Rock mass permeability coefficient reliability evaluation method and system based on multi-source data

Abstract

This invention relates to a method and system for evaluating the reliability of rock mass permeability coefficients based on multi-source data. The invention utilizes pre-established physical correspondences, taking fracture geometric and integrity characteristic parameters as input variables, to deduce the expected permeability range of the test rock mass section under current rock mass integrity and fracture development conditions. This constructs a theoretical reference benchmark for rock mass permeability characteristics based on the intrinsic physical properties of the rock mass itself, completely independent of the actual pressure water test results. Furthermore, by comparing the actual permeability characteristic parameters measured by the pressure water test with the aforementioned expected permeability range, the degree of contradiction between the two is detected. Thus, without any external construction intervention, relying solely on the inherent logical consistency between the multi-source physical information of the rock mass itself, it is possible to objectively and quantitatively determine whether the pressure water test data truly reflects the rock mass permeability characteristics.

Description

Technical Field

[0001] This invention relates to the field of geotechnical engineering investigation and hydrogeological testing technology, specifically to a method and system for evaluating the reliability of rock mass permeability coefficient based on multi-source data. Background Technology

[0002] Water pressure testing is a primary in-situ testing method for obtaining the permeability coefficient and evaluating the permeability characteristics of rock masses. Its results serve as a direct basis for seepage prevention design and seepage analysis in water conservancy and hydropower projects. However, water pressure test results are subject to various factors, including experimental operation and the heterogeneity of the rock mass, resulting in a degree of uncertainty. Therefore, verifying the reliability of the water pressure test results is a prerequisite for ensuring the safety of subsequent engineering analysis and design.

[0003] Currently, the industry has begun to explore the use of integrated geophysical exploration methods for rock mass quality evaluation. For example, in engineering practices such as the Yinjiang Hydropower Station, there have been cases where single-hole acoustic wave methods, borehole panoramic imaging, and water pressure tests have been combined to evaluate the construction quality of consolidation grouting and curtain grouting. These methods effectively verify the treatment effect of grouting projects by comparing changes in rock mass wave velocity before and after grouting, observing the filling of cement-aggregate in borehole wall fissures, and detecting the permeability of the rock mass after grouting.

[0004] However, the evaluation object of the aforementioned existing technologies is the engineering effect, and its logic is based on a comparative analysis before and after grouting. This method does not address the reliability of the original water pressure test data itself. In actual exploration, engineers still face an unresolved problem: how to determine whether the obtained water pressure test data truly reflects the permeability characteristics of the rock mass without any external construction intervention. For example, when a test section has a high acoustic velocity, indicating rock mass integrity, but the water pressure test shows strong permeability, engineers find it difficult to determine whether this is due to genuine permeability caused by localized hidden fractures or false data caused by problems such as embolization failure during the test. Currently, there is a lack of a systematic method that deeply integrates acoustic testing, borehole television, and water pressure testing in terms of physical logic, and uses the inherent contradictions among the three to reverse-verify the reliability of the water pressure test results. Summary of the Invention

[0005] The main objective of this invention is to provide a method and system for evaluating the reliability of rock mass permeability coefficient based on multi-source data, in order to solve the technical problem of how to determine whether the currently obtained pressure water test data truly reflects the permeability characteristics of the rock mass without any external construction intervention.

[0006] To achieve the above objectives, this invention provides a method for evaluating the reliability of rock mass permeability coefficient based on multi-source data, comprising the following steps: S1. Obtain water pressure test data, borehole image data, and acoustic test data for the same rock mass test section within the same depth range of the same borehole. S2. Extract a first characteristic parameter reflecting the permeability characteristics of the rock mass from the water pressure test data, extract a second characteristic parameter reflecting the geometric characteristics of the fracture from the borehole image data, and extract a third characteristic parameter reflecting the integrity characteristics of the rock mass from the acoustic test data. S3. Based on the pre-established physical correspondence between rock mass integrity characteristics, fracture geometry characteristics and rock mass permeability characteristics, determine the expected permeability range of the rock mass test section under the current rock mass integrity conditions and fracture development conditions according to the second characteristic parameter and the third characteristic parameter. S4. Compare and verify the measured permeability characteristics characterized by the first feature parameter with the expected permeability range to obtain the comparison results, and generate the reliability evaluation results of the pressure water test data of the rock mass test section based on the comparison results.

[0007] Furthermore, step S1 specifically includes the following steps: The test data were obtained by in-situ water pressure test device in the same borehole for the same rock mass test section, the borehole image data were obtained by continuously photographing the borehole wall before and after the water pressure test by borehole television imaging device, and the longitudinal wave velocity data were obtained by acoustic wave testing device at preset point distances in the rock mass test section. The water pressure test data, the borehole image data, and the longitudinal wave velocity data have the same depth calibration benchmark, so that the first characteristic parameter, the second characteristic parameter, and the third characteristic parameter all correspond to the same rock mass test section.

