A tunnel advance geological prediction automation evaluation and feedback correction method and system

Abstract

The present application belongs to the technical field of geophysical exploration, and provides a tunnel advanced geological prediction automation evaluation and feedback correction method and system, which performs weighted summation on interval coincidence degree indexes, geological risk consistency indexes, geological type consistency indexes, hydrogeological consistency indexes and surrounding rock grade consistency indexes to determine a comprehensive score; adjusts the index weight according to a preset adjustment instruction, and corrects the evaluation benchmark; determines a unified standardized evaluation index, establishes an automatic feedback link from exposure to evaluation to correction, changes the traditional post-event static analysis mode, enables the prediction system to quickly learn from historical errors, forms self-adaptive correction capability, effectively prevents the repeated occurrence of similar errors, and significantly improves the timeliness of construction safety guarantee and the continuous optimization capability of the prediction model.

Description

Technical Field

[0001] This invention belongs to the field of geophysical exploration technology, and in particular relates to an automated assessment and feedback correction method and system for tunnel advanced geological prediction. Background Technology

[0002] To effectively identify adverse geological formations ahead of the tunnel face and ensure construction safety and schedule, advanced geological forecasting has become a crucial aspect of tunnel engineering construction. Current advanced geological forecasting work typically focuses on the forecasting stage, relying on various methods such as seismic wave analysis, resistivity analysis, ground-penetrating radar, and drilling, resulting in a relatively rich body of forecasting data.

[0003] Current advanced geological forecasting methods struggle to quantify the match between forecasts and actual exposure results. They lack unified calculation methods and standardized evaluation indicators, primarily focusing on the spatial overlap between forecast and exposure intervals. They lack systematic analysis of surrounding rock grades, disaster-causing structural types, the development of karst caves or fracture zones, and differences in water content. The exposure information does not form a dynamic correction loop, and errors or deviations in historical forecast conclusions cannot be systematically captured and corrected, potentially leading to similar prediction errors in subsequent sections. Furthermore, some data may contain missing data, noise, or anomalies, directly affecting the accuracy of comparing forecast results with exposure information. Current evaluation of historical forecast results heavily relies on human experience and subjective judgment, resulting in a large workload, low efficiency, and susceptibility to differences in the experience of assessment personnel, leading to unstable and inconsistent results. There is a lack of intelligent calculation and automated analysis methods. Summary of the Invention

[0004] To address the aforementioned problems, this invention proposes an automated evaluation and feedback correction method and system for tunnel advanced geological forecasting. This invention establishes unified and standardized evaluation indicators and sets up an automated feedback chain from discovery to evaluation and correction. It changes the traditional post-event static analysis mode, enabling the forecasting system to quickly learn from historical errors, form adaptive correction capabilities, effectively prevent the recurrence of similar errors, and significantly improve the timeliness of construction safety assurance and the continuous optimization capability of the forecasting model.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solution: In a first aspect, the present invention provides an automated assessment and feedback correction method for tunnel advanced geological prediction, comprising: Obtain historical forecast conclusions and excavation exposure information, as well as images of the tunnel excavation face; Based on historical forecast conclusions, excavation exposure information, and tunnel excavation face images, the interval consistency index, geological risk consistency index, geological type consistency index, hydrogeological consistency index, and surrounding rock grade consistency index are determined. The interval consistency index, geological risk consistency index, geological type consistency index, hydrogeological consistency index, and surrounding rock grade consistency index are weighted and summed to determine the comprehensive score; the index weights are adjusted according to the preset adjustment instructions, and the evaluation benchmark is revised. The evaluation is based on the overall score.

[0006] Furthermore, the historical forecast conclusions include the location of the forecast interval, adverse geological types, five geological category levels, hydrogeological characteristics, and forecast surrounding rock level; the excavation and exposure data correspond to the historical forecast conclusions.

[0007] Furthermore, the interval matching index S 1 is: ; Among them, for the excavation and exposure indicators, the advanced geological prediction interval is set as [ P i , P i+1 The actual excavation and exposure record range is []. P j , P j+1 If ], then the length of the overlapping interval is: ; The forecast interval length is: .

[0008] Furthermore, the geological risk consistency index S 2 is: ; in, G i To predict geological types; G j To reveal geological types.

[0009] Furthermore, the geological type consistency index S 3 is: ; For the consistency of the five geological category levels, including rock mass stability (L1), rock hardness (L2), rock mass integrity (L3), joint and fracture development (L4), and surrounding rock grade (L5), the five geological category level vectors are as follows: ; ; Geological type consistency score S 3k for, ; in, L i Forecast level vector; L j This is the excavation and exposure level vector.

[0010] Furthermore, the aforementioned hydrogeological consistency index S 4 is: ; in, W i To forecast water outflow; W j To reveal the situation regarding the water discharge.

[0011] Furthermore, the determination of the surrounding rock grade consistency index includes: acquiring images or video frames of cracks at the tunnel excavation face, performing grayscale conversion, histogram equalization or adaptive histogram equalization and noise filtering on the images to obtain preprocessed images, comparing the crack grade index with historical prediction results based on the preprocessed images to obtain a surrounding rock grade consistency index score.

