A method and device for exploring medium-deep underground hot water resources based on multi-source data fusion

Abstract

The application provides a kind of based on multi-source data fusion's middle-deep layer underground hot water resource exploration method and device, belong to underground hot water resource exploration technical field, wherein, method includes following steps: step S1: the multi-source data of target area is collected, including geological data, geophysical data, geochemical data and remote sensing data;Step S2: the multi-source data is preprocessed, including wavelet packet transform frequency decomposition, principal component analysis noise reduction and normalization processing.The application effectively handles data noise, conflict and uncertainty problem through the systematic integration and dynamic optimization of multi-source data, improves the accuracy and reliability of middle-deep layer underground hot water resource exploration;The three-dimensional coupled model constructed and the improved heat conduction equation solving method accurately depict the thermal storage characteristics under the complex geology-thermodynamic environment, combined with the automatic hardware device to realize the whole process efficient operation, which provides an integrated solution for the scientific development of underground hot water resources.

Description

Technical Field

[0001] This invention relates to the field of underground hot water resource exploration technology, specifically to a method and apparatus for exploring medium-deep underground hot water resources based on multi-source data fusion. Background Technology

[0002] As an important clean energy source, the exploration of medium-deep geothermal resources requires comprehensive consideration of multi-dimensional data, including geological, geophysical, geochemical, and remote sensing data. Traditional exploration methods suffer from problems such as low data fusion efficiency, rigid weight allocation, insufficient handling of conflicting evidence, and poor model adaptability, making it difficult to accurately characterize the distribution of geothermal reservoirs under complex geological conditions. This invention addresses these issues by proposing an exploration method and apparatus based on multi-source data fusion. Through advanced data processing, fusion modeling, and dynamic optimization techniques, it improves the accuracy and reliability of resource exploration.

[0003] To address this, a method and apparatus for exploring medium-deep geothermal resources based on multi-source data fusion are proposed. Summary of the Invention

[0004] The present invention aims to solve the problems mentioned in the background art by providing a method and apparatus for exploring medium-deep underground hot water resources based on multi-source data fusion.

[0005] The specific technical solution is as follows:

[0006] A method for exploring medium-deep geothermal resources based on multi-source data fusion includes the following steps:

[0007] Step S1: Collect multi-source data of the target area, including geological data, geophysical data, geochemical data and remote sensing data;

[0008] Step S2: Preprocess the multi-source data, including wavelet packet transform frequency division decomposition, principal component analysis for noise reduction, and normalization.

[0009] Step S3: Construct a Bayesian network model to represent the probabilistic dependencies between data sources, and combine fuzzy inference and information entropy weight allocation to achieve multi-level data fusion;

[0010] Step S4: Generate a comprehensive anomaly map with confidence rating and delineate potential resource areas, and dynamically optimize model parameters by combining drilling verification and machine learning.

[0011] The above-mentioned method for exploring medium-deep geothermal resources based on multi-source data fusion, wherein in step S1:

[0012] The geological data includes stratigraphic lithology, fault structures, and hydrogeological parameters;

[0013] The geophysical data includes resistivity data, gravity data, and seismic shear wave velocity structure data.

[0014] The geochemical data includes the chemical composition and isotopic composition of groundwater.

[0015] The remote sensing data includes thermal infrared radiation images and multispectral vegetation index data.

[0016] The above-mentioned method for exploring medium-deep geothermal resources based on multi-source data fusion, wherein the preprocessing in step S2 includes:

[0017] Step S2a: Use dual-tree complex wavelet transform to decompose the original data into frequency bands, and filter out high-frequency noise and low-frequency drift;

[0018] Step S2b: Perform principal component analysis on the data of each frequency band and extract principal component reconstruction data with a contribution rate ≥ 85%;

[0019] Step S2c: Normalize the multi-source data based on the improved Z-score method to eliminate dimensional differences.

[0020] The above-mentioned method for exploring medium-deep geothermal resources based on multi-source data fusion includes step S3:

[0021] The fuzzy inference uses a Gaussian membership function to quantize feature mapping;

[0022] The information entropy weights are dynamically adjusted by calculating the uncertainty of each data source, and the weight allocation formula is as follows:

[0023] in,

[0024] in:

[0025] H i : Information entropy of the i-th data source, representing data uncertainty;

[0026] λ: Fusion sensitivity coefficient (value ranges from 0.5 to 2.0), which controls the degree of concentration in weight allocation;

[0027] σ 2 The noise variance of the data source is calculated through the preprocessing module;

[0028] ∈: Smoothing constant (default 1e-5) to prevent the denominator from being zero.

