A method for monitoring deformation of tunnel surrounding rock based on virtual target

Abstract

The present invention discloses a method for monitoring deformation of tunnel surrounding rocks based on virtual targets; when the method is applied, a camera device, a virtual target device and a PC device are required. The specific monitoring method includes: S1, arranging the camera device and the virtual target device at intervals along the length direction of the tunnel pavement; S2, controlling the virtual target device to form an annular aperture on the cross section in the tunnel; S3, using the camera device to delay shooting the annular aperture, and sending the collected data to the PC device; the PC device converts the data of the annular aperture into a displacement and time relationship curve of the aperture point, and forms a surrounding rock deformation time curve based on this. The present invention does not require excessive interference processing by the staff during the entire process, and the equipment can detect and feedback the monitoring data for a long time, which can reduce manpower consumption; the method can realize deformation monitoring of any point of the entire section, combined with long-term monitoring and data feedback, to achieve continuous surrounding rock deformation observation.

Description

Technical Field

[0001] The present invention belongs to the technical field of tunnel detection, and in particular relates to a tunnel surrounding rock deformation monitoring method based on a virtual target. Background Art

[0002] As an important part of infrastructure such as transportation and water conservancy, the safety and stability of tunnel engineering are of great significance. Tunnel deformation is one of the common problems during the construction and later operation of tunnels, which may lead to the destruction of tunnel structures and risks to operational safety. Therefore, tunnel deformation monitoring has become an indispensable part of tunnel engineering. Tunnel section shape changes and convergence deformation monitoring are important contents of tunnel deformation monitoring.

[0003] Traditional high-precision tunnel deformation measurement technology often uses manual testing with a total station. This test method requires manual setting of targets (i.e. installing several reflective targets on the test tunnel section, and then measuring the target coordinates one by one), and the test process will consume a lot of manpower. Currently, the more advanced measurement equipment is the laser scanner, which also measures and fits the entire section point by point. Obviously, this type of monitoring method is difficult to achieve deformation monitoring of any point on the entire section, and it is difficult to achieve long-term continuous deformation observation. Therefore, it is of great significance to seek a method that is convenient, fast, low-cost, high-precision, and can achieve real-time continuous deformation monitoring for tunnel deformation monitoring. Summary of the invention

[0004] The purpose of the present invention is to provide a tunnel surrounding rock deformation monitoring method based on virtual targets, so as to solve the technical problem of high manpower consumption caused by the existing monitoring method of setting up physical targets, and the technical problem of being unable to achieve long-term continuous deformation observation in the tunnel surrounding rock monitoring method using laser scanners.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] A tunnel surrounding rock deformation monitoring method based on virtual targets.

[0007] Providing a camera device;

[0008] A virtual target device is provided, which includes a light emitting device capable of forming an annular laser aperture;

[0009] Provide a PC device;

[0010] Monitoring methods include:

[0011] S1. Arrange the camera device and the virtual target device at intervals along the length direction of the tunnel pavement;

[0012] S2, controlling the virtual target device to form a ring-shaped laser aperture on the cross section in the tunnel;

[0013] S3. Use a camera device to delay shooting of the annular laser aperture, and send the collected data to a PC device; the PC device converts the data of the annular laser aperture into a displacement and time relationship curve of the aperture point, and forms a surrounding rock deformation time curve based on this.

[0014] Furthermore, in step S3, the PC device processes the collected data in the following process to obtain the surrounding rock deformation time curve:

[0015] s1. Use low-light image enhancement algorithm to reconstruct image frame sequence from the collected image data;

[0016] s2, DeepLabV3+ network based on deep learning performs object segmentation on the image enhanced by s1;

[0017] s3, processing the surrounding rock small deformation motion amplification based on the phase of the image frame sequence segmented by s2;

[0018] s4. Further process the image processed by s3 and obtain the surrounding rock deformation time curve.

[0019] Furthermore, in step s1, after receiving the image data transmitted by the camera device, the PC device enhances the brightness of each frame of the image through a zero-reference depth curve estimation algorithm.

[0020] Furthermore, in step s2, the PC device uses the deep learning algorithm DeepLabV3+ network to segment the annular laser aperture and the image calibration target in the image, and binarizes the foreground and background of the segmented and extracted image.

[0021] Furthermore, in step s3, the previously segmented annular laser aperture image frame sequence is subjected to motion amplification with a slight displacement.

[0022] Furthermore, in step s4, the pixel displacement of the image calibration target and the annular laser aperture is extracted, and the pixel displacement is restored to the real physical displacement to form a surrounding rock deformation time curve.

