Tunnel face construction information automatic monitoring method and device and storage medium
Through the process identification network model and support component detection model, combined with image and point cloud data processing, the problem of low automation of tunnel palm surface construction information collection is solved, and the automatic collection and intelligent monitoring of information is realized, reducing costs and Energy consumption improves the accuracy and efficiency of construction information.
Patent Information
- Application Number
- CN202411549425.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-01
- Publication Date
- 2024-12-03
AI Technical Summary
The existing technology has low automation, high cost and low intelligence in the construction information collection of tunnel palm surfaces, and the monitoring of rock surface and spray concrete surface flatness problems are huge and the processing time is long.
The process identification network model and support component detection model are adopted, and through tunnel palm surface construction image acquisition and point cloud data processing, the construction process is automatically identified and supported component information is collected, combined with inverse distance weighted interpolation method and sliding window standard deviation calculation , obtain construction information.
It realizes the automatic collection of tunnel palm surface construction information, reduces monitoring costs, improves intelligence level, saves energy consumption, and improves the accuracy and efficiency of construction information.
Smart Images

Figure CN119068421A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of tunnel construction information technology, and in particular to a method, a device and a storage medium for automatically monitoring tunnel face construction information. Background Art
[0002] The construction information of the railway tunnel face mainly refers to the construction time, construction parameters and construction quality of each construction process, including the construction timing of each construction process, construction refinement information, over-excavation and under-excavation, rock surface flatness, shotcrete flatness, etc. By collecting construction information, the management and control level of the tunnel construction process can be further improved, and basic data can be provided for drilling and blasting optimization, intelligent shotcrete and support optimization in railway tunnel face construction.
[0003] At present, the method of collecting the timing of tunnel support component construction is mainly based on manual recording, which is labor-intensive, has a low degree of automation, is difficult to apply data, and has low reliability. The flatness of rock surfaces and shotcrete surfaces is generally collected using three-dimensional laser point clouds, but collecting three-dimensional laser point clouds requires the purchase of special equipment, which is expensive, takes a long time to access the face, and the amount of data is very large, and the transmission and processing takes a long time.
[0004] Therefore, how to realize the automatic collection of construction information in the tunnel face area, reduce monitoring costs, improve monitoring intelligence, and save energy consumption is an important issue in current research. Summary of the invention
[0005] The main technical problem solved by the present invention is to provide a method, device and storage medium for automatic monitoring of tunnel face construction information, so as to solve the problem of how to realize automatic collection of tunnel face area construction information, reduce monitoring costs, improve monitoring intelligence level and save energy consumption.
[0006] In order to solve the above technical problems, a technical solution adopted by the present invention is to provide a method for automatically monitoring construction information of a tunnel face, comprising the steps of: Collect tunnel face construction images.
[0007] The tunnel face construction image is input into the process recognition network model, and the process recognition network model is used to identify and monitor the construction process to obtain the data of each construction process.
[0008] If it is monitored that the construction process is the construction of supporting components, the corresponding construction process data is input into the supporting component detection model to obtain the supporting component construction information.
[0009] In some embodiments, it also includes: collecting point cloud data of the rock surface in the tunnel face construction site before and after the concrete is sprayed.
[0010] The point cloud data is preprocessed to obtain grid point cloud data.
[0011] The flatness information is obtained based on the grid point cloud data.
[0012] In some embodiments, after the step of preprocessing the point cloud data to obtain the grid point cloud data, the method further includes: interpolating the grid points using an inverse distance weighted interpolation method, the method being: ; in, represents the estimated value of the interpolation point, Represents known grid points The value of represents the inverse of the distance between the estimated grid point and the known grid point, Indicates the number of grid points.
[0013] In some embodiments, obtaining flatness information according to grid point cloud data includes: Expand the grid point cloud data along the tunnel centerline to obtain an orthogonal image; The height standard deviation of all pixels in each window is calculated by sliding the orthogonal image. The height standard deviation is used to measure the flatness information.
[0014] In some embodiments, the process identification network model includes: a backbone module, a detection module, and a first classification module.
[0015] The input layer of the backbone module is used to input the tunnel face construction image. The backbone module is used to extract the features of the tunnel face construction image, obtain the feature map, and input the feature map into the detection module and the first classification module respectively.