[0029] The above-mentioned method for exploring medium-deep geothermal resources based on multi-source data fusion, wherein the multi-level data fusion in step S3 includes:

[0030] Step S3a: Infer the probabilistic relationships between data sources based on Bayesian networks;

[0031] Step S3b: Use the Dempster-Shafer evidence theory to resolve the problem of merging conflicting evidence;

[0032] Step S3c: Dynamically adjust the fusion weight coefficients using the particle swarm optimization algorithm;

[0033] The conflicting evidence fusion employs a modified Dempster combination rule:

[0034]

[0035] in:

[0036] ψ(B,C)=exp(-β·|conf(B,C)|): Conflict attenuation factor, where β is the attenuation coefficient;

[0037] conf(B,C): The degree of conflict between evidence B and evidence C;

[0038] K: The number of conflicts in the traditional Dempster rule;

[0039] δ: Conflict compensation term (δ=0.1K).

[0040] The above-mentioned method for exploring medium-deep geothermal resources based on multi-source data fusion, wherein the dynamic optimization model parameters in step S4 include:

[0041] Step S4a: Obtain underground rock core and water sample data through drilling;

[0042] Step S4b: Use the support vector machine algorithm to compare the fusion result with the actual data and calculate the error rate;

[0043] Step S4c: If the error rate exceeds the threshold, adjust the Bayesian network parameters and fuzzy membership function in reverse.

[0044] The above-mentioned method for exploring medium-deep geothermal resources based on multi-source data fusion, wherein step S4 further includes:

[0045] A three-dimensional geological-thermodynamic coupled model is constructed based on the fusion results. The model includes:

[0046] Three-dimensional mesh model of stratigraphy and lithology;

[0047] Three-dimensional distribution model of fracture structure;

[0048] Dynamic distribution field of rock thermal conductivity.

[0049] The aforementioned method for exploring medium-deep geothermal resources based on multi-source data fusion, wherein the three-dimensional geological-thermodynamic coupling model solves the non-Darcy flow heat conduction equation using the finite volume method, and combines Monte Carlo simulation to assess the uncertainty of resource quantity prediction, the solution of the non-Darcy flow heat conduction equation adopts the following modified form:

[0050]

[0051] Parameter description:

[0052] Nonlinear flow coefficient, where φ is porosity, μ is fluid viscosity, and k is permeability;

[0053] γ: Temperature decay coefficient;

[0054] P: Fluid pressure field.

[0055] This invention also provides a device for exploring medium-deep geothermal resources based on multi-source data fusion, used to execute the aforementioned method for exploring medium-deep geothermal resources based on multi-source data fusion, comprising:

[0056] The data acquisition module integrates a high-precision gravimeter, a distributed fiber optic temperature measurement system, and an UAV-borne multispectral thermal infrared imager.

[0057] The data preprocessing module is configured with an FPGA-accelerated wavelet packet transform unit and a GPU-parallel principal component analysis unit.

[0058] The data fusion module includes a Bayesian network modeling unit, an evidence theory reasoning engine, and a dynamic weight optimization unit;

[0059] The verification and correction module achieves closed-loop optimization through the drilling equipment control unit and machine learning algorithms.

[0060] The aforementioned multi-source data fusion-based medium-deep geothermal resource exploration device further includes, in its data acquisition module:

[0061] A downhole three-component microseismic monitoring array is used to acquire seismic wave data at depths ≥3km;

[0062] An automated online fluid chemical composition analyzer can detect the chemical components and isotope content of groundwater in real time.

[0063] The present invention has the following beneficial effects:

[0064] This invention overcomes the limitations of single-data exploration by systematically integrating and dynamically optimizing multi-source data, effectively handling data noise, conflicts, and uncertainties, and improving the accuracy and reliability of exploration for medium-deep geothermal resources. The constructed three-dimensional coupled model and the improved heat conduction equation solution method accurately characterize the geothermal reservoir characteristics under complex geological-thermodynamic environments. Combined with automated hardware devices, it achieves efficient operation throughout the entire process, providing an integrated solution for the scientific development of geothermal resources. Attached Figure Description

[0065] Figure 1 A flowchart illustrating the method for exploring medium-deep geothermal resources based on multi-source data fusion provided in this embodiment of the invention;

[0066] Figure 2 This is an architectural diagram of a device for exploring medium-deep underground hot water resources based on multi-source data fusion, provided in an embodiment of the present invention.