[0012] Furthermore, determine the proportion of fractures. F r Crack degree D r Surface porosity A r and rock mass integrity index Q f : ; ; ; ; in, N crack The number of pixels in the crack. N total This represents the total number of pixels. N c The number of connected segments in the fracture. ω i For the first i Average width of the segmental crack l i For the first i Segment crack length; Q f for li Rock mass integrity index; α , β, γ These are the weighting coefficients; based on the index Q fThe surrounding rock is classified into 5 levels:

[0013] Compared with historical forecast fracture levels L pre Comparison yields consistent indicators of surrounding rock grade S 5: ; The overall score is: : in, S 1 represents the interval fit index; S 2 represents the consistency index of geological risk; S 3 represents the consistency index of geological types; S 4 represents the hydrogeological consistency index. S 5 represents the consistency index of surrounding rock grade; α 1, α 2, α 3, α 4, α 5 represents the weight of a single item's score.

[0014] Furthermore, an automated quality assessment is performed based on the acquired multi-source raw forecast data, assigning a quantitative quality score to each raw forecast data; the weighting coefficients of each evaluation indicator are dynamically adjusted based on the quality score of the corresponding raw forecast data.

[0015] Secondly, the present invention also provides an automated assessment and feedback correction system for tunnel advanced geological prediction, comprising: The data acquisition module is configured to: acquire historical forecast conclusions and excavation exposure information, as well as images of the tunnel excavation face; The indicator determination module is configured to determine the interval consistency index, geological risk consistency index, geological type consistency index, hydrogeological consistency index, and surrounding rock grade consistency index based on historical forecast conclusions, excavation exposure information, and tunnel excavation face images. The weighted summation module is configured to: perform weighted summation on the interval consistency index, geological risk consistency index, geological type consistency index, hydrogeological consistency index, and surrounding rock grade consistency index to determine the comprehensive score; adjust the index weights according to preset adjustment instructions; and revise the evaluation benchmark. The evaluation module is configured to evaluate based on the overall score.

[0016] Thirdly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the automated assessment and feedback correction method for tunnel advanced geological prediction as described in the first aspect.

[0017] Fourthly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the program to implement the steps of the automated assessment and feedback correction method for tunnel advanced geological prediction described in the first aspect.

[0018] Fifthly, the present invention also provides a computer program product, the computer program product comprising a computer program, which, when executed by a processor, implements the steps of the automated assessment and feedback correction method for tunnel advanced geological prediction described in the first aspect.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention first determines the interval consistency index, geological risk consistency index, geological type consistency index, hydrogeological consistency index, and surrounding rock grade consistency index based on historical forecast conclusions, excavation exposure information, and tunnel excavation face images. Then, it performs a weighted summation of these indices to determine a comprehensive score. The weights of the indices are adjusted according to preset adjustment instructions, and the evaluation benchmark is corrected. A unified and standardized evaluation index is established, creating an automated feedback loop from exposure to evaluation and then to correction. This changes the traditional post-hoc static analysis model, enabling the forecasting system to quickly learn from historical errors, develop adaptive correction capabilities, effectively prevent the recurrence of similar errors, and significantly improve the timeliness of construction safety assurance and the continuous optimization capability of the forecasting model.

[0020] 2. This invention addresses the problem of not fully considering the impact of data quality on forecast results, thus improving the scientific rigor and reliability of the evaluation results. Traditional methods fail to differentiate data quality, treating all comparison results equally, resulting in an unreliable evaluation basis. This invention introduces an intelligent data quality assessment and dynamic weight allocation mechanism, incorporating data quality as a core variable into the evaluation system. The evaluation results more accurately reflect the effectiveness of the forecasting method itself, rather than being contaminated by low-quality data. This allows the comprehensive scoring and index analysis to more accurately reflect the actual geological conditions, providing a scientific basis for determining which forecasting method is more reliable under what data conditions. Attached Figure Description

[0021] The accompanying drawings, which form part of this embodiment, are used to provide a further understanding of this embodiment. The illustrative embodiments and their descriptions are used to explain this embodiment and do not constitute an improper limitation of this embodiment.

[0022] Figure 1 This is a technical roadmap of the method and system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the forecast interval, the disclosure recording interval, and the matching area in an embodiment of the present invention; Figure 3 These are images of the working face as described in an embodiment of the present invention. Figure 4 This is a grayscale image of the working face in an embodiment of the present invention; Figure 5 This is a grayscale image after noise reduction processing according to an embodiment of the present invention; Figure 6 The background and the image after binarization of the crack are shown in the embodiment of the present invention. Figure 7 This is a comparison diagram of the crack identification effect in an embodiment of the present invention. Detailed Implementation

[0023] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0024] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0025] Example 1: To effectively identify unfavorable geological formations ahead of the tunnel face and ensure construction safety and schedule, advanced geological forecasting has become a crucial aspect of tunnel engineering construction. Through forecasting, the surrounding rock grade, structural type, water content, and hazard risks of the strata ahead can be predicted before excavation, thus providing a basis for construction plans and support design.

