CN115359447A - Highway tunnel remote monitering system - Google Patents
Highway tunnel remote monitering system Download PDFInfo
- Publication number
- CN115359447A CN115359447A CN202210991425.8A CN202210991425A CN115359447A CN 115359447 A CN115359447 A CN 115359447A CN 202210991425 A CN202210991425 A CN 202210991425A CN 115359447 A CN115359447 A CN 115359447A
- Authority
- CN
- China
- Prior art keywords
- information
- image
- tunnel
- image information
- machine learning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000012544 monitoring process Methods 0.000 claims abstract description 45
- 238000010801 machine learning Methods 0.000 claims abstract description 36
- 238000004088 simulation Methods 0.000 claims abstract description 31
- 239000011435 rock Substances 0.000 claims description 29
- 238000001514 detection method Methods 0.000 claims description 26
- 230000011218 segmentation Effects 0.000 claims description 17
- 238000012549 training Methods 0.000 claims description 16
- 238000007689 inspection Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 3
- 238000005315 distribution function Methods 0.000 claims 1
- 239000000203 mixture Substances 0.000 claims 1
- 239000002994 raw material Substances 0.000 claims 1
- 238000005259 measurement Methods 0.000 abstract description 12
- 238000012423 maintenance Methods 0.000 abstract description 6
- 230000005284 excitation Effects 0.000 abstract description 3
- 238000010276 construction Methods 0.000 description 12
- 238000000034 method Methods 0.000 description 11
- 230000006870 function Effects 0.000 description 9
- 238000009412 basement excavation Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 230000005540 biological transmission Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 5
- 239000011159 matrix material Substances 0.000 description 5
- 239000000523 sample Substances 0.000 description 5
- 229910000831 Steel Inorganic materials 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 4
- 210000002569 neuron Anatomy 0.000 description 4
- 238000004062 sedimentation Methods 0.000 description 4
- 239000010959 steel Substances 0.000 description 4
- 229910001294 Reinforcing steel Inorganic materials 0.000 description 3
- 230000004913 activation Effects 0.000 description 3
- 238000009826 distribution Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 208000037656 Respiratory Sounds Diseases 0.000 description 2
- 239000004567 concrete Substances 0.000 description 2
- 238000005520 cutting process Methods 0.000 description 2
- 238000013499 data model Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 239000011378 shotcrete Substances 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000009933 burial Methods 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 229910052738 indium Inorganic materials 0.000 description 1
- APFVFJFRJDLVQX-UHFFFAOYSA-N indium atom Chemical compound [In] APFVFJFRJDLVQX-UHFFFAOYSA-N 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 239000011150 reinforced concrete Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 238000003466 welding Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21F—SAFETY DEVICES, TRANSPORT, FILLING-UP, RESCUE, VENTILATION, OR DRAINING IN OR OF MINES OR TUNNELS
- E21F17/00—Methods or devices for use in mines or tunnels, not covered elsewhere
- E21F17/18—Special adaptations of signalling or alarm devices
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30204—Marker
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mining & Mineral Resources (AREA)
- Multimedia (AREA)
- Geology (AREA)
- Medical Informatics (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Geochemistry & Mineralogy (AREA)
- Quality & Reliability (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
- Image Analysis (AREA)
Abstract
The remote monitoring system for the road tunnel comprises a central workstation; a network connected to the central workstation in a wireless or wired manner; the system comprises a vehicle and an image acquisition device arranged on the vehicle; the position mark and/or the position sensor are/is arranged on the surface of the highway tunnel or nearby the highway tunnel, and the position sensor sends position information or measured position information to the network through the transmitting terminal; the central workstation can receive image information sent by the image acquisition device through a network; the central workstation can receive location information over a network. The vehicle-mounted image acquisition device acquires image information in time, screens the image information to obtain image information with a broken contour line, indexes the image information and sends out the image information, and the central workstation performs machine learning of big data and performs excitation of a learning model by combining with measurement simulation data; when operation and maintenance monitoring is carried out, identification of high-risk cracks is carried out through index image information obtained by polling of an image acquisition device of a vehicle.
Description
Technical Field
The invention relates to the technical field of highway tunnel monitoring, in particular to a highway tunnel remote monitoring system.