[0067] Figure 3 Information entropy comparison curve in the method for exploring medium-deep underground hot water resources based on multi-source data fusion provided in this embodiment of the invention;

[0068] Figure 4 A comparison curve of noise levels in the method for exploring medium-deep underground hot water resources based on multi-source data fusion provided in this embodiment of the invention. Detailed Implementation

[0069] The technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0070] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual images. They should not be construed as limiting the scope of this patent. To better illustrate the embodiments of the present invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual dimensions of the product. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0071] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "inner," and "outer" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, 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. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present patent. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0072] In the description of this invention, unless otherwise explicitly specified and limited, the term "connection" or similar designation indicating a connection between components should be interpreted broadly. For example, it can refer to a fixed connection, a detachable connection, or an integral part; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can refer to the internal communication between two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0073] The method for exploring medium-deep geothermal resources based on multi-source data fusion provided in this embodiment, such as... Figures 1-4 As shown, where: Figure 3 The information entropy comparison curve in the multi-source data fusion-based medium-deep geothermal resource exploration method provided in this embodiment shows the changes in information entropy of different data sources before and after preprocessing, helping to understand the impact of the preprocessing step on the amount of data information. Figure 4 This embodiment presents a noise level comparison curve in the multi-source data fusion-based method for exploring medium-deep geothermal resources. The curve illustrates the changes in noise variance before and after preprocessing for different data sources, helping to evaluate the noise suppression effect of the preprocessing step. The method includes the following steps:

[0074] Step S1: Collect multi-source data of the target area, including geological data, geophysical data, geochemical data and remote sensing data;

[0075] Step S2: Preprocess the multi-source data, including wavelet packet transform frequency division decomposition, principal component analysis for noise reduction, and normalization.

[0076] Step S3: Construct a Bayesian network model to represent the probabilistic dependencies between data sources, and combine fuzzy inference and information entropy weight allocation (based on dynamic weight adjustment formula) to achieve multi-level data fusion;

[0077] Step S4: Generate a comprehensive anomaly map with confidence rating and delineate potential resource areas, and dynamically optimize model parameters by combining drilling verification and machine learning.

[0078] This method for exploring medium-deep geothermal resources based on multi-source data fusion can integrate the advantages of different types of data through the fusion processing of multi-source data, comprehensively characterize the potential features of geothermal resources in the target area, improve the systematicness of the exploration process and the accuracy of resource identification; dynamic optimization of model parameters can make the method adaptable to different exploration scenarios, enhancing the adaptability and reliability of the overall process.

[0079] In step S1:

[0080] Geological data includes stratigraphy, fault structures, and hydrogeological parameters;

[0081] Geophysical data includes resistivity data (accuracy ≤ 0.1 Ω·m), gravity data (resolution ≤ 0.1 mGal), and seismic shear wave velocity structure data;

[0082] Geochemical data include the chemical composition and isotopic composition of groundwater;

[0083] The remote sensing data includes thermal infrared radiation images (spatial resolution ≤1m) and multispectral vegetation index data.

[0084] This plan clarifies the specific content and accuracy requirements of various types of data, ensuring that the collected data covers multiple dimensions of information such as geology, physical properties, chemical properties, and remote sensing, providing comprehensive and high-precision basic data support for subsequent fusion analysis, and avoiding exploration deviations caused by missing data or insufficient accuracy.

[0085] The preprocessing in step S2 includes:

[0086] Step S2a: Use dual-tree complex wavelet transform to decompose the original data into frequency bands, and filter out high-frequency noise and low-frequency drift;

[0087] Step S2b: Perform principal component analysis on the data of each frequency band and extract principal component reconstruction data with a contribution rate ≥ 85%;

[0088] Step S2c: Normalize the multi-source data based on the improved Z-score method to eliminate dimensional differences.

[0089] This scheme employs wavelet packet transform for frequency division and noise reduction, principal component analysis for dimensionality reduction, and normalization in the preprocessing stage. This effectively removes noise interference from the data, reduces data redundancy, and unifies the dimensions, thereby improving the quality and compatibility of multi-source data and laying a high-quality data foundation for subsequent fusion modeling.

[0090] In step S3:

[0091] Fuzzy inference uses a Gaussian membership function to quantize feature mapping;

[0092] The information entropy weights are dynamically adjusted based on the uncertainty of each data source. The weight allocation formula is as follows:

[0093] in,

[0094] in:

[0095] H i : Information entropy of the i-th data source, representing data uncertainty;

[0096] λ: Fusion sensitivity coefficient (value ranges from 0.5 to 2.0), which controls the degree of concentration in weight allocation;

[0097] σ 2 The noise variance of the data source is calculated through the preprocessing module;

[0098] ∈: Smoothing constant (default 1e-5) to prevent the denominator from being zero.

