Comprehensive detection method and device for tunnel defects and storage medium

Abstract

The invention discloses a comprehensive detection method and device for tunnel defects and a storage medium, and relates to the technical field of tunnel defect detection.The method comprises the steps that a corresponding first tunnel defect detection result is determined based on a laser grey-scale map, and a corresponding second tunnel defect detection result is determined based on a visible light image; performing consistency matching on the first tunnel defect detection result and the second tunnel defect detection result, and determining a surface defect fusion detection result of the tunnel; determining a tunnel internal defect detection result based on the radar oscillogram data; and performing spatial position consistency matching on the tunnel internal defect detection result and the surface defect fusion result, and fusing the tunnel internal defect detection result and the surface defect fusion detection result to generate a multi-source fusion defect detection result of the tunnel. Therefore, the technical problem of low recognition reliability caused by isolation of multi-source tunnel defect detection data in the prior art is solved, and the reliability of overall tunnel defect recognition is improved.

Description

Technical Field

[0001] This application relates to the field of tunnel defect detection technology, and in particular to a comprehensive detection method, equipment and storage medium for tunnel defects. Background Technology

[0002] In current tunnel defect detection methods, visible surface defects such as cracks, water accumulation, and seepage are mainly detected using visible light cameras or manual inspections, with some cracks supplemented by laser detection. For invisible surface defects such as lining voids, insufficient density, and inadequate thickness, radar detection is primarily used. Stress detection is employed to detect potential rockfalls at tunnel entrances. These detection methods operate independently, acquiring data for only one type of defect. They lack effective temporal and spatial correlation, forming isolated "data silos" and hindering data integration and analysis. Surface defect detection, relying on images or laser data from a single device, is prone to false positives and false negatives due to factors such as overhead contact line obstruction, environmental contamination, glare, and icing within the tunnel. Furthermore, it cannot cross-validate defect authenticity through multi-source data. Simultaneously, the detection data for internal and surface defects are fragmented, making it difficult to determine spatial correlations and potential composite risks, directly impacting the accuracy and comprehensiveness of tunnel defect detection.

[0003] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main objective of this application is to provide a comprehensive method, device, and storage medium for detecting tunnel defects, aiming to solve the technical problem of low reliability in tunnel defect identification caused by the isolation of multi-source tunnel defect detection data in the prior art.

[0005] To achieve the above objectives, this application proposes a comprehensive detection method for tunnel defects, which includes: Acquire the laser grayscale image of the tunnel output by the laser camera, the visible light image of the tunnel output by the high-definition camera, and the radar waveform data of the tunnel output by the radar detection device; The laser grayscale image is input into the first tunnel surface defect recognition model to generate the corresponding first tunnel defect detection result, and the visible light image is input into the second tunnel surface defect recognition model to generate the corresponding second tunnel defect detection result. The consistency of the first tunnel defect detection result and the second tunnel defect detection result is matched, and when it is determined that the first tunnel defect detection result and the second tunnel defect detection result belong to the same tunnel defect, the surface defect fusion detection result of the tunnel is determined; The radar waveform data is input into the tunnel internal defect identification model to generate the corresponding tunnel internal defect detection results. Spatial position consistency matching is performed on the tunnel internal defect detection results and the surface defect fusion results. When it is determined that there are overlapping defect areas in the tunnel defects corresponding to the tunnel internal defect detection results and the surface defect fusion detection results, the multi-source fusion defect detection results of the tunnel are fused to generate the tunnel.

[0006] In one embodiment, the step of matching the consistency between the first tunnel defect detection result and the second tunnel defect detection result includes: Extract the first defect category and the first defect location coordinates of the first tunnel defect from the first tunnel defect detection result, and extract the second defect category and the second defect location coordinates of the second tunnel defect from the second tunnel defect detection result; Determine whether the first defect category and the second defect category are the same; If they are the same, map the first defect location coordinates and the second defect location coordinates to a unified three-dimensional coordinate system to obtain the third defect location coordinates of the first tunnel defect and the fourth defect location coordinates of the second tunnel defect. Calculate the distance difference between the coordinates of the third defect location and the coordinates of the fourth defect location. If the distance difference is less than a preset distance threshold, determine that the first tunnel defect detection result and the second tunnel defect detection result belong to the same tunnel defect.

