A structural surface strength prediction system and method based on residual network

Abstract

The present invention discloses a structural surface strength prediction system and method based on a residual network, which relates to the field of geological engineering technology and aims to solve the technical problems that current methods rely on on-site measurements and empirical formulas, and have low efficiency and insufficient precision. The system comprises a residual network feature extraction unit that uses a residual network to extract features from GPR images, a spatially dilated convolutional neural network and a dynamic DBSCAN clustering unit that combine spatially dilated convolution to enhance the receptive field to capture a wider range of contextual information, a structural surface strength prediction unit, and a "random-prediction-deterministic" structural surface network algorithm generation unit. The system and method for predicting rock structural surface strength based on GPR three-instantaneous features and residual networks proposed in the present invention achieve accurate prediction of rock structural surface strength and intelligent generation of structural surface networks by fusing multi-source data, applying deep learning technology and optimizing algorithm processes, providing strong support for the design and construction of rock engineering.

Description

Technical Field

[0001] The present invention relates to the field of geological engineering technology, and more particularly to a structural surface strength prediction system and method based on a residual network. Background Art

[0002] The distribution of structural planes in a rock mass, such as joints and fissures, plays a crucial role in the stability of the surrounding rock and excavation face during tunnel construction. Numerous researchers, both domestically and internationally, have used statistical methods to establish structural plane network models of unexcavated rock masses and to calculate rock mass stability using discrete element methodologies and key block theory. The accuracy of rock mass stability analysis results is determined by the degree to which the structural plane network model closely matches the actual situation. However, due to the randomness of the spatial distribution of structural planes, constructing an accurate model based solely on probability theory and statistics is difficult. Some researchers have improved the accuracy of the model by adding constraints to the generation of the structural plane network model based on known structural plane distribution information. In practical applications, traditional manual structural plane survey methods are time-consuming, have limited coverage, and are inherently risky. Furthermore, the excavation face provides limited information, lacking reliable data along the depth direction.

[0003] Traditional methods rely on field measurements and empirical formulas, resulting in low efficiency and insufficient precision. The development of deep learning technology, particularly the successful application of residual networks in image processing, has provided new insights for the automatic and efficient prediction of structural surface strength. To address this, we propose a structural surface strength prediction system and method based on residual networks. Summary of the Invention

[0004] The purpose of the present invention is to provide a structural surface strength prediction system and method based on residual network to solve the technical problems that the current method relies on field measurement and empirical formulas, has low efficiency and insufficient accuracy.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: a residual network-based structural surface strength prediction system, comprising a residual network feature extraction unit that utilizes a residual network to extract features from GPR images, a spatially dilated convolutional neural network and a dynamic DBSCAN clustering unit that combine spatially dilated convolution to enhance the receptive field to capture broader contextual information, a structural surface strength prediction unit that extracts structural surface features to construct a strength prediction model, and a "random-prediction-deterministic" structural surface network algorithm generation unit that optimizes the accuracy of rock mass structural surface network description through prediction data;

[0006] The residual network feature extraction unit is connected to the spatial expansion convolutional neural network and the dynamic DBSCAN clustering unit, and the spatial expansion convolutional neural network and the dynamic DBSCAN clustering unit are connected to the structural surface strength prediction unit, and the "random-prediction-deterministic" structural surface network algorithm generation unit is connected to the structural surface strength prediction unit.

[0007] Preferably, the residual network feature extraction unit includes a data preprocessing module for normalizing the input GPR image, a residual network construction module for designing the residual network architecture, and a feature extraction module for extracting feature maps. The data preprocessing module is connected to the residual network construction module, and the residual network construction module is connected to the feature extraction module.

[0008] Preferably, the spatially dilated convolutional neural network and the dynamic DBSCAN clustering unit include a spatially dilated convolution module for capturing contextual features and a dynamic DBSCAN clustering module for identifying different structural surface areas, and the spatially dilated convolution module is connected to the dynamic DBSCAN clustering module.

[0009] Preferably, the structural surface strength prediction unit includes a feature integration module for forming a complete feature vector, a prediction model construction module for learning the mapping relationship between the feature vector and the structural surface strength, and a strength prediction module for calculating the structural surface strength prediction value. The feature integration module is connected to the prediction model construction module, and the prediction model construction module is connected to the strength prediction module.

[0010] Preferably, the "random-prediction-determination" structural surface network algorithm generation unit includes a random generation module for generating multiple groups of candidate structural surface network samples, a prediction and evaluation module for predicting and evaluating candidate samples, and an optimal solution determination module for selecting the optimal candidate samples. The random generation module is connected to the prediction and evaluation module, and the prediction and evaluation module is connected to the optimal solution determination module.

[0011] The rock mass structural surface strength prediction system and method based on GPR three-instantaneous characteristics and residual network proposed in this invention achieves accurate prediction of rock mass structural surface strength and intelligent generation of structural surface network by integrating multi-source data, applying deep learning technology and optimizing algorithm process, providing strong support for the design and construction of rock engineering.

