Tunnel deformation monitoring method based on chained inertial vision measurement

Abstract

The invention relates to the technical field of tunnel deformation monitoring, in particular to a tunnel deformation monitoring method based on chained inertial vision measurement, and the method comprises the steps: dividing a to-be-monitored reservoir tunnel region into a plurality of monitoring regions; acquiring image data and camera inertial data of each monitoring area, and determining target displacement data of the corresponding monitoring area according to the image data; determining deformation quantity data of the corresponding monitoring area based on the target displacement data and the camera inertia data; and generating a deformation monitoring result of the corresponding monitoring area according to the deformation quantity data. The method has the effect of improving the accuracy of reservoir tunnel deformation monitoring calculation and the adaptability of deformation monitoring.

Description

Technical Field

[0001] This application relates to the technical field of tunnel deformation monitoring, and in particular to a tunnel deformation monitoring method based on chain inertial vision measurement. Background Art

[0002] At present, as an important part of water conservancy projects, reservoir tunnels undertake functions such as water conveyance, flood discharge, and maintenance channels. In order to ensure the safe operation of reservoir tunnels, it is necessary to conduct long-term monitoring of the structural deformation of the tunnels to determine whether there are risks such as abnormal displacement, settlement, and deformation, so as to take maintenance measures in a timely manner to prevent serious structural damage and safety accidents.

[0003] For the existing deformation monitoring of reservoir tunnels, single-point measurement methods, total station measurement methods, etc. are mainly used. Among them, the single-point measurement method usually relies on manual regular measurement using a level, total station or static GPS equipment. This method has a low measurement frequency, is difficult to achieve continuous monitoring of deformation amounts, and manual operation is easily affected by environmental factors, resulting in large measurement errors; although the total station measurement method can provide high-precision data, due to its dependence on single-point fixed positions, it is difficult to measure at the bent places of the tunnel. At the same time, the equipment cost is high and the measurement period is long, so there is room for improvement. Summary of the Invention

[0004] This application provides a tunnel deformation monitoring method based on chain inertial vision measurement, which can improve the accuracy of deformation monitoring calculation and the adaptability of deformation monitoring for reservoir tunnels.

[0005] The first above-mentioned invention object of this application is achieved through the following technical solutions: A tunnel deformation monitoring method based on chain inertial vision measurement, the tunnel deformation monitoring method based on chain inertial vision measurement includes: Dividing the area of the reservoir tunnel to be monitored into several monitoring areas; Obtaining the image data and camera inertial data of each monitoring area, and determining the target displacement data of the corresponding monitoring area according to the image data; Based on the target displacement data and the camera inertial data, determining the deformation amount data of the corresponding monitoring area; Generating a deformation monitoring result for the corresponding monitoring area according to the deformation amount data.

[0006] By adopting the above technical solution, the reservoir tunnel area to be monitored is divided into several monitoring areas, which can ensure that the monitoring system can conduct refined monitoring for different areas, avoid data mixing caused by overall monitoring, thereby improving the monitoring accuracy. By acquiring the image data and camera inertial data of each monitoring area and determining the target displacement data of the corresponding monitoring area according to the image data, the displacement information of the reservoir tunnel can be obtained and analyzed in real time, avoiding data errors caused by a single measurement method, thereby enhancing the stability and reliability of deformation monitoring. By determining the deformation amount data of the corresponding monitoring area based on the target displacement data and camera inertial data, the data of visual measurement and inertial measurement can be comprehensively considered, reducing the measurement errors caused by external environmental interference, thereby improving the accuracy of deformation amount calculation and making the monitoring results more credible.

[0007] In a preferred example of the present application, it can be further configured that: the tunnel deformation monitoring method based on chain inertial vision measurement further includes: Acquire the target images of different monitoring areas of the reservoir tunnel at any moment and the corresponding camera inertial data, wherein the camera inertial data is collected by the inertial measurement unit of the inertial camera; Input the corresponding target image and the camera inertial data into the initial model, and optimize the model parameters of the initial model through backpropagation and model evaluation metrics; When the model evaluation metric reaches a preset threshold, complete the training of the initial model to obtain a deformation model.

