A rock mass irrigability rapid evaluation method and system based on drilling parameters

Abstract

This invention discloses a rapid evaluation method and system for rock mass groutability based on drilling parameters. The method includes: collecting drilling parameters of the rock mass, filtering out invalid data such as idle drilling and tripping, and obtaining a drilling parameter depth sequence by interpolating and aligning valid data. Based on this sequence, a comprehensive drilling index characterizing rock mass integrity and a discrete index reflecting fracture development are calculated. These two types of indices are input into an inversion model, which outputs engineering indicators such as rock mass quality and equivalent permeability. Quantitative evaluation is conducted in conjunction with grouting conditions such as grouting pressure and grout properties to classify the rock mass groutability level, and a three-dimensional distribution of regional groutability is generated based on borehole spatial information. The measured grouting data are compared with the rock mass engineering indicators. If the deviation exceeds the limit, the inversion model is fine-tuned or the groutability classification standard is updated using the measured data, achieving dynamic correction of the evaluation system. This provides reliable technical support for improving the quality of grouting construction.

Description

Technical Field

[0001] This invention belongs to the field of geotechnical engineering technology, and more specifically, relates to a method and system for rapid evaluation of rock mass groutability based on drilling parameters. Background Technology

[0002] Grouting is a crucial construction method in water conservancy and hydropower projects for forming seepage prevention systems and improving the engineering properties of rock masses. Among these methods, the gritability of the rock mass is a core fundamental condition determining the grouting effect and the rationality of the construction plan. However, the gritability of rock masses exhibits significant spatial heterogeneity and concealment. Traditional gritability evaluation methods often rely on post-hoc or indirect detection methods such as borehole water pressure tests, core sampling, and sonic testing. These methods suffer from problems such as long testing cycles, limited testing points, significant construction disturbance, and difficulty in reflecting the continuous changes in the rock mass.

[0003] In practical engineering, the grouting feasibility of rock mass can often only be determined after drilling is completed, based on water pressure tests or trial grouting results. This makes it difficult to provide timely decision-making basis for grouting parameter selection, zoning and grading, and construction optimization. At the same time, with the continuous increase in project scale, traditional point-based detection methods are no longer sufficient to meet the needs of precise and intelligent construction for rapid identification of rock mass groutability.

[0004] In recent years, digital drilling and monitoring-while-drilling (WWD) technologies have been gradually applied in water conservancy and hydropower projects. During drilling, various construction parameters such as drilling speed, drilling pressure, rotational speed, and torque can be acquired in real time. These WWD parameters directly reflect the interaction characteristics between the drill string and the rock mass during drilling and are intrinsically related to the degree of rock fragmentation, structural integrity, and permeability. However, current technologies lack an effective method for systematically processing WWD parameters and using them for rapid evaluation of rock mass groutability. The application value of WWD data in grouting engineering has not yet been fully explored.

[0005] Therefore, it is urgent to propose a scientific and feasible technical solution to realize the forward identification and continuous evaluation of rock mass groutability, and to provide technical support for the optimization of grouting construction parameters and the control of engineering quality. Summary of the Invention

[0006] This invention addresses the shortcomings of traditional rock mass groutability testing, such as lag, limited measuring points, and inability to conduct continuous evaluation. It constructs dual drilling indices based on real-time drilling parameters and uses neural networks to invert rock mass engineering indices. Combined with grouting conditions, it completes groutability classification and spatial modeling. Then, it uses on-site grouting data to dynamically correct the model and evaluation standards, enabling rapid and continuous pre-assessment of rock mass groutability and providing support for intelligent grouting construction and quality control.

[0007] To address the aforementioned deficiencies or improvement needs of existing technologies, as a first aspect of this invention, the present invention provides a rapid evaluation method for rock mass groutability based on drilling parameters, comprising: S1. Obtain the drilling parameter data of the target rock mass during the drilling process, remove the non-drilling state data from the drilling parameter data, and align the remaining valid data according to the drilling depth to construct the drilling parameter depth sequence; S2. Based on the drilling parameter depth sequence, construct a comprehensive drilling index characterizing the integrity of the rock mass structure and a drilling discreteness index characterizing the degree of rock mass fracture development; input the comprehensive drilling index and the drilling discreteness index into the inversion model and map out the rock mass engineering index. S3. Obtain grouting condition data, combine the rock mass engineering indicators and the grouting condition data to evaluate groutability, generate rock mass groutability classification results, and generate groutability spatial distribution characteristics based on the rock mass groutability classification results; S4. Obtain actual grouting data during the actual grouting construction process, and calculate the deviation between the actual grouting data and the rock mass engineering indicators; when the deviation exceeds the preset range, use the deviation to correct the mapping relationship of the inversion model or update the grading limit of the groutability evaluation to complete the dynamic correction.