[0012] A structural surface strength prediction method based on residual network includes the following steps:

[0013] S1, data acquisition and sample set construction;

[0014] GPR image data is collected on the rock surface or in a borehole using GPR equipment. Three instantaneous features, namely instantaneous amplitude (A), instantaneous frequency (F), and instantaneous phase (Φ), are extracted from the GPR images. The GPR images, their three instantaneous features, and spatial position information are then combined to construct a multi-domain feature vector sample set for training and testing.

[0015] S2, multi-domain feature vector extraction of radar images;

[0016] The three instantaneous features (A, F, Φ) of the GPR image are fused with the spatial position information (x, y) to form a multi-domain feature vector V, where t is time;

[0017] V(t,x,y)=[A(t,x,y),F(t,x,y),Φ(t,x,y),x,y]

[0018] S3, network structure design and residual network model construction;

[0019] S301. Residual Network Design: Design a deep residual network model containing multiple residual units;

[0020]

[0021] Among them, x l is the input of the lth layer;

[0022] and is the weight of the convolutional layer;

[0023] Conv is the convolution operation;

[0024] BN stands for batch normalization;

[0025] ReLU is the activation function;

[0026] S302, residual neural network features: extracting high-level features through residual learning mechanism;

[0027] S4, numerical simulation setup and deep learning training;

[0028] Numerical simulation experiments were set up under different geological conditions to generate a large number of GPR images and their corresponding three-instantaneous characteristic data. The simulated data and some measured data were then used to train the residual network model to make a preliminary prediction of the structural surface strength.

[0029] S5. Analysis of deep learning results;

[0030] S501, Model evaluation: Evaluate the prediction performance of the model through cross-validation and confusion matrix;

[0031] S502. Feature Importance Analysis: Analyze the impact of different features on prediction results and optimize feature selection and model structure;

[0032] S6, intelligent extraction algorithm for rock mass structural surface information;

[0033] S601, spatial multi-scale convolutional point cloud classification: using spatial multi-scale convolutional neural network to classify point cloud data;

[0034] S602, dynamic DBSCAN clustering: performing dynamic DBSCAN clustering on the candidate regions;

[0035] Clusters=DBSCAN(V,∈,MinPts)

[0036] Among them, V is a set of multi-domain feature vectors;

[0037] ∈ is the neighborhood radius;

[0038] MinPts is the minimum number of points;

[0039] S7, structural surface segmentation and information extraction;

[0040] S701, structural surface segmentation: In the structural surface segmentation stage, an image segmentation algorithm is used, such as threshold-based segmentation, edge detection, or image segmentation technology. Taking threshold-based segmentation as an example, a threshold function is defined to distinguish between structural surface and non-structural surface areas.

[0041]

[0042] Where V(x,y) is the multi-domain eigenvector at point (x,y);

[0043] Threshold is the preset segmentation threshold;

[0044] Mask(x,y) is the binary mask after segmentation;

[0045] 1 indicates the structural surface area;

[0046] 0 represents the non-structural surface area;

[0047] S702, information extraction: After the structural surface is segmented, its geometric and physical properties are extracted, such as the area A of the structural surface. Area , inclination θ and dip φ;

[0048]

[0049] Among them, (x1, y1) and (x2, y2) are the coordinates of adjacent pixels on the structure surface;

[0050] PixelArea is the actual area represented by each pixel after calibration based on image resolution and scanning distance;

[0051] Weight is a weight based on pixel position or feature importance;

[0052] S8, structural surface model test and effect analysis;

[0053] S801. Structural surface model test: Construct a physical or digital model in the laboratory to simulate the structural surface of the actual rock mass. By controlling variables such as the number, spacing, and inclination of the structural surfaces, observe and record the model's behavior, such as strength and deformation.

[0054] S802, Structural Surface Effect Analysis: Use statistical methods to evaluate the effects of different structural surface parameters on model behavior;

[0055]

[0056] Among them, SS Between is the between-group sum of squares, i.e., the difference between different combinations of structural surface parameters;

[0057] SS Total is the total sum of squares;

[0058] S9, case application and effect analysis;

[0059] S901. Application: Apply the algorithm to actual rock mass engineering projects, such as tunnels and slopes, to perform structural surface identification and strength prediction;

[0060] S902, Example Results Analysis: Evaluate the accuracy and reliability of the algorithm by comparing the prediction results with field measured data, such as drilling coring and in-situ testing;

[0061]

[0062] Among them, Accuracy is the accuracy;

[0063] RMSE is the root mean square error;

[0064] N is the sample size;

[0065] y' i is the predicted value;

[0066] y i is the true value;

[0067] S10, bidirectional constrained random structure surface generation;