Background
With the increasing difficulty of tunnel construction, it is particularly important to detect the stability of the road tunnel structure. In the traditional highway tunnel stability detection, visual, broken detection or manual nondestructive detection methods are often adopted to detect the tunnel, and the detection of the methods is unstable and low in detection efficiency, so that how to quickly and accurately detect the tunnel stability is a research hotspot. In the prior art, for example, chinese patent application with publication number CN114296079A, a system and a method for detecting stability of a road tunnel structure based on radar include a radar detection module composed of a plurality of radars and a data analysis unit; the method comprises the steps that a plurality of radars are arranged in a tunnel, distance data of target detection points are collected regularly or in real time through the radars, then a target detection sequence generated according to the distance data is compared with a standard detection sequence to obtain a stability analysis result, and detection precision and detection efficiency can be improved; the radar detection module in the invention can integrate and associate to generate a radar detection system; the radar detection system can accurately detect a plurality of highway tunnels in a monitoring area, and detection results can be fed back to workers in time, so that the detection range is expanded, and the detection efficiency of the highway tunnels in a large-range area is improved. However, the prior art has the problems that various sensors need to be arranged in a large range, and the construction and operation and maintenance costs are high.
Disclosure of Invention
The invention aims to provide a road tunnel remote monitoring system which can continuously and intelligently monitor potential crack risks, has small transmission information redundancy, realizes image recognition road tunnel detection instead of a sensor, and greatly reduces construction and operation and maintenance costs.
A remote monitoring system for a road tunnel comprises a road tunnel remote monitoring system,
a central workstation;
a network connected to the central workstation in a wireless or wired manner;
the system comprises a vehicle and an image acquisition device arranged on the vehicle;
the position mark and/or the position sensor are/is arranged on the surface or the vicinity of the highway tunnel, and the position sensor sends position information or measured position information to the network through the transmitting terminal;
the central workstation can receive the image information sent by the image acquisition device through a network; the central workstation can receive location information over a network.
The vehicle-mounted image acquisition device acquires image information in time, screens the image information to obtain image information with a broken contour line, indexes the image information and sends out the image information, and the central workstation performs machine learning of big data and performs excitation of a learning model by combining with measurement simulation data; when operation and maintenance monitoring is carried out, identification of high-risk cracks is carried out through index image information obtained by polling of an image acquisition device of a vehicle.
In order to further optimize the technical scheme, the adopted measures further comprise:
the central workstation performs machine learning through image information training;
after numerical simulation is carried out on the position information, generating maximum offset distance information or surrounding rock deformation cloud picture information or deformation settlement numerical information, and using the maximum offset distance information or the surrounding rock deformation cloud picture information or the deformation settlement numerical information as reward information for machine learning to form a machine learning model;
the machine learning model is used for risk detection of image information. And calculating the maximum offset distance information or the surrounding rock deformation cloud picture information or the deformation settlement numerical information, acquiring basic data through a transmitting terminal by adopting daily measurement of a manual embedded position mark and a real-time position sensor, and performing numerical simulation by inputting a local mountain data model of the bridge tunnel.
The image acquisition device on the vehicle acquires image information and then adopts the following steps:
1) An information capturing step: reading image information obtained by a road tunnel inspection or fixed camera;
2) Information indexing step: extracting a fracture contour line in the image information, and storing the fracture contour line as a segmentation image; indexing the segmentation image with the fracture contour line; and then, transmitting the indexed segmentation image. The data transmission pressure can be reduced by screening and indexing the broken contour line of the image information to be inspected. The fixed camera is fixedly arranged under the artificial and on-demand conditions, and the cracks with greater risks are monitored uninterruptedly.
The central workstation carries out remote monitoring on the image information by adopting the following steps:
3) A machine learning step: converting the received indexed segmentation image into tensor as a training set, and taking numerical simulation result information of section deformation corresponding to the indexed segmentation image as reward information; training the training set by using a machine learning model, wherein the reward information is used for adjusting the machine learning model;
taking numerical simulation result information of the section deformation corresponding to the indexed segmented image as reward information, and taking tensor of the indexed segmented image as a training set; until the preset accuracy is met;
4) A detection step: and (3) processing and segmenting the inspection photos of the tunnel to be detected according to the step 2), and then processing and identifying each image by using the step 3). Through machine learning, through the fracture contour line of effective discernment crackle and under the mode that section deformation information regarded as reward, make the fracture contour line of discernment have certain tunnel road bridge mechanics numerical simulation relevance to reach the technical purpose that carries out tunnel road bridge deformation monitoring through machine discernment contour line.