[0023] Furthermore, the virtual target device also includes:

[0024] An electric hand-pushed flatbed cart comprises a flatbed support platform, on which are provided three positioning rods arranged in a triangle and threadedly matched with the flatbed support platform;

[0025] The camera support frame includes a multi-section electric push rod arranged horizontally and fixed on the flat support platform, and a support rod arranged vertically at one end of the multi-section electric push rod; the camera device adopts a camera and is connected to the support rod;

[0026] The light source support frame includes columns arranged parallel to and spaced apart from the support rods, a flat base rotatably connected to the top of the columns, a U-shaped frame fixed on the upper side of the base, a flip rod arranged transversely and hinged on the inner side of the U-shaped frame, a torsion spring connected between the flip rod and the U-shaped frame, an electronic level arranged on the flip rod, and an adjusting screw threadedly matched with the base and arranged vertically; the top end of the adjusting screw is supported on the lower side of the flip rod, and the electronic level is wirelessly connected to the PC device; one side of the base is provided with a first top screw threadedly matched and acting on the adjusting screw.

[0027] Furthermore, the light-emitting device includes a disk with a groove on its outer circumference, an annular laser light strip embedded in the groove, and a laser controller arranged on one end surface of the disk and connected to the annular laser light strip; the one end surface of the disk is arranged vertically and fixedly connected to one end of the flip rod.

[0028] Furthermore, in step S3, the PC device also receives data from an electronic level, and uses data changes of the level to eliminate significant abnormal data or issue an alarm.

[0029] The beneficial effects of the present invention are:

[0030] The present invention uses the light-emitting device on the virtual target device to form an annular laser aperture on the tunnel surrounding rock section, thereby forming a virtual target, and then uses the camera device to collect image data of the annular laser aperture from time to time, and uses the PC device to receive and analyze the data, and finally forms a surrounding rock deformation time curve; the whole process does not require excessive intervention by the staff, the equipment can detect and feedback the monitoring data for a long time, and can reduce manpower consumption; the method can realize deformation monitoring of any point in the whole section, combined with long-term monitoring and data feedback, to achieve continuous surrounding rock deformation observation, and provide data support for the convergence of surrounding rock deformation during tunnel construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is a flow chart of the monitoring method of the present invention;

[0032] Figure 2 This is a flow chart of the PC device processing the collected data in the present invention;

[0033] Figure 3 It is a schematic diagram of the calibration target image processing parameters;

[0034] Figure 4 It is a schematic diagram of the deformation time curve of the tunnel surrounding rock during the construction process;

[0035] Figure 5 It is a structural schematic diagram of the virtual target device in the present invention;

[0036] Figure 6It is a schematic diagram of the top view of the light emitting device in the present invention.

[0037] Among them, 1-PC equipment; 2-camera; 3-electric hand-pushed flatbed cart; 4-flat support table; 5-positioning rod; 6-multi-section electric push rod; 7-support rod; 8-column; 9-flat base; 10-U-shaped frame; 11-flip rod; 12-torsion spring; 13-electronic level; 14-adjusting screw; 15-second top screw; 16-ring; 17-third top screw; 18-first top screw; 19-disc; 20-ring laser light belt; 21-laser controller. DETAILED DESCRIPTION

[0038] like Figures 1 to 6 As shown, a method for monitoring deformation of tunnel surrounding rock based on virtual target is provided in advance, wherein a camera device, a virtual target device and a PC device 1 are provided. The virtual target device comprises a light emitting device capable of forming an annular laser aperture, which is used to form an annular laser aperture on the tunnel surrounding rock section; the camera device is used to collect image data of the annular laser aperture, and the PC device 1 is used to process the image data and form a time curve of surrounding rock deformation.

[0039] The camera device adopts a camera 2, which has a timed recording function and a wired or wireless data transmission function component.

