Method for rapidly evaluating stress state of underground roadway surrounding rock anchor rod

Abstract

The invention discloses an underground roadway surrounding rock anchor rod stress state rapid assessment method, which comprises the following steps that a strain gauge is connected to a convergence node of an underground roadway surrounding rock anchor rod, surface stress information of the surrounding rock anchor rod is monitored, and a stress edge calculation gateway module is connected; the stress edge calculation gateway module collects the unit time stress of the anchor rod; the stress edge computing gateway module collects the information and uploads the information to the remote computer control platform after unified data denoising; the uploaded information is summarized to a stress edge calculation gateway module for deformation size and stress state evaluation; and the stress edge calculation gateway module performs deformation size and stress state evaluation, including deformation residence time analysis and evaluation, stress state grade evaluation and giving of maintenance measures. The surrounding rock anchor rod deformation type performance monitoring system realizes surrounding rock anchor rod deformation type performance monitoring, and realizes rapid evaluation through model prediction; timely support state feedback is provided for engineers, and safe operation of the roadway is guaranteed.

Description

Technical Field

[0001] The invention relates to the fields of rock mass engineering and tunnel engineering, and in particular to a method for quickly evaluating the stress state of surrounding rock anchors in underground tunnels. Background Art

[0002] During the construction and operation of underground tunnels, the stability of the surrounding rock is directly related to the safety and benefits of the project. As a common support method, anchor bolts improve the stability of the surrounding rock by anchoring it. However, the stress state of the anchor bolts will change during long-term stress, which may lead to support failure and affect the safety of the tunnel. Therefore, real-time, rapid and accurate evaluation of the stress state of the anchor bolts is of great significance to ensure the safety of the tunnel.

[0003] Traditional anchor stress monitoring methods usually rely on fixed-point measurement and manual recording, which is time-consuming and labor-intensive and cannot provide global stress distribution information. At the same time, due to the complex environment of underground tunnels, the installation and maintenance of measurement equipment also face many challenges. In addition, existing methods mostly rely on post-processing analysis, which makes it difficult to achieve real-time evaluation and early warning, and there is a lag. Therefore, a simple, fast and reliable anchor stress state assessment method is developed. Summary of the invention

[0004] In order to overcome the shortcomings and deficiencies of the prior art, the present invention provides a method for quickly evaluating the stress state of surrounding rock anchors in underground tunnels.

[0005] The technical solution adopted by the present invention is a method for quickly evaluating the stress state of surrounding rock bolts in underground tunnels, comprising the following steps:

[0006] S1. Connect strain gauges at the junction nodes of the surrounding rock anchors in the underground tunnel to monitor the surface stress information of the surrounding rock anchors in the underground tunnel, and connect to the stress edge computing gateway module through the wireless sensor network; use the anchor puller to carry out indoor anchor pull-out tests to obtain the deformation data of the anchors under different pull-out forces;

[0007] S2. The stress edge computing gateway module collects the unit time stress of the anchor rod. The stress monitoring platform cleans and normalizes the unit time stress of the anchor rod and performs clustering processing on the unit time stress of the anchor rod. The high-precision pressure sensor is used to measure the axial force value of the anchor rod in the process of load deformation in real time and record the force change of the anchor rod.

[0008] S3, the signal transmission module of the stress edge computing gateway module uses the communication device module to clean and normalize the unit time stress of the anchor rod on the stress monitoring platform, cluster the unit time stress of the anchor rod, and collect the surrounding stress information of the surrounding rock anchor rod, and then upload it to the remote computer control platform after unified data denoising; use a multi-angle high-definition camera to shoot the anchor rod tray from multiple angles and viewpoints, and reconstruct the three-dimensional point cloud model of the anchor rod tray from the two-dimensional image sequence through the SFM (Structure from Motion) algorithm;

[0009] S4. The uploaded stress monitoring platform cleans and normalizes the anchor stress per unit time, clusters the anchor stress per unit time, and aggregates the surrounding rock anchor stress information to the stress edge computing gateway module for deformation size and stress state evaluation; the initial anchor tray point cloud model is accurately aligned with the loaded point cloud model using the ICP (Iterative Closest Point) point cloud pairing algorithm;

[0010] S5. Use the aligned point cloud model to calculate the distance, use the Euclidean distance formula to calculate the deformation, and use the deformation field drawing software to generate the deformation field map of the anchor tray; the stress edge computing gateway module evaluates the deformation size and stress state, including deformation residence time analysis and evaluation;

[0011] S6. Construct a CNN convolutional neural network model, and use the acquired axial force values ​​and generated deformation field data for training and prediction. Input the newly acquired deformation field data into the CNN prediction model to obtain the corresponding axial force value, and evaluate the stress state of the anchor according to the predicted axial force value. The stress edge computing gateway module evaluates the stress state level and provides maintenance measures for the deformed surrounding rock anchor.

