A method, system, equipment, and medium for evaluating hydraulic fractures based on multi-source data.

Abstract

This invention discloses a method, system, equipment, and medium for evaluating hydraulic fractures based on multi-source data, relating to the field of hydraulic fracturing fracture characterization and reconstruction technology. The method includes: acquiring distributed fiber optic monitoring data and laser scanning data from a hydraulic fracturing physical simulation experiment; spatially aligning the two types of data; performing multi-source data cleaning; performing fracture reconstruction on the laser scanning data using a multi-source radial interpolation algorithm; fitting and recombining the fracture morphology waterfall plot and the laser scanning reconstructed image; extracting fracture information features and location information features from the two images; constructing a feature fusion network model; performing image fusion to obtain a fused feature representation; based on the fused feature representation, performing image reconstruction on the laser scanning reconstructed image to obtain the final fused image; and using evaluation metrics to evaluate the hydraulic fractures in the final fused image to obtain the evaluation result. This method achieves high-precision, multi-dimensional dynamic characterization of fracture morphology.

Description

Technical Field

[0001] This invention relates to the field of hydraulic fracturing fracture characterization and reconstruction technology, specifically to a hydraulic fracture evaluation method, system, equipment, and medium based on multi-source data. Background Technology

[0002] In oil and gas field fracturing operations, directly monitoring and characterizing the fracture network morphology after fracturing presents significant challenges due to the fact that reservoirs are typically buried deep underground. Under these conditions, obtaining crucial information on fracture morphology and propagation patterns through on-site operations has significant limitations. Therefore, conducting large-scale, true triaxial hydraulic fracturing physical simulation experiments in the laboratory can effectively compensate for the lack of on-site data, providing a scientific basis for a deeper understanding of fracture propagation patterns and optimization of fracturing processes.

[0003] In actual field operations, reservoir depths typically exceed 2000 meters. Conventional microseismic monitoring methods suffer from fracture location errors of 10-30 meters due to complex geological conditions and signal attenuation. Furthermore, well logging interpretation relies on empirical models, which struggle to accurately reflect the spatial distribution characteristics of complex fractures. The quantitative coupling relationship between multiple parameters—such as formation heterogeneity in tight reservoirs (e.g., differences in rock elastic modulus of 2-5 times), horizontal stress differences (typically 3-10 MPa), and fracturing fluid viscosity (20-200 mPa·s)—and fracture morphology has not yet been fully established. Existing laboratory experiments often employ simplified two-dimensional models, which cannot accurately reproduce the triaxial stress state of the reservoir (vertical stress is typically 1.2-2 times that of horizontal stress), leading to significant errors in the scaling-down conversion of construction parameters.

[0004] These limitations indicate that existing methods are significantly inadequate in characterizing fracture morphology. To address this issue, there is an urgent need for high-precision dynamic monitoring systems, as well as fracture 3D morphology reconstruction techniques and millimeter-level precision in capturing and reconstructing fracture propagation processes. Currently used fracture characterization methods, such as JRC parameters, fractal theory, and 3D surface photometers, while providing some fracture characteristic information, cannot comprehensively and accurately characterize complex fracture morphologies. Real-world reservoir fracture morphologies are complex and diverse, including natural fractures, artificial fractures, and complex fracture networks formed by their interactions. The morphology, orientation, and connectivity of these fractures are difficult to accurately characterize using existing experimental methods.

[0005] Therefore, how to fully consider the complex situation of tight reservoirs and characterize the hydraulic fracture morphology in real time and with high precision, so as to more comprehensively clarify the influence of each parameter on the fracture morphology and spatial distribution scale, has become an important technical problem that needs to be solved in the indoor true triaxial hydraulic fracturing physical simulation experiment. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method, system, device, and medium for evaluating hydraulic fractures based on multi-source data, thereby resolving the technical problems existing in the prior art.

[0007] In a first aspect, the first embodiment of the present invention provides a method for evaluating hydraulic fractures based on multi-source data, comprising the following steps:

[0008] Acquire distributed fiber optic monitoring data and laser scanning data in fracturing physical simulation experiments;

[0009] The distributed fiber optic monitoring data and laser scanning data are spatially aligned to obtain the aligned data.

[0010] The aligned distributed optical fiber monitoring data is cleaned to obtain cleaned distributed optical fiber monitoring data.

[0011] The laser scanning data is coupled with a multi-source radial interpolation algorithm to reconstruct the crack, resulting in a laser scanning reconstruction image;

[0012] The crack morphology waterfall map and laser scan reconstruction map formed from the cleaned distributed optical fiber monitoring data are fitted and reconstructed. Crack information features and location information features are extracted from the crack morphology waterfall map and laser scan reconstruction map, and encoded separately to obtain crack feature vectors and location feature vectors.

[0013] A feature fusion network model is constructed, and the crack feature vector and the location feature vector are input into the feature fusion network model for fusion to obtain the fused feature representation. Based on the fused feature representation, image reconstruction is performed on the laser scanning reconstruction map to obtain the final fused image.

[0014] The final fused image was evaluated for hydraulic fractures using evaluation metrics, and the evaluation results were obtained.

[0015] Furthermore, the specific method for spatially aligning the distributed optical fiber monitoring data and laser scanning data includes:

[0016] The coordinate systems of distributed optical fiber monitoring data and laser scanning data are subjected to matrix eigenvalue decomposition, and the spatial transformation matrix is ​​calculated.

[0017] The translation vector is calculated using the spatial transformation matrix, the centroid of the distributed optical fiber monitoring data, and the centroid of the laser scanning data.

[0018] An iterative calculation method is used to spatially align distributed fiber optic monitoring data and laser monitoring data.