[0016] The detection module receives the feature map, and detects the construction target according to the feature map to obtain a construction target map.
[0017] The first classification module receives the feature map, classifies the construction processes according to the feature map, and outputs the data of each construction process.
[0018] In some embodiments, the support member detection model includes a segmentation module, a hole position analysis module and a second classification module connected in sequence.
[0019] The input layer of the segmentation module is used to receive the construction process data corresponding to the support component construction, extract the features of the construction process data corresponding to the support component construction, and output the detection segmentation map to the hole position analysis module.
[0020] The hole position analysis module receives the detection segmentation map, detects the position of the support component according to the detection segmentation map, and obtains the point position information of the support component.
[0021] The second classification module classifies and identifies the point information to obtain the support component information.
[0022] In some embodiments, the total loss function of the process identification network model is for: ; in, represents the loss function of construction target detection, represents the construction target detection box loss function, represents the construction process classification loss function.
[0023] Total loss function of the support component detection model for: ; in, represents the loss function of support component detection, represents the support component detection box loss function, represents the segmentation loss function.
[0024] In some embodiments, the method further includes: monitoring whether the construction workers are wearing safety helmets effectively, the method being: Identify the number of pixels in the pixel set corresponding to the construction personnel , identify the number of pixels in the pixel set corresponding to the hard hat adjacent to the construction worker , then the helmet wearing index Calculated by the following formula: ; in, The number of pixels that overlap between the pixel set corresponding to the construction worker and the pixel set corresponding to the helmet. It represents the total number of pixels after the union of the pixel set corresponding to the construction worker and the pixel set corresponding to the helmet. represents the first threshold, Greater than 0.5; If the helmet wearing index If it is greater than or equal to 0, it means that the construction workers are wearing safety helmets effectively.
[0025] In order to solve the above technical problems, the present application also provides an electronic device, including a processor and a memory, wherein the memory stores a computer program that can be executed by the processor, and when the computer program is executed by the processor, any of the above-mentioned methods for automatically monitoring tunnel face construction information is implemented.
[0026] In order to solve the above technical problems, the present application also provides a computer-readable storage medium on which a computer program is stored, and the computer program is used to implement any of the above-mentioned methods for automatically monitoring tunnel face construction information.
[0027] Beneficial effects: The present invention discloses a method, device and storage medium for automatically monitoring construction information of a tunnel face, the method comprising the steps of: collecting tunnel face construction images; inputting tunnel face construction images into a process identification network model, using the process identification network model to identify and monitor construction processes, and obtaining data of each construction process; if the construction process is monitored to be support component construction, the corresponding construction process data is input into a support component detection model to obtain support component construction information. The method of the present application can realize the automatic collection of construction information in the tunnel face area, reduce monitoring costs, and improve the level of intelligent monitoring. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Figure 1 It is a flow chart of an embodiment of a method for automatically monitoring construction information of a tunnel face according to the present invention; Figure 2 It is a structural schematic diagram of an embodiment of a process identification network model according to the present invention; Figure 3 It is a structural schematic diagram of an embodiment of a support component detection model according to the present invention; Figure 4 A schematic diagram of the detection segmentation map is given; Figure 5 yes Figure 4 An enlarged schematic diagram of Figure 6 It is a schematic diagram of the construction position of the anchor bolt construction stage drawn according to the support component information; Figure 7 is a schematic structural diagram of an electronic device according to the present invention; Figure 8 It is a schematic diagram of the structure of a computer-readable storage medium according to the present invention. DETAILED DESCRIPTION
[0029] In order to facilitate the understanding of the present invention, the present invention is described in more detail below in conjunction with the accompanying drawings and specific embodiments. Preferred embodiments of the present invention are provided in the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described in this specification. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present invention more thorough and comprehensive.
[0030] It should be noted that, unless otherwise defined, all technical and scientific terms used in this specification have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used in this specification includes any and all combinations of one or more related listed items.
[0031] It should be noted that the “tunnel face” referred to in this application refers to an excavation working face in an underground engineering or mining engineering.
[0032] Figure 1 A flow chart showing an embodiment of a method for automatically monitoring tunnel face construction information of the present invention includes the following steps: Step S1, collecting tunnel face construction images.
[0033] Step S2: input the tunnel face construction image into the process recognition network model, use the process recognition network model to identify and monitor the construction process, and obtain the data of each construction process.
