Tunnel expansion rock stress detection system based on remote sensing data

Abstract

The invention discloses a tunnel expansion rock stress detection system based on remote sensing data, which relates to the field of stress detection technology, and includes a detection center, wherein the detection center is communicatively connected with a remote sensing data acquisition module, a remote sensing data processing module, a geological information analysis module, a stress state detection module, and an early warning intervention module. The remote sensing data acquisition module collects remote sensing data of the surface around the tunnel and the expansion rock mass, and generates a remote sensing data set. The remote sensing data processing module processes the remote sensing data set to generate a surface feature layer and a rock mass feature layer. The geological information analysis module performs geological information analysis on the surface feature layer and the rock mass feature layer, and generates geological identification information. The stress state detection module constructs a rock mass mechanical model according to the geological identification information to perform stress state analysis on the expansion rock mass. The early warning intervention module generates corresponding early warning information according to the result of the stress state analysis, and then executes corresponding intervention measures.

Description

Technical Field

[0001] The invention relates to the technical field of stress detection, in particular to a tunnel expansion rock stress detection system based on remote sensing data. Background Art

[0002] Tunnel swelling rock stress detection refers to the process of monitoring and evaluating the stress state of swelling rock around the tunnel. In tunnel engineering, the underground rock mass will be affected by the forces from the surface and surrounding rock formations. Swelling rock refers to rock that will expand in volume or produce stress response when subjected to external forces. During tunnel construction and operation, the stress state of swelling rock is one of the important factors affecting the safety and stability of tunnel structure.

[0003] In tunnel engineering, traditional swelling rock stress detection relies too much on geological exploration and on-site monitoring, which leads to high data acquisition cost, poor real-time performance, and data accuracy that often does not meet requirements. With the rapid development of various remote sensing technologies, a more convenient means is provided for the acquisition of swelling rock stress detection data. How to obtain more accurate data for data analysis, and then obtain the stress condition of swelling rock in real time, when the stress condition of swelling rock is in an abnormal state, timely collect corresponding measures to intervene, so as to avoid the problems in tunnel engineering as much as possible, this is the problem that needs to be solved urgently in tunnel swelling rock stress detection. Summary of the invention

[0004] In order to solve the above problems, the object of the present invention is to provide a tunnel expansion rock stress detection system based on remote sensing data.

[0005] The object of the present invention can be achieved by the following technical solutions: a tunnel expansion rock stress detection system based on remote sensing data, comprising a detection center, wherein the detection center is communicatively connected with a remote sensing data acquisition module, a remote sensing data processing module, a geological information analysis module, a stress state detection module and an early warning intervention module;

[0006] The remote sensing data acquisition module is used to collect remote sensing data of the surface around the tunnel and the swelling rock mass, and generate a corresponding remote sensing data set;

[0007] The remote sensing data processing module is used to obtain remote sensing data sets to perform abnormal data calibration, image correction, feature data extraction, feature layer construction and feature layer fusion, and then generate corresponding surface feature layers and rock mass feature layers;

[0008] The geological information analysis module is used to analyze geological information based on the surface characteristic layer and the rock mass characteristic layer, and generate corresponding geological identification information;

[0009] The stress state detection module is used to construct a rock mass mechanics model according to geological identification information, and to perform stress state analysis on the swelling rock mass through the rock mass mechanics model;

[0010] The early warning intervention module is used to generate corresponding early warning information according to the result of stress state analysis, and then execute corresponding intervention measures.

[0011] Furthermore, the process of collecting remote sensing data of the surface around the tunnel and the swelling rock mass and generating the corresponding remote sensing data set includes:

[0012] Select a collection area, arrange and start remote sensing collection equipment in the collection area. Remote sensing collection equipment includes satellite remote sensing imaging equipment, drone remote sensing imaging equipment and aerial remote sensing imaging equipment. Determine whether the equipment collection range of each remote sensing collection equipment can cover the collection area;

[0013] If yes, remote sensing data of the surface around the tunnel and the swelling rock mass are collected by different remote sensing acquisition equipment, wherein the remote sensing data collected by the satellite remote sensing imaging equipment, the UAV remote sensing imaging equipment and the aerial remote sensing imaging equipment are respectively marked as Data1, Data2 and Data3, and Data1, Data2 and Data3 are merged to generate a remote sensing data set of the acquisition area;

[0014] If not, adjust the device collection range of each remote sensing collection device until the device collection range is able to cover the collection area, and collect the corresponding remote sensing data.

[0015] Furthermore, the process of obtaining remote sensing data sets for abnormal data calibration includes:

[0016] Acquire a remote sensing data set, and perform abnormal data calibration on Data1, Data2 and Data3 in the remote sensing data set in turn. The data type of the remote sensing data corresponding to Data1, Data2 and Data3 is remote sensing image, and the remote sensing image includes the remote sensing image of the surface around the tunnel and the remote sensing image of the swelling rock mass. Perform pixel segmentation on the remote sensing images corresponding to Data1, Data2 and Data3, and then generate a number of pixel areas with fixed area sizes.

[0017] Abnormal data calibration is used to calibrate the pixel area in abnormal state, obtain a number of pixel units corresponding to each pixel area, and obtain the pixel exposure, pixel clarity and pixel unit distortion information corresponding to the pixel unit;

[0018] If the pixel exposure of the pixel unit in the pixel area is less than or equal to the preset exposure threshold, the pixel clarity of the pixel unit is greater than or equal to the preset clarity threshold, and it is determined through the pixel unit distortion information that the current pixel unit does not have distortion, then the pixel unit is marked as a qualified pixel unit, otherwise, the pixel unit is marked as an abnormal pixel unit;

[0019] When the number of abnormal pixel units in the pixel area is greater than or equal to 1 / 3 of the total number of pixel units, the corresponding pixel area is marked as an image area to be corrected in an abnormal state, otherwise, the corresponding pixel area is marked as a normal image area in a normal state;

[0020] The image area to be errata includes the image area to be errata corresponding to the surface around the tunnel and the image area to be errata corresponding to the swelling rock mass, and the normal image area also includes the normal image areas corresponding to the surface around the tunnel and the swelling rock mass.