[0026] Current advanced geological forecasting work typically focuses on the forecasting stage, relying on various methods such as seismic wave analysis, resistivity analysis, ground-penetrating radar, and drilling, resulting in relatively abundant forecasting results. However, the evaluation stage, i.e., the verification and feedback of historical forecast conclusions after excavation and exposure, remains significantly inadequate. Current practices often remain at the level of qualitative description, such as comparing the consistency between forecast results and exposure results to give conclusions of basic agreement, partial agreement, or discrepancies. This approach lacks systematic and quantitative standards, making it difficult to form a reusable and traceable evaluation system.

[0027] Meanwhile, excavation, as a direct means of obtaining real underground geological information, can provide multi-dimensional information such as surrounding rock grade, structural location, spatial extension range, and water outflow. If this information can be quantitatively compared with historical forecast results, it can not only improve the objectivity and accuracy of the evaluation but also provide a basis for dynamic correction of forecasts for subsequent sections.

[0028] Currently, advanced geological prediction methods lack a unified and quantifiable evaluation index system. Specifically, existing tunnel geological prediction evaluation methods mostly rely on manual comparison or qualitative description, typically only providing vague levels such as conformity, non-conformity, or high / medium / low, making it difficult to quantify the degree of matching between predictions and actual exposure results. The lack of unified calculation methods and standardized evaluation indicators makes it impossible to directly compare evaluation results between different sections and projects, and also hinders the formation of traceable historical records. This limits the operability of prediction method optimization and model calibration. The lack of quantitative indicators not only reduces the objectivity of evaluations but also hinders the formation of a data-driven dynamic prediction system. Furthermore, the evaluation dimensions are singular, failing to fully reflect the complexity of geological attributes. Specifically, traditional evaluation methods... The primary focus is on the spatial overlap between the forecast and exposure intervals, lacking a systematic analysis of surrounding rock grade, disaster-causing structural types, the development degree of karst caves or fractured zones, and differences in water content. This singular evaluation dimension leads to insufficient characterization of actual geological complexity, causing the judgment of forecast results to be biased towards local information rather than comprehensively reflecting the safety risks of the entire tunnel section. The lack of multi-dimensional evaluation also affects the rationality of the comprehensive score, resulting in a singular and one-sided basis for engineering decisions. Furthermore, the use of exposure information for follow-up evaluation is untimely or nonexistent, leading to delayed feedback on forecast effects. Specifically, in current engineering practice, data exposed during tunnel face excavation is typically only used for post-hoc analysis and fails to be promptly fed back into the forecasting system or construction plan adjustments. This is because the exposure information is not... The formation of a dynamic correction loop means that errors or deviations in historical forecast conclusions cannot be captured and corrected by the system, leading to the possibility of similar prediction errors in subsequent sections. This lag feedback not only affects forecast accuracy but also reduces construction safety assurance and the continuous optimization capability of the forecast model, making it difficult for the forecast system to form an adaptive learning mechanism. Furthermore, insufficient attention is paid to the impact of inconsistent quality of multi-source raw data on forecast conclusions. Specifically, in existing technologies and engineering applications, the raw data relied upon for forecasting often comes from diverse sources, including borehole data, geophysical data, and construction records. Data from different sources vary in accuracy, completeness, and reliability; some data may contain missing measurements, noise, or outliers, directly affecting forecast results and reveal information. The accuracy of information comparison is currently hampered by insufficient consideration of the impact of data quality in the current evaluation system. It fails to effectively distinguish the contribution of high-quality and low-quality data to the evaluation results, which may lead to deviations in the comprehensive score and indicator analysis from the actual geological conditions. The lack of intelligent means and automated analysis results in low evaluation efficiency. Specifically, the current evaluation of historical forecast results relies heavily on human experience and subjective judgment, which is labor-intensive and inefficient. It is also susceptible to the influence of differences in the experience of the evaluators, resulting in unstable and inconsistent results. The lack of intelligent computing and automated analysis means that it is difficult to process and analyze large-scale, multi-segment forecast data quickly and in batches, and it also limits the calculation and visualization of complex indicators (such as multi-weight comprehensive scores and credibility curves).

[0029] To solve at least one of the above problems, such as Figure 1 As shown in the figure, this embodiment provides an automated evaluation and feedback correction method for tunnel advanced geological prediction, which can realize the scientific evaluation of historical prediction results and continuous optimization of the prediction system. Specifically, it includes the following steps: S1. Obtain historical forecast conclusions and excavation reveal information: In this embodiment, geological sketching, raw forecast data inspection, imaging result inspection, and advanced geological forecast report sorting are carried out at the tunnel site to collect effective early forecast information. Forecast conclusions, parameter values, geological risk classifications, etc. are processed reasonably and accurately. Combined with information revealed by excavation and human experience, specific case information is supplemented from multiple perspectives.

[0030] In some embodiments, geological survey data, advanced drilling, borehole television, and pilot tunnel information can be utilized to fully leverage available prior geological information, providing accurate and reliable information for accurate automated assessment. Alternatively, the excavation reveal information can be corrected based on on-site construction procedures, rock physics test conclusions, TBM tunneling parameter variation curves, and expert experience.

[0031] S2. Combine the forecast report and the excavation and exposure form to generate a scoring interval table. The forecast report and the excavation and exposure form include information such as the start and end chainage of the forecast and excavation and exposure, adverse geological types, five geological category levels (rock mass stability, rock hardness, rock mass integrity, degree of joint and fissure development, and geological risk type), hydrogeological characteristics, and the level of the exposed surrounding rock.