[0012] The stress edge computing gateway module performs deformation size and stress state evaluation including the following steps:

[0013] When the clustering process of the anchor stress per unit time changes, the deformation type characteristic analysis and evaluation program starts;

[0014] Adopting the object stress feature extraction algorithm, the collected stress information around the surrounding rock anchor is analyzed to evaluate the real-time collected stress-strain relationship, stress accumulation effect, and peak stress fluctuation of the surrounding rock anchor; evaluate the changes in the peak-to-peak value of the surrounding rock anchor transmission support load; evaluate the partial deformation damage of the surrounding rock anchor and the degree of influence of the damage on the surrounding rock anchor;

[0015] The surrounding stress of the surrounding rock anchor, the changes in the surface stress of the rock mass, and the clustering of the stress per unit time of the anchor are used to comprehensively calculate the support and stress dispersion efficiency of the surrounding rock anchor. Based on the deformation surface stress and support and stress dispersion efficiency of the surrounding rock anchor, the deformation surface stress of the surrounding rock anchor is analyzed and extracted from the surface stress of the rock mass. The overall distribution of the deformation of the surrounding rock anchor is calculated using the stress evaluation algorithm, and the use environment of the surrounding rock anchor is evaluated.

[0016] Based on the above assessment results, the status of surrounding rock anchors is graded and evaluated, and different maintenance and protection plans are given.

[0017] The specific steps of the object stress feature extraction algorithm are as follows:

[0018] Establish a database of surrounding rock bolt status under different weather conditions, including a database of normal waveforms of surrounding rock bolts under different levels of support loads, a database of historical waveforms of tested surrounding rock bolts, and a database of waveforms of various surrounding rock bolt defects;

[0019] Preprocess the collected surface stress waveform information of the surrounding rock bolts in underground tunnels;

[0020] Extract and select features from the collected surface stress waveform information of the surrounding rock bolts in underground tunnels;

[0021] The surrounding rock anchor characteristic support load curve is generated, and the characteristic matching and stress comparison analysis are performed. By quantifying the difference between the measured waveform and the normal waveform and the historical waveform, the type and state of deformation are evaluated. By matching with the defect waveform characteristics, the deformation defects and their causes are evaluated.

[0022] The specific steps of using the stress assessment algorithm to calculate the rock mass crack distribution of the surrounding rock anchor and assess the use environment of the surrounding rock anchor are as follows:

[0023] Monitor the maximum deformation of the surrounding rock bolts in underground tunnels and the crack depth parameters of the rock mass;

[0024] The surrounding rock anchor support and stress dispersion efficiency are comprehensively calculated by using the surrounding rock anchor stress information, rock surface stress changes, and clustering of anchor stress per unit time.

[0025] Based on the deformation surface stress of surrounding rock bolts and the support and stress dispersion efficiency, the deformation surface stress of surrounding rock bolts is extracted from the rock surface stress, and the overall distribution of surrounding rock bolt deformation is calculated using the stress evaluation algorithm.

[0026] The overall distribution of surrounding rock bolt deformation is used to evaluate the use environment of the bolt and predict the erosion rate of the surrounding rock bolt.

[0027] The surface stress information of the surrounding rock anchor of the underground tunnel monitored by the strain gauge is converted into clustered stress by a mechanical stress converter, and then connected to the stress edge computing gateway module through a wireless sensor network.

[0028] The rapid assessment method of the stress state of surrounding rock bolts in underground tunnels is implemented through different modules, including a stress monitoring device module, a communication device module, and a stress edge computing gateway module.

[0029] The strain gauge, stress edge computing gateway module and remote signaling module are connected to the confluence node of the surrounding rock anchor of the underground tunnel. The strain gauge is used to monitor the surface stress information of the surrounding rock anchor of the underground tunnel and transmit it to the stress edge computing gateway module through the wireless sensor network. The stress edge computing gateway module is used to clean and normalize the unit time stress of the anchor using the stress monitoring platform. The remote signaling module is used to collect the unit time stress of the anchor from the stress edge computing gateway module and perform clustering processing.

