Tunnel structure displacement deformation measurement system and method based on image recognition
By deploying targets and image displacement sensors inside the tunnel, and combining digital image correlation algorithms and improved tangent angle algorithms, the problems of low measurement frequency and inaccurate prediction in traditional tunnel deformation monitoring methods have been solved. This has enabled efficient and continuous absolute displacement measurement and deformation early warning, ensuring tunnel safety.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-18
- Publication Date
- 2026-03-20
AI Technical Summary
Traditional tunnel deformation monitoring methods have low measurement frequency, rely on manual operation, and have inaccurate risk prediction. They are difficult to achieve continuous real-time monitoring and absolute displacement measurement, especially in complex construction environments where data continuity and accuracy are insufficient.
An image recognition-based tunnel structure displacement and deformation measurement system is adopted. By deploying multiple targets and image displacement sensors inside the tunnel, combined with digital image correlation algorithms and cascaded target design, absolute displacement measurement is achieved, and an improved tangent angle algorithm is used for deformation trend prediction and early warning.
It enables efficient, continuous, and accurate absolute displacement measurement and deformation early warning within the tunnel, reduces interference from the construction environment on data acquisition, improves the accuracy of data processing and the reliability of prediction, and ensures the safe operation of the tunnel.
Smart Images
Figure CN121702296A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of tunnel engineering and structural health monitoring technology, specifically to a tunnel structure displacement and deformation measurement system and method based on image recognition. Background Technology
[0002] During tunnel construction, underground excavation disrupts the original equilibrium of the rock mass, leading to stress redistribution and structural deformation. Excessive or rapid deformation can easily trigger collapses and other safety accidents. Therefore, real-time and accurate deformation monitoring and early warning systems for tunnels under construction are crucial.
[0003] Currently, traditional tunnel deformation monitoring methods mainly include convergence meters, levels, and total stations. These methods generally have the following limitations: (1) low measurement frequency, making it impossible to achieve continuous real-time monitoring and difficult to capture sudden deformations; (2) reliance on manual on-site operation, resulting in low efficiency and personal safety risks, and the tunnel deformation monitoring technology is not intelligent enough; (3) difficulty in setting up benchmark points, especially in long tunnels, making it difficult to achieve true absolute displacement measurement and risk prediction is not accurate enough.
[0004] In recent years, automated monitoring technologies such as 3D laser scanning and measuring robots have emerged. However, these technologies are expensive, require professional maintenance, and in some cases, the field of view is still obstructed by construction equipment (such as trolleys and vehicles). While image recognition-based monitoring technologies have advantages such as non-contact and full-field measurement, ensuring data continuity in complex construction environments, transmitting stable benchmarks from the rear to the front excavation face, and accurately predicting and warning from massive amounts of monitoring data remain pressing technical challenges in long-distance, multi-section tunnel monitoring applications. Summary of the Invention
[0005] To address the aforementioned shortcomings in existing technologies, this invention provides a tunnel structure displacement and deformation measurement system and method based on image recognition, which solves the problems of low measurement efficiency, reliance on manual on-site operation, and inaccurate risk prediction in traditional tunnel deformation monitoring methods.
[0006] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows: On the one hand, the present invention provides a tunnel structure displacement deformation measurement system based on image recognition, including multiple targets, multiple image displacement sensors, a data processing unit and a network transmission module; The plurality of targets include at least one reference target, a plurality of transfer targets and a plurality of monitoring targets. At least one of the reference targets is arranged in the secondary lining stability zone inside the tunnel. One of the transfer targets is arranged on the primary support trolley and the secondary lining trolley respectively. A plurality of the monitoring targets are arranged on the monitoring section behind the excavation face. Multiple image displacement sensors are used to acquire images of the reference target, the transmission target, and the monitoring target; The data processing unit is used to receive image data collected by the image displacement sensor, calculate the absolute displacement based on the image data, and make predictions and early warnings. The network transmission module is used to establish communication links between multiple image displacement sensors and data processing units.
[0007] Furthermore, each of the image displacement sensors is surrounded by a supplementary light. The supplementary light is an infrared supplementary light, used to illuminate each target in order to acquire high-contrast target images in a dimly lit tunnel environment.
