Method and device for monitoring initial tunnel vault settlement based on machine vision calibration

Abstract

The present invention provides a method and device for monitoring initial tunnel vault settlement based on machine vision calibration, relating to the technical field of tunnel engineering monitoring. This method utilizes an adaptive ambient light compensation algorithm and an infrared-assisted calibration device to obtain real-time settlement data from a tunnel vault target in the excavated area behind the tunnel face. This settlement is used as a benchmark correction value and input into a displacement rope data correction model. The displacement rope system, deployed within a pre-drilled borehole, enables continuous monitoring of tunnel vault settlement. A multi-sensor data spatiotemporal registration module ensures accurate fusion of data from the visual displacement system and the displacement rope system, thereby outputting corrected settlement measurements for the entire excavation section ahead of the tunnel face. This method not only improves measurement accuracy and efficiency, but also, through the benchmark correction algorithm, eliminates drift errors in the displacement rope system's benchmark points caused by stratum disturbances, enhancing the adaptability and reliability of the monitoring system.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel engineering monitoring, and in particular to a method and device for monitoring initial vault settlement of a tunnel based on machine vision calibration. Background Art

[0002] Currently, monitoring vault settlement is crucial during tunnel construction, directly impacting construction safety and project quality. However, existing monitoring technologies have numerous shortcomings. Traditional total station monitoring relies on manual operation, resulting in low efficiency. Most critically, monitoring points can only be deployed after tunnel excavation, typically taking 1-3 days after initial support is established. Consequently, they are unable to capture settlement changes from the initial tunnel state to the excavation phase, making them inadequate for dynamic construction environments. While machine vision-based monitoring solutions offer certain automation advantages, they are also limited by the requirement to deploy targets only after excavation. Furthermore, recognition accuracy significantly decreases in complex environments such as construction dust, making it difficult to ensure accurate monitoring data. To address the measurement of vault settlement from the initial tunnel state to excavation, Chinese Patent Application No. 201910257595.1 discloses a device and method for monitoring vault settlement during tunnel construction. This method uses inclinometer sensors embedded in pre-drilled holes above the tunnel face to calculate the total displacement of the vault settlement. Although the scheme of pre-embedded displacement meters in advance drilling can monitor the settlement of the arch crown in real time throughout the entire process before and after the tunnel face excavation, this scheme requires a fixed point in the stratum as the starting reference point. The stratum disturbance within the influence range of the tunnel face excavation will cause the reference point to drift, and the displacement of the reference point in the stratum cannot be verified and corrected through measurement, which in turn affects the reliability of the measurement results of this scheme.

[0003] More importantly, there is no existing technology that integrates the machine vision displacement system with the physical sensor system for monitoring vault settlement, uses machine vision dynamic calibration as the basis for correcting the reference points of the physical sensor for monitoring vault settlement, and effectively solves the time and space reference synchronization problem between the visual displacement system and the physical sensor, thereby effectively fusing multi-source data to accurately and dynamically measure the vault settlement of the entire tunnel excavation process.

[0004] Therefore, developing a new technology that can realize accurate dynamic measurement of vault settlement from the initial unexcavated state of the tunnel to the entire excavation construction process is of great practical significance for improving the level of tunnel engineering monitoring.

[0005] In view of this, this application is filed. Summary of the Invention

[0006] The present invention provides a method and device for monitoring initial vault settlement of a tunnel based on machine vision calibration, which can at least partially improve the above-mentioned problems.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] A method for monitoring initial tunnel vault settlement based on machine vision calibration, comprising:

[0009] Acquire real-time dynamic visual displacement data collected by a preset visual displacement system, and obtain tunnel vault settlement using displacement rope system data collected by a pre-buried displacement rope system in the tunnel, and pre-process the visual displacement data and displacement rope system data to obtain corrected visual displacement data and vault settlement;

[0010] Performing data fusion and spatiotemporal registration processing based on the corrected visual displacement data and the vault settlement, so that the visual displacement data and the displacement rope system data are aligned in space and synchronized in time;

[0011] The visual displacement data is input into the displacement rope system data correction model as the reference point correction value, and the displacement rope system data correction model is used to dynamically correct the aligned displacement rope system data to obtain the settlement measurement result of the excavation section in front of the tunnel face.