[0021] Furthermore, the process of performing the image correction includes:

[0022] Obtain all the image areas to be errata, locate the abnormal pixel units in the image areas to be errata, and analyze the distortion types of the abnormal pixel units. The distortion types include geometric transformation distortion and lens distortion. For geometric transformation distortion, build a convolutional neural network model through deep learning, and obtain the remote sensing image corresponding to the image area to be errata corresponding to the geometric transformation distortion as sample data. Input the sample data into the convolutional neural network model for training, and use the convolutional neural network model to correct the geometric transformation distortion. For lens distortion, reconfigure the lens parameters of the corresponding image lens of the remote sensing acquisition equipment, and re-acquire the remote sensing image until there is no lens distortion in the remote sensing image.

[0023] When the distortion processing of the abnormal pixel unit is completed, the pixel exposure of the abnormal pixel unit is adjusted to the exposure threshold and below the exposure threshold through the PS software, and the pixel clarity of the pixel unit is adjusted to the clarity threshold and above through the image sharpening or image enhancement algorithm, thereby converting the entire image area to be calibrated into a normal image area.

[0024] Furthermore, the process of extracting the feature data includes:

[0025] The remote sensing images after converting all the image areas to be corrected corresponding to the remote sensing data of Data1, Data2 and Data3 in the remote sensing data set into normal image areas are respectively marked as the first image, the second image and the third image, and feature data of the first image, the second image and the third image are respectively extracted by the remote sensing image processing algorithm to obtain the feature data corresponding to the first image, the second image and the third image, and the feature data includes several groups of image texture features, image color features and image semantic features.

[0026] Furthermore, the process of constructing the feature layers and fusing the feature layers to generate corresponding surface feature layers and rock mass feature layers includes:

[0027] Set up mapping layers and shared layers;

[0028] The mapping layer is used to construct a feature layer, and several groups of image texture features, image color features, and image semantic features corresponding to the first image, the second image, and the third image are used as mapping elements of the mapping layer, and the mapping elements are mapped to the mapping layer, and then after the first image, the second image, and the third image are mapped to the mapping layer, the corresponding feature layer 1, feature layer 2, and feature layer 3 are constructed;

[0029] The feature layers and corresponding feature data corresponding to the remote sensing images of the surface around the tunnel in the feature layer 1, the feature layer 2 and the feature layer 3 are used as the shared element 1, and the feature layers and corresponding feature data corresponding to the remote sensing images of the swelling rock mass in the feature layer 1, the feature layer 2 and the feature layer 3 are used as the shared element 2;

[0030] The shared layer is used to perform feature layer fusion, input shared element one into the shared layer, and then the shared layer marks the fusion point of each surface layer element included in the shared element one, and all surface layer elements are associated with the corresponding fusion points in the shared layer, and then the corresponding surface feature layer is generated, and shared element two is input into the shared layer, and then the shared layer marks the fusion point of each rock layer element included in the shared element two, and all rock layer elements are associated with the corresponding fusion points in the shared layer, and then the corresponding rock feature layer is generated.

[0031] Furthermore, the process of analyzing geological information based on the surface characteristic layer and the rock mass characteristic layer and generating corresponding geological identification information includes:

[0032] Set the standard reference layers corresponding to the surface feature layer and the rock mass feature layer, enter the corresponding surface analysis data items for the standard reference layers corresponding to the surface feature layers, the surface analysis data items include the regional geological structure type, regional soil settlement degree and regional surface stress degree of several surface points, enter the corresponding rock mass analysis data items for the standard reference layers corresponding to the rock mass feature layers, the rock mass type, rock mass tensile strength, rock mass compressive strength, rock mass shear strength and Poisson's ratio of several rock mass points;

[0033] Divide the surface feature layer into a number of surface points, compare the surface feature layer with the corresponding standard reference layer, and then match the surface analysis data items corresponding to each surface point in the surface feature layer at the surface point in the corresponding standard reference layer, and use all the acquired surface analysis data items as the surface identification information corresponding to the current surface feature layer; divide the rock mass feature layer into a number of rock mass points, compare the rock mass feature layer with the corresponding standard reference layer, and then match the rock mass analysis data items corresponding to each rock mass point in the rock mass feature layer at the rock mass point in the corresponding standard reference layer, and use all the acquired rock mass analysis data items as the rock mass identification information corresponding to the current rock mass feature layer;

[0034] The surface identification information and the rock mass identification information are summarized as the geological identification information, which includes the regional soil settlement degree and the regional surface stress degree at different surface points around the tunnel, as well as the rock mass tensile strength, rock mass compressive strength, rock mass shear strength and Poisson's ratio at different rock mass points of the expansive rock mass.