[0008] Furthermore, each of the image displacement sensors is mounted within an IP67-rated protective enclosure via a universal joint mounting bracket, and the front of each enclosure is a shock-resistant acrylic plate. The protective enclosures prevent damage to the image displacement sensors from flying debris during tunnel blasts.
[0009] Furthermore, as a specific arrangement of multiple image displacement sensors, the multiple image displacement sensors are respectively a first image displacement sensor, a second image displacement sensor, a third image displacement sensor, and a fourth image displacement sensor; The first image displacement sensor, as a monitoring section device, is placed at the stable location of the channel steel on the right or left side of the initial support platform roof to observe multiple monitoring targets on the monitoring section; the monitoring section is 5m to 10m away from the working face, with one monitoring target arranged at the top of the monitoring section and one monitoring target arranged on each of the two sides of the arch. The second image displacement sensor, as a transfer device for the primary support trolley, is placed on the left side of the primary support trolley channel to observe the transfer target on the primary support trolley and the monitoring target at the arch of the monitoring section; the transfer target on the primary support trolley is placed on the left side of the primary support trolley channel at a distance of about 4.2m from the ground. The third image displacement sensor, as a transfer device for the secondary lining trolley, is placed on the right side of the secondary lining trolley channel to observe the transfer target on the secondary lining trolley and the transfer target on the primary support trolley; the transfer target on the secondary lining trolley is placed on the right side of the secondary lining trolley channel at a height of about 2.2m above the ground. The fourth image displacement sensor, serving as a reference point device, is placed to the left of the stable secondary lining location; the reference target is arranged in the stable concrete area 10m to 20m in front of the fourth image displacement sensor.
[0010] By using a cascaded target design, the rear stability benchmark is accurately transmitted to the front excavation face, enabling absolute displacement measurement in the tunnel's longitudinal direction. The use of a mobile trolley as a transfer platform for deploying instruments and targets cleverly avoids obstructing the monitoring line of sight by the trolley and other equipment, ensuring the continuity of data acquisition.
[0011] On the other hand, the present invention also provides a method for measuring the displacement and deformation of tunnel structures based on image recognition, including: Step 1, constructing a tunnel structure displacement and deformation measurement system based on image recognition: setting up at least one reference target in the secondary lining stable zone inside the tunnel, setting up transfer targets on the primary support trolley and the secondary lining trolley respectively, and setting up monitoring targets on the monitoring section behind the excavation face; correspondingly, setting up multiple image displacement sensors on the top or platform of the primary support trolley, the secondary lining trolley passage and the secondary lining stable zone to collect images of the reference target, transfer target and monitoring target; Step 2, Data Acquisition and Transmission: Each image displacement sensor acquires image data of the corresponding target in real time and transmits it to the data processing unit through a network consisting of 4G / 5G routers, optical fibers, wireless bridges and switches. Step 3, Displacement Calculation and Reference Transfer: The data processing unit processes each image data based on a digital image correlation algorithm, identifies the sub-pixel coordinates of the target center, calculates the two-dimensional displacement of each target relative to its corresponding image displacement sensor, and transfers the absolute coordinate reference to the monitoring target step by step through the displacement data of the reference target and the transfer targets at each level, and calculates the absolute displacement of the monitoring section, including the crown settlement value and the surrounding convergence value. Step 4, Deformation Prediction and Early Warning: Based on the absolute displacement, the data processing unit uses an improved tangent angle algorithm to predict the deformation trend and issues an early warning when the displacement, displacement rate, or predicted value exceeds a preset threshold.
[0012] Furthermore, in step 3, the formula for calculating the absolute displacement is:
[0013] in, For absolute displacement, ( x 1, y 1, z 1) and ( x 2, y 2, z 2) These are the image coordinates of the monitoring target at different times.