[0007] In one embodiment, the step of determining the surface defect fusion detection result of the tunnel when the first tunnel defect detection result and the second tunnel defect detection result belong to the same tunnel defect includes: If it is determined that the first tunnel defect detection result and the second tunnel defect detection result belong to the same tunnel defect, either the first tunnel defect detection result or the second tunnel defect detection result shall be determined as the surface defect fusion detection result of the tunnel. After the step of matching the spatial location consistency between the first tunnel defect detection result and the second tunnel defect detection result, the method further includes: If it is determined that the first tunnel defect detection result and the second tunnel defect detection result do not belong to the same tunnel defect, a laser defect detection report of the tunnel is generated based on the first tunnel defect detection result, which includes the first defect category, the third defect location coordinates and the first defect level of the first tunnel defect. Based on the second tunnel defect detection results, an image defect detection report of the tunnel is generated, which includes the second defect category, the fourth defect location coordinates, and the second defect level.

[0008] In one embodiment, the step of fusing and generating a multi-source fusion defect detection result for the tunnel when it is determined that the tunnel interior defect detection result and the surface defect fusion detection result correspond to overlapping defect areas includes: If it is determined that there is an overlapping defect area between the tunnel defects corresponding to the tunnel interior defect detection result and the surface defect fusion detection result, both the tunnel interior defect detection result and the surface defect fusion detection result are determined as the multi-source fusion defect detection result; After the step of matching the spatial location consistency of the tunnel interior defect detection results and the surface defect fusion results, the method further includes: If it is determined that the tunnel defects corresponding to the internal tunnel defect detection results and the surface defect fusion detection results do not overlap in the defect area, a joint surface defect detection report of the tunnel is generated based on the surface defect fusion detection results, and an internal tunnel defect detection report of the tunnel is generated based on the internal tunnel defect detection results.

[0009] In one embodiment, after determining that the tunnel defects corresponding to the internal tunnel defect detection results and the surface defect fusion detection results respectively overlap in a defect region, the method further includes: The system acquires real-time stress data from wireless stress sensors deployed in the overlapping defect area, and retrieves historical stress data, historical visible light image data, and historical radar waveform data of the overlapping defect area from the historical database. By combining the real-time stress data, the historical stress data, the historical visible light image data, and the historical radar waveform data, a stress degradation judgment result is generated; After the step of generating the multi-source fusion defect detection results of the tunnel, the method further includes: When the stress degradation judgment result indicates that degradation exists, a comprehensive degradation alarm initial detection report is generated based on the multi-source fusion defect detection result, which includes defect location coordinates, defect category, defect level and degradation trend warning information. A three-dimensional point cloud model of the tunnel output by the laser camera is obtained, and the tunnel defects in the three-dimensional point cloud model are labeled with the defect location coordinates, defect category and defect level in the multi-source fusion defect detection results of the tunnel, so as to generate a three-dimensional visualization defect model. The three-dimensional visualization defect model is embedded into the initial detection report of the comprehensive degradation alarm, and the comprehensive degradation alarm detection report is output.

[0010] In one embodiment, the step of generating a stress degradation judgment result by combining the real-time stress data, the historical stress data, the historical visible light image data, and the historical radar waveform data, and combining the real-time stress data and the historical stress data of the overlapping defect area, includes: Based on the historical stress data, the stress change time series data of the overlapping defect area are determined; Based on the historical visible light image data, extract the time-series data of the crack width in the overlapping defect region; Based on the historical radar waveform data, the internal defect time series data of the overlapping defect region is obtained by inversion. Based on the stress change time series data and the crack width time series data, a first correlation curve is constructed, and based on the stress change rate data and the void or non-dense volume time series data, a second correlation curve is constructed. Based on the first correlation curve and the second correlation curve, a defect degradation prediction model is established. The real-time stress data and the historical stress data are input into the defect deterioration prediction model, and the stress change rate prediction value of the overlapping defect area is predicted by the defect deterioration prediction model. If the predicted stress change rate exceeds the preset degradation threshold, the stress degradation judgment result is determined to be that degradation exists; otherwise, the stress degradation judgment result is determined to be that there is no degradation.