[0008] Further, step S2 involves extracting a second feature parameter reflecting the geometric characteristics of the fracture from the image data within the borehole, including the following steps: The second characteristic parameter includes the equivalent fracture aperture and the number of volumetric fractures in the rock mass; The image data inside the hole is input into a trained deep learning network, and the deep learning network performs crack instance segmentation on the image data inside the hole, outputting a pixel-level mask for each crack. The average width, filling coefficient, and connectivity coefficient of each fracture are calculated based on the pixel-level mask. The equivalent fracture aperture is calculated based on the average width, filling coefficient, and connectivity coefficient of each fracture, and the number of fractures per unit volume of rock mass is obtained by counting the number of fractures in the rock mass.

[0009] More preferably, the physical correspondence in step S3 includes a pre-constructed first correspondence table and a second correspondence table. The first correspondence table records the correspondence between the rock mass integrity index and the permeability, and the second correspondence table records the correspondence between the equivalent fracture aperture, the number of volumetric fractures in the rock mass, and the permeability.

[0010] More preferably, determining the expected permeability range in step S3 includes the following steps: The third characteristic parameter includes the rock mass integrity index; Using the equivalent fracture aperture and rock mass volume fracture number in the second characteristic parameter, and the rock mass integrity index in the third characteristic parameter as input variables, dual-parameter interpolation is performed based on the first correspondence table and the second correspondence table to calculate the expected permeability value of the rock mass test section under the current rock mass integrity conditions and fracture development conditions; Based on the expected permeability value, and in conjunction with a preset range coefficient, the expected permeability range is determined.

[0011] More preferably, step S3 further includes the following steps: Construct comprehensive fracture parameters, wherein the comprehensive fracture parameters are the product of the equivalent fracture aperture and the number of fractures in the rock mass volume; The dual-parameter interpolation uses the rock mass integrity index and the fracture comprehensive parameter as two input variables, extracts interpolation base points from the first correspondence table and the second correspondence table, and calculates the expected permeability value through bilinear interpolation.

[0012] Furthermore, step S4 specifically includes the following steps: First-level verification: Obtain the pre-set suspiciousness determination rules, and determine whether the first feature parameter, the second feature parameter, and the third feature parameter meet the suspiciousness determination rules; if so, determine that there is a contradiction between the first feature parameter, the second feature parameter, and the third feature parameter, and mark the rock mass test section as highly suspicious; Second layer of verification: For rock mass test sections that fail the first layer of verification and are marked as highly suspicious, the measured permeability characteristics are compared with the expected permeability characteristic range. When the measured permeability characteristics fall within the expected permeability characteristic range, the rock mass test section is determined to have passed the verification. When the measured permeability characteristics deviate from the expected permeability characteristic range, the rock mass test section is determined to have low reliability. The third layer of verification involves inputting the first feature parameter, the second feature parameter, and the third feature parameter into a pre-trained comprehensive rating classification model. The comprehensive rating classification model then outputs the probability that the rock mass test section belongs to a high reliability level, a medium reliability level, or a low reliability level, and the final reliability level is determined based on the highest probability.

[0013] More preferably, the comprehensive rating classification model employs five-fold cross-validation during the training phase, specifically including the following steps: Historical sample data labeled with reliability levels were randomly divided into five data subsets. One data subset was used as the validation set and the other four data subsets were used as the training set for five rounds of training and validation. The average performance index of the five rounds of validation was used as the performance evaluation result of the comprehensive rating classification model.

[0014] Furthermore, the following steps are included after step S4: S5. Based on the reliability evaluation results, in the borehole columnar section or three-dimensional geological model, the rock mass test section is marked with different colors according to the reliability level, and suggestions for retesting or intensified testing are generated for the rock mass test section with low reliability level.

[0015] This invention also provides a rock mass permeability coefficient reliability evaluation system based on multi-source data, which applies the rock mass permeability coefficient reliability evaluation method based on multi-source data as described above, including: The data acquisition module is used to acquire water pressure test data, borehole image data, and acoustic test data collected for the same rock mass test section within the same depth range of the same borehole. The feature extraction module is used to extract a first feature parameter reflecting the permeability characteristics of the rock mass from the water pressure test data, a second feature parameter reflecting the geometric characteristics of the fracture from the borehole image data, and a third feature parameter reflecting the integrity characteristics of the rock mass from the acoustic test data. The reliability evaluation module is used to determine the expected permeability range of the rock mass test section under the current rock mass integrity conditions and fracture development conditions based on the pre-established physical correspondence between rock mass integrity characteristics, fracture geometry characteristics and rock mass permeability characteristics, according to the second characteristic parameter and the third characteristic parameter; and to compare and verify the measured permeability characteristics characterized by the first characteristic parameter with the expected permeability range, obtain the comparison result, and generate the reliability evaluation result of the pressure water test data of the rock mass test section based on the comparison result.