[0014] Furthermore, during the entire monitoring process of tunnel displacement, the slope of the cumulative displacement S-time t curve (hereinafter referred to as the St curve) in the time series graph of displacement monitoring data is constantly changing. Based on the changing characteristics of the displacement St curve, tunnel risk warning can be made by analyzing the tangent angle between displacement and time. However, since the dimensions of the vertical and horizontal axes of the tunnel St curve are significantly different, if either the vertical or horizontal axis is stretched or compressed, the St curve can still maintain its changing characteristics, but the displacement tangent angle at the same moment will change due to the stretching or compression transformation. In other words, if the St curve of the same displacement monitoring data is drawn using different coordinate scales, the displacement tangent angle measured at the same moment will be different. That is, directly using the St curve to define the displacement tangent angle will have uncertainty. Therefore, the traditional criterion for tunnel risk warning by analyzing the tangent angle between displacement and time is naturally not rigorous. In order to solve the problem of inconsistent dimensions of the vertical and horizontal axes of the St curve, the St coordinate system can be appropriately transformed to make the dimensions of its vertical and horizontal axes consistent. Specifically, in step 4, the improved tangent angle algorithm includes: Step 41: Accumulate displacement S -time t Perform coordinate transformation on the curve and define new coordinates. New coordinates The expression is:
[0015] in, For the first i The change in displacement within a unit time period v The displacement rate is represented by the constant velocity deformation stage; by transforming the ordinate of the St curve to the same time dimension as the abscissa, a new Tt curve is obtained. Step 42: Calculate the first step based on the new Tt curve. i Improved tangent angle at time : ; in, For improved tangent angle; For a specific monitoring time; For calculation The corresponding unit of time; Within a unit of time period The change in; Step 4.3, based on the improved tangent angle The magnitude and trend of displacement are used to determine the deformation stage of the displacement, and different prediction algorithms are dynamically selected to fit the displacement data and predict future displacement.
[0016] The proposed improved tangent angle algorithm eliminates the influence of dimensions through mathematical transformation, enabling the prediction model to adapt to deformation stages and data quality, accurately predicting the final deformation value of the cross-section, and providing a scientific basis for construction. It also improves the accuracy and efficiency of data processing, and provides a scientific basis for construction safety and optimized design.
[0017] Furthermore, the dynamic selection of different prediction algorithms in step S43 specifically involves: When improving the tangent angle If the displacement remains stable and small over a period of time, it is determined to be in the uniform deformation stage. A prediction algorithm based on tangent angle increment is used, and the calculation formula of the prediction algorithm is as follows: ; in, The predicted value at time t; express( t -1), ( t -2), ..., ( t - n The monitoring value at that time. n It is a sliding window; When improving the tangent angle When the displacement shows a continuous increasing trend, it is determined that the displacement is in the accelerated deformation stage, and an autoregressive model is used for prediction. When monitoring data is continuously missing for more than the set duration of 5 days, an autoregressive moving average model is used for data compensation and prediction; the expression for the autoregressive moving average model is: ; in, for t The predicted value at any given time; The change in displacement within a unit of time period; for t -1 represents the improved tangent angle within a unit time interval, where v is the displacement rate during the constant velocity deformation stage. For calculation The corresponding unit time period. A method that flexibly switches between multiple different data prediction algorithms to predict monitoring data can improve the low fitting degree of traditional single prediction algorithms, thereby enhancing the accuracy and reliability of tunnel deformation risk prediction.
[0018] Furthermore, in step 3, the method of progressively transferring the absolute coordinate reference to the monitoring target through the reference target and various levels of transfer targets includes: Step A: Set the reference target placed in the secondary lining stability zone as the absolute reference point, with zero displacement; Step B: Let the sensor observing the reference target be the fourth sensor, and the sensor observing the target on the secondary lining trolley be the third sensor; by comparing the displacement of the target observed by the fourth sensor with the displacement of the same target observed by the third sensor, calculate the absolute displacement offset of the third sensor itself. Step C: Following this logic, the absolute displacement reference is passed forward to the second sensor on the observation support trolley, and finally to the first sensor on the observation monitoring section target, thereby calculating the absolute displacement of the monitoring section target relative to the absolute reference point.
[0019] Image recognition-based tunnel structure displacement and deformation measurement reduces interference from the construction environment on data acquisition, ensures data continuity and quality, and achieves high-precision, high-efficiency, and stable and reliable monitoring, thus providing a guarantee for the safe operation of tunnels.
[0020] Furthermore, in step 3, the initial in-situ stress lateral pressure coefficient of the tunnel surrounding rock can be inferred from the absolute displacement. λ Equivalent elastic modulus of rock mass E The calculation formula is as follows: λ = f ( , E ); in, This is the absolute displacement. f This is the inverse analysis function. In the current analysis of tunnel surrounding rock stability, the two most important and influential design parameters are the initial in-situ stress lateral pressure coefficient. λ Equivalent elastic modulus of rock mass E The lateral pressure coefficient λ and the rock mass elastic modulus E are calculated by absolute displacement, and compared with the analysis of design and other test stresses and internal forces to determine the current stability of the surrounding rock of the tunnel.
