Method and equipment for evaluating and improving tunnel blasting effect and medium

Abstract

The invention discloses an evaluation and improvement method and device for a tunnel blasting effect and a medium, and relates to the technical field of tunnel engineering.The evaluation and improvement method comprises the steps that a tunnel coupling data set is collected, space registration and three-dimensional shape reconstruction are conducted, fracture space distribution is formed, risk probability mapping is conducted on the fracture space distribution, and a fracture risk thermodynamic diagram is output; the fracture risk thermodynamic diagram and tunnel rock mass data are input into a digital twin blasting model, a load mapping layer carries out load gradient distribution, a rock mass response deduction layer carries out stress wave propagation simulation and damage accumulation quantification, an optimized blasting number instruction set is generated, and a drill jumbo is driven to execute blasting operation and vibration monitoring according to the optimized blasting instruction set. And actual blasting effect parameters are obtained. According to the method, accurate simulation of geological conditions and optimal adjustment of blasting parameters are realized through the digital twin blasting model, and meanwhile, refined and interpretable evaluation of the blasting effect is realized through vibration monitoring and error propagation modeling of the actual blasting effect.

Description

Technical Field

[0001] This invention relates to the field of tunnel engineering technology, and in particular to a method, equipment, and medium for evaluating and improving the effectiveness of tunnel blasting. Background Technology

[0002] In the field of tunnel engineering, blasting operations are a core technology, and their effectiveness directly impacts construction safety, tunneling efficiency, and support costs. With the development of intelligent technologies, traditional experience-based blasting design is gradually evolving towards a data-driven intelligent control model. In recent years, methods for optimizing blasting parameters based on the combination of geological survey data and digital twins have made some progress, enabling visualized simulations of the blasting construction process. Simultaneously, sensor technology is being used for blasting vibration monitoring and damage assessment, improving the objectivity of blasting effect evaluation.

[0003] Current methods for evaluating and improving tunnel blasting effects suffer from two main shortcomings: First, existing blasting response prediction models largely rely on simplified geological assumptions or static numerical simulations, resulting in insufficient consideration of the spatial heterogeneity of geological discontinuities in blasting load distribution, thus affecting the accuracy of blasting effect optimization. Second, existing technologies only focus on pre-blasting design optimization or post-blasting apparent effect monitoring, failing to establish a reverse tracing mechanism from actual blasting response deviations to blasting design parameters, thus hindering effective evaluation of blasting effects. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for improving the evaluation of tunnel blasting effects, addressing the problems of low accuracy in optimizing blasting effects and insufficient evaluation capabilities.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for evaluating and improving the effect of tunnel blasting, which includes: collecting a tunnel coupling dataset, performing spatial registration and three-dimensional morphological reconstruction to form a spatial distribution of fractures, mapping the spatial distribution of fractures to risk probabilities, and outputting a fracture risk heat map. The heat map of fracture risk and tunnel rock mass data are input into the digital twin blasting model. The load mapping layer performs load gradient allocation, and the rock mass response simulation layer performs stress wave propagation simulation and damage accumulation quantification to generate an optimized blasting instruction set. Based on the optimized blasting instruction set, the rock drilling rig is driven to perform blasting operations and vibration monitoring to obtain actual blasting effect parameters; error propagation modeling and root cause tracing are performed on the actual blasting effect parameters to form a response deviation quantification matrix; Perform global sensitivity analysis on the response deviation quantization matrix to generate a blasting sensitivity cloud map; perform crack damage region clustering and risk level mapping on the blasting sensitivity cloud map to output a blasting damage assessment report.

[0007] As a preferred embodiment of the method for evaluating and improving the tunnel blasting effect described in this invention, the tunnel coupling dataset includes infrared images of the tunnel face, tunnel laser point cloud data, and tunnel rock mass data.

[0008] As a preferred embodiment of the method for evaluating and improving the tunnel blasting effect described in this invention, the output fracture risk heat map specifically includes the following steps: spatially registering the infrared image of the tunnel face and the tunnel laser point cloud data to obtain temperature-geometric joint features; and reconstructing the three-dimensional morphology based on the temperature-geometric joint features to form the spatial distribution of fractures. Kriging space interpolation is performed on the spatial distribution of fractures to generate a fracture density field; the fracture density field is then mapped to risk probability to output a fracture risk heat map.