[0008] By adopting the above technical solution, by acquiring the target images of different monitoring areas of the reservoir tunnel at any moment and the corresponding camera inertial data and inputting them into the initial model, the initial model is trained to obtain a deformation model, enabling the deformation calculation model to have an adaptive adjustment ability and still maintain a high calculation accuracy under different monitoring environments, different tunnel structures, and different external influencing factors. The influence degree of the deformation amount at different positions of the target and the camera is fully considered, and the corresponding deformation amount is output, thereby improving the accuracy of the deformation amount.

[0009] In a preferred example of the present application, it can be further configured that: the acquiring the image data and camera inertial data of each monitoring area and determining the target displacement data of the corresponding monitoring area according to the image data specifically includes: Collect the image data and camera inertial data of each monitoring area according to a preset time period; Based on the image processing algorithm and PnP algorithm, perform image processing and pose calculation on the image data to obtain the target displacement data.

[0010] By adopting the above technical solution, image data and camera inertial data of each monitoring area are collected, which can ensure the time synchronization of the monitoring data and avoid the inconsistency caused by time drift during the data acquisition process, thereby improving the temporal continuity and integrity of the monitoring data. Then, by obtaining the target displacement data based on the image processing algorithm and the PnP algorithm, the influence of factors such as the camera installation angle, illumination changes, and image distortion on the measurement accuracy can be effectively reduced, thereby improving the stability of the target displacement calculation and making the measurement results more reliable.

[0011] In a preferred example, the present application may be further configured as follows: performing image processing and pose calculation on the image data based on the image processing algorithm and the PnP algorithm to obtain the target displacement data specifically includes: Based on the image processing algorithm, target area recognition processing is performed on the image data to obtain target highlight image data; Based on the PnP algorithm, the target pose is calculated for the target highlighted image data to obtain the target displacement data.

[0012] By adopting the above technical solution, with the help of image processing algorithms and PnP algorithms, the target area can be accurately identified, the reliability of target positioning can be improved, and the position changes of the target in different camera frames can be combined to calculate the real displacement data of the target, thereby improving the stability of the target displacement calculation.

[0013] In a preferred example, the present application may be further configured as follows: determining the deformation data of the corresponding monitoring area based on the target displacement data and the camera inertial data, specifically including: Inputting the target displacement data and the camera inertia data into the deformation model to obtain a deformation variable; The deformation variables are fitted based on a time series filtering algorithm and a least square method to obtain deformation variable data of the corresponding monitoring area.

[0014] By adopting the above technical solution, the target displacement data and camera inertial data are input into the deformation model to obtain the deformation variable, which fully combines the advantages of visual measurement and inertial measurement, reduces the data fluctuation caused by a single measurement method, and improves the accuracy of deformation variable calculation. The deformation variable is then fitted to obtain the deformation variable data of the corresponding monitoring area, which can effectively eliminate outliers caused by short-term noise and improve the trend analysis capability of deformation data.

[0015] In a preferred example, the present application may be further configured as follows: generating a deformation monitoring result corresponding to the monitoring area according to the deformation amount data specifically includes: Based on the finite element analysis method, the benchmark deformation data of different monitoring areas are obtained; Calculate the deviation value between the deformation data of the current monitoring area and the reference deformation data, and generate a corresponding deformation monitoring result according to the deviation value.

[0016] By adopting the above technical solution, based on the finite element analysis method, the reference deformation data of different monitoring areas can be obtained, and the theoretical reference deformation data can be calculated by combining the structural characteristics and mechanical model of the reservoir tunnel, so as to provide a reliable comparison standard for the actual monitoring data, avoid misjudgment caused by measurement errors or environmental changes, and then through the deviation value between the deformation data and the reference deformation data, it can be judged whether the deformation of the current monitoring area is within the normal range, thereby improving the credibility of the deformation monitoring result and enabling the monitoring data to be directly used for decision-making analysis.