[0019] Furthermore, the method for performing matrix eigenvalue decomposition on the coordinate system of the distributed optical fiber monitoring data and the coordinate system of the laser scanning data to calculate the spatial transformation matrix includes:

[0020] Considering the difference between the number of distributed fiber optic monitoring data points and the number of laser scanning data points, based on nearest neighbor matching, a corresponding laser scanning point is found for each distributed fiber optic monitoring point, resulting in multiple matching point pairs.

[0021] Based on each matching point pair, extract the coordinates of the fiber optic monitoring point and the laser scanning point;

[0022] A relationship matrix is ​​constructed using the coordinate matrix of fiber optic monitoring data and the coordinate matrix of laser scanning data. The distributed fiber optic monitoring data and laser scanning data are then centralized to obtain the coordinate matrix of the distributed fiber optic data and the coordinate matrix of the laser scanning data after centralization.

[0023] Extract the centered coordinates of the matched fiber optic monitoring points and the centered coordinates of the laser scanning points;

[0024] The centralized coordinates of the fiber optic monitoring point and the centralized coordinates of the laser scanning point are combined into a centralized relation matrix. Each row of the centralized relation matrix contains the centralized fiber optic monitoring coordinates and the centralized laser scanning coordinates of a matching point pair.

[0025] The singular value decomposition of the centered relation matrix is ​​performed to obtain the decomposition result, and the spatial transformation matrix is ​​determined based on the decomposition result.

[0026] Furthermore, the specific method for spatially aligning distributed optical fiber monitoring data and laser monitoring data using iterative calculation includes:

[0027] Initialize the spatial transformation matrix and translation vector. In each iteration, based on the current spatial transformation matrix and the current translation vector, transform the distributed optical fiber monitoring data into the coordinate space of the laser scanning data to obtain the transformed distributed optical fiber monitoring data.

[0028] Calculate the root mean square error between the converted distributed fiber optic monitoring data and the laser scanning data;

[0029] The gradient descent optimization algorithm is used, and the parameters of the current spatial transformation matrix and the current translation vector are adjusted according to the root mean square error to obtain a new spatial transformation matrix and translation vector, which are then used to replace the current spatial transformation matrix and the current translation vector.

[0030] By setting the iteration termination condition, the optimized spatial transformation matrix and translation vector are finally obtained.

[0031] Furthermore, the feature fusion network model includes an input layer, a feature extraction layer, a fusion layer, and an output layer. The input layer is used to input crack feature vectors and location feature vectors. The feature extraction layer uses convolutional layers to extract local features from the crack feature vectors and location feature vectors. The fusion layer fuses the local features extracted by the feature extraction layer to obtain a fused feature vector. The output layer outputs the fused feature representation.

[0032] Furthermore, the evaluation metrics include image sharpness, contrast, information entropy, and structural similarity.

[0033] Secondly, another embodiment of the present invention provides a hydraulic fracture evaluation system based on multi-source data, comprising: a data acquisition module, a data processing module, an image fusion module, and an evaluation module.

[0034] The data acquisition module is used to acquire distributed fiber optic monitoring data and laser scanning data in the fracturing physical simulation experiment;

[0035] The data processing module is used to clean the aligned distributed optical fiber monitoring data to obtain cleaned distributed optical fiber monitoring data, and to perform crack reconstruction on the laser scanning data according to the multi-source radial interpolation algorithm to obtain the laser scanning reconstruction map. The crack morphology waterfall map and the laser scanning reconstruction map formed from the cleaned distributed optical fiber monitoring data are fitted and recombined. Crack information features and location information features are extracted from the crack morphology waterfall map and the laser scanning reconstruction map, and encoded respectively to obtain crack feature vectors and location feature vectors.

[0036] The image fusion module is used to construct a feature fusion network model. The crack feature vector and the location feature vector are input into the feature fusion network model for fusion to obtain the fused feature representation. Based on the fused feature representation, image reconstruction is performed on the laser scanning reconstruction image to obtain the final crack morphology image.

[0037] The evaluation module is used to evaluate the hydraulic fractures in the final fracture morphology image using evaluation indicators, and obtain the evaluation results.

[0038] Furthermore, the data processing module includes a spatial alignment unit, which is used to perform matrix eigenvalue decomposition on the coordinate system of the distributed optical fiber monitoring data and the coordinate system of the laser scanning data, and calculate the spatial transformation matrix; calculate the translation vector through the spatial transformation matrix, the centroid of the distributed optical fiber monitoring data and the centroid of the laser scanning data; and use an iterative calculation method to align the distributed optical fiber monitoring data and the laser monitoring data in space.

[0039] Thirdly, another embodiment of the present invention provides an electronic device comprising: a processor, an input device, an output device, and a memory, wherein the processor, the input device, the output device, and the memory are interconnected, the memory is used to store a computer program, the computer program including program instructions, wherein the processor is configured to invoke the program instructions to execute the method described in the first embodiment above.

[0040] Fourthly, another embodiment of the present invention provides a computer-readable storage medium storing a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the method described in the first embodiment above.

[0041] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0042] This invention provides a method, system, device, and medium for evaluating hydraulic fractures based on multi-source data. It combines distributed fiber optic monitoring technology with three-dimensional laser scanning technology to accurately characterize hydraulic fracture morphology based on multi-source data. This breakthrough overcomes the technical bottlenecks of traditional fracture monitoring by integrating the real-time performance and data diversity of distributed fiber optic monitoring with the intuitive visualization capabilities of three-dimensional laser scanning technology, achieving high-precision, multi-dimensional dynamic characterization of fracture morphology during hydraulic fracturing. Distributed fiber optic monitoring can capture real-time changes in multiple parameters such as temperature and strain during fracture propagation, while three-dimensional laser scanning provides an intuitive representation of fracture geometry. The synergistic effect of both significantly improves the accuracy and reliability of fracture characterization. This multi-technology integration not only provides a comprehensive description of fracture morphology from multiple angles and parameters but also significantly enhances the ability of indoor simulation experiments to reproduce complex fracture networks in the field, providing strong technical support for hydraulic fracturing optimization design and fracture network modeling. Attached Figure Description

[0043] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0044] Figure 1 A schematic diagram of the structure of a wellbore spiral distributed optical fiber experimental frame;

[0045] Figure 2 This is a schematic diagram of an internal winding wiring device.

