Intelligent prevention and control system for construction disaster risk of water-rich weathered granite gneiss tunnel

Abstract

The application discloses a water-rich weathered granite gneiss tunnel construction disaster risk intelligent prevention and control system and relates to the field of tunnel construction.The system comprises a collection module, a feature extraction module and an iteration module.The collection module is used for collecting image data in a tunnel and storing the image data in the tunnel.The feature extraction module is used for receiving the image data stored in the collection module and extracting feature images of the image data.The iteration module is used for obtaining the feature images of the image data extracted by the feature extraction module, transmitting the feature images to the collection module and iterating each image data corresponding to each feature image.Through visual monitoring technology, the application identifies the contour of the image data after collecting the image data in the construction tunnel in a targeted manner, and then, based on contour image variability analysis, the application monitors the disaster risk in the construction tunnel in real time, synchronously issues an early warning based on the identification result, ensures that construction personnel know the disaster risk state in the construction tunnel in real time and maintains stable tunnel construction under the water-rich weathered granite gneiss geological condition.

Description

Technical Field

[0001] This invention relates to the field of tunnel construction technology, specifically to an intelligent disaster risk prevention and control system for water-rich weathered granite gneiss tunnel construction. Background Technology

[0002] Construction of tunnels in water-rich weathered granite gneiss is extremely challenging. Because the rock is rich in moisture, it is prone to water inrush and collapse. Careful geological surveys are required during construction, along with the use of appropriate methods, such as strict control of blasting parameters in drill-and-blast methods. Simultaneously, advanced support, drainage and waterproofing measures, and real-time monitoring and measurement are essential to comprehensively ensure construction safety and project quality.

[0003] Patent application No. 202310843463.3 discloses an intelligent analysis method for geological disaster prevention and control in tunnel construction. The method includes: constructing a knowledge graph model in the field of geological disaster prevention and control in tunnel construction; structurally representing the knowledge related to disaster prevention and control; conducting disaster prevention and control correlation analysis; mining the potential correlation between prevention and control strategies and geological disasters in tunnel construction through visualization graph analysis technology based on the constructed knowledge graph model; and selecting prevention and control strategies based on the correlation. This application aims to solve the problems that "under complex geological conditions; geological disasters in tunnel construction are affected by the coupling effect of multiple disaster sources; the disaster evolution mechanism is extremely complex; the principles, measures, timing, materials, equipment, procedures and other elements related to prevention and control strategies are closely related to the engineering geological and hydrogeological conditions of the disaster occurrence, as well as the characteristics of the disaster type, location, scale, form, and nature; and the selection of prevention and control measures based on past experience or standard specifications has great uncertainty and blindness."

[0004] However, the construction of tunnels in water-rich weathered granite gneiss is more difficult than that of tunnels in ordinary geological structures.

[0005] To address this, an intelligent prevention and control system for disaster risks during the construction of water-rich weathered granite gneiss tunnels was proposed. Summary of the Invention

[0006] To address the aforementioned shortcomings of existing technologies, this invention provides an intelligent risk prevention and control system for tunnel construction in water-rich weathered granite gneiss, which can effectively solve the problems of existing technologies.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions;

[0008] This invention discloses an intelligent disaster risk prevention and control system for tunnel construction in water-rich weathered granite gneiss, comprising:

[0009] The system comprises the following modules: an acquisition module for collecting and storing image data within the tunnel; a feature extraction module for receiving the stored image data from the acquisition module and extracting feature images from the image data; an iteration module for acquiring the feature images extracted by the feature extraction module and transmitting them back to the acquisition module to iterate through the image data corresponding to each feature image; an analysis module for retrieving the feature images stored in the acquisition module based on the iteration operations and analyzing the tunnel construction hazard risk based on these feature images; and an early warning module for receiving the tunnel construction hazard risk analysis results from the analysis module and deciding whether to issue an early warning message to the user's mobile computer device based on the analysis results.

[0010] Furthermore, the acquisition module is equipped with a construction unit and a selection unit. The construction unit is used to upload the detection information of water-rich weathered granite gneiss in the tunnel construction area, construct a distribution model of water-rich weathered granite gneiss based on the detection information, and simultaneously configure the real coordinates of the distribution model of water-rich weathered granite gneiss so that the position coordinates of any position on the distribution model of water-rich weathered granite gneiss are consistent with its position coordinates in reality. The selection unit is used to traverse the distribution model of water-rich weathered granite gneiss and select image data acquisition points on the distribution model of water-rich weathered granite gneiss.