[0017] In a preferred example of the present application, it can be further configured that: after obtaining the deformation monitoring result, if the deviation value exceeds a preset warning threshold, an alarm mechanism is triggered, and the deformation monitoring result is visually processed to obtain a three-dimensional deformation heat map.

[0018] By adopting the above technical solution, after obtaining the deformation monitoring result, if the deviation value exceeds the preset warning threshold, an alarm mechanism is triggered, which can immediately issue an alarm when the deformation is abnormal, improve the guarantee ability of the structural safety of the reservoir tunnel, and through visual processing of the deformation monitoring result, enable the monitoring personnel to intuitively view the overall deformation of the reservoir tunnel, quickly identify high-risk areas, thereby improving the readability and visual analysis ability of the monitoring result, and making the monitoring data easier to understand and analyze.

[0019] In summary, the present application includes at least one of the following beneficial technical effects: 1. By adopting the above technical solution, the reservoir tunnel area to be monitored is divided into several monitoring areas, which can ensure that the monitoring system can conduct refined monitoring for different areas, avoid data mixing caused by overall monitoring, thereby improving the monitoring accuracy. By obtaining the image data and camera inertial data of each monitoring area, and determining the target displacement data of the corresponding monitoring area according to the image data, the displacement information of the reservoir tunnel can be obtained and analyzed in real time, avoiding data errors caused by a single measurement method, thereby improving the stability and reliability of deformation monitoring. By determining the deformation data of the corresponding monitoring area based on the target displacement data and camera inertial data, the data of visual measurement and inertial measurement can be comprehensively considered, reducing measurement errors caused by external environmental interference, thereby improving the accuracy of deformation calculation and making the monitoring result more credible; 2. By adopting the above technical solution, by obtaining the target images of different monitoring areas of the reservoir tunnel at any moment and the corresponding camera inertial data, and inputting them into the initial model, the initial model is trained to obtain a deformation model, enabling the deformation calculation model to have an adaptive adjustment ability, and still being able to maintain a high calculation accuracy under different monitoring environments, different tunnel structures and different external influencing factors. The influence degree of the deformation amount at different positions of the target and the camera is fully considered, and the corresponding deformation amount is output, thereby improving the accuracy of the deformation amount; 3. After obtaining the deformation monitoring result, if the deviation value exceeds the preset warning threshold, the alarm mechanism is triggered, which can immediately send an alarm when the deformation appears abnormally, improving the guarantee ability of the safety of the reservoir tunnel structure. And by visualizing the deformation monitoring result, the monitoring personnel can intuitively view the overall deformation situation of the reservoir tunnel, quickly identify high-risk areas, thereby improving the readability and visual analysis ability of the monitoring result, making the monitoring data easier to understand and analyze. Description of the Drawings

[0020] Figure 1 is the implementation flowchart of the tunnel deformation monitoring method based on chain inertial vision measurement in an embodiment of the present application; Figure 2 is another implementation flowchart of the tunnel deformation monitoring method based on chain inertial vision measurement in an embodiment of the present application; Figure 3 is the implementation flowchart of step S20 in the tunnel deformation monitoring method based on chain inertial vision measurement in an embodiment of the present application; Figure 4 is the implementation flowchart of step S22 in the tunnel deformation monitoring method based on chain inertial vision measurement in an embodiment of the present application; Figure 5 is the implementation flowchart of step S30 in the tunnel deformation monitoring method based on chain inertial vision measurement in an embodiment of the present application; Figure 6 is the implementation flowchart of step S40 in the tunnel deformation monitoring method based on chain inertial vision measurement in an embodiment of the present application; Figure 7 is the setting schematic diagram of the tunnel deformation monitoring method based on chain inertial vision measurement in an embodiment of the present application. Detailed Embodiments

[0021] The following embodiments will help those skilled in the art to further understand the role of the present application, but do not limit the present application in any form. It should be noted that those of ordinary skill in the art can make several deformations and improvements without departing from the concept of the present application. These all belong to the protection scope of the present application.