[0046] Figure 3 This is a schematic diagram illustrating the principle of distributed optical fiber monitoring.

[0047] Figure 4 A flowchart of a hydraulic fracture evaluation method based on multi-source data provided in the first embodiment of the present invention;

[0048] Figure 5 This is a schematic diagram of a technical route for a hydraulic fracture evaluation method based on multi-source data provided in the first embodiment of the present invention;

[0049] Figure 6 (a) is a diagram of the crack morphology after compression of a certain group of samples;

[0050] Figure 6 (b) Reconstructed image of crack morphology after compression in a certain group of samples;

[0051] Figure 7 This is a schematic diagram of a hydraulic fracture evaluation system based on multi-source data, provided as another embodiment of the present invention. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0054] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0055] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes such combinations.

[0056] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if [described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [described condition or event] is detected," or "in response to detection of [described condition or event]."

[0057] Currently, distributed fiber optic monitoring cabling often employs a U-shaped wrapping method within the rock. However, due to the high brittleness of the fiber optic monitoring lines, they are prone to breakage during solidification within artificial rock samples; furthermore, the position of individual fiber optic lines is easily altered, leading to sample preparation failures or measurement point deviations. Therefore, a distributed fiber optic cabling device was designed for use in indoor experiments. Based on experimental requirements, a suitable fiber optic cabling method was selected, including wellbore spiral and internal winding methods. After debugging, the experiment began, and fiber optic monitoring data was recorded. After the fracturing experiment, a spatial coordinate system was established at the experimental site. The wellbore location was set as the origin O (0, 0, 0), the wellhead location as the vertex in the Z-direction (0, 0, 450), the X-direction coordinates as (500, 0, 0), and the Y-direction coordinates as (0, 500, 0); all units are mm. Under this coordinate system, laser scanners were placed at four points in the space, scanning from bottom to top, with each movement distance being 10 mm, and the highest position the scanner reached being 450 mm.

[0058] Among them, the well-tube spiral distributed optical fiber experimental rig is an improvement on the traditional well-tube structure, as shown in the structural diagram. Figure 1 As shown, the wellbore consists of three parts: the wellbore injection end, connected to the injection pipeline; the wellbore sealing end, a cylindrical section that completely covers the entire wellbore and bears pressure during fracturing; and the lower wellbore section, a hollow, thickened pressure-bearing steel pipe. A 5mm diameter fiber optic extension hole is made at the wellbore sealing end, allowing the fiber optic cable to be inserted into the wellbore section. Grooves 3-5mm deep are etched into the wellbore section, spiraling to the bottom. These grooves facilitate the placement of distributed fiber optic cables. If coated fiber optics are used as the monitoring device, the wellbore does not need to be wrapped with a thermoplastic film; if bare fiber optics are used, a thermoplastic film must be wrapped around the outside of the wellbore to prevent the fiber optic cable from being scratched within the rock.

[0059] During hydraulic pressure testing, crack network extension often exhibits a main network plus branch network distribution. To better understand the location and scale of crack extension, a single-line U-shaped fiber optic cable arrangement is insufficient for crack monitoring needs. Therefore, an internally wound cabling method is required. Internally wound cabling devices, such as... Figure 2 As shown.

[0060] The internal spiral cabling device features a helical steel frame structure, approximately 8mm in diameter. A groove, about 3mm deep, is cut into the upper surface of the frame, facilitating the embedding of coated optical fibers into the experimental fixture. Because the grooves are all open on one side, the impact of the experimental fixture on the signal is significantly reduced. The experimental fixture demonstrated using this method is 180mm long with a helical bend of 270°. The length and helical bend of the experimental fixture can also be designed according to experimental requirements.

[0061] The light source in a distributed fiber optic monitoring system can be divided into two paths: a reference light and a probe light. During the transmission of the probe light, Brillouin scattering occurs continuously. The scattered and reflected signals are coupled together by a coupler to form beat frequency interference, which is ultimately detected by a photodetector. The principle is explained in [link to principle]. Figure 3 The relationship between Brillouin scattering and fiber strain is as follows:

[0062]

[0063] In equation (1), For fiber strain; In order to adapt to Brillouin shift at that time; The Brillouin frequency shift is when the fiber strain is 0. This is the strain coefficient, which is related to the fiber type and the frequency of the probe light.

[0064] like Figure 4 As shown, the first embodiment of the present invention provides a hydraulic fracture evaluation method based on multi-source data, which includes the following steps:

[0065] Acquire distributed fiber optic monitoring data and laser scanning data in fracturing physical simulation experiments;

[0066] The distributed fiber optic monitoring data and laser scanning data are spatially aligned to obtain the aligned data.

[0067] The aligned distributed optical fiber monitoring data is cleaned to obtain cleaned distributed optical fiber monitoring data.

[0068] The laser scanning data is coupled with a multi-source radial interpolation algorithm to reconstruct the crack, resulting in a laser scanning reconstruction image;

[0069] The crack morphology waterfall map and laser scan reconstruction map formed from the cleaned distributed optical fiber monitoring data are fitted and reconstructed. Crack information features and location information features are extracted from the crack morphology waterfall map and laser scan reconstruction map, and encoded separately to obtain crack feature vectors and location feature vectors.