[0068] S1001, random simulation: Generate the initial structural surface network based on geostatistical principles;

[0069] S1002, Bidirectional Constraint: Constrain the generated structural surface network based on geological laws and engineering requirements, using geological structural lines as constraints to ensure that the direction of the structural surface is consistent with the geological structural lines;

[0070] ConstrainedNetwork=RandomNetwork∩GeologicalConstraints

[0071] Among them, ConstrainedNetwork is the constrained structural surface network;

[0072] RandomNetwork is a randomly generated structural surface network;

[0073] GeologicalConstraints are geological constraints;

[0074] S11, statistical characteristics of structural surface distribution and “random-predictive-deterministic” network generation;

[0075] Analyze the distribution statistical characteristics of the structural surface, such as density and directionality, and then generate the final structural surface network through random simulation, predictive optimization and deterministic adjustment;

[0076] S12, field test verification;

[0077] S1201. Field test: Drill holes and conduct in-situ tests at the rock mass engineering site to obtain actual structural surface data;

[0078] S1202. Result analysis: Compare the field measured data with the predicted results, analyze the sources of errors, and propose improvement measures;

[0079] S13, analysis of structural surface network inversion results;

[0080] Based on field test data, an inversion analysis of the structural surface network is performed to verify the accuracy of the prediction model. The model parameters and algorithm process are then adjusted according to the inversion results to achieve continuous improvement in prediction accuracy.

[0081] Preferably, the S1 data acquisition and sample set construction further includes the following steps:

[0082] S101, GPR data acquisition: Use GPR equipment to collect GPR image data on the rock surface or in the borehole;

[0083] S102, three-instantaneous feature extraction: extracting three-instantaneous features of instantaneous amplitude (A), instantaneous frequency (F) and instantaneous phase (Φ) from the GPR image;

[0084] A(t,x,y)=Amplitude(I(t,x,y))

[0085] F(t,x,y)=Frequency(I(t,x,y))

[0086] Φ(t,x,y)=Phase(I(t,x,y))

[0087] Where I(t,x,y) is the GPR image data;

[0088] t is time;

[0089] x, y are spatial coordinates;

[0090] S103, sample set construction: Combine the GPR image and its three-moment features and spatial position information to construct a multi-domain feature vector sample set for training and testing.

[0091] Preferably, the S4 numerical simulation setting and deep learning training further includes the following steps:

[0092] S401, numerical simulation: set up numerical simulation experiments under different geological conditions to generate a large number of GPR images and their corresponding three-instantaneous characteristic data;

[0093] S402, deep learning training: using simulated data and some measured data to train the residual network model to make a preliminary prediction of the structural surface strength;

[0094]

[0095] Where N is the number of samples;

[0096] y' i is the predicted value;

[0097] y i is the true value.

[0098] Preferably, the S11 structural surface distribution statistical characteristics and the "random-predictive-deterministic" network generation further include the following steps:

[0099] S1101. Statistical feature analysis: Analyze the distribution characteristics of the structural surface, such as density and direction distribution;

[0100]

[0101] Among them, Density is the structural surface density;

[0102] Area of ​​Interest is the area of ​​the region of interest;

[0103] N Areas is the number of structural surfaces;

[0104] φi is the tendency of the i-th structural surface;

[0105] S1102, "random-predictive-deterministic" network generation;

[0106] S1102a, random simulation: generating an initial network;

[0107] S1102b, prediction optimization: adjusting network parameters based on prediction data, such as GPR images and geological data;

[0108] S1102c, Deterministic Adjustment: Finalize the network based on geological laws and engineering requirements.

[0109] Preferably, the S13 structural surface network inversion result analysis further includes the following steps:

[0110] S1301, Inversion Analysis: Based on the field test data, inversion analysis is performed on the structural surface network, that is, the actual situation of the underground structural surface is inferred based on the observed data;

[0111] S1302, optimization iteration: adjusting model parameters and algorithm flow according to the inversion results to improve prediction accuracy;

[0112]

[0113] Among them, UpdatedParameters is the updated parameter;

[0114] OriginalParameters is the original parameter;

[0115] α is the learning rate;

[0116] is the gradient of the loss function with respect to the parameters.

[0117] Compared with the prior art, the present invention has the following beneficial effects:

[0118] 1. The rock mass structural surface strength prediction system and method based on GPR three-instantaneous characteristics and residual network proposed in this invention achieves accurate prediction of rock mass structural surface strength and intelligent generation of structural surface network by integrating multi-source data, applying deep learning technology and optimizing algorithm process, providing strong support for the design and construction of rock engineering.

[0119] 2. The present invention is also verified through field tests. Tests are carried out at multiple actual rock engineering sites to test the practicality and generalization ability of the algorithm and system, and to compare the field measured data with the prediction results, analyze the source of errors, and propose improvement measures. Based on the field test data, the structural surface network is inverted and analyzed to verify the accuracy of the prediction model. Then, the model parameters and algorithm flow are adjusted according to the inversion results to achieve continuous improvement in prediction accuracy. This realizes the verification and improvement of the model through forward modeling and inversion to further improve the accuracy of rock structural surface strength prediction.