[0008] Further, in step S1, non-drilling state data is removed from the drilling parameter data, and the remaining valid data is aligned according to the drilling depth, including: Obtain drilling pressure, torque, rotational speed and drilling speed from the drilling parameter data, and calculate the mechanical specific energy during the drilling process by combining the drill bit cross-sectional area; When the drilling pressure is lower than the preset drilling pressure threshold and the drilling speed is lower than the preset drilling speed threshold, or when the mechanical specific energy deviates from the normal drilling baseline by more than the preset energy consumption threshold, it is identified as a non-drilling state such as tripping, connecting a single joint, or idling and is rejected. Obtain the collection timestamps corresponding to the retained valid data and the synchronously recorded continuous drilling depths, and establish a time-depth mapping relationship; Based on the preset target depth resolution, equally spaced depth grid nodes are set. Based on the time-depth mapping relationship, the drilling parameter values ​​at each depth grid node are calculated using an interpolation algorithm, and the drilling parameter data acquired based on time is converted into a drilling parameter depth sequence triggered by depth.

[0009] Furthermore, in S2, the construction of a comprehensive drilling index characterizing the integrity of the rock mass structure and a drilling discreteness index characterizing the degree of rock mass fracture development includes: Set a sliding depth window in the drilling parameter depth sequence; The ratio of the average mechanical specific energy to the average drilling speed within the sliding depth window is used as the comprehensive drilling index to quantify the energy consumption per unit volume of rock mass breaking. The ratio of the fluctuation amplitude of the drill bit torque to the average torque within the sliding depth window, and the ratio of the fluctuation amplitude of the drilling pressure to the average drilling pressure are combined to form the drilling dispersion index, which is used to quantify the degree of uneven stress on the drill bit caused by rock fractures and weak interlayers.

[0010] Furthermore, the training and mapping output process of the inversion model in S2 includes: Based on the test section depth of the water pressure test and the cycle depth of the core logging, the drilling parameter depth sequence of historical boreholes is divided into intervals, and the characteristics of the drilling parameter sequence in each interval are extracted. The drilling parameter sequence features are used as input, and the core logging index and water pressure test permeability of the corresponding interval are used as labels to construct training samples. The water pressure test permeability is then logarithmically transformed to smooth the data distribution. Construct a multi-task neural network model that includes a shared feature extraction layer and two independent task output layers; During training, a joint loss function containing classification and regression loss terms is constructed. Based on the rate of decrease of prediction error of the output layers of the two independent tasks on the validation set, the proportion of the classification and regression loss terms is dynamically adjusted. The model is updated using the joint loss function so that the trained model can map the output rock mass quality index and equivalent permeability respectively.

[0011] Further, in step S3, grouting condition data is acquired, and groutability is evaluated by combining the rock mass engineering indicators and the grouting condition data to generate a rock mass groutability classification result, including: The grouting design pressure, maximum particle size of the grout, and grout viscosity are obtained as the grouting condition data. Calculate the average aperture of rock mass fractures based on the equivalent permeability. Calculate the ratio of the average opening to the maximum particle size of the slurry. When the ratio is less than a preset clogging threshold, it is determined that the slurry cannot be grouted. When the ratio is greater than or equal to the preset blockage threshold, the theoretical diffusion distance of the grout in the crack is calculated based on the average opening, the grouting design pressure, and the grout viscosity. Obtain the design reinforcement radius of the current borehole location, calculate the ratio of the theoretical diffusion distance to the design reinforcement radius, and based on the ratio and the rock mass quality index, match the preset grading standard to generate the rock mass groutability grading result.

[0012] Further, in step S3, generating spatial distribution characteristics of irrigability based on the rock mass irrigability classification results includes: Obtain the borehole coordinates and borehole inclination data for each borehole, and convert the rock mass groutability classification results along the borehole depth into scatter data with three-dimensional spatial coordinates. Obtain the dominant strike of the rock strata in the exploration area; perform spatial interpolation based on the scattered data, and set the search distance parallel to the dominant strike of the rock strata to be greater than the search distance perpendicular to the dominant strike of the rock strata during the interpolation calculation; Based on the interpolation results, a three-dimensional irrigation availability distribution model of the surveyed area is generated.

[0013] Further, in step S4, correcting the mapping relationship of the inversion model or updating the classification boundary of the irrigationability evaluation using the deviation includes: When the deviation exceeds the preset range, the actual grouting data and drilling parameter depth sequence corresponding to the deviation are extracted as incremental samples. The inversion model is fine-tuned online using the incremental samples, and the network structure of the shared feature extraction layer is updated to correct the mapping relationship. Alternatively, the distribution range of the deviation under different irrigation availability levels can be statistically analyzed, and the distribution range can be used to widen or narrow the grading boundary of the irrigation availability evaluation.

[0014] Furthermore, the step of using the incremental samples to fine-tune the inversion model online and updating the network structure of the shared feature extraction layer to correct the mapping relationship specifically includes: Keep the weight parameters of the original backbone network in the shared feature extraction layer unchanged, and add a feature compensation branch at the end of the shared feature extraction layer; A portion of the samples are extracted from the historical training samples and mixed with the incremental samples to construct a fine-tuning dataset, in order to prevent the model from forgetting historical geological patterns; The fine-tuning dataset is input into a shared feature extraction layer containing the feature compensation branch, and only the weight parameters of the feature compensation branch and the weight parameters of the two independent task output layers are updated to complete the online fine-tuning of the inversion model.