[0012] The present invention also provides a tunnel initial vault settlement monitoring device based on machine vision calibration, which includes:

[0013] a preprocessing unit for acquiring visual displacement data collected by a preset visual displacement system and displacement rope system data collected by a displacement rope system, and preprocessing the visual displacement data and displacement rope system data to obtain corrected visual displacement data and vault settlement;

[0014] an alignment unit for performing data fusion and spatiotemporal registration processing based on the corrected visual displacement data and the vault settlement, so that the visual displacement data and the displacement rope system data are spatially aligned and temporally synchronized;

[0015] The correction unit is used to input the visual displacement data as a reference point correction amount into the displacement rope data correction model, and use the displacement rope data correction model to correct the aligned displacement rope system data to obtain the settlement measurement result of the excavation section in front of the tunnel face.

[0016] In summary, the proposed method for monitoring initial tunnel vault settlement based on machine vision calibration addresses the drift of pre-drilled displacement meter reference points due to ground disturbance during tunnel excavation by integrating machine vision with displacement meter data correction technology and spatiotemporal registration of multi-source monitoring data. This method enables high-precision dynamic monitoring from the initial tunnel state to the start of construction. Specifically, this method utilizes machine vision dynamic calibration as its core, combined with extended monitoring of displacement meters, to construct a novel two-layer monitoring architecture. The machine vision system acquires tunnel vault settlement data in real time through ambient light compensation and infrared-assisted calibration, using this data as a reference point correction to correct the displacement meter data. Spatiotemporal registration of multi-source data resolves the issue of temporal and spatial reference synchronization between the vision system and the displacement meter, ensuring the accuracy and consistency of the monitoring data. Furthermore, this method incorporates a reference point correction algorithm that, through a dynamic adjustment mechanism, assesses the reliability of the visual data in real time and dynamically adjusts and corrects the displacement meter data, effectively eliminating the impact of ground disturbances on the reference points. The machine vision calibration-based tunnel initial vault settlement monitoring method not only improves monitoring accuracy, but also enhances adaptability and reliability. It is suitable for tunnel vault settlement monitoring in complex construction environments and provides a strong technical guarantee for the safe construction of tunnel projects. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 1 is a flow chart of a method for monitoring initial vault settlement of a tunnel based on machine vision calibration provided by the first embodiment of the present invention;

[0018] Figure 2 This is a flowchart of the overall process of the method for monitoring initial tunnel vault settlement based on machine vision calibration provided by the first embodiment of the present invention;

[0019] Figure 3 Schematic diagram of the structure of the visual displacement system and the displacement rope system provided by an embodiment of the present invention;

[0020] Figure 4 It is a module schematic diagram of a tunnel initial vault settlement monitoring device based on machine vision calibration provided by the second embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0022] refer to Figures 1 to 3As shown, the first embodiment of the present invention discloses a method for monitoring initial tunnel vault settlement based on machine vision calibration. The method can be performed by a tunnel initial vault settlement monitoring device based on machine vision calibration (hereinafter referred to as the monitoring device), and in particular, by one or more processors in the monitoring device to implement the following method:

[0023] S1, obtaining real-time dynamic visual displacement data collected by a preset visual displacement system and displacement rope system data collected by a displacement rope system pre-buried in the tunnel to obtain tunnel vault settlement, and pre-processing the visual displacement data and displacement rope system data to obtain corrected visual displacement data and vault settlement;

[0024] Preferably, the visual displacement system is a visual displacement camera, which includes an infrared anti-interference module. The visual displacement camera is arranged in a stable area within a preset length range behind the tunnel face to be monitored where tunnel support has been formed and no displacement occurs. Its corresponding target is arranged at the arch position within the excavated area behind the tunnel face to be monitored, wherein the preset length range is 20~50m.

[0025] The displacement rope system is a displacement rope fixed on a steel pipe. The displacement rope is pre-buried in the stratum above the tunnel by advance drilling. The displacement rope is composed of a gradient distribution of node sensors with equal or unequal spacing, wherein the position of the target coincides with the position of the drilling hole of the displacement rope system.

