A tunnel three-dimensional fault detection and identification method and system

Abstract

The present invention discloses a tunnel three-dimensional fault detection and identification method and system, which relates to the technical field of tunnel defect detection, and comprises the following steps: acquiring tunnel three-dimensional data, tunnel 2D image data and current positioning information in real time; associating the tunnel three-dimensional data, tunnel 2D image data and current positioning information to form a current identification, and recording the current identification in real time; extracting suspected defects from the tunnel three-dimensional data and tunnel 2D image data respectively, and generating a corresponding tunnel three-dimensional suspected defect list and tunnel 2D image defect list; associating and reviewing the tunnel three-dimensional suspected defect list with the tunnel 2D image defect list through the current identification to confirm the authenticity of the three-dimensional suspected defect; performing ellipse fitting calculation on the tunnel three-dimensional data corresponding to the real three-dimensional suspected defect to obtain suspected defect tunnel parameters, and comparing the suspected defect tunnel parameters with the reference parameters to generate a tunnel parameter comparison report.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel defect detection, and more particularly to a method and system for detecting and identifying three-dimensional faults in tunnels. Background Art

[0002] A tunnel is a passage excavated through an existing building or earth-rock structure. It is an engineering structure buried in the ground and represents a form of human utilization of underground space. Tunnels can be categorized as transportation tunnels, hydraulic tunnels, municipal tunnels, and mining tunnels. During tunnel operation, structural deformation detection is particularly important. Over long-term operation, tunnel walls can experience overall settlement and localized cross-sectional deformation due to vibration loads, temperature, and humidity, posing significant risks to the normal operation and safety of vehicles within the tunnel.

[0003] Existing technologies primarily rely on manual inspections for railway tunnel defects. However, manual inspections suffer from slow speed, low efficiency, high rates of false positives and missed detections, and limited auxiliary information. While some tunnel inspection equipment has been commercially available both domestically and internationally, research reveals limitations such as limited inspection points or incomplete coverage. Furthermore, even more comprehensive tunnel inspection systems are prohibitively expensive.

[0004] A Chinese invention patent application, publication number CN113431631A, discloses a method and system for determining tunnel convergence deformation. The method includes: a preset step of dividing the tunnel into sections based on its construction data, providing a scanning unit and an environmental monitoring unit in each section, and installing a wind direction monitor at each tunnel entrance; an environmental monitoring step of collecting environmental data from each section through the environmental monitoring unit and determining whether there are any interference items exceeding the standard in each section; if so, proceeding to a section analysis step; if not, proceeding to a scanning step; a section analysis step of determining whether the distance between the section with the interference item exceeding the standard and the tunnel entrance is less than a preset distance; if not, activating the ventilation system of the section for a first preset time and then returning to the environmental monitoring step; if so, proceeding to a tunnel entrance analysis step. This application can improve the stability of non-contact detection while saving energy required to purify interference items exceeding the standard.

[0005] However, the method and system disclosed in the above patents have the following defects:

[0006] 1. A single method is used to judge tunnel deformation, and only one judgment is made at a time, resulting in low fault detection and identification accuracy.

[0007] 2. In the later detection, it is impossible to understand the changing trend of various parameters of the tunnel, which is not conducive to the later maintenance.

[0008] 3. When using conventional ellipse fitting for modeling, in actual fitting, there is usually noise of varying degrees in the data set, and even a large number of outliers, which will greatly reduce the fitting effect; at the same time, there is no guarantee that the fitting result will be an ellipse. Summary of the Invention

[0009] In order to overcome the defects existing in the above-mentioned prior art, the present invention discloses a three-dimensional tunnel fault detection and identification method and system. The purpose of the present invention is to solve the problems in the prior art such as low fault detection and identification accuracy, inability to understand the changing trends of various parameters of the tunnel, and low ellipse fitting effect.

[0010] In order to achieve the above objectives, the present invention adopts the following technical solutions:

[0011] A method for detecting and identifying three-dimensional faults in a tunnel comprises the following steps:

[0012] S1, Scanning and Positioning Association

[0013] Real-time acquisition of tunnel 3D data, tunnel 2D image data and current positioning information;

[0014] The tunnel three-dimensional data, the tunnel 2D image data and the current positioning information are associated to form a current identification, and the current identification is recorded in real time.

[0015] In this step, the tunnel is scanned to obtain three-dimensional data of the tunnel, and the current identification of the scan is recorded in real time, and the value of the current identification is identified in real time; at the same time, positioning information is obtained in real time, and the value of the current identification is associated with the positioning information, and the associated data is stored after association.

[0016] The above steps are implemented using a system including a 3D scanning module, a control module, and a positioning module. The 3D scanning module and the positioning module move synchronously within the tunnel. At each scanning point, the 3D scanning module scans the tunnel to obtain 3D data for that point, records the current identifier for that scan in real time, and sends the current identifier to the control module. The control module queries the value of the current identifier in real time. The current identifier can be an encoder value. Both the scanned 3D data and the recorded current identifier are stored. At each scanning point, the positioning module locates the current scan's positioning information and sends this positioning information to the control module. The control module associates the current identifier with the positioning information to determine the current position of the 3D scanning module, and further determines the actual tunnel location corresponding to the 3D data collected by each scan. This associated information is recorded in a local database to facilitate later data restoration. For example, after a defect is identified, its associated location, i.e., the specific location of the actual defect within the tunnel, can be determined based on its corresponding identifier.