[0032] Specifically, each geological classification includes several tags: 3 tags for rock mass stability, 5 tags for rock hardness, 9 tags for rock mass integrity, 4 tags for joint and fracture development, 10 tags for unfavorable geological types, 10 tags for geological risk types, and 7 tags for hydrogeological characteristics, as shown in Table 1. Table 1. Correspondence between categories and labels Label Name Classification Rock mass stability Unstable, Relatively Stable, Stable rock hardness Extremely soft rock, soft rock, relatively soft rock, relatively hard rock, hard rock Rock mass integrity Broken, somewhat broken ~ Broken, somewhat broken, poorly intact ~ somewhat broken, poorly intact, relatively intact ~ poorly intact, relatively intact, intact ~ relatively intact, complete Degree of joint and fracture development Very developed, developed, somewhat developed, not developed Unfavorable geological types Karst, well-developed groundwater, weak strata, fault fracture zones, densely jointed and fractured zones, none, mined-out areas, aquifers, water-retaining structures, high stress Hydrogeological characteristics gushing water, stream of water, linear flow of water, dripping water, seepage, dampness, dryness Geological risk types Mudslide, collapse + increased groundwater, landslide, soft rock deformation, water inrush, increased groundwater, collapse + landslide, rockfall, no, hard rock rockburst S3. Calculate the excavation exposure tracking index, adverse geological type index, five geological category levels and hydrogeological characteristic comparison scores. The excavation exposure form is shown in Table 2, and the advanced geological forecast report form is shown in Table 3.

[0033] Table 2 Mileage range Rock mass stability rock hardness Rock mass integrity Degree of joint and fracture development Unfavorable geological types Surrounding rock grade Geological risk types Hydrogeological characteristics K129+865.2~K129+890.2 Unstable hard rock Poor integrity - relatively fragile Relatively developed none Ⅲ Dropping blocks - K129+880~K129+886 Unstable hard rock Poor integrity - relatively fragile Relatively developed none Ⅲ Dropping blocks - K129+890.2~K129+930.2 Unstable hard rock Poor integrity Relatively developed none Ⅲ Dropping blocks - Table 3 Mileage range Rock mass stability rock hardness Rock mass integrity Degree of joint and fracture development Unfavorable geological types Surrounding rock grade Geological risk types Hydrogeological characteristics K129+865.2~K129+895.2 Relatively stable hard rock Poor integrity - relatively fragile Relatively developed none Ⅲ Dropping blocks - K129+895.2~K129+915.2 Unstable harder Poor integrity development none Ⅲ Dropping blocks - K129+915.2~K129+930.2 Unstable hard rock Poor integrity Relatively developed none Ⅲ Dropping blocks - Specifically, the extracted data is analyzed to calculate the following two scores: the excavation and exposure tracking index and the single-item level difference index. An example of the corresponding relationship is shown below. Figure 2 As shown in section a). The specific process is as follows: The formula for calculating interval overlap is: ; The formula for calculating the length of the forecast interval is: ; The excavation and exposure tracking indicators S The calculation formula is as follows: ; in, P i , P i+1 These are the starting and ending station numbers of the forecast interval, respectively. P j , P j+1 These are the starting and ending station numbers of the excavation and exposure tracking section, respectively.

[0034] S4. Calculate the consistency score for geological risk types. An example of the corresponding relationship is shown below. Figure 2 As shown in section b).

[0035] Consistency regarding geological risk types. The predicted geological type is: G i The geological type revealed is G j Its individual score S 2 is the value, and the calculation formula is: ; S5. Calculate the consistency of the five geological categories and assign weights to calculate the comprehensive score for this item. An example of the corresponding relationships is shown below. Figure 2 As shown in section b).

[0036] The five geological categories are consistent in their classification. These include rock mass stability (L1), rock hardness (L2), rock mass integrity (L3), joint and fracture development (L4), and surrounding rock grade (L). 5。

[0037] The five geological category level vectors are: ; ; Among them, L i L is the forecast level vector; j This is the excavation and exposure level vector.

[0038] Geological type consistency score S 3k for, ; Geological type consistency scoreS 3 is, ; Consistency regarding hydrogeological characteristics. Forecast water outflow. W i The situation regarding the water discharge was revealed. W j Its individual score S 4 is, ; S6, Evaluation index for predicted surrounding rock grade.

[0039] Specifically, the evaluation quality indicators are obtained by identifying images of the working face or surrounding rock exposed during on-site excavation, delineating the distribution of fractures through image processing and binary classification methods, and further calculating the number of fracture connections, fracture degree, and cracking degree.

[0040] Obtain images or video frames of cracks at the tunnel excavation face, and convert them to RGB format, such as... Figure 3 As shown.

[0041] Let the original RGB image be The image after grayscale conversion is as follows Figure 4 As shown, the processing formula is, ; Further histogram equalization, contrast enhancement, and noise filtering are performed. The cumulative distribution function of the grayscale image is CDF(g), and the preprocessed grayscale values ​​are: ; After Gaussian denoising, it becomes: ; ; in, i , j This represents the offset of the convolution kernel; its value range is... k ≤ i , j ≤ k ,in, k Determine the size of the filtering window, typically 2 k +1 represents the window side length; σ The standard deviation of the Gaussian function is used to obtain the final preprocessed image, such as... Figure 5 As shown.