[0009] As a preferred embodiment of the method for evaluating and improving the tunnel blasting effect described in this invention, the step of generating an optimized blasting instruction set specifically includes the following steps: using a Transformer architecture to build a load mapping layer, and using a GNN-CNN hybrid network to build a rock mass response inference layer. Parameter sharing and gradient coupling are performed on the load mapping layer and the rock mass response inference layer to construct a digital twin blasting model; The fracture risk heat map and tunnel rock mass data are input into the digital twin blasting model. The load mapping layer is used to distribute the load gradient through the dynamic response spectrum inversion method to generate the borehole load distribution matrix. The rock mass response simulation layer uses the PLAXIS finite element engine to simulate stress wave propagation and quantify damage accumulation, and obtains over-excavation cloud map and rockburst risk value. The borehole load distribution matrix, over-excavation cloud map, and rock burst risk value are optimized through multi-objective optimization to generate an optimized blasting instruction set.

[0010] As a preferred embodiment of the method for evaluating and improving the tunnel blasting effect described in this invention, the formation of the response deviation quantification matrix specifically includes the following steps: according to the optimized blasting instruction set, driving the rock drilling rig to perform automatic drilling and detonation networking, and simultaneously applying a triaxial vibration sensor for vibration monitoring to obtain actual blasting effect parameters; Error propagation modeling is performed on the actual blasting effect parameters to obtain error distribution characteristics, and root cause analysis is conducted to form a response deviation quantification matrix.

[0011] As a preferred embodiment of the method for evaluating and improving the tunnel blasting effect described in this invention, the generation of the blasting sensitivity cloud map specifically refers to applying the Sobol operator to perform global sensitivity analysis on the response deviation quantization matrix, obtaining the first-order sensitivity index, and projecting the first-order sensitivity index using a Gaussian kernel function to generate the blasting sensitivity cloud map.

[0012] As a preferred embodiment of the method for evaluating and improving the tunnel blasting effect described in this invention, the output of the blasting damage assessment report specifically includes the following steps: performing connected component analysis on the blasting sensitivity cloud map to output the fracture damage area; performing density peak clustering on the fracture damage area to output the damage cluster distribution. The risk level of the damage cluster distribution is mapped, and a blasting damage assessment report is output.

[0013] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the method for evaluating and improving the tunnel blasting effect as described in the first aspect of the present invention.

[0014] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the method for evaluating and improving the tunnel blasting effect as described in the first aspect of the present invention.

[0015] The beneficial effects of this invention are as follows: By using a digital twin blasting model to achieve load gradient distribution and dynamic extrapolation of rock mass response, accurate simulation of geological conditions and optimized adjustment of blasting parameters are realized, thereby improving the accuracy and adaptability of blasting parameter optimization. Simultaneously, through vibration monitoring and error propagation modeling of actual blasting effects, a closed-loop inversion mechanism from execution feedback to design parameters is formed, enabling refined and interpretable evaluation of blasting effects. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart for evaluating and improving methods for tunnel blasting effectiveness.

[0018] Figure 2 A flowchart for optimizing blasting parameters.

[0019] Figure 3 This is a flowchart for a blasting sensitivity cloud map.

[0020] Figure 4 This is a flowchart for a blasting damage assessment report. Detailed Implementation

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0022] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0023] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0024] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a method for evaluating and improving the effect of tunnel blasting, comprising the following steps: S1. Collect the tunnel coupling dataset, perform spatial registration and three-dimensional morphological reconstruction to form the spatial distribution of fractures, map the spatial distribution of fractures to risk probability, and output a fracture risk heat map.

[0025] Specifically, the steps include the following: S1.1 Collect the tunnel coupling dataset and perform preprocessing.

[0026] In specific operations, The tunnel coupling dataset includes infrared images of the tunnel face, laser point cloud data of the tunnel, and tunnel rock mass data; the infrared images of the tunnel face were acquired using an infrared thermal imager; the laser point cloud data of the tunnel was acquired using a 3D laser scanner; and the tunnel rock mass data was acquired using ground-penetrating radar and a core drill.

[0027] Next, the tunnel coupling dataset is preprocessed. Furthermore, for the infrared image of the tunnel face, noise suppression is performed through non-uniformity correction to eliminate the influence of noise, and histogram equalization is applied to enhance contrast and improve the recognizability of crack edges. Simultaneously, median filtering is used to perform smoothing processing to reduce thermal noise caused by environmental interference. For tunnel laser point cloud data, voxel filtering is used for downsampling, and neighbor interpolation is used to remove outliers, thereby improving the processing efficiency and purity of tunnel laser point cloud data. At the same time, the PTP timestamp synchronization protocol is used to perform time alignment to ensure the spatiotemporal consistency of tunnel laser point cloud data. For tunnel rock mass data, Min-Max normalization is used to unify the dimensions and eliminate the dimensional differences between tunnel rock mass data. Linear interpolation is applied to fill in missing values, and a moving exponential average is performed simultaneously using a sliding window to enhance the stability of tunnel rock mass data.