[0015] As a second aspect of the present invention, a rapid evaluation system for rock mass groutability based on drilling parameters is also provided, comprising: The drilling parameter sequence construction unit is used to acquire the drilling parameter data of the target rock mass during the drilling process, remove the non-drilling state data in the drilling parameter data, and align the remaining valid data according to the drilling depth to construct the drilling parameter depth sequence. The rock mass engineering index inversion unit is used to construct a comprehensive drilling index characterizing the integrity of the rock mass structure and a drilling discreteness index characterizing the degree of rock mass fracture development based on the drilling parameter depth sequence; the comprehensive drilling index and the drilling discreteness index are input into the inversion model and the rock mass engineering index is mapped and output. The rock mass groutability classification evaluation unit is used to acquire grouting condition data, combine the rock mass engineering indicators and the grouting condition data to evaluate groutability, generate rock mass groutability classification results, and generate groutability spatial distribution characteristics based on the rock mass groutability classification results. The evaluation model dynamic correction unit is used to acquire actual grouting data during the actual grouting construction process, calculate the deviation between the actual grouting data and the rock mass engineering indicators; when the deviation exceeds the preset range, the mapping relationship of the inversion model is corrected or the grading limit of the groutability evaluation is updated to complete the dynamic correction.

[0016] As a third aspect of the invention, a computer-readable storage medium is also provided, on which a computer program is stored, which is executed by a processor as described in any one of the claims, a method for rapid evaluation of rock mass groutability based on drilling parameters.

[0017] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects: 1. The present invention provides a rapid evaluation method for rock mass groutability based on drilling parameters. This method collects drilling parameters throughout the drilling process and filters out data from non-drilling conditions. It then generates a standardized parameter sequence aligned to the same depth using time-depth mapping and interpolation. During on-site drilling, a large amount of invalid data is generated from tripping in and out of the drill string, idle drilling, and single-joint drilling. This data cannot reflect the true mechanical characteristics of the rock mass, and directly using it in calculations would distort the judgment of rock mass characteristics. This method uses multiple thresholds for drilling pressure, drilling speed, and mechanical energy to automatically remove invalid data. It uses borehole depth as the data benchmark, eliminating the problem of uneven sampling intervals caused by time-series sensor acquisition. Preprocessing is completed entirely based on data collected synchronously during drilling, eliminating the need for additional borehole inspection work. Data regularization can be completed synchronously during the borehole construction phase, providing continuous and interference-free basic data for subsequent quantitative calculations of rock mass characteristics, thus avoiding the limitations of traditional detection methods that lag in obtaining geological data.

[0018] 2. The rapid evaluation method for rock mass groutability based on drilling parameters of this invention calculates comprehensive drilling indicators and discrete drilling indicators through a sliding depth window. These two types of indicators are then input into a multi-task neural network inversion model to output rock mass engineering indicators. The sliding window filters out random noise generated by drilling rig vibration and sensor data acquisition. The two types of indicators correspond to the integrity of the rock mass structure and the development of internal fractures, respectively, with the physical representation logic closely aligned with the interaction between the drilling tool and the rock mass. The multi-task neural network shares the underlying feature extraction structure, simultaneously completing rock mass quality classification and permeability regression calculation. The permeability data distribution is optimized through logarithmic transformation, and an adaptive joint loss function balances the two training tasks. No manual fixing of weight parameters is required. Model training is completed based on historical borehole water pressure and core logging samples, continuously outputting rock mass quality indicators and equivalent permeability distributed along the borehole depth, achieving continuous quantitative inversion of rock mass permeability and integrity characteristics.

[0019] 3. The rapid evaluation method for rock mass groutability based on drilling parameters of this invention performs groutability classification and constructs a spatial distribution model by combining grouting condition parameters, and simultaneously relies on on-site grouting measurement data to achieve dynamic correction of the evaluation system. The average fracture aperture is derived based on the equivalent permeability, and the risk of fracture blockage and the grout diffusion range are judged by combining grout particle size, grouting pressure, and grout viscosity. The groutability level is obtained by matching the corresponding classification standard, and then spatial interpolation is completed to generate a three-dimensional distribution model by combining borehole coordinates, borehole inclination, and rock strata strike. During the construction phase, the measured grouting data and the rock mass indicators output by the model are continuously compared. When the deviation exceeds the preset range, incremental sample fine-tuning of the inversion model feature compensation branch is used, or the groutability classification boundary is adjusted according to the deviation distribution. A data closed loop of exploration evaluation and on-site construction is formed throughout the process, continuously adapting to the heterogeneous characteristics of regional geology and steadily improving the degree to which the groutability evaluation results conform to the actual on-site grouting conditions. Attached Figure Description