[0034] Step S3: if it is monitored that the construction process is the support component construction, the corresponding construction process data is input into the support component detection model to obtain the support component construction information.
[0035] The method of the present application monitors the construction process through the process identification model, and obtains the support component information through the support component detection model, so as to realize the automatic collection of construction information of the tunnel face area, reduce the monitoring cost and improve the level of intelligent monitoring.
[0036] Specifically, the construction process includes: blasting, slag removal, face clearing, steel frame transportation, steel frame installation, anchor bolt construction, spray-mixing construction, and lower step excavation. Through the process recognition model, the construction process displayed in the construction image of the tunnel face can be automatically identified without manual recording and recognition, thus having the effect of intelligent monitoring of the construction process.
[0037] In some embodiments, in combination Figure 2 The process identification network model includes: a backbone module 1, a detection module 2 and a first classification module 3.
[0038] The input layer 101 of the backbone module 1 is used to input the tunnel face construction image. The backbone module 1 is used to extract the features of the tunnel face construction image, obtain a feature map, and input the feature map to the detection module 2 and the first classification module 3 respectively.
[0039] The detection module 2 receives the feature map, and detects the construction target according to the feature map to obtain a construction target map.
[0040] The first classification module 3 receives the feature map, classifies the construction processes according to the feature map, and outputs the data of each construction process.
[0041] Combination Figure 2 The backbone module 1 includes an input layer 101, a first convolutional layer 102, a second convolutional layer 103, a first c2f layer 104, a third convolutional layer 105, a second c2f layer 106, a fourth convolutional layer 107, a third c2f layer 108, a fifth convolutional layer 109, a fourth c2f layer 110 and a pyramid pooling layer 111, which are connected in sequence. Among them, the first convolution layer 102, the second convolution layer 103, the third convolution layer 105, the fourth convolution layer 107 and the fifth convolution layer 109 are all used to extract the features of the tunnel face construction image, and generate feature maps of different sizes respectively; the first c2f layer 104, the second c2f layer 106, the third c2f layer 108 and the fourth c2f layer 110 are all used to receive feature maps, and fuse feature maps of different scales to improve the accuracy and robustness of target detection, and generate intermediate feature maps of different sizes respectively; the pyramid pooling layer 111 receives the intermediate feature map output by the fourth c2f layer 110, and performs a pooling operation on the intermediate feature map output by the fourth c2f layer 110 to improve the detection capability of targets of different sizes, and finally outputs the pooled feature map to the detection module 2 and the first classification module 3.
[0042] The first classification module 3 includes a Neck layer 31, a Head layer 32 and a frame loss layer 33 connected in sequence, and the output end of the Head layer 32 is also connected to the input end of the first classification loss layer 34; the detection module 2 includes a Conv layer 21, a first Fc layer 22, a second Fc layer 23 and a second classification loss layer 24 connected in sequence.
[0043] Furthermore, the intermediate feature maps output by the second c2f layer 106 and the third c2f layer 108 are respectively input to the Neck layer 31 of the first classification module 3; the Neck layer 31 of the first classification module 3 receives the intermediate feature maps output by the second c2f layer 106 and the third c2f layer 108, and the pooled feature maps output by the pyramid pooling layer 111, and fuses the feature maps of different sizes to enhance the feature representation capability, and then outputs the first fused map to the Head layer 32; the Head layer 32 is used to detect and identify the construction process in the first fused map, and output the target detection map; the box loss layer 33 is used to calculate the difference between the predicted bounding box and the true bounding box in the target detection map, and output the bounding box of the construction process; the first classification loss layer 34 is used to measure the difference between the predicted category distribution and the true label in the target detection map, and output the category of the construction process, and the construction process data includes the category and bounding box of the construction process.
[0044] Specifically, in some embodiments, the construction target may include construction vehicles, construction workers, and whether the construction workers are wearing helmets. For the safety of the construction workers, the construction workers must wear helmets when entering the tunnel. However, since the construction workers may take off their helmets during the construction process, the construction workers and helmets may be paired to identify whether the construction workers are wearing helmets effectively.