Because the invention adopts the position mark and/or the position sensor which are arranged on the surface of the highway tunnel or nearby, the position sensor sends the position information or the measured position information to the network through the transmitting terminal; the central workstation can receive the image information sent by the image acquisition device through a network; the central workstation can receive the position information through the network, thereby achieving the advantages of continuously and intelligently monitoring potential crack risks, having small transmission information redundancy, realizing the detection of image recognition highway tunnels instead of sensors, and greatly reducing the construction and operation and maintenance cost.
Drawings
FIG. 1 is a schematic diagram of the connection logic of the device according to the embodiment of the present invention;
FIG. 2 is a schematic diagram of the method steps of an embodiment of the present invention;
FIG. 3 is a schematic cross-sectional plot of a sensor according to an embodiment of the present invention;
FIG. 4 is a cloud of vertical deformation of surrounding rocks according to an embodiment of the invention;
FIG. 5 is a graph of a vertical interval of surface subsidence measurements according to an embodiment of the present invention;
FIG. 6 is a schematic diagram of deformation and settlement values for different top plate thicknesses according to an embodiment of the present invention;
FIG. 7 is a diagram of the arrangement of the measuring points of the surface subsidence cross section in the embodiment of the invention;
FIG. 8 is a schematic diagram of a machine learning process according to an embodiment of the present invention;
FIG. 9 illustrates maximum offset distance information of a tunnel according to an embodiment of the present invention;
FIG. 10 is a schematic diagram of a tunnel crack identification result according to an embodiment of the present invention;
FIG. 11 is a comparison of recognition accuracy for different input schemes according to embodiments of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the following examples.
Description of the reference numerals: the system comprises a central workstation 1, a network 2, a vehicle 3, an image acquisition device 31, a transmitting terminal 4, a real-time position sensor 5 and a position mark 6.
Example (b): the remote monitoring system of the road tunnel, it includes, the central work station 1; a network 2 connected to the central station 1 in a wireless or wired manner; a vehicle 3, an image capturing device 31 provided on the vehicle 3; and position mark 6 and/or position sensor 5 set up on the surface of the highway tunnel or nearby, the position sensor 5 sends the position information or position information measured to the network 2 through the transmitting terminal 4; the central workstation 1 can receive the image information sent by the image acquisition device 31 through the network 2; the central station 1 can receive location information via the network 2. The vehicle 3 is regularly patrolled and examined in excavation, the daily use in tunnel, scans and shoots. The vehicle 3 carries an image acquisition device 31 to acquire image information in time, screens the image information to obtain image information with a broken contour line, indexes the image information and sends the image information, and the central workstation 1 performs machine learning of big data and performs excitation of a learning model by combining measurement simulation data; during operation and maintenance monitoring, identification of high-risk cracks is carried out through the indexing image information obtained by inspection of the image acquisition device 31 of the vehicle 3.
In order to further optimize the technical scheme, the adopted measures further comprise:
the central workstation 1 performs machine learning through image information training; after numerical simulation of the position information, generating maximum offset distance information or surrounding rock deformation cloud picture information or deformation settlement numerical information, and using the maximum offset distance information or the surrounding rock deformation cloud picture information or the deformation settlement numerical information as reward information for machine learning to form a machine learning model; the machine learning model is used for risk detection of image information. And calculating the maximum offset distance information or the surrounding rock deformation cloud picture information or the deformation settlement numerical information, acquiring basic data through a transmitting terminal 4 by adopting daily measurement of a manual embedded position marker 3 and a real-time position sensor 5, and performing numerical simulation through inputting a local mountain data model of the bridge tunnel.
The road tunnel remote monitoring system comprises the following steps:
1) An information capturing step: reading image information of road tunnel inspection;
2) Information indexing step: extracting a fracture contour line in the image information, and storing the fracture contour line as a segmentation image; indexing the segmentation image with the fracture contour line;
the steps 1 and 2 are completed in the inspection stage of the vehicle 3, so that the data transmission quantity can be greatly reduced. Through screening and indexing of broken contour lines of the image information to be inspected, the data transmission pressure can be reduced. The fixed camera is fixedly arranged under the condition of manual operation and on-demand operation, and the cracks with higher risks are monitored uninterruptedly.