[0020] Figure 1 This is a flowchart of a rapid evaluation method for rock mass groutability based on drilling parameters according to an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the linear relationship between drilling pressure and torque during the drilling process according to an embodiment of the present invention; M represents torque, which is the rotational torque transmitted from the drill rod to the drill bit during the drilling process; F represents drilling pressure, which is the axial pressure applied to the drill bit during the drilling process; Figure 3 This is a schematic diagram of the system units in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0022] Example 1 Please refer to Figure 1 This embodiment 1 provides a rapid evaluation method for rock mass groutability based on drilling parameters, including: S1. Obtain the drilling parameter data of the target rock mass during the drilling process, remove the non-drilling state data from the drilling parameter data, and align the remaining valid data according to the drilling depth to construct the drilling parameter depth sequence; S2. Based on the drilling parameter depth sequence, construct a comprehensive drilling index characterizing the integrity of the rock mass structure and a drilling discreteness index characterizing the degree of rock mass fracture development; input the comprehensive drilling index and the drilling discreteness index into the inversion model and map out the rock mass engineering index. S3. Obtain grouting condition data, combine the rock mass engineering indicators and the grouting condition data to evaluate groutability, generate rock mass groutability classification results, and generate groutability spatial distribution characteristics based on the rock mass groutability classification results; S4. Obtain actual grouting data during the actual grouting construction process, and calculate the deviation between the actual grouting data and the rock mass engineering indicators; when the deviation exceeds the preset range, use the deviation to correct the mapping relationship of the inversion model or update the grading limit of the groutability evaluation to complete the dynamic correction.

[0023] This embodiment 1 further elaborates on the above steps.

[0024] (1) Construction of drilling parameter sequence During rock drilling, digital drilling rigs or drilling monitoring systems are used to collect real-time data on drilling parameters and the cross-sectional area of ​​the drill bit currently in use. Drilling parameters include: drilling pressure Torque Rotation speed and drilling speed And the drilling parameters are recorded synchronously according to the drilling depth. For example, please refer to... Figure 2This study demonstrates a typical linear relationship model between drilling pressure and torque during drilling. This strong correlation verifies the reliability of the parameter acquisition and also reveals the influence of changes in rock mass properties on drilling parameters—this linear relationship deviates significantly when the integrity of the rock mass structure changes. Based on the obtained physical quantities, the mechanical specific energy during drilling is calculated. This parameter variable represents the total energy consumed per unit volume of rock mass that is broken.

[0025] After acquiring multi-dimensional physical quantities and calculating mechanical specific energy, a data filtering mechanism needs to be established to eliminate invalid data generated outside of drilling operations. The specific quantitative rejection rule is as follows: when drilling pressure is detected... Below the preset drilling pressure threshold And drilling speed Below the preset drilling speed threshold When or when the calculated mechanical specific energy The deviation from the normal drilling baseline exceeds the preset energy consumption threshold. When such conditions are met, the drill string is identified as being pulled up or down, connected to a single joint, or idling, and is therefore rejected.

[0026] After obtaining clean and valid data, it is necessary to create spatial location indexes for these discrete data points to transform the time-dimensional data into spatial-dimensional data. First, extract the collection timestamps corresponding to the retained valid data. And the continuous drilling depth recorded simultaneously ,in Representing the 1 valid data point. By matching data points at the same time... and Establish a continuous time-depth mapping relationship to ensure that each valid acquisition moment can uniquely correspond to a subsurface spatial depth location.

[0027] After establishing a spatial location index, the non-uniformly spaced time-series data is resampled into equally spaced depth sequences to eliminate the uneven data density caused by variations in drilling speed. Equally spaced depth grid nodes are set according to a preset target depth resolution. Based on the established time-depth mapping relationship, an interpolation algorithm is used to calculate the drilling parameter values ​​at each depth grid node. These parameter values ​​represent the interpolated results of drill pressure, torque, rotational speed, drilling rate, or mechanical specific energy at a specific depth location after resampling. All drilling parameter interpolation results are arranged and encapsulated in ascending order of depth grid nodes, and redundant timestamp information in the original data is removed, ultimately generating a depth-triggered drilling parameter depth sequence.

[0028] (2) Rock mass engineering index inversion After obtaining the drilling parameter depth sequence aligned along the borehole depth, a sliding depth window is set in the drilling parameter depth sequence to eliminate interference from high-frequency vibrations of the drilling rig and random noise from sensor acquisition, and to accurately reflect lithological changes within local borehole sections. This window slides downwards with a preset step size as the drilling depth progresses, capturing parameter data within a local depth range. Within the sliding depth window, the mechanical specific energy of each sampling point needs to be calculated first, using the following formula: ,in, For mechanical specific energy, For drilling pressure, For drilling speed, For drill bit torque, For rotational speed, The cross-sectional area of ​​the drill bit is given. Mechanical specific energy combines the axial and rotational work done by the drill bit during rock breaking, and can directly represent the total energy consumption required for the drill bit to break a unit volume of rock.

[0029] Within the sliding depth window, calculate the arithmetic mean of the mechanical specific energy. and the arithmetic mean of drilling speed Dividing the two yields the overall drilling performance index. Its calculation formula is Because drilling consumes a lot of energy and has a slow drilling speed in hard, intact rock masses, while it consumes less energy and has a faster drilling speed in fractured or weak rock masses, the overall drilling performance is better. This method can effectively quantify the macroscopic structural integrity and hardness of rock masses. To further quantify the uneven stress and vibration of the drill string caused by internal fracture development, karst caves, or weak interlayers within the rock mass, drill string torque and drill pressure data are extracted within the same sliding depth window, and the standard deviation of the drill string torque is calculated separately. Arithmetic mean of drill string torque The ratio, and the standard deviation of drill pressure. arithmetic mean of drilling pressure The ratio of these two ratios essentially reflects the degree of parameter variation. To objectively synthesize these two fluctuations without introducing an artificially set scaling factor, the two ratios are squared and the root is taken to calculate the drilling dispersion index. Its calculation formula is When the drill bit passes through fractured zones or interlayers of varying hardness, the drill string experiences severe vibrations and blockages, leading to a significant increase in both torque and drilling pressure fluctuations, and consequently, increased drilling dispersion. This physical phenomenon is used to objectively characterize the degree of rock mass fracture development. Through the above calculation process, the originally high-frequency and chaotic drilling time-series data is transformed into comprehensive drilling indicators that are distributed along the depth and can characterize the physical and mechanical properties of the rock mass. and drilling dispersion index sequence.