[0011] Among them, the detection information of water-rich weathered granite gneiss in the tunnel construction area is the distribution location information of water-rich weathered granite gneiss. The closed three-dimensional graphic obtained by connecting the distribution location information of water-rich weathered granite gneiss is recorded as the distribution model of water-rich weathered granite gneiss.

[0012] Among them, the detection information is obtained through high-precision ground-penetrating radar to conduct comprehensive detection of the tunnel construction area.

[0013] Furthermore, after constructing the distribution model of water-rich weathered granite gneiss, the construction unit simultaneously segments the distribution model of water-rich weathered granite gneiss based on the preset tunnel construction area, so that the corresponding area of ​​the tunnel construction area in the distribution model of water-rich weathered granite gneiss is separated from the distribution model of water-rich weathered granite gneiss. The water-rich weathered granite gneiss distribution model after the separation operation is completed is selected by the system user in the selection unit to select the image data acquisition point.

[0014] The acquisition module has a preset image data acquisition frequency by the system user. Based on the image data acquisition frequency, the acquisition module performs continuous image data acquisition operations at the acquisition points. When storing image data, the acquisition module distinguishes and stores the data based on the acquisition point of the image data source and sorts and stores the data based on the acquisition time of the image data source.

[0015] Furthermore, the image data acquisition points also include:

[0016] A custom section capture spacing is used. On the water-rich weathered granite gneiss distribution model that has been cut, the section capture spacing is used to capture the section of the tunnel construction area starting from one end of the tunnel construction area. The water-rich weathered granite gneiss distribution model is divided based on the section to obtain several segmentation surfaces. The area of ​​each segmentation surface is calculated. Based on the area calculation results, each segmentation surface is sorted in descending order. Image data acquisition points are selected again on the segmentation surfaces corresponding to the second half of the descending order queue.

[0017] In the section capture spacing setting stage, the following rules apply: the larger the volume of the water-rich weathered granite gneiss distribution model, the larger the section capture spacing; the smaller the volume of the water-rich weathered granite gneiss distribution model, the smaller the section capture spacing. Furthermore, the machine learning-based prediction model uses the volume of the water-rich weathered granite gneiss distribution model, the complexity of the geological structure, and historical construction disaster data as input features. Through training and learning, a nonlinear relationship model between the volume and the section capture spacing is established to complete the final setting of the section capture spacing.

[0018] Furthermore, the logic for calculating the area of ​​the segmented surface is as follows:

[0019] S split =S out -S in ;

[0020] In the formula: S split S represents the area of ​​the divided surface; out S defines the area of ​​the region bounded by the outer contour of the segmented surface. in The cross-sectional area of ​​the tunnel construction area;

[0021] In this process, when selecting image data acquisition points on the segmented surface, the distances from each point on the cross-sectional contour of the tunnel construction area to each point on the inner contour of the segmented surface are identified. The point on the cross-sectional contour of the tunnel construction area corresponding to the shortest distance is used as the image data acquisition point. When calculating the area of ​​the segmented surface, a method based on pixel statistics and geometric calculation is used to convert it into a pixel matrix for pixel statistics to obtain an approximate area, and then a geometric algorithm is applied for precise correction. When selecting acquisition points, a target detection algorithm based on deep neural networks is used to automatically identify the cross-sectional contour of the tunnel construction area and the inner contour of the segmented surface. The shortest distance point is determined by calculating the Euclidean distance between the point cloud data, which serves as the basic parameter for calculating the area of ​​the segmented surface.

[0022] Furthermore, during the acquisition phase, the acquisition distance and angle of image data acquired at the same acquisition point remain consistent, and the feature image extracted by the feature extraction module is the contour image.

[0023] In the feature image extraction stage, the image data is first optimized, and then the feature image extraction operation is performed.

[0024]

[0025] In the formula: O(x,y) is the optimized image data, i.e., the image data after contour enhancement; I(x,y) is the original image data; α is the weighting coefficient; For Laplace operator; * denotes convolution operation; K σ (x,y) is the Gaussian blur kernel function;

[0026] Where the weighting coefficient α > 0, (x,y) represents the coordinate position in the kernel function, and σ represents the standard deviation of the Gaussian distribution;

[0027] In the optimization process, an adaptive weight adjustment mechanism is introduced. Based on the grayscale distribution and texture complexity of the image, the adaptive weight adjustment mechanism dynamically adjusts the weight coefficients by constructing a weight prediction model based on a convolutional neural network.