[0022] In the following description, specific details such as specific system structures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0023] It should be understood that when used in the specification of the present application and the appended claims, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0024] The present application will be further described in detail below with reference to the accompanying drawings.

[0025] In one embodiment, as Figure 1 shown, the present application discloses a tunnel deformation monitoring method based on chain inertial vision measurement, which specifically includes the following steps: S10: Divide the reservoir tunnel area to be monitored into several monitoring areas.

[0026] Specifically, the reservoir tunnel area to be monitored is divided according to its structural characteristics, length, bending conditions, and monitoring requirements. The division of the monitoring areas can be based on established structural units, such as tunnel segments, segment construction joints, key structural support points, etc. At the same time, the field of view range and installation position of the camera need to be considered to ensure that the camera can completely cover each monitoring area. The number and boundaries of the monitoring areas can be dynamically adjusted according to historical deformation conditions and structural health assessment results. After the division is completed, the numbers, coordinate ranges, and corresponding camera monitoring ranges of each monitoring area need to be recorded in the system database for subsequent automated execution of data collection and analysis.

[0027] Exemplarily, as Figure 7 shown, a total of nine cross-sections are selected in the tunnel to install targets. The numbers of the nine cross-sections are a - i. There is one monitoring area between each cross-section. Six targets are installed along the wall of each cross-section, and it is ensured that the three targets on the left and right sides are symmetrically distributed and evenly distributed. To monitor all cross-sections, a total of five measuring stations are set up in the tunnel. Each measuring station consists of two inertial cameras facing in opposite directions and a target, and is used to monitor the cross-section targets on both sides of each measuring station. Among them, the target installed on cross-section c is a cooperative target. Due to the limitation of the field of view, the targets on cross-section c are only installed on the outer wall of the turn, and considering that each target can only reflect light in one direction, three points are selected on the outer wall, and the targets are installed in a "back-to-back" manner at each point to ensure that they can be observed by the second measuring station before entering the bend and the third measuring station after entering the bend at the same time.

[0028] S20: Obtain the image data and camera inertial data of each monitoring area, and determine the target displacement data of the corresponding monitoring area according to the image data.

[0029] Specifically, within each monitoring period, the image data of each monitoring area is synchronously collected by multiple inertial cameras, and the camera attitude data is synchronously recorded by the inertial measurement unit (IMU). The image data is stored in the database through the image transmission protocol and time-stamped alignment is performed to ensure that all data sources can be correctly matched and associated. Then, the image data is processed by the built-in algorithm to determine the displacement of the target, so as to obtain the target displacement data of the corresponding monitoring area.

[0030] S30: Determine the deformation data of the corresponding monitoring area based on the target displacement data and camera inertial data.

[0031] Specifically, a preset calculation model is adopted. The target displacement data and camera inertial data corresponding to the monitoring area are input into the calculation model. After obtaining the target displacement data, the calculation model uses the transformation matrix to convert it into a unified global coordinate system, and combines the camera inertial data to calculate the displacement distribution of each target within the monitoring area, so as to output the deformation data of the corresponding monitoring area.

[0032] S40: Generate the deformation monitoring result of the corresponding monitoring area according to the deformation data.

[0033] Specifically, by comparing the calculated deformation data with the reference deformation data of the corresponding monitoring area, and calculating the deformation growth rate based on the time-series trend of the deformation to determine whether there is a trend of accelerating development of the deformation. During the calculation of the deformation monitoring result, different deformation thresholds are set, including the normal state threshold, the warning threshold, and the danger threshold, and the current monitoring data is compared. If the deformation data is within the normal range, it is marked as "normal"; if it exceeds the warning threshold, it is marked as "warning"; if it exceeds the danger threshold, it is marked as "abnormal".