[0040] The virtual target device includes an electric hand-pushed flatbed trolley 3, a camera support frame, a light source support frame and a light-emitting device. The electric hand-pushed flatbed trolley 3 includes a flatbed support platform 4; the flatbed support platform 4 is provided with three positioning rods 5 arranged in a triangle and threadedly matched with the flatbed support platform 4, and the positioning rods 5 are specifically arranged on the front side and the left and right sides of the flatbed support platform 4. By rotating the positioning rods 5, the wheels of the electric hand-pushed flatbed trolley 3 are all separated from the ground, which can improve the stability of the electric hand-pushed flatbed trolley 3 when parked. The camera support frame includes a multi-section electric push rod 6 and a support rod 7; the multi-section electric push rod 6 is arranged horizontally and fixed on the flatbed support platform 4, and the support rod 7 is perpendicular to the flatbed support platform 4 and arranged vertically; the camera 2 is connected to the support rod 7. The light source support frame includes a column 8, a flat base 9, a U-shaped frame 10, a flip rod 11, a torsion spring 12, an electronic level 13 and an adjusting screw 14; the column 8 is arranged parallel to the support rod 7 and spaced apart, and the column 8 is fixed on the flat support table 4; a sleeve is provided on the lower side of the flat base 9, the sleeve is sleeved on the top of the column 8, and a second top screw 15 is threadedly matched on the outer wall of the sleeve, and the second top screw 15 is supported on the column 8; the U-shaped frame 10 is fixed on the upper side of the flat base 9; the flip rod 11 is arranged horizontally and hinged on the inner side of the U-shaped frame 10, and the specific hinge structure is that the flip rod 11 is radially symmetrically provided with two rotating shafts, the rotating shaft passes through the two side arms of the U-shaped frame 10 and rotates together; the torsion spring 12 is sleeved on the two rotating shafts, and the two connecting ends of the torsion spring 12 are respectively connected and fixed to the flip rod 11 and the U-shaped frame 10, and the flip rod 11 The free end flips toward the base side; the electronic level 13 is connected to the flip rod 11 through a ring 16 sleeved on the flip rod 11, and a third top screw 17 with a threaded fit is provided on the ring 16; the third top screw 17 acts on the flip rod 11, and is used to adjust the circumferential position of the electronic level 13 on the flip rod 11, so that the electronic level 13 can be adjusted to measure accurately. The electronic level 13 is wirelessly connected to the PC device 1 and transmits horizontal data; the adjusting screw 14 is threaded with the base and is arranged vertically, and the top end of the adjusting screw 14 is supported on the lower side of the flip rod 11. The adjusting screw 14 is operated and with the cooperation of the torsion spring 12, the horizontal adjustment of the flip rod 11 is realized, and a first top screw 18 with a threaded fit and acting on the adjusting screw 14 is provided on one side of the flat base 9, and the adjusting screw 14 is stopped by the first top screw 18. The light-emitting device includes a disk 19 with a groove on the outer circumference, a ring-shaped laser light strip 20 embedded in the groove, and a laser controller 21 arranged on one end surface of the disk 19 and connected to the ring-shaped laser light strip 20; one end surface of the disk 19 is arranged vertically and fixedly connected to the other end of the flip rod 11.

[0041] The PC device 1 can be a notebook computer or a smart phone.

[0042] A controller with a touch screen is installed on the electric hand-pushed flatbed cart 3. The camera device, laser controller 21, multi-section electric push rod 6 and electronic level 13 are electrically connected to the controller respectively. The controller is powered by the power supply of the electric hand-pushed flatbed cart 3, and the original control system of the electric hand-pushed flatbed cart 3 remains unchanged.

[0043] The specific monitoring methods of tunnel surrounding rock deformation monitoring based on virtual targets include:

[0044] S1, control the electric hand-pushed flatbed truck 3 to the tunnel monitoring area, use the controller to control the multi-section electric push rod 6 to move horizontally for a certain distance, so that the camera device and the virtual target device are separated by more than 1.5m along the length direction of the tunnel pavement; rotate the three positioning rods 5 to make the tires of the electric hand-pushed flatbed truck 3 leave the ground; operate the adjustment screw 14 and use the electronic level 13 to keep the flip rod 11 in a horizontal state, at this time the disc 19 is vertical, and the formed laser aperture is irradiated on the same section of the tunnel surrounding rock, which is conducive to improving the accuracy of surrounding rock deformation monitoring; by rotating the flat plate base plate and using the second top screw 15, the disc 19 is relatively parallel to the surrounding rock section;

[0045] S2, controlling the laser controller on the virtual target device so that the annular laser light strip 20 forms an annular laser aperture on the cross section in the tunnel;

[0046] S3, using a camera device to delay and continuously shoot the annular laser aperture, and send the collected data to the PC device 1; the PC device 1 converts the data of the annular laser aperture into a displacement and time relationship curve of the aperture point, and forms a surrounding rock deformation time curve based on this. The PC device 1 also receives data from the electronic level 13, and uses the data changes of the level to eliminate significant abnormal data or implement an alarm.

[0047] In the above step S3, the PC device 1 processes the collected data in the following process to obtain the surrounding rock deformation time curve:

[0048] s1. Use low-light image enhancement algorithm to reconstruct the image frame sequence from the collected image data; due to the weak light intensity during tunnel construction, it is necessary to use the deep learning network zero-reference deep curve estimation algorithm (ZeroDCE Zero-reference Deep Curve Estimation) to enhance the brightness of each frame of the image, and transmit the enhanced image to the video analysis and processing end. The ZeroDCE deep learning network is used mainly because it does not require any paired data for training, and it uses a lightweight network to predict the high-order pixels of pixels, so that it can run in real time on mobile phones, thus meeting the real-time monitoring of tunnel surrounding rock deformation.