[0028] Furthermore, the feature extraction module runs continuously, and each time the feature extraction module runs, the target of receiving image data is all image data stored in the same storage interval in the acquisition module;

[0029] In the feature image retrieval stage, the analysis module retrieves all feature images stored in the same storage interval within the acquisition module.

[0030] Furthermore, the logic for analyzing the disaster risk of tunnel construction in the analysis module is expressed as follows:

[0031]

[0032] In the formula: D represents the difference between the two feature images, i.e., the tunnel construction disaster risk represented by the two feature images; ω1 and ω2 are weights; S is the number of successfully matched key point pairs; M is the set of matched point pairs; (x 1i y 1i (x) represents the coordinates of k1i in the feature image; (x) 2i y 2i ) is k 2i Coordinates in the feature image; D curve The similarity distance between the contour curves of two feature images;

[0033] Among them, the two feature images are used to extract key point sets based on the SIFT algorithm, denoted as k1={k 11 ,k 12 ,...k 1n}、k2={k 21 ,k 22 ,...k 2n}, the set of matching point pairs is obtained through keypoint matching, M = {(k 1i ,k 2i )|i=1,...,s;j=1,...,s};

[0034] The contour curves of the two feature images are discretized into point sets, and D is calculated using the dynamic time warping algorithm. curve ;

[0035] Based on the above formula, the differences are calculated in pairs for all feature images corresponding to each collection point. The maximum difference in the calculation results is identified, a risk judgment threshold is set, and the maximum difference is compared with the risk judgment threshold. Collection points with a maximum difference greater than or equal to the risk judgment threshold are recorded as construction disaster risk points.

[0036] Furthermore, the warning information issued by the warning module is the coordinates of the construction disaster risk points. The warning information is displayed in real time through the system interface and pushed to the system user through a preset communication method.

[0037] When pushing early warning information, a priority-based message push strategy is adopted. The priority of early warning information is determined according to the risk level, with high-risk early warning information being pushed first. At least one communication method is used for redundant push.

[0038] Furthermore, the acquisition module is interconnected with a construction unit and a selection unit via a wireless network, and is also interconnected with a feature extraction module and an iteration module via a wireless network. The acquisition module is also interconnected with an analysis module via a wireless network, and the analysis module is interconnected with an early warning module via a wireless network.

[0039] Compared with the known prior art, the technical solution provided by this invention has the following beneficial effects:

[0040] This invention provides an intelligent disaster risk prevention and control system for tunnel construction in water-rich weathered granite gneiss. During operation, the system uses visual monitoring technology to collect image data inside the construction tunnel, identifies the contours of the image data, and then analyzes the variability of the contour images to monitor the disaster risks inside the construction tunnel in real time. Simultaneously, it issues early warnings based on the identification results, ensuring that construction personnel are aware of the disaster risk status inside the construction tunnel in real time, maintaining stable tunnel construction under the geological conditions of water-rich weathered granite gneiss, and ensuring the stable progress of tunnel construction. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0042] Figure 1 A schematic diagram of the intelligent disaster risk prevention and control system for tunnel construction in water-rich weathered granite gneiss;

[0043] Figure 2 This is a schematic diagram illustrating the area of ​​the segmented surface of the water-rich weathered granite gneiss distribution model in this invention. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0045] The present invention will be further described below with reference to embodiments.

[0046] Example:

[0047] The intelligent prevention and control system for disaster risks in water-rich weathered granite gneiss tunnel construction in this embodiment, such as... Figure 1 As shown, it includes:

[0048] The acquisition module is used to acquire and store image data inside the tunnel.

[0049] The acquisition module has a construction unit and a selection unit at the lower level. The construction unit is used to upload the detection information of water-rich weathered granite gneiss in the tunnel construction area, build a distribution model of water-rich weathered granite gneiss based on the detection information, and simultaneously configure the real coordinates of the distribution model of water-rich weathered granite gneiss so that the position coordinates of any position on the distribution model of water-rich weathered granite gneiss are consistent with its position coordinates in reality. The selection unit is used to traverse the distribution model of water-rich weathered granite gneiss and select image data acquisition points on the distribution model of water-rich weathered granite gneiss.