[0026] Specifically, step S1 further includes: determining whether the current ambient light intensity is less than a preset value;

[0027] If so, activate the infrared anti-interference module, perform infrared fill light, and perform adaptive environmental compensation processing on the collected visual displacement data to obtain corrected visual data;

[0028] If not, the collected visual displacement data is directly subjected to adaptive environmental compensation processing to obtain corrected visual data.

[0029] Based on the nonlinear compensation model, the corrected visual displacement data is calculated using the formula: ,in, is the camera grayscale value, is the ambient light intensity, is the dark current reference value, and are all fitting parameters.

[0030] Obtain the inclination change data at the location collected by the node sensor, use the preset reference point as the starting point, perform trapezoidal integration processing, and calculate the arch settlement at any location of the tunnel to be monitored. The calculation formula is: ,in, is the settlement of the vault at point i, is the i-th original tilt change data, is the i-1th original tilt change data, is the arrangement spacing of node sensors.

[0031] In this embodiment, the monitoring equipment must be properly arranged. Figure 3 As can be seen, the visual displacement camera is installed in an area of ​​existing tunnel support 20 to 50 meters behind the tunnel face to be monitored. This location was chosen based on a comprehensive consideration of monitoring range and data accuracy. The camera's target is precisely positioned at the arch of the excavated area behind the tunnel face, coinciding with the borehole of the displacement rope system, ensuring spatial consistency of the data. The displacement rope system consists of a displacement rope fixed to a steel pipe and pre-buried into the stratum above the tunnel through advance drilling. The node sensors on the displacement rope are distributed in a gradient with equal or unequal spacing. This distribution can be flexibly adjusted according to the specific geological conditions and monitoring requirements of the tunnel to achieve more accurate settlement monitoring. The spacing between the equally spaced node sensors can be 0.5m.

[0032] During monitoring, the visual displacement camera first collects visual displacement data and then determines in real time whether the current ambient light intensity is below a preset value (e.g., 50 lux). If the ambient light intensity is insufficient, the infrared anti-interference module is activated to provide infrared fill light to ensure image clarity and data accuracy. Subsequently, the collected visual displacement data undergoes adaptive environmental compensation. This process, based on a nonlinear compensation model, effectively eliminates the impact of ambient light variations on monitoring results by calculating corrected visual data, thereby improving data reliability. This compensation process enables the visual displacement camera to consistently output high-quality monitoring data even in complex construction environments, where interference such as dust and water mist may be present.

[0033] At the same time, the displacement rope system also operates synchronously; node sensors collect real-time data on inclination changes at their locations. Using a preset reference point as the starting point, the system uses a trapezoidal integration method to calculate the vault settlement at any location in the monitored tunnel. This accurately reflects the settlement of the tunnel vault at different locations, providing real-time data support for construction safety.

[0034] S2, performing data fusion and spatiotemporal registration processing based on the corrected visual displacement data and the vault settlement, so that the visual displacement data and the displacement rope system data are spatially aligned and temporally synchronized;

[0035] Specifically, step S2 further includes: performing Euler rotation matrix transformation processing on the corrected visual displacement data so that the coordinates of the corrected visual data are consistent with the coordinates of the vault settlement;

[0036] The data sampling frequency of the displacement rope system is adjusted according to the change of the vault settlement. When the settlement change of the vault settlement is within the preset normal state, the data sampling frequency of the displacement rope system is the initial setting value. When the settlement change of the vault settlement exceeds the preset warning value, the acquisition frequency will be automatically adjusted dynamically and the acquisition frequency will be encrypted. The displacement rope system will collect the data of the displacement rope system according to the real-time dynamic acquisition frequency. Perform cubic spline interpolation to match the timestamp of the visual displacement system ;

[0037] Set the timeline node sequence to , define the jth cubic B-spline basis function The recursive formula is: , , where k=3 represents cubic spline, and the final basis function is obtained ,in, It is the 0th order (constant) B-spline basis function, which serves as the starting point of recursion and defines the basic time unit of data sampling. It is the basis for constructing all higher-order basis functions. is the j-th k-order B-spline basis function, used to construct the cubic spline interpolation curve, is the j-th k-1 order B-spline basis function, is the j+1th k-1th order B-spline basis function; here, in order to construct a cubic spline function, k=3 represents a cubic B-spline, and the final basis function is obtained ,The cubic B-spline basis function is composed of a weighted combination of two adjacent quadratic basis functions.,The high-order smooth basis function is gradually constructed by recursively combining low-order basis functions,,and finally a cubic B-spline curve is generated;