[0030] The communication device module includes an information collection module and an optical fiber or wireless signal transmission module. The information collection module is used to collect the surface stress information of the surrounding rock anchors of the underground tunnel, clean and normalize the unit time stress of the anchors, and cluster the unit time stress of the anchors. The optical fiber or wireless signal transmission module is used to upload the monitoring information collected by the information collection module to the remote computer control platform to the stress edge computing gateway module after unified data denoising;

[0031] The stress edge computing gateway module is used to analyze and process the aggregated monitoring information, and provide deformation residence time evaluation results and different maintenance and protection solutions.

[0032] The strain gauge is connected to a mechanical stress converter for converting the collected analog signal of the surface stress information of the surrounding rock anchor of the underground tunnel into a digital signal for transmission through a wireless sensor network.

[0033] Technical effect of the present invention: The present invention makes full use of the basic functions of the stress edge computing gateway module, installs a simple strain gauge in the surrounding rock anchor, and combines the communication channel of the automated stress monitoring platform, the strain gauge, and clusters the stress per unit time of the anchor to achieve remote online monitoring and evaluation of the deformation type performance. The monitoring efficiency is high and the cost is low, which ensures the safety of the surrounding rock anchor and improves the targeted equipment maintenance and control. The present invention proposes a comprehensive evaluation method combining pull-out test, three-dimensional reconstruction and machine learning, which can efficiently obtain the stress state information of the anchor during the loading process, and realize rapid evaluation through model prediction, providing timely support status feedback to engineering personnel and ensuring the safe operation of the tunnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a schematic diagram of the process of the method for rapid assessment of stress state of surrounding rock anchors in underground tunnels of the present invention;

[0035] Figure 2 It is a schematic diagram of the structural composition of the modules for realizing the method for rapid assessment of stress state of surrounding rock anchors in underground tunnels of the present invention. DETAILED DESCRIPTION

[0036] The present invention is further described below in conjunction with the accompanying drawings:

[0037] like Figure 1 As shown in the figure, the method for rapid assessment of stress state of surrounding rock bolts in underground tunnels includes the following steps:

[0038] S1. Connect strain gauges at the junction nodes of the surrounding rock anchors in the underground tunnel to monitor the surface stress information of the surrounding rock anchors in the underground tunnel, and connect to the stress edge computing gateway module through the wireless sensor network; use the anchor puller to carry out indoor anchor pull-out tests to obtain the deformation data of the anchors under different pull-out forces;

[0039] S2. The stress edge computing gateway module collects the unit time stress of the anchor rod. The stress monitoring platform cleans and normalizes the unit time stress of the anchor rod and performs clustering processing on the unit time stress of the anchor rod. The high-precision pressure sensor is used to measure the axial force value of the anchor rod in the process of load deformation in real time and record the force change of the anchor rod.

[0040] S3, signal transmission of the stress edge computing gateway module uses a communication device module to clean and normalize the unit time stress of the anchor rod on the stress monitoring platform, cluster the unit time stress of the anchor rod, collect the surrounding stress information of the surrounding rock anchor rod, and then upload it to the remote computer control platform after unified data denoising; use a multi-angle high-definition camera to shoot the anchor rod tray from multiple angles and viewpoints, and reconstruct the three-dimensional point cloud model of the anchor rod tray from the two-dimensional image sequence through the SFM (Structure from Motion) algorithm;

[0041] Data cleaning: The stress monitoring platform will obtain the anchor stress data per unit time, which may contain noise, outliers or missing values. The purpose of data cleaning is to remove or correct these invalid or inaccurate data through algorithms, thereby improving the accuracy of subsequent analysis. Cleaning methods can include filtering, interpolation, smoothing and other technologies to ensure the quality of input data.

[0042] Normalization: Normalization refers to converting stress data of different magnitudes into a unified scale range (for example, between 0 and 1). This step is to eliminate the scale differences due to different anchors, sensors or data sources, and ensure that different data can be effectively compared and clustered. Common normalization methods include minimum-maximum normalization, Z-score normalization, etc.

[0043] Clustering: Clustering is to divide the stress data of anchor rods into different groups (clusters), and the anchor rod stress in each group has similar characteristics. This process often uses unsupervised learning algorithms, such as K-Means clustering or DBSCAN (density-based spatial clustering), etc. Through cluster analysis of similar stress data, anchor rods with the same stress change pattern can be identified. The purpose of clustering is to simplify subsequent analysis and integrate similar anchor rod stress characteristics, thereby reducing the complexity of data processing.

[0044] Gathering stress information around surrounding rock anchors: The stress monitoring platform not only monitors the stress changes of a single anchor, but also aggregates the stress information of multiple anchors in the surrounding rock area. Through such aggregation, the overall stress distribution around the surrounding rock can be obtained, which is of great significance for judging landslides, rock deformation, etc.