[0028] S1.2. Spatial registration and three-dimensional morphological reconstruction are performed on the preprocessed infrared image of the tunnel face and the laser point cloud data of the tunnel to form the spatial distribution of the fracture.

[0029] In specific operations, Spatial registration is performed on the preprocessed tunnel face infrared image and tunnel laser point cloud data using the SIFT (Scale Invariant Feature Transform) algorithm. Furthermore, feature decomposition is performed on the preprocessed tunnel face infrared image to obtain infrared image feature points, while voxel uniform sampling is performed on the preprocessed tunnel laser point cloud data to obtain point cloud feature points. The infrared image feature points and point cloud feature points are then integrated to form a SIFT feature descriptor. Bidirectional nearest neighbor search is performed on the SIFT feature descriptor pairs to obtain matching feature point pairs; affine transformation is performed on the matching feature point pairs to generate temperature-geometric registration data; feature space projection is performed on the temperature-geometric registration data to output temperature-geometric joint features; It should be noted that affine transformation refers to the process of aligning and calibrating the position and orientation of matching feature point pairs through spatial coordinate transformation.

[0030] After registration, the moving least squares method is used to reconstruct the three-dimensional morphology of the temperature-geometric joint features. Furthermore, the temperature-geometric joint features are mapped into phase space to obtain crack feature data. The crack feature data are then fitted with a surface and connected in space to generate a three-dimensional reconstructed crack structure. The crack space parameters of the three-dimensional reconstructed crack structure are extracted and Gaussian kernel density is estimated to generate the crack space distribution. It should be noted that phase space mapping refers to the process of coordinate transformation and feature dimensionality reduction of the temperature-geometric joint features in order to project the high-dimensional temperature-geometric joint features into three-dimensional space; Gaussian kernel density estimation refers to the process of describing the probability density distribution of the fracture space parameters and smoothing the data.

[0031] S1.3. Perform Kriging space interpolation on the spatial distribution of fractures to generate a fracture density field; perform risk probability mapping on the fracture density field to output a fracture risk heat map.

[0032] In specific operations, Kriging spatial interpolation is performed on the spatial distribution of fractures. Further, the interpolation coefficients are initialized, for example, by setting the range parameter to 5 meters and the nugget effect to 0.1. The spatial location coordinates, density parameters, and timestamps of the fracture spatial distribution are extracted. A spatial interpolation grid is constructed using the spatial location coordinates as the X-axis in 3D space, the density parameters as the Y-axis, and the timestamps as the Z-axis. Based on the interpolation coefficients, continuous interpolation is performed on the spatial interpolation grid to generate a preliminary density field. Boundary constraint optimization is applied to the preliminary density field to output the fracture density field. It should be noted that boundary constraint optimization refers to the process of spatial constraint correction and outlier removal of the initial density field.

[0033] Next, risk probability mapping is performed on the fracture density field. Furthermore, the parameters of the fracture distribution probability field are normalized to generate a standard fracture density vector. The standard fracture density vector is then weighted by spatial correlation to form a risk correlation matrix. The feature dimension of the risk correlation matrix is ​​compressed to form a risk feature space. The risk feature space is then projected by probability density to output a fracture risk heat map. It should be noted that spatial correlation weighting refers to the process of assigning weights and associating risks to the standard fracture density vector through risk impact factors (defined based on the statistical regularity of historical fracture risk heat maps, with an exemplary value range of 0.5-0.7); probability density projection refers to the process of reducing the dimensionality of the risk feature space and transforming its density distribution.

[0034] S2. Input the fracture risk heat map and tunnel rock mass data into the digital twin blasting model. The load mapping layer performs load gradient allocation, and the rock mass response simulation layer performs stress wave propagation simulation and damage accumulation quantification to generate an optimized blasting instruction set.

[0035] Specifically, the steps include the following: S2.1 Construct and train a digital twin blasting model.