[0016] Compared with the prior art, the present invention has the following beneficial effects: In this invention, the water pressure test data, borehole image data, and acoustic wave test data all originate from the same test section of the same rock mass in the same borehole, ensuring a strict spatial correspondence for subsequent cross-validation of multi-source data. Utilizing a pre-established physical correspondence, and using fracture geometric and integrity characteristic parameters as input variables, the expected permeability range of the test section under current rock mass integrity and fracture development conditions is derived in reverse. This constructs a theoretical reference benchmark for rock mass permeability characteristics, completely independent of the actual water pressure test results and based on the intrinsic physical properties of the rock mass itself. Furthermore, by comparing the actual permeability characteristic parameters from the water pressure test with the aforementioned expected permeability range, the degree of contradiction between the two is detected. This allows for objective and quantitative determination of whether the water pressure test data truly reflects the rock mass permeability characteristics, effectively distinguishing between the true permeability response of the rock mass and false data caused by deviations in the test process, without any external construction intervention and solely relying on the inherent logical consistency between the multi-source physical information of the rock mass itself. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a method for evaluating the reliability of rock mass permeability coefficient based on multi-source data, according to an embodiment of the present invention.

[0019] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0021] 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 a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0022] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0023] Furthermore, the use of terms such as "first" and "second" in this invention is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Additionally, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed by this invention.

[0024] The following embodiments aim to address the challenges of existing rock mass permeability evaluation technologies, such as the reliance on single methods for verifying the reliability of pressure water test results, strong subjectivity, and a lack of collaborative physical-logical analysis based on multi-source data. Specifically, the objectives are: to overcome the uncertainty of single pressure water test data and provide a systematic reliability evaluation method integrating borehole television and sonic testing data; to establish and quantify the physical correspondence between rock mass integrity, fracture development degree, and permeability, ensuring it conforms to engineering specifications and empirical knowledge, transforming implicit expert experience into explicit logical rules; and to automate and intelligently classify the credibility of the results from each pressure water test section, providing engineers with clear and quantifiable decision-making basis for data adoption, enhanced testing, or the use of other methods, thereby improving the reliability of engineering survey results.

[0025] Please see Figure 1 This embodiment proposes a method for evaluating the reliability of rock mass permeability coefficient based on multi-source data, including the following steps: S1. Obtain water pressure test data, borehole image data, and acoustic test data for the same rock mass test section within the same depth range of the same borehole. Furthermore, step S1 specifically includes the following steps: The test data were obtained by in-situ water pressure test device in the same borehole for the same rock mass test section, the borehole image data were obtained by continuously photographing the borehole wall before and after the water pressure test by borehole television imaging device, and the longitudinal wave velocity data were obtained by acoustic wave testing device at preset point distances in the rock mass test section. The water pressure test data, the borehole image data, and the longitudinal wave velocity data have the same depth calibration benchmark, so that the first characteristic parameter, the second characteristic parameter, and the third characteristic parameter all correspond to the same rock mass test section.

[0026] Specifically, in the same borehole, water pressure test, in-hole television imaging and acoustic wave test are carried out in sequence according to the following methods to ensure the spatial correspondence of the data.

[0027] I. Water pressure test: Test standards: Strictly in accordance with the "Specifications for Borehole Pressure Water Test in Water Conservancy and Hydropower Engineering" (SL 31-2003). A three-level pressure, five-stage cyclic test method is adopted.

[0028] Key parameters: test pressure P (0.3~1.0MPa), injection flow rate Q (L / min), test section length L (typically 5m).

[0029] Output index: permeability q, calculated using the following formula: The unit is Lu; the pressure-flow curve, i.e., the PQ curve; the permeability coefficient K (cm / s).

[0030] II. Drilling television imaging: Before and after the water pressure test, borehole television imaging equipment was used to probe the same borehole section as the water pressure test.

[0031] Equipment requirements: High-definition drilling television camera, with a resolution of no less than 1920×1080 and a viewing angle of no less than 150°.

[0032] Acquisition method: uniform lifting speed ≤3cm / s, continuous shooting to ensure image distortion-free images.

[0033] Output data: 360° borehole wall unfolded image, intelligent recognition results of structural surfaces, including location, depth, orientation, width and filling status.

[0034] III. Sound Wave Testing: Single-hole or cross-hole acoustic wave testers were used to obtain the acoustic wave parameters of the rock mass in the same borehole section.

[0035] Equipment requirements: Single-hole acoustic wave tester, transmission frequency 10~50kHz, sampling frequency ≥1MHz.

[0036] Acquisition parameters: point spacing 0.2~0.5m, transmission interval 0.5~1.0m.

[0037] Output Specification: P-wave velocity (m / s), shear wave velocity (m / s).

[0038] S2. Extract a first characteristic parameter reflecting the permeability characteristics of the rock mass from the water pressure test data, extract a second characteristic parameter reflecting the geometric characteristics of the fracture from the borehole image data, and extract a third characteristic parameter reflecting the integrity characteristics of the rock mass from the acoustic test data.

[0039] Specifically, the collected multi-source data is processed to extract feature parameters for reliability evaluation.

[0040] I. Data Processing of Water Pressure Test: a) Analyze the PQ curve type, which includes laminar flow, turbulent flow, expansion, erosion, and filling types (E), and eliminate test sections with obvious abnormalities; b) Obtain the permeability q and permeability coefficient K of the test section; Extracting the permeation feature vector: ; in Encoding for PQ curve type.