[0038] Determine the time non-uniformity point allocation rule that adapts to the excavation process and dynamically adjust the data collection time interval. The allocation rule is: ,in, The time interval for dynamic data collection of the displacement rope system;

[0039] Based on the final basis function and the distribution rule, the data point sequence of the displacement rope system , construct the difference function: , where m is the number of original data points of the displacement rope system, For the interpolation coefficient to be determined, expand the two control points to meet the boundary conditions;

[0040] The coefficients are solved according to the difference function, and the formula is: , and marked as Ac=S, where B m+2 (t m ) is the m+2th cubic B-spline basis function at time node t m The value at is the interpolation coefficient of the m+2th basis function to be determined, is the displacement rope system at time stamp t m The actual measured settlement data, t m is the time label of the mth data point collected by the displacement rope system; A is the value of the basis function at the time node B1(t1), B2(t1)...B m+2 (t m ), c is a coefficient matrix composed of c1, c2...c m+2 The coefficient vector to be solved is composed of S, which is the settlement data actually collected by the displacement rope system. ... The vector formed;

[0041] According to the constraints , and obtain the constrained least squares solution , , where the constraint matrix C corresponds to the second-order derivative condition and is solved using the Lagrange multiplier method;

[0042] Based on the constrained least squares solution, high-frequency data synchronization processing is performed to time stamp the visual displacement system. , and the corresponding synchronized displacement rope system data is: ,in, is the interpolation coefficient of the jth basis function to be determined.

[0043] In this embodiment, the purpose of performing data fusion spatiotemporal registration processing is to ensure that the visual displacement data and the displacement rope system data are aligned in space and synchronized in time, thereby providing an accurate and reliable data basis for subsequent settlement analysis. First, the Euler rotation matrix change processing is performed on the corrected visual data. This processing process uses mathematical transformation to align the coordinate system of the visual data with the coordinate system of the vault settlement, thereby solving the problem of inconsistent spatial references that may exist between the visual system and the displacement meter system. This spatial alignment method not only improves the accuracy of data fusion, but also provides a unified spatial reference framework for subsequent settlement analysis, allowing data from different sources to be effectively compared and analyzed in the same spatial dimension.

[0044] Secondly, the displacement rope system's data sampling frequency is dynamically adjusted based on the amount of vault settlement. When the vault settlement changes within a preset normal range, the displacement rope system's data sampling frequency remains at the initial setting. This initial setting is typically predetermined based on the general conditions of tunnel construction and monitoring requirements and meets routine monitoring requirements. However, when the vault settlement changes exceed the preset warning value, this indicates a possible abnormality in the tunnel vault settlement. In this case, the displacement rope system data needs to be sampled more frequently to capture the details of the settlement changes. To achieve this, cubic spline interpolation is performed on the displacement rope system data to match the timestamps of the visual displacement system. Cubic spline interpolation is a mathematical interpolation method that generates a smooth curve from existing data points, thereby temporally encrypting the displacement rope system data and synchronizing it with the high-frequency data of the visual displacement system.

[0045] When performing cubic spline interpolation processing, the time axis node sequence is first set to a specific distribution form. The recursive formula for the j-th cubic B-spline basis function is defined, where k=3 represents cubic spline. Through the recursive formula, the final basis function can be obtained. The determination of this basis function is the basis for interpolation processing, which determines the shape and properties of the interpolation curve. Next, the data collection time interval is dynamically adjusted according to the point allocation rule that adapts to the temporal non-uniformity of the excavation process. This dynamic adjustment rule can flexibly adjust the data collection time interval according to the actual situation during tunnel excavation, thereby better adapting to the temporal non-uniformity of the excavation process. Through this dynamic adjustment, the data collection frequency can be increased when the settlement changes greatly, and the collection frequency can be kept low when the settlement changes are small, thereby improving monitoring efficiency while ensuring data accuracy.