[0070] A feature fusion network model is constructed, and the crack feature vector and the location feature vector are input into the feature fusion network model for fusion to obtain the fused feature representation. Based on the fused feature representation, image reconstruction is performed on the laser scanning reconstruction map to obtain the final fused image.

[0071] The final fused image was evaluated for hydraulic fractures using evaluation metrics, and the evaluation results were obtained.

[0072] In this embodiment, fiber optic monitoring and three-dimensional laser scanning are used to acquire hydraulic fracture data. For example... Figure 5 As shown, the data points obtained by fiber optic monitoring and 3D laser scanning methods all have corresponding spatial coordinates. The format of fiber optic monitoring data is: spatial coordinates are... The data points are The data format for laser scanning cracks is: spatial coordinates. The data points are .

[0073] The spatial coordinates of each point in the distributed fiber optic monitoring data are spatially unified with those of the laser scanning points. Coordinate spatial unification typically involves two steps: coarse spatial coordinate calibration and fine spatial coordinate calibration. Since distributed fiber optic monitoring and laser scanning monitoring are performed within the same coordinate system, coordinate system transformation is unnecessary; that is, coarse calibration is not required. Only the second step, fine spatial coordinate calibration, needs to be performed.

[0074] In this embodiment, the specific method for spatially aligning distributed fiber optic monitoring data and laser scanning data includes:

[0075] The coordinate systems of distributed optical fiber monitoring data and laser scanning data are subjected to matrix eigenvalue decomposition to calculate the spatial transformation matrix.

[0076] The translation vector is calculated using the spatial transformation matrix, the centroid of the distributed optical fiber monitoring data, and the centroid of the laser scanning data.

[0077] An iterative calculation method is used to spatially align distributed fiber optic monitoring data and laser monitoring data.

[0078] Specifically, matrix eigenvalue decomposition is performed based on the coordinates of the two coordinate systems. The spatial transformation matrix K is calculated through the eigenvalue decomposition, and a threshold for the spatial transformation matrix is ​​set.

[0079] Suppose there are p points in the distributed optical fiber monitoring data, and the coordinate matrix F of the distributed optical fiber monitoring data is:

[0080]

[0081] The laser scan data has q points, and the coordinate matrix S of the laser scan data is:

[0082]

[0083] In equations (2)-(3), These represent the X, Y, and Z dimensions of the fiber optic monitoring data coordinate matrix. , , These represent the X, Y, and Z dimensions of the laser scanning data coordinate matrix, respectively; R only represents the dimension of the matrix and has no physical meaning. This represents the coordinate matrix of fiber optic monitoring data, where F has dimensions of 1. p represents the number of fiber optic monitoring points; The dimension of the laser scanning data coordinate matrix S is . q represents the number of laser scanning points.

[0084] Considering that the two types of data points have different numbers (p points for fiber optic monitoring data and n points for laser scanning data), the nearest neighbor matching method is used to find the corresponding laser scanning point for each distributed fiber optic monitoring point, resulting in k matching point pairs.

[0085] For each pair of matching points, extract the coordinate matrix of the fiber optic monitoring points. and laser scanning point coordinate matrix They are respectively:

[0086]

[0087] In equations (4)-(5), k is the number of successfully matched point pairs, usually k ≤ min(p,q). These represent the X, Y, and Z dimensions of the fiber optic monitoring point coordinate matrix. These represent the X, Y, and Z dimensions of the laser scanning point coordinate matrix, respectively.

[0088] A relationship matrix is ​​constructed using the coordinate matrices of the two sets of data, and then multiplied by the centered coordinate matrices. The distributed fiber optic monitoring data F and the laser scanning data S are centered separately, as shown in the following formula:

[0089]

[0090] In equations (6)-(7), This represents the coordinate matrix after centralized processing of distributed fiber optic monitoring data. This represents the coordinate matrix after the laser scanning data has been centered. It is a p-dimensional column vector. It is a q-dimensional column vector with all elements being 1.

[0091] Centered coordinates of the matched fiber optic monitoring points The centered coordinates of the laser scanning point coordinates The formula is:

[0092]

[0093] In equations (8)-(9), and These represent the centered X-axis coordinates of the k-th fiber optic monitoring point and the laser scanning point, respectively. and These represent the centered Y-axis coordinates of the k-th fiber optic monitoring point and the laser scanning point, respectively. and These represent the Z-axis coordinates of the k-th fiber optic monitoring point and the laser scanning point, respectively, after centralization.

[0094] The centered coordinates are combined into a centered relation matrix M:

[0095]

[0096] Each row of the centralized relation matrix M contains the centralized fiber optic monitoring coordinates and the centralized laser scanning coordinates of a matching point pair.

[0097] Perform singular value decomposition (SVD) on the centered relation matrix M:

[0098]

[0099] In the formula, It is an orthogonal matrix. It is a diagonal matrix, with diagonal elements being singular values, arranged in order of size.

[0100] Based on the results of SVD, the spatial transformation matrix K can be determined. Consider the transformation relationships of rotation, scaling, and translation:

[0101]

[0102] In the formula, K is a 3×3 transformation matrix that contains rotation and scaling information. It is a translation vector.

[0103] To solve for K and b, we can use the results of SVD:

[0104]

[0105] However, this only considers rotation. If scaling is taken into account, K can be expressed as:

[0106]

[0107] In the formula, D is a diagonal matrix, and the diagonal elements are scaling factors, which can be determined according to actual needs.

[0108] The translation vector b is obtained by calculating the centroid difference between the two sets of data:

[0109]

[0110] In the formula, It is the centroid of the distributed optical fiber monitoring data, that is It is the centroid of the laser scanning data, that is .

[0111] Because the data cloud from distributed fiber optic monitoring differs in quantity from the data cloud from laser monitoring, or because of noise, iterative calculations are needed to align the two sets of data spatially as much as possible.