[0034] In one embodiment, as Figure 2 shown, the tunnel deformation monitoring method based on chain inertial vision measurement further includes: S1: Obtain the target images of different monitoring areas of the reservoir tunnel at any moment and the corresponding camera inertial data, where the camera inertial data is collected by the inertial measurement unit of the inertial camera.

[0035] Specifically, during the calculation of the deformation amount, historical data is required for model training. Therefore, it is necessary to retrieve the target images at different monitoring times from the database and simultaneously obtain the camera inertial data at the corresponding time points. To ensure the integrity of the data, a temporal alignment operation of the data is first performed, that is, the camera images and IMU data are matched according to the timestamps, and linear interpolation or time synchronization adjustment is performed to ensure that all data points are aligned in the time dimension for the accuracy of subsequent model training.

[0036] S2: Input the corresponding target images and camera inertial data into the initial model, and optimize the model parameters of the initial model through backpropagation and model evaluation metrics.

[0037] Specifically, when training the deformation model, it is first necessary to extract the features of the target image data and fuse it with the inertial data to form the input features of the model. During backpropagation, the gradient descent method is used to continuously optimize the model parameters to minimize the value of the loss function. At the same time, model evaluation metrics, such as the prediction error or the value of the loss function, are used to measure the fitting effect of the model. If the error converges within the set range, it is considered that the model training is completed; otherwise, the parameters are continuously optimized until the preset threshold is reached.

[0038] S3: When the model evaluation metrics reach the preset threshold, complete the training of the initial model to obtain the deformation model.

[0039] Specifically, during the model training process, the parameters will be continuously adjusted to optimize the accuracy of the deformation model. When the model evaluation metrics, such as the prediction error or the value of the loss function, are stable within the set threshold range and there is no obvious optimization after several consecutive rounds of iteration, it is considered that the model training is completed, and thus the trained deformation model is obtained, and the deformation model is stored for subsequent deformation amount calculation and trend prediction.

[0040] In one embodiment, as Figure 3 shown, in step S20, that is, obtaining the image data and camera inertial data of each monitoring area, and determining the target displacement data of the corresponding monitoring area according to the image data, specifically including: S21: Collect the image data and camera inertial data of each monitoring area according to the preset time period.

[0041] Specifically, in each monitoring cycle, an inertial camera synchronization acquisition instruction is triggered to make all cameras perform image acquisition at a set sampling interval. The image data collected by each camera is automatically associated with a timestamp and stored in a database. At the same time, an inertial measurement unit (IMU) collects attitude data such as the acceleration and pitch angle of the camera for use in pose compensation in subsequent processing. After the acquisition is completed, all camera image data and inertial data are classified and stored according to the monitoring area, and a data integrity check is performed. If data loss occurs, a re-acquisition mechanism is triggered to make the camera perform additional re-acquisition in the next time cycle to ensure data integrity and continuity. Among them, the sampling frequency of the inertial camera is set to 60 Hz, the synchronization error of multiple inertial cameras is 20 ms, the field of view angle is 3.5°×2.7°, and the coverage range is ≤100 m.

[0042] S22: Based on the image processing algorithm and the PnP algorithm, perform image processing and pose calculation on the image data to obtain the target displacement data.

[0043] Specifically, the collected image data is first denoised and enhanced to improve the image quality and the recognizability of the target. Gaussian filtering or bilateral filtering can be used for denoising, and brightness equalization can be performed through adaptive histogram equalization for enhancement. Then, a feature point detection algorithm such as SIFT, ORB, or SURF is used to extract the features of the target in the image, and the FLANN matching algorithm or the optical flow method is used to track the target to calculate the displacement of the target between adjacent frames. Then, through the Perspective-n-Point (PnP) algorithm, the spatial pose change of the target is solved according to the internal and external parameters of the camera, and finally the three-dimensional displacement data of the target is obtained and converted to a unified coordinate system for subsequent calculation of the deformation amount.