[0046] Based on the final basis functions and the allocation rule, a difference function is constructed from the data point sequence of the displacement rope system. This difference function can be used to connect the data points of the displacement rope system with the nodes on the time axis, thereby achieving data interpolation. Next, the coefficients are solved based on the difference function. By solving this system of equations, the required coefficients can be obtained, thus determining the specific form of the interpolation function.

[0047] Furthermore, to ensure the smoothness of the interpolation curve, constraints were introduced and solved using the Lagrange multiplier method. The constraints correspond to the second-order derivative conditions, and the Lagrange multiplier method yields a constrained least-squares solution. This constrained least-squares solution not only ensures the smoothness of the interpolation curve but also, to a certain extent, suppresses overfitting during the interpolation process, thereby improving the reliability and stability of the interpolation results. Based on the constrained least-squares solution, high-frequency data synchronization processing is performed. This synchronization process achieves precise temporal alignment of the visual displacement system and the displacement rope system data, providing high-quality synchronized data for subsequent settlement analysis.

[0048] S3, inputting the visual displacement data as a reference point correction amount into a displacement rope system data correction model, and using the displacement rope system data correction model to dynamically correct the aligned displacement rope system data to obtain a settlement measurement result of the excavation section in front of the tunnel face.

[0049] Specifically, step S3 further includes: the visual displacement system detects the current measured visual displacement data and the visual displacement data at the previous moment in real time, and automatically triggers continuous data collection when the visual displacement system measures a sudden change in the data set, and determines whether the current sudden change in data is a measurement error or sudden settlement through a standard deviation evaluation method based on the continuously measured data set;

[0050] When a measurement error occurs, the erroneous data value is eliminated and the stable data is updated using continuous measurement alternative treatments;

[0051] When sudden settlement occurs, the arch settlement is used as the reference point correction amount. ;

[0052] When the visual displacement system lens is blocked and the system fails, data is dynamically generated through the interpolation algorithm. , in order to smoothly modify the displacement rope system data curve, the formula for dynamically generating the benchmark point data using the interpolation algorithm is: ,in, The displacement measurement at the current moment Displacement measurement at the previous moment time interval, Displacement measurement for the next moment Displacement measurement at the previous moment time interval;

[0053] Correction amount based on reference point To correct the vault settlement, the displacement rope data correction model formula is: ,in, is the settlement of the vault at point i.

[0054] In this embodiment, the visual displacement data is input into the displacement rope data correction model as the reference point correction value to correct the data of the displacement rope system, thereby obtaining the settlement measurement result of the excavation section in front of the tunnel face.

[0055] Specifically, the visual displacement data at the current moment is monitored in real time and compared with the visual displacement data at the previous moment. The standard deviation evaluation method is used to determine whether the current visual displacement data has measurement errors or sudden settlements. This real-time monitoring and analysis mechanism can quickly identify data anomalies and ensure the reliability of the monitoring data. If the system detects measurement errors, such as abnormal visual displacement data caused by dust, water mist, or other interference factors, these erroneous data values ​​will be automatically eliminated. Subsequently, stable data from continuous measurements is used for update and replacement processing. This process not only avoids the impact of erroneous data on monitoring results, but also ensures the continuity and stability of the data, thereby improving the robustness of the monitoring system.

[0056] When sudden settlement is detected, the current vault settlement is used as a benchmark correction and input into the displacement rope data correction model. Based on this correction method, the method dynamically adjusts the displacement rope system data to eliminate benchmark drift errors caused by ground disturbances or other factors, thereby improving the accuracy of monitoring results. During the correction process, the method also dynamically generates data based on the corrected vault settlement and an interpolation algorithm to smooth the correction curve of the displacement rope system data. This dynamic interpolation method enables the system to generate a smooth correction data curve even in the event of visual displacement system failure or data discontinuity, ensuring the integrity and continuity of the monitoring data. Not only can it effectively eliminate benchmark drift errors, but the dynamic interpolation algorithm can also generate reliable correction data in the event of visual displacement system failure.