[0099] This scheme dynamically adjusts weights based on information entropy and noise level, and can adaptively allocate fusion weights according to the uncertainty of the data, so that low-noise, high-determinism data dominates the fusion process, avoiding the limitations of traditional fixed-weight methods and improving the fusion results' resistance to data fluctuations.

[0100] Example:

[0101] During the data fusion phase, the weights of each data source are dynamically adjusted based on their information entropy and noise levels.

[0102] For example:

[0103] Geophysical data has low noise variance (σ) 2 =0.02), Information entropy H i =0.3, then the weight is relatively high;

[0104] High noise variance in remote sensing data σ 2 =0.15), Information entropy H i =0.8, then the weight is low.

[0105] Technical effects:

[0106] By combining the exponential function with noise variance, higher weights are given to low-noise, high-deterministic data.

[0107] The weight allocation is adaptively adjusted, avoiding the rigidity problem of traditional weighted averages and improving the robustness of the fusion results.

[0108] Working principle and process:

[0109] 1. The preprocessing module calculates σ for each data source. 2 ;

[0110] 2. The information entropy module calculates H. i ;

[0111] 3. Dynamically generate weight w based on the formula. i Input fusion model.

[0112] The multi-level data fusion in step S3 includes:

[0113] Step S3a: Infer the probabilistic relationships between data sources based on Bayesian networks;

[0114] Step S3b: Use the Dempster-Shafer evidence theory to resolve the problem of merging conflicting evidence;

[0115] Step S3c: Dynamically adjust the fusion weight coefficients using the particle swarm optimization algorithm;

[0116] The fusion of conflicting evidence employs a modified Dempster combination rule:

[0117]

[0118] in:

[0119] ψ(B,C)=exp(-β·|conf(B,C)|): Conflict attenuation factor, β is the attenuation coefficient (value 1~3);

[0120] conf(B,C): The degree of conflict between evidence B and C, calculated by the fuzzy inference engine;

[0121] K: The number of conflicts in the traditional Dempster rule;

[0122] δ: Conflict compensation term (δ=0.1K), to prevent the denominator from being too small.

[0123] This scheme combines multi-level data fusion with Bayesian network probabilistic reasoning, improved evidence theory conflict handling, and particle swarm optimization dynamic weighting. It can effectively handle conflicts and uncertainties among multi-source data, alleviate the sensitivity of traditional fusion methods to highly conflicting evidence, and improve the stability and credibility of fusion results in complex data scenarios.

[0124] Example:

[0125] When resistivity data (evidence B) conflicts with seismic data (evidence C) (conf(B,C)=0.6), the weight of the conflicting evidence is reduced by an attenuation factor ψ. For example:

[0126] When β = 2, ψ = exp(-2 × 0.6) ≈ 0.30, which significantly suppresses the influence of conflicting evidence.

[0127] Technical effects:

[0128] An exponential conflict decay factor is introduced to alleviate the sensitivity of the traditional Dempster rule to highly conflicting evidence.

[0129] The conflict compensation term δ avoids numerical instability under extreme conflict scenarios and improves fusion reliability.

[0130] Working principle and process:

[0131] The fuzzy inference engine calculates the degree of conflict between pieces of evidence, conf(B,C).

[0132] Adjust the weights of the evidence combination based on ψ(B,C);

[0133] Output the corrected fusion result m(A).

[0134] The dynamic optimization of model parameters in step S4 includes:

[0135] Step S4a: Obtain underground rock core and water sample data through drilling;

[0136] Step S4b: Compare the fusion results with the actual data using the Support Vector Machine (SVM) algorithm, and calculate the error rate;

[0137] Step S4c: If the error rate exceeds the threshold, adjust the Bayesian network parameters and fuzzy membership function in reverse.

[0138] This approach utilizes drilling data and machine learning algorithms to dynamically optimize model parameters, and uses actual exploration data to correct the theoretical model, forming a closed loop of "exploration-verification-optimization" to ensure that the fused model matches the actual geological conditions and continuously improve the accuracy of resource exploration.

[0139] Step S4 further includes:

[0140] A three-dimensional geological-thermodynamic coupled model was constructed based on the fusion results. The model includes:

[0141] Three-dimensional mesh model of stratigraphy and lithology (resolution ≤ 50m);

[0142] Three-dimensional distribution model of fracture structure;

[0143] Dynamic distribution field of rock thermal conductivity.

[0144] This scheme constructs a three-dimensional geological-thermodynamic coupled model to model the spatial distribution of strata lithology, fracture structure and thermal conductivity, and intuitively presents the occurrence environment of underground hot water resources, providing a three-dimensional spatial dimension quantitative analysis basis for resource potential assessment.