3) A machine learning step: converting the indexed segmented image into tensor to be used as a training set, and using numerical simulation result information of the corresponding section deformation of the indexed segmented image as reward information; training the training set by using a machine learning model, wherein the reward information is used for adjusting the machine learning model;
taking numerical simulation result information of the corresponding section deformation of the segmentation image with the index as reward information, and taking a tensor of the segmentation image with the index as a training set; until the preset accuracy is met;
4) A detection step: processing and segmenting the inspection photo of the tunnel to be detected according to the step 2), and then processing and identifying each image by using the step 3).
Through machine learning, the fracture contour line of the crack is effectively identified, and the identified fracture contour line has certain tunnel road and bridge mechanical numerical simulation relevance under the mode that the section deformation information is used as reward, so that the technical purpose of monitoring the deformation of the tunnel road and bridge through the machine identification contour line is achieved.
And the information of the settlement monitoring point is required to be input for obtaining the numerical simulation result of the corresponding section deformation. The settlement monitoring point is a common monitoring means, and a settlement monitoring probe or a mark is arranged at a monitoring position as required such as a road bridge, a tunnel, a mountain and the like, so that the mechanical condition of an object to be monitored as required is simulated in a numerical simulation mode, and the effects of predicting and monitoring results to achieve risk monitoring are obtained. The crack images are identified through machine learning, and under the action of a reward mechanism, identification of the at-risk cracks is improved. By testing a plurality of cross sections by providing a plurality of measuring points, i.e. position markers 6, as in the cross section of fig. 2, a simulation can be obtained by numerical simulation. The continuous settlement section monitoring can also be carried out by adopting a real-time position sensor 5 connected with a transmitting terminal 4.
The numerical simulation result of the corresponding section deformation is that the maximum offset distance information is contained, as shown in fig. 8, the maximum offset distance is obtained through detection, that is, the position of the inner surface of the tunnel section is calibrated by detecting a preset monitoring probe or laser scanning, then the calibration points are fitted, finally, the deviation distance between each point and the fitted curve is compared, and the maximum deviation distance is searched. A method for monitoring a tunnel section and obtaining a maximum deviation distance belongs to the prior art, and specifically can be researched by referring to Liu Guanlan, a subway tunnel deformation monitoring key technology and an analysis and forecast method, wuhan university, 2013. According to the technical scheme, the segmentation image is learned by adopting the maximum deviation distance of the section, so that the characteristics of the fracture contour line are associated with the deformation of the tunnel, and the risk is judged and predicted under the conditions of separation from a monitoring probe and laser scanning through certain machine learning. The result of crack recognition is shown in fig. 10.
The numerical simulation result of the corresponding section deformation contains the information of the surrounding rock deformation cloud chart (namely, contains continuous simulation of the surrounding rock deformation numerical value), as shown in fig. 3. The acquisition of the information of the surrounding rock deformation cloud picture is obtained through numerical simulation. Specifically, sensors are arranged on a plurality of sections of a tunnel, and numerical simulation information of each section is obtained after a numerical simulation model is established by taking monitoring information of settlement and deformation as input. For example, in a standard text such as "urban rail transit engineering measurement specification" GB50308-2008, there are regulations on monitoring tunnel deformation in both construction and operation stages. The monitoring projects in the construction stage comprise supporting structures, the structures, earth surfaces in deformation areas, buildings, pipelines and other peripheral environments. The monitoring projects in the operation stage comprise tracks, ballast beds, building structures and surrounding environments such as ground surfaces, buildings and pipelines which are influenced by operation or surrounding construction. The deformation of the subway along with the road in two stages of construction and operation is comprehensively considered, and the deformation mainly comprises the deformation of the structure along with the road in the construction and operation periods, the deformation of a supporting structure in the construction period and the deformation of a track and a track bed in the operation period.
The numerical simulation result of the corresponding section deformation contains deformation settlement numerical information. As shown in fig. 4, the deformation settlement numerical information can reflect the deformation state of the tunnel and define risky deformation and non-risky deformation according to the preset.