[0030] When constructing the training samples for the inversion model, the depth range of the test sections for completed water pressure tests and the depth range of core logging runs in historical boreholes were obtained. Based on this, the drilling parameter depth sequence of historical boreholes was divided into intervals, resulting in multiple independent data intervals. For each data interval, the statistical features of its internal comprehensive drilling index sequence and discrete drilling index sequence were extracted and combined into drilling parameter sequence features, which served as input variables for the inversion model. Simultaneously, the core logging index and water pressure test permeability corresponding to this data interval were used as supervision labels for the model. The core logging index represents discrete rock mass quality classification levels, serving as labels for the classification task. Because the water pressure test permeability values ​​under different geological conditions vary greatly and exhibit a significant long-tail distribution, directly using them for regression training would lead to a decrease in the model's prediction accuracy for low permeability intervals. Therefore, the water pressure test permeability... Perform a logarithmic transformation to convert it into a continuously distributed regression task label. Its calculation formula is This is done to smooth the data distribution and improve the model's ability to fit the full range of permeability.

[0031] A multi-task neural network model was constructed, comprising a shared feature extraction layer and two independent task output layers. Since rock mass quality indicators and equivalent permeability are highly correlated in a geophysical sense, both controlled by the development of fractures and structural integrity of the rock mass, the multi-task architecture allows the shared feature extraction layer to learn the underlying rock mechanical characteristics shared by both, thereby improving the overall generalization ability of the model. The features of the drilling parameter sequence are input to the shared feature extraction layer, and high-dimensional latent features of the input data are extracted through a multi-layer network structure. These extracted high-dimensional latent features are then simultaneously passed to the two independent task output layers. The first independent task output layer is a classification network used to predict core logging indicators based on the high-dimensional latent features, ultimately mapping and outputting the rock mass quality indicators. These rock mass quality indicators refer to the rock mass engineering geological classification results output by the classification network based on core logging indicators (such as rock integrity, joint development degree, and strength characteristics), specifically represented as rock mass quality grades (such as Grade I, Grade II, Grade III, and Grade IV), used to quantify the macroscopic structural integrity and hardness of the rock mass. The second independent task output layer is a regression network, used to predict the log-transformed permeability label based on high-dimensional latent features. The equivalent permeability is then restored through an inverse exponential transform and finally mapped to output. Its reduction calculation formula is: ,in, This represents the predicted permeability value output by the regression network.

[0032] During model training, in order to balance the learning progress of classification and regression tasks and avoid relying on human experience to set a fixed scaling factor, a joint loss function based on homoscedasticity uncertainty is constructed. To update network parameters, the calculation formula is: ,in, For the category loss item, For the regression loss term, For the uncertainty parameters of the classification task, These are the uncertainty parameters for the regression task. In the initial training phase, and The parameters are initialized to the same baseline value. Within each training iteration, the model automatically calculates the gradient using backpropagation, increasing the uncertainty parameter for tasks with larger prediction errors. This mathematically reduces the proportion of that task in the total loss, preventing a single task from dominating the gradient update direction. The joint loss function is used to calculate the total gradient and simultaneously update the shared feature extraction layer, the output layers of the two independent tasks, and the uncertainty parameter. and The network parameters are adjusted until the model converges, thereby completing the training of the inversion model and enabling it to output rock mass engineering indicators, including rock mass quality indicators and equivalent permeability, based on real-time drilling parameters.

[0033] (3) Evaluation of the groutability of rock mass When conducting a rock mass groutability assessment, it is necessary to extract grouting condition data from the grouting engineering design plan, specifically including obtaining the grouting design pressure. Maximum particle size of slurry and slurry viscosity At the same time, the equivalent permeability obtained through inversion was studied. This variable represents the rock mass permeability per unit length of borehole under a unit water head. To convert permeability into fracture geometry, the average aperture of the rock mass fractures needs to be calculated. This variable represents the equivalent average width of the fracture network within the rock mass, and its quantitative calculation formula is set as follows: ,in Represents the dynamic viscosity coefficient of water at the test water temperature. The length of the test section representing the pressure test. It represents the density of water.