[0036] In specific operations, In the PyTorch framework, the Transformer architecture is invoked via the nn.Transformer parameter and initialized: for example, the number of attention heads is set to 8, the hidden layer dimension is set to 512, and the feedforward network dimension is set to 2048. The dynamic response spectrum inversion method is then applied after the Transformer architecture for load gradient allocation, and feature normalization is performed using LayerNorm layers to complete the construction of the load mapping layer. The nn.GraphConv and nn.Conv3d functions are used to invoke the GNN-CNN (Graph Neural Network-Convolutional Neural Network) hybrid network. The GNN-CNN hybrid network is initialized: for example, the number of graph convolutional layers is set to 3, the kernel size is set to 5×5×5, and the feature concatenation dimension is set to 256. The PLAXIS finite element engine is embedded into the GNN-CNN hybrid network to simulate stress wave propagation and damage accumulation, completing the construction of the rock mass response inference layer. Skip connections are used to transfer features across the load mapping layer and the rock mass response simulation layer to obtain a multi-source feature matrix. The multi-source feature matrix is ​​then parameter-shared to generate a joint feature representation. The joint feature representation is then projected into the feature space to obtain coupling weights. The coupling weights are then sparsified to generate regularized weights. Based on the regularized weights, the load mapping layer and the rock mass response simulation layer are then gradient-coupled to complete the construction of the digital twin blasting model. Next, the digital twin blasting model is trained. Further, historical fracture risk heatmaps and tunnel rock mass data are divided into a sample set, a training set, and a validation set. On the sample set, a data augmenter is used to amplify features, and batch sampling is performed using a sliding window to form training batch data. On the training set, forward propagation is used to update the parameters of the training batch data, and a learning rate scheduler is applied synchronously for dynamic adjustment to obtain optimized digital twin blasting model parameters. On the validation set, an early stopping mechanism is used to monitor the performance of the optimized digital twin blasting model parameters, obtaining the validation set loss. When the validation set loss exceeds the convergence threshold for several consecutive rounds (e.g., 5 times), training terminates, and the trained digital twin blasting model is output synchronously. It should be noted that the convergence threshold is defined based on the historical fluctuation range of the validation set loss, with an exemplary range of 0.05-0.1.

[0037] S2.2 The load mapping layer uses the dynamic response spectrum inversion method to distribute the load gradient and generate the borehole load distribution matrix.

[0038] In specific operations, The fracture risk heatmap and tunnel rock mass data are input into the load mapping layer of the digital twin blasting model through the TorchVision interface and the Geometric data pipeline, respectively. Feature interaction and fusion are performed through the Transformer architecture. Furthermore, feature alignment and feature association are performed on the fracture risk heatmap and tunnel rock mass data to obtain multi-source coupling features. The GeLU activation function is used to perform nonlinear transformation on the multi-source coupling features to obtain a high-dimensional feature space. The high-dimensional feature space is then weighted and fused to output the risk-mechanical coupling matrix. The risk-mechanical coupling matrix is ​​transformed to the frequency domain using Fourier transform. Then, the risk-mechanical coupling matrix is ​​filtered in the frequency domain and the energy spectrum is normalized to obtain the filtered spectrum. The filtered spectrum is then decomposed to obtain the fundamental frequency component. Finally, the fundamental frequency component is reconstructed to output the reconstructed spectrum.

[0039] Load gradient allocation is performed on the reconstructed spectrum using the dynamic response spectrum inversion method. Furthermore, frequency domain energy integration and weighted norm calculations are performed on the reconstructed spectrum to obtain the equivalent static load value. The specific mathematical format is as follows. ; in, This represents the equivalent static load value. This represents the total number of frequency bands in the reconstructed spectrum. This represents the frequency band index of the reconstructed spectrum. Indicates the first Energy integral value of each frequency band Indicates the first Engineering weighting coefficients for each frequency band; It should be noted that the engineering weight coefficient is used to adjust the contribution weight of different frequency bands to the blasting load. It is based on the statistical definition of the frequency band energy ratio of historical equivalent static load values, and the exemplary value range is 0.2–0.8.

[0040] The equivalent static load values ​​are spatially discretely sampled to form load distribution nodes, and interpolation is performed on the load distribution nodes to obtain the spatial load distribution. LayerNorm normalization is applied to perform standardization scaling on the spatial load distribution to eliminate dimensional differences and output the borehole load distribution matrix. The row elements of the borehole load distribution matrix represent the axial position of the borehole, and the column elements represent the charge linear density, which can accurately guide the design and optimization of blasting parameters.