The deformation of the tunnel and the road bridge has obvious space-time effect. According to the general actual measurement, a two-dimensional plane strain model is established, according to the Saint-Venn principle, tunnel excavation influences stress strain in rock bodies within 3-5 times of the width of the periphery of a tunnel, the upper calculation range is taken to the ground in combination with the structural size of a tunnel main tunnel, the lower calculation range is taken from the center of the tunnel to the lower 30m, the left and right boundary calculation range is 150m, the road surface loading is 100t of standard automobile axle load, in order to simplify calculation and not consider the influence of advance support, a steel frame and sprayed concrete are equivalent to primary support, and full-section excavation is adopted. In order to analyze the influence of the thickness of the tunnel roof on the stability of the tunnel, on the premise that the elevation of the roadbed is not changed, the thickness of the tunnel roof is respectively 5m, 8m, 11m, 14m and 18m according to the actual use thickness or selected, and modeling is respectively carried out on the thicknesses. The model is gridded by 1m in consideration of the accuracy of the simulation.
The model adopts a two-dimensional plane strain model, the rock constitutive model adopts a Mokolun model, and the primary support and the secondary lining are simulated by beam units.
The highway automobile load grade adopts a first I grade of a highway, the vehicle load is adopted in the calculation of the numerical value according to the highway engineering technical standard, the standard gravity value of each vehicle is 550kN, the vehicle is uniformly distributed on each lane, the width of each lane is 3.75m, the model automobile is calculated to be uniformly loaded, and the size q =550/3.75=146.67kN/m.
And (3) combining the actual condition of the supporting and monitoring project, carrying out simulation by using finite element software, wherein the surrounding rock and tunnel supporting parameters are shown in a table 1. The primary support is equivalent parameters of a steel frame and sprayed concrete.
TABLE 1 computational model mechanics parameters Table
It can be seen from the deformation cloud picture that, when the difference of tunnel roof thickness, tunnel country rock is different with cutting slope deformation trend, and when tunnel roof thickness was 14m and 18m, tunnel country rock warp comparatively independent with cutting slope deformation trend, for the deformation condition of the tunnel country rock under each operating mode of clearer contrast, arrange 6 characteristic points in left hole, arrange 5 characteristic points on the road surface. And obtaining the deformation settlement value of any point of the actual measured tunnel section according to the simulation.
Surface subsidence
(1) Test method
Observation points, namely position marks 6 are arranged at positions where surface subsidence is likely to occur in the construction process, 3 temporary leveling base points are buried in an area 3 times of the hole diameter outside the expected section by referring to a standard leveling point burying method, and the subsidence of each observation point is calculated by taking the temporary leveling base points as a standard for the elevation measurement of each observation point. And retesting the datum points every 3 months, and judging the stability of the datum points.
(2) Laying of cross sections
The surface subsidence measuring point section is arranged at the shallow buried section of the opening and is a bias section which has influence on the tunnel construction.
The surface subsidence measuring points are preferably arranged on a transverse section v where the net width convergence measuring points in the tunnel are located, the longitudinal spacing can be adopted according to the specification of the following table, and 4 longitudinal measuring sections are arranged in each tunnel.
TABLE 2 vertical spacing of surface subsidence fracture
Depth of tunnel burial | Longitudinal spacing of cross-section |
2D<h<3D | V-VI grade surrounding rock: 20m |
D<h<2D | V-VI grade surrounding rock: 10m; II-IV grade surrounding rock: 20m |
h<D | V-VI grade surrounding rock: 5m; II-IV grade surrounding rock: 10m |
Note: d is the excavation width, and h is the tunnel buried depth.
The vertical measurement interval of the surface subsidence is shown in fig. 6.
The advance distance of the measuring points arranged on the longitudinal section is h + h1, and the longitudinal measuring interval is (h + h 1) + h' + (2-5) D; the surface subsidence measurement can arrange 5-10 measuring points on the cross section, the number of the measuring points can be properly increased particularly for the weak surrounding rock section, and the distance between the two measuring points is 2-5 m. The measuring points near the tunnel center line should be arranged densely, and the measuring points far away from the tunnel center line should be sparse. The surface subsidence cross-sectional test point arrangement is shown in figure 7.
(3) Station burying
The measuring line on each section is vertical to the central line of the tunnel, the central monitoring point is arranged on the ground surface of the tunnel axis when the measuring points are buried, other monitoring points are symmetrically arranged along the central line, and the distance between the measuring points is from the central point to the point farthest from the axis of the tunnel on the ground surface from dense to sparse.
And 2 temporary leveling base points are buried in the areas outside 3 times of the hole diameter in the longitudinal and transverse directions of the tunnel excavation according to a standard leveling point burying method. The datum point requires good visibility, the measurement is convenient, the foundation is firm, and the protection is easy. One datum point serves as a test station and the other as a check point.