[0049] s2. The DeepLabV3+ network based on deep learning performs target segmentation on the image enhanced in s1; the foreground and background of the segmented and extracted image are binarized; in this step, some laser projection and target images of the tunnel surrounding rock need to be manually marked as training sets and test sets, and trained using the DeepLab V3+ network to obtain model parameters such as a reasonable learning rate; the trained model is then integrated into the image analysis device to quickly realize the image segmentation of the laser projection and calibration target of the tunnel surrounding rock deformation image frame sequence; in order to reduce the amount of subsequent image processing calculations, the pixel value of the extracted foreground image is set to 1, and the remaining background pixel values ​​are set to 0, so that the image is converted into a binary image.

[0050] S3, the image frame sequence after segmentation in S2 is processed by phase-based motion amplification of small deformation of surrounding rock; due to the slow deformation rate of tunnel surrounding rock, the real-time monitoring of surrounding rock deformation faces great challenges; the time-lapse shooting technology can speed up the structural deformation to a certain extent on the video frame, but some important details of deformation may be ignored; for this reason, in this step, the video frame sequence of the laser projection circle image segmented in the previous step is amplified by the motion of small displacement, so that small deformation can be easily and accurately extracted; the main process to achieve this purpose is described as follows:

[0051] 1) Convert the video space domain of the selected time period into the frequency domain. Select the video frames of the segmented foreground image area during the time period of interest (e.g., select the video every 1 hour after the excavation is completed), perform 2D Fourier transform, realize the image space domain to frequency domain conversion, and obtain the image amplitude spectrum and phase spectrum;

[0052] 2) Phase-based amplification of small deformations. Based on 1), the first frame of the video in the selected period is used as a reference, and the phase of the subsequent frames is subtracted from the first frame to obtain the phase difference of each image. Since time-lapse photography is used to obtain the tunnel deformation video, it is easy to obtain the frequency of laser projection changes. , as shown in formula ①. Set an appropriate magnification factor λ and The phase difference of the frequency baseband is multiplied by λ to achieve the amplification of small deformations;

[0053] ①

[0054] In the formula, is the delay interval (s);

[0055] 3) Image reconstruction of the magnified video: The processed frequency baseband in 2) is added to the low-pass residual image sequence, and the image sequence is reconstructed using a complex controllable pyramid to output the processed video. At the same time, motion magnification of a specific frequency band can effectively reduce the impact of noise outside the frequency band and improve the signal-to-noise ratio.

[0056] s4. Further process the image processed by s3 and obtain the deformation time curve of the surrounding rock. The specific description is as follows:

[0057] 1) Obtain the coordinate base point and conversion factor of each frame image. Based on the reconstructed video, select the calibration target area, use the Canny edge detection algorithm to extract the circular contour of the calibration target, and calculate the centroid coordinates of the contour line. and the mean radius of the contour line (Formulas ② and ③). On this basis, the scale conversion factor of each frame image is obtained according to the ratio of the actual size of the calibration target to the pixel size. (Formula ④).

[0058] ②

[0059] ③

[0060] ④

[0061] Where: and represents the center of the calibration target in the j-th frame image; and represents the coordinates of the i-th pixel of the calibrated target contour line in the j-th frame image (such as Figure 3 ), B is the number of contour line pixels; and represents the radius of the calibration target circle and the area of ​​the pixels in the j-th frame image. is the area of ​​a certain pixel in the calibration target of the j-th frame image, H is the number of pixels contained in the calibration target; R is the actual radius of the calibration target circle.

[0062] 2) Obtain the coordinate difference of the relative base point. Based on the reconstructed video, a thinning algorithm is used for the laser projection circle of the video frame sequence image to obtain the pixel-level contour skeleton curve of the laser projection in each frame image. Taking the calibration target in 1) as the base point, calculate the sum of the coordinate differences between the pixel center and the base point of any part to be analyzed in the contour skeleton curve of the image.