[0042] Texture features that characterize the differences between rock mass and fracture regions are extracted from the preprocessed image to generate a texture feature map F_tex; the texture features can be multi-scale texture response of Gaussian filter or depth features extracted based on convolutional neural network (CNN).

[0043] Using the texture feature map F_tex as input, and setting the number of clusters K=2, the K-means clustering algorithm is used to cluster the image pixels into two categories: fractures and rock background, generating a binarized fracture segmentation image, such as... Figure 6 As shown.

[0044] The binarized crack segmentation image is subjected to morphological post-processing, followed by calculation of crack parameters, specifically including: By scanning a binary image using a connected component labeling algorithm, all independent fracture regions are identified; the total number of these regions represents the number of fracture connected segments. N c The crack is binarized and converted into a mask matrix.

[0045] For each labeled connected component (i.e., the first...) i (segment crack), calculate its skeleton length as the length of the segment crack. l i Based on the total pixel area of ​​the connected component A i With length l i Calculate its average width using the following formula: ω i = A i / l i Using the original image as a background, the crack mask matrix is ​​input into the RGB red channel and superimposed on the original image to form a comparison image, such as... Figure 7 As shown.

[0046] Based on the above parameters, the fracture ratio is further calculated ( F r ), cracking degree ( D r ), surface porosity ( A r ) and rock mass integrity index ( Q f The calculation formula is as follows: ; ; ; ; in, N crack This represents the number of pixels in the crack. N total This represents the total number of pixels. N c This represents the number of connected segments in the fracture. ω i For the first i Average width of the segmental crack; l i For the first i Segment crack length; Q f for li Rock mass integrity index; α , β , γ These are weighting coefficients, typically α + β + γ =1.

[0047] According to the index Q f The surrounding rock is classified into 5 levels, as follows: ; Compared with historical forecast fracture levels L pre Comparison yields consistent indicators of surrounding rock grade S 5: ; The final overall score is: : in, S 1 represents the interval fit index; S 2 represents the consistency index of geological risk; S 3 represents the consistency index of geological types; S 4 represents the hydrogeological consistency index. S 5 represents the consistency index of surrounding rock grade; α 1, α 2, α 3, α 4, α 5 represents the weight of a single item's score (which can be set based on engineering experience). α 1+ α 2+ α 3+ α 4+ α 5=1. For railway tunnels, the suggested weights for this invention patent are 0.15, 0.25, 0.2, 0.25, and 0.15, respectively.

[0048] The image processing parameters for implementing the above embodiments are shown in Table 4: Table 4 Image Processing Parameters Parameter type variable numerical values Remark Pixel physical size pixel_size 4.2mm If there are actual dimensions, they can be set (e.g., how many millimeters each pixel represents). Contrast Limiting Parameters clipLimit 8 Control the degree of contrast enhancement Block size gridSize (8，8) The image is divided into an 8x8 grid for local contrast enhancement. Gaussian filter kernel gauss_ksize 5 The size of the Gaussian filter kernel must be an odd number for noise reduction. Gaussian filter standard deviation gauss_sigma 1 The standard deviation of the Gaussian filter controls the smoothness. Number of clusters kmeans_K 2 The number of clusters in KMeans clustering is divided into two categories: gaps and background. Minimum Crack Size min_area 80 Minimum crack area (in pixels); areas smaller than this value will be filtered out. Fr weight alpha 0.7 Weighting coefficient of porosity Dr weight beta 0.3 Weighting coefficient of cracking degree Example 2: Based on the method in Example 1, this example provides a comprehensive index output and feedback correction method and system, such as... Figure 1 As shown, the specific steps include the following: Taking the section from K45+300 to K45+360 of a tunnel as an example, this paper illustrates the practical application process of the comprehensive index output and feedback correction of the present invention. This embodiment uses a specific engineering section (fictional but strictly conforming to engineering logic) as the background, providing the entire process from data input → data quality calculation → index calculation → expert intervention → dynamic correction → report output, further enhancing the feasibility, repeatability, and engineering credibility of the solution.

[0049] In this embodiment, three prediction methods are used in this section: seismic wave advance prediction, resistivity imaging, and advance drilling. At the same time, excavation and exposure data are available, which can realize a complete comparison and evaluation process.

[0050] S1. Data Acquisition and Preprocessing: Through a preset data interface, the module automatically extracts raw data from different sources. The module has a built-in data cleaning rule library that can automatically identify and process missing values, obvious outliers, and non-standard formats in the raw data, and unify the data into a standard format.

[0051] S1.1 The system backend detected the addition of three types of forecast data and tunnel face exposure information for this section to the engineering database, and automatically started the data acquisition process. The collected raw data is shown in Table 5: Table 5 Raw Data Collection Data types Summary Seismic wave prediction The reflective interface is expected to be located at K45+323, and the estimated width of the fracture zone is approximately 6m. Resistivity imaging The abnormally low resistance range is predicted to be K45+320~K45+330. Advanced drilling The Class IV surrounding rock fracture zone is located between K45+325 and K45+340, and actual drilling revealed weak water content. Disclosure Information The actual fracture zone is located between K45+324 and K45+339, with a maximum width of 15.5m and localized weak water inrush. S1.2 The preprocessing module performs the following operations on the above data: S1.2.1 Missing Point Repair: Two sampling points are missing in the seismic wave reflection data. The system automatically uses cubic spline interpolation to fill in the missing points.