[0035] Furthermore, the process of constructing a rock mass mechanics model based on geological identification information includes:

[0036] The regional soil settlement degree and regional surface stress degree at different surface points in the geological identification information are used as the first modeling data for constructing the rock mass mechanics model, and the rock mass tensile strength, rock mass compressive strength, rock mass shear strength and Poisson's ratio at different rock mass points are used as the second modeling data for constructing the rock mass mechanics model;

[0037] A convolutional neural network is constructed by deep learning technology, the first modeling data and the second modeling data are defined as input parameters of two input nodes corresponding to the convolutional neural network, three output nodes of the convolutional neural network are defined, and the three output nodes are marked as Output1, Output2 and Output3 respectively, and then a rock mass mechanics model is constructed by the convolutional neural network;

[0038] Among them, the output node Output1 is used to output the surface stress coefficient corresponding to the rock mechanics model, the output node Output2 is used to output the swelling rock stress coefficient corresponding to the rock mechanics model, and the output node Output3 is used to output the comprehensive stress coefficient corresponding to the rock mechanics model. The surface stress coefficient, swelling rock stress coefficient and comprehensive stress coefficient are marked as γ1, γ2 and γ3 respectively.

[0039] Furthermore, the process of analyzing the stress state of the expansive rock mass by the rock mass mechanics model includes:

[0040] The coefficient thresholds corresponding to the surface stress coefficient, swelling rock stress coefficient and comprehensive stress coefficient output by the rock mass mechanics model are set, and they are recorded as ρ1, ρ2 and ρ3 respectively. By comparing the numerical relationships between γ1 and ρ1, γ2 and ρ2, and γ3 and ρ3, the stress state of the current swelling rock mass is analyzed.

[0041] When γ1<ρ1, γ2<ρ2 and γ3<ρ3 are simultaneously established, the stress state of the corresponding swelling rock mass is judged to be stress safe;

[0042] When at least one of the conditions γ1<ρ1, γ2<ρ2 and γ3<ρ3 is not satisfied, the stress state corresponding to the current swelling rock mass is judged to be stress dangerous.

[0043] Furthermore, the process of generating corresponding warning information according to the result of stress state analysis and then executing corresponding intervention measures includes:

[0044] When the result of the stress state analysis of the swelling rock mass is stress safety, the swelling rock mass is currently in a safe stress environment and no warning information is generated;

[0045] When the result of the stress state analysis of the swelling rock mass is stress danger, the swelling rock mass is currently in a dangerous stress environment, and a corresponding warning message is generated. The content of the warning message is: "The current swelling rock mass stress is abnormal, please arrange the corresponding maintenance personnel to perform the corresponding intervention measures in time";

[0046] The intervention measures include decompression measures, reinforcement support measures, measures to reduce void water pressure and filling measures. After the corresponding intervention measures are executed, the rock mass mechanics model continues to analyze the stress state of the current expansive rock mass until the stress state is stress safe, and then the execution of the intervention measures is stopped.

[0047] Compared with the prior art, the invention has the following beneficial effects: remote sensing data of the surface around the tunnel and the swelling rock mass are collected by the remote sensing data collection module to generate a remote sensing data set, the remote sensing data processing module processes the data to generate a surface feature layer and a rock mass feature layer, the geological information analysis module performs geological information analysis on the surface feature layer and the rock mass feature layer, and generates corresponding geological identification information, the stress state detection module constructs a rock mass mechanics model according to the geological identification information, and the rock mass mechanics model is used to analyze the stress state of the swelling rock mass, the early warning intervention module generates corresponding early warning information according to the result of the stress state analysis, and then corresponding intervention measures are executed according to the early warning information, the remote sensing data of the surface around the tunnel and the swelling rock mass are comprehensively processed and analyzed, and the accurate acquisition of stress-related data of the swelling rock mass in the tunnel is realized, the constructed rock mass mechanics model is used to analyze the stress state of the swelling rock mass, the stress condition of the tunnel swelling rock is timely monitored, and corresponding intervention measures are taken to intervene in the tunnel swelling rock under abnormal stress conditions, thereby ensuring the efficient and safe progress of the tunnel project. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a schematic diagram of the present invention. DETAILED DESCRIPTION

[0049] like Figure 1 As shown, a tunnel expansion rock stress detection system based on remote sensing data includes a detection center, which is communicatively connected to a remote sensing data acquisition module, a remote sensing data processing module, a geological information analysis module, a stress state detection module, and an early warning intervention module;

[0050] The remote sensing data acquisition module is used to collect remote sensing data of the surface around the tunnel and the swelling rock mass, and generate a corresponding remote sensing data set;

[0051] The remote sensing data processing module is used to obtain remote sensing data sets to perform abnormal data calibration, image correction, feature data extraction, feature layer construction and feature layer fusion, and then generate corresponding surface feature layers and rock mass feature layers;

[0052] The geological information analysis module is used to analyze geological information based on the surface characteristic layer and the rock mass characteristic layer, and generate corresponding geological identification information;

[0053] The stress state detection module is used to construct a rock mass mechanics model according to geological identification information, and to perform stress state analysis on the swelling rock mass through the rock mass mechanics model;

[0054] The early warning intervention module is used to generate corresponding early warning information according to the result of stress state analysis, and then execute corresponding intervention measures.

[0055] It should be further explained that, in the specific implementation process, the process of collecting remote sensing data of the surface around the tunnel and the swelling rock mass and generating the corresponding remote sensing data set includes:

[0056] Selecting a collection area where remote sensing collection operations are required, arranging and starting remote sensing collection equipment in the collection area, wherein the remote sensing collection equipment includes satellite remote sensing imaging equipment, drone remote sensing imaging equipment, and aerial remote sensing imaging equipment, and determining whether the equipment collection range of each remote sensing collection equipment can cover the collection area;

[0057] If yes, remote sensing data of the surface around the tunnel and the swelling rock mass are collected by different remote sensing collection equipment, wherein the remote sensing data collected by the satellite remote sensing imaging equipment is marked as Data1, the remote sensing data collected by the drone remote sensing imaging equipment is marked as Data2, and the remote sensing data collected by the aerial remote sensing imaging equipment is marked as Data3;