[0011] In one embodiment, the step of acquiring radar waveform data of the tunnel output by the radar detection device includes: The radar antenna of the radar detection device is controlled to transmit electromagnetic wave signals to the tunnel lining and receive the echo signals reflected by the tunnel lining based on the electromagnetic wave signals. During the transmission of the electromagnetic wave signal, the contact pressure between the radar antenna and the lining surface is monitored in real time by a pressure detection sensor, and the air pressure of the controllable airbag connected to the radar antenna is dynamically adjusted according to the feedback signal of the contact pressure to maintain the contact pressure within a preset range. The radar waveform data is generated based on the echo signal.

[0012] In one embodiment, the step of controlling the radar antenna of the radar detection device to transmit electromagnetic wave signals toward the tunnel lining includes: The operating frequency of the radar antenna is dynamically configured according to the type of the tunnel lining to be detected; wherein, if the type of the tunnel lining to be detected is an arch, the operating frequency of the radar antenna is configured to be 900MHz and 400MHz, and if the radar antenna is an inverted arch, the operating frequency of the radar antenna is configured to be 270MHz.

[0013] Furthermore, to achieve the above objectives, this application also proposes a comprehensive detection device for tunnel defects, the comprehensive detection device for tunnel defects comprising: The data acquisition module is used to acquire the laser grayscale image of the tunnel output by the laser camera, the visible light image of the tunnel output by the high-definition camera, and the radar waveform data of the tunnel output by the radar detection device. The surface defect detection module is used to input the laser grayscale image into the first tunnel surface defect recognition model to generate the corresponding first tunnel defect detection result, and to input the visible light image into the second tunnel surface defect recognition model to generate the corresponding second tunnel defect detection result. The matching module is used to perform consistency matching between the first tunnel defect detection result and the second tunnel defect detection result, and when it is determined that the first tunnel defect detection result and the second tunnel defect detection result belong to the same tunnel defect, the surface defect fusion detection result of the tunnel is determined. The internal defect detection module is used to input the radar waveform data into the tunnel internal defect identification model and generate the corresponding tunnel internal defect detection results. The fusion module is used to perform spatial position consistency matching between the tunnel internal defect detection results and the surface defect fusion results, and when it is determined that there are overlapping defect areas in the tunnel defects corresponding to the tunnel internal defect detection results and the surface defect fusion detection results, the module fuses and generates the multi-source fusion defect detection results of the tunnel.

[0014] Furthermore, to achieve the above objectives, this application also proposes a comprehensive detection device for tunnel defects, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the comprehensive detection method for tunnel defects as described above.

[0015] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the comprehensive detection method for tunnel defects as described above.

[0016] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the comprehensive detection method for tunnel defects as described above.