[0052] S1.2.2, Outlier Removal: The resistivity curve has two single-point spikes (which do not conform to the segment trend), which are automatically eliminated by median filtering.

[0053] S1.2.3 Unified Dimensions and Formats: Unify the unit of resistivity to Ωm, and unify the station number of all types of data to K+ meter format.

[0054] S1.2.4 Interval pairing: The forecast interval is uniformly set as K45+300~K45+360, and is automatically associated with the revealed interval to form a comparable data group.

[0055] After step S1.2 is completed, all data will meet the unified format requirements and can automatically enter the data quality assessment stage.

[0056] S2, Intelligent Data Quality Assessment: The system performs automated and quantitative quality assessments on the multi-source raw forecast data imported into the system. It operates based on a pre-defined multi-dimensional quality evaluation index system, which typically includes automated evaluation of multiple data parameters. Ultimately, each piece of raw data participating in the forecast is assigned a quantitative quality score Q. Score Rating labels.

[0057] The system is based on a multi-dimensional indicator system, mainly including completeness, reliability, correlation, and sensor spatial deployment effectiveness, and calculates the QScore for the three types of forecast data respectively: S2.1 Seismic wave prediction data quality: Integrity C-score=0.92; Signal-to-noise ratio T-score=0.85; Trend consistency R-score=0.88; Detector spatial deployment effectiveness F-score=0.93; Overall QScore=0.89 (high quality).

[0058] S2.2 Resistivity data quality: Completeness C-score=0.86; Noise is relatively high T-score=0.72; Good trend R-score=0.81; Electrode spatial layout effectiveness F-score=0.91; Overall QScore=0.80 (medium quality).

[0059] S2.3, Quality of Advanced Drilling Data: Small data volume, C-score=0.60; High reliability, T-score=0.94; Insufficient section coverage; Overall QScore=0.73 (medium to low quality).

[0060] The system automatically labels the three types of data as: high quality, medium quality, and low coverage data, and generates weighting factors for the next weighted calculation.

[0061] S3. Calculation of comprehensive forecast effectiveness index: The preprocessed and quality-rated forecast data is compared with its corresponding actual revealed information in detail to calculate a series of basic indicators. Through weighted comprehensive evaluation, dynamic weight factors are set to automatically draw visualization charts such as the forecast accuracy change curve with data quality and the method effectiveness comparison chart under different geological conditions.

[0062] The system automatically performs quantitative comparisons based on various forecast and disclosure data to obtain the basic indicators shown in Table 6: Table 6 Basic Indicators index definition result M1: Spatial bias Difference between predicted center point and revealed center point Seismic wave: +1.0m; Resistivity: -2.0m; Drilling depth: +0.5m M2: Scale Deviation Difference between predicted width and actual width Seismic wave: +0.5m; Resistivity: +1.5m; Drilling: +0.2m M3: Type Consistency Geological category prediction accuracy Seismic wave: 0.90; Resistivity: 0.75; Drilling: 0.95 M4: Risk Matching Degree Covering the accuracy of inrush level prediction 0.85 (overall value) The system automatically adjusts the weights of each indicator based on the QScore from step S2, forming the following weighted comprehensive index: seismic wave weighted contribution coefficient: 0.42; resistivity weighted contribution coefficient: 0.33; drilling weighted contribution coefficient: 0.25; finally, the system calculates the comprehensive evaluation index of this section: EIndex=0.87, which is rated as high accuracy.

[0063] As an optional implementation, the system can be set to automatically generate charts as needed, including but not limited to data quality-accuracy coupling curves, bar charts comparing the deviations of various methods, and radar charts of comprehensive indicators.

[0064] S4. Overall Evaluation of Human-Computer Interaction: By achieving deep integration of intelligence and expert input, fully automated initial evaluation, expert review, and intervention can be completed. Experts can annotate special geological anomalies that the system fails to identify, adjust the automatically assigned indicator weights under specific circumstances, or correct the benchmark labels used for comparison. Expert interventions are recorded and used as feedback samples to fine-tune the data quality assessment model and indicator weight allocation strategy. The system will send the preliminary evaluation report to geological experts, who will then make the following judgments: S4.1 Expert identification of abnormal information: Experts found a typical tensile dissolution zone near K45+327. This feature is prone to causing excessive expansion of resistivity anomalies. Therefore, the resistivity width deviation of the system (+1.5m) is an interpretable error.

[0065] S4.2 Adjusting the weights: Experts adjusted the weight of resistivity data from 0.33 to 0.28 and increased the weight of seismic waves to 0.45 to better reflect the reliability distribution under these geological conditions.

[0066] S4.3 Supplementary Qualitative Explanation: Experts added: The water inflow in this section is relatively weak, and the drilling data is highly reliable. It is recommended to increase the weight of drilling data in similar sections.