[0017] In the present invention, when scanning a tunnel or when the positioning module is performing positioning, an identifier (i.e., a current identifier) ​​can be generated simultaneously, or one of the devices / modules can generate an identifier (i.e., a current identifier). This identifier can be any information used for positioning, such as time, kilometer markers, or a code value (if an encoder is not used). Based on the current identifier, the positioning module's positioning information is associated with the 3D data scanned by the scanner.

[0018] In the present invention, the above steps also utilize a 2D high-definition module. The 2D high-definition module, 3D scanner, and positioning module operate simultaneously. Positioning primarily involves associating 3D data with positioning information. Simultaneously, the 2D high-definition module is used to obtain a 2D image, which can be compared with suspected defects extracted from the 3D data in subsequent steps to confirm their authenticity. Suspected defects can also be extracted from the 2D image and a list of suspected defects can be generated.

[0019] S2. Calculation review of suspected defects

[0020] extracting suspected defects from the three-dimensional tunnel data and the 2D tunnel image data respectively, and generating a corresponding three-dimensional tunnel suspected defect list and a 2D tunnel image defect list;

[0021] The three-dimensional suspected defect list of the tunnel is associated with the 2D image defect list of the tunnel through the current identification to verify the authenticity of the three-dimensional suspected defect.

[0022] In the above steps, the 3D suspected defect list of the tunnel is correlated and reviewed with the 2D image defect list of the tunnel. The two areas are compared. If the two are identical or similar, they are considered to be defects. The suspected defect areas are reviewed again to confirm the authenticity of the suspected defects and improve the recognition accuracy of the defect areas.

[0023] The process of associating and reviewing the 3D tunnel suspected defect list with the 2D tunnel image defect list includes calculating the suspected defects corresponding to each type based on different types of data features in the 3D data, generating a corresponding suspected defect list, and then associating and reviewing the list with the corresponding types of features in the 2D tunnel image. The different types of data features include tunnel mutation area features, depth data features, and limit contour data features, as follows:

[0024] S21. Characteristics of tunnel mutation area

[0025] The tunnel mutation area is compared with the suspected defect measured by infrared temperature measurement to determine the area of ​​the tunnel with mutation and the size of the mutation area, and the mutation area is marked as follows:

[0026] S211. The three-dimensional tunnel data obtained in the scanning and positioning association step is cyclically taken out for a certain length for calculation, and the area of ​​the tunnel with a sudden change and the size of the sudden change area are determined based on the calculation results, and the sudden change area is marked.

[0027] S212: Measure the temperature of the object in the mutation area and the objects around the mutation area in real time, and record the temperature measurement area and the temperature measurement results.

[0028] S213. Based on the mutation region and the temperature measurement result, it is inferred whether there is a suspected defect in the mutation region, and a suspected defect list is generated for the region with suspected defects.

[0029] In the above steps, if it is a protruding steel bar defect, the temperature at the protruding steel bar defect area is different from the temperature of the surrounding cables or pipes. Based on the different temperature measurement results, it can be judged that there is a suspected defect in the mutation area, and a suspected defect list can be generated based on the inference results.

[0030] S214 , associating the suspected defects in the suspected defect list with the 2D high-definition image to confirm the authenticity of the suspected defects.

[0031] The aforementioned tunnel mutation region characterization is implemented using a system consisting of a control module, a 3D scanning module, a temperature measurement module, a calculation module, and a 2D HD module. The control module controls the 3D scanning module to scan the tunnel and obtain 3D data. During scanning, the collected 3D data and the current identification are stored simultaneously. The calculation module uses the 3D data to determine the tunnel mutation region and its size, and then marks it. The control module also controls the temperature measurement module to measure the temperature of objects in the mutation region and surrounding objects in real time. During temperature measurement, the collected regional and temperature information is stored simultaneously. The calculation module then uses the temperature and mutation region information to infer whether there are suspected defects in the mutation region and generates a list of suspected defects based on the inference results. Finally, the suspected defects in the list are linked to the high-definition image generated by the 2D HD module to confirm their authenticity.

[0032] S22. Deep Data Features

[0033] Based on the depth data features in the three-dimensional tunnel data, a tunnel depth map is generated, and regions of different color blocks are extracted from the tunnel depth map, and a corresponding suspected defect list is generated, as follows:

[0034] S221, generating a tunnel depth map using the depth data in the three-dimensional tunnel data obtained in the scanning positioning association step;

[0035] S222. Find regions of different color blocks in the tunnel depth map, and generate corresponding suspected defect lists based on the regions of different color blocks.

[0036] S223 , associating the suspected defects in the suspected defect list with the 2D high-definition image to confirm the authenticity of the suspected defects.