[0112] Initialize the spatial transformation matrix K0 and the translation vector b0. For example, the initial transformation matrix can be the identity matrix and the translation vector can be the zero vector. In each iteration, based on the current transformation matrix K0... i Translation vector b i The distributed fiber optic monitoring data is converted into the coordinate space of the laser scanning data:

[0113]

[0114] Calculate the converted distributed fiber optic monitoring data The root mean square error between the laser scanning data S and the data is:

[0115]

[0116] In equations (16)-(17), k1 is the number of matching point pairs. It is the j-th converted distributed optical fiber monitoring point. These are the corresponding laser scanning points.

[0117] Based on the gradient descent optimization algorithm and adjusting the transformation matrix K according to the error, i Translation vector b i The parameters,

[0118]

[0119] Then update the transformation matrix and translation vector:

[0120]

[0121] In equations (20)-(21), It's the learning rate, which controls the step size.

[0122] During actual monitoring and scanning, due to limitations in equipment accuracy, environmental factors, or data processing methods, multiple data points may be generated near the same physical location. While these data points may have slight differences in coordinates, they essentially belong to the same physical location or area. For points with a large volume of data at the same location, separate sets are set up for management. Multiple points at the same location from the distributed fiber optic monitoring data are placed into these sets. Multiple points at the same location in the laser scanning data are placed into a set. When calculating the error, for each position e, the average coordinates or other statistics of the points within the set can be calculated before performing the error calculation, so as to more accurately assess the correspondence and error.

[0123] Set an iteration termination condition, such as the error being less than a certain threshold. (For example (or reaching the maximum number of iterations) (For example The iteration stops when the time is reached, and the optimized spatial transformation matrix is ​​finally obtained. Translation vector .

[0124] The evaluation of monitoring data includes the following two aspects:

[0125] (1) Relative error

[0126] If the actual temperature or strain values ​​at certain locations are known, the monitoring data can be compared with the true values ​​to calculate the relative error. The calculation formula is as follows:

[0127]

[0128] In the formula, To monitor the obtained temperature or strain values, This represents the actual temperature or strain value at that location.

[0129] (2) Stability

[0130] For monitoring data from the same location at different times, calculate their mean and standard deviation to assess data stability. A smaller standard deviation indicates more stable data. The formula for calculating the standard deviation is:

[0131]

[0132] In the formula, is the th Secondary monitoring data, is the mean of the monitoring data, and n is the number of monitoring data.

[0133] Simultaneously, the cracking and opening of rocks produces the sound of rock breaking. This sound is transmitted through the solid to the optical fiber, where the vibration of the sound waves causes slight deformation and varying degrees of scattering. Because the experiment involves a variety of background sounds, a specific acoustic signal library needs to be established.

[0134] To address the different frequency characteristics of sound signals, a Fast Fourier Transform (FFT) is used to perform spectral analysis, converting the time-domain signal into a frequency-domain signal to determine its main frequency components. Let the sound signal be x(t), and its FFT is:

[0135]

[0136] In the formula, It is a frequency domain representation, calculated The spectral amplitude of the signal can be obtained, thereby determining the frequency characteristics of the signal.

[0137] Different types of sound signals during fracturing are collected, including rock fracturing sounds, fluid flow sounds, machine operation sounds, and laboratory personnel noise. Feature extraction and annotation are performed on these sound signals to construct an acoustic signal library. Simultaneously, characteristic parameters for each sound signal, such as frequency range and waveform characteristics, are recorded. For example, feature extraction of rock fracturing sound signals reveals that their frequency range is mainly concentrated around 100-150Hz; feature extraction of water flow sound signals within pipelines reveals that their frequency range is mainly concentrated around 500-900Hz; and feature extraction of laboratory noise signals reveals that their frequency range is mainly concentrated around 300-800Hz.

[0138] As the monitoring process progresses, new sound signals are continuously collected, and the acoustic signal database is updated and optimized. For newly emerging unknown sound signals, clustering algorithms can be used for classification and identification. Based on the clustering results, their category is determined, and the corresponding category in the acoustic signal database is updated.

[0139] In actual monitoring, the acquired sound signals may contain multiple sound components. Using feature information from an acoustic signal database, mixed sound signals can be separated and filtered. For example, a bandpass filter can be used to retain the frequency range of the target sound signal while filtering out noise at other frequencies. For monitoring rock fracture sounds, a bandpass filter with a center frequency of 125Hz and a bandwidth of 50Hz can be designed to filter out high-frequency noise above 500Hz and low-frequency noise below 50Hz. Let the frequency response of the filter be... Then the filtered signal y(t) is:

[0140]

[0141] In the formula, This represents the inverse Fourier transform.

[0142] The acquired sound signals are compared with standard sound signals in an acoustic signal library using correlation analysis. The similarity between the acquired signal and each sound signal in the library is calculated; if the similarity exceeds a certain threshold, the acquired signal is considered to contain that sound component. Based on the correlation analysis results, the acquired sound signals are classified and identified to eliminate background noise contamination and improve the accuracy of crack interpretation. Let the acquired sound signal be... The standard sound signals in the acoustic signal library are Then their correlation coefficient r can be expressed as:

[0143]

[0144] In the formula, These are the mean values ​​of the acquired signal and the standard signal, respectively. If r is greater than the set threshold (usually 0.7), the acquired signal is considered to contain that sound component.

[0145] Compared with temperature and sound signal characteristics, deformation signal characteristics have the advantages of clear features and less data interference, but the thickness of the monitored object is relatively limited. Laser scanning data processing mainly utilizes scalar function methods to decompose and reconstruct the data.