[0057] In summary, the machine vision calibration-based method for monitoring initial tunnel vault settlement, by integrating dynamic machine vision calibration technology with displacement meter data correction, creates an innovative two-layer monitoring architecture. This enables continuous, high-precision monitoring from pre-construction to construction. This approach addresses the existing inability to accurately measure tunnel vault settlement throughout the entire process, from the initial, unexcavated state to excavation and construction.

[0058] Specifically, this method first uses a visual displacement camera to collect visual displacement data of the tunnel vault. It then utilizes an adaptive ambient light compensation algorithm and infrared-assisted calibration to ensure high-quality monitoring data, even in complex construction environments (such as those subject to interference from dust and water mist). The visual displacement camera is installed in a supported area 20 to 50 meters behind the tunnel face. Its target coincides with the borehole of the displacement rope system, ensuring consistency of the spatial reference. Simultaneously, the displacement rope system is pre-drilled into the stratum above the tunnel, with node sensors distributed in a gradient pattern with equal or unequal spacing, enabling continuous settlement monitoring of the excavation section ahead of the tunnel face.

[0059] Secondly, the coordinates of the visual displacement data and the displacement rope system data were unified through Euler rotation matrix transformations, resolving any potential spatial datum inconsistencies between the two. Furthermore, the displacement rope system data sampling frequency was dynamically adjusted based on the amount of vault settlement. When the settlement exceeded the warning value, cubic spline interpolation was used to match the timestamps of the visual displacement system, ensuring temporal synchronization. This spatiotemporal registration method not only improved the accuracy of data fusion but also provided a unified spatiotemporal datum for subsequent settlement analysis.

[0060] Finally, regarding data correction, visual displacement data is input into the displacement rope data correction model as a benchmark correction. By monitoring changes in visual displacement data in real time, measurement errors or sudden settlement can be quickly identified and corrective measures can be taken. When a measurement error is detected, the erroneous data is discarded and replaced with stable data from continuous measurements. When sudden settlement is detected, the benchmark correction is used to correct the displacement rope system data, eliminating benchmark drift errors caused by formation disturbances. Furthermore, a dynamic interpolation algorithm generates a smooth correction data curve in the event of a visual displacement system failure, ensuring the integrity and continuity of the monitoring data.

[0061] Simply put, the machine vision calibration-based method for monitoring the initial vault settlement of a tunnel not only enables continuous monitoring of the entire process from the initial state of the tunnel to excavation and construction, but also significantly improves the monitoring accuracy and system adaptability through a dynamic benchmark establishment and transmission mechanism. Compared with existing technologies, the machine vision calibration-based method for monitoring the initial vault settlement of a tunnel solves problems such as the low efficiency of traditional total station monitoring, the insufficient accuracy of single machine vision monitoring in complex environments, and the drift of displacement meter benchmark points. It also further enhances the reliability and stability of the monitoring system through multi-source data cross-validation and spatiotemporal benchmark synchronization. This high-precision, high-reliability monitoring system provides strong technical support for the safety and quality of tunnel construction, and has significant practical value and broad application prospects.

[0062] See also Figure 4A second embodiment of the present invention provides a tunnel initial vault settlement monitoring device based on machine vision calibration, which includes:

[0063] A preprocessing unit 201 is configured to obtain visual displacement data collected by a preset visual displacement system and displacement rope system data collected by a displacement rope system, and preprocess the visual displacement data and displacement rope system data to obtain corrected visual displacement data and vault settlement;

[0064] an alignment unit 202 for performing data fusion and spatiotemporal registration processing based on the corrected visual displacement data and the vault settlement, so that the visual displacement data and the displacement rope system data are spatially aligned and temporally synchronized;

[0065] The correction unit 203 is used to input the visual displacement data as a reference point correction amount into the displacement rope data correction model, and use the displacement rope data correction model to correct the aligned displacement rope system data to obtain the settlement measurement result of the excavation section in front of the tunnel face.