[0044] In one embodiment, as Figure 4 shown, in step S22, that is, based on the image processing algorithm and the PnP algorithm, perform image processing and pose calculation on the image data to obtain the target displacement data, which specifically includes: S221: Based on the image processing algorithm, perform target area recognition processing on the image data to obtain the target highlighted image data.

[0045] Specifically, first perform target area recognition. By pre-calibrating the target image area range, draw the region of interest; then use median filtering to denoise the image within the region, arrange the pixels within a certain neighborhood range of a certain pixel according to the gray level, and use the middle value as the output object to eliminate abnormal pixel values. Finally, use the position of the pixel with the maximum weighted pixel value as the measurement target position for recognition, thereby forming the highlighted target image data to provide input for subsequent pose calculation.

[0046] S222: Calculate the pose of the target based on the PnP algorithm for the highlighted image data of the target to obtain the target displacement data.

[0047] Specifically, use the PnP algorithm to solve the pose change of the target. First, extract the feature points of the target from the highlighted target image data, and find the corresponding point pairs between adjacent frames through the matching algorithm. Then, based on the internal parameter matrix, distortion coefficient, and external parameters of the camera, solve the spatial pose change of the target to obtain the three-dimensional displacement data of the target, and convert it to a unified coordinate system. Finally, obtain the displacement data of the target at different time points and store it in the database for subsequent calculation of the deformation amount. The robust optimization method can also be used for error compensation to remove abnormal data points caused by occlusion and lighting changes.

[0048] In one embodiment, as Figure 5 shown, in step S30, that is, based on the target displacement data and the camera inertial data, determine the deformation amount data of the corresponding monitoring area, specifically including: S31: Input the target displacement data and the camera inertial data into the deformation model to obtain the deformation amount.

[0049] Specifically, during the calculation of the deformation amount, input the target displacement data and the camera inertial data into the deformation model. The deformation model considers the influence degree of the deformation amount at different positions of the target and the camera, and thus outputs the corresponding deformation amount.

[0050] S32: Fit the deformation amount based on the time series filtering algorithm and the least squares method to obtain the deformation amount data of the corresponding monitoring area.

[0051] Specifically, after the calculation of the deformation amount is completed, in order to remove the short-term measurement error and data fluctuation, use the time series filtering algorithm for data smoothing. The time series filtering uses the Kalman filter or the exponential weighted moving average method to ensure the smoothness of the deformation amount data. Then, use the least squares method to perform curve fitting on the deformation amount to obtain the deformation amount data of the corresponding monitoring area.

[0052] In one embodiment, as Figure 6 shown, in step S40, that is, generate the deformation monitoring result of the corresponding monitoring area according to the deformation amount data, specifically including: S41: Based on the finite element analysis method, obtain the reference deformation amount data of different monitoring areas.

[0053] Specifically, in order to obtain the reference deformation amount data, first establish a finite element model of the reservoir tunnel. This model considers factors such as material properties, boundary conditions, and external loads. Then, based on this finite element model, perform static analysis and dynamic analysis, calculate the deformation amount distribution under normal conditions, and thus obtain the reference deformation amount data of different monitoring areas.

[0054] S42: Calculate the deviation value between the deformation data of the current monitoring area and the reference deformation data, and generate a corresponding deformation monitoring result according to the deviation value.

[0055] Specifically, by comparing the deformation data of the current monitoring area with the reference deformation data, the deformation situation of the current monitoring area can be judged, and thus the deformation monitoring result can be obtained. If the deformation amount exceeds the set threshold, it is determined that there is abnormal deformation in the monitoring area.

[0056] In one embodiment, after obtaining the deformation monitoring result, when the deviation value exceeds the preset warning threshold, an alarm mechanism is triggered, and the deformation monitoring result is visually processed to obtain a three-dimensional deformation heat map.