[0120] 3. The present invention also generates a large number of GPR images and their corresponding three-instant feature data through the settings of numerical simulation, deep learning training and deep learning result analysis, by setting up numerical simulation experiments under different geological conditions, and then uses the simulated data and part of the measured data to train the residual network model to make a preliminary prediction of the structural surface strength, thereby evaluating the prediction performance of the model through cross-validation, confusion matrix and other methods, and analyzing the impact of different features on the prediction results, optimizing feature selection and model structure, so as to provide the model with deep learning training and further optimize the accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0121] Figure 1 This is a schematic block diagram of a structural surface strength prediction system based on a residual network in Example 1 of the present invention. DETAILED DESCRIPTION

[0122] To facilitate those skilled in the art to understand the technical solution of the present invention, the technical solution of the present invention is further described with reference to the accompanying drawings.

[0123] Example 1:

[0124] like Figure 1 As shown, the present invention provides a structural surface strength prediction system based on a residual network, comprising a residual network feature extraction unit for extracting features from GPR images using a residual network, a spatial dilation convolutional neural network and a dynamic DBSCAN clustering unit for combining spatial dilation convolution to enhance the receptive field to capture broader contextual information, a structural surface strength prediction unit for extracting structural surface features to construct a strength prediction model, and a "random-prediction-deterministic" structural surface network algorithm generation unit for optimizing the accuracy of rock mass structural surface network description through prediction data;

[0125] The residual network feature extraction unit is connected to the spatial expansion convolutional neural network and the dynamic DBSCAN clustering unit, and the spatial expansion convolutional neural network and the dynamic DBSCAN clustering unit are connected to the structural surface strength prediction unit, and the "random-prediction-deterministic" structural surface network algorithm generation unit is connected to the structural surface strength prediction unit.

[0126] In an embodiment of the present invention, the residual network feature extraction unit includes a data preprocessing module for normalizing the input GPR image, a residual network construction module for designing the residual network architecture, and a feature extraction module for extracting feature maps. The data preprocessing module is connected to the residual network construction module, and the residual network construction module is connected to the feature extraction module.

[0127] In an embodiment of the present invention, the spatially dilated convolutional neural network and the dynamic DBSCAN clustering unit include a spatially dilated convolution module for capturing contextual features and a dynamic DBSCAN clustering module for identifying different structural surface areas, and the spatially dilated convolution module is connected to the dynamic DBSCAN clustering module.

[0128] In an embodiment of the present invention, the structural surface strength prediction unit includes a feature integration module for forming a complete feature vector, a prediction model construction module for learning the mapping relationship between the feature vector and the structural surface strength, and a strength prediction module for calculating the structural surface strength prediction value. The feature integration module is connected to the prediction model construction module, and the prediction model construction module is connected to the strength prediction module.

[0129] In an embodiment of the present invention, the "random-prediction-determination" structural surface network algorithm generation unit includes a random generation module for generating multiple groups of candidate structural surface network samples, a prediction and evaluation module for predicting and evaluating candidate samples, and an optimal solution determination module for selecting the optimal candidate samples. The random generation module is connected to the prediction and evaluation module, and the prediction and evaluation module is connected to the optimal solution determination module.

[0130] Example 2:

[0131] The present invention provides a structural surface strength prediction method based on a residual network according to embodiment 1, comprising the following steps:

[0132] S1, data acquisition and sample set construction;

[0133] GPR image data is collected on the rock surface or in a borehole using GPR equipment. Three instantaneous features, namely instantaneous amplitude (A), instantaneous frequency (F), and instantaneous phase (Φ), are extracted from the GPR images. The GPR images, their three instantaneous features, and spatial position information are then combined to construct a multi-domain feature vector sample set for training and testing.

[0134] S2, multi-domain feature vector extraction of radar images;

[0135] The three instantaneous features (A, F, Φ) of the GPR image are fused with the spatial position information (x, y) to form a multi-domain feature vector V, where t is time;

[0136] V(t,x,y)=[A(t,x,y),F(t,x,y),Φ(t,x,y),x,y]

[0137] S3, network structure design and residual network model construction;

[0138] S301. Residual Network Design: Design a deep residual network model containing multiple residual units;

[0139]

[0140] Among them, x l is the input of the lth layer;

[0141] and is the weight of the convolutional layer;

[0142] Conv is the convolution operation;

[0143] BN stands for batch normalization;

[0144] ReLU is the activation function;

[0145] S302, residual neural network features: extracting high-level features through residual learning mechanism;

[0146] S4, numerical simulation setup and deep learning training;

[0147] Numerical simulation experiments were set up under different geological conditions to generate a large number of GPR images and their corresponding three-instantaneous characteristic data. The simulated data and some measured data were then used to train the residual network model to make a preliminary prediction of the structural surface strength.