[0045] In some embodiments, the number of pixels in the pixel set corresponding to the construction personnel is identified. , identify the number of pixels in the pixel set corresponding to the hard hat adjacent to the construction worker , then the helmet wearing index Calculated by the following formula: ; in, The number of pixels that overlap between the pixel set corresponding to the construction worker and the pixel set corresponding to the helmet. It represents the total number of pixels after the union of the pixel set corresponding to the construction worker and the pixel set corresponding to the helmet. represents the first threshold, Greater than 0.5.
[0046] If the helmet wearing index If it is greater than or equal to 0, it means that the construction workers are wearing safety helmets effectively.
[0047] Furthermore, when the helmet wearing index When it is not negative (i.e. greater than or equal to zero), it is recognized that the construction worker is wearing a helmet effectively; otherwise, it is submitted to the data platform for manual review. It should be noted that when the process recognition network model is used to identify construction workers, the calculation is always based on the helmet closest to the construction worker. Generally, the Euclidean distance is used to calculate the distance between the construction worker and the helmet.
[0048] In some embodiments, the total loss function of the process identification network model is for: ; in, represents the loss function of construction target detection, represents the construction target detection box loss function, represents the construction process classification loss function.
[0049] in, , Represents the loss function of the first classification module 3 in the process identification network model.
[0050] In some embodiments, the loss function of the first classification module 3 in the process identification network model is , the calculation method is: ; in, represents the total sample size, Indicates the construction process stage of the face. Indicates Is the true label of the sample kind, Represents the corresponding predicted category probability.
[0051] In some embodiments, it also includes: collecting point cloud data of the rock surface in the tunnel face construction site before and after the spraying of concrete; preprocessing the point cloud data to obtain grid point cloud data; and obtaining flatness information based on the grid point cloud data.
[0052] In some embodiments, binocular camera grid structured light can be used to collect sparse grid point clouds of rock surfaces or sprayed concrete surfaces in the tunnel face construction site. "Point cloud" refers to a data set of all measurement points about the measurement target in a certain coordinate system, including the three-dimensional coordinates of each measurement point. Binocular camera grid structured light includes two cameras and a projector. This measurement method is an improvement and extension of traditional binocular vision measurement technology. It has both a theoretical basis for binocular vision measurement, which can save the complex calibration of the projector, and structured light to increase the geometric feature information of the object surface. It can avoid the problem of difficulty in matching weak texture or repeated texture areas in binocular stereo vision, thereby having higher measurement accuracy than traditional binocular vision measurement methods.
[0053] In some embodiments, the point cloud data is preprocessed, and the preprocessing may include operations such as denoising, sampling, and reconstruction. Preprocessing point cloud data is an important part of processing point cloud data, and its purpose is to improve the quality and processing efficiency of point cloud data. By denoising (removing outliers), sampling (reducing point cloud density), and enhancing features, noise can be eliminated, the amount of calculation can be reduced, and the accuracy and robustness of the algorithm can be improved, thereby providing a higher quality data foundation for subsequent point cloud data processing and analysis steps (such as reconstruction).
[0054] In some embodiments, after the step of preprocessing the point cloud data to obtain the grid point cloud data, the method further includes: interpolating the grid points using an inverse distance weighted interpolation method to increase the grid point cloud density in the grid point cloud data, the method being: ; in, represents the estimated value of the interpolation point, Represents known grid points The value of represents the inverse of the distance between the estimated grid point and the known grid point, Indicates the number of grid points.
[0055] Specifically, Inverse Distance Weighting (IDW) is a common spatial interpolation method that can use the values of known data points to estimate the values of unknown points.
[0056] In some embodiments, flatness information is obtained based on grid point cloud data, and the method includes: unfolding the grid point cloud data along the center line of the tunnel to obtain an orthogonal image map; slidingly calculating the height standard deviation of all pixel points in each window based on the orthogonal image map, and the height standard deviation is used to measure the flatness information.
[0057] In some embodiments, the height standard deviation The calculation method is: ; in, Indicates The height of the samples, represents the mean height within the window, Indicates the number of pixels in the window.
[0058] In some embodiments, concrete thickness information is obtained based on grid point cloud data, and the method is as follows: the grid point cloud data before and after shotcrete spraying are expanded along the center line of the tunnel to obtain an orthogonal image; the height mean of all pixel points in each window is calculated by sliding the orthogonal image, and the difference in the height mean is compared, specifically the difference between the three-dimensional left side after shotcrete spraying and the three-dimensional coordinates before shotcrete spraying, which is used as the point cloud thickness in the window to measure the flatness information.