[0041] S2.3. The rock mass response simulation layer uses the PLAXIS finite element engine to simulate stress wave propagation and quantify damage accumulation, and obtains over-excavation cloud maps and rockburst risk values.

[0042] In specific operations, The borehole load distribution matrix and tunnel rock mass data are input into a GNN-CNN hybrid network for joint feature processing. Further, the borehole load distribution matrix and tunnel rock mass data are dimensionally aligned to unify dimensions, and dynamic neighborhood aggregation is performed through a graph convolutional network (GNN) to obtain load-rock mass coupling features. Global average pooling is then applied to the load-rock mass coupling features to generate optimized feature representations. Local mechanical feature extraction is performed on the optimized feature representations through a three-dimensional convolutional network (CNN) to form dynamic rock mass response features. Finally, the dynamic rock mass response features are parameterized to output rock mass response parameters.

[0043] The rock mass response parameters are transmitted to the PLAXIS finite element engine to construct a three-dimensional geological simulation structure for the tunnel. Further, geological properties are mapped onto the rock mass response parameters to obtain a set of geological material properties. Simultaneously, spatial features are extracted and geometrically reconstructed from the rock mass response parameters to obtain a three-dimensional geological geometry. Boundary conditions are applied to the rock mass response parameter set to generate an initial mechanical state. Using the geological material property set as the basis for material definition, the three-dimensional geological geometry as the geometric foundation, and the initial mechanical state as the mechanical benchmark, a three-dimensional geological simulation structure for the tunnel is constructed. It should be noted that geological attribute mapping refers to the process of quantitatively correlating rock mass response parameters and geomechanical parameters (acquired through a triaxial testing instrument); boundary condition loading refers to the process of setting the stress state and applying displacement constraints to rock mass response parameters based on engineering geological conditions (obtained through on-site geostress testing).

[0044] The PLAXIS finite element engine was used to simulate stress wave propagation in the three-dimensional geological structure of the tunnel, obtaining stress variation data. The Hoek-Brown strength criterion was then used to calculate the rock mass damage index from the stress variation data to obtain the rockburst risk value. The specific mathematical format is as follows. ; in, Indicates the rockburst risk value. Indicates the maximum principal stress. Indicates the minimum principal stress. Indicates the uniaxial compressive strength of the rock mass. Indicates the influence index of confining pressure; It should be noted that the maximum principal stress is obtained by extracting the peak values ​​from the stress variation data; the minimum principal stress is obtained by extracting the valley values ​​from the stress variation data. The uniaxial compressive strength of rock mass refers to the maximum axial pressure that tunnel rock mass can withstand under unconfined conditions, which is obtained by conducting uniaxial compression tests on the tunnel rock mass; the confining pressure influence index is defined based on the geological strength index (GSI) of tunnel rock mass, with an exemplary value range of 0.3-0.7.

[0045] Damage accumulation is quantified based on rockburst risk value. Further, the rockburst risk value is integrated over time to obtain the cumulative damage index. Based on the cumulative damage index, damage is marked on the three-dimensional geological simulation structure of the tunnel to obtain the loss and damage area. The loss and damage area is clustered and the boundary is extracted to generate the damage area distribution. The damage area distribution is rendered in three dimensions to output the over-excavation cloud map. It should be noted that regional clustering refers to the process of performing local density statistics on damaged or destroyed areas and merging adjacent areas.

[0046] S2.4. Perform multi-objective optimization on the borehole load distribution matrix, over-excavation cloud map, and rock burst risk value to generate an optimized blasting instruction set.

[0047] In specific operations, Based on the borehole load distribution matrix, over-excavation cloud map, and rockburst risk value, an optimization framework is built using the NSGA-II multi-objective genetic algorithm. Furthermore, the borehole load distribution matrix is ​​parametrically sampled to obtain load distribution parameters, and the over-excavation cloud map is volume integrated to obtain the risk zone volume. The load distribution parameters, risk zone volume, and rockburst risk value are used as three optimization objectives to construct the optimization framework. In the optimization framework, load distribution parameters, risk zone volume, and rockburst risk value are randomly combined to generate multiple sets of blasting parameters. These multiple sets of blasting parameters are then input into the PLAXIS finite element engine for rapid simulation to obtain a multi-objective response dataset.

