion in less than 1 second. The geological knowledge base is stored in a graph database structure, describing the topological relationships and sedimentary evolution logic between geological entities. The knowledge base logic checking unit verifies the judgment results output by the deep learning inference engine using stratigraphic sequence law and tectonic mechanics principles, and removes anomalies that do not conform to geological logic.

8. The integrated geophysical exploration system with automatic discrimination according to claim 7, characterized in that, The multi-model fusion judgment unit is based on an integrated method of game theory and machine learning optimization. It assigns weights to the judgment results of different deep learning models and solves the overfitting problem under imbalanced sample conditions by calculating the Nash equilibrium point between the output features of each model. The uncertainty quantification module uses Monte Carlo sampling or variational reasoning methods to evaluate the confidence level of the automatically determined conclusions and outputs probability distribution maps of the location, size, and nature of geological anomalies.

9. The integrated geophysical exploration system with automatic discrimination according to claim 1, characterized in that, In the process of constructing the three-dimensional geological model, the mesh generation technique under topological constraints and the non-uniform rational B-spline NURBS technique are used to fit the complex and curved stratigraphic interfaces.

10. The integrated geophysical exploration system with automatic discrimination according to claim 1, characterized in that, The wireless communication backbone in the distributed data acquisition and real-time transmission system supports traffic scheduling based on software-defined networking (SDN), and dynamically allocates bandwidth according to the real-time data priority of each exploration module.

Images