[0059] In some embodiments, the height average is calculated as follows: .
[0060] In some embodiments, in combination Figure 3 The support component detection model includes a segmentation module 4, a hole position analysis module 5 and a second classification module 6 which are connected in sequence.
[0061] The segmentation module 4 receives the construction process data corresponding to the construction of the support component, extracts the features of the construction process data corresponding to the construction of the support component, and outputs the detection segmentation map to the hole position analysis module 5.
[0062] The hole position analysis module 5 receives the detection segmentation map, and detects the position of the support component according to the detection segmentation map to obtain the point position information of the support component.
[0063] The second classification module 6 classifies and identifies the point information to obtain the support component information.
[0064] Specifically, the "support member" referred to in the present application may refer to an anchor rod, which is a rod structure used for rock surface reinforcement.
[0065] In some embodiments, the total loss function of the support member detection model is for: ; in, represents the loss function of support component detection, represents the support component detection box loss function, represents the segmentation loss function.
[0066] Specifically, when the process identification network model detects that the construction process is support component construction, the corresponding construction process data is input into the support component detection model to obtain support component information. Exemplarily, the segmentation module 4 can be a network model based on the Yolov8 network, which is a prior art and will not be described in detail.
[0067] Specifically, the construction process data corresponding to the support component construction is first input into the segmentation module 4 of the support component detection model. The segmentation module 4 detects and segments the construction process data and outputs a detection segmentation map. Figure 4 A schematic diagram of the detection segmentation map. Figure 4 The detection segmentation frame 701 and the detection segmentation frame 702 are included. When the segmentation module 4 detects and segments, the identified support member (such as an anchor rod) is separated by a detection segmentation frame (such as Figure 4 The first detection segmentation box 701 and the second detection segmentation box 702 in the image are marked.
[0068] When anchor bolt construction is detected, it is necessary to further determine whether the current robot arm is drilling anchor bolt holes or installing anchor bolts. The determination method is as follows: Vector Indicates the first Frame image The location of the anchor bolt, where is the vertex coordinate of the anchor rod, is the pixel length of the anchor. By detecting the position changes of the anchors in consecutive frames, a detection result queue is formed, that is, If the following two conditions are met, it means that the current anchor is in the process of drilling the anchor hole or installing the anchor. Figure 5 Explanation ( Figure 5 for Figure 4 : (1) The vertex of the anchor rod (vertex coordinates are ), on the elliptical fitting contour line 731 of the tunnel face, or the distance from the contour line 731 is within the range of the second threshold T, where the second threshold T represents the distance between the vertex of the anchor and the contour line 731, and the value of the second threshold T is one quarter of the diameter of the tunnel face. Specifically, when the vertex of the anchor (the vertex coordinate is ), on the tunnel face ellipse fitting contour line 731, or the distance from the contour line 731 is within one quarter of the diameter of the tunnel face, indicating that the current anchor is in the anchor hole drilling position or the anchor is installed.
[0069] (2) As the robot arm moves during anchor drilling, installation, or withdrawal, the pixel length of the anchor It shows a trend of changing from long to short, or from short to long; If the above two conditions are not met, it means that the robot arm anchor is in a static or adjusted position; if the above two conditions are met, the corresponding anchor construction vector will be recorded, and the corresponding image sequence will form a construction video summary and construction record.
[0070] Specifically, Figure 5 For example, the vertex coordinates of the vertex 703 of the anchor located at the first detection segmentation frame 701 are , the pixel length of the anchor is ; The vertex coordinates of the vertex 704 of the anchor located at the second detection segmentation frame 702 are , the pixel length of the anchor is .
[0071] Furthermore, by inputting the above-mentioned detection segmentation map into the hole position analysis module in the support component detection model, the point information of the support component, that is, the hole position coordinates, can be obtained; and the hole position coordinates are input into the second classification module to obtain the support component information, which includes the number, position and other information of the anchor construction during the construction process.