[0048] The multi-target response dataset is sorted non-dominated and crowded to obtain the Pareto front solution set; simulated binary crossover and polynomial mutation are performed on the Pareto front solution set to generate the offspring population; elite retention iteration is performed on the offspring population to output the Pareto optimal solution set; the Pareto optimal solution set is used as the optimal blasting parameter combination, and parameter formatting encoding is performed on the optimal blasting parameter combination to generate the optimized blasting instruction set; It should be noted that non-dominated sorting refers to the process of performing dominance comparison and frontier hierarchy sorting on a multi-objective response dataset; crowding measure refers to the process of performing neighborhood density statistics and optimizing the distribution of the solution set on a multi-objective response dataset; and elite retention iteration refers to the process of updating the offspring population and retaining dominant individuals.

[0049] S3. Based on the optimized blasting instruction set, drive the rock drilling rig to perform blasting operation and vibration monitoring to obtain actual blasting effect parameters; perform error propagation modeling and root cause tracing on the actual blasting effect parameters to form a response deviation quantification matrix.

[0050] Specifically, the steps include the following: S3.1. Based on the optimized blasting instruction set, drive the rock drilling rig to perform automatic drilling and detonation networking, and simultaneously use a three-axis vibration sensor to monitor vibration and obtain actual blasting effect parameters.

[0051] In specific operations, The optimized blasting command set is transmitted to the control center of the drilling rig, and the drilling coordinate parameters, charge parameters, and detonation timing parameters are extracted. Based on the drilling coordinate parameters, a total station and an IMU inertial measurement unit are used for combined navigation to complete high-precision positioning drilling. Based on the charge parameters, the amount of emulsion explosive is precisely controlled by a charging robotic arm to achieve charge gradient control. At the same time, based on the detonation timing parameters, electronic detonators are used for millisecond-delay detonation of each hole to ensure the controllability of blasting energy release. During the blasting process, triaxial vibration sensors are used to collect peak vibration velocity (PPV), dominant frequency, and vibration duration in real time, and integrate them to generate actual blasting effect parameters.

[0052] S3.2. Model the error propagation of the actual blasting effect parameters, obtain the error distribution characteristics, trace the root cause, and form a response deviation quantification matrix.

[0053] In specific operations, Error propagation modeling is performed on the actual blasting effect parameters. Furthermore, the actual blasting effect parameters and the target blasting effect parameters are compared differentially to generate a blasting error vector. Kernel density estimation is performed on the blasting error vector to obtain the error propagation probability distribution. The feature space of the error propagation probability distribution is reconstructed to output the error distribution features. It should be noted that the target blasting effect parameters are defined based on the theoretical expected values ​​of the optimized blasting instruction set; kernel density estimation refers to the process of bandwidth optimization and probability density surface fitting of the blasting error vector; feature space reconstruction refers to the process of dimensional compression and feature reorganization of the error propagation probability distribution.

[0054] Independent component analysis (ICA) is used to trace the root causes of the error distribution characteristics. Further, eigenvalue decomposition is performed on the error distribution characteristics, and the top B eigenvalues ​​are extracted and sorted by variance to obtain the independent error components. The Pearson correlation coefficient formula is used to calculate the correlation strength of the independent error components, yielding the error influence coefficient. The specific mathematical format is as follows. ; in, Indicates the error influence coefficient. Represents the target explosion effect vector. Represents independent error components. Represents independent error components The mean, This represents the actual explosion effect vector. Represents the explosion effect vector The mean, Represents independent error components standard deviation Represents the actual explosion effect vector Standard deviation; It should be noted that the target blasting effect vector is obtained by vectorizing the target blasting effect parameters; the actual blasting effect vector is obtained by vectorizing the actual blasting effect parameters. The error influence coefficients are weighted and normalized to obtain normalized influence weights; the normalized influence weights are then integrated into a matrix structure to form a response deviation quantification matrix.

[0055] S4. Perform global sensitivity analysis on the response deviation quantization matrix to generate a blasting sensitivity cloud map; perform crack damage region clustering and risk level mapping on the blasting sensitivity cloud map, and output a blasting damage assessment report.

[0056] Specifically, the steps include the following: S4.1 Apply the Sobol operator to perform global sensitivity analysis on the response deviation quantization matrix to obtain the first-order sensitivity index. Project the first-order sensitivity index using a Gaussian kernel function to generate a blasting sensitivity cloud map.