[0045] Data denoising: The collected stress data may contain some environmental noise or measurement errors. Data denoising algorithms, such as wavelet denoising or Kalman filtering, can effectively reduce the impact of noise on data analysis, thereby obtaining more accurate and stable stress information. The denoising process is an important step in signal processing to ensure the accuracy and reliability of data.

[0046] After data cleaning, normalization, clustering and denoising, all processed stress data are uploaded to the remote computer control platform through the communication device module. The platform is used to centrally manage, analyze and store stress monitoring data for further model analysis and stress monitoring results display. Usually, data transmission uses a stable wireless or wired network protocol to ensure that the data of large-scale, multi-point sensor networks can be transmitted to the remote server in a timely and reliable manner.

[0047] The multi-angle HD camera is used to shoot the anchor tray from multiple different angles and viewpoints. In this way, a series of 2D image sequences are obtained, which capture the details and shape changes of the anchor tray. Each camera shoots at a different angle, and the information obtained from different directions can provide more geometric details, which is conducive to the subsequent 3D reconstruction.

[0048] The SFM (Structure from Motion) algorithm is an algorithm that reconstructs a 3D structure from a 2D image sequence. The basic principle of the algorithm is to infer the 3D geometric structure of an object by analyzing the changes in different viewpoints of the same object in multiple images. In the monitoring of anchor pallets, the SFM algorithm matches and associates the feature points in the 2D images captured by the camera, and then derives the position and shape of the anchor pallet in 3D space through these matching points. After processing, the result generated by SFM is a 3D point cloud model of the anchor pallet, which is composed of many 3D points and accurately reflects the geometric structure of the anchor pallet.

[0049] The 3D point cloud model can be used for further shape analysis, stress state assessment or visualization. The point cloud data can be used to more intuitively observe the physical changes of the anchor tray, such as displacement and deformation.

[0050] S4. Use the stress monitoring platform to clean and normalize the anchor stress per unit time, cluster the anchor stress per unit time, and summarize the surrounding rock anchor stress information to the stress edge computing gateway module to evaluate the deformation size and stress state; use the ICP (Iterative Closest Point) point cloud pairing algorithm to accurately align the initial anchor tray point cloud model with the loaded point cloud model;

[0051] S5. Use the aligned point cloud model to calculate the distance, use the Euclidean distance formula to calculate the deformation, and use the deformation field drawing software to generate the deformation field map of the anchor tray; the stress edge computing gateway module evaluates the deformation size and stress state, including deformation residence time analysis and evaluation; construct an RNN recurrent neural network model, and use the acquired axial force value and generated deformation field data for training and prediction; input the newly acquired deformation field data into the CNN prediction model to obtain the corresponding axial force value, and evaluate the stress state of the anchor according to the predicted axial force value;

[0052] S6. Construct a CNN convolutional neural network model, and use the acquired axial force values ​​and generated deformation field data for training and prediction. Input the newly acquired deformation field data into the CNN prediction model to obtain the corresponding axial force value, and evaluate the stress state of the anchor according to the predicted axial force value. The stress edge computing gateway module evaluates the stress state level and provides maintenance measures for the deformed surrounding rock anchor.

[0053] ANSYS is a widely used engineering simulation software suitable for simulation of multi-physics fields such as structure, fluid, electromagnetic and heat transfer. It is powerful in finite element analysis (FEA) and can accurately calculate and visualize the deformation field of the structure under different external forces.

[0054] The anchor rod puller is an electronic puller capable of providing accurate control and measurement of the pulling force.

[0055] The pressure sensor is a high-precision resistance strain type pressure sensor, which can achieve accurate measurement of tiny force changes.

[0056] The multi-angle high-definition camera is an industrial camera with high-resolution and high-speed shooting functions, which can capture the slight deformation of the anchor tray.

[0057] The ICP point cloud pairing algorithm includes multiple alignment strategies such as point-to-point and point-to-plane to improve the accuracy and stability of point cloud model alignment.

[0058] The distance calculation not only uses the Euclidean distance formula, but also includes other suitable geometric calculation methods to accurately draw the deformation field map.

[0059] The CNN convolutional neural network model improves the generalization ability and prediction accuracy of the model by optimizing hyperparameters and adjusting the network structure.