[0037] The aforementioned depth data features are implemented using a control module, a 3D scanning module, a calculation module, and a 2D HD module. The control module controls the 3D scanning module to scan the tunnel and obtain 3D data. During scanning, the acquired 3D data and the current identification are stored simultaneously. The calculation module generates a tunnel depth map based on the depth information contained in the 3D data. Within the tunnel depth map, different depth blocks have inconsistent color blocks. The calculation module combines these blocks to generate a corresponding list of suspected defects (e.g., missing blocks). Finally, the suspected defects in the suspected defect list are correlated with the high-definition image generated by the 2D HD module to confirm their authenticity.

[0038] S23, limit profile data features

[0039] According to the data features of the limit contour in the three-dimensional tunnel data, an ellipse fitting calculation is performed to obtain the limit contour, and the limit contour is compared with the standard contour to generate a corresponding suspected defect list.

[0040] The ellipse fitting calculation is performed to obtain the bounding contour, comprising the following steps:

[0041] Preset ellipse fitting model;

[0042] By optimizing the iterative ellipse fitting model, the parameters of the ellipse fitting model are calculated;

[0043] Based on the parameters, a bounding contour is obtained.

[0044] The optimized iterative ellipse fitting model includes:

[0045] Add adaptive weights to reduce the impact of noise in three-dimensional data on fitting results;

[0046] Add ellipse fitting constraints to ensure that the fitting result is an ellipse;

[0047] The iteration error threshold is preset, and the iteration is stopped when the error obtained by the iteration is less than the error threshold.

[0048] The steps of bounded contour data feature are as follows:

[0049] S231, performing ellipse fitting on the three-dimensional tunnel data obtained in the scanning and positioning association step.

[0050] The process of ellipse fitting is as follows:

[0051] Step 1: Initialize W = {1};

[0052] in, , represents the diagonal matrix operator, w i represents the adaptive weight, ,in, is a hyperparameter, and the general equation of the ellipse is:

[0053] ,

[0054] in, , and order:

[0055] ,

[0056] In the prior art, the parameters can be obtained by solving the following model: .

[0057] ,

[0058] in, .

[0059] However, there may be the following two problems in actual fitting:

[0060] 1) There is usually varying degrees of noise in the data set, and even a large number of outliers, which will greatly reduce the fitting effect.

[0061] 2) There is no guarantee that the fitting result will be an ellipse.

[0062] In the present invention, the general equation of the ellipse should be constrained: , to ensure that the fitting result is an ellipse rather than other quadratic curves. At the same time, the above-mentioned adaptive weights are introduced to reduce the impact of noise in the data on the fitting results.

[0063] Step 2: Solve the optimization model and get ;

[0064] The optimization model is as follows:

[0065] ,

[0066] In the optimization model:

[0067] ,

[0068] Among the above parameters, represents the anti-diagonal matrix operator, is a negative real number;

[0069] The above optimized mathematical model can be solved by semi-positive programming, matrix decomposition or gradient descent methods, and finally the parameter vector .

[0070] Step 3: Get Substitute into Step 1 and update W;

[0071] Step 4: Substitute the updated W into Step 2 to get ;

[0072] Step 5: Repeat Step 1-Step 4 above, and the results of the two iterations are When the error is less than the threshold, the iteration stops.

[0073] S232. After ellipse fitting, the bounding contour is calculated and compared with the standard contour. After the comparison, the comparison result is displayed as a contour in the interface, and a suspected defect list is generated for the part where the comparison result indicates a suspected defect;

[0074] S233 , associating the suspected defects in the suspected defect list with the 2D high-definition image to confirm the authenticity of the suspected defects.

[0075] The aforementioned limit contour data features are implemented using a control module, a 3D scanning module, a calculation module, and a 2D HD module. The control module controls the 3D scanning module to scan the tunnel and obtain 3D data. During the scan, the acquired 3D data and the current identification are stored simultaneously. The calculation module uses the 3D data to perform ellipse fitting, calculates the limit contour, compares it with the standard contour, and automatically generates a defect list (such as line intrusion defects). Finally, suspected defects in the suspected defect list are linked to the high-definition image of the 2D HD module to confirm the authenticity of the suspected defects.

[0076] S3. Tunnel parameter comparison test

[0077] An ellipse fitting calculation is performed on the three-dimensional tunnel data corresponding to the actual three-dimensional suspected defect to obtain suspected defect tunnel parameters, and the suspected defect tunnel parameters are compared with the reference parameters to generate a tunnel parameter comparison report.

[0078] In the above steps, the three-dimensional tunnel data corresponding to the real three-dimensional suspected defects obtained in the suspected defect calculation and review step are subjected to ellipse fitting. After the ellipse fitting is completed, the parameters of the tunnel are calculated, and a measurement curve is established with the kilometer mark as the horizontal axis and the tunnel parameters as the vertical axis. The measurement curve is then compared with the post-completion benchmark curve. The parameters whose difference with the benchmark curve is greater than the threshold are used to generate a tunnel parameter comparison report.

[0079] The parameters of the tunnel include horizontal axis, minor axis, out-of-roundness and major axis parameters.