[0050] Among them, the detection information of water-rich weathered granite gneiss in the tunnel construction area is the distribution location information of water-rich weathered granite gneiss. The closed three-dimensional graphic obtained by connecting the distribution location information of water-rich weathered granite gneiss is recorded as the distribution model of water-rich weathered granite gneiss.

[0051] Among them, the detection information is obtained by using high-precision ground-penetrating radar to conduct all-round detection of the tunnel construction area;

[0052] After constructing the distribution model of water-rich weathered granite gneiss, the construction unit simultaneously segments the distribution model of water-rich weathered granite gneiss based on the preset tunnel construction area. This separates the corresponding area of ​​the tunnel construction area in the distribution model of water-rich weathered granite gneiss from the distribution model of water-rich weathered granite gneiss. After the separation operation is completed, the system user selects the image data acquisition points in the selection unit.

[0053] The acquisition module has a preset image data acquisition frequency by the system user. The acquisition module performs continuous image data acquisition operations at the acquisition points based on the image data acquisition frequency. When storing image data, the acquisition module distinguishes and stores the data based on the acquisition point of the image data source and sorts and stores the data based on the acquisition time of the image data source.

[0054] Image data acquisition points also include:

[0055] A custom section capture spacing is used. On the water-rich weathered granite gneiss distribution model that has been cut, the section capture spacing is used to capture the section of the tunnel construction area starting from one end of the tunnel construction area. The water-rich weathered granite gneiss distribution model is divided based on the section to obtain several segmentation surfaces. The area of ​​each segmentation surface is calculated. Based on the area calculation results, each segmentation surface is sorted in descending order. Image data acquisition points are selected again on the segmentation surfaces corresponding to the second half of the descending order queue.

[0056] In the section capture spacing setting stage, the following applies: the larger the volume of the water-rich weathered granite gneiss distribution model, the larger the section capture spacing; the smaller the volume of the water-rich weathered granite gneiss distribution model, the smaller the section capture spacing. Furthermore, the machine learning-based prediction model uses the volume of the water-rich weathered granite gneiss distribution model, the complexity of the geological structure, and historical construction disaster data as input features. Through training and learning, a nonlinear relationship model between the volume and the section capture spacing is established to complete the final setting of the section capture spacing.

[0057] The logic for calculating the area of ​​a segmented surface is as follows:

[0058] S split =S out -S in ;

[0059] In the formula: S split S represents the area of ​​the divided surface; out S defines the area of ​​the region bounded by the outer contour of the segmented surface. in The cross-sectional area of ​​the tunnel construction area;

[0060] Specifically, when selecting image data acquisition points on the segmented surface, the distances from each point on the cross-sectional contour of the tunnel construction area to each point on the inner contour of the segmented surface are identified. The point on the cross-sectional contour of the tunnel construction area corresponding to the shortest distance is used as the image data acquisition point. When calculating the area of ​​the segmented surface, a method based on pixel statistics and geometric calculation is used to convert it into a pixel matrix for pixel statistics to obtain an approximate area, and then a geometric algorithm is applied for precise correction. When selecting acquisition points, a target detection algorithm based on deep neural networks is used to automatically identify the cross-sectional contour of the tunnel construction area and the inner contour of the segmented surface. The shortest distance point is determined by calculating the Euclidean distance between the point cloud data, which serves as the basic parameter for calculating the area of ​​the segmented surface.

[0061] The feature extraction module is used to receive image data stored in the acquisition module and extract feature images from the image data;

[0062] Image data collected from the same collection point maintains consistent collection distance and angle during the collection phase. The feature extraction module extracts the feature image, i.e., the contour image, from the image data.

[0063] In the feature image extraction stage, the image data is first optimized, and then the feature image extraction operation is performed.

[0064]

[0065] In the formula: O(x,y) is the optimized image data, i.e., the image data after contour enhancement; I(x,y) is the original image data; α is the weighting coefficient; For Laplace operator; * denotes convolution operation; K σ (x,y) is the Gaussian blur kernel function;

[0066] Where the weighting coefficient α > 0, (x,y) represents the coordinate position in the kernel function, and σ represents the standard deviation of the Gaussian distribution;

[0067] In the optimization process, an adaptive weight adjustment mechanism is introduced. The adaptive weight adjustment mechanism dynamically adjusts the weight coefficients based on the gray-scale distribution, texture complexity and other features of the image by constructing a weight prediction model based on a convolutional neural network.