[0060] The performance of Convolutional Neural Network (CNN) models can be improved by optimizing hyperparameters and adjusting the network structure to improve generalization ability and prediction accuracy. Here are the main aspects of how CNN models can be improved through these methods:

[0061] 1. Optimizing hyperparameters includes:

[0062] Hyperparameters refer to the parameters that need to be set in advance during model training. Optimizing these hyperparameters can significantly improve the performance of the model.

[0063] Learning Rate: The learning rate determines the step size of the model parameter update. If the learning rate is too high, the training may be unstable or even fail to converge; if the learning rate is too low, the training speed may be too slow or fall into a local optimum. Using a learning rate scheduler (such as learning rate decay or adaptive learning rate algorithms such as Adam and RMSprop) can dynamically adjust the learning rate to help the model converge faster and achieve better generalization effects.

[0064] Batch Size: Batch size affects the training efficiency and performance of the model. Small batches allow the model to be updated more frequently, but may lead to unstable gradients; large batches are more stable but require more computing resources. Reasonable adjustment of batch size, combined with hardware resources, can accelerate training and improve the generalization ability of the model.

[0065] Regularization: Regularization is a key method to avoid model overfitting. Common regularization methods include L2 regularization (weight decay) and L1 regularization. In addition, Dropout technology is also commonly used in CNN. By randomly discarding the output of some neurons, the network does not rely on specific neurons, thereby improving the generalization ability of the model.

[0066] Optimizer: Different optimizers perform differently in different data sets and tasks. Commonly used optimizers include SGD, Adam, RMSprop, etc. SGD can accelerate convergence, but in complex problems, Adam is usually able to better adapt the gradient, thereby improving the performance of the model.

[0067] Data Augmentation: During the training process, more training samples can be generated by performing data augmentation (such as rotation, translation, scaling, cropping, color change, etc.) on the training samples to enhance the generalization ability of the model. Data augmentation is particularly effective for CNN models in image data and can alleviate the overfitting problem.

[0068] 2. Adjust the network structure including:

[0069] The design of the CNN network structure is crucial to the performance of the model. Adjusting the appropriate structure can improve the learning ability and generalization effect of the model.

[0070] The number and size of convolutional layers: The number of convolutional layers and the size of each convolution kernel directly affect the model's feature extraction capabilities. Increasing the depth of the convolutional layer (such as using a deeper ResNet, VGG, etc. architecture) can extract more complex features, but care must be taken to prevent overfitting. The size of the convolution kernel also needs to be reasonably selected based on the feature distribution of the data. Usually, a 3x3 convolution kernel is more commonly used. A larger convolution kernel can capture a wider range of features, but the amount of calculation is larger.

[0071] Pooling: Common pooling methods include Max Pooling and Average Pooling. The pooling layer can reduce the size of the feature map, reduce computational complexity, and reduce overfitting. A suitable pooling strategy can retain important features while reducing unnecessary noise.

[0072] Residual Networks (ResNet): By introducing skip connections, residual networks can effectively solve the gradient vanishing problem in deep neural networks, allowing the network to be trained deeper and more stably. The introduction of the ResNet structure can improve the accuracy of the model, especially in deep learning tasks.

[0073] Batch Normalization: Adding BatchNormalization after the output of each layer of the neural network can make the network more stable during training, reduce the risk of gradient vanishing, and enable the network to be trained with a higher learning rate, thereby speeding up training and improving generalization ability.

[0074] The trade-off between network depth and width: Increasing the depth (more convolutional layers) or width (more convolution kernels per layer) of the network can improve the expressiveness of the model, but excessively increasing the depth or width may lead to overfitting or excessive computational overhead. Therefore, it is necessary to select the appropriate depth and width according to the specific task to find a balance between expressiveness and generalization ability.

[0075] Design of fully connected layers: At the end of CNN, there are usually 1-2 fully connected layers. In order to avoid overfitting, it may be necessary to reduce the size of the fully connected layers, and even reduce the use of fully connected layers in some tasks, and use global average pooling (GAP) instead of traditional fully connected layers.

[0076] 3. Improve the generalization ability of the model including:

[0077] Improving the generalization ability of the CNN model so that it can perform well outside the training data is the core goal of model optimization. The generalization ability can be further improved by the following methods:

[0078] Cross-Validation: By using cross-validation techniques, especially K-fold cross-validation, the performance of the model on different data partitions can be more comprehensively evaluated to prevent the model from overfitting to a specific data set.

[0079] Early Stopping: Monitor the performance of the validation set during training. When the loss of the validation set no longer decreases, stop training early to avoid overfitting the training set.

[0080] Ensemble Learning: Combining the prediction results of multiple models can further improve generalization capabilities. For example, by training multiple CNN models with different architectures and then weighted averaging or voting their outputs, the final prediction accuracy can be improved.