[0145] The three-dimensional geological-thermodynamic coupling model also includes a fracture network spatial distribution model, a geothermal heat flow spatial distribution model, a thermodynamic model, and a fluid flow model, enabling the three-dimensional geological-thermodynamic coupling model to achieve the following technical improvements on the original basis:

[0146] 1. Slit network modeling:

[0147] Precisely characterizing the spatial distribution of underground fissures reveals key channels for hot water migration and the distribution of reservoir spaces, providing a structural basis for the analysis of resource endowment patterns.

[0148] 2. Analysis of Earth's heat flow distribution:

[0149] Quantifying regional geothermal anomaly characteristics and combining them with dynamic thermal conductivity fields can locate heat flow enrichment areas and enhance the spatial orientation of resource potential assessment.

[0150] 3. Simulation of thermodynamic behavior:

[0151] By coupling the temperature field with rock thermal properties, the heat transfer process and thermal equilibrium state of the reservoir are simulated to assess the long-term stability and sustainable development potential of the reservoir.

[0152] 4. Fluid flow dynamics prediction:

[0153] Based on the non-Darcy flow equation and fracture network model, the pressure distribution, flow path and dynamic response of underground hot water after extraction are simulated to optimize the development scheme design.

[0154] Overall technical effect

[0155] The newly added sub-model, through multiphysics coupling, achieves a comprehensive analysis of the geothermal resource occurrence environment, including geological structure, thermodynamic properties, and fluid dynamic behavior. Its technical effects are manifested in:

[0156] Refined modeling: expanding from single geological structures to multi-field coupling of heat, fluid and force, improving the ability to characterize complex reservoirs;

[0157] Dynamic prediction capability: Combining fluid flow models and thermodynamic simulations, it supports visualized prediction and risk assessment of the resource extraction process;

[0158] Decision support upgrade: By integrating geothermal flow and fracture network data, scientific basis is provided for target area selection, well location deployment and development strategy formulation.

[0159] In summary, the supplementary scheme, through the combination of multi-dimensional modeling and dynamic simulation, significantly enhances the comprehensive analysis and prediction capabilities of the three-dimensional coupled model, providing more comprehensive technical support for the accurate exploration and efficient development of underground hot water resources.

[0160] The three-dimensional geological-thermodynamic coupled model solves the non-Darcy flow heat conduction equation using the finite volume method. Combined with Monte Carlo simulation to assess the uncertainty in resource quantity prediction, the solution to the non-Darcy flow heat conduction equation adopts the following modified form:

[0161]

[0162] Parameter description:

[0163] Nonlinear flow coefficient, where φ is porosity, μ is fluid viscosity, and k is permeability;

[0164] γ: Temperature decay coefficient (value range 0.01 to 0.1), characterizing the effect of high temperature on flow resistance;

[0165] P: Fluid pressure field, obtained through fluid dynamics simulation.

[0166] This scheme improves the solution method for non-Darcy flow heat conduction equations by introducing a temperature-dependent energy dissipation term, accurately characterizing the coupling relationship between underground fluid flow and heat conduction under high-temperature conditions, and combining Monte Carlo simulation to assess prediction uncertainties, thereby enhancing the scientificity and accuracy of resource quantity calculations under complex thermal reservoir conditions.

[0167] Example:

[0168] In high-temperature reservoirs (T>150℃), e -γT This significantly reduces flow energy loss. For example:

[0169] When γ = 0.05 and T = 200℃, e -0.05×200 ≈0.0067, reflecting the inhibitory effect of high temperature on flow.

[0170] Technical effects:

[0171] By introducing a temperature-dependent exponential term, the energy dissipation of non-Darcy flow under high-temperature conditions can be characterized more accurately.

[0172] By combining a porosity-permeability dynamic correlation model, the accuracy of resource quantity inversion can be improved.

[0173] Working principle and process:

[0174] 1. The fluid dynamics module calculates the pressure field P;

[0175] 2. Input temperature field T into the thermodynamics module;

[0176] 3. Solve the equations simultaneously to output the three-dimensional thermal reservoir distribution.

[0177] This embodiment also provides a device for exploring medium-deep geothermal resources based on multi-source data fusion, used to execute the above-described method for exploring medium-deep geothermal resources based on multi-source data fusion. The device includes: a data acquisition module, a data preprocessing module, a data fusion module, and a verification and correction module. The data acquisition module is connected to the data preprocessing module, the data fusion module is connected to the data preprocessing module, and the verification and correction module is connected to the data fusion module.