[0068] The image data collected from the acquisition points is optimized by calculating the above logical formula to highlight the contour information in the image data, thereby improving the accuracy of the system's output results.

[0069] The feature extraction module runs continuously. Each time the feature extraction module runs, it receives image data with the target of all image data stored in the same storage interval in the acquisition module.

[0070] During the feature image retrieval phase, the analysis module retrieves all feature images stored in the same segmentation storage area within the acquisition module.

[0071] The iteration module is used to obtain the feature images of the image data extracted by the feature extraction module, and transmit the feature images to the acquisition module to iterate the image data corresponding to each feature image.

[0072] The analysis module is used to retrieve feature images stored based on iterative operations from the acquisition module, and to analyze the disaster risk of tunnel construction based on the feature images;

[0073] The logic for analyzing the disaster risk in tunnel construction within the analysis module is as follows:

[0074]

[0075] In the formula: D represents the difference between the two feature images, i.e., the tunnel construction disaster risk represented by the two feature images; ω1 and ω2 are weights; S is the number of successfully matched key point pairs; M is the set of matched point pairs; (x 1i y 1i ) is k 1i Coordinates in the feature image; (x 2i y 2i ) is k 2i Coordinates in the feature image; D curve The similarity distance between the contour curves of two feature images;

[0076] Among them, the two feature images are used to extract key point sets based on the SIFT algorithm, denoted as k1={k 11 ,k 12 ,...k 1n}、k2={k 21 ,k 22 ,...k 2n}, the set of matching point pairs is obtained through keypoint matching, M = {(k 1i ,k 2i )|i=1,...,s;j=1,...,s};

[0077] The contour curves of the two feature images are discretized into point sets, and D is calculated using the dynamic time warping algorithm. curve ;

[0078] The above logical formula further defines the calculation logic for the differences between feature images in the analysis module, providing support for the further operation of the early warning module of the system in this embodiment.

[0079] Based on the above formula, the difference is calculated in pairs for all feature images corresponding to each collection point. The maximum difference in the calculation result is identified, a risk judgment threshold is set, and the maximum difference is compared with the risk judgment threshold. Collection points with a maximum difference greater than or equal to the risk judgment threshold are recorded as construction disaster risk points.

[0080] The early warning module is used to receive the tunnel construction disaster risk analysis results from the analysis module and decide whether to issue early warning information to the mobile computer devices of system users based on the analysis results.

[0081] The warning information issued by the warning module is the coordinates of the construction disaster risk points. The warning information is displayed in real time through the system interface and pushed to users on the system side through preset communication methods.

[0082] When pushing early warning information, a priority-based message push strategy is adopted. The priority of early warning information is determined according to the risk level, with high-risk early warning information being pushed first. At least one communication method is used for redundant push.

[0083] The acquisition module has a construction unit and a selection unit that are interconnected via a wireless network. The acquisition module also has a feature extraction module and an iteration module that are interconnected via a wireless network. The acquisition module also has an analysis module that is interconnected via a wireless network. The analysis module also has an early warning module that is interconnected via a wireless network.

[0084] In this embodiment, the acquisition module collects and stores image data inside the tunnel. The construction unit simultaneously uploads the detection information of water-rich weathered granite gneiss in the tunnel construction area and constructs a distribution model of water-rich weathered granite gneiss based on the detection information. Simultaneously, it configures real-world coordinates for the distribution model, ensuring that the coordinates of any position on the model match its real-world coordinates. The selection unit iterates through the distribution model in real time, selecting image data acquisition points. The feature extraction module receives the image data stored in the acquisition module and extracts feature images from the image data. The feature extraction module further receives the image data stored in the acquisition module and extracts feature images from the image data. The analysis module then retrieves the feature images stored in the acquisition module based on iterative operations, analyzes the tunnel construction disaster risk based on the feature images, and finally receives the tunnel construction disaster risk analysis results from the analysis module through the early warning module. Based on the analysis results, it decides whether to issue an early warning message to the user's mobile computer device.

[0085] Through the system operation in the above embodiments, intelligent disaster risk prevention and control is brought to the construction scenario of water-rich weathered granite gneiss tunnels, ensuring the stable progress of the construction of water-rich weathered granite gneiss tunnels.