[0057] Specifically, after the deformation monitoring result is generated, if the deviation value exceeds the set threshold, the alarm mechanism is triggered. The alarm methods include real-time alarm, remote notification, and automatic report generation. The real-time alarm can remind the monitoring personnel through an audible and visual alarm device. The remote notification can be sent to relevant personnel by means of text messages, emails, or push notifications from the monitoring system. At the same time, the deformation result is visually processed using a three-dimensional deformation heat map. The three-dimensional heat map uses colors to map the deformation amounts of different regions. Red indicates a severely deformed area, yellow indicates a warning area, and green indicates a normal area, so that the monitoring personnel can intuitively analyze the deformation situation.

[0058] It should be understood that the sequence numbers of the steps in the above embodiments do not mean the order of execution. The execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0059] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the above division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the system is divided into different functional units or modules to complete all or part of the functions described above.

[0060] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A tunnel deformation monitoring method based on chain inertial vision measurement, characterized in that The tunnel deformation monitoring method based on chain inertial vision measurement includes: Dividing the reservoir tunnel area to be monitored into several monitoring areas; Obtaining the image data and camera inertial data of each monitoring area, and determining the target displacement data of the corresponding monitoring area according to the image data; Determining the deformation amount data of the corresponding monitoring area based on the target displacement data and the camera inertial data; Generating a deformation monitoring result of the corresponding monitoring area according to the deformation amount data.

2. The tunnel deformation monitoring method based on chain inertial vision measurement according to claim 1, characterized in that, The tunnel deformation monitoring method based on chain inertial vision measurement further includes: Obtaining the target images and corresponding camera inertial data of different monitoring areas of the reservoir tunnel at any moment, wherein the camera inertial data is collected by the inertial measurement unit of an inertial camera; Inputting the corresponding target images and the camera inertial data into an initial model, and optimizing the model parameters of the initial model through backpropagation and model evaluation metrics; When the model evaluation metric reaches a preset threshold, completing the training of the initial model to obtain a deformation model.

3. The tunnel deformation monitoring method based on chain inertial vision measurement according to claim 1, characterized in that The obtaining the image data and camera inertial data of each monitoring area, and determining the target displacement data of the corresponding monitoring area according to the image data specifically includes: Collecting the image data and camera inertial data of each monitoring area at a preset time period; Performing image processing and pose calculation on the image data based on an image processing algorithm and a PnP algorithm to obtain the target displacement data.

4. The tunnel deformation monitoring method based on chain inertial vision measurement according to claim 3, wherein The performing image processing and pose calculation on the image data based on the image processing algorithm and the PnP algorithm to obtain the target displacement data specifically includes: Performing target area recognition processing on the image data based on the image processing algorithm to obtain target highlighted image data; Performing target pose calculation on the target highlighted image data based on the PnP algorithm to obtain the target displacement data.

5. The tunnel deformation monitoring method based on chain inertial vision measurement according to claim 2, wherein, The determining the deformation amount data of the corresponding monitoring area based on the target displacement data and the camera inertial data specifically includes: Inputting the target displacement data and the camera inertial data into the deformation model to obtain a deformation amount; Fitting the deformation amount based on a time series filtering algorithm and the least squares method to obtain the deformation amount data of the corresponding monitoring area.

6. The tunnel deformation monitoring method based on chain inertial vision measurement according to claim 5, characterized in that, The generating a deformation monitoring result of the corresponding monitoring area according to the deformation amount data specifically includes: Obtaining the reference deformation amount data of different monitoring areas based on a finite element analysis method; Calculating the deviation value between the deformation amount data of the current monitoring area and the reference deformation amount data, and generating a corresponding deformation monitoring result according to the deviation value.

7. The tunnel deformation monitoring method based on chain inertial vision measurement according to claim 6, characterized in that, After obtaining the deformation monitoring result, if the deviation value exceeds a preset warning threshold, triggering an alarm mechanism and performing visualization processing on the deformation monitoring result to obtain a three-dimensional deformation heat map.

Images