[0066] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for monitoring initial tunnel vault settlement based on machine vision calibration, characterized in that: include: Acquire real-time dynamic visual displacement data collected by a preset visual displacement system, and obtain tunnel vault settlement using displacement rope system data collected by a pre-buried displacement rope system in the tunnel, and pre-process the visual displacement data and displacement rope system data to obtain corrected visual displacement data and vault settlement; Performing data fusion and spatiotemporal registration processing based on the corrected visual displacement data and the vault settlement, so that the visual displacement data and the displacement rope system data are aligned in space and synchronized in time; The visual displacement data is input into the displacement rope system data correction model as the reference point correction value, and the displacement rope system data correction model is used to dynamically correct the aligned displacement rope system data to obtain the settlement measurement results of the excavation section in front of the tunnel face, specifically: The visual displacement system detects the current visual displacement data and the visual displacement data of the previous moment in real time. When the visual displacement system measures data suddenly, it automatically triggers continuous data collection. Based on the continuously measured data set, the standard deviation evaluation method is used to determine whether the current sudden change data is a measurement error or sudden settlement. When a measurement error occurs, the erroneous data value is eliminated and the continuous measurement stable data is updated. Alternative treatment: when sudden settlement occurs, the vault settlement is used as the reference point correction amount When the visual displacement system lens is blocked and the system fails, data is dynamically generated through the interpolation algorithm. , in order to smoothly modify the displacement rope system data curve, the formula for dynamically generating the benchmark point data using the interpolation algorithm is: ,in, The displacement measurement at the current moment Displacement measurement at the previous moment time interval, Displacement measurement for the next moment Displacement measurement at the previous moment time interval; Correction amount based on reference point To correct the vault settlement, the displacement rope data correction model formula is: ,in, is the settlement of the vault at point i.

2. The method for monitoring initial tunnel vault settlement based on machine vision calibration according to claim 1, characterized in that: The visual displacement system is a visual displacement camera, which includes an infrared anti-interference module. The visual displacement camera is configured in a stable area within a preset length range behind the tunnel face to be monitored, where tunnel support has been formed and no displacement occurs. Its corresponding target is arranged at the arch position within the excavated area behind the tunnel face to be monitored, wherein the preset length range is 20~50m.

3. The method for monitoring initial tunnel vault settlement based on machine vision calibration according to claim 2, characterized in that: The displacement rope system is a displacement rope fixed on a steel pipe. The displacement rope is pre-buried in the stratum above the tunnel by advance drilling. The displacement rope is composed of a gradient distribution of node sensors with equal or unequal spacing, wherein the position of the target coincides with the position of the drilling hole of the displacement rope system.

4. The method for monitoring initial tunnel vault settlement based on machine vision calibration according to claim 2, characterized in that: Preprocessing the visual displacement data to obtain corrected visual data is performed, specifically: Determine whether the current ambient light intensity is less than the preset value; If so, activate the infrared anti-interference module, perform infrared fill light, and perform adaptive environmental compensation processing on the collected visual displacement data to obtain corrected visual data; If not, the collected visual displacement data is directly subjected to adaptive environmental compensation processing to obtain corrected visual data.

5. The method for monitoring initial tunnel vault settlement based on machine vision calibration according to claim 4 is characterized in that: The adaptive environmental compensation process is specifically as follows: Based on the nonlinear compensation model, the corrected visual displacement data is calculated using the formula: ,in, is the camera grayscale value, is the ambient light intensity, is the dark current reference value, and are all fitting parameters.

6. The method for monitoring initial tunnel vault settlement based on machine vision calibration according to claim 3, characterized in that: The displacement rope system data is preprocessed to obtain the vault settlement, specifically: Obtain the inclination change data at the location collected by the node sensor, use the preset reference point as the starting point, perform trapezoidal integration processing, and calculate the arch settlement at any location of the tunnel to be monitored. The calculation formula is: ,in, is the settlement of the vault at point i, is the i-th original tilt change data, is the i-1th original tilt change data, is the arrangement spacing of node sensors.