[0041] II. Processing of image data inside the hole: Step S2 involves extracting a second feature parameter reflecting the geometric characteristics of the fracture from the image data inside the borehole, including the following steps: The second characteristic parameter includes the equivalent fracture aperture and the number of volumetric fractures in the rock mass; The image data inside the hole is input into a trained deep learning network, and the deep learning network performs crack instance segmentation on the image data inside the hole, outputting a pixel-level mask for each crack. The average width, filling coefficient, and connectivity coefficient of each fracture are calculated based on the pixel-level mask. The equivalent fracture aperture is calculated based on the average width, filling coefficient, and connectivity coefficient of each fracture, and the number of fractures per unit volume of rock mass is obtained by counting the number of fractures in the rock mass.

[0042] Specifically, the Mask R-CNN deep learning network is used to automatically identify cracks and extract parameters from the unfolded images of the borehole image data. The specific process is as follows: a) Image preprocessing; The original images of the in-hole image data are normalized for brightness and enhanced for contrast to eliminate the effects of uneven illumination; geometric correction is performed to ensure accurate correspondence between depth and image position.

[0043] b) Crack instance segmentation; The preprocessed image is input into the trained Mask R-CNN network, which uses ResNet-101 as the backbone network and combines a Feature Pyramid Network (FPN) to extract multi-scale features. The network outputs the bounding box, confidence score, and pixel-level mask for each crack instance. A confidence threshold of ≥0.85 is set, and the recognition result is retained when the confidence score meets the confidence threshold.

[0044] c) Calculation of fracture geometric parameters; Crack location: determined by the depth coordinates of the center point of the pixel-level mask; Crack width: The pixel width within the pixel-level mask is statistically analyzed along the crack normal direction and converted into physical width (mm) based on the image resolution. The average and maximum values ​​are taken for cracks with varying widths. Crack length: The pixel length of the pixel-level mask along the crack direction, converted into physical length (m). Equivalent fracture aperture Taking into account a comprehensive index of crack width, filling degree, and connectivity, the calculation formula is as follows: ; in, is the average width (mm) of the i-th crack; The filling coefficient is determined based on image recognition to assess the filling degree, and its value ranges from 0 to 1. Let be the connectivity coefficient of the i-th fracture, determined based on the fracture extension length; Number of volumetric fractures in rock mass Number of fractures per unit volume of rock mass (fractures / m) 3 ); Connectivity Index : A quantitative index reflecting the connectivity of a fracture network, with a value ranging from 0 to 1, and the calculation formula is: ; in, The number of fractures that intersect with at least one other fracture; This represents the total number of cracks. The total length of the cracks involved in the cross-cutting; The total length of all cracks is automatically obtained through image topology analysis of the pixel-level mask of the cracks segmented by MaskR-CNN. Fracture orientation: Combining borehole azimuth and depth information, the dip angle of the dominant fracture is inverted from the fracture morphology in the unfolded image. and dominant fracture tendency .

[0045] d) Feature vector output; The extracted crack geometric features are combined into a feature vector: ; III. Processing of Acoustic Wave Test Data: a) Based on the longitudinal wave velocity Calculate the rock mass integrity index: ; in, The longitudinal wave velocity of the intact rock block is determined by indoor rock physics and mechanics tests. If no actual measurement conditions are available, the empirical values ​​of the wave velocity of intact rock blocks of various types of rocks can be referred to, or the highest value of the acoustic wave velocity in the borehole can be taken as an approximate substitute.

[0046] For all measuring points within the test section The values ​​were statistically analyzed to obtain the average completeness index. and minimum integrity index .

[0047] b) Calculate the average P-wave velocity within the test section : ; in, This represents the total number of acoustic measurement points within the test section. Let be the longitudinal wave velocity at the i-th measuring point.

[0048] c) Calculate the wave speed ratio ; d) Extracting acoustic feature vectors: .

[0049] S3. Based on the pre-established physical correspondence between rock mass integrity characteristics, fracture geometry characteristics and rock mass permeability characteristics, determine the expected permeability range of the rock mass test section under the current rock mass integrity conditions and fracture development conditions according to the second characteristic parameter and the third characteristic parameter. S4. Compare and verify the measured permeability characteristics characterized by the first feature parameter with the expected permeability range to obtain the comparison results, and generate the reliability evaluation results of the pressure water test data of the rock mass test section based on the comparison results.

[0050] In this embodiment, the water pressure test data, borehole image data, and acoustic wave test data are all derived from the same test section of the same rock mass in the same borehole. This ensures that the subsequent cross-validation of multi-source data is strictly spatially correlated. Using the pre-established physical correspondence, with the geometric characteristic parameters of fractures and the characteristic parameters of rock integrity as input variables, the expected range of permeability characteristics that the test section of rock mass should exhibit under the current rock mass integrity conditions and fracture development conditions is deduced in reverse. This constructs a theoretical reference benchmark for permeability characteristics that is completely independent of the actual water pressure test results and is based on the intrinsic physical properties of the rock mass itself. Furthermore, by comparing and verifying the actual permeability characteristic parameters of the water pressure test with the above-mentioned expected permeability characteristic range, the degree of contradiction between the two is detected. Thus, without any external construction intervention, relying solely on the inherent logical consistency between the multi-source physical information of the rock mass itself, it is possible to objectively and quantitatively determine whether the water pressure test data truly reflects the permeability characteristics of the rock mass, effectively distinguishing between the true permeability response of the rock mass and false data caused by deviations in the test process.