[0178] The data acquisition module integrates a high-precision gravimeter (resolution ≤0.1mGal), a distributed fiber optic temperature measurement system (temperature measurement accuracy ±0.1℃), and an UAV-borne multispectral thermal infrared imager.

[0179] The data preprocessing module is configured with an FPGA-accelerated wavelet packet transform unit and a GPU-parallel principal component analysis unit.

[0180] The data fusion module includes a Bayesian network modeling unit, an evidence theory reasoning engine, and a dynamic weight optimization unit;

[0181] The verification and correction module achieves closed-loop optimization through the drilling equipment control unit and machine learning algorithms.

[0182] This exploration device integrates high-precision acquisition equipment, FPGA / GPU accelerated processing units, and a multi-module collaborative architecture to achieve full automation and efficiency from data acquisition and preprocessing to fusion verification, reducing the cost of manual intervention and improving the efficiency and real-time performance of exploration operations.

[0183] The data acquisition module also includes: a downhole three-component microseismic monitoring array and an automated online fluid chemical composition analyzer, wherein:

[0184] A downhole three-component microseismic monitoring array is used to acquire seismic wave data at depths ≥3km;

[0185] An automated online fluid chemical composition analyzer can detect the chemical components and isotope content of groundwater in real time.

[0186] The data acquisition module has been upgraded with a downhole microseismic monitoring array and an online chemical analyzer, which can acquire deep seismic wave data and real-time fluid composition information, expanding the depth and real-time nature of data acquisition and providing richer raw data for deep geothermal reservoir structure analysis and hydrochemical characteristic research.

[0187] The data acquisition module also includes: a borehole fracture imaging system and a distributed fiber optic water-rich monitoring instrument, wherein:

[0188] The borehole fracture imaging system uses a downhole high-definition camera and laser scanning technology to acquire three-dimensional spatial distribution data of fracture width, density, and orientation on the borehole wall;

[0189] Technical parameters: Imaging resolution ≤1mm, detection depth ≥3km, supports real-time transmission of fracture network 3D model.

[0190] Technical benefits: It directly reveals the geometry and connectivity of fractures in deep thermal reservoirs, providing high-precision measured data for the construction of fracture network models.

[0191] The distributed fiber optic water-rich monitoring instrument uses fiber optic sensing technology to deploy distributed temperature-strain sensors along the borehole to monitor changes in formation temperature gradient, pore water pressure and permeability in real time, and invert the water-rich distribution.

[0192] Technical parameters: Spatial resolution ≤0.5m, temperature measurement accuracy ±0.1℃, strain sensitivity ≤1με.

[0193] Technical benefits: Dynamically captures groundwater activity characteristics and spatial distribution of water-rich areas, and assesses thermal reservoir recharge capacity by combining chemical composition data.

[0194] In summary:

[0195] Fracture Imaging System: Breaking through the indirect inference of fractures by traditional geophysical methods, it directly obtains the three-dimensional spatial distribution of fracture networks through in-situ imaging, improving modeling accuracy by ≥40%.

[0196] It supports refined identification of deep (≥3km) fractures, avoiding reservoir evaluation bias caused by missing data.

[0197] Fiber optic water-rich monitor: Enables real-time dynamic monitoring of water-rich parameters, and accurately locates high-permeability areas by combining temperature and strain data;

[0198] In conjunction with a chemical analyzer, it reveals the spatiotemporal evolution patterns of hot water transport pathways and chemical components, providing data support for the sustainable development of resources.

[0199] The new scheme fills the gap in traditional exploration techniques for direct measurement of fracture networks and water-bearing properties, forming a three-in-one data acquisition system of "structure-physical properties-fluid", which significantly enhances the comprehensive analysis capability of medium-deep geothermal reservoir systems.

[0200] The workflow of the multi-source data fusion-based medium-deep geothermal resource exploration device is as follows:

[0201] 1. Data Acquisition Phase:

[0202] Gravity and geothermal data: High-precision gravimeters and distributed fiber optic temperature measurement systems simultaneously collect underground density and temperature field data;

[0203] Remote sensing data: The distribution of surface thermal anomalies and vegetation indices was obtained by a UAV-borne thermal infrared imager;

[0204] Deep data: The downhole microseismic monitoring array transmits seismic wave data at depths of ≥3km in real time, and the automatic fluid analyzer detects the chemical composition of groundwater online.

