Tunnel Excavation Management System

Abstract

To provide a tunnel excavation management system that can accurately predict dangerous locations in tunnel excavation work without depending on a constructor.SOLUTION: A tunnel excavation management system includes a 3D scanner (S1) that acquires 3D tunnel shape data, a thermography camera (S2) that acquires temperature distribution image data of a face, a sound level meter (S3) that acquires noise data in the tunnel, a spring water meter (S4) that acquires spring water amount data in the tunnel, and an output unit (110) that displays dangerous areas and issues an alarm. A processor analyzes the hardness and softness distribution of the face (105), extracts the face shape and tunnel side shape (103, 104), detects the deformed state of the tunnel (106), and analyzes the noise data (107). An AI analysis unit (108) uses machine learning models (M1, M2) to predict dangerous locations where tunnel deformation may occur.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to an excavation management system for tunnel excavation work for roads, railways, waterways, etc. [Background technology]

[0002] In tunnel excavation work, detecting deformations such as face displacement and collapse (collapse) is extremely important for safely proceeding with tunnel construction. From this perspective, Patent Document 1 discloses a system that uses a digital camera to photograph the contour line irradiated onto the face with a line laser, and detects changes in the contour line as deformations on the face from the photographed data. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2018-159640 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the system disclosed in the above-mentioned Patent Document 1 only detects deformation of the face at that time by irradiating a special contour line with a laser and photographing it. Therefore, the only way to predict areas at risk of collapse was for an experienced builder to judge by looking at changes in the photographed contour line. As a result, the prediction results of dangerous areas depend on the builder's experience, etc., and accurate predictions that do not depend on humans are not possible. In addition, automation is difficult and variations depending on the builder cannot be avoided.

[0005] The present invention has been devised in view of the above circumstances, and an object of the present invention is to provide a tunnel excavation management system that can accurately predict dangerous areas during tunnel excavation work without relying on the contractor. [Means for solving the problem]

[0006] In order to achieve the above object, according to one embodiment of the present invention, there is provided a tunnel excavation management system comprising: a three-dimensional scanner for acquiring three-dimensional tunnel shape data; a thermographic camera for acquiring temperature distribution image data of a face; a sound level meter for acquiring noise data within the tunnel; a spring water meter for acquiring data on the amount of spring water within the tunnel; at least one processor for inputting the three-dimensional tunnel shape data, the temperature distribution image data, the noise data, and the spring water amount data and predicting dangerous locations when a deformation occurs; and an output unit for displaying dangerous locations and issuing an alarm. The processor analyzes the hardness distribution of the tunnel face from the thermal distribution image data, extracts the tunnel face shape and the tunnel side shape from the three-dimensional tunnel shape data to detect the deformation state of the tunnel, analyzes the noise data to determine the presence or absence of sound components that indicate the possibility of face collapse, and predicts dangerous areas of the tunnel where deformation may occur using a machine learning model constructed based on the hardness distribution of the tunnel face, the location and amount of spring water estimated based on the hardness distribution of the tunnel face, the deformation state of the tunnel, and the analysis results of the sound components that indicate the possibility of face collapse. According to one embodiment of the present invention, the processor can identify areas of the temperature distribution image data of the working face that are displayed with a density above a predetermined standard level as spring water locations, and estimate the amount of water based on the density of the area. Furthermore, according to one embodiment of the present invention, the system further includes a memory unit that stores, in chronological order, face shape and tunnel side shape data obtained from the three-dimensional tunnel shape data, face hardness distribution data, data on the amount of spring water, and data on the estimated location and amount of spring water and predicted areas at risk of deformation, and each of the time-series data stored in the memory unit can be used for machine learning of the machine learning model. Furthermore, according to one embodiment of the present invention, the output unit can display dangerous areas of the tunnel due to deformation in accordance with instructions from the processor, and issue an alarm according to the danger level. [Effects of the Invention]