[0051] In one embodiment, the physical correspondence in step S3 includes a pre-constructed first correspondence table and a second correspondence table. The first correspondence table records the correspondence between the rock mass integrity index and the permeability, and the second correspondence table records the correspondence between the equivalent fracture aperture, the number of volumetric fractures in the rock mass, and the permeability.

[0052] As a further preferred embodiment, step S3, determining the expected permeability range, includes the following steps: The third characteristic parameter includes the rock mass integrity index; Using the equivalent fracture aperture and rock mass volume fracture number in the second characteristic parameter, and the rock mass integrity index in the third characteristic parameter as input variables, dual-parameter interpolation is performed based on the first correspondence table and the second correspondence table to calculate the expected permeability value of the rock mass test section under the current rock mass integrity conditions and fracture development conditions; Based on the expected permeability value, and in conjunction with a preset range coefficient, the expected permeability range is determined.

[0053] More preferably, step S3 further includes the following steps: Construct comprehensive fracture parameters, wherein the comprehensive fracture parameters are the product of the equivalent fracture aperture and the number of fractures in the rock mass volume; The dual-parameter interpolation uses the rock mass integrity index and the fracture comprehensive parameter as two input variables, extracts interpolation base points from the first correspondence table and the second correspondence table, and calculates the expected permeability value through bilinear interpolation.

[0054] In this embodiment, step S4 specifically includes the following steps: First-level verification: Obtain the pre-set suspiciousness determination rules, and determine whether the first feature parameter, the second feature parameter, and the third feature parameter meet the suspiciousness determination rules; if so, determine that there is a contradiction between the first feature parameter, the second feature parameter, and the third feature parameter, and mark the rock mass test section as highly suspicious; Second layer of verification: For rock mass test sections that fail the first layer of verification and are marked as highly suspicious, the measured permeability characteristics are compared with the expected permeability characteristic range. When the measured permeability characteristics fall within the expected permeability characteristic range, the rock mass test section is determined to have passed the verification. When the measured permeability characteristics deviate from the expected permeability characteristic range, the rock mass test section is determined to have low reliability. The third layer of verification involves inputting the first feature parameter, the second feature parameter, and the third feature parameter into a pre-trained comprehensive rating classification model. The comprehensive rating classification model then outputs the probability that the rock mass test section belongs to a high reliability level, a medium reliability level, or a low reliability level, and the final reliability level is determined based on the highest probability.

[0055] Specifically, this embodiment establishes a three-tiered progressive verification rule system to achieve reliability evaluation from rapid screening to detailed analysis and then to comprehensive rating. The verification rule system is based on a pre-established physical correspondence table: Table 1 serves as the first correspondence table, and Table 2 serves as the second correspondence table. These tables are derived from existing engineering specifications and statistical regression of multiple engineering measured data.

[0056] Table 1 Rock Mass Integrity Index Table of correspondence between permeability q and permeability Table 2. Relationship between fracture geometry characteristics and permeability q Based on the above physical correspondence table, a three-layer verification model is constructed: First layer: Rapid screening of basic contradictions: When the rock mass integrity index is greater than the first threshold and the permeability is greater than the second threshold, or when the rock mass integrity index is less than the third threshold and the permeability is less than the second threshold, the rock mass test section is marked as highly suspicious. Alternatively, when the equivalent fracture aperture is less than the fourth threshold, the number of volumetric fractures in the rock mass is less than the fifth threshold, and the permeability is greater than the second threshold, or when the equivalent fracture aperture is greater than the sixth threshold, the number of volumetric fractures in the rock mass is greater than the seventh threshold, and the permeability is less than the eighth threshold, the rock mass test section is marked as highly suspicious.

[0057] Specifically, test sections with obvious physical contradictions are quickly identified using the above rules and marked as "highly suspicious".

[0058] Rule 1 (Integrity-Permeability Contradiction): If and If so, it is marked as "highly suspicious"; if and If so, it will be marked as "highly suspicious"; Rule 2 (Fracture-Permeability Contradiction): If and and ,or and and If it is, then it is marked as "highly suspicious".

[0059] Second layer: Detailed parameter relationship verification: Based on the physical correspondence established in Tables 1 and 2, this layer constructs a function for the expected permeability range, thereby enabling refined quantitative verification of the results of the water pressure test.

[0060] a) Calculation of comprehensive fracture parameters: To comprehensively reflect the influence of fracture aperture and the number of volumetric fractures on permeability, a comprehensive fracture parameter is defined. .

[0061] ; b) Calculation of expected permeability value: Based on the discrete correspondence between Tables 1 and 2, the expected permeability is constructed using a two-parameter interpolation method. Functions: ; For any measured point The expected permeability value is calculated using bilinear interpolation: Interpolation base points are extracted from Tables 1 and 2 to construct... The three-dimensional discrete point set is calculated using bilinear interpolation.