[0205] 2. Preprocessing stage:

[0206] Noise reduction: The FPGA-accelerated wavelet packet transform unit performs frequency division decomposition on the original data to filter out high-frequency noise and low-frequency drift;

[0207] Feature extraction: GPU parallel principal component analysis unit extracts principal components with a contribution rate ≥85% and reconstructs standardized data;

[0208] Normalization: The improved Z-score method eliminates dimensional differences and outputs data in a uniform format to the storage system.

[0209] 3. Data Fusion Stage:

[0210] Hierarchical integration:

[0211] Data layer: Bayesian network modeling units infer probabilistic dependencies among multi-source data;

[0212] Feature layer: The fuzzy inference engine maps multi-source features to the same fuzzy space, and the dynamic weight optimization unit assigns information entropy weights;

[0213] Decision-making level: The evidence theory reasoning engine processes conflicting evidence and outputs a comprehensive anomaly graph with confidence rating.

[0214] 4. Verification and Correction Phase:

[0215] Drilling verification: The target area is delineated based on the anomaly map, and the drilling equipment control unit drives the drilling rig to take samples and obtain core and water sample data;

[0216] Machine learning optimization: The support vector machine (SVM) is compared with the fusion results and drilling data, and the Bayesian network parameters and fuzzy membership function are adjusted in reverse.

[0217] Dynamic iteration: The corrected parameters are fed back to the fusion module, and the optimization is carried out in a loop until the error rate reaches the target.

[0218] 5. Output Results:

[0219] 3D modeling: Based on the fusion results, a geological-thermodynamic coupled model is constructed to visualize the distribution of thermal reservoirs and predict resource volume;

[0220] Decision support: Providing drilling target area recommendations and geothermal reservoir recoverable quantity assessment reports to guide actual exploration and development.

[0221] Through modular design and intelligent feedback mechanisms, the entire chain of multi-source data acquisition and decision-making is automated. The expansion of the data acquisition module enhances the monitoring capabilities of deep geological and fluid dynamics; the heterogeneous computing architecture of the preprocessing and fusion modules improves processing efficiency; and the closed-loop optimization mechanism of the verification and correction module significantly improves exploration accuracy and reliability. The overall solution demonstrates high adaptability and engineering practicality under complex geological conditions.

[0222] In summary, the method and apparatus for exploring medium-deep geothermal resources based on multi-source data fusion provided in this embodiment have the following advantages:

[0223] This invention overcomes the limitations of single-data exploration by systematically integrating and dynamically optimizing multi-source data, effectively handling data noise, conflicts, and uncertainties, and improving the accuracy and reliability of exploration for medium-deep geothermal resources. The constructed three-dimensional coupled model and the improved heat conduction equation solution method accurately characterize the geothermal reservoir characteristics under complex geological-thermodynamic environments. Combined with automated hardware devices, it achieves efficient operation throughout the entire process, providing an integrated solution for the scientific development of geothermal resources.

[0224] Overall Workflow

[0225] 1. Data Acquisition: High-precision equipment is used to acquire geological, geophysical, geochemical and remote sensing data of the target area, covering multiple types of information such as stratigraphic lithology, resistivity, groundwater chemical composition, and thermal infrared images.

[0226] 2. Preprocessing: Data quality is improved and the units are standardized by frequency division and noise reduction through dual-tree complex wavelet transform, extraction of high contribution features by principal component analysis, and improved Z-score normalization.

[0227] 3. Multi-level fusion: Construct a Bayesian network to represent the probabilistic dependencies of data, combine fuzzy inference and dynamic weight allocation (based on information entropy and noise level), and use improved evidence theory to handle conflicting evidence to achieve hierarchical fusion of multi-source data.

[0228] 4. Validation and Optimization: Generate a comprehensive anomaly map with confidence rating to delineate potential areas, obtain actual data through drilling and sampling, compare and fuse the results using support vector machine, and adjust the model parameters in reverse; construct a three-dimensional geological-thermodynamic coupling model, solve the improved non-Darcy flow heat conduction equation, assess the uncertainty of resource quantity prediction, and form a closed-loop optimization.