[0021] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention discloses an image recognition-based system and method for measuring displacement and deformation of tunnel structures. Within the tunnel, benchmark targets, multi-level transfer targets, and monitoring section targets are deployed from the secondary lining stability zone to the excavation face, along with corresponding image displacement sensors. The sensors acquire images of each target, calculate the two-dimensional displacement of the target using a digital image correlation algorithm, and utilize the cascade transfer principle to progressively transfer the absolute coordinates of the benchmark points to the monitoring sections, thereby achieving absolute measurement of tunnel arch settlement and perimeter convergence. For displacement prediction, this invention proposes an improved tangent angle algorithm based on coordinate transformation and, based on the improved tangent angle… The invention features a flexible prediction model that can be switched between different sizes, effectively improving the accuracy and reliability of early warnings when data is missing or fluctuating. It solves the technical challenges of field-of-view obstruction, difficulties in benchmark transfer, and the inaccuracies of traditional prediction methods in tunnel construction environments. Attached Figure Description
[0022] Figure 1 This is a schematic diagram of the on-site layout of an image recognition-based tunnel structure displacement and deformation measurement system.
[0023] Figure 2 This is a schematic diagram of the monitoring section.
[0024] Figure 3 This is a schematic diagram of the field of view of the first image displacement sensor.
[0025] Figure 4 This is a schematic diagram of the field of view of the second image displacement sensor.
[0026] Figure 5 This is a schematic diagram of the field of view of the third image displacement sensor.
[0027] Figure 6 This is a schematic diagram of the field of view of the fourth image displacement sensor.
[0028] Figure 7 This is a flowchart of a method for measuring displacement and deformation of tunnel structures based on image recognition.
[0029] Among them, 1. reference target; 2. transfer target; 3. monitoring target; 4. primary support trolley; 5. secondary lining trolley; 6. first image displacement sensor; 7. second image displacement sensor; 8. third image displacement sensor; 9. fourth image displacement sensor; 10. monitoring section; 11. secondary lining stable zone. Detailed Implementation
[0030] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.
[0031] refer to Figure 1 , Figure 1This is a schematic diagram of the on-site layout of an image recognition-based tunnel structure displacement and deformation measurement system. The system includes multiple targets, multiple image displacement sensors, a data processing unit, and a network transmission module. The multiple targets include at least one reference target 1, multiple transfer targets 2, and multiple monitoring targets 3. At least one reference target 1 is arranged in the secondary lining stability zone 11 within the tunnel. One transfer target 2 is arranged on the initial support trolley 4 and the secondary lining trolley 5, respectively. Multiple monitoring targets 3 are arranged on the monitoring section 10 behind the excavation face. The multiple image displacement sensors are used to acquire images of the reference target 1, transfer targets 2, and monitoring targets 3. The data processing unit is used to receive the image data acquired by the image displacement sensors, calculate the absolute displacement based on the image data, and perform prediction and early warning. The network transmission module is used to establish a communication link between the multiple image displacement sensors and the data processing unit.
[0032] Preferably, but not exclusively, each of the image displacement sensors is surrounded by a supplementary light. The supplementary light is an infrared supplementary light used to illuminate each target, independent of external lighting conditions, to acquire high-contrast target images in dim tunnel environments. Specifically, the surface of each target is specially treated to have high reflectivity to infrared light, for example. The concentrated illumination of the supplementary lights greatly improves the brightness and contrast of the target in the image, making it significantly distinguishable from the dark, complex tunnel background. This effectively reduces image noise, providing a high-quality image foundation for subsequent digital image correlation algorithms to achieve sub-pixel-level high-precision positioning.
[0033] Preferably, each image displacement sensor is mounted within an IP67-rated protective cover via a universal joint mounting bracket, and the front of each protective cover is an impact-resistant acrylic plate. The protective cover prevents flying rocks from blasting inside the tunnel from damaging the image displacement sensor.
[0034] Specifically, such as Figures 1-6 As shown, as a specific arrangement of multiple image displacement sensors, the multiple image displacement sensors are respectively a first image displacement sensor 6, a second image displacement sensor 7, a third image displacement sensor 8, and a fourth image displacement sensor 9.