[0017] One or more technical solutions proposed in this application have at least the following technical effects: acquiring the laser grayscale image of the tunnel output by the laser camera, the visible light image of the tunnel output by the high-definition camera, and the radar waveform data of the tunnel output by the radar detection device, thereby realizing multi-source data acquisition covering the surface and interior. By fully leveraging the sensing advantages of different modalities, laser grayscale images are input into a first tunnel surface defect recognition model to generate corresponding first tunnel defect detection results, and visible light images are input into a second tunnel surface defect recognition model to generate corresponding second tunnel defect detection results. Next, the first and second tunnel defect detection results are matched for consistency. When it is determined that the first and second tunnel defect detection results belong to the same tunnel defect, the surface defect fusion detection result is determined, and radar waveform data is input into a tunnel internal defect recognition model to generate corresponding tunnel internal defect detection results. Spatial location consistency matching is performed between the tunnel internal defect detection results and the surface defect fusion results to associate surface and internal defects. When it is determined that the tunnel defects corresponding to the tunnel internal defect detection results and the surface defect fusion detection results have overlapping defect areas, a complete and highly reliable multi-source fusion defect detection result for the tunnel is generated. This upgrades the originally isolated and fragile single-point judgment to a robust, interpretable, and highly confident comprehensive defect detection, thereby solving the technical problem of low recognition reliability caused by isolated multi-source tunnel defect detection data in existing technologies. This improves the overall reliability of tunnel defect identification and provides a highly reliable basis for risk assessment and maintenance priority ranking. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart illustrating the first embodiment of the comprehensive detection method for tunnel defects in this application; Figure 2 A schematic diagram of the architecture of a comprehensive tunnel defect detection system provided in this application; Figure 3 A schematic diagram of the architecture of a wireless stress sensor provided in this application; Figure 4 A functional block diagram of a computer (integrated testing subsystem) provided in this application; Figure 5 This application provides a schematic diagram of the structure of a radar detection device. Figure 6 A schematic diagram of the structure of a pressure-controlled and angle-adjustable radar antenna provided in this application; Figure 7 A side view of a pressure-controlled and angle-adjustable radar antenna provided in this application; Figure 8 A top view of a pressure-controlled and angle-adjustable radar antenna provided in this application; Figure 9 A flowchart illustrating another comprehensive detection method for tunnel defects provided in this application; Figure 10 This is a schematic diagram of the modular structure of the integrated detection device for tunnel defects according to an embodiment of this application; Figure 11 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the comprehensive detection method for tunnel defects in this application embodiment. Detailed Implementation

[0021] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0022] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0023] The main solution of this application embodiment is as follows: acquiring a laser grayscale image of the tunnel output by a laser camera, a visible light image of the tunnel output by a high-definition camera, and radar waveform data of the tunnel output by a radar detection device; inputting the laser grayscale image into a first tunnel surface defect recognition model to generate a corresponding first tunnel defect detection result, and inputting the visible light image into a second tunnel surface defect recognition model to generate a corresponding second tunnel defect detection result; performing consistency matching on the first tunnel defect detection result and the second tunnel defect detection result, and determining the surface defect fusion detection result of the tunnel when it is determined that the first tunnel defect detection result and the second tunnel defect detection result belong to the same tunnel defect; inputting the radar waveform data into a tunnel internal defect recognition model to generate a corresponding tunnel internal defect detection result; performing spatial position consistency matching on the tunnel internal defect detection result and the surface defect fusion result, and fusing to generate a multi-source fusion defect detection result of the tunnel when it is determined that the tunnel defects corresponding to the tunnel internal defect detection result and the surface defect fusion detection result have overlapping defect areas.

[0024] In this embodiment, for ease of description, the following description will focus on the main body of the comprehensive detection system for tunnel defects.

[0025] Current tunnel defect detection technologies primarily rely on visible light cameras or manual inspections for surface-visible defects such as cracks, water accumulation, and seepage, with laser detection used for some cracks. For invisible surface defects like lining voids, loose lining, and insufficient thickness, radar detection is mainly used. Stress detection is employed to detect potential rockfalls at tunnel entrances. These detection methods operate independently, acquiring data for only one type of defect. They lack effective temporal and spatial correlation, forming isolated "data silos" and hindering data integration and analysis. Surface defect detection, in particular, relies on images or laser data from single devices, making it susceptible to false positives and false negatives due to factors such as overhead contact lines, environmental contamination, glare, and icing within the tunnel. Furthermore, it cannot cross-validate defect authenticity through multi-source data. Simultaneously, the detection data for internal and surface defects are fragmented, making it difficult to determine spatial correlations and potential composite risks, directly impacting the accuracy and comprehensiveness of tunnel defect detection.