[0229] The above are merely preferred embodiments of the present invention and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should recognize that any equivalent substitutions and obvious changes made based on the description and illustrations of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for exploring medium-deep geothermal resources based on multi-source data fusion, characterized in that, Includes the following steps: Step S1: Collect multi-source data of the target area, including geological data, geophysical data, geochemical data and remote sensing data; Step S2: Preprocess the multi-source data, including wavelet packet transform frequency division decomposition, principal component analysis for noise reduction, and normalization. Step S3: Construct a Bayesian network model to represent the probabilistic dependencies between data sources, and combine fuzzy inference and information entropy weight allocation to achieve multi-level data fusion; Step S4: Generate a comprehensive anomaly map with confidence rating and delineate potential resource areas, and dynamically optimize model parameters by combining drilling verification and machine learning; In step S1: The geological data includes stratigraphic lithology, fault structures, and hydrogeological parameters; The geophysical data includes resistivity data, gravity data, and seismic shear wave velocity structure data. The geochemical data includes the chemical composition and isotopic composition of groundwater. The remote sensing data includes thermal infrared radiation images and multispectral vegetation index data; The preprocessing described in step S2 includes: Step S2a: Use dual-tree complex wavelet transform to decompose the original data into frequency bands, and filter out high-frequency noise and low-frequency drift; Step S2b: Perform principal component analysis on the data of each frequency band and extract principal component reconstruction data with a contribution rate ≥ 85%; Step S2c: Normalize the multi-source data based on the improved Z-score method to eliminate dimensional differences; In step S3: The fuzzy inference uses a Gaussian membership function to quantize feature mapping; The information entropy weights are dynamically adjusted by calculating the uncertainty of each data source, and the weight allocation formula is as follows: ； in: x i : Information entropy of the i-th data source, representing data uncertainty; λ: Fusion sensitivity coefficient, ranging from 0.5 to 2.0, controls the degree of concentration in weight allocation; σ 2 The noise variance of the data source is calculated through the preprocessing module; : Smoothing constant, defaults to 1e-5, to prevent the denominator from being zero.

2. The method for exploring medium-deep geothermal resources based on multi-source data fusion according to claim 1, characterized in that, The multi-level data fusion mentioned in step S3 includes: Step S3a: Infer the probabilistic relationships between data sources based on Bayesian networks; Step S3b: Use the Dempster-Shafer evidence theory to resolve the problem of merging conflicting evidence; Step S3c: Dynamically adjust the fusion weight coefficients using the particle swarm optimization algorithm; The conflicting evidence fusion employs a modified Dempster combination rule: ； in: : Conflict attenuation factor, where β is the attenuation coefficient; conf(B,C): The degree of conflict between evidence B and evidence C; K: The number of conflicts in the traditional Dempster rule; δ: Conflict compensation term, δ=0.1K.

3. The method for exploring medium-deep geothermal resources based on multi-source data fusion according to claim 1, characterized in that, The dynamic optimization model parameters mentioned in step S4 include: Step S4a: Obtain underground rock core and water sample data through drilling; Step S4b: Use the support vector machine algorithm to compare the fusion result with the actual data and calculate the error rate; Step S4c: If the error rate exceeds the threshold, adjust the Bayesian network parameters and fuzzy membership function in reverse.

4. The method for exploring medium-deep geothermal resources based on multi-source data fusion according to claim 1, characterized in that, Step S4 also includes: A three-dimensional geological-thermodynamic coupling model is constructed based on the fusion results. The three-dimensional geological-thermodynamic coupling model includes: Three-dimensional mesh model of stratigraphy and lithology; Three-dimensional distribution model of fracture structure; Dynamic distribution field of rock thermal conductivity.

5. The method for exploring medium-deep geothermal resources based on multi-source data fusion according to claim 4, characterized in that, The three-dimensional geological-thermodynamic coupled model solves the non-Darcy flow heat conduction equation using the finite volume method. Combined with Monte Carlo simulation to assess the uncertainty in resource quantity prediction, the solution to the non-Darcy flow heat conduction equation adopts the following modified form: ； Parameter description: : Nonlinear flow coefficient, φ is porosity, μ is fluid viscosity, k is permeability; γ: Temperature decay coefficient; P: Fluid pressure field.

6. A device for exploring medium-deep geothermal resources based on multi-source data fusion, characterized in that, The method for exploring medium-deep geothermal resources based on multi-source data fusion as described in any one of claims 1-5 includes: The data acquisition module integrates a high-precision gravimeter, a distributed fiber optic temperature measurement system, and an UAV-borne multispectral thermal infrared imager. The data preprocessing module is configured with an FPGA-accelerated wavelet packet transform unit and a GPU-parallel principal component analysis unit. The data fusion module includes a Bayesian network modeling unit, an evidence theory reasoning engine, and a dynamic weight optimization unit; The verification and correction module achieves closed-loop optimization through the drilling equipment control unit and machine learning algorithms.

7. The device for exploring medium-deep underground hydrothermal resources based on multi-source data fusion according to claim 6, characterized in that, The data acquisition module also includes: A downhole three-component microseismic monitoring array is used to acquire seismic wave data at depths ≥3km; An automated online fluid chemical composition analyzer can detect the chemical components and isotope content of groundwater in real time.

Images