[0146] Based on data alignment and standardization, crack reconstruction is performed using a coupled multi-source radial interpolation intelligent algorithm. The interpolation method used here is radial basis function interpolation (RBF), which does not require any assumptions about the data structure and can be expressed as follows through polynomial enhancement:

[0147]

[0148] In the formula, As the interpolation center; These are radial basis functions; To enhance the polynomial; is the weighting coefficient; n is the dimension of the m-1 order polynomial space.

[0149] First, two key features—crack information and location information—are extracted from two types of images (a waterfall image of crack morphology generated by fiber optic monitoring and a laser scan reconstruction image) using a flattened circle detection algorithm. For crack information, geometric features such as crack length, width, height, and morphology are extracted; for location information, parameters such as the crack's coordinates and orientation within the image are extracted. Then, an image registration algorithm is used to determine the crack's position coordinates within the image.

[0150] A fitting and reconstruction model is constructed to fit the waterfall plot of crack morphology from fiber optic monitoring with the reconstructed plot from laser scanning. The mean square error function of the crack information and location information is used as the objective function, expressed as:

[0151]

[0152] In the formula, It is the i-th feature value of the crack in the laser scan reconstruction image (such as crack length, location coordinates, etc.). It is the predicted value of the crack feature corresponding to the crack morphology waterfall plot monitored by optical fiber.

[0153] The crack information and location information extracted from the two types of monitoring images are encoded separately. For example, the crack information is feature-encoded and converted into a crack feature vector. : The crack information includes geometric features such as length l, width w, and height h; the location information is encoded into coordinates and converted into a location coordinate vector. .

[0154] Design a feature fusion network to fuse encoded crack and location features. The network employs a convolutional neural network architecture for automatic feature extraction, encoding, and fusion. During the fusion process, the network's learning mechanism mines the correlation and complementarity between the two feature types to generate a fused feature representation.

[0155] The feature fusion network architecture is as follows:

[0156] (1) Input layer

[0157] The crack feature vector and the location feature vector are used as inputs to the convolutional neural network.

[0158] (2) Feature extraction layer

[0159] Convolutional layers are used to extract local features for both types of features. For crack features, a one-dimensional convolutional layer is used to extract their local features; for location features, a fully connected layer is used to map them to a high-dimensional space.

[0160] (3) Fusion layer

[0161] The extracted local features are then fused. This can be done by concatenating the local feature vectors of the two types of features together to form a fused feature vector.

[0162] (4) Output layer

[0163]

[0164] In the formula, This represents the original crack feature data; This represents the convolutional neural network portion used to extract crack features; This represents the output features obtained after processing by a convolutional neural network; This represents a fully connected layer used to extract location features; Indicates input to the fully connected layer The original location feature data used to extract location features; Output features for fully connected layers; Indicates feature concatenation operation; This represents the feature representation after fusion.

[0165] Based on the fused feature representation Image reconstruction is performed based on laser-scanned images. The high-resolution pixel information from the laser scan is used to compensate for potential pixel loss during the fusion process. A bilinear interpolation algorithm can be employed to improve the resolution and sharpness of the fused image, compensating for pixel loss and generating the final fused image. Assuming the input image is I, and the magnified image is... The scaling factor is s. The formula for bilinear interpolation is as follows:

[0166]

[0167] In the formula, These are the coordinates of the four surrounding pixels.

[0168] Multiple evaluation metrics are used to evaluate the fused image, such as image sharpness, contrast, information entropy, and structural similarity, to obtain evaluation results. The quality and effect of the fused image are assessed by comparing it with the original image or the real scene. This embodiment uses structural similarity for evaluation, and the formula for calculating the structural similarity index is:

[0169]

[0170] In the formula, x represents the fused image; y represents the original image. These are the means of the x and y values ​​of the image, respectively. These are the variances of the images x and y, respectively. It is the covariance of the images x and y. , It is a constant.

[0171] The output fused image is compared with the experimental sample photograph to capture structural deficiencies in the sample. The parameters and structure of the feature fusion network are adjusted, and the process is continuously improved through iterative iterations, gradually refining the quality and effect of the fused image. The results are as follows: Figure 6 As shown, Figure 6(a) is a diagram of the crack morphology after compression in a certain group of samples. Figure 6 (b) Reconstruction diagram of crack morphology after compression of a certain group of samples.

[0172] This invention provides a method for evaluating hydraulic fractures based on multi-source data. It innovatively combines distributed fiber optic monitoring technology with three-dimensional laser scanning technology to accurately characterize hydraulic fracture morphology based on multi-source data. This method overcomes the technical bottlenecks of traditional fracture monitoring by integrating the real-time performance and data diversity of distributed fiber optic monitoring with the intuitive visualization capabilities of three-dimensional laser scanning technology. This achieves high-precision, multi-dimensional dynamic characterization of fracture morphology during hydraulic fracturing. Distributed fiber optic monitoring can capture real-time changes in multiple parameters such as temperature and strain during fracture propagation, while three-dimensional laser scanning provides an intuitive representation of fracture geometry. The synergistic effect of both significantly improves the accuracy and reliability of fracture characterization. This multi-technology integration not only provides a comprehensive description of fracture morphology from multiple angles and parameters but also significantly enhances the ability of indoor simulation experiments to reproduce complex fracture networks in the field, providing strong technical support for hydraulic fracturing optimization design and fracture network modeling.

[0173] like Figure 7 As shown, another embodiment of the present invention provides a hydraulic fracture evaluation system based on multi-source data, comprising: a data acquisition module, a data processing module, an image fusion module, and an evaluation module.

[0174] The data acquisition module is used to acquire distributed fiber optic monitoring data and laser scanning data in the fracturing physical simulation experiment;

[0175] The data processing module is used to clean the aligned distributed optical fiber monitoring data to obtain cleaned distributed optical fiber monitoring data, and to perform crack reconstruction on the laser scanning data according to the multi-source radial interpolation algorithm to obtain the laser scanning reconstruction map. The crack morphology waterfall map and the laser scanning reconstruction map formed from the cleaned distributed optical fiber monitoring data are fitted and recombined. Crack information features and location information features are extracted from the crack morphology waterfall map and the laser scanning reconstruction map, and encoded respectively to obtain crack feature vectors and location feature vectors.