[0148] S5. Analysis of deep learning results;

[0149] S501, Model Evaluation: Evaluate the prediction performance of the model through cross-validation and confusion matrix;

[0150] S502. Feature Importance Analysis: Analyze the impact of different features on prediction results and optimize feature selection and model structure;

[0151] S6, intelligent extraction algorithm for rock mass structural surface information;

[0152] S601, spatial multi-scale convolutional point cloud classification: using spatial multi-scale convolutional neural network to classify point cloud data;

[0153] S602, dynamic DBSCAN clustering: performing dynamic DBSCAN clustering on the candidate regions;

[0154] Clusters=DBSCAN(V,∈,MinPts)

[0155] Among them, V is a set of multi-domain feature vectors;

[0156] ∈ is the neighborhood radius;

[0157] MinPts is the minimum number of points;

[0158] Through the settings of numerical simulation, deep learning training, and deep learning result analysis, by setting up numerical simulation experiments under different geological conditions, a large number of GPR images and their corresponding three-instantaneous feature data are generated. Then, the residual network model is trained using the simulated data and some measured data to make a preliminary prediction of the structural surface strength. The prediction performance of the model is evaluated through cross-validation, confusion matrix, and other methods. The impact of different features on the prediction results is analyzed, and the feature selection and model structure are optimized, so that the model can be deeply learned and trained, and the accuracy of the model is further optimized.

[0159] S7, structural surface segmentation and information extraction;

[0160] S701, structural surface segmentation: In the structural surface segmentation stage, an image segmentation algorithm is used, such as threshold-based segmentation, edge detection, or image segmentation technology. Taking threshold-based segmentation as an example, a threshold function is defined to distinguish between structural surface and non-structural surface areas.

[0161]

[0162] Where V(x,y) is the multi-domain eigenvector at point (x,y);

[0163] Threshold is the preset segmentation threshold;

[0164] Mask(x,y) is the binary mask after segmentation;

[0165] 1 indicates the structural surface area;

[0166] 0 represents the non-structural surface area;

[0167] S702, information extraction: After the structural surface is segmented, its geometric and physical properties are extracted, such as the area A of the structural surface. Area , inclination θ and dip φ;

[0168]

[0169] Among them, (x1, y1) and (x2, y2) are the coordinates of adjacent pixels on the structure surface;

[0170] PixelArea is the actual area represented by each pixel after calibration based on image resolution and scanning distance;

[0171] Weight is a weight based on pixel position or feature importance;

[0172] S8, structural surface model test and effect analysis;

[0173] S801. Structural surface model test: Construct a physical or digital model in the laboratory to simulate the structural surface of the actual rock mass. By controlling variables such as the number, spacing, and inclination of the structural surfaces, observe and record the model's behavior, such as strength and deformation.

[0174] S802, Structural Surface Effect Analysis: Use statistical methods to evaluate the effects of different structural surface parameters on model behavior;

[0175]

[0176] Among them, SS Between is the between-group sum of squares, i.e., the difference between different combinations of structural surface parameters;

[0177] SS Total is the total sum of squares;

[0178] S9, case application and effect analysis;

[0179] S901. Application: Apply the algorithm to actual rock mass engineering projects, such as tunnels and slopes, to perform structural surface identification and strength prediction;

[0180] S902, Example Results Analysis: Evaluate the accuracy and reliability of the algorithm by comparing the prediction results with field measured data, such as drilling coring and in-situ testing;

[0181]

[0182] Among them, Accuracy is the accuracy;

[0183] RMSE is the root mean square error;

[0184] N is the sample size;

[0185] y' i is the predicted value;

[0186] y i is the true value;

[0187] S10, bidirectional constrained random structure surface generation;

[0188] S1001, random simulation: Generate the initial structural surface network based on geostatistical principles;

[0189] S1002, Bidirectional Constraint: Constrain the generated structural surface network based on geological laws and engineering requirements, using geological structural lines as constraints to ensure that the direction of the structural surface is consistent with the geological structural lines;

[0190] ConstrainedNetwork=RandomNetwork∩GeologicalConstraints

[0191] Among them, ConstrainedNetwork is the constrained structural surface network;

[0192] RandomNetwork is a randomly generated structural surface network;

[0193] GeologicalConstraints are geological constraints;

[0194] S11, statistical characteristics of structural surface distribution and “random-predictive-deterministic” network generation;

[0195] Analyze the distribution statistical characteristics of the structural surface, such as density and directionality, and then generate the final structural surface network through random simulation, predictive optimization and deterministic adjustment;

[0196] S12, field test verification;

[0197] S1201. Field test: Drill holes and conduct in-situ tests at the rock mass engineering site to obtain actual structural surface data;

[0198] S1202. Result analysis: Compare the field measured data with the predicted results, analyze the sources of errors, and propose improvement measures;

[0199] S13, analysis of structural surface network inversion results;

[0200] Based on field test data, an inversion analysis of the structural surface network is performed to verify the accuracy of the prediction model. The model parameters and algorithm process are then adjusted according to the inversion results to achieve continuous improvement in prediction accuracy.