[0007] According to one embodiment of the present invention, it is possible to predict areas at risk of deformation using face hardness distribution data, estimated locations and amounts of spring water, deformation state data, and noise analysis results. For example, if a large amount of water is gushing out at an area where deformation has occurred, it is possible to determine that there is a high risk of collapse. Furthermore, if sound components that indicate a precursor to collapse are detected in the noise analysis, it is possible to determine that there is an extremely high risk of collapse at that spring location. Therefore, by notifying dangerous areas early and issuing an alarm, it is possible to significantly improve the safety of construction work. Furthermore, according to one embodiment of the present invention, it is possible to easily identify the hardness / softness distribution and the location of spring water and estimate the amount of spring water from the temperature distribution image data of the face. Furthermore, according to one embodiment of the present invention, by accumulating face shape and tunnel side shape data, face hardness distribution data, spring water volume data, estimated spring water locations and volumes, and predicted data on dangerous areas of deformation in a chronological order, these can be used for machine learning of a machine learning model, enabling rapid and accurate prediction of dangerous areas. Furthermore, according to one embodiment of the present invention, the output unit displays dangerous areas where deformation of the tunnel may occur and issues an alarm according to the danger level, thereby providing early notification of dangerous areas and issuing an alarm, thereby significantly improving the safety of construction work. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a cross-sectional view of a tunnel for explaining a tunnel excavation management system according to one embodiment of the present invention. FIG. [Figure 2] FIG. 2 is a block diagram showing an example of the functional configuration of a tunnel excavation management system according to the present embodiment. [Figure 3] 1 is a schematic diagram for explaining the function of predicting the location and amount of spring water based on a temperature distribution image of the tunnel excavation management system according to this embodiment. FIG. [Figure 4] 1 is a flowchart illustrating an example of a tunnel excavation management method according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, a tunnel excavation management system according to one embodiment of the present invention will be described in detail, taking a tunnel excavation work as shown in FIG. 1 as an example.

[0010] As shown in Figure 1, a 3D scanner S1, a thermographic camera S2, a sound level meter S3, and a spring water meter S4 are installed inside the tunnel 10. The 3D scanner S1, thermographic camera S2, and sound level meter S3 are mounted on a movable cart. The 3D scanner S1 can capture the three-dimensional shape of the area near the tunnel face 12. The thermographic camera S2 is an infrared camera that uses infrared light to capture an image of the temperature distribution at the tunnel face 12. The sound level meter S3 detects noise inside the tunnel 10, particularly sound from the tunnel face 12. As will be described later, analyzing the noise can determine the possibility of collapse in front of the tunnel face and on the side of the tunnel. The spring water meter S4 is installed in the drainage equipment 21, which discharges water that springs from the tunnel inner wall 11 and the tunnel face 12 outside the tunnel, and measures the amount of spring water being discharged.

[0011] The tunnel excavation management system 100 according to this embodiment can accurately predict dangerous areas during tunnel excavation work using data acquired by the above-mentioned 3D scanner S1, thermographic camera S2, sound level meter S3 and spring water meter S4.

[0012] The configuration and operation of the tunnel excavation management system according to this embodiment will be described below with reference to Figures 2 to 4. The tunnel excavation management system according to this embodiment can be constructed on a general-purpose server or personal computer having one or more processors such as a CPU (Central Processing Unit) or an MPU (Micro Processor Unit).

[0013] 1. System Configuration As shown in Figure 2, a tunnel excavation management system 100 according to one embodiment of the present invention has an input unit 101 for inputting various data. The input unit 101 is connected by wire or wirelessly to a 3D scanner S1, a thermographic camera S2, a sound level meter S3, and a spring water volume meter S4 installed inside the tunnel, and inputs 3D scan data, temperature distribution image data, noise data, and spring water volume data, respectively. The input unit 101 is also connected to input devices such as a keyboard and pointing device of the system, and various setting data can be input.