[0058] Furthermore, Data1 = {d1 隧道 , d1 岩体}, Data2 = {d2 隧道 , d2 岩体}, Data3 = {d3 隧道 , d3 岩体}, where d1 隧道 and d1 岩体 They are the remote sensing data of the surface around the tunnel and the remote sensing data of the swelling rock mass collected by satellite remote sensing imaging equipment. Similarly, d2 隧道 and d2 岩体 They are the remote sensing data of the surface around the tunnel and the remote sensing data of the swelling rock mass collected by the UAV remote sensing imaging equipment, d3 隧道 and d3 岩体 They are the remote sensing data of the surface around the tunnel and the remote sensing data of the swelling rock mass collected by aerial remote sensing imaging equipment;

[0059] Then, Data1, Data2 and Data3 are merged to generate the remote sensing data set of the acquisition area;

[0060] If not, the device collection range of each remote sensing collection device is adjusted until the device collection range is able to cover the collection area, and remote sensing data of the corresponding collection area is collected.

[0061] It should be further explained that, in the specific implementation process, the process of obtaining remote sensing data sets for abnormal data calibration and image correction includes:

[0062] The remote sensing data processing module obtains the remote sensing data set collected by the remote sensing data collection module, and performs abnormal data calibration on Data1, Data2 and Data3 in the remote sensing data set in turn. The data type of the remote sensing data corresponding to the Data1, Data2 and Data3 is a remote sensing image, and the remote sensing image includes a remote sensing image of the surface around the tunnel and a remote sensing image of the swelling rock mass. The remote sensing images corresponding to Data1, Data2 and Data3 are pixelated and segmented to generate a plurality of pixel areas with fixed area sizes.

[0063] The abnormal data calibration is used to calibrate the pixel area in an abnormal state, obtain a number of pixel units corresponding to each pixel area, and obtain the pixel exposure, pixel clarity and pixel unit distortion information corresponding to the pixel unit;

[0064] If the pixel exposure of the pixel unit in the pixel area is less than or equal to the preset exposure threshold, the pixel clarity of the pixel unit is greater than or equal to the preset clarity threshold, and it is determined through the pixel unit distortion information that the current pixel unit does not have distortion, then the pixel unit is marked as a qualified pixel unit, otherwise, the pixel unit is marked as an abnormal pixel unit;

[0065] When the number of abnormal pixel units in the pixel area is greater than or equal to 1 / 3 of the total number of pixel units, the corresponding pixel area is marked as an image area to be corrected in an abnormal state, otherwise, the corresponding pixel area is marked as a normal image area in a normal state;

[0066] The image area to be errata includes the image area to be errata corresponding to the surface around the tunnel and the image area to be errata corresponding to the swelling rock mass. Similarly, the normal image area also includes the normal image area corresponding to the surface around the tunnel and the swelling rock mass.

[0067] The contents of image correction are as follows: obtain all the image areas to be corrected, and then locate the abnormal pixel units in the image areas to be corrected, and analyze the distortion types of the abnormal pixel units, which include geometric transformation distortion and lens distortion. For geometric transformation distortion, a convolutional neural network model is constructed through deep learning, and the remote sensing image corresponding to the image area to be corrected corresponding to the geometric transformation distortion is obtained as sample data and input into the convolutional neural network model for training. The convolutional neural network model adaptively learns and corrects the geometric transformation distortion. For lens distortion, the lens parameters of the corresponding image lens of the remote sensing acquisition equipment are reconfigured, and the remote sensing image is re-acquired until the remote sensing image has no lens distortion.

[0068] When the distortion processing of the abnormal pixel unit is completed, the pixel exposure of the abnormal pixel unit is adjusted to the exposure threshold and below the exposure threshold through the PS software, and the pixel clarity of the pixel unit is adjusted to the clarity threshold and above through the image sharpening or image enhancement algorithm, thereby converting the entire image area to be calibrated into a normal image area.

[0069] It should be further explained that, in a specific implementation process, the process of extracting the feature data includes:

[0070] The remote sensing image obtained by converting all the image areas to be errata corresponding to the remote sensing data of Data1 in the remote sensing data set into normal image areas is labeled as the first image, the remote sensing image obtained by converting all the image areas to be errata corresponding to the remote sensing data of Data2 into normal image areas is labeled as the second image, and the remote sensing image obtained by converting all the image areas to be errata corresponding to the remote sensing data of Data3 into normal image areas is labeled as the third image;

[0071] Extracting feature data from the first image, the second image, and the third image using a remote sensing image processing algorithm, thereby obtaining feature data corresponding to each of the first image, the second image, and the third image, wherein the feature data includes image texture features, image color features, and image semantic features;

[0072] The content of feature data extraction by remote sensing image processing algorithm is as follows: the grayscale co-occurrence feature, RGB value and semantic keyword of each pixel unit in the remote sensing image are obtained through remote sensing image processing algorithm, the pixel units under the same grayscale co-occurrence feature are marked as a group, and all the grayscale co-occurrence features of the group are summarized as the image texture feature of the group;

[0073] Mark all pixel units whose RGB values ​​are in a preset RGB grouping interval as a group, and summarize the color distribution, hue contrast, and color saturation of all pixel units in the group as image color features of the pixel units in the group;

[0074] A semantic bag-of-words model is established. The semantic bag-of-words model predefines the semantic scene type and semantic object category of the remote sensing image, and the semantic keywords corresponding to all pixel units of the remote sensing image are input into the semantic bag-of-words model. Then, the pixel units in the same semantic object category and semantic scene type are marked as a group, and the semantic keywords of all pixel units in the group are summarized as the image semantic features of the group.