[0080] The tunnel parameter comparison and detection steps described above are implemented using a control module, a 3D scanning module, and a calculation module. The control module controls the 3D scanning module to scan the tunnel and obtain 3D data. During the scan, the collected 3D data and the current identification are stored simultaneously. The calculation module uses the 3D data to perform ellipse fitting, using the same ellipse fitting method used in the aforementioned limit contour suspected defect calculation and verification steps. After ellipse fitting, the calculation module calculates the tunnel's horizontal axis, minor axis, out-of-roundness, and major axis parameters, and establishes a measurement curve with the kilometer mark as the horizontal axis and the aforementioned parameters as the vertical axis. The measurement curve is then compared with the post-completion benchmark curve. The degree of overlap allows the changing trends of various tunnel parameters to be viewed, and a parameter comparison report is generated (according to the standard, with a certain range of deviation from the standard). This parameter comparison report allows staff to quickly identify abnormalities in various tunnel parameters and their changing trends.

[0081] Based on the above-mentioned tunnel 3D fault detection and identification method, the present invention also provides a tunnel 3D fault detection and identification system, which includes a 3D scanning module, a comprehensive positioning module, a control module, a 2D high-definition module and a calculation module;

[0082] The three-dimensional scanning module is connected to the control module and the calculation module, and is used to obtain three-dimensional data of the tunnel in real time;

[0083] The integrated positioning module is connected to the control module and is used to obtain current positioning information in real time;

[0084] The control module is configured to associate the three-dimensional tunnel data and the 2D tunnel image data with the current positioning information to form a current identifier, and record the current identifier in real time;

[0085] The 2D high-definition module is connected to the computing module and is used to obtain 2D image data of the tunnel; and at the same time, through the current identifier, the 3D suspected defect list of the tunnel is associated with the 2D image defect list of the tunnel to verify the authenticity of the 3D suspected defect;

[0086] The calculation module is used to extract suspected defects from the tunnel three-dimensional data and the tunnel 2D image data respectively, and generate a corresponding tunnel three-dimensional suspected defect list and tunnel 2D image defect list; at the same time, after the 2D high-definition module confirms the authenticity of the three-dimensional suspected defects, the three-dimensional tunnel data corresponding to the real three-dimensional suspected defects are subjected to ellipse fitting calculation to obtain suspected defect tunnel parameters, and the suspected defect tunnel parameters are compared with the benchmark parameters to generate a tunnel parameter comparison report.

[0087] Furthermore, the detection and identification system also includes a temperature measurement module, which is connected to the control module and the calculation module respectively, and is used to receive the control signal of the control module, measure the temperature of objects in the mutation area and objects around the mutation area in real time, and send the measured temperature information to the calculation module.

[0088] Beneficial effects of the present invention:

[0089] 1. The three-dimensional fault detection and identification method for tunnels provided by the present invention uses tunnel mutation area features to detect and judge steel bar protrusion defects, depth data features to detect and judge block drop defects, and limit contour data features to detect and judge line intrusion defects. Multiple methods are used to infer suspected defects with a high recognition rate. After each suspected defect is inferred, the suspected defect is associated with a 2D high-definition image for defect review to confirm the authenticity of the suspected defect and improve the recognition accuracy of the defect area.

[0090] 2. The three-dimensional tunnel fault detection and identification method provided by the present invention performs ellipse fitting on the three-dimensional tunnel data. After the ellipse fitting is completed, the horizontal axis, minor axis, out-of-roundness and major axis parameters of the tunnel are calculated, and a measurement curve is established with the kilometer mark as the horizontal axis and the above parameters as the vertical axis. The measurement curve is then compared with the completed benchmark curve. The parameters that differ from the benchmark curve by more than a threshold are used to generate a tunnel parameter comparison report. By viewing the parameter comparison report, it is possible to quickly locate whether the various tunnel parameters are abnormal and the changing trends of the various parameters, thereby facilitating later tunnel maintenance based on the changing trends.

[0091] 3. The three-dimensional tunnel fault detection and identification method provided by the present invention introduces adaptive weights during ellipse fitting, which greatly reduces the impact of data noise on the fitting results; and adds constraints on the curve form of the fitting results in the fitting model to ensure that the fitting results are always elliptical. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] Figure 1 Schematic diagram of the method of the present invention;

[0093] Figure 2 A schematic diagram of the scanning and positioning association of the present invention;

[0094] Figure 3 Schematic diagram of the characteristics of the tunnel mutation area of ​​the present invention;

[0095] Figure 4 Schematic diagram of the depth data feature of the present invention;

[0096] Figure 5 Schematic diagram of the bounded contour data feature of the present invention;

[0097] Figure 6 Schematic diagram of tunnel parameter comparison detection according to the present invention. DETAILED DESCRIPTION

[0098] The following will provide a clear and complete description of the concept, specific structure and technical effects of the present invention in conjunction with the embodiments and drawings, so as to fully understand the purpose, features and effects of the present invention.