[0067] S4.4 The system records all the above operations as feedback samples and inputs them into the next step for model correction.

[0068] S4.5 After expert confirmation, the final comprehensive score for this section is revised to: EIndex_final=0.89.

[0069] S5. Dynamic Feedback and Model Correction: The final evaluation conclusions, confirmed through comprehensive human-computer interaction, are transformed into specific optimization actions. The module continuously monitors the comprehensive evaluation results, and when a systematic deviation is detected in a certain forecasting method or data source under specific geological conditions, a correction mechanism can be automatically triggered.

[0070] Based on the manual intervention record from step S4, the system initiates the model correction process, including: S5.1 Adjust the geological scene weight model: The system updates the preset weight matrix of the karst development zone, reduces the weight of the seismic wave method by 5%, and increases the weight of the resistivity method by 6%.

[0071] S5.2 Correcting data quality model parameters: The system identifies that resistivity is prone to abnormal expansion in this type of area and automatically increases the weight of resistivity trend consistency in the QScore.

[0072] S5.3 Update the forecast method credibility coefficient library: The system adjusts the credibility coefficient of drilling in this type of section from 0.78 to 0.84 to reflect expert opinions.

[0073] The revised model will be used for the next automatic evaluation of the segment, enabling the system to have dynamic learning capabilities.

[0074] S6. Automated report generation and visualization output: Based on preset templates and rules, the processing results of all the aforementioned modules (data quality distribution, forecast-reveal comparison details, comprehensive scoring trends, manual intervention records, and model correction suggestions) can be integrated with one click to automatically generate a tunnel advance forecast quality evaluation and feedback report, improving the efficiency and standardization of deliverables.

[0075] The system automatically generates the "Forecast Quality Evaluation and Feedback Report for the K45+300~K45+360 Section", which includes: a list of various raw forecast data, a data quality scoring table and distribution map, quantitative indicators for the comparison between forecasts and revelations, deviation analysis charts for each method, the final comprehensive score (EIndex_final=0.89), automatic recording of expert adjustments, and automatically triggered model correction summaries.

[0076] The report can be exported with one click and used directly for project data archiving or subsequent construction decisions.

[0077] The working method of the system is the same as that of the automated assessment and feedback correction method for tunnel advanced geological prediction in Example 1, and will not be repeated here.

[0078] This invention addresses the issue of feedback lag during tunnel excavation by establishing a real-time automated feedback chain from exposure to evaluation and correction. Once excavation is completed, the system immediately initiates evaluation and generates feedback, promptly correcting model parameters. This changes the traditional post-hoc static analysis model, enabling the forecasting system to quickly learn from historical errors, develop adaptive correction capabilities, effectively prevent the recurrence of similar errors, and significantly improve the timeliness of construction safety assurance and the continuous optimization capability of the forecasting model.

[0079] This invention addresses the problem of insufficiently considering the impact of data quality on forecast results, thereby improving the scientific rigor and reliability of the evaluation. Traditional methods fail to differentiate data quality, treating all comparison results equally, leading to an unreliable evaluation basis. This invention introduces an intelligent data quality assessment module and a dynamic weight allocation mechanism, incorporating data quality as a core variable into the evaluation system. The evaluation results more accurately reflect the effectiveness of the forecasting method itself, rather than being contaminated by low-quality data. This allows the comprehensive scoring and index analysis to more accurately reflect the actual geological conditions, providing a scientific basis for determining which forecasting method is more reliable under what data conditions.

[0080] This invention proposes a novel integrated evaluation model combining intelligence and human expertise by introducing a collaborative approach that integrates automated calculation, human-computer interaction, and automated report generation. The system handles a large amount of repetitive computation, ensuring efficiency and objectivity; experts focus on handling abnormal and complex situations, contributing their professional judgments and providing guidance to the system. This mechanism leverages both the computational efficiency of computers and the professional wisdom of humans, making the evaluation results not only efficient and consistent but also more flexible, interpretable, and credible in engineering. The automatic generation of the visualized report significantly improves the efficiency of results presentation and decision support capabilities.

[0081] Example 3: This embodiment provides an automated assessment and feedback correction system for tunnel advanced geological prediction, including: The data acquisition module is configured to: acquire historical forecast conclusions and excavation exposure information, as well as images of the tunnel excavation face; The indicator determination module is configured to determine the interval consistency index, geological risk consistency index, geological type consistency index, hydrogeological consistency index, and surrounding rock grade consistency index based on historical forecast conclusions, excavation exposure information, and tunnel excavation face images. The weighted summation module is configured to: perform weighted summation on the interval consistency index, geological risk consistency index, geological type consistency index, hydrogeological consistency index, and surrounding rock grade consistency index to determine the comprehensive score; adjust the index weights according to preset adjustment instructions; and revise the evaluation benchmark. The evaluation module is configured to evaluate based on the overall score.

[0082] The working method of the system is the same as that of the automated assessment and feedback correction method for tunnel advanced geological prediction in Example 1, and will not be repeated here.

[0083] Example 4: This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the automated assessment and feedback correction method for tunnel advanced geological prediction described in Embodiment 1.

[0084] Example 5: This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the program, it implements the steps of the automated evaluation and feedback correction method for tunnel advanced geological prediction described in Embodiment 1.