[0035] The first image displacement sensor 6 serves as the monitoring section 10 device and is placed on the right or left side of the channel steel of the top platform of the initial support trolley 4. The first image displacement sensor 6 uses a 25mm lens to observe all monitoring section 10 targets within a range of 30-80m. The monitoring section 10 is 5m to 10m away from the working face. One monitoring target 3 is arranged on the arch top of the monitoring section 10, and one monitoring target 3 is arranged on each of the two sides of the arch waist.
[0036] The second image displacement sensor 7 serves as a transfer device for the primary support trolley 4 and is placed on the left side of the primary support trolley 4 channel. The second image displacement sensor 7 uses a 35mm lens to observe the transfer target 2 on the primary support trolley 4 and the monitoring target 3 at the arch of the monitoring section 10. The transfer target 2 on the primary support trolley 4 is placed on the left side of the primary support trolley 4 channel at a distance of about 4.2m from the ground.
[0037] The third image displacement sensor 8, as a transfer device for the secondary lining trolley 5, is placed on the right side of the secondary lining trolley 5 channel. The third image displacement sensor 8 uses a 12mm lens to observe the transfer target 2 on the secondary lining trolley 5 and the transfer target 2 on the primary support trolley 4. The transfer target 2 on the secondary lining trolley 5 is placed on the right side of the secondary lining trolley 5 channel at a distance of about 2.2m from the ground.
[0038] The fourth image displacement sensor 9 serves as a reference point device and is placed on the left side of the stable secondary lining location. The third image displacement sensor 8 uses a 25mm lens to observe the reference target 1 and the transfer target 2 on the secondary lining trolley 5. The reference target 1 is arranged in the stable concrete area 10m to 20m in front of the fourth image displacement sensor 9.
[0039] By using a cascaded target design, the rear stability benchmark is accurately transmitted to the front excavation face, enabling absolute displacement measurement in the tunnel's longitudinal direction. The use of a mobile trolley as a transfer platform for deploying instruments and targets cleverly avoids obstructing the monitoring line of sight by the trolley and other equipment, ensuring the continuity of data acquisition.
[0040] like Figure 1 and Figure 7 As shown, on the other hand, the present invention also provides a method for measuring the displacement and deformation of tunnel structures based on image recognition, including: Step 1, constructing a tunnel structure displacement and deformation measurement system based on image recognition: at least one reference target 1 is set up in the secondary lining stable zone 11 in the tunnel, transfer targets 2 are set up on the primary support trolley 4 and the secondary lining trolley 5 respectively, and monitoring targets 3 are set up on the monitoring section 10 behind the excavation face; correspondingly, multiple image displacement sensors are set up on the top or platform of the primary support trolley 4, the channel of the secondary lining trolley 5 and the secondary lining stable zone 11 to collect images of the reference target 1, transfer targets 2 and monitoring targets 3; Step 2, Data Acquisition and Transmission: Each image displacement sensor acquires image data of the corresponding target in real time and transmits it to the data processing unit through a network consisting of 4G / 5G routers, optical fibers, wireless bridges and switches. Step 3, Displacement Calculation and Reference Transfer: The data processing unit processes each image data based on a digital image correlation algorithm, identifies the sub-pixel coordinates of the target center, calculates the two-dimensional displacement of each target relative to its corresponding image displacement sensor, and transfers the absolute coordinate reference to the monitoring target 3 step by step through the displacement data of the reference target 1 and the various levels of transfer targets 2, and calculates the absolute displacement of the monitoring section 10, including the crown settlement value and the surrounding convergence value. Step 4, Deformation Prediction and Early Warning: Based on the absolute displacement, the data processing unit uses an improved tangent angle algorithm to predict the deformation trend and issues an early warning when the displacement, displacement rate, or predicted value exceeds a preset threshold.
[0041] Furthermore, in step 3, the formula for calculating the absolute displacement is:
[0042] in, For absolute displacement, ( x 1, y 1, z 1) and ( x 2, y 2, z 2) These are the image coordinates of the monitoring target at different times.