[0099] Example 1

[0100] A tunnel three-dimensional fault detection and identification method, such as Figure 1 As shown, the following steps are included:

[0101] S1, Scanning and Positioning Association

[0102] Real-time acquisition of tunnel 3D data, tunnel 2D image data and current positioning information;

[0103] The tunnel three-dimensional data, the tunnel 2D image data and the current positioning information are associated to form a current identification, and the current identification is recorded in real time.

[0104] In this embodiment, Figure 2 As shown, the scanning and positioning association steps are implemented using a system including a 3D scanning module, a control module, and a positioning module. The 3D scanning module and the positioning module move synchronously within the tunnel. At each scanning point, the 3D scanning module scans the tunnel to obtain 3D data for that point, records the current identifier for that scan in real time, and sends the current identifier to the control module. The control module queries the value of the current identifier in real time. The current identifier can be an encoder value. Both the scanned 3D data and the recorded current identifier need to be stored. At each scanning point, the positioning module locates the positioning information for the current scan and sends the positioning information to the control module. The control module associates the current identifier with the positioning information to determine the current position of the 3D scanning module, and further determines the actual tunnel location corresponding to the 3D data collected by each scan. The associated information is recorded in a local database to facilitate later data restoration. For example, after a defect is identified, its associated location, i.e., the specific location of the actual defect within the tunnel, can be determined based on its corresponding identifier.

[0105] S2. Calculation review of suspected defects

[0106] extracting suspected defects from the three-dimensional tunnel data and the 2D tunnel image data respectively, and generating a corresponding three-dimensional tunnel suspected defect list and a 2D tunnel image defect list;

[0107] The three-dimensional suspected defect list of the tunnel is associated with the 2D image defect list of the tunnel through the current identification to verify the authenticity of the three-dimensional suspected defect.

[0108] The correlating and reviewing of the three-dimensional suspected defect list of the tunnel with the defect list of the 2D image of the tunnel includes calculating the suspected defects corresponding to each type based on different types of data features in the three-dimensional data, generating a corresponding suspected defect list, and then correlating and reviewing them with the corresponding types of features in the 2D image of the tunnel.

[0109] In this embodiment, Figure 3-5 As shown, by associating suspected defects with 2D high-definition images for defect review, the authenticity of the suspected defects is confirmed and the recognition accuracy of the defective area is improved.

[0110] S3. Tunnel parameter comparison test

[0111] An ellipse fitting calculation is performed on the three-dimensional tunnel data corresponding to the actual three-dimensional suspected defect to obtain suspected defect tunnel parameters, and the suspected defect tunnel parameters are compared with the reference parameters to generate a tunnel parameter comparison report.

[0112] In the above steps, the three-dimensional tunnel data corresponding to the real three-dimensional suspected defects obtained in the suspected defect calculation and review step are subjected to ellipse fitting. After the ellipse fitting is completed, the parameters of the tunnel are calculated, and a measurement curve is established with the kilometer mark as the horizontal axis and the tunnel parameters as the vertical axis. The measurement curve is then compared with the post-completion benchmark curve. The parameters whose difference with the benchmark curve is greater than the threshold are used to generate a tunnel parameter comparison report.

[0113] In this embodiment, Figure 6 As shown, the tunnel parameter comparison and detection steps described above are implemented using a control module, a 3D scanner, and a calculation module. The control module controls the 3D scanner to scan the tunnel and obtain 3D data. During scanning, the collected 3D data and the current identification are stored simultaneously. The calculation module uses the 3D data to perform ellipse fitting, using the same ellipse fitting method used in the aforementioned limit contour suspected defect calculation and verification steps. After ellipse fitting, the calculation module calculates the tunnel's horizontal axis, minor axis, out-of-roundness, and major axis parameters, and establishes a measurement curve with the kilometer mark as the horizontal axis and the aforementioned parameters as the vertical axis. This measurement curve is then compared with the completed baseline curve. The degree of overlap allows the changing trends of various tunnel parameters to be viewed, and a parameter comparison report is generated (according to the standard, with a certain range of deviation from the standard). This parameter comparison report allows personnel to quickly identify any abnormalities in the tunnel parameters and their changing trends.

[0114] Example 2

[0115] This embodiment is further improved on the basis of embodiment 1, as shown in FIG. Figure 3 As shown, the different types of data features, including tunnel mutation area features, are as follows:

[0116] S211. The three-dimensional tunnel data obtained in the scanning and positioning association step is cyclically taken out for a certain length for calculation, and the area of ​​the tunnel with a sudden change and the size of the sudden change area are determined based on the calculation results, and the sudden change area is marked.

[0117] S212: Measure the temperature of the object in the mutation area and the objects around the mutation area in real time, and record the temperature measurement area and the temperature measurement results.

[0118] S213. Based on the mutation region and the temperature measurement result, it is inferred whether there is a suspected defect in the mutation region, and a suspected defect list is generated for the region with suspected defects.

[0119] In the above steps, if it is a protruding steel bar defect, the temperature at the protruding steel bar defect area is different from the temperature of the surrounding cables or pipes. Based on the different temperature measurement results, it can be judged that there is a suspected defect in the mutation area, and a suspected defect list can be generated based on the inference results.