[0086] See Figure 2 As shown in the figure, the shaded area represents the area of ​​the segmentation surface of the distribution model of water-rich weathered granite gneiss.

[0087] In summary, during operation, the system in the above embodiments uses visual monitoring technology to collect image data inside the construction tunnel, identifies the contours of the image data, and then monitors the disaster risks inside the construction tunnel in real time based on contour image variability analysis. Simultaneously, it issues early warnings based on the identification results, ensuring that construction personnel are aware of the disaster risk status inside the construction tunnel in real time, maintaining stable tunnel construction under the geological scenario of water-rich weathered granite gneiss, and ensuring the stable progress of tunnel construction.

[0088] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent prevention and control system for disaster risks in the construction of water-rich weathered granite gneiss tunnels, characterized in that: include: The acquisition module is used to acquire and store image data inside the tunnel. The acquisition module is equipped with a construction unit and a selection unit. The construction unit is used to upload the detection information of water-rich weathered granite gneiss in the tunnel construction area, construct a distribution model of water-rich weathered granite gneiss based on the detection information, and simultaneously configure the real coordinates of the distribution model of water-rich weathered granite gneiss so that the position coordinates of any position on the distribution model of water-rich weathered granite gneiss are consistent with its position coordinates in reality. The selection unit is used to traverse the distribution model of water-rich weathered granite gneiss and select image data acquisition points on the distribution model of water-rich weathered granite gneiss. The feature extraction module is used to receive image data stored in the acquisition module and extract feature images from the image data; The iteration module is used to obtain the feature images of the image data extracted by the feature extraction module, and transmit the feature images to the acquisition module to iterate the image data corresponding to each feature image. The analysis module is used to retrieve feature images stored based on iterative operations from the acquisition module, and to analyze the disaster risk of tunnel construction based on the feature images; The early warning module is used to receive the tunnel construction disaster risk analysis results from the analysis module and decide whether to issue early warning information to the mobile computer device of the system user based on the analysis results; The image data acquisition points also include: A custom section capture spacing is used. On the water-rich weathered granite gneiss distribution model that has been cut, the section capture spacing is used to capture the section of the tunnel construction area starting from one end of the tunnel construction area. The water-rich weathered granite gneiss distribution model is divided based on the section to obtain several segmentation surfaces. The area of ​​each segmentation surface is calculated. Based on the area calculation results, each segmentation surface is sorted in descending order. Image data acquisition points are selected again on the segmentation surfaces corresponding to the second half of the descending order queue. In the section capture spacing setting stage, the following rule applies: the larger the volume of the water-rich weathered granite gneiss distribution model, the larger the section capture spacing; the smaller the volume of the water-rich weathered granite gneiss distribution model, the smaller the section capture spacing. Furthermore, the machine learning-based prediction model uses the volume of the water-rich weathered granite gneiss distribution model, geological structural complexity, historical construction disaster data, etc., as input features. Through training and learning, a nonlinear relationship model between volume and section capture spacing is established to complete the final setting of the section capture spacing. The logic for calculating the area of ​​the segmented surface is as follows: ； In the formula: For the area of ​​the divided surface; The area defined by the outer contour of the segmented surface; The cross-sectional area of ​​the tunnel construction area; Specifically, when selecting image data acquisition points on the segmented surface, the distances from each point on the cross-sectional contour of the tunnel construction area to each point on the inner contour of the segmented surface are identified. The point on the cross-sectional contour of the tunnel construction area corresponding to the shortest distance is used as the image data acquisition point. When calculating the area of ​​the segmented surface, a method based on pixel statistics and geometric calculation is used to convert it into a pixel matrix for pixel statistics to obtain an approximate area, and then a geometric algorithm is applied for precise correction. When selecting acquisition points, a target detection algorithm based on deep neural networks is used to automatically identify the cross-sectional contour of the tunnel construction area and the inner contour of the segmented surface. The shortest distance point is determined by calculating the Euclidean distance between the point cloud data, which serves as the basic parameter for calculating the area of ​​the segmented surface. Image data collected from the same collection point maintains consistent collection distance and angle during the collection phase. The feature image extracted by the feature extraction module from the image data is the contour image. In the feature image extraction stage, the image data is first optimized, and then the feature image extraction operation is performed. ； In the formula: To optimize the processed image data, i.e., the image data with highlighted contours; This is the original image data; These are the weighting coefficients; For the Laplace operator; This represents the convolution operation; This is the Gaussian blur kernel function; The weighting coefficient >0, , This represents the coordinate position in the kernel function. This represents the standard deviation of a Gaussian distribution. In the optimization process, an adaptive weight adjustment mechanism is introduced. The adaptive weight adjustment mechanism dynamically adjusts the weight coefficients based on the gray-scale distribution, texture complexity and other features of the image by constructing a weight prediction model based on a convolutional neural network. The logic for analyzing the disaster risk of tunnel construction in the analysis module is expressed as follows: ； In the formula: The difference between the two feature images is used to represent the tunnel construction disaster risk represented by the two feature images; As weight; The number of keypoint pairs that were successfully matched; For matching point pairs; for Coordinates in the feature image; for Coordinates in the feature image; The similarity distance between the contour curves of two feature images; Specifically, a set of key points is extracted from two feature images using the SIFT algorithm, denoted as […]. The set of matching point pairs is obtained through key point matching. ; The contour curves of the two feature images are discretized into point sets, and the results are calculated using the dynamic time warping algorithm. ,Will Participate in the process of disaster risk analysis during tunnel construction; Based on the above formula, the differences are calculated in pairs for all feature images corresponding to each collection point. The maximum difference in the calculation results is identified, a risk judgment threshold is set, and the maximum difference is compared with the risk judgment threshold. Collection points with a maximum difference greater than or equal to the risk judgment threshold are recorded as construction disaster risk points.