[0034] The average aperture of rock mass fractures was obtained. Next, the risk of physical blockage of grout particles in the cracks needs to be assessed, which serves as the primary criterion for determining groutability. First, the average aperture is calculated. With the maximum particle size of the slurry The ratio, and compare this ratio with a preset blockage threshold. For comparison, this threshold represents the critical size ratio that prevents slurry particles from forming an arching effect and clogging at the fracture inlet. When the calculated ratio is less than the preset clogging threshold... If the rock mass fissures in the current depth range are deemed too narrow, preventing effective grout injection, an "uninjectable" evaluation result is generated. If this ratio is greater than or equal to a preset grouting threshold... When the slurry meets the geometric conditions to enter the crack, it indicates that the slurry has entered the next stage of quantitative evaluation of diffusion capacity.

[0035] After confirming that the grout possesses the geometric conditions for entering the fractures, it is necessary to predict the actual penetration range of the grout within the rock fracture network to assess the effectiveness of the grouting. Based on previously obtained parameters, the theoretical diffusion distance of the grout within the fractures is calculated. This variable represents the farthest effective penetration radius that the grout can reach under a specific grouting pressure, and its quantitative calculation formula is set as follows: ,in This represents the equivalent radius of the grouting hole. After calculating the theoretical permeability range, the design reinforcement radius for the current hole location is obtained. This variable represents the radial range within which the borehole requires reinforcement according to engineering design requirements. Subsequently, the theoretical diffusion distance is calculated. With design reinforcement radius The ratio, and combined with the rock mass quality indicators obtained from previous inversion. (This variable represents the overall integrity and mechanical quality grade of the rock mass), and is input into a preset grading standard matrix for matching, ultimately generating a rock mass groutability grading result containing excellent, good, medium, and poor grades.

[0036] After completing the stepwise evaluation of a single borehole along its depth, the one-dimensional evaluation results need to be extended to three-dimensional geological space to reveal the spatial distribution characteristics of groutability throughout the entire exploration area. First, the borehole coordinates and inclination data are acquired. Using spatial geometric trigonometric relationships, the one-dimensional groutability classification results along the borehole depth are mapped and converted into scattered data with three-dimensional spatial coordinates. To accurately reflect the anisotropy of rock mass permeability characteristics, the dominant strike of the rock strata in the exploration area needs to be obtained. Based on this, spatial interpolation calculations are performed. In the search ellipsoid parameter settings of the interpolation algorithm, the search distance parallel to the dominant strike of the rock strata is set to be greater than the search distance perpendicular to the dominant strike. This aligns with the physical law that groundwater and grout have low permeability resistance and long diffusion distance along the strike in layered rock masses. Finally, based on the interpolation calculation results, a three-dimensional groutability distribution model is generated that intuitively reflects the differences in grouting difficulty within the exploration area.

[0037] (4) Dynamic correction of the evaluation model After the grouting construction is carried out, the actual grouting data is obtained through an automatic grouting recorder and converted into the actual equivalent permeability. This variable represents the actual permeability of the rock mass calculated based on the actual amount of grout absorbed. Simultaneously, the equivalent permeability prediction value output from previous models is retrieved. And calculate the absolute deviation between the two. Therefore, this deviation needs to be compared with a preset deviation threshold. The comparison is performed, and this threshold represents the upper limit of the allowable prediction error for the project. When the calculated deviation... Greater than the preset deviation threshold When this occurs, it indicates that the model's prediction accuracy has decreased under the current geological conditions. The system automatically extracts the actual grouting data and the synchronously acquired drilling parameter depth sequence corresponding to this deviation, and packages them into an incremental sample set. .

[0038] After obtaining the latest real geological feedback data, the mapping relationship of the inversion model is corrected by reconstructing the neural network feature extraction path. During the specific implementation of the network structure update, the data processing system maintains the weight parameters of the original backbone network in the shared feature extraction layer. The model is frozen to prevent disruption of the learned fundamental laws of rock fracture mechanics. A new feature compensation branch is added in parallel at the end of this shared feature extraction layer, with its initialized weight parameters denoted as... To compensate for the information gaps caused by new geological features, data was collected from historical training sample sets. A representative sample is drawn from the sample using a stratified sampling strategy, and then compared with the incremental sample set. Hybrid construction of fine-tuning dataset This hybrid mechanism prevents the model from forgetting historical geological patterns when learning new stratigraphic features.

[0039] After constructing a fine-tuned dataset containing both old and new geological features, the computing device can initiate an online fine-tuning program for the inversion model to complete local updates of parameter weights. The data processing system will then fine-tune the dataset. The input is fed into a shared feature extraction layer that includes a feature compensation branch. During backpropagation gradient computation, only the weight parameters of the feature compensation branch are allowed to be updated. and the weight parameters of the two independent task output layers downstream (i.e., the classification network and the regression network). and Through this local parameter update strategy, the model can use the feature compensation branch to specifically fit the special geological noise and unmodeled features contained in the incremental samples without changing the original main feature extraction logic, thereby correcting the nonlinear mapping relationship between drilling parameters and rock mass engineering indicators.

[0040] In addition to structurally correcting the neural network model, the data processing system can also adaptively adapt to on-site construction deviations by adjusting the evaluation standard thresholds. The computing equipment comprehensively analyzes the distribution range of historical predicted deviations under different irrigation feasibility levels, calculating the mean and standard deviation of the deviations at each level. Based on this statistical distribution characteristic, the data processing system can dynamically widen or narrow the original irrigation feasibility evaluation grading boundaries using the distribution range. For example, when the actual deviations at a certain level generally lean towards the better side, the lower limit threshold of that level will be automatically widened, thereby quickly completing the on-site adaptive calibration of the irrigation feasibility evaluation system without increasing the computational power consumption of the model.