[0176] The image fusion module is used to construct a feature fusion network model. The crack feature vector and the location feature vector are input into the feature fusion network model for fusion to obtain the fused feature representation. Based on the fused feature representation, image reconstruction is performed on the laser scanning reconstruction image to obtain the final crack morphology image.

[0177] The evaluation module is used to evaluate the hydraulic fractures in the final fracture morphology image using evaluation indicators, and obtain the evaluation results.

[0178] The data processing module includes a spatial alignment unit, which performs matrix eigenvalue decomposition on the coordinate system of the distributed optical fiber monitoring data and the coordinate system of the laser scanning data, calculates the spatial transformation matrix, calculates the translation vector using the spatial transformation matrix, the centroid of the distributed optical fiber monitoring data and the centroid of the laser scanning data, and uses an iterative calculation method to align the distributed optical fiber monitoring data and the laser monitoring data in space.

[0179] The data processing module also includes a spatial transformation matrix processing unit. This unit considers the difference in the number of distributed fiber optic monitoring data points and laser scanning data points, and, based on nearest neighbor matching, finds a corresponding laser scanning point for each distributed fiber optic monitoring point, resulting in multiple matching point pairs. Based on each matching point pair, it extracts the coordinates of the fiber optic monitoring point and the laser scanning point. It then constructs a relation matrix using the coordinate matrices of the fiber optic monitoring data and the laser scanning data, performing centering processing on both to obtain the centered coordinate matrices of the distributed fiber optic data and the centered coordinate matrices of the laser scanning data. It extracts the centered coordinates of the matched fiber optic monitoring points and the centered coordinates of the laser scanning points. Finally, it combines these coordinates into a centered relation matrix, where each row contains the centered fiber optic monitoring coordinates and the centered laser scanning coordinates of a matching point pair. Singular value decomposition is performed on the centered relation matrix to obtain the decomposition results, and the spatial transformation matrix is ​​determined based on these results.

[0180] The data processing module also includes an iteration unit, which initializes the spatial transformation matrix and translation vector. In each iteration, based on the current spatial transformation matrix and translation vector, the distributed optical fiber monitoring data is transformed into the coordinate space of the laser scanning data to obtain the transformed distributed optical fiber monitoring data. The root mean square error between the transformed distributed optical fiber monitoring data and the laser scanning data is calculated. The gradient descent optimization algorithm is used, and the parameters of the current spatial transformation matrix and translation vector are adjusted according to the root mean square error to obtain a new spatial transformation matrix and translation vector, which replaces the current spatial transformation matrix and translation vector. The iteration termination condition is set, and finally the optimized spatial transformation matrix and translation vector are obtained.

[0181] The feature fusion network model includes an input layer, a feature extraction layer, a fusion layer, and an output layer. The input layer is used to input crack feature vectors and location feature vectors. The feature extraction layer uses convolutional layers to extract local features from the crack feature vectors and location feature vectors. The fusion layer fuses the local features extracted by the feature extraction layer to obtain a fused feature vector. The output layer outputs the fused feature representation.

[0182] The hydraulic fracture evaluation system and method based on multi-source data provided in this invention are based on the same inventive concept and have the same beneficial effects, and will not be described again here.

[0183] Another embodiment of the present invention provides an electronic device, which includes a processor, an input device, an output device, and a memory. The processor, the input device, the output device, and the memory are interconnected. The memory is used to store a computer program, which includes program instructions. The processor is configured to call the program instructions to execute the method described in the first embodiment above.

[0184] It should be understood that, in the embodiments of the present invention, the processor may be a Central Processing Unit (CPU), but it may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0185] Input devices may include touchpads, fingerprint sensors (used to collect the user's fingerprint information and fingerprint orientation information), microphones, etc., while output devices may include displays (LCDs, etc.), speakers, etc.

[0186] The memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of the memory may also include non-volatile random access memory. For example, the memory may also store information about the device type.

[0187] In specific implementations, the processor, input device, and output device described in the embodiments of the present invention can execute the implementation of the method embodiments described in the embodiments of the present invention, or they can execute the implementation of the system embodiments described in the embodiments of the present invention, which will not be repeated here.

[0188] The present invention also provides an embodiment of a computer-readable storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, cause the processor to perform the method described in the first embodiment above.

[0189] The computer-readable storage medium can be an internal storage unit of the terminal described in the foregoing embodiments, such as the terminal's hard drive or memory. The computer-readable storage medium can also be an external storage device of the terminal, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the terminal. Furthermore, the computer-readable storage medium can include both internal storage units and external storage devices of the terminal. The computer-readable storage medium is used to store the computer program and other programs and data required by the terminal. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0190] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0191] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the terminals and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0192] In the several embodiments provided in this application, it should be understood that the disclosed terminals and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices or units, or may be electrical, mechanical or other forms of connection.