[0120] S214 , associating the suspected defects in the suspected defect list with the 2D high-definition image to confirm the authenticity of the suspected defects.

[0121] In this embodiment, Figure 3 As shown, the aforementioned tunnel mutation region characterization is implemented using a system comprising a control module, a 3D scanner, an infrared thermometer, a computing module, and a 2D HD module. The control module controls the 3D scanner to scan the tunnel, obtaining 3D data. During scanning, the acquired 3D data and the current identification are stored simultaneously. The computing module uses the 3D data to determine the tunnel mutation region and its size, and then marks it. The control module also controls the infrared thermometer to measure the temperature of objects in the mutation region and surrounding areas in real time, storing the acquired region and temperature information simultaneously. The computing module then uses the temperature and mutation region information to infer whether a suspected defect exists in the mutation region and generates a list of suspected defects based on the inference results. Finally, the suspected defects in the list are correlated with the high-definition image from the 2D HD module to confirm their authenticity.

[0122] Example 3

[0123] This embodiment is further improved on the basis of embodiment 2, as shown in FIG. Figure 4 As shown, the different types of data features also include deep data features, which are as follows:

[0124] S221, generating a tunnel depth map using the depth data in the three-dimensional tunnel data obtained in the scanning positioning association step;

[0125] S222. Find regions of different color blocks in the tunnel depth map, and generate corresponding suspected defect lists based on the regions of different color blocks.

[0126] S223 , associating the suspected defects in the suspected defect list with the 2D high-definition image to confirm the authenticity of the suspected defects.

[0127] In this embodiment, Figure 4 As shown, the aforementioned depth data features are implemented using a control module, a 3D scanner, a computing module, and a 2D HD module. The control module controls the 3D scanner to scan the tunnel and obtain 3D data. During scanning, the acquired 3D data and the current identification are stored simultaneously. The computing module generates a tunnel depth map based on the depth information contained in the 3D data. Within the tunnel depth map, different depth blocks have inconsistent color. The computing module combines these blocks to generate a corresponding list of suspected defects (e.g., missing blocks). Finally, the suspected defects in the suspected defect list are correlated with the high-definition image generated by the 2D HD module to confirm their authenticity.

[0128] Example 4

[0129] This embodiment is further improved on the basis of embodiment 3, as shown in FIG. Figure 5 As shown, the different types of data features also include bounded outline data features, which are as follows:

[0130] S231. Perform ellipse fitting on the three-dimensional tunnel data obtained by scanning and positioning association.

[0131] S232. After ellipse fitting, the limit contour is calculated and compared with the standard contour. After the comparison, the comparison result is displayed as a contour in the interface, and a suspected defect list is generated for the part that is a suspected defect according to the comparison result.

[0132] S233 , associating the suspected defects in the suspected defect list with the 2D high-definition image to confirm the authenticity of the suspected defects.

[0133] In this embodiment, Figure 5 As shown, the aforementioned limit contour data features are implemented using a control module, a 3D scanner, a calculation module, and a 2D HD module. The control module controls the 3D scanner to scan the tunnel and obtain 3D data. During scanning, the acquired 3D data and the current identification are stored simultaneously. The calculation module uses the 3D data to perform ellipse fitting, calculates the limit contour, compares it with the standard contour, and automatically generates a defect list (such as line intrusion defects). Finally, suspected defects in the suspected defect list are linked to the high-definition image of the 2D HD module to confirm the authenticity of the suspected defects.

[0134] Example 5

[0135] This embodiment is further improved on the basis of embodiment 4, and the process of ellipse fitting is as follows:

[0136] Let the general equation of the ellipse be:

[0137] ,

[0138] make:

[0139] ,

[0140] Then the parameters can be obtained by optimizing the following model ,

[0141]

[0142] However, in actual fitting, there may be the following two problems:

[0143] 1) There is usually varying degrees of noise in the data set, and even a large number of outliers, which will greatly reduce the fitting effect.

[0144] 2) Constraints should be added to the general equation of the ellipse. , to ensure that the fitting result is an ellipse rather than other quadratic curves.

[0145] To solve the above problem, this embodiment first introduces adaptive weights to reduce the impact of noise in the data on the fitting results. The weights are designed as follows:

[0146] ,

[0147] in It is a hyperparameter and is set manually.

[0148] Further combined with the ellipse constraint, the original optimization model is rewritten as follows:

[0149] ,

[0150] In the optimization model:

[0151] ,

[0152] Among the above parameters, represents the diagonal matrix operator, represents the anti-diagonal matrix operator, is a very small negative real number, such as .

[0153] The above optimized mathematical model can be solved by semi-positive programming, matrix decomposition or gradient descent methods, and finally the parameter vector .

[0154] The specific fitting process is:

[0155] Step 1: Initialize W = {1};

[0156] Step 2: Solve the optimization model and get ;

[0157] Step 3: Get Substitute into Step 1 and update W;

[0158] Step 4: Substitute the updated W into Step 2 to get ;

[0159] Step 5: Repeat Step 1-Step 4 above, and the results of the two iterations are When the error is less than the threshold, the iteration stops.