[0072] Further, such as Figure 6 As shown in Figure 1, a schematic diagram of the construction position of the anchor bolt construction phase can be drawn based on the support component information, and the corresponding image sequence can be used to form a construction video summary and construction record. Figure 6 As shown, the outer side of the tunnel contour line 731 is the first hole position 11 to the sixth hole position 16, and the inner side of the tunnel contour line 731 is the anchor rod 732 installed corresponding to each hole position. It should be noted that Figure 6 The diagram shows the relative position relationship of the anchor rod 732 when operating the first hole position 11 to the sixth hole position 16. Figure 6The hole position information is displayed in the figure, including the pixel coordinates corresponding to each hole position and the inclination angle of the anchor rod 732 installed at the hole position. When measuring the inclination angle of the anchor rod 732, the horizontal line is used as the reference direction to the left. When the anchor rod 732 rotates to the right along the reference direction, the angle increases continuously. For example: the first hole position 11, the pixel coordinates are (432, 311), and the inclination angle is 37°; the second hole position 12, the pixel coordinates are (494, 247), and the inclination angle is 44°; the third hole position 13, the pixel coordinates are (1353, 266), and the inclination angle is 160°; the fourth hole position 14, the pixel coordinates are (1436, 330), and the inclination angle is 173°; the fifth hole position 15, the pixel coordinates are (1533, 484), and the inclination angle is 178°; the sixth hole position 16, the pixel coordinates are (1580, 598), and the inclination angle is 186°.
[0073] Furthermore, by identifying and detecting the above-mentioned construction procedures and construction targets, calculating the flatness information and concrete thickness information, and obtaining the support component information and other construction information, the safety of tunnel face construction can be monitored. These construction information can also help technicians grasp the progress of tunnel construction in real time and standardize the rationality of tunnel construction. In addition, by automatically calculating the concrete volume, the amount of concrete is reduced, and by identifying the support construction process, the construction process is reasonably optimized and the construction efficiency is improved, which is conducive to reducing the overall energy consumption of tunnel face excavation and has the advantages of energy saving and emission reduction.
[0074] Combination Figure 7 The present application also provides an electronic device, including a processor 72 and a memory 71, wherein the memory 71 stores a computer program that can be executed by the processor, and when the computer program is executed by the processor 72, any one of the above-mentioned methods for automatically monitoring tunnel face construction information is implemented.
[0075] Specifically, the memory 71 is used to store computer programs or data. The memory 71 can be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc.
[0076] The processor 72 is used to read / write data or computer programs stored in the memory 71 and execute corresponding functions. For example, when the computer program stored in the memory 71 is executed by the processor 72, a method for automatically monitoring tunnel face construction information provided in an embodiment of the present application can be implemented.
[0077] Combination Figure 8 The present application also provides a computer-readable storage medium on which a computer program 801 is stored. The computer program 801 is used to implement any of the above-mentioned methods for automatically monitoring tunnel face construction information.
[0078] In some embodiments, the functions or modules included in the device provided by the embodiments of the present disclosure can be used to execute the method described in the above method embodiments. The specific implementation can refer to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.
[0079] The above description of various embodiments tends to emphasize the differences between the various embodiments. The same or similar aspects can be referenced to each other, and for the sake of brevity, they will not be repeated herein.
[0080] In the several embodiments provided in the present application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device implementation described above is only schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation, such as units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.
[0081] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0082] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of each implementation method of the present application. The aforementioned storage medium can be non-transient or transient. Specifically, it includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), disk or optical disk and other media that can store program code.
[0083] In summary, the present invention discloses a method, device and storage medium for automatic monitoring of tunnel face construction information, the method comprising the steps of: collecting tunnel face construction images; inputting tunnel face construction images into a process recognition network model, using the process recognition network model to identify and monitor construction processes, and obtaining data of each construction process; if the construction process is monitored to be support component construction, the corresponding construction process data is input into a support component detection model to obtain support component construction information. The method of the present application can realize the automatic collection of construction information in the tunnel face area, reduce monitoring costs, and improve the level of intelligent monitoring.
[0084] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structural transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A method for automatically monitoring tunnel face construction information, characterized in that: Includes steps: Collect tunnel face construction images; Inputting the tunnel face construction image into a process recognition network model, using the process recognition network model to identify and monitor the construction process, and obtaining data of each construction process; If it is monitored that the construction process is the support component construction, the corresponding construction process data is input into the support component detection model to obtain the support component construction information.