[0081] The CNN prediction model can process different types of deformation field inputs, output corresponding axial force values, and provide efficient and accurate anchor stress state assessment. The stress edge computing gateway module performs deformation size and stress state assessment including the following steps:

[0082] S10. When the clustering process of the anchor stress per unit time changes, the deformation type characteristic analysis and evaluation program starts;

[0083] S20. Use the object stress feature extraction algorithm to analyze the collected stress information around the surrounding rock anchor, evaluate the real-time collected surrounding rock anchor stress-strain relationship, stress accumulation effect, and peak stress fluctuation; evaluate the changes in the peak-to-peak value of the surrounding rock anchor transmission support load; evaluate the partial deformation damage of the surrounding rock anchor and the degree of influence of the damage on the surrounding rock anchor;

[0084] S30. Comprehensively calculate the support and stress dispersion efficiency of the surrounding rock anchor by using the surrounding stress of the surrounding rock anchor and the changes in the surface stress of the rock mass, and cluster the stress per unit time of the anchor. Based on the deformation surface stress and support and stress dispersion efficiency of the surrounding rock anchor, analyze and extract the deformation surface stress of the surrounding rock anchor from the surface stress of the rock mass, calculate the overall distribution of the deformation of the surrounding rock anchor by using the stress evaluation algorithm, and evaluate the use environment of the surrounding rock anchor;

[0085] S40. Based on the above assessment results, the surrounding rock anchor status is graded and evaluated, and different maintenance and protection plans are given.

[0086] The specific steps of the object stress feature extraction algorithm are as follows:

[0087] S100. Establish a database of surrounding rock bolt status under different weather conditions, including a normal waveform database of surrounding rock bolts under different levels of support loads, a historical waveform database of tested surrounding rock bolts, and a waveform database of various surrounding rock bolt defects;

[0088] S200. Preprocessing the collected surface stress waveform information of the surrounding rock bolt of the underground tunnel;

[0089] S300. Extract and select features of the collected surface stress waveform information of the surrounding rock bolts of the underground tunnel;

[0090] S400. Generate characteristic support load curve of surrounding rock anchor, and conduct characteristic matching and stress comparison analysis. By quantifying the difference between the measured waveform and the normal waveform and historical waveform, the type and state of deformation are evaluated. By matching with the defect waveform characteristics, the deformation defects and their causes are evaluated.

[0091] The specific steps of using the stress assessment algorithm to calculate the rock mass crack distribution of the surrounding rock anchor and assess the use environment of the surrounding rock anchor are as follows:

[0092] S01. Monitor the maximum deformation of the surrounding rock bolts in underground tunnels and the crack depth parameters of the rock mass;

[0093] S02. Comprehensively calculate the surrounding rock anchor support and stress dispersion efficiency by using the surrounding rock anchor stress information, rock surface stress changes, and clustering the anchor stress per unit time;

[0094] S03. Based on the deformation surface stress of the surrounding rock bolt and the support and stress dispersion efficiency, the deformation surface stress of the surrounding rock bolt is analyzed and extracted from the surface stress of the rock mass, and the overall distribution of the deformation of the surrounding rock bolt is calculated using the stress evaluation algorithm;

[0095] S04. Evaluate the use environment of the anchor bolt based on the overall distribution of the surrounding rock anchor bolt deformation and predict the erosion rate of the surrounding rock anchor bolt.

[0096] The surface stress information of the surrounding rock anchor of the underground tunnel monitored by the strain gauge is converted into clustered stress by a mechanical stress converter, and then connected to the stress edge computing gateway module through a wireless sensor network.

[0097] like Figure 2 As shown, the rapid assessment method of the stress state of surrounding rock bolts in underground tunnels is implemented through different modules, including a stress monitoring device module, a communication device module, and a stress edge computing gateway module.

[0098] The strain gauge, stress edge computing gateway module and remote signaling module in the surrounding rock anchor of the underground tunnel are connected. The strain gauge is used to monitor the surface stress information of the surrounding rock anchor of the underground tunnel and transmit it to the information collection module of the stress edge computing gateway module through the wireless sensor network. The stress edge computing gateway module is used to clean and normalize the unit time stress of the anchor using the stress monitoring platform. The remote signaling module is used to collect the unit time stress of the anchor from the stress edge computing gateway module and perform clustering processing.