And the monitoring network observation mark adopts reinforced concrete observation mark piers. The standard pier foundation is strived to be stable, or a surface weathered layer is removed to pour the standard pier on the fresh bedrock; or when the surface covering layer is thick, a foundation pit is dug, the depth is not less than 1m, and the mark pier is cast in place.
Digging a pit with the length, width and depth of 200mm at the position of a measuring point, then placing the pit into a self-made embedded part, adopting a flat round-head reinforcing steel bar with the diameter of 20-30 mm and the length of 1m at the measuring point, welding a steel plate with the side length of about 5cm square at one end of the reinforcing steel bar, placing the other end of the reinforcing steel bar into the pit, using a hammer to knock the patch to be 30cm away from the ground surface, adjusting the direction of the measuring point, and using concrete to fill the pit until the concrete is solidified, thus measuring.
(4) Monitoring instrument
And the surface settlement adopts a high-precision Chelidica LS10 level gauge and an indium steel ruler.
(5) Monitoring frequency
When the excavation face distance measuring section is less than 2B, 1-2 times/day, when the excavation face distance measuring section is less than 5B, 1 time/2 day, when the excavation face distance measuring section is more than 5B, 1 time/week.
Machine learning:
in the embodiment, a parting identification algorithm of an opportunistic CNN-SVM (convolutional neural network-support vector machine) is adopted to perform blocking, feature extraction and classification on the image.
And (3) rolling layers:
firstly, selecting a position (x, y) on the graph A to be convolved, then calculating the coiling machine by taking the position as the center, and outputting the value of the position (x, y) on the graph H by the coiling machine. The jth channel 4 of the 1-layer convolution output is computed. The convolution kernel is represented by a heart, and the convolution kernel and the X of the 1 st layer to the 1 st layer are calculated each time 1-1 The parts are connected. The input in the field of view is denoted as x i Output quantity z of jth channel before nonlinear mapping j . Wherein z is j Is a scalar quantity, representing the matrix Z j A point of (a), b j Is neuron biased. X is to be j And stretching the vector and calculating an inner product with the convolution kernel vector to obtain a convolution result.
The discrete convolution can be converted to a matrix multiplication. And multiplying the convolution kernel matrix with the transformed feature map matrix to realize one-time feature map convolution.
A pooling layer:
pooling is performed using, for example, maxporoling.
Full connection layer:
at this level, the neurons at level 1 and each neuron at levels 1-1 are linked, and the connection weight matrix is W 1 。
Activation function:
the activation function is the reward. In the model, the response relationship is determined by the activation function. In the embodiment, a sigmoid function is adopted, and a tanh function can also be adopted. As an optimization scheme, training can be accelerated by using a corrective linear unit (ReLUs).
Softmax layer:
the fully connected layer is followed by the softmax layer, the sum of the outputs of which is 1. The technical scheme adopts probability distribution. For example:
loss function:
the error J and the sample number of the technical scheme are m, and the superscript (i) represents the sample serial number, and the relation is as follows:
the back propagation phase is in a classical way and will not be described in detail here.
The technical scheme is further improved on the softmax layer by adjusting the probability function so as to improve the corresponding learning effect. The specific adjustment is as follows:
preferably, the probability distribution of the softmax layer is adjusted, and experimental simulation shows that the learning speed can be further improved by about 2.7 +/-0.12% on the basis of the probability distribution.
In order to enable neuron learning to have better risk prediction capability, the maximum offset distance information of the settlement monitoring points, the surrounding rock deformation cloud picture information and the maximum value and the minimum value of the deformation settlement numerical value information after traversal of all the points are normalized and then are brought into x and y.