[0043] Furthermore, during the entire monitoring process of tunnel displacement, the slope of the cumulative displacement S-time t curve (hereinafter referred to as the St curve) in the time series graph of displacement monitoring data is constantly changing. Based on the changing characteristics of the displacement St curve, tunnel risk warning can be made by analyzing the tangent angle between displacement and time. However, since the dimensions of the vertical and horizontal axes of the tunnel St curve are significantly different, if either the vertical or horizontal axis is stretched or compressed, the St curve can still maintain its changing characteristics, but the displacement tangent angle at the same moment will change due to the stretching or compression transformation. In other words, if the St curve of the same displacement monitoring data is drawn using different coordinate scales, the displacement tangent angle measured at the same moment will be different. That is, directly using the St curve to define the displacement tangent angle will have uncertainty. Therefore, the traditional criterion for tunnel risk warning by analyzing the tangent angle between displacement and time is naturally not rigorous. In order to solve the problem of inconsistent dimensions of the vertical and horizontal axes of the St curve, the St coordinate system can be appropriately transformed to make the dimensions of its vertical and horizontal axes consistent. Specifically, in step 4, the improved tangent angle algorithm includes: Step 41: Accumulate displacement S -time t Perform coordinate transformation on the curve and define new coordinates. New coordinates The expression is:
[0044] in, For the first i The change in displacement within a unit time period v The displacement rate is represented by the constant velocity deformation stage; by transforming the ordinate of the St curve to the same time dimension as the abscissa, a new Tt curve is obtained. Step 42: Calculate the first step based on the new Tt curve. i Improved tangent angle at time : ; in, For improved tangent angle; For a specific monitoring time; For calculation The corresponding unit of time; Within a unit of time period The change in; Step 4.3, based on the improved tangent angle The magnitude and trend of displacement are used to determine the deformation stage of the displacement, and different prediction algorithms are dynamically selected to fit the displacement data and predict future displacement.
[0045] The proposed improved tangent angle algorithm eliminates the influence of dimensions through mathematical transformation, enabling the prediction model to adapt to deformation stages and data quality, accurately predicting the final deformation value of the cross-section, and providing a scientific basis for construction. It also improves the accuracy and efficiency of data processing, and provides a scientific basis for construction safety and optimized design.
[0046] Furthermore, the dynamic selection of different prediction algorithms in step S43 specifically involves: When improving the tangent angle If the displacement remains stable and small over a period of time, it is determined to be in the uniform deformation stage. A prediction algorithm based on tangent angle increment is used, and the calculation formula of the prediction algorithm is as follows: ; in, The predicted value at time t; express( t -1), ( t -2), ..., ( t - n The monitoring value at that time. n It is a sliding window; When improving the tangent angle When the displacement shows a continuous increasing trend, it is determined that the displacement is in the accelerated deformation stage, and an autoregressive model is used for prediction. When monitoring data is continuously missing for more than the set duration of 5 days, an autoregressive moving average model is used for data compensation and prediction; the expression for the autoregressive moving average model is: ; in, for t The predicted value at any given time; The change in displacement within a unit of time period; for t -1 represents the improved tangent angle within a unit time interval, where v is the displacement rate during the constant velocity deformation stage. For calculation The corresponding unit time period. A method that flexibly switches between multiple different data prediction algorithms to predict monitoring data can improve the low fitting degree of traditional single prediction algorithms, thereby enhancing the accuracy and reliability of tunnel deformation risk prediction.
[0047] Furthermore, in step 3, the method of progressively transferring the absolute coordinate reference to the monitoring target 3 through the reference target 1 and the various levels of transfer targets 2 includes: Step A: Set the reference target 1 placed in the secondary lining stability zone 11 as the absolute reference point, and its displacement is zero; Step B: Let the sensor for observing the reference target 1 be the fourth sensor, and the sensor for observing the transfer target 2 on the secondary lining trolley 5 be the third sensor; by comparing the displacement of the transfer target 2 observed by the fourth sensor with the displacement of the same transfer target 2 observed by the third sensor, calculate the absolute displacement offset of the third sensor itself. Step C: Following this logic, the absolute displacement reference is passed forward to the second sensor of the target 2 on the observation support trolley 4, and finally to the first sensor of the target of the observation monitoring section 10, thereby calculating the absolute displacement of the target of the monitoring section 10 relative to the absolute reference point.
[0048] Image recognition-based tunnel structure displacement and deformation measurement reduces interference from the construction environment on data acquisition, ensures data continuity and quality, and achieves high-precision, high-efficiency, and stable and reliable monitoring, thus providing a guarantee for the safe operation of tunnels.