[0099] The communication device module includes an information collection module and an optical fiber or wireless signal transmission module. The information collection module is used to collect the surface stress information of the surrounding rock anchors of the underground tunnel, clean and normalize the unit time stress of the anchors, and cluster the unit time stress of the anchors. The optical fiber or wireless signal transmission module is used to upload the monitoring information collected by the information collection module to the remote computer control platform to the stress edge computing gateway module after unified data denoising;

[0100] The stress edge computing gateway module is used to analyze and process the aggregated monitoring information, and provide deformation residence time evaluation results and different maintenance and protection solutions.

[0101] The strain gauge is connected to a mechanical stress converter for converting the collected analog signal of the surface stress information of the surrounding rock anchor of the underground tunnel into a digital signal for transmission through a wireless sensor network.

[0102] The present invention makes full use of the basic functions of the stress edge computing gateway module. By installing a simple strain gauge in the surrounding rock anchor, and combining the communication channel of the automated stress monitoring platform and the strain gauge, the stress per unit time of the anchor is clustered to achieve remote online monitoring and evaluation of the type and performance of deformation. The monitoring efficiency is high and the cost is low, which ensures the safety of the surrounding rock anchor, improves the pertinence of equipment maintenance and control, and effectively expands the application of stress algorithms and processing in the operation and maintenance of surrounding rock anchors.

[0103] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", "connect", and "fix" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood by specific circumstances.

[0104] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A rapid assessment method for the stress state of surrounding rock bolts in underground tunnels, characterized in that: The method comprises the following steps: S1. Connect strain gauges at the junction nodes of the surrounding rock anchors in the underground tunnel to monitor the surface stress information of the surrounding rock anchors in the underground tunnel, and connect to the stress edge computing gateway module through the wireless sensor network; use the anchor puller to carry out indoor anchor pull-out tests to obtain the deformation data of the anchors under different pull-out forces; S2. The stress edge computing gateway module collects the unit time stress of the anchor rod. The stress monitoring platform cleans and normalizes the unit time stress of the anchor rod and performs clustering processing on the unit time stress of the anchor rod. The high-precision pressure sensor is used to measure the axial force value of the anchor rod in the process of load deformation in real time and record the force change of the anchor rod. S3, the signal transmission of the stress edge computing gateway module uses the communication device module to clean and normalize the unit time stress of the anchor rod on the stress monitoring platform, cluster the unit time stress of the anchor rod, collect the surrounding stress information of the surrounding rock anchor rod, and then upload it to the remote computer control platform after unified data denoising; use a multi-angle high-definition camera to shoot the anchor rod tray from multiple angles and viewpoints, and reconstruct the three-dimensional point cloud model of the anchor rod tray from the two-dimensional image sequence through the SFM algorithm; S4. The uploaded stress monitoring platform cleans and normalizes the anchor stress per unit time, clusters the anchor stress per unit time, and aggregates the surrounding rock anchor stress information to the stress edge computing gateway module for deformation size and stress state evaluation; the initial anchor tray point cloud model is accurately aligned with the loaded point cloud model using the ICP point cloud pairing algorithm; S5. Use the aligned point cloud model to calculate the distance, use the Euclidean distance formula to calculate the deformation, and use the deformation field drawing software to generate the deformation field map of the anchor tray; the stress edge computing gateway module evaluates the deformation size and stress state, including deformation residence time analysis and evaluation; S6. Construct a CNN convolutional neural network model, and use the acquired axial force values ​​and generated deformation field data for training and prediction. Input the newly acquired deformation field data into the CNN prediction model to obtain the corresponding axial force value, and evaluate the stress state of the anchor according to the predicted axial force value. The stress edge computing gateway module evaluates the stress state level and provides maintenance measures for the deformed surrounding rock anchor.

2. The method for rapid assessment of stress state of surrounding rock bolts in underground tunnels according to claim 1 is characterized in that: The stress edge computing gateway module performs deformation size and stress state evaluation including the following steps: When the clustering process of the anchor stress per unit time changes, the deformation type characteristic analysis and evaluation program starts; Adopting the object stress feature extraction algorithm, the collected stress information around the surrounding rock anchor is analyzed to evaluate the real-time collected stress-strain relationship, stress accumulation effect, and peak stress fluctuation of the surrounding rock anchor; evaluate the changes in the peak-to-peak value of the surrounding rock anchor transmission support load; evaluate the partial deformation damage of the surrounding rock anchor and the degree of influence of the damage on the surrounding rock anchor; The surrounding stress of the surrounding rock anchor, the changes in the surface stress of the rock mass, and the clustering of the stress per unit time of the anchor are used to comprehensively calculate the support and stress dispersion efficiency of the surrounding rock anchor. Based on the deformation surface stress and support and stress dispersion efficiency of the surrounding rock anchor, the deformation surface stress of the surrounding rock anchor is analyzed and extracted from the surface stress of the rock mass. The overall distribution of the deformation of the surrounding rock anchor is calculated using the stress evaluation algorithm, and the use environment of the surrounding rock anchor is evaluated. Based on the above assessment results, the status of surrounding rock anchors is graded and evaluated, and different maintenance and protection plans are given.