[0041] Example 2 Please refer to Figure 3 This embodiment 2 provides a rapid evaluation system for rock mass groutability based on drilling parameters, including: The drilling parameter sequence construction unit is used to acquire the drilling parameter data of the target rock mass during the drilling process, remove the non-drilling state data in the drilling parameter data, and align the remaining valid data according to the drilling depth to construct the drilling parameter depth sequence. The rock mass engineering index inversion unit is used to construct a comprehensive drilling index characterizing the integrity of the rock mass structure and a drilling discreteness index characterizing the degree of rock mass fracture development based on the drilling parameter depth sequence; the comprehensive drilling index and the drilling discreteness index are input into the inversion model and the rock mass engineering index is mapped and output. The rock mass groutability classification evaluation unit is used to acquire grouting condition data, combine the rock mass engineering indicators and the grouting condition data to evaluate groutability, generate rock mass groutability classification results, and generate groutability spatial distribution characteristics based on the rock mass groutability classification results. The evaluation model dynamic correction unit is used to acquire actual grouting data during the actual grouting construction process, calculate the deviation between the actual grouting data and the rock mass engineering indicators; when the deviation exceeds the preset range, the mapping relationship of the inversion model is corrected or the grading limit of the groutability evaluation is updated to complete the dynamic correction.

[0042] Example 3 This embodiment 3 also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it can implement any step of a rapid evaluation method for rock mass groutability based on drilling parameters.

[0043] The computer-readable storage medium may include various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0044] For a description of the computer-readable storage medium provided in this application, please refer to the above method embodiments; further details will not be repeated here.

[0045] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A rapid evaluation method for the groutability of rock mass based on drilling parameters, characterized in that, include: S1. Obtain the drilling parameter data of the target rock mass during the drilling process, remove the non-drilling state data from the drilling parameter data, and align the remaining valid data according to the drilling depth to construct the drilling parameter depth sequence; S2. Based on the drilling parameter depth sequence, construct a comprehensive drilling index characterizing the integrity of the rock mass structure and a drilling discreteness index characterizing the degree of rock mass fracture development; input the comprehensive drilling index and the drilling discreteness index into the inversion model and map out the rock mass engineering index. S3. Obtain grouting condition data, combine the rock mass engineering indicators and the grouting condition data to evaluate groutability, generate rock mass groutability classification results, and generate groutability spatial distribution characteristics based on the rock mass groutability classification results; S4. Obtain actual grouting data during the actual grouting construction process, and calculate the deviation between the actual grouting data and the rock mass engineering indicators; when the deviation exceeds the preset range, use the deviation to correct the mapping relationship of the inversion model or update the grading limit of the groutability evaluation to complete the dynamic correction.

2. The rapid evaluation method for rock mass groutability based on drilling parameters according to claim 1, characterized in that, In step S1, non-drilling state data is removed from the drilling parameter data, and the remaining valid data is aligned according to the drilling depth, including: Obtain drilling pressure, torque, rotational speed and drilling speed from the drilling parameter data, and calculate the mechanical specific energy during the drilling process by combining the drill bit cross-sectional area; When the drilling pressure is lower than the preset drilling pressure threshold and the drilling speed is lower than the preset drilling speed threshold, or when the mechanical specific energy deviates from the normal drilling baseline by more than the preset energy consumption threshold, it is identified as a non-drilling state such as tripping, connecting a single joint, or idling and is rejected. Obtain the collection timestamps corresponding to the retained valid data and the synchronously recorded continuous drilling depths, and establish a time-depth mapping relationship; Based on the preset target depth resolution, equally spaced depth grid nodes are set. Based on the time-depth mapping relationship, the drilling parameter values ​​at each depth grid node are calculated using an interpolation algorithm, and the drilling parameter data acquired based on time is converted into a drilling parameter depth sequence triggered by depth.

3. The rapid evaluation method for rock mass groutability based on drilling parameters according to claim 1, characterized in that, In S2, a comprehensive drilling index characterizing the integrity of the rock mass structure and a drilling dispersion index characterizing the degree of rock mass fracture development are constructed, including: Set a sliding depth window in the drilling parameter depth sequence; The ratio of the average mechanical specific energy to the average drilling speed within the sliding depth window is used as the comprehensive drilling index to quantify the energy consumption per unit volume of rock mass breaking. The ratio of the fluctuation amplitude of the drill bit torque to the average torque within the sliding depth window, and the ratio of the fluctuation amplitude of the drilling pressure to the average drilling pressure are combined to form the drilling dispersion index, which is used to quantify the degree of uneven stress on the drill bit caused by rock fractures and weak interlayers.