[0049] Furthermore, in step 3, the initial in-situ stress lateral pressure coefficient of the tunnel surrounding rock can be inferred from the absolute displacement. λ Equivalent elastic modulus of rock mass E The calculation formula is as follows: λ = f ( , E ); in, This is the absolute displacement.f This is the inverse analysis function. In the current analysis of tunnel surrounding rock stability, the two most important and influential design parameters are the initial in-situ stress lateral pressure coefficient. λ Equivalent elastic modulus of rock mass E The lateral pressure coefficient λ and the rock mass elastic modulus E are calculated by absolute displacement, and compared with the analysis of design and other test stresses and internal forces to determine the current stability of the surrounding rock of the tunnel.
[0050] In summary, an image recognition-based system and method for measuring displacement and deformation of tunnel structures solves the technical challenges of field-of-view obstruction, difficulty in reference transfer, and inaccuracy of traditional prediction methods in tunnel construction environments.
Claims
1. A displacement and deformation measurement system for tunnel structures based on image recognition, characterized in that, It includes multiple targets, multiple image displacement sensors, a data processing unit, and a network transmission module; The plurality of targets include at least one reference target, a plurality of transfer targets and a plurality of monitoring targets. At least one of the reference targets is arranged in the secondary lining stability zone inside the tunnel. One of the transfer targets is arranged on the primary support trolley and the secondary lining trolley respectively. A plurality of the monitoring targets are arranged on the monitoring section behind the excavation face. Multiple image displacement sensors are used to acquire images of the reference target, the transmission target, and the monitoring target; The data processing unit is used to receive image data collected by the image displacement sensor, calculate the absolute displacement based on the image data, and make predictions and early warnings. The network transmission module is used to establish communication links between multiple image displacement sensors and data processing units.
2. The tunnel structure displacement and deformation measurement system based on image recognition according to claim 1, characterized in that, Each of the image displacement sensors is matched with a supplementary light.
3. The tunnel structure displacement and deformation measurement system based on image recognition according to claim 2, characterized in that, Each of the image displacement sensors is mounted in an IP67-rated protective cover via a universal joint mounting bracket, and the front of each protective cover is an impact-resistant acrylic plate.
4. The tunnel structure displacement and deformation measurement system based on image recognition according to claim 3, characterized in that, The plurality of image displacement sensors are a first image displacement sensor, a second image displacement sensor, a third image displacement sensor, and a fourth image displacement sensor; The first image displacement sensor, as a monitoring section device, is placed at the stable location of the channel steel on the right or left side of the initial support platform roof to observe multiple monitoring targets on the monitoring section; the monitoring section is 5m to 10m away from the working face, with one monitoring target arranged at the top of the monitoring section and one monitoring target arranged on each of the two sides of the arch. The second image displacement sensor, as a transfer device for the primary support trolley, is placed on the left side of the primary support trolley channel to observe the transfer target on the primary support trolley and the monitoring target at the arch of the monitoring section; the transfer target on the primary support trolley is placed on the left side of the primary support trolley channel at a distance of about 4.2m from the ground. The third image displacement sensor, as a transfer device for the secondary lining trolley, is placed on the right side of the secondary lining trolley channel to observe the transfer target on the secondary lining trolley and the transfer target on the primary support trolley; the transfer target on the secondary lining trolley is placed on the right side of the secondary lining trolley channel at a height of about 2.2m above the ground. The fourth image displacement sensor, serving as a reference point device, is placed to the left of the stable secondary lining location; the reference target is arranged in the stable concrete area 10m to 20m in front of the fourth image displacement sensor.