[0193] 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 or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for evaluating hydraulic fractures based on multi-source data, characterized in that, Includes the following steps: Acquire distributed fiber optic monitoring data and laser scanning data in fracturing physical simulation experiments; The distributed fiber optic monitoring data and laser scanning data are spatially aligned to obtain the aligned data. The specific method for spatially aligning distributed optical fiber monitoring data and laser scanning data includes: The coordinate systems of distributed optical fiber monitoring data and laser scanning data are subjected to matrix eigenvalue decomposition, and the spatial transformation matrix is ​​calculated. The translation vector is calculated using the spatial transformation matrix, the centroid of the distributed optical fiber monitoring data, and the centroid of the laser scanning data. An iterative calculation method is used to spatially align distributed fiber optic monitoring data and laser monitoring data; The method for calculating the spatial transformation matrix by performing matrix eigenvalue decomposition on the coordinate systems of distributed optical fiber monitoring data and laser scanning data includes: Considering the difference between the number of distributed fiber optic monitoring data points and the number of laser scanning data points, based on nearest neighbor matching, a corresponding laser scanning point is found for each distributed fiber optic monitoring point, resulting in multiple matching point pairs. Based on each pair of matching points, extract the coordinate matrix of the fiber optic monitoring point and the coordinate matrix of the laser scanning point; A relationship matrix is ​​constructed using the coordinate matrix of fiber optic monitoring data and the coordinate matrix of laser scanning data. The distributed fiber optic monitoring data and laser scanning data are then centralized to obtain the coordinate matrix of the distributed fiber optic data and the coordinate matrix of the laser scanning data after centralization. Extract the centered coordinates of the matched fiber optic monitoring point coordinates and the centered coordinates of the laser scanning point coordinates; The centralized coordinates of the fiber optic monitoring point and the centralized coordinates of the laser scanning point are combined into a centralized relation matrix. Each row of the centralized relation matrix contains the centralized fiber optic monitoring coordinates and the centralized laser scanning coordinates of a matching point pair. Singular value decomposition is performed on the centered relation matrix to obtain the decomposition result, and the spatial transformation matrix is ​​determined based on the decomposition result. The specific method for spatially aligning distributed optical fiber monitoring data and laser monitoring data using iterative calculation includes: Initialize the spatial transformation matrix and translation vector. In each iteration, based on the current spatial transformation matrix and the current translation vector, transform the distributed optical fiber monitoring data into the coordinate space of the laser scanning data to obtain the transformed distributed optical fiber monitoring data. Calculate the root mean square error between the converted distributed fiber optic monitoring data and the laser scanning data; The gradient descent optimization algorithm is used, and the parameters of the current spatial transformation matrix and the current translation vector are adjusted according to the root mean square error to obtain a new spatial transformation matrix and translation vector, which are then used to replace the current spatial transformation matrix and the current translation vector. By setting the iteration termination condition, the optimized spatial transformation matrix and translation vector are finally obtained. The aligned distributed optical fiber monitoring data is cleaned to obtain cleaned distributed optical fiber monitoring data. The laser scanning data is coupled with a multi-source radial interpolation algorithm to reconstruct the crack, resulting in a laser scanning reconstruction image; The crack morphology waterfall map and laser scan reconstruction map formed from the cleaned distributed optical fiber monitoring data are fitted and reconstructed. Crack information features and location information features are extracted from the crack morphology waterfall map and laser scan reconstruction map, and encoded separately to obtain crack feature vectors and location feature vectors. A feature fusion network model is constructed, and the crack feature vector and the location feature vector are input into the feature fusion network model for fusion to obtain the fused feature representation. Based on the fused feature representation, image reconstruction is performed on the laser scanning reconstruction map to obtain the final fused image. The final fused image was evaluated for hydraulic fractures using evaluation metrics, and the evaluation results were obtained.

2. The hydraulic fracture evaluation method based on multi-source data as described in claim 1, characterized in that, The feature fusion network model includes an input layer, a feature extraction layer, a fusion layer, and an output layer. The input layer is used to input crack feature vectors and location feature vectors. The feature extraction layer uses convolutional layers to extract local features of the crack feature vectors and location feature vectors. The fusion layer fuses the local features extracted by the feature extraction layer to obtain a fused feature vector. The output layer outputs the fused feature representation.

3. The hydraulic fracture evaluation method based on multi-source data as described in claim 1, characterized in that, The evaluation metrics include image sharpness, contrast, information entropy, and structural similarity.

4. A hydraulic fracture evaluation system based on multi-source data, characterized in that, To implement the method as described in any one of claims 1-3, the system comprises: a data acquisition module, a data processing module, an image fusion module, and an evaluation module. The data acquisition module is used to acquire distributed fiber optic monitoring data and laser scanning data in the fracturing physical simulation experiment; The data processing module is used to clean the aligned distributed optical fiber monitoring data to obtain cleaned distributed optical fiber monitoring data, and to perform crack reconstruction on the laser scanning data according to the multi-source radial interpolation algorithm to obtain the laser scanning reconstruction map. The crack morphology waterfall map and the laser scanning reconstruction map formed from the cleaned distributed optical fiber monitoring data are fitted and recombined. Crack information features and location information features are extracted from the crack morphology waterfall map and the laser scanning reconstruction map, and encoded respectively to obtain crack feature vectors and location feature vectors. The image fusion module is used to construct a feature fusion network model. The crack feature vector and the location feature vector are input into the feature fusion network model for fusion to obtain the fused feature representation. Based on the fused feature representation, image reconstruction is performed on the laser scanning reconstruction image to obtain the final crack morphology image. The evaluation module is used to evaluate the hydraulic fractures in the final fracture morphology image using evaluation indicators, and obtain the evaluation results. The data processing module includes a spatial alignment unit, which is used to perform matrix eigenvalue decomposition on the coordinate system of the distributed optical fiber monitoring data and the coordinate system of the laser scanning data, and calculate the spatial transformation matrix; calculate the translation vector using the spatial transformation matrix, the centroid of the distributed optical fiber monitoring data and the centroid of the laser scanning data; and use an iterative calculation method to align the distributed optical fiber monitoring data and the laser monitoring data in space.

5. An electronic device, characterized in that, include: The system includes a processor, an input device, an output device, and a memory, which are interconnected. The memory is used to store a computer program, which includes program instructions. The processor is configured to invoke the program instructions to execute the method as described in any one of claims 1-3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the method as described in any one of claims 1-3.

Images