[0201] The proposed rock mass structural surface strength prediction system and method based on GPR three-instantaneous characteristics and residual networks achieves accurate prediction of rock mass structural surface strength and intelligent generation of structural surface networks by integrating multi-source data, applying deep learning technology, and optimizing algorithm processes, providing strong support for the design and construction of rock engineering projects.

[0202] And through field test verification, test verification was carried out on multiple actual rock engineering sites to test the practicality and generalization ability of the algorithm and system, and compare the field measured data with the prediction results, analyze the source of error, and propose improvement measures. Based on the field test data, the structural surface network was inverted and analyzed to verify the accuracy of the prediction model. Then, according to the inversion results, the model parameters and algorithm process were adjusted to achieve continuous improvement in prediction accuracy. The verification and improvement of the model through forward and inversion were realized to further improve the accuracy of rock structural surface strength prediction.

[0203] As another embodiment of the present invention, the S1 data acquisition and sample set construction further includes the following steps:

[0204] S101, GPR data acquisition: Use GPR equipment to collect GPR image data on the rock surface or in the borehole;

[0205] S102, three-instantaneous feature extraction: extracting three-instantaneous features of instantaneous amplitude (A), instantaneous frequency (F) and instantaneous phase (Φ) from the GPR image;

[0206] A(t,x,y)=Amplitude(I(t,x,y))

[0207] F(t,x,y)=Frequency(I(t,x,y))

[0208] Φ(t,x,y)=Phase(I(t,x,y))

[0209] Where I(t,x,y) is the GPR image data;

[0210] t is time;

[0211] x, y are spatial coordinates;

[0212] S103, sample set construction: Combine the GPR image and its three-moment features and spatial position information to construct a multi-domain feature vector sample set for training and testing.

[0213] As another embodiment of the present invention, the S4 numerical simulation setting and deep learning training further includes the following steps:

[0214] S401, numerical simulation: set up numerical simulation experiments under different geological conditions to generate a large number of GPR images and their corresponding three-instantaneous characteristic data;

[0215] S402, deep learning training: using simulated data and some measured data to train the residual network model to make a preliminary prediction of the structural surface strength;

[0216]

[0217] Where N is the number of samples;

[0218] y' i is the predicted value;

[0219] y i is the true value.

[0220] As another embodiment of the present invention, the S11 structural surface distribution statistical characteristics and the "random-predictive-deterministic" network generation further include the following steps:

[0221] S1101. Statistical feature analysis: Analyze the distribution characteristics of the structural surface, such as density and direction distribution;

[0222]

[0223] Among them, Density is the structural surface density;

[0224] Area of ​​Interest is the area of ​​the region of interest;

[0225] N Areas is the number of structural surfaces;

[0226] φ i is the tendency of the i-th structural surface;

[0227] S1102, "random-predictive-deterministic" network generation;

[0228] S1102a, random simulation: generating an initial network;

[0229] S1102b, prediction optimization: adjusting network parameters based on prediction data, such as GPR images and geological data;

[0230] S1102c, Deterministic Adjustment: Finalize the network based on geological laws and engineering requirements.

[0231] As another embodiment of the present invention, the S13 structural surface network inversion result analysis further includes the following steps:

[0232] S1301, Inversion Analysis: Based on the field test data, inversion analysis is performed on the structural surface network, that is, the actual situation of the underground structural surface is inferred based on the observed data;

[0233] S1302, optimization iteration: adjusting model parameters and algorithm flow according to the inversion results to improve prediction accuracy;

[0234]

[0235] Among them, UpdatedParameters is the updated parameter;

[0236] OriginalParameters is the original parameter;

[0237] α is the learning rate;

[0238] is the gradient of the loss function with respect to the parameters.

[0239] The embodiments disclosed in the present invention are preferred embodiments, but are not limited to them. Ordinary technicians in this field can easily understand the spirit of the present invention based on the above embodiments and make different extensions and changes. As long as they do not deviate from the spirit of the present invention, they are all within the scope of protection of the present invention.