[0085] Example 6: This embodiment provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the steps of the automated assessment and feedback correction method for tunnel advanced geological prediction described in Embodiment 1.

[0086] The above description is merely a preferred embodiment of this practice and is not intended to limit the scope of this practice. Various modifications and variations can be made to this practice by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this practice should be included within the protection scope of this practice.

Claims

1. An automated assessment and feedback correction method for tunnel advanced geological prediction, characterized in that, include: Obtain historical forecast conclusions and excavation exposure information, as well as images of the tunnel excavation face; Based on historical forecast conclusions, excavation exposure information, and tunnel excavation face images, the interval consistency index, geological risk consistency index, geological type consistency index, hydrogeological consistency index, and surrounding rock grade consistency index are determined. The interval consistency index, geological risk consistency index, geological type consistency index, hydrogeological consistency index, and surrounding rock grade consistency index are weighted and summed to determine the comprehensive score; the index weights are adjusted according to the preset adjustment instructions, and the evaluation benchmark is revised. The evaluation is based on the overall score.

2. The automated assessment and feedback correction method for tunnel advanced geological prediction as described in claim 1, characterized in that, The historical forecast conclusions include the location of the forecast interval, the type of adverse geological conditions, the levels of five geological categories, the hydrogeological characteristics, and the forecast surrounding rock level; the excavation data corresponds to the historical forecast conclusions.

3. The automated assessment and feedback correction method for tunnel advanced geological prediction as described in claim 1, characterized in that, The interval matching index S 1 is: ； Among them, for the excavation and exposure indicators, the advanced geological prediction interval is set as [ P i , P i+1 The actual excavation and exposure record range is []. P j , P j+1 If ], then the length of the overlapping interval is: ； The forecast interval length is: 。 4. The automated assessment and feedback correction method for tunnel advanced geological prediction as described in claim 1, characterized in that, The geological risk consistency index S 2 is: ； in, G i To predict geological types; G j To reveal geological types.

5. The automated assessment and feedback correction method for tunnel advanced geological prediction as described in claim 1, characterized in that, Geological type consistency index S 3 is: ； For the consistency of the five geological category levels, including rock mass stability (L1), rock hardness (L2), rock mass integrity (L3), joint and fracture development (L4), and surrounding rock grade (L5), the five geological category level vectors are as follows: ； ； Geological type consistency score S 3k for, ； in, L i Forecast level vector; L j This is the excavation and exposure level vector.

6. The automated assessment and feedback correction method for tunnel advanced geological prediction as described in claim 1, characterized in that, The hydrogeological consistency index S 4 is: ； in, W i To forecast water outflow; W j To reveal the situation regarding the water discharge.

7. The automated assessment and feedback correction method for tunnel advanced geological prediction as described in claim 1, characterized in that, The determination of the surrounding rock grade consistency index includes: acquiring images or video frames of cracks at the tunnel excavation face, performing grayscale conversion, histogram equalization or adaptive histogram equalization and noise filtering on the images to obtain preprocessed images, comparing the crack grade index with historical prediction results based on the preprocessed images to obtain the surrounding rock grade consistency index score.

8. The automated assessment and feedback correction method for tunnel advanced geological prediction as described in claim 7, characterized in that, Determine the proportion of cracks F r Crack degree D r Surface porosity A r and rock mass integrity index Q f : ； ； ； ； in, N crack The number of pixels in the crack. N total This represents the total number of pixels. N c The number of connected segments in the fracture. ω i For the first i Average width of the segmental crack l i For the first i Segment crack length; Q f for li Rock mass integrity index; α , β, γ These are the weighting coefficients; based on the index Q f The surrounding rock is classified into 5 levels: Compared with historical forecast fracture levels L pre Comparison yields consistent indicators of surrounding rock grade S 5: ； The overall score is: ： in, S 1 represents the interval fit index; S 2 represents the consistency index of geological risk; S 3 represents the consistency index of geological types; S 4 represents the hydrogeological consistency index. S 5 represents the consistency index of surrounding rock grade; α 1, α 2, α 3, α 4, α 5 represents the weight of a single item's score.

9. The automated evaluation and feedback correction method for tunnel advanced geological prediction as described in claim 7, characterized in that, An automated quality assessment is performed based on the acquired multi-source raw forecast data, assigning a quantitative quality score to each raw forecast data. The weighting coefficients for each evaluation indicator are dynamically adjusted based on the quality score of the corresponding raw forecast data.

10. An automated assessment and feedback correction system for tunnel advanced geological prediction, characterized in that, include: The data acquisition module is configured to: acquire historical forecast conclusions and excavation exposure information, as well as images of the tunnel excavation face; The indicator determination module is configured to determine the interval consistency index, geological risk consistency index, geological type consistency index, hydrogeological consistency index, and surrounding rock grade consistency index based on historical forecast conclusions, excavation exposure information, and tunnel excavation face images. The weighted summation module is configured to: perform weighted summation on the interval consistency index, geological risk consistency index, geological type consistency index, hydrogeological consistency index, and surrounding rock grade consistency index to determine the comprehensive score; adjust the index weights according to preset adjustment instructions; and revise the evaluation benchmark. The evaluation module is configured to evaluate based on the overall score.

Images