[0075] It should be noted that the RGB grouping interval is a numerical interval that limits a class of RGB value ranges. Pixel units are grouped into different image color features through different RGB grouping intervals. Grayscale co-occurrence feature is a means to describe image texture features. It captures the texture information of the image by analyzing the spatial relationship between the grayscale levels of pixels in the image. Semantic keywords describe the semantic information and high-level semantic structure in the image, and are used for tasks such as image understanding, automatic annotation, and image retrieval.

[0076] It should be further explained that, in the specific implementation process, the process of constructing the feature layer and fusing the feature layer to generate the corresponding surface feature layer and rock mass feature layer includes:

[0077] Set up mapping layers and shared layers;

[0078] The mapping layer is used to construct the feature layer, and the working period of the feature layer construction is set, and the working period is recorded as T 构建 , T 构建 =[t1, t2], where t1 is the start time of the working period and t2 is the end time of the working period;

[0079] At the start time corresponding to t1, several groups of image texture features, image color features and image semantic features corresponding to the first image are used as mapping elements of the mapping layer, and the mapping elements are mapped to the corresponding mapping layer. Similarly, several groups of image texture features, image color features and image semantic features corresponding to the second image and the third image are used as mapping elements of their respective mapping layers, and are mapped to their respective corresponding mapping layers.

[0080] After the first image, the second image and the third image are mapped to the mapping layer, the corresponding feature layer 1, feature layer 2 and feature layer 3 are constructed. At the end time corresponding to t2, the partial feature layers and corresponding feature data corresponding to the remote sensing images of the surface around the tunnel in the feature layer 1, feature layer 2 and feature layer 3 are used as the shared element 1, and the partial feature layers and corresponding feature data corresponding to the remote sensing images of the swelling rock mass in the feature layer 1, feature layer 2 and feature layer 3 are used as the shared element 2;

[0081] The shared layer is used to perform feature layer fusion, and a working period for feature layer fusion is set, which is recorded as T 融合 , T 融合 =[t3, t4], where t3 is the start time of the working period and t4 is the end time of the working period;

[0082] At the start time corresponding to t3, the shared element one is input into the shared layer, and then the shared layer marks the fusion point of each surface layer element included in the shared element one, and all the surface layer elements are associated with the corresponding fusion points in the shared layer, and then the corresponding surface feature layer is generated. Similarly, the shared element two is input into the shared layer, and then the shared layer marks the fusion point of each rock layer element included in the shared element two, and all the rock layer elements are associated with the corresponding fusion points in the shared layer, and then the corresponding rock feature layer is generated;

[0083] At the end time corresponding to t4, it is determined whether the surface feature layer and the rock mass feature layer meet the corresponding layer integrity. If so, the construction of the surface feature layer and the rock mass feature layer is completed. If not, the number and position of the fusion points in the shared layer are adjusted until the surface feature layer and the rock mass feature layer meet the corresponding layer integrity.

[0084] It should be noted that the surface layer elements and the rock layer elements respectively represent the characteristic data of the remote sensing image corresponding to the surface around the tunnel and the remote sensing image corresponding to the swelling rock mass, which are used to reflect the relevant information of their respective geological characteristics.

[0085] It should be further explained that, in the specific implementation process, the process of analyzing geological information based on the surface characteristic layer and the rock mass characteristic layer and generating corresponding geological identification information includes:

[0086] Set the corresponding standard reference layers of the surface characteristic layer and the rock mass characteristic layer for geological information analysis;

[0087] The content of geological information analysis is: entering corresponding surface analysis data items for the standard reference layer corresponding to the surface feature layer, wherein the surface analysis data items include the regional geological structure type, regional soil settlement degree and regional surface stress degree of several surface points;

[0088] Enter corresponding rock mass analysis data items for the standard reference layer corresponding to the rock mass characteristic layer, wherein the rock mass analysis data items include rock mass type, rock mass tensile strength, rock mass compressive strength, rock mass shear strength and Poisson's ratio of several rock mass points;

[0089] Divide the surface feature layer into a number of surface points, compare the surface feature layer with its corresponding standard reference layer, and then match the surface analysis data items corresponding to each surface point in the surface feature layer at the surface point in the corresponding standard reference layer, and use all the acquired surface analysis data items as the surface identification information corresponding to the current surface feature layer, wherein the surface points of the surface feature layer and its standard reference layer at the same regional geological structure type are matched as a matching object;

[0090] Similarly, the rock mass characteristic layer is divided into several rock mass points, and the rock mass characteristic layer is compared with its corresponding standard reference layer, so as to match the rock mass analysis data items corresponding to each rock mass point in the rock mass characteristic layer at the rock mass point in the corresponding standard reference layer, and all the obtained rock mass analysis data items are used as the rock mass identification information corresponding to the current rock mass characteristic layer, wherein the surface points of the rock mass characteristic layer and its standard reference layer at the same rock mass type are matched as a matching object;

[0091] The surface identification information and rock mass identification information obtained through geological information analysis are then summarized as corresponding geological identification information, that is, the geological identification information includes the regional soil settlement degree and regional surface stress degree at different surface points around the tunnel, as well as the rock mass tensile strength, rock mass compressive strength, rock mass shear strength and Poisson's ratio at different rock mass points of the expansive rock mass.