[0057] In specific operations, The Sobol operator is applied to measure the variance contribution of the response bias quantization matrix to obtain the global sensitivity parameter. The probability density integral of the global sensitivity parameter is then performed to output the first-order sensitivity exponent. The specific mathematical formula is as follows. ; in, This represents the first-order sensitivity index. Represents the response bias quantization matrix. Indicates the index of the global sensitivity parameter. Represents global sensitivity parameters The response deviation quantization matrix, This represents the mean of the response bias quantization matrix; It should be noted that the Sobol operator is called directly through the sobol.analyze function in the SALib library.

[0058] Next, Gaussian kernel density projection is applied to the first-order sensitivity index. Further, the first-order sensitivity index is discretized and sampled to obtain a sensitivity sampling point set. Surface fitting is performed on the sensitivity sampling point set to generate a sensitivity density curve. Contour mapping is performed on the sensitivity density curve to generate a sensitivity cloud map. It should be noted that surface fitting refers to the process of spatial interpolation and surface smoothing of the sensitivity sampling point set through radial basis functions; contour mapping refers to the process of contour extraction and feature space projection of the sensitivity density curve.

[0059] S4.2 Perform connected component analysis on the blasting sensitivity cloud map and output the fracture damage area; perform density peak clustering on the fracture damage area and output the damage cluster distribution; In specific operations, The `findContours` function from the OpenCV library is used to perform connected component analysis on the blast sensitivity contour map. Further, binarization segmentation is performed on the blast sensitivity contour map. For example, the sensitivity index of the blast sensitivity contour map is extracted. When the sensitivity index exceeds the binarization threshold, the corresponding region in the blast sensitivity contour map is classified as the foreground region; when the sensitivity index is below the binarization threshold, the corresponding region in the blast sensitivity contour map is classified as the background region. Contours are extracted from the divided foreground and background regions to obtain the damage region contour. Polygon approximation is performed on the damage region contour to generate the fracture boundary vector. Spatial coordinate transformation is performed on the fracture boundary vector to output the fracture damage region. It should be noted that the binarization threshold is defined based on the sensitivity distribution law of the blast sensitivity cloud map, and the exemplary value range is 0.3-0.5; polygon approximation refers to the process of simplifying the vertices and optimizing the geometry of the damaged area contour.

[0060] Spatial distribution nodes of the fracture damage region are extracted and weighted to obtain candidate cluster centers. These candidate cluster centers are then screened, for example, by performing neighborhood statistics and interval detection to obtain local density values ​​and proximity distance values. When the local density value exceeds a reasonable density range (e.g., 40–80 points / cubic meter) and the proximity distance value exceeds the standard distance range (e.g., 0.5–2 meters), the candidate cluster center is defined as the cluster center. Based on the cluster centers, neighborhood expansion and density clustering are performed on the fracture damage region to generate a damage cluster distribution. It should be noted that the reasonable density range is defined based on the spatial density distribution of the fracture damage area; the standard distance range is defined based on the spatial distribution pattern of the fracture damage area; and density clustering refers to the process of local density fusion and cluster boundary optimization in the fracture damage area.

[0061] S4.3 Map the risk level of the damage cluster distribution and output a blasting damage assessment report.

[0062] In specific operations, Geometric parameters of the damage cluster distribution are extracted and weighted to obtain geometric feature vectors. Based on the geometric feature vectors, the risk level of the damage cluster distribution is classified using a three-level threshold. For example, the range of the first-level threshold is set to 0.8-1.0, the range of the second-level threshold is 0.5-0.8, and the range of the third-level threshold is 0-0.5. When the geometric feature vector is within the range of the first-level threshold, the corresponding damage cluster is defined as a high-risk area; when the geometric feature vector is within the range of the second-level threshold, the corresponding damage cluster is defined as a medium-risk area; and when the geometric feature vector is within the range of the third-level threshold, the corresponding damage cluster is defined as a low-risk area. It should be noted that the third-level threshold is defined based on the density distribution characteristics of the damage cluster distribution.

[0063] The risk levels are mapped using feature rendering to generate a risk level distribution map; the risk level distribution map is then integrated into a report structure to output a blasting damage assessment report.

[0064] This embodiment also provides a computer device applicable to the method for evaluating and improving the effect of tunnel blasting, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method for evaluating and improving the effect of tunnel blasting as proposed in the above embodiment.

[0065] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0066] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for evaluating and improving the tunnel blasting effect as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0067] In summary, this invention achieves accurate simulation of geological conditions and optimized adjustment of blasting parameters by using a digital twin blasting model to realize load gradient distribution and dynamic extrapolation of rock mass response, thereby improving the accuracy and adaptability of blasting parameter optimization. Simultaneously, through vibration monitoring and error propagation modeling of actual blasting effects, a closed-loop inversion mechanism from execution feedback to design parameters is formed, enabling refined and interpretable evaluation of blasting effects.