3. The method for rapid assessment of stress state of surrounding rock bolts in underground tunnels according to claim 2 is characterized in that: The specific steps of the object stress feature extraction algorithm are as follows: Establish a database of surrounding rock bolt status under different weather conditions, including a database of normal waveforms of surrounding rock bolts under different levels of support loads, a database of historical waveforms of tested surrounding rock bolts, and a database of waveforms of various surrounding rock bolt defects; Preprocess the collected surface stress waveform information of the surrounding rock bolts in underground tunnels; Extract and select features from the collected surface stress waveform information of the surrounding rock bolts in underground tunnels; The surrounding rock anchor characteristic support load curve is generated, and the characteristic matching and stress comparison analysis are performed. By quantifying the difference between the measured waveform and the normal waveform and the historical waveform, the type and state of deformation are evaluated. By matching with the defect waveform characteristics, the deformation defects and their causes are evaluated.

4. The method for rapid assessment of stress state of surrounding rock bolts in underground tunnels according to claim 2, characterized in that: The specific steps of using the stress assessment algorithm to calculate the rock mass crack distribution of the surrounding rock anchor and assess the use environment of the surrounding rock anchor are as follows: Monitor the maximum deformation of the surrounding rock bolts in underground tunnels and the crack depth parameters of the rock mass; The surrounding rock anchor support and stress dispersion efficiency are comprehensively calculated by using the surrounding rock anchor stress information, rock surface stress changes, and clustering of anchor stress per unit time. Based on the deformation surface stress of surrounding rock bolts and the support and stress dispersion efficiency, the deformation surface stress of surrounding rock bolts is extracted from the rock surface stress, and the overall distribution of surrounding rock bolt deformation is calculated using the stress evaluation algorithm. The overall distribution of surrounding rock bolt deformation is used to evaluate the use environment of the bolt and predict the erosion rate of the surrounding rock bolt.

5. The method for rapid assessment of stress state of surrounding rock bolts in underground tunnels according to claim 1, characterized in that: The surface stress information of the surrounding rock anchor of the underground tunnel monitored by the strain gauge is converted into clustered stress by a mechanical stress converter, and then connected to the stress edge computing gateway module through a wireless sensor network.

6. The method for rapid assessment of stress state of surrounding rock bolts in underground tunnels according to any one of claims 1 to 5, characterized in that: The method is implemented through different modules, including a stress monitoring device module, a communication device module and a stress edge computing gateway module. The strain gauge, stress edge computing gateway module and remote signaling module are connected to the confluence node of the surrounding rock anchor of the underground tunnel. The strain gauge is used to monitor the surface stress information of the surrounding rock anchor of the underground tunnel and transmit it to the stress edge computing gateway module through the wireless sensor network. The stress edge computing gateway module is used to clean and normalize the unit time stress of the anchor by using the stress monitoring platform. The remote signaling module is used to collect the unit time stress of the anchor from the stress edge computing gateway module and perform clustering processing.

7. The method for rapid assessment of stress state of surrounding rock bolts in underground tunnels according to claim 6, characterized in that: The communication device module includes an information collection module and an optical fiber or wireless signal transmission module. The information collection module is used to summarize the surface stress information of the surrounding rock anchors of the underground tunnel, clean and normalize the unit time stress of the anchors, and cluster the unit time stress of the anchors. The optical fiber or wireless signal transmission module is used to upload the monitoring information summarized by the information collection module to the remote computer control platform to the stress edge computing gateway module after unified data denoising.

8. The method for rapid assessment of stress state of surrounding rock bolts in underground tunnels according to claim 6, characterized in that: The stress edge computing gateway module is used to analyze and process the aggregated monitoring information, and provide deformation residence time evaluation results and different maintenance and protection solutions.

9. The method for rapid assessment of stress state of surrounding rock bolts in underground tunnels according to claim 6, characterized in that: The strain gauge is connected to a mechanical stress converter for converting the collected analog signal of the surface stress information of the surrounding rock anchor of the underground tunnel into a digital signal for transmission through a wireless sensor network.

Images