The model mode comparison is formed after the maximum deviation distance information, the surrounding rock deformation cloud picture information and the deformation and sedimentation numerical value information are learned and rewarded through three machines, as shown in the figure 11, the recognition accuracy rate is the best that the learning model is adjusted through the maximum deviation distance, risk cracks can be recognized accurately to the maximum degree, the figure 11 also shows that the two machine learning rewards of the surrounding rock deformation cloud picture information and the deformation and sedimentation numerical value information obtain excellent recognition accuracy rate, and the comparison analysis is characterized in that the optimization of a probability function and a loss function is beneficial to improving the recognition efficiency of the machine learning model, so that the applicable mode of the machine learning rewards is effectively expanded, the recognition accuracy of the method under the condition that partial reward modes are lacked is improved, the model after the machine learning rewards through the maximum deviation distance information, the surrounding rock deformation cloud picture information or the deformation and sedimentation numerical value information can recognize the risk cracks accurately to the great degree, the model after the three machine learning rewards are trained through the cross comparison analysis, the error recognition of the risk cracks through the maximum deviation distance information, the surrounding rock deformation cloud picture information or the deformation and sedimentation numerical value information can be avoided, and the model trained through the single reward mode can be used for recognizing the potential risk cracks accurately, and the accuracy of the monitoring system can be improved.
Through the implementation of this technical scheme, can realize detecting the crackle risk with 3 daily modes of patrolling and examining of vehicle, avoid setting up expensive sensors such as radar in a large number, obvious can reduce implementation cost.
While the invention has been described in connection with a preferred embodiment, it is not intended to limit the invention, since various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (5)
1. Highway tunnel remote monitering system, characterized by: comprises the steps of (a) preparing a mixture of a plurality of raw materials,
a central workstation (1);
a network (2) connected to said central station (1) in a wireless or wired manner;
a vehicle (3) is provided with a vehicle body,
the image acquisition device (31) is arranged on the vehicle (3) and is used for acquiring the inspection image information of the vehicle, screening and indexing part of the image information with the fracture contour line;
and position mark (6) and/or position sensor (5) set up on the surface of the highway tunnel or nearby, the said position sensor (5) sends the position information or position information measured to the said network (2) through the transmitting terminal (4);
the central workstation (1) can receive the image information sent by the image acquisition device (31) through a network (2); the central workstation (1) can receive the position information through a network (2).
2. The road tunnel remote monitoring system according to claim 1, wherein: the central workstation (1) performs machine learning through the image information training; after the position information is subjected to numerical simulation, generating maximum offset distance information or surrounding rock deformation cloud picture information or deformation settlement numerical information, and taking the maximum offset distance information or the surrounding rock deformation cloud picture information or the deformation settlement numerical information as reward information of the machine learning to form a machine learning model; and using the machine learning model for risk detection of image information.
3. The road tunnel remote monitoring system of claim 1, wherein: the image acquisition device (31) on the vehicle (3) acquires image information and then adopts the following steps:
1) An information capturing step: reading the image information acquired by the road tunnel inspection or fixed camera;
2) Information indexing step: extracting a fracture contour line in the image information, and storing the fracture contour line into a segmentation image; indexing the segmentation image with the fracture contour line; and then, transmitting the indexed segmentation image.
4. The road tunnel remote monitoring system of claim 1, wherein: the central workstation (1) carries out remote monitoring on the image information and adopts the following steps:
3) A machine learning step: converting the received indexed segmentation image into tensor as a training set, and taking numerical simulation result information of section deformation corresponding to the indexed segmentation image as reward information; training a training set by using a machine learning model, wherein the reward information is used for adjusting the machine learning model;
taking numerical simulation result information of the corresponding section deformation of the segmentation image with the index as reward information, and taking the tensor of the segmentation image with the index as a training set; until the preset accuracy is met;