[0092] It should be further explained that, in the specific implementation process, the process of constructing a rock mass mechanics model according to geological identification information and analyzing the stress state of the expansive rock mass through the rock mass mechanics model includes:

[0093] The regional soil settlement degree and regional surface stress degree at different surface points in the geological identification information are used as the first modeling data for constructing the rock mass mechanics model, and the rock mass tensile strength, rock mass compressive strength, rock mass shear strength and Poisson's ratio at different rock mass points are used as the second modeling data for constructing the rock mass mechanics model;

[0094] A convolutional neural network is constructed by deep learning technology, the first modeling data and the second modeling data are defined as input parameters of two input nodes corresponding to the convolutional neural network, three output nodes of the convolutional neural network are defined, and the three output nodes are marked as Output1, Output2 and Output3 respectively, and then a rock mass mechanics model is constructed by the convolutional neural network;

[0095] Among them, the output node Output1 is used to output the surface stress coefficient corresponding to the rock mass mechanics model, the output node Output2 is used to output the swelling rock stress coefficient corresponding to the rock mass mechanics model, and the output node Output3 is used to output the comprehensive stress coefficient corresponding to the rock mass mechanics model. The surface stress coefficient, swelling rock stress coefficient and comprehensive stress coefficient are marked as γ1, γ2 and γ3 respectively;

[0096] Then there is, Among them, Ave 沉降 The average value of the settlement depth corresponding to the regional soil settlement degree of several surface points is: Several surface points are numbered, and the number is recorded as i, i = 1, 2, 3, ..., n, n is a natural number greater than 0, h[i] represents the settlement depth corresponding to the i-th surface point, Ave 受力 It represents the average value of the pressure value corresponding to the regional surface stress degree of several surface points. P[i] represents the pressure value corresponding to the i-th surface point, 1 / 2σ1 2 Indicates Ave 沉降 and Ave 受力 The mean of the sum of their variances, σ 2 1 represents Ave 沉降 and Ave 受力 The sum of the variances of

[0097]

[0098] Among them, Ave 抗拉 、Ave 抗压 、Ave 抗剪切 and Ave 泊松比 They represent the average tensile strength, compressive strength, shear strength and Poisson's ratio of several rock mass points of the expansive rock, and 1 / 4σ2 represents Ave 抗拉 、Ave 抗压 、Ave 抗剪切 and Ave 泊松比 The mean of the sum of their standard deviations, σ2, represents Ave 抗拉 、Ave 抗压 、Ave 抗剪切 and Ave 泊松比 The sum of their respective standard deviations;

[0099] Number several rock mass points, denoted by j, where j = 1, 2, 3, ..., m, where m is a natural number greater than 0, then KL[j] represents the tensile strength of the rock mass at the jth rock mass point. KL[j] represents the compressive strength of the rock mass at the jth rock mass point. KJ[j] represents the shear strength of the rock mass at the jth rock mass point. represents the Poisson's ratio of the jth rock mass point;

[0100] , where V1 is the surface volume corresponding to the ground surface around the tunnel where the swelling rock mass is located, and V2 is the volume of the swelling rock mass itself;

[0101] The coefficient thresholds corresponding to the surface stress coefficient, swelling rock stress coefficient and comprehensive stress coefficient output by the rock mass mechanics model are set, and they are recorded as ρ1, ρ2 and ρ3 respectively. By comparing the numerical relationships between γ1 and ρ1, γ2 and ρ2, and γ3 and ρ3, the stress state of the current swelling rock mass is analyzed.

[0102] When γ1<ρ1, γ2<ρ2 and γ3<ρ3 are simultaneously established, the stress state of the corresponding swelling rock mass is judged to be stress safe;

[0103] When at least one of the conditions γ1<ρ1, γ2<ρ2 and γ3<ρ3 is not satisfied, the stress state corresponding to the current swelling rock mass is judged to be stress dangerous.

[0104] It should be noted that the numerical meanings of the coefficient thresholds ρ1, ρ2 and ρ3 corresponding to the surface stress coefficient, the swelling rock stress coefficient and the comprehensive stress coefficient are: the swelling rock mass is in a stress-safe state, the surface stress coefficient, the swelling rock stress coefficient and the comprehensive stress coefficient have their respective maximum limit values, and when any one of the values ​​exceeds the maximum limit value, the swelling rock mass is in a stress-dangerous state.

[0105] It should be further explained that, in the specific implementation process, the process of generating corresponding early warning information according to the results of stress state analysis and then executing corresponding intervention measures includes:

[0106] When the result of the stress state analysis of the swelling rock mass is stress safety, it means that the swelling rock mass is currently in a safe stress environment, and there is no stress-related risk problem for the time being, and no early warning information is generated;

[0107] When the result of the stress state analysis of the swelling rock mass is stress danger, it means that the swelling rock mass is currently in a dangerous stress environment and risk problems caused by stress may occur at any time. At this time, the corresponding warning information is generated, and the content of the warning information is: "The current swelling rock mass stress is abnormal, please arrange the corresponding maintenance personnel to perform the corresponding intervention measures in time";

[0108] The intervention measures include decompression measures, reinforcement support measures, measures to reduce void water pressure and filling measures. After all or part of the intervention measures are executed, the rock mechanics model continues to analyze the stress state of the current expansive rock mass until the stress state is stress safe, and then the execution of the intervention measures is stopped.

[0109] It should be noted that the decompression measures are: by controlling the mining speed of the tunnel and adopting appropriate mining methods and mining processes, the stress impact on the swelling rock mass and the stress level are reduced; the reinforcement support measures are: setting up support structures around the swelling rock mass, such as anchor rods, shotcrete nets and steel frame supports, to enhance the stability and bearing capacity of the swelling rock mass; the measures to reduce the pore water pressure are: taking precipitation and drainage measures to reduce the pore water pressure inside the swelling rock mass and alleviate the pressure state of the swelling rock mass, so as to improve the stability of the swelling rock mass; the filling measures are: filling inside the swelling rock mass to fill the rock caves or cracks, reduce the stress concentration and deformation inside the swelling rock mass, and improve the overall stability of the swelling rock mass.