[0062] It should be noted that: in this embodiment, permeability is used as the permeability characteristic of the rock mass, and the reasonable range of expected permeability is used as the expected range of permeability characteristics.

[0063] c) Determining the reasonable range of expected permeability: Considering the natural variability of the rock mass and testing errors, a reasonable range for the expected permeability is defined: ; Wherein, λ and μ are range coefficients, determined according to the engineering grade and exploration stage. This embodiment recommends using: λ=0.5, μ=2.0, that is, the reasonable range of expected permeability is 0.5 to 2.0 times the expected permeability value.

[0064] d) Fine-grained validation rules: The measured permeability value q is compared with the reasonable range of the expected permeability. Comparison: like If the test section passes the second layer verification, the results of the water pressure test are basically consistent with the geometric characteristics and integrity of the rock mass fractures; if If the permeability is abnormally low, it indicates possible that the cracks are filled, the test section is partially blocked, or the pressure is not sufficiently stable. If the permeability is abnormally high, it is considered an anomaly, indicating possible concentrated seepage channels, failure of plugging for water sealing, or disturbance of the rock mass due to construction. For test sections judged as abnormal (too high or too low), they are marked as "low reliability," and technicians are advised to conduct a comprehensive assessment in conjunction with the results of the first layer of screening.

[0065] Third layer: Multi-source data collaborative comprehensive rating: This layer integrates all extracted features and uses a random forest classification model for comprehensive rating, outputting the reliability probability of each test segment.

[0066] a) Feature fusion: The feature vectors extracted in step 2 are concatenated to form a comprehensive feature vector. : ; in, For the penetration feature vector, The geometric feature vector of the crack. This is the characteristic vector of the sound wave.

[0067] b) Model training: Random Forest was adopted as the comprehensive rating classification model. Random Forest is a machine learning algorithm based on the Bagging ensemble strategy, which achieves classification by constructing multiple decision trees and integrating their voting results.

[0068] This embodiment uses 150 decision trees to construct a random forest model. At each node split, a random selection is made. We consider n features (where n is the total number of features) to enhance model diversity. We use five-fold cross-validation to evaluate model performance and ensure the model's robustness and generalization ability.

[0069] The five-fold cross-validation specifically includes the following steps: Historical sample data labeled with reliability levels were randomly divided into five data subsets. One data subset was used as the validation set and the other four data subsets were used as the training set for five rounds of training and validation. The average performance index of the five rounds of validation was used as the performance evaluation result of the comprehensive rating classification model.

[0070] c) Tag definition: Based on expert experience and historical data, reliability is categorized into three types: Category I (High Reliability): The geometric characteristics of fractures, rock mass integrity index and permeability are highly matched, and the data from multiple sources corroborate each other; Category II (Moderately Reliable): Some parameters have slight anomalies or uncertainties, but there are no fundamental contradictions; Category III (Low Reliability): There are obvious conflicts between multiple data sources, or the data is seriously inconsistent with the first and second level rules.

[0071] d) Training data preparation: Collect verified historical data from multiple projects, including: composite feature vectors. The final reliability label given by experts; the sample size should cover different lithologies and different engineering types to ensure the generalization ability of the model.

[0072] e) Reliability output: After training, the random forest model outputs the predicted probability for each level: ; The final reliability level is determined by the maximum probability. Simultaneously, the output probability value can be used as the confidence level of the evaluation result.

[0073] f) Model interpretability analysis: To enhance engineers' confidence in the model's decisions, eigenvalue importance analysis is employed to reveal the contribution of each input parameter to the reliability assessment. Based on the Mean Decrease Gini, eigenvalue importance can visually demonstrate the relative importance of parameters such as fracture geometry, rock mass integrity index, and permeability in the model's decision-making process.

[0074] Furthermore, the following steps are included after step S4: S5. Based on the reliability evaluation results, in the borehole columnar section or three-dimensional geological model, the rock mass test section is marked with different colors according to the reliability level, and suggestions for retesting or intensified testing are generated for the rock mass test section with low reliability level.

[0075] Specifically, a standardized "Rock Mass Permeability Coefficient Reliability Evaluation Report" is generated, which includes: Project information, borehole number, test section depth; Original results of the water pressure test (permeability, permeability coefficient); Various characteristic parameters (fracture aperture, number of volumetric fractures in rock mass, rock mass integrity index, etc.); Verification results at each layer (whether the first-layer contradiction is triggered, and whether the second-layer deviation from expectations); Final reliability class (I / II / III) and probability; Description of the main contradiction indicators.

[0076] Visual output: In the borehole column or cross-sectional diagram, the reliability of each test section is marked with colors: green (high reliability), yellow (medium reliability), and red (low reliability). Low-reliability sections are highlighted in the 3D geological model, and a recommendation is automatically generated: "Retest or intensified testing is recommended."