[0068] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for evaluating and improving the effectiveness of tunnel blasting, characterized in that: include, Collect a coupled tunnel dataset, perform spatial registration and 3D morphological reconstruction to form a spatial distribution of fractures, map the spatial distribution of fractures to risk probabilities, and output a fracture risk heat map. The heat map of fracture risk and tunnel rock mass data are input into the digital twin blasting model. The load mapping layer performs load gradient allocation, and the rock mass response simulation layer performs stress wave propagation simulation and damage accumulation quantification to generate an optimized blasting instruction set. Based on the optimized blasting instruction set, the rock drilling rig is driven to perform blasting operations and vibration monitoring to obtain actual blasting effect parameters; error propagation modeling and root cause tracing are performed on the actual blasting effect parameters to form a response deviation quantification matrix; Perform global sensitivity analysis on the response deviation quantization matrix to generate a blasting sensitivity cloud map; perform crack damage region clustering and risk level mapping on the blasting sensitivity cloud map to output a blasting damage assessment report.

2. The method for evaluating and improving the tunnel blasting effect as described in claim 1, characterized in that: The tunnel coupling dataset includes infrared images of the tunnel face, tunnel laser point cloud data, and tunnel rock mass data.

3. The method for evaluating and improving the tunnel blasting effect as described in claim 2, characterized in that: The output fracture risk heat map specifically includes the following steps. Spatial registration is performed between infrared images of the tunnel face and laser point cloud data of the tunnel to obtain temperature-geometric joint features; three-dimensional morphology is reconstructed based on temperature-geometric joint features to form the spatial distribution of fractures; Kriging space interpolation is performed on the spatial distribution of fractures to generate a fracture density field; the fracture density field is then mapped to risk probability to output a fracture risk heat map.

4. The method for evaluating and improving the tunnel blasting effect as described in claim 1, characterized in that: The specific construction process of the digital twin blasting model is as follows. The load mapping layer is built using the Transformer architecture, and the rock mass response inference layer is built using a GNN-CNN hybrid network. A digital twin blasting model is constructed by sharing parameters and coupling gradients between the load mapping layer and the rock mass response inference layer.

5. The method for evaluating and improving the tunnel blasting effect as described in claim 4, characterized in that: The process of generating the optimized blasting instruction set specifically includes the following steps. The fracture risk heat map and tunnel rock mass data are input into the digital twin blasting model. The load mapping layer is used to distribute the load gradient through the dynamic response spectrum inversion method to generate the borehole load distribution matrix. The rock mass response simulation layer uses the PLAXIS finite element engine to simulate stress wave propagation and quantify damage accumulation, and obtains over-excavation cloud map and rockburst risk value. The borehole load distribution matrix, over-excavation cloud map, and rock burst risk value are optimized through multi-objective optimization to generate an optimized blasting instruction set.

6. The method for evaluating and improving the tunnel blasting effect as described in claim 1, characterized in that: The formation of the response deviation quantization matrix specifically includes the following steps. Based on the optimized blasting instruction set, the rock drilling rig is driven to perform automatic drilling and detonation networking, and a triaxial vibration sensor is used simultaneously for vibration monitoring to obtain actual blasting effect parameters. Error propagation modeling is performed on the actual blasting effect parameters to obtain error distribution characteristics, and root cause analysis is conducted to form a response deviation quantification matrix.

7. The method for evaluating and improving the tunnel blasting effect as described in claim 1, characterized in that: The generation of the blasting sensitivity cloud map specifically refers to applying the Sobol operator to perform global sensitivity analysis on the response deviation quantization matrix, obtaining the first-order sensitivity index, and projecting the first-order sensitivity index using a Gaussian kernel function to generate the blasting sensitivity cloud map.

8. The method for evaluating and improving the tunnel blasting effect as described in claim 1, characterized in that: The output blast damage assessment report specifically includes the following steps. Connectivity analysis is performed on the blast sensitivity cloud map to output the fracture damage area; density peak clustering is performed on the fracture damage area to output the damage cluster distribution; The risk level of the damage cluster distribution is mapped, and a blasting damage assessment report is output.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the method for evaluating and improving the tunnel blasting effect according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the method for evaluating and improving the tunnel blasting effect as described in any one of claims 1 to 8.

Images