[0014] The tunnel excavation management system 100 has a control unit 102 that controls the operation of the entire system, and the control unit 102 is connected to an input unit 101, a face shape extraction unit 103, a tunnel side surface shape extraction unit 104, a face hardness / softness distribution analysis unit 105, a deformation state detection unit 106, a noise analysis unit 107, an AI (Artificial Intelligence) analysis unit 108, a data accumulation unit 109, and an output unit 110. The face shape extraction unit 103 and the tunnel side surface shape extraction unit 104 extract the face shape and the tunnel side surface shape, respectively, from the 3D scan data acquired by the 3D scanner S1, and store them in chronological order in the tunnel shape data accumulation unit 201.

[0015] The face hardness distribution analysis unit 105 analyzes the temperature distribution image data of the face 12 input from the thermographic camera S2, classifies the temperature region according to predetermined temperature level standards, and determines that parts with relatively low temperature levels are soft rock. The face hardness distribution data analyzed in this way is stored in chronological order in the hardness distribution data storage unit 202. As will be described later, it is also possible to estimate the locations of spring water and the amount of water at the face 12 from the temperature distribution image data. The face hardness distribution analysis unit 105 can also be configured using a machine learning model.

[0016] The deformation state detection unit 106 detects the collapsed state (deformed state) of the tunnel using the current face shape and tunnel side face shape data acquired from the face shape extraction unit 103 and the tunnel side face shape extraction unit 104, and the face shape and tunnel side face shape data accumulated in the tunnel shape data accumulation unit 201. This deformation state detection unit 106 can also be configured using a machine learning model.

[0017] The noise analysis unit 107 receives noise data from the sound level meter S3 and detects sounds that indicate a precursor to a face collapse through audio analysis. The sounds that indicate a precursor to a collapse may be registered in advance or may be learned by a machine learning model.

[0018] The data storage unit 109 includes a tunnel shape data storage unit 201, a hardness distribution data storage unit 202, a water volume data storage unit 203, and a dangerous area estimation data storage unit 204. As described above, the tunnel shape data storage unit 201 stores the tunnel face shape and tunnel side face shape data acquired by the tunnel face shape extraction unit 103 and the tunnel side face shape extraction unit 104 in chronological order. As described above, the hardness distribution data storage unit 202 stores the hardness distribution data of the tunnel face in chronological order using the temperature distribution image data of the tunnel face 12 input from the thermographic camera S2. The water volume data storage unit 203 stores the water volume data acquired by the water spring meter S4 and the estimated water volume data described below in chronological order. The dangerous area estimation data storage unit 204 stores the data on the locations and volume of water springs estimated by the AI ​​analysis unit 108 described below and the predicted deformation dangerous areas in chronological order. In addition, these stored data can be used to build a machine learning model in the AI ​​analysis unit 108.

[0019] The AI ​​analysis unit 108 inputs the time-series tunnel shape data, time-series face hardness / softness distribution data, and time-series spring water volume data stored in the above-mentioned data storage unit 109, the deformation state data obtained from the deformation state detection unit 106, and the analysis results obtained from the noise analysis unit 107, and predicts spring water locations and deformation risk locations on the face of the tunnel using a spring water location estimation model M1 and a deformation risk location prediction model M2, which are machine learning models.

[0020] The spring water location estimation model M1 predicts the location and amount of spring water using time-series face hardness / softness distribution data and time-series spring water volume data as input data. The deformation risk location prediction model M2 inputs the spring water location and amount predicted by the spring water location estimation model M1, the deformation state data obtained from the deformation state detection unit 106, and the analysis results obtained from the noise analysis unit 107, and predicts deformation risk locations.

[0021] Machine learning models can be constructed using deep learning, for example. Deep learning technology can construct machine learning models with sufficiently small errors by providing appropriate training data and adjusting neuron parameters, even when a large amount of data is interacting with each other. The seepage location estimation model M1 and the deformation risk area prediction model M2 can each be constructed using predetermined training data, but they can also be constructed using past data, such as changes in tunnel shape over time, changes in the amount of seepage over time, changes in hardness and softness distribution over time, changes in the location and amount of seepage over time, and changes in deformation areas over time.