[0160] In this embodiment, adaptive weights are introduced during fitting, which greatly reduces the impact of data noise on the fitting results; constraints on the curve form of the fitting results are added to the fitting model to ensure that the fitting results are always elliptical.

[0161] In the ellipse equation described above, x and y are the coordinates of the scanned points in the 3D data (z is a constant value for each scan, so the Z coordinate is not considered during fitting). ABCDEF are the coefficients of a general ellipse (quadratic curve) equation. The goal of ellipse fitting is to calculate accurate and fixed coefficients of the ellipse equation. The optimization model is an iterative process until the ellipse equation equals 0. The purpose of the iteration is to fit the ellipse and determine the ellipse parameter values. During this iterative process, adaptive weights are introduced. The larger the weight, the more effective the fitting process becomes. The weight ranges from 0 to 1.

[0162] Example 6

[0163] Based on the above embodiment 5, this embodiment provides a tunnel three-dimensional fault detection and identification system, such as Figure 2-6 As shown, it includes a 3D scanning module, a comprehensive positioning module, a control module, a 2D high-definition module, a calculation module and a temperature measurement module;

[0164] The three-dimensional scanning module is connected to the control module and the calculation module, and is used to obtain three-dimensional data of the tunnel in real time;

[0165] The integrated positioning module is connected to the control module and is used to obtain current positioning information in real time;

[0166] The control module is configured to associate the three-dimensional tunnel data and the 2D tunnel image data with the current positioning information to form a current identifier, and record the current identifier in real time;

[0167] The 2D high-definition module is connected to the computing module and is used to obtain 2D image data of the tunnel; and at the same time, through the current identifier, the 3D suspected defect list of the tunnel is associated with the 2D image defect list of the tunnel to verify the authenticity of the 3D suspected defect;

[0168] The calculation module is used to extract suspected defects from the three-dimensional tunnel data and the 2D tunnel image data, respectively, and generate a corresponding three-dimensional tunnel suspected defect list and a tunnel 2D image defect list; and after the 2D high-definition module confirms the authenticity of the three-dimensional suspected defects, perform ellipse fitting calculation on the three-dimensional tunnel data corresponding to the real three-dimensional suspected defects to obtain suspected defect tunnel parameters, and compare the suspected defect tunnel parameters with the reference parameters to generate a tunnel parameter comparison report;

[0169] The temperature measurement module is connected to the control module and the calculation module respectively, and is used to receive the control signal of the control module, measure the temperature of objects in the mutation area and objects around the mutation area in real time, and send the measured temperature information to the calculation module.

[0170] The above is a detailed description of the embodiments of the present invention, but the present invention is not limited to the embodiments. Those skilled in the art may make various equivalent modifications or substitutions without departing from the spirit of the present invention. These equivalents or substitutions are all included in the scope defined by the claims of the present invention.

Claims

1. A three-dimensional tunnel fault detection and identification method, characterized in that: The following steps are involved: Real-time acquisition of tunnel 3D data, tunnel 2D image data and current positioning information; Associating the tunnel 3D data and tunnel 2D image data with the current positioning information to form a current identifier, and recording the current identifier in real time; extracting suspected defects from the three-dimensional tunnel data and the 2D tunnel image data respectively, and generating a corresponding three-dimensional tunnel suspected defect list and a 2D tunnel image defect list; By using the current identifier, the three-dimensional suspected defect list of the tunnel is correlated and reviewed with the 2D image defect list of the tunnel to confirm the authenticity of the three-dimensional suspected defect; Performing ellipse fitting calculation on the three-dimensional tunnel data corresponding to the actual three-dimensional suspected defect to obtain suspected defect tunnel parameters, and comparing the suspected defect tunnel parameters with the reference parameters to generate a tunnel parameter comparison report; The correlating and reviewing the three-dimensional suspected defect list of the tunnel with the defect list of the 2D image of the tunnel includes calculating the suspected defects corresponding to each type based on different types of data features in the three-dimensional data, generating a corresponding suspected defect list, and then correlating and reviewing them with the corresponding types of features in the 2D image of the tunnel; wherein the different types of data features include tunnel mutation area features, depth data features, and limit contour data features.

2. The tunnel three-dimensional fault detection and identification method according to claim 1, characterized in that: The different types of data features include: bounded contour data features; According to the data features of the limit contour in the three-dimensional tunnel data, an ellipse fitting calculation is performed to obtain the limit contour, and the limit contour is compared with the standard contour to generate a corresponding suspected defect list.

3. The tunnel three-dimensional fault detection and identification method according to claim 2, characterized in that: The ellipse fitting calculation is performed to obtain the bounding contour, comprising the following steps: Preset ellipse fitting model; By optimizing the iterative ellipse fitting model, the parameters of the ellipse fitting model are calculated; Based on the parameters, a bounding contour is obtained.

4. The method for detecting and identifying three-dimensional faults in tunnels according to claim 3, characterized in that: The optimized iterative ellipse fitting model includes: Add adaptive weights to reduce the impact of noise in three-dimensional data on fitting results; Add ellipse fitting constraints to ensure that the fitting result is an ellipse; The iteration error threshold is preset, and the iteration is stopped when the error obtained by the iteration is less than the error threshold.