4. The rapid evaluation method for rock mass groutability based on drilling parameters according to claim 1, characterized in that, The training and mapping output process of the inversion model in S2 includes: Based on the test section depth of the water pressure test and the cycle depth of the core logging, the drilling parameter depth sequence of historical boreholes is divided into intervals, and the characteristics of the drilling parameter sequence in each interval are extracted. The drilling parameter sequence features are used as input, and the core logging index and water pressure test permeability of the corresponding interval are used as labels to construct training samples. The water pressure test permeability is then logarithmically transformed to smooth the data distribution. Construct a multi-task neural network model that includes a shared feature extraction layer and two independent task output layers; During training, a joint loss function containing classification and regression loss terms is constructed. Based on the rate of decrease of prediction error of the output layers of the two independent tasks on the validation set, the proportion of the classification and regression loss terms is dynamically adjusted. The model is updated using the joint loss function so that the trained model can map the output rock mass quality index and equivalent permeability respectively.

5. The rapid evaluation method for rock mass groutability based on drilling parameters according to claim 1, characterized in that, In step S3, grouting condition data is acquired, and groutability is evaluated by combining the rock mass engineering indicators and the grouting condition data to generate a rock mass groutability classification result, including: The grouting design pressure, maximum particle size of the grout, and grout viscosity are obtained as the grouting condition data. Calculate the average aperture of rock mass fractures based on the equivalent permeability. Calculate the ratio of the average opening to the maximum particle size of the slurry. When the ratio is less than a preset clogging threshold, it is determined that the slurry cannot be grouted. When the ratio is greater than or equal to the preset blockage threshold, the theoretical diffusion distance of the grout in the crack is calculated based on the average opening, the grouting design pressure, and the grout viscosity. Obtain the design reinforcement radius of the current borehole location, calculate the ratio of the theoretical diffusion distance to the design reinforcement radius, and based on the ratio and the rock mass quality index, match the preset grading standard to generate the rock mass groutability grading result.

6. The rapid evaluation method for rock mass groutability based on drilling parameters according to claim 1, characterized in that, In step S3, the spatial distribution characteristics of irrigability are generated based on the rock mass irrigability classification results, including: Obtain the borehole coordinates and borehole inclination data for each borehole, and convert the rock mass groutability classification results along the borehole depth into scatter data with three-dimensional spatial coordinates. Obtain the dominant strike of the rock strata in the exploration area; perform spatial interpolation based on the scattered data, and set the search distance parallel to the dominant strike of the rock strata to be greater than the search distance perpendicular to the dominant strike of the rock strata during the interpolation calculation; Based on the interpolation results, a three-dimensional irrigation availability distribution model of the surveyed area is generated.

7. The rapid evaluation method for rock mass groutability based on drilling parameters according to claim 1, characterized in that, In step S4, correcting the mapping relationship of the inversion model or updating the classification boundary of the irrigationability evaluation using the deviation includes: When the deviation exceeds the preset range, the actual grouting data and drilling parameter depth sequence corresponding to the deviation are extracted as incremental samples. The inversion model is fine-tuned online using the incremental samples, and the network structure of the shared feature extraction layer is updated to correct the mapping relationship. Alternatively, the distribution range of the deviation under different irrigation availability levels can be statistically analyzed, and the distribution range can be used to widen or narrow the grading boundary of the irrigation availability evaluation.

8. The rapid evaluation method for rock mass groutability based on drilling parameters according to claim 7, characterized in that, The step of using the incremental samples to fine-tune the inversion model online and updating the network structure of the shared feature extraction layer to correct the mapping relationship specifically includes: Keep the weight parameters of the original backbone network in the shared feature extraction layer unchanged, and add a feature compensation branch at the end of the shared feature extraction layer; A portion of the samples are extracted from the historical training samples and mixed with the incremental samples to construct a fine-tuning dataset, in order to prevent the model from forgetting historical geological patterns; The fine-tuning dataset is input into a shared feature extraction layer containing the feature compensation branch, and only the weight parameters of the feature compensation branch and the weight parameters of the two independent task output layers are updated to complete the online fine-tuning of the inversion model.

9. A rapid evaluation system for the groutability of rock mass based on drilling parameters, characterized in that, include: The drilling parameter sequence construction unit is used to acquire the drilling parameter data of the target rock mass during the drilling process, remove the non-drilling state data in the drilling parameter data, and align the remaining valid data according to the drilling depth to construct the drilling parameter depth sequence. The rock mass engineering index inversion unit is used to construct a comprehensive drilling index characterizing the integrity of the rock mass structure and a drilling discreteness index characterizing the degree of rock mass fracture development based on the drilling parameter depth sequence; the comprehensive drilling index and the drilling discreteness index are input into the inversion model and the rock mass engineering index is mapped and output. The rock mass groutability classification evaluation unit is used to acquire grouting condition data, combine the rock mass engineering indicators and the grouting condition data to evaluate groutability, generate rock mass groutability classification results, and generate groutability spatial distribution characteristics based on the rock mass groutability classification results. The evaluation model dynamic correction unit is used to acquire actual grouting data during the actual grouting construction process, calculate the deviation between the actual grouting data and the rock mass engineering indicators; when the deviation exceeds the preset range, the mapping relationship of the inversion model is corrected or the grading limit of the groutability evaluation is updated to complete the dynamic correction.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer program is executed by a processor according to any one of claims 1-8, a rapid evaluation method for rock mass groutability based on drilling parameters.

Images