[0022] The output unit 110 is a display unit such as a computer monitor, which displays the tunnel shape, the hardness distribution of the face, the deformed areas, the locations and amounts of spring water predicted by the AI ​​analysis unit 108, and the predicted areas at risk of deformation, and issues alarms etc. according to the danger level of the predicted areas at risk of deformation.

[0023] The functions of the control unit 102, face shape extraction unit 103, tunnel side surface shape extraction unit 104, face hardness / softness distribution analysis unit 105, deformation state detection unit 106, noise analysis unit 107 and AI analysis unit 108 can be realized by executing a program stored in a memory (not shown) on a processor.

[0024] 2. Estimation of spring locations As shown in FIG. 3, the face hardness distribution analysis unit 105 analyzes the temperature distribution image IMG of the face 12 input from the thermographic camera S2 and divides it into temperature regions according to a predetermined density level 105a. The spring water location estimation model M1 estimates the spring water locations by referring to the temperature distribution image IMG and the temperature region divisions. Here, it is assumed that areas in the temperature distribution image IMG with a density higher than a predetermined reference level are identified as spring water regions 12a, 12b, and 12c. In this embodiment, as an example, it is assumed that the spring water from the face 12 has a lower temperature than the surrounding area, and areas with lower temperatures are displayed in darker blue. Conversely, if the spring water is hotter than the surrounding area, areas with higher temperatures are displayed in darker red. In either case, high-density areas in the temperature distribution image can be identified as spring water locations.

[0025] The face hardness distribution analysis unit 105 can also estimate the amount of spring water by calculating the area of ​​the spring water area with a concentration higher than a predetermined reference level and the maximum concentration value in that area. The amount of spring water is estimated to be larger the larger the area of ​​the spring water area and the higher the concentration. The location information of the spring water area estimated in this way and the estimated water volume are stored in the water volume data accumulation unit 203.

[0026] 3.Prediction of areas at risk of deformation The deformation risk location prediction model M2 predicts deformation risk locations by inputting the face hardness distribution data obtained by the face hardness distribution analysis unit 105, the location and amount of spring water predicted by the spring water location estimation model M1, deformation state data indicating the degree of collapse of the tunnel obtained from the deformation state detection unit 106, and the analysis results obtained from the noise analysis unit 107. In other words, if a large amount of water is gushing out at a location where the shape of the tunnel has changed, there is a high risk of collapse, and if sound components indicating a precursor to collapse are detected in the noise data, it can be determined that there is an extremely high risk of collapse at that spring water location.

[0027] 4.Operation Hereinafter, the operation of the tunnel excavation management system 100 in this embodiment will be described with reference to FIG.

[0028] As shown in Figure 4, the control unit 102 receives 3D scan data, temperature distribution image data, noise data, and spring water volume data from the 3D scanner S1, thermographic camera S2, sound level meter S3, and spring water volume meter S4, respectively (operation 301). The face hardness distribution analysis unit 105 analyzes the temperature distribution image (IMG) of the tunnel face 12 received from the thermographic camera S2 and divides it into temperature regions according to a predetermined density level 105a (operation 302). The deformation detection unit 106 detects the deformation of the tunnel using the face shape and tunnel side shape acquired from the 3D scan data by the face shape extraction unit 103 and the tunnel side shape extraction unit 104, respectively, and the time-series data acquired from the tunnel shape data storage unit 201 (operation 303). The noise analysis unit 107 analyzes the noise data from the sound level meter S3 and determines whether or not there are any sound components that indicate a collapse precursor that have been previously registered or learned (operation 304).