7. The method for monitoring initial tunnel vault settlement based on machine vision calibration according to claim 6, characterized in that: Data fusion and spatiotemporal registration processing is performed based on the corrected visual displacement data and the vault settlement, specifically as follows: The corrected visual displacement data is processed by Euler rotation matrix transformation so that the coordinates of the corrected visual data are consistent with the coordinates of the vault settlement; The data sampling frequency of the displacement rope system is adjusted according to the change of the vault settlement. When the settlement change of the vault settlement is within the preset normal state, the data sampling frequency of the displacement rope system is the initial setting value. When the settlement change of the vault settlement exceeds the preset warning value, the acquisition frequency will be automatically adjusted dynamically and the acquisition frequency will be encrypted. The displacement rope system will collect the data of the displacement rope system according to the real-time dynamic acquisition frequency. Perform cubic spline interpolation to match the timestamp of the visual displacement system ; Set the timeline node sequence to , define the jth cubic B-spline basis function The recursive formula is: , , where k=3 represents cubic spline, is the 0-order B-spline basis function, which serves as the starting point of the recursion. is the j-th k-order B-spline basis function, is the j-th k-1 order B-spline basis function, For the j+1th k-1th order B-spline basis function, construct a cubic spline interpolation curve, k=3, and obtain the final basis function ; Determine the time non-uniformity point allocation rule that adapts to the excavation process and dynamically adjust the data collection time interval. The allocation rule is: ,in, The time interval for dynamic data collection of the displacement rope system; Based on the final basis function and the distribution rule, the data point sequence of the displacement rope system , construct the difference function: , where m is the number of original data points of the displacement rope system, For the interpolation coefficient to be determined, expand the two control points to meet the boundary conditions; The coefficients are solved according to the difference function, and the formula is: , and marked as Ac=S, where is the m+2th cubic B-spline basis function at time node The value at is the m+2th cubic B-spline basis function at time node The value at is the m+2th cubic B-spline basis function at time node The value at is the interpolation coefficient of the first basis function to be determined, is the interpolation coefficient of the second basis function to be determined, is the interpolation coefficient of the m+2th basis function to be determined, is the displacement rope system at the time node The settlement data actually measured at is the displacement rope system at the time node The settlement data actually measured at is the displacement rope system at the time node The settlement data actually measured at is the time label of the mth data point collected by the displacement rope system, A is the coefficient matrix composed of the values ​​of the basis function at each time node, , c is the coefficient vector to be solved, , S is a vector consisting of the settlement data actually collected by the displacement rope system, ; According to the constraints , and obtain the constrained least squares solution , , where the constraint matrix C corresponds to the second-order derivative condition and is solved using the Lagrange multiplier method; Based on the constrained least squares solution, high-frequency data synchronization processing is performed to time stamp the visual displacement system. , and the corresponding synchronized displacement rope system data is: ,in, is the interpolation coefficient of the jth basis function to be determined.

8. A tunnel initial vault settlement monitoring device based on machine vision calibration, characterized in that: include: a preprocessing unit for acquiring visual displacement data collected by a preset visual displacement system and displacement rope system data collected by a displacement rope system, and preprocessing the visual displacement data and displacement rope system data to obtain corrected visual displacement data and vault settlement; an alignment unit for performing data fusion and spatiotemporal registration processing based on the corrected visual displacement data and the vault settlement, so that the visual displacement data and the displacement rope system data are spatially aligned and temporally synchronized; The correction unit is used to input the visual displacement data as a reference point correction amount into the displacement rope data correction model, and use the displacement rope data correction model to correct the aligned displacement rope system data to obtain the settlement measurement results of the excavation section in front of the tunnel face, specifically: The visual displacement system detects the current visual displacement data and the visual displacement data of the previous moment in real time. When the visual displacement system measures data suddenly, it automatically triggers continuous data collection. Based on the continuously measured data set, the standard deviation evaluation method is used to determine whether the current sudden change data is a measurement error or sudden settlement. When a measurement error occurs, the erroneous data value is eliminated and the continuous measurement stable data is updated. Alternative treatment: when sudden settlement occurs, the vault settlement is used as the reference point correction amount When the visual displacement system lens is blocked and the system fails, data is dynamically generated through the interpolation algorithm. , in order to smoothly modify the displacement rope system data curve, the formula for dynamically generating the benchmark point data using the interpolation algorithm is: ,in, The displacement measurement at the current moment Displacement measurement at the previous moment time interval, Displacement measurement for the next moment Displacement measurement at the previous moment time interval; Correction amount based on reference point To correct the vault settlement, the displacement rope data correction model formula is: ,in, is the settlement of the vault at point i.

Images