[0077] This invention also provides a rock mass permeability coefficient reliability evaluation system based on multi-source data, which applies the rock mass permeability coefficient reliability evaluation method based on multi-source data as described above, including: The data acquisition module is used to acquire water pressure test data, borehole image data, and acoustic test data collected for the same rock mass test section within the same depth range of the same borehole. The feature extraction module is used to extract a first feature parameter reflecting the permeability characteristics of the rock mass from the water pressure test data, a second feature parameter reflecting the geometric characteristics of the fracture from the borehole image data, and a third feature parameter reflecting the integrity characteristics of the rock mass from the acoustic test data. The reliability evaluation module is used to determine the expected permeability range of the rock mass test section under the current rock mass integrity conditions and fracture development conditions based on the pre-established physical correspondence between rock mass integrity characteristics, fracture geometry characteristics and rock mass permeability characteristics, according to the second characteristic parameter and the third characteristic parameter; and to compare and verify the measured permeability characteristics characterized by the first characteristic parameter with the expected permeability range, obtain the comparison result, and generate the reliability evaluation result of the pressure water test data of the rock mass test section based on the comparison result.

[0078] The results output module is used to generate reports and visualizations.

[0079] This invention achieves physical logic verification through deep fusion of source data: unlike simple data overlay, this invention establishes for the first time a quantitative correspondence table among rock mass integrity indicators, fracture geometry characteristics, and rock mass permeability, and constructs a logical reasoning chain from screening to verification based on this table. This gives the reliability evaluation a solid foundation in rock mechanics and hydrogeology, rather than a black-box statistical correlation, making the results more interpretable and credible.

[0080] This invention also significantly improves the objectivity and intelligence of the evaluation: by identifying cracks through deep learning, automatically calculating feature parameters, and using machine learning models for comprehensive rating, it greatly reduces the subjectivity and experience dependence of manual interpretation, and realizes the batch, standardization and intelligent processing of massive drilling data.

[0081] This invention directly addresses core engineering issues and is highly practical: it directly answers the question of "whether water pressure test data is reliable." The output results are clear and intuitive, directly guiding engineering practice. High-reliability data can be directly used in design; medium-reliability data requires comprehensive evaluation in conjunction with other data; and low-reliability data necessitates re-performing or re-intensifying the test, thus effectively mitigating engineering risks.

[0082] This invention fills a gap in technical standards and lays the foundation for intelligent geotechnical exploration: It provides a standardized set of operating tools that can be embedded into existing exploration processes, filling the current gap in the systematic and quantitative verification of water pressure test results. More importantly, it successfully transforms the implicit experience of senior engineers in judging the reliability of water pressure tests into explicit rules and algorithms, which is a key step in promoting the digital and intelligent transformation of geotechnical engineering exploration.

[0083] The present invention is open and has good scalability: the fusion framework and logical verification model proposed in this invention have good scalability and can be easily integrated with more new detection data in the future, such as borehole radar, nuclear magnetic resonance, optical imaging, etc., to form a more powerful and complete rock mass quality and parameter reliability evaluation system.

[0084] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for evaluating the reliability of rock mass permeability coefficient based on multi-source data, characterized in that, Includes the following steps: S1. Obtain water pressure test data, borehole image data, and acoustic test data for the same rock mass test section within the same depth range of the same borehole. S2. Extract a first characteristic parameter reflecting the permeability characteristics of the rock mass from the water pressure test data, extract a second characteristic parameter reflecting the geometric characteristics of the fracture from the borehole image data, and extract a third characteristic parameter reflecting the integrity characteristics of the rock mass from the acoustic test data. S3. Based on the pre-established physical correspondence between rock mass integrity characteristics, fracture geometry characteristics and rock mass permeability characteristics, determine the expected permeability range of the rock mass test section under the current rock mass integrity conditions and fracture development conditions according to the second characteristic parameter and the third characteristic parameter. S4. Compare and verify the measured permeability characteristics characterized by the first feature parameter with the expected permeability range to obtain the comparison results, and generate the reliability evaluation results of the pressure water test data of the rock mass test section based on the comparison results.

2. The method for evaluating the reliability of rock mass permeability coefficient based on multi-source data according to claim 1, characterized in that, Step S1 specifically includes the following steps: The test data were obtained by in-situ water pressure test device in the same borehole for the same rock mass test section, the borehole image data were obtained by continuously photographing the borehole wall before and after the water pressure test by borehole television imaging device, and the longitudinal wave velocity data were obtained by acoustic wave testing device at preset point distances in the rock mass test section. The water pressure test data, the borehole image data, and the longitudinal wave velocity data have the same depth calibration benchmark, so that the first characteristic parameter, the second characteristic parameter, and the third characteristic parameter all correspond to the same rock mass test section.

3. The method for evaluating the reliability of rock mass permeability coefficient based on multi-source data according to claim 1, characterized in that, Step S2 involves extracting a second feature parameter reflecting the geometric characteristics of the fracture from the image data inside the borehole, including the following steps: The second characteristic parameter includes the equivalent fracture aperture and the number of volumetric fractures in the rock mass; The image data inside the hole is input into a trained deep learning network, and the deep learning network performs crack instance segmentation on the image data inside the hole, outputting a pixel-level mask for each crack. The average width, filling coefficient, and connectivity coefficient of each fracture are calculated based on the pixel-level mask. The equivalent fracture aperture is calculated based on the average width, filling coefficient, and connectivity coefficient of each fracture, and the number of fractures per unit volume of rock mass is obtained by counting the number of fractures in the rock mass.