[0029] Next, the AI ​​analysis unit 108 estimates the location and amount of water springing from the temperature distribution image IMG and temperature region classification according to the water spring location estimation model M1, and predicts the deformation risk area according to the deformation risk area prediction model M2 by inputting the hardness / softness distribution data on the face, the location and amount of water springing predicted by the water spring location estimation model M1, the deformation state data indicating the degree of collapse of the tunnel obtained from the deformation state detection unit 106, and the analysis results obtained from the noise analysis unit 107 (operation 305).

[0030] The control unit 102 displays the locations and amounts of spring water predicted by the AI ​​analysis unit 108 and the areas at risk of deformation on the output unit 110, and issues an alarm inside the tunnel and on the surface according to the danger level (operation 306). For example, if the degree of collapse of the tunnel indicated by the deformation state data exceeds a predetermined level, the estimated amount of spring water in the soft rock area exceeds a predetermined value, and the sound inside the tunnel contains components that indicate the possibility of collapse, the control unit 102 issues an alarm from the output unit 110.

[0031] As described above, according to this embodiment, it is possible to predict dangerous areas during tunnel excavation work based on the occurrence of deformation, the location or increase in the amount of spring water, abnormal sounds inside the tunnel, etc., and issue an alarm. This makes it possible to accurately predict dangerous areas and issue an alarm without relying on the construction worker, thereby significantly improving safety. [Explanation of symbols]

[0032] 10. Tunnel 11 Tunnel inner wall 12 Cutting edge 21 Drainage equipment 22 Drill Jumbo 100 Tunnel Excavation Management System 101 Input section 102 Control section 103 Face shape extraction section 104 Tunnel side shape extraction section 105 Face hardness distribution analysis section 106 Deformation detection unit 107 Noise Analysis Section 108 AI analysis department 109 Data Storage Unit 110 Output section 201 Tunnel shape data storage unit 202 Hardness / Softness Distribution Data Storage Unit 203 Water volume data storage unit 204 Hazardous Area Estimation Data Storage Unit M1 Spring location estimation model M2 Deformation risk area prediction model S1 3D Scanner S2 Thermal Imaging Camera S3 Sound Level Meter S4 Spring water meter

Claims

1. A tunnel excavation management system, a three-dimensional scanner for acquiring three-dimensional tunnel shape data; a thermography camera that acquires temperature distribution image data of the tunnel face; A sound level meter that collects noise data inside the tunnel, A spring water meter that collects data on the amount of spring water inside the tunnel; At least one processor that inputs the three-dimensional tunnel shape data, the temperature distribution image data, the noise data, and the spring water volume data and predicts dangerous locations when a deformation occurs; an output unit that displays dangerous locations and issues alarms; wherein the processor: Analyzing the hardness distribution of the face from the temperature distribution image data; Extracting a face shape and a tunnel side shape from the three-dimensional tunnel shape data to detect a deformation state of the tunnel; Analyzing the noise data to determine whether or not there is a sound component indicating the possibility of a face collapse; Predicting dangerous locations of deformation of the tunnel using a machine learning model constructed based on the hardness distribution of the tunnel face, the location and amount of spring water estimated based on the hardness distribution of the tunnel face, the deformation state of the tunnel, and the analysis results of sound components indicating the possibility of the tunnel face collapsing. A tunnel excavation management system characterized by:

2. The tunnel excavation management system described in claim 1, characterized in that the processor identifies areas of the temperature distribution image data of the face that are displayed with a density above a predetermined standard level as soft rock areas and spring water areas, and estimates the amount of water based on the density in the areas.

3. Further, a memory unit is provided for storing, in chronological order, face shape and tunnel side shape data acquired from the three-dimensional tunnel shape data, face hardness distribution data, data on the amount of spring water, and data on the estimated location and amount of spring water and predicted deformation risk areas, The tunnel excavation management system according to claim 1, characterized in that each piece of time-series data stored in the memory unit is used for machine learning of the machine learning model.

4. A tunnel excavation management system as described in any one of claims 1 to 3, characterized in that the output unit displays dangerous areas of the tunnel due to deformation in accordance with instructions from the processor and issues an alarm depending on the danger level.

Images