4) A detection step: processing and segmenting the inspection photo of the tunnel to be detected according to the step 2), and then processing and identifying each image by using the step 3).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210991425.8A CN115359447B (en) | 2022-08-01 | 2022-08-01 | Highway tunnel remote monitoring system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210991425.8A CN115359447B (en) | 2022-08-01 | 2022-08-01 | Highway tunnel remote monitoring system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115359447A true CN115359447A (en) | 2022-11-18 |
CN115359447B CN115359447B (en) | 2023-06-20 |
Family
ID=84003626
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210991425.8A Active CN115359447B (en) | 2022-08-01 | 2022-08-01 | Highway tunnel remote monitoring system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115359447B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112528059A (en) * | 2021-02-08 | 2021-03-19 | 南京理工大学 | Deep learning-based traffic target image retrieval method and device and readable medium |
WO2021223124A1 (en) * | 2020-05-06 | 2021-11-11 | 深圳市大疆创新科技有限公司 | Position information obtaining method and device, and storage medium |
CN114022448A (en) * | 2021-11-05 | 2022-02-08 | 南京睿捷智慧交通科技研究院有限公司 | Tunnel wall crack and bulge detection method and multifunctional inspection system |
CN114419421A (en) * | 2022-01-21 | 2022-04-29 | 中国地质大学(北京) | Subway tunnel crack identification system and method based on images |
CN114663840A (en) * | 2022-05-13 | 2022-06-24 | 浙江有色地球物理技术应用研究院有限公司 | Tunnel environment inspection equipment monitoring method and system |
-
2022
- 2022-08-01 CN CN202210991425.8A patent/CN115359447B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2021223124A1 (en) * | 2020-05-06 | 2021-11-11 | 深圳市大疆创新科技有限公司 | Position information obtaining method and device, and storage medium |
CN112528059A (en) * | 2021-02-08 | 2021-03-19 | 南京理工大学 | Deep learning-based traffic target image retrieval method and device and readable medium |
CN114022448A (en) * | 2021-11-05 | 2022-02-08 | 南京睿捷智慧交通科技研究院有限公司 | Tunnel wall crack and bulge detection method and multifunctional inspection system |
CN114419421A (en) * | 2022-01-21 | 2022-04-29 | 中国地质大学(北京) | Subway tunnel crack identification system and method based on images |
CN114663840A (en) * | 2022-05-13 | 2022-06-24 | 浙江有色地球物理技术应用研究院有限公司 | Tunnel environment inspection equipment monitoring method and system |
Also Published As
Publication number | Publication date |
---|---|
CN115359447B (en) | 2023-06-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Solla et al. | A review of GPR application on transport infrastructures: Troubleshooting and best practices | |
CN112945139A (en) | Shield engineering auxiliary system combining three-dimensional scanning with BIM technology | |
CN111895911B (en) | Method for monitoring hidden danger of ground collapse of shallow sand layer | |
CN114353876A (en) | Loess highway slope health monitoring method | |
CN111473779B (en) | Method for identifying and monitoring deformation of landslide-tunnel system in linkage manner | |
CN114548278A (en) | In-service tunnel lining structure defect identification method and system based on deep learning | |
Mazzanti | Toward transportation asset management: what is the role of geotechnical monitoring? | |
CN112948952B (en) | Evolution prediction method for cavity behind shield tunnel lining | |
CN112541665A (en) | Slope stability refined evaluation method based on multi-source information fusion | |
CN113936228A (en) | High slope monitoring and early warning system based on unmanned aerial vehicle data check | |
CN114663840B (en) | Tunnel environment inspection equipment monitoring method and system | |
CN116203559A (en) | Intelligent recognition and early warning system and method for underground rock and soil disease body | |
CN112213722A (en) | Side slope monitoring method combining InSAR and deep learning | |
CN113742827B (en) | Highway slope monitoring network system construction method based on finite difference analysis | |
Qin et al. | Development and application of an intelligent robot for rock mass structure detection: A case study of Letuan tunnel in Shandong, China | |
Koohmishi et al. | Advancing railway track health monitoring: Integrating GPR, InSAR and machine learning for enhanced asset management | |
CN108614307B (en) | Inverted arch quality detection method and system by using comprehensive geophysical prospecting technology | |
CN113111411A (en) | System and method for reinforcing deep silt foundation | |
Zhao et al. | A study of landslide deformation fields with a digital correlation method | |
Huang et al. | Research on void signal recognition algorithm of 3D ground-penetrating radar based on the digital image | |
CN115116198B (en) | Air-ground integrated road collapse monitoring method and system | |
CN115359447B (en) | Highway tunnel remote monitoring system | |
Huang et al. | A photogrammetric system for tunnel underbreak and overbreak detection | |
CN116794651A (en) | Landslide feature interpretation method and system based on aerial survey ground penetrating radar and image data | |
Martínez-Sánchez et al. | Automatic thickness and volume estimation of sprayed concrete on anchored retaining walls from terrestrial lidar data |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
CB03 | Change of inventor or designer information | ||
CB03 | Change of inventor or designer information |
Inventor after: Xu Zheng Inventor after: Zhang Jianhua Inventor after: Zhang Wei Inventor after: Han Weiwei Inventor after: Wang Haiming Inventor before: Wang Haiming Inventor before: Xu Zheng Inventor before: Zhang Jianhua Inventor before: Zhang Wei Inventor before: Han Weiwei |
|
GR01 | Patent grant | ||
GR01 | Patent grant |