4. The method for evaluating the reliability of rock mass permeability coefficient based on multi-source data according to claim 3, characterized in that, The physical correspondence in step S3 includes a pre-constructed first correspondence table and a second correspondence table. The first correspondence table records the correspondence between the rock mass integrity index and the permeability, and the second correspondence table records the correspondence between the equivalent fracture aperture, the number of volumetric fractures in the rock mass, and the permeability.

5. The method for evaluating the reliability of rock mass permeability coefficient based on multi-source data according to claim 4, characterized in that, Step S3, determining the expected permeability range, includes the following steps: The third characteristic parameter includes the rock mass integrity index; Using the equivalent fracture aperture and rock mass volume fracture number in the second characteristic parameter, and the rock mass integrity index in the third characteristic parameter as input variables, dual-parameter interpolation is performed based on the first correspondence table and the second correspondence table to calculate the expected permeability value of the rock mass test section under the current rock mass integrity conditions and fracture development conditions; Based on the expected permeability value, and in conjunction with a preset range coefficient, the expected permeability range is determined.

6. The method for evaluating the reliability of rock mass permeability coefficient based on multi-source data according to claim 5, characterized in that, Step S3 also includes the following steps: Construct comprehensive fracture parameters, wherein the comprehensive fracture parameters are the product of the equivalent fracture aperture and the number of fractures in the rock mass volume; The dual-parameter interpolation uses the rock mass integrity index and the fracture comprehensive parameter as two input variables, extracts interpolation base points from the first correspondence table and the second correspondence table, and calculates the expected permeability value through bilinear interpolation.

7. The method for evaluating the reliability of rock mass permeability coefficient based on multi-source data according to claim 1, characterized in that, Step S4 specifically includes the following steps: First-level verification: Obtain the pre-set suspiciousness determination rules, and determine whether the first feature parameter, the second feature parameter, and the third feature parameter meet the suspiciousness determination rules; if so, determine that there is a contradiction between the first feature parameter, the second feature parameter, and the third feature parameter, and mark the rock mass test section as highly suspicious; Second layer of verification: For rock mass test sections that fail the first layer of verification and are marked as highly suspicious, the measured permeability characteristics are compared with the expected permeability characteristic range. When the measured permeability characteristics fall within the expected permeability characteristic range, the rock mass test section is determined to have passed the verification. When the measured permeability characteristics deviate from the expected permeability characteristic range, the rock mass test section is determined to have low reliability. The third layer of verification involves inputting the first feature parameter, the second feature parameter, and the third feature parameter into a pre-trained comprehensive rating classification model. The comprehensive rating classification model then outputs the probability that the rock mass test section belongs to a high reliability level, a medium reliability level, or a low reliability level, and the final reliability level is determined based on the highest probability.

8. The method for evaluating the reliability of rock mass permeability coefficient based on multi-source data according to claim 7, characterized in that, The comprehensive rating classification model employs five-fold cross-validation during the training phase, specifically including the following steps: Historical sample data labeled with reliability levels were randomly divided into five data subsets. One data subset was used as the validation set and the other four data subsets were used as the training set for five rounds of training and validation. The average performance index of the five rounds of validation was used as the performance evaluation result of the comprehensive rating classification model.

9. The method for evaluating the reliability of rock mass permeability coefficient based on multi-source data according to claim 1, characterized in that, Step S4 is followed by the following steps: S5. Based on the reliability evaluation results, in the borehole columnar section or three-dimensional geological model, the rock mass test section is marked with different colors according to the reliability level, and suggestions for retesting or intensified testing are generated for the rock mass test section with low reliability level.

10. A rock mass permeability coefficient reliability evaluation system based on multi-source data, employing the rock mass permeability coefficient reliability evaluation method based on multi-source data as described in any one of claims 1-9, characterized in that, include: The data acquisition module is used to acquire water pressure test data, borehole image data, and acoustic test data collected for the same rock mass test section within the same depth range of the same borehole. The feature extraction module is used to extract a first feature parameter reflecting the permeability characteristics of the rock mass from the water pressure test data, a second feature parameter reflecting the geometric characteristics of the fracture from the borehole image data, and a third feature parameter reflecting the integrity characteristics of the rock mass from the acoustic test data. The reliability evaluation module is used to determine the expected permeability range of the rock mass test section under the current rock mass integrity conditions and fracture development conditions based on the pre-established physical correspondence between rock mass integrity characteristics, fracture geometry characteristics and rock mass permeability characteristics, according to the second characteristic parameter and the third characteristic parameter; and to compare and verify the measured permeability characteristics characterized by the first characteristic parameter with the expected permeability range, obtain the comparison result, and generate the reliability evaluation result of the pressure water test data of the rock mass test section based on the comparison result.

Images