2. The method for automatically monitoring tunnel face construction information according to claim 1, characterized in that: Also includes: Collect point cloud data of rock surface before and after shotcrete spraying at the tunnel face construction site; Preprocessing the point cloud data to obtain grid point cloud data; The flatness information is obtained according to the grid point cloud data.
3. The method for automatic monitoring of tunnel face construction information according to claim 2, characterized in that: After the step of preprocessing the point cloud data to obtain grid point cloud data, the method further includes: The inverse distance weighted interpolation method is used to interpolate the grid points. The method is: ; in, represents the estimated value of the interpolation point, Represents known grid points The value of represents the inverse of the distance between the estimated grid point and the known grid point, Indicates the number of grid points.
4. The method for automatically monitoring tunnel face construction information according to claim 3, characterized in that: The method for obtaining the flatness information according to the grid point cloud data comprises: Expand the grid point cloud data along the tunnel centerline to obtain an orthogonal image; The height standard deviation of all pixels in each window is calculated based on the sliding of the orthogonal image, and the height standard deviation is used to measure the flatness information.
5. The method for automatic monitoring of tunnel face construction information according to claim 1, characterized in that: The process identification network model includes: a backbone module, a detection module and a first classification module; The input layer of the backbone module is used to input the tunnel face construction image, and the backbone module is used to extract features of the tunnel face construction image to obtain a feature map, and input the feature map to the detection module and the first classification module respectively; The detection module receives the feature map, and detects the construction target according to the feature map to obtain a construction target map; The first classification module receives the feature map, classifies the construction processes according to the feature map, and outputs data of each construction process.
6. The method for automatic monitoring of tunnel face construction information according to claim 1, characterized in that: The support member detection model includes a segmentation module, a hole position analysis module and a second classification module connected in sequence; The input layer of the segmentation module is used to receive the construction process data corresponding to the construction of the support member, extract the features of the construction process data corresponding to the construction of the support member, and output the detection segmentation map to the hole position analysis module; The hole position analysis module receives the detection segmentation map, and detects the position of the support component according to the detection segmentation map to obtain the point position information of the support component; The second classification module classifies and identifies the point information to obtain the supporting component information.
7. The method for automatically monitoring tunnel face construction information according to claim 1, characterized in that: The process identifies the total loss function of the network model for: ; in, represents the loss function of construction target detection, represents the construction target detection box loss function, represents the construction process classification loss function; The total loss function of the support component detection model for: ; in, represents the loss function of support component detection, represents the support component detection box loss function, represents the segmentation loss function.
8. The method for automatic monitoring of tunnel face construction information according to claim 1, characterized in that: Also includes: Monitor whether construction workers are wearing helmets effectively by: Identify the number of pixels in the pixel set corresponding to the construction personnel , identify the number of pixels in the pixel set corresponding to the hard hat adjacent to the construction worker , then the helmet wearing index Calculated by the following formula: ; in, The number of pixels that overlap between the pixel set corresponding to the construction worker and the pixel set corresponding to the helmet. It represents the total number of pixels after the union of the pixel set corresponding to the construction worker and the pixel set corresponding to the helmet. represents the first threshold, Greater than 0.5; If the helmet wearing index If it is greater than or equal to 0, it means that the construction workers are wearing safety helmets effectively.
9. An electronic device, characterized in that: It comprises a processor and a memory, wherein the memory stores a computer program that can be executed by the processor, and when the computer program is executed by the processor, the method for automatically monitoring the construction information of a tunnel face according to any one of claims 1 to 8 is implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The computer program is used to implement the method for automatically monitoring tunnel face construction information as described in any one of claims 1 to 8.
Citation Information
Patent Citations
A method for detecting personnel and a safety helmet in a monitoring video
CN109948501A
Safety protector wearing detection method and system
CN113792584A
Method and system for automatically collecting multivariate information of tunnel in construction period
CN116858098A
Tunnel face crack feature extraction method, device, equipment and medium
CN118196432A
Safety early warning method and device for full-section tunneling of tunnel featuring dynamic water and weak surrounding rock
US20220112806A1
Cited By
Construction process management method and system based on AI
CN119358981A
A construction process management method and system based on AI
CN119358981B
Tunnel construction hidden project intelligent detection equipment and method based on machine vision
CN119832200A
Tunnel construction efficacy management method and system based on generative visual large model
CN121684715A