Claims

1. A structural surface strength prediction method based on residual network, characterized in that: The following steps are involved: S1, data acquisition and sample set construction; GPR image data is collected on the rock surface or in a borehole using GPR equipment. Three instantaneous features, namely instantaneous amplitude A, instantaneous frequency F, and instantaneous phase Φ, are extracted from the GPR image. The GPR image, its three instantaneous features, and spatial position information are then combined to construct a multi-domain feature vector sample set for training and testing. S2, multi-domain feature vector extraction of radar images; The three instantaneous features (A, F, Φ) of the GPR image are fused with the spatial position information (x, y) to form a multi-domain feature vector set V, where t is time; V(t,x,y)=pA(t,x,y),F(t,x,y),Φ(t,x,y),x,y] S3, network structure design and residual network model construction; S301. Residual Network Design: Design a deep residual network model containing multiple residual units; Among them, x l is the input of the lth layer; W l 1 and W l 2 is the weight of the convolutional layer; Conv is the convolution operation; BN stands for batch normalization; ReLU is the activation function; S302, residual neural network features: extracting high-level features through residual learning mechanism; S4, numerical simulation setup and deep learning training; Numerical simulation experiments were set up under different geological conditions to generate a large number of GPR images and their corresponding three-instantaneous characteristic data. The simulated data and some measured data were then used to train the residual network model to make a preliminary prediction of the structural surface strength. S5. Analysis of deep learning results; S501, Model evaluation: Evaluate the prediction performance of the model through cross-validation and confusion matrix; S502. Feature Importance Analysis: Analyze the impact of different features on prediction results and optimize feature selection and model structure; S6, intelligent extraction algorithm for rock mass structural surface information; S601, spatial multi-scale convolutional point cloud classification: using spatial multi-scale convolutional neural network to classify point cloud data; S602, dynamic DBSCAN clustering: performing dynamic DBSCAN clustering on the candidate regions; Clusters=DBSCAN(V,∈,MinPts) Among them, V is a set of multi-domain feature vectors; ∈ is the neighborhood radius; MinPts is the minimum number of points; S7, structural surface segmentation and information extraction; S701, structural surface segmentation: In the structural surface segmentation stage, an image segmentation algorithm is used, such as threshold-based segmentation, edge detection, or image segmentation technology. Taking threshold-based segmentation as an example, a threshold function is defined to distinguish between structural surface and non-structural surface areas. Where V(x,y) is the multi-domain eigenvector at point (x,y); Threshold is the preset segmentation threshold; Mask(x,y) is the binary mask after segmentation; 1 indicates the structural surface area; 0 represents the non-structural surface area; S702, information extraction: After the structural surface is segmented, its geometric and physical properties are extracted, and the area A of the structural surface Area , inclination θ and dip φ; Among them, (x1, y1) and (x2, y2) are the coordinates of adjacent pixels on the structure surface; PixelArea is the actual area represented by each pixel after calibration based on image resolution and scanning distance; Weight is a weight based on pixel position or feature importance; S8, structural surface model test and effect analysis; S801. Structural surface model test: Construct a physical or digital model in the laboratory to simulate the structural surface of the actual rock mass. By controlling variables such as the number, spacing, and inclination of the structural surfaces, observe and record the model's behavior, such as strength and deformation. S802, Structural Surface Effect Analysis: Use statistical methods to evaluate the effects of different structural surface parameters on model behavior; Among them, SS Between is the between-group sum of squares, i.e., the difference between different combinations of structural surface parameters; SS Total is the total sum of squares; S9, case application and effect analysis; S901. Application: Apply the algorithm to actual rock mass engineering to perform structural plane identification and strength prediction; S902. Example effect analysis: Evaluate the accuracy and reliability of the algorithm by comparing the prediction results with the field measured data; Among them, Accuracy is the accuracy; RMSE is the root mean square error; N is the sample size; y' i is the predicted value; y i is the true value; S10, bidirectional constrained random structure surface generation; S1001, random simulation: Generate the initial structural surface network based on geostatistical principles; S1002, Bidirectional Constraint: Constrain the generated structural surface network based on geological laws and engineering requirements, using geological structural lines as constraints to ensure that the direction of the structural surface is consistent with the geological structural lines; ConstrainedNetwork=RandomNetwork∩GeologicalConstraints Among them, ConstrainedNetwork is the constrained structural surface network; RandomNetwork is a randomly generated structural surface network; GeologicalConstraints are geological constraints; S11, statistical characteristics of structural surface distribution and "random-predictive-deterministic" network generation; Analyze the distribution statistics, density, and directionality of the structural surface, and then generate the final structural surface network through random simulation, predictive optimization, and deterministic adjustment; S12, field test verification; S1201. Field test: Drill holes and conduct in-situ tests at the rock mass engineering site to obtain actual structural surface data; S1202. Result analysis: Compare the field measured data with the predicted results, analyze the sources of errors, and propose improvement measures; S13, analysis of structural surface network inversion results; Based on field test data, an inversion analysis of the structural surface network is performed to verify the accuracy of the prediction model. The model parameters and algorithm process are then adjusted according to the inversion results to achieve continuous improvement in prediction accuracy.