5. A measurement method for a tunnel structure displacement and deformation measurement system based on image recognition according to any one of claims 1 to 4, characterized in that, The measurement method includes: Step 1: Construct a tunnel structure displacement and deformation measurement system based on image recognition: At least one reference target is set up in the secondary lining stable zone inside the tunnel, transfer targets are set up on the primary support trolley and the secondary lining trolley respectively, and monitoring targets are set up at the monitoring section behind the excavation face; Correspondingly, multiple image displacement sensors are set up on the top or platform of the primary support trolley, the secondary lining trolley passage and the secondary lining stable zone to collect images of the reference target, transfer target and monitoring target; Step 2, Data Acquisition and Transmission: Each image displacement sensor acquires image data of the corresponding target in real time and transmits it to the data processing unit through a network consisting of 4G / 5G routers, optical fibers, wireless bridges and switches. Step 3, Displacement Calculation and Reference Transfer: The data processing unit processes each image data based on a digital image correlation algorithm, identifies the sub-pixel coordinates of the target center, calculates the two-dimensional displacement of each target relative to its corresponding image displacement sensor, and transfers the absolute coordinate reference to the monitoring target step by step through the displacement data of the reference target and the transfer targets at each level, and calculates the absolute displacement of the monitoring section, including the crown settlement value and the surrounding convergence value. Step 4, Deformation Prediction and Early Warning: Based on the absolute displacement, the data processing unit uses an improved tangent angle algorithm to predict the deformation trend and issues an early warning when the displacement, displacement rate, or predicted value exceeds a preset threshold.
6. The measurement method of the tunnel structure displacement and deformation measurement system based on image recognition according to claim 5, characterized in that, In step 3, the formula for calculating the absolute displacement is: in, For absolute displacement, ( x 1, y 1, z 1) and ( x 2, y 2, z 2) These are the image coordinates of the monitoring target at different times.
7. The measurement method of the tunnel structure displacement and deformation measurement system based on image recognition according to claim 5, characterized in that, In step 4, the improved tangent angle algorithm includes: Step 41: Accumulate displacement S -time t Perform coordinate transformation on the curve and define new coordinates. New coordinates The expression is: in, For the first i The change in displacement within a unit time period v The displacement rate is represented by the constant velocity deformation stage; by transforming the ordinate of the St curve to the same time dimension as the abscissa, a new Tt curve is obtained. Step 42: Calculate the first step based on the new Tt curve. i Improved tangent angle at time : ； in, For improved tangent angle; For a specific monitoring time; For calculation The corresponding unit of time; Within a unit of time period The change in; Step 4.3, based on the improved tangent angle The magnitude and trend of displacement are used to determine the deformation stage of the displacement, and different prediction algorithms are dynamically selected to fit the displacement data and predict future displacement.
8. The measurement method of the tunnel structure displacement and deformation measurement system based on image recognition according to claim 7, characterized in that, The dynamic selection of different prediction algorithms in step S43 specifically refers to: When improving the tangent angle If the displacement remains stable and small over a period of time, it is determined to be in the uniform deformation stage. A prediction algorithm based on tangent angle increment is used, and the calculation formula of the prediction algorithm is as follows: ； in, The predicted value at time t; express( t -1), ( t -2), ..., ( t - n The monitoring value at that time. n It is a sliding window; When improving the tangent angle When the displacement shows a continuous increasing trend, it is determined that the displacement is in the accelerated deformation stage, and an autoregressive model is used for prediction. When monitoring data is continuously missing for more than the set duration of 5 days, an autoregressive moving average model is used for data compensation and prediction; the expression for the autoregressive moving average model is: ； in, for t The predicted value at any given time; The change in displacement within a unit of time period; for t -1 represents the improved tangent angle within a unit time interval, where v is the displacement rate during the constant velocity deformation stage. For calculation The corresponding unit of time period.
9. The measurement method of the tunnel structure displacement and deformation measurement system based on image recognition according to claim 6, characterized in that, In step 3, the method of progressively transferring the absolute coordinate reference to the monitoring target through the reference target and various levels of transfer targets includes: Step A: Set the reference target placed in the secondary lining stability zone as the absolute reference point, with zero displacement; Step B: Let the sensor observing the reference target be the fourth sensor, and the sensor observing the target on the secondary lining trolley be the third sensor; by comparing the displacement of the target observed by the fourth sensor with the displacement of the same target observed by the third sensor, calculate the absolute displacement offset of the third sensor itself. Step C: Following this logic, the absolute displacement reference is passed forward to the second sensor on the observation support trolley, and finally to the first sensor on the observation monitoring section target, thereby calculating the absolute displacement of the monitoring section target relative to the absolute reference point.
10. The measurement method of the tunnel structure displacement and deformation measurement system based on image recognition according to claim 9, characterized in that, In step 3, the initial geostress lateral pressure coefficient of the tunnel surrounding rock can also be inferred from the absolute displacement. λ Equivalent elastic modulus of rock mass E The calculation formula is as follows: λ = f （ ， E ）； in, This is the absolute displacement. f This is an inverse analysis function.