[0110] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A tunnel expansion rock stress detection system based on remote sensing data, including a detection center, characterized in that: The detection center is communicatively connected to a remote sensing data acquisition module, a remote sensing data processing module, a geological information analysis module, a stress state detection module, and an early warning intervention module; The remote sensing data acquisition module is used to collect remote sensing data of the surface around the tunnel and the swelling rock mass, and generate a corresponding remote sensing data set; The remote sensing data processing module is used to obtain remote sensing data sets to perform abnormal data calibration, image correction, feature data extraction, feature layer construction and feature layer fusion, and then generate corresponding surface feature layers and rock mass feature layers; The geological information analysis module is used to analyze geological information based on the surface characteristic layer and the rock mass characteristic layer, and generate corresponding geological identification information; The stress state detection module is used to construct a rock mass mechanics model according to geological identification information, and to perform stress state analysis on the swelling rock mass through the rock mass mechanics model; The early warning intervention module is used to generate corresponding early warning information according to the results of the stress state analysis, and then execute corresponding intervention measures; The process of analyzing geological information based on surface feature layers and rock mass feature layers and generating corresponding geological identification information includes: Set the standard reference layers corresponding to the surface feature layer and the rock mass feature layer, enter the corresponding surface analysis data items for the standard reference layers corresponding to the surface feature layers, the surface analysis data items include the regional geological structure type, regional soil settlement degree and regional surface stress degree of several surface points, enter the corresponding rock mass analysis data items for the standard reference layers corresponding to the rock mass feature layers, the rock mass type, rock mass tensile strength, rock mass compressive strength, rock mass shear strength and Poisson's ratio of several rock mass points; Divide the surface feature layer into a number of surface points, compare the surface feature layer with the corresponding standard reference layer, and then match the surface analysis data items corresponding to each surface point in the surface feature layer at the surface point in the corresponding standard reference layer, and use all the acquired surface analysis data items as the surface identification information corresponding to the current surface feature layer; divide the rock mass feature layer into a number of rock mass points, compare the rock mass feature layer with the corresponding standard reference layer, and then match the rock mass analysis data items corresponding to each rock mass point in the rock mass feature layer at the rock mass point in the corresponding standard reference layer, and use all the acquired rock mass analysis data items as the rock mass identification information corresponding to the current rock mass feature layer; Summarize the surface identification information and the rock mass identification information as geological identification information, which includes the regional soil settlement degree and regional surface stress degree at different surface points around the tunnel, and the rock mass tensile strength, rock mass compressive strength, rock mass shear strength and Poisson's ratio at different rock mass points of the expansive rock mass; The process of constructing a rock mass mechanics model based on geological identification information includes: The regional soil settlement degree and regional surface stress degree at different surface points in the geological identification information are used as the first modeling data for constructing the rock mass mechanics model, and the rock mass tensile strength, rock mass compressive strength, rock mass shear strength and Poisson's ratio at different rock mass points are used as the second modeling data for constructing the rock mass mechanics model; A convolutional neural network is constructed by deep learning technology, the first modeling data and the second modeling data are defined as input parameters of two input nodes corresponding to the convolutional neural network, three output nodes of the convolutional neural network are defined, and the three output nodes are marked as Output1, Output2 and Output3 respectively, and then a rock mass mechanics model is constructed by the convolutional neural network; Among them, the output node Output1 is used to output the surface stress coefficient corresponding to the rock mechanics model, the output node Output2 is used to output the swelling rock stress coefficient corresponding to the rock mechanics model, and the output node Output3 is used to output the comprehensive stress coefficient corresponding to the rock mechanics model. The surface stress coefficient, swelling rock stress coefficient and comprehensive stress coefficient are marked as γ1, γ2 and γ3 respectively.

2. The tunnel expansion rock stress detection system based on remote sensing data according to claim 1 is characterized in that: The process of collecting remote sensing data of the surface and swelling rock mass around the tunnel and generating the corresponding remote sensing data set includes: Select a collection area, arrange and start remote sensing collection equipment in the collection area. Remote sensing collection equipment includes satellite remote sensing imaging equipment, drone remote sensing imaging equipment and aerial remote sensing imaging equipment. Determine whether the equipment collection range of each remote sensing collection equipment can cover the collection area; If yes, remote sensing data of the surface around the tunnel and the swelling rock mass are collected by different remote sensing acquisition equipment, wherein the remote sensing data collected by the satellite remote sensing imaging equipment, the UAV remote sensing imaging equipment and the aerial remote sensing imaging equipment are respectively marked as Data1, Data2 and Data3, and Data1, Data2 and Data3 are merged to generate a remote sensing data set of the acquisition area; If not, adjust the device collection range of each remote sensing collection device until the device collection range is able to cover the collection area, and collect the corresponding remote sensing data.

3. The tunnel expansion rock stress detection system based on remote sensing data according to claim 2 is characterized in that: The process of obtaining remote sensing data sets for abnormal data calibration includes: Acquire a remote sensing data set, and perform abnormal data calibration on Data1, Data2 and Data3 in the remote sensing data set in turn. The data type of the remote sensing data corresponding to Data1, Data2 and Data3 is remote sensing image, and the remote sensing image includes the remote sensing image of the surface around the tunnel and the remote sensing image of the swelling rock mass. Perform pixel segmentation on the remote sensing images corresponding to Data1, Data2 and Data3, and then generate a number of pixel areas with fixed area sizes. Abnormal data calibration is used to calibrate the pixel area in abnormal state, obtain a number of pixel units corresponding to each pixel area, and obtain the pixel exposure, pixel clarity and pixel unit distortion information corresponding to the pixel unit; If the pixel exposure of the pixel unit in the pixel area is less than or equal to the preset exposure threshold, the pixel clarity of the pixel unit is greater than or equal to the preset clarity threshold, and it is determined through the pixel unit distortion information that the current pixel unit does not have distortion, then the pixel unit is marked as a qualified pixel unit, otherwise, the pixel unit is marked as an abnormal pixel unit; When the number of abnormal pixel units in the pixel area is greater than or equal to 1 / 3 of the total number of pixel units, the corresponding pixel area is marked as an image area to be corrected in an abnormal state, otherwise, the corresponding pixel area is marked as a normal image area in a normal state; The image area to be errata includes the image area to be errata corresponding to the surface around the tunnel and the image area to be errata corresponding to the swelling rock mass, and the normal image area also includes the normal image areas corresponding to the surface around the tunnel and the swelling rock mass.