2. The prediction method according to claim 1, characterized in that The S1 data acquisition and sample set construction further include the following steps: S101, GPR data acquisition: Use GPR equipment to collect GPR image data on the rock surface or in the borehole; S102, three-instantaneous feature extraction: extracting three-instantaneous features of instantaneous amplitude A, instantaneous frequency F and instantaneous phase Φ from the GPR image; A(t,x,y)=Amplitude(I(t,x,y)) F(t,x,y)=Frequency(I(t,x,y)) Φ(t,x,y)=Phase(I(t,x,y)) Where I(t,x,y) is the GPR image data; t is time; x, y are spatial coordinates; S103, sample set construction: Combine the GPR image and its three-moment features and spatial position information to construct a multi-domain feature vector sample set for training and testing.

3. The prediction method according to claim 1, wherein: The S4 numerical simulation setup and deep learning training also include the following steps: S401, numerical simulation: set up numerical simulation experiments under different geological conditions to generate a large number of GPR images and their corresponding three-instantaneous characteristic data; S402, deep learning training: using simulated data and some measured data to train the residual network model to make a preliminary prediction of the structural surface strength; Where N is the number of samples; y' i is the predicted value; y i is the true value.

4. The prediction method according to claim 1, wherein: The generation of the statistical characteristics of the S11 structural surface distribution and the "random-predictive-deterministic" network also includes the following steps: S1101. Statistical feature analysis: Analyze the distribution characteristics of the structural surface, density, and direction distribution; Among them, Density is the structural surface density; Area of ​​Interest is the area of ​​the region of interest; N Areas is the number of structural surfaces; φ i is the tendency of the i-th structural surface; S1102, "random-predictive-deterministic" network generation; S1102a, random simulation: generating an initial network; S1102b, prediction optimization: adjusting network parameters based on prediction data, GPR images, and geological data; S1102c, Deterministic Adjustment: Finalize the network based on geological laws and engineering requirements.

5. The prediction method according to claim 1, wherein: The S13 structural surface network inversion result analysis also includes the following steps: S1301, Inversion Analysis: Based on the field test data, inversion analysis is performed on the structural surface network, that is, the actual situation of the underground structural surface is inferred based on the observed data; S1302, optimization iteration: adjusting model parameters and algorithm flow according to the inversion results to improve prediction accuracy; Among them, UpdatedParameters is the updated parameter; OriginalParameters is the original parameter; α is the learning rate; is the gradient of the loss function with respect to the parameters.

6. A prediction system using the structural surface strength prediction method based on residual network according to any one of claims 1 to 5, characterized in that: It includes a residual network feature extraction unit that uses a residual network to extract features from GPR images; a spatially dilated convolutional neural network and dynamic DBSCAN clustering unit that combines spatially dilated convolution to enhance the receptive field to capture broader contextual information; a structural surface strength prediction unit that extracts structural surface features to construct a strength prediction model; and a "random-prediction-deterministic" structural surface network algorithm generation unit that optimizes the accuracy of rock mass structural surface network descriptions through predicted data; The residual network feature extraction unit is connected to the spatial expansion convolutional neural network and the dynamic DBSCAN clustering unit, and the spatial expansion convolutional neural network and the dynamic DBSCAN clustering unit are connected to the structural surface strength prediction unit, and the "random-prediction-deterministic" structural surface network algorithm generation unit is connected to the structural surface strength prediction unit.

7. The structural surface strength prediction system based on residual network according to claim 6, characterized in that: The residual network feature extraction unit includes a data preprocessing module for normalizing the input GPR image, a residual network construction module for designing the residual network architecture, and a feature extraction module for extracting feature maps. The data preprocessing module is connected to the residual network construction module, and the residual network construction module is connected to the feature extraction module.

8. The structural surface strength prediction system based on residual network according to claim 6, characterized in that: The spatial dilation convolutional neural network and dynamic DBSCAN clustering unit include a spatial dilation convolution module for capturing contextual features and a dynamic DBSCAN clustering module for identifying different structural surface areas. The spatial dilation convolution module is connected to the dynamic DBSCAN clustering module.

9. The structural surface strength prediction system based on residual network according to claim 6, characterized in that: The structural surface strength prediction unit includes a feature integration module for forming a complete feature vector, a prediction model construction module for learning the mapping relationship between the feature vector and the structural surface strength, and a strength prediction module for calculating the structural surface strength prediction value. The feature integration module is connected to the prediction model construction module, and the prediction model construction module is connected to the strength prediction module.

10. The structural surface strength prediction system based on residual network according to claim 6, characterized in that: The "random-prediction-determination" structural surface network algorithm generation unit includes a random generation module for generating multiple groups of candidate structural surface network samples, a prediction and evaluation module for predicting and evaluating candidate samples, and an optimal solution determination module for selecting the optimal candidate samples. The random generation module is connected to the prediction and evaluation module, and the prediction and evaluation module is connected to the optimal solution determination module.

Images