[0063] 3) Obtain the displacement time curve of the analyzed part. Based on the first frame image, use formula ⑤ to obtain the actual displacement of the calibration target in the subsequent frame image sequence. and At the same time, according to the relative displacement relationship between the analysis point and the base point of the laser image contour skeleton curve, the true displacement of the analysis part can be obtained by formula ⑥ and According to the time interval of time-lapse shooting and the aperture displacement of laser projection, the deformation displacement time curve can be obtained as follows: Figure 4As shown (only a linear illustration). By merging and loading the video frames of each time period, a video of the circumferential deformation of the tunnel surrounding rock can be obtained, which is convenient for intuitive analysis and judgment of the deformation convergence. When the equipment is significantly deformed due to vibration or other reasons, the electronic level 10 will be abnormal, and the alarm function will be realized with the help of the PC device 3 to prompt the staff to conduct inspection and maintenance; when the abnormal range fluctuates slightly, the PC device 3 can be used to eliminate the video data of this period to improve the overall detection accuracy. This operation can be set in step s1, or it can be placed after step 3) of s4;

[0064] ⑤

[0065] ⑥.

Claims

1. A method for monitoring tunnel surrounding rock deformation based on virtual targets, characterized in that: Providing a camera device; A virtual target device is provided, which includes a light emitting device capable of forming an annular laser aperture; Provide a PC device; Monitoring methods include: S1. Arrange the camera device and the virtual target device at intervals along the length direction of the tunnel pavement; S2, controlling the virtual target device to form a ring-shaped laser aperture on the cross section in the tunnel; S3, using a camera device to delay shooting of the annular laser aperture, and sending the collected data to a PC device; the PC device converts the data of the annular laser aperture into a displacement and time relationship curve of the aperture point, and forms a surrounding rock deformation time curve based on this; wherein the PC device processes the collected data in the following process to obtain the surrounding rock deformation time curve: s1. Use low-light image enhancement algorithm to reconstruct image frame sequence from the collected image data; s2, DeepLabV3+ network based on deep learning performs object segmentation on the image enhanced by s1; s3, processing the surrounding rock small deformation motion amplification based on the phase of the image frame sequence segmented by s2; s4. Further process the image processed by s3 and obtain the surrounding rock deformation time curve.

2. A method for monitoring deformation of surrounding rock of a tunnel based on virtual target according to claim 1, characterized in that: In the step s1, after receiving the image data transmitted by the camera device, the PC device enhances the brightness of each frame of the image through a zero-reference depth curve estimation algorithm.

3. The method for monitoring deformation of tunnel surrounding rock based on virtual targets according to claim 1, characterized in that: In step s2, the PC device uses the deep learning algorithm DeepLabV3+ network to segment the annular laser aperture and the image calibration target in the image, and binarizes the foreground and background of the segmented and extracted image.

4. The method for monitoring deformation of surrounding rock of a tunnel based on virtual target according to claim 1, characterized in that: In step s3, the previously segmented annular laser aperture image frame sequence is subjected to motion amplification with a slight displacement.

5. The method for monitoring deformation of tunnel surrounding rock based on virtual targets according to claim 1, characterized in that: In step s4, the pixel displacement of the image calibration target and the annular laser aperture is extracted, and the pixel displacement is restored to the real physical displacement to form a surrounding rock deformation time curve.

6. A method for monitoring deformation of surrounding rock of a tunnel based on virtual targets according to any one of claims 1 to 5, characterized in that: The virtual target device also includes: An electric hand-pushed flatbed cart comprises a flatbed support platform, on which are provided three positioning rods arranged in a triangle and threadedly matched with the flatbed support platform; The camera support frame includes a multi-section electric push rod arranged horizontally and fixed on the flat support platform, and a support rod arranged vertically at one end of the multi-section electric push rod; the camera device adopts a camera and is connected to the support rod; The light source support frame includes columns arranged parallel to and spaced apart from the support rods, a flat base rotatably connected to the top of the columns, a U-shaped frame fixed on the upper side of the base, a flip rod arranged transversely and hinged on the inner side of the U-shaped frame, a torsion spring connected between the flip rod and the U-shaped frame, an electronic level arranged on the flip rod, and an adjusting screw threadedly matched with the base and arranged vertically; the top end of the adjusting screw is supported on the lower side of the flip rod, and the electronic level is wirelessly connected to the PC device; one side of the base is provided with a first top screw threadedly matched and acting on the adjusting screw.

7. A method for monitoring deformation of surrounding rock of a tunnel based on virtual target according to claim 6, characterized in that: The light-emitting device comprises a disc with a groove on the outer circumference, an annular laser light strip embedded in the groove, and a laser controller arranged on one end surface of the disc and connected to the annular laser light strip; the one end surface of the disc is arranged vertically and fixedly connected to one end of a flip rod.

8. The method for monitoring deformation of surrounding rock of a tunnel based on virtual target according to claim 6, characterized in that: In step S3, the PC device also receives data from the electronic level, and uses data changes of the level to eliminate significant abnormal data or issue an alarm.

Images