4. The tunnel expansion rock stress detection system based on remote sensing data according to claim 3 is characterized in that: The process of performing the image errata includes: Obtain all the image areas to be errata, locate the abnormal pixel units in the image areas to be errata, and analyze the distortion types of the abnormal pixel units. The distortion types include geometric transformation distortion and lens distortion. For geometric transformation distortion, build a convolutional neural network model through deep learning, and obtain the remote sensing image corresponding to the image area to be errata corresponding to the geometric transformation distortion as sample data. Input the sample data into the convolutional neural network model for training, and use the convolutional neural network model to correct the geometric transformation distortion. For lens distortion, reconfigure the lens parameters of the corresponding image lens of the remote sensing acquisition equipment, and re-acquire the remote sensing image until there is no lens distortion in the remote sensing image. When the distortion processing of the abnormal pixel unit is completed, the pixel exposure of the abnormal pixel unit is adjusted to the exposure threshold and below the exposure threshold through the PS software, and the pixel clarity of the pixel unit is adjusted to the clarity threshold and above through the image sharpening or image enhancement algorithm, thereby converting the entire image area to be calibrated into a normal image area.

5. The tunnel expansion rock stress detection system based on remote sensing data according to claim 4 is characterized in that: The process of extracting the feature data includes: The remote sensing images after converting all the image areas to be corrected corresponding to the remote sensing data of Data1, Data2 and Data3 in the remote sensing data set into normal image areas are respectively marked as the first image, the second image and the third image, and feature data of the first image, the second image and the third image are respectively extracted by the remote sensing image processing algorithm to obtain the feature data corresponding to the first image, the second image and the third image, and the feature data includes several groups of image texture features, image color features and image semantic features.

6. The tunnel expansion rock stress detection system based on remote sensing data according to claim 5 is characterized in that: The process of constructing the feature layer and fusing the feature layer to generate the corresponding surface feature layer and rock mass feature layer includes: Set up mapping layers and shared layers; The mapping layer is used to construct a feature layer, and several groups of image texture features, image color features, and image semantic features corresponding to the first image, the second image, and the third image are used as mapping elements of the mapping layer, and the mapping elements are mapped to the mapping layer, and then after the first image, the second image, and the third image are mapped to the mapping layer, the corresponding feature layer 1, feature layer 2, and feature layer 3 are constructed; The feature layers and corresponding feature data corresponding to the remote sensing images of the surface around the tunnel in the feature layer 1, the feature layer 2 and the feature layer 3 are used as the shared element 1, and the feature layers and corresponding feature data corresponding to the remote sensing images of the swelling rock mass in the feature layer 1, the feature layer 2 and the feature layer 3 are used as the shared element 2; The shared layer is used to perform feature layer fusion, input shared element one into the shared layer, and then the shared layer marks the fusion point of each surface layer element included in the shared element one, and all surface layer elements are associated with the corresponding fusion points in the shared layer, and then the corresponding surface feature layer is generated, and shared element two is input into the shared layer, and then the shared layer marks the fusion point of each rock layer element included in the shared element two, and all rock layer elements are associated with the corresponding fusion points in the shared layer, and then the corresponding rock feature layer is generated.

7. The tunnel expansion rock stress detection system based on remote sensing data according to claim 6 is characterized in that: The process of analyzing the stress state of expansive rock mass using the rock mass mechanics model includes: The coefficient thresholds corresponding to the surface stress coefficient, swelling rock stress coefficient and comprehensive stress coefficient output by the rock mass mechanics model are set, and they are recorded as ρ1, ρ2 and ρ3 respectively. By comparing the numerical relationships between γ1 and ρ1, γ2 and ρ2, and γ3 and ρ3, the stress state of the current swelling rock mass is analyzed. When γ1<ρ1, γ2<ρ2 and γ3<ρ3 are simultaneously established, the stress state of the corresponding swelling rock mass is judged to be stress safe; When at least one of the conditions γ1<ρ1, γ2<ρ2 and γ3<ρ3 is not satisfied, the stress state corresponding to the current swelling rock mass is judged to be stress dangerous.

8. The tunnel expansion rock stress detection system based on remote sensing data according to claim 7 is characterized in that: The process of generating corresponding early warning information based on the results of stress state analysis and then executing corresponding intervention measures includes: When the result of the stress state analysis of the swelling rock mass is stress safety, the swelling rock mass is currently in a safe stress environment and no warning information is generated; When the result of the stress state analysis of the swelling rock mass is stress danger, the swelling rock mass is currently in a dangerous stress environment, and a corresponding warning message is generated. The content of the warning message is: "The current swelling rock mass stress is abnormal, please arrange the corresponding maintenance personnel to perform the corresponding intervention measures in time"; The intervention measures include decompression measures, reinforcement support measures, measures to reduce void water pressure and filling measures. After the corresponding intervention measures are executed, the rock mass mechanics model continues to analyze the stress state of the current expansive rock mass until the stress state is stress safe, and then the execution of the intervention measures is stopped.

Images