5. The method for detecting and identifying three-dimensional faults in tunnels according to claim 4, characterized in that: The process of ellipse fitting is as follows: Step 1: Initialize W = {1}; in, , represents the diagonal matrix operator, w i represents the adaptive weight, ,in, is a hyperparameter, and the general equation of the ellipse is: ， in, , and order: ， Step 2: Solve the optimization model and get ; The optimization model is as follows: ， In the optimization model: ， Among the above parameters, represents the anti-diagonal matrix operator, is a negative real number; Step 3: Get Substitute into Step 1 and update W; Step 4: Substitute the updated W into Step 2 to get ; Step 5: Repeat Step 1-Step 4 above, and the results of the two iterations are When the error is less than the threshold, the iteration stops.

6. The method for detecting and identifying three-dimensional faults in tunnels according to claim 1, characterized in that: The different types of data features include: tunnel mutation area features; The tunnel mutation area is compared with the suspected defect measured by infrared temperature measurement to determine the area of ​​the tunnel with mutation and the size of the mutation area, and the mutation area is marked as follows: S211, looping and calculating a certain length of the three-dimensional tunnel data obtained in the scanning and positioning association step, determining the area of ​​the tunnel with a sudden change and the size of the sudden change area based on the calculation results, and marking the sudden change area; S212, measuring the temperature of the object in the mutation area and the objects around the mutation area in real time, and recording the temperature measurement area and the temperature measurement results; S213. Inferring whether there is a suspected defect in the mutation area based on the mutation area and the temperature measurement result, and generating a suspected defect list for the area with suspected defects; S214 , associating the suspected defects in the suspected defect list with the 2D high-definition image to confirm the authenticity of the suspected defects.

7. The method for detecting and identifying three-dimensional faults in tunnels according to claim 1, characterized in that: The different types of data features include: deep data features; Based on the depth data features in the three-dimensional tunnel data, a tunnel depth map is generated, and regions of different color blocks are extracted from the tunnel depth map, and a corresponding suspected defect list is generated, as follows: S221, generating a tunnel depth map using the depth data in the three-dimensional tunnel data obtained in the scanning positioning association step; S222. Find regions of different color blocks in the tunnel depth map, and generate corresponding suspected defect lists based on the regions of different color blocks. S223 , associating the suspected defects in the suspected defect list with the 2D high-definition image to confirm the authenticity of the suspected defects.

8. The method for detecting and identifying three-dimensional faults in tunnels according to claim 1, characterized in that: The three-dimensional tunnel data corresponding to the actual three-dimensional suspected defect is subjected to ellipse fitting calculation to obtain suspected defect tunnel parameters, and the suspected defect tunnel parameters are compared with the reference parameters to generate a tunnel parameter comparison report, including: The three-dimensional tunnel data corresponding to the real three-dimensional suspected defects obtained in the suspected defect calculation and review step are subjected to ellipse fitting. After the ellipse fitting is completed, the tunnel parameters are calculated, and a measurement curve is established with the kilometer mark as the horizontal axis and the tunnel parameters as the vertical axis. The measurement curve is then compared with the post-completion benchmark curve. The parameters whose difference with the benchmark curve is greater than a threshold are used to generate a tunnel parameter comparison report.

9. A tunnel three-dimensional fault detection and identification system according to the tunnel three-dimensional fault detection and identification method according to any one of claims 1 to 8, characterized in that: Including 3D scanning module, integrated positioning module, control module, 2D high-definition module and calculation module; The three-dimensional scanning module is connected to the control module and the calculation module, and is used to obtain three-dimensional data of the tunnel in real time; The integrated positioning module is connected to the control module and is used to obtain current positioning information in real time; The control module is configured to associate the three-dimensional tunnel data and the 2D tunnel image data with the current positioning information to form a current identifier, and record the current identifier in real time; The 2D high-definition module is connected to the computing module and is used to obtain 2D image data of the tunnel; and at the same time, through the current identifier, the 3D suspected defect list of the tunnel is associated with the 2D image defect list of the tunnel to verify the authenticity of the 3D suspected defect; The calculation module is used to extract suspected defects from the tunnel three-dimensional data and the tunnel 2D image data respectively, and generate a corresponding tunnel three-dimensional suspected defect list and tunnel 2D image defect list; at the same time, after the 2D high-definition module confirms the authenticity of the three-dimensional suspected defects, the three-dimensional tunnel data corresponding to the real three-dimensional suspected defects are subjected to ellipse fitting calculation to obtain suspected defect tunnel parameters, and the suspected defect tunnel parameters are compared with the benchmark parameters to generate a tunnel parameter comparison report.

10. The three-dimensional tunnel fault detection and identification system according to claim 9, characterized in that: The detection and identification system also includes a temperature measurement module, which is connected to the control module and the calculation module respectively, and is used to receive control signals from the control module, measure the temperature of objects in the mutation area and objects around the mutation area in real time, and send the measured temperature information to the calculation module.

Images