2. The intelligent prevention and control system for disaster risks in the construction of water-rich weathered granite gneiss tunnels according to claim 1, characterized in that, The detection information of water-rich weathered granite gneiss in the tunnel construction area is the distribution location information of water-rich weathered granite gneiss. The closed three-dimensional graphic obtained by connecting the distribution location information of water-rich weathered granite gneiss is recorded as the distribution model of water-rich weathered granite gneiss. Among them, the detection information is obtained through high-precision ground-penetrating radar to conduct all-round detection of the tunnel construction area.

3. The intelligent prevention and control system for disaster risks in the construction of water-rich weathered granite gneiss tunnels according to claim 2, characterized in that, After constructing the distribution model of water-rich weathered granite gneiss, the construction unit simultaneously segments the distribution model of water-rich weathered granite gneiss based on the preset tunnel construction area. This separates the corresponding area of ​​the tunnel construction area in the distribution model of water-rich weathered granite gneiss from the distribution model of water-rich weathered granite gneiss. After the separation operation is completed, the system user selects the image data acquisition points in the selection unit of the distribution model of water-rich weathered granite gneiss. The acquisition module has a preset image data acquisition frequency by the system user. Based on the image data acquisition frequency, the acquisition module performs continuous image data acquisition operations at the acquisition points. When storing image data, the acquisition module distinguishes and stores the data based on the acquisition point of the image data source and sorts and stores the data based on the acquisition time of the image data source.

4. The intelligent prevention and control system for disaster risks in the construction of water-rich weathered granite gneiss tunnels according to claim 1, characterized in that, The feature extraction module runs continuously, and each time the feature extraction module runs, the target of receiving image data is to collect all image data stored in the same storage interval in the acquisition module. In the feature image retrieval stage, the analysis module retrieves all feature images stored in the same storage interval within the acquisition module.

5. The intelligent prevention and control system for disaster risks in the construction of water-rich weathered granite gneiss tunnels according to claim 1, characterized in that, The warning information issued by the warning module is the coordinates of the construction disaster risk points. The warning information is displayed in real time through the system interface and pushed to the system user through a preset communication method. When pushing early warning information, a priority-based message push strategy is adopted. The priority of early warning information is determined according to the risk level, with high-risk early warning information being pushed first. At least one communication method is used for redundant push.

6. The intelligent prevention and control system for disaster risks in the construction of water-rich weathered granite gneiss tunnels according to claim 1, characterized in that, The acquisition module is interconnected with a construction unit and a selection unit via a wireless network. The acquisition module is also interconnected with a feature extraction module and an iteration module via a wireless network. The acquisition module is interconnected with an analysis module via a wireless network. The analysis module is interconnected with an early warning module via a wireless network.

Images