[0026] This application provides a solution that firstly, simultaneously acquires laser grayscale images, visible light images, and radar waveform data of the tunnel, breaking the traditional fragmented detection data and laying the foundation for fusion analysis. Then, a dedicated model is used to extract preliminary detection results for surface and internal defects respectively. Subsequently, the fusion detection results of the two types of surface defects are matched for consistency. Only when the same tunnel defect is confirmed, the fusion detection results of the two types of surface defects are fused to generate the tunnel's surface defect fusion detection result. Furthermore, the surface defect fusion detection results are correlated with the tunnel's internal defect detection results through spatial location matching. If overlapping areas exist, the results are fused to generate the tunnel's multi-source fusion defect detection result, allowing data from different dimensions to corroborate each other. This overcomes the limitations of traditional methods where sensor data are fragmented. Through data collaboration and correlation fusion, it solves the technical problem of low identification reliability caused by isolated multi-source tunnel defect detection data in existing technologies, significantly improving the reliability and comprehensiveness of tunnel defect identification.

[0027] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a comprehensive tunnel defect detection device capable of performing the above functions. The following description uses a comprehensive tunnel defect detection system as an example to illustrate this embodiment and the subsequent embodiments.

[0028] Based on this, embodiments of this application provide a comprehensive detection method for tunnel defects, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the comprehensive detection method for tunnel defects in this application.

[0029] In this embodiment, the comprehensive detection method for tunnel defects includes steps 101-105: Step 101: Obtain the laser grayscale image of the tunnel output by the laser camera, the visible light image of the tunnel output by the high-definition camera, and the radar waveform data of the tunnel output by the radar detection device.

[0030] Specifically, this application applies a novel comprehensive detection system for tunnel defects, such as... Figure 2 As shown, a comprehensive tunnel defect detection system can include a computer (comprehensive detection subsystem), a ring-deployed high-definition camera system, a laser camera system, radar detection equipment, a wireless stress sensor system, and a wireless communication subsystem. The computer (comprehensive detection subsystem) consists of a computer and related algorithms, performing real-time analysis of image data acquired by the ring-deployed high-definition cameras. Related model training can be completed offline, identifying and analyzing cracks and seepage on the lining surface. Simultaneously, the comprehensive detection subsystem controls each subsystem (high-definition camera, laser camera, radar detection equipment, and wireless communication subsystem), including adjusting the angle and focus of the ring-deployed high-definition camera system; controlling the angle of the laser camera; and controlling the angle of the radar antenna and the pressure on the contact surface in the radar detection equipment. The ring-deployed high-definition camera system consists of multiple sets of high-resolution industrial visible light cameras (typically ≥5 megapixels) with automatic focusing and angle adjustment. The laser camera system consists of multiple sets of adjustable-angle laser cameras. The laser camera is a dedicated device integrating a laser light source and an imaging sensor, typically using a line scanning method. It generates images by emitting near-infrared laser light and receiving its diffuse reflection intensity, unaffected by ambient light. The radar detection equipment proposed in this application includes radar detection signal transmitting and receiving equipment, a corresponding radar antenna, and a retractable pneumatic radar antenna support rod, which can achieve multi-angle adaptive direction adjustment. The wireless stress sensor consists of a wireless stress detection sensor pre-installed at the tunnel entrance or a designated area, a wireless communication module, and a power supply module. (Reference) Figure 3 The wireless stress sensor system includes a stress detection module (strain gauge / strain rosette), a data receiving module, a data transmission module, a cloud server, and a monitoring platform. Strain gauges or strain rosettes are used to detect stress changes around cracks and are typically deployed around the crack. They are linked to a data acquisition card, which transmits the data changes from the strain gauges or rosettes to a receiving card. The receiving card connects to a 4G, 5G, or 5G-R mobile communication card via a RS-485 interface. The data is then transmitted to the cloud server via a public or private mobile network, and finally to the monitoring platform, which can be vehicle-mounted or fixed at a monitoring center. By monitoring stress changes around cracks over a long period, the system can provide early warnings of crack deterioration trends. The wireless communication subsystem consists of a wireless transceiver system used to communicate with the wireless stress detection sensor subsystem.

[0031] Furthermore, the functional block diagram of the computer (integrated testing subsystem) is as follows: Figure 4It includes functions such as human-computer interaction, current detection analysis and result display, historical data query and comparison, and exporting detection reports. The human-computer interaction function can include the display of configuration information, search information, and analysis information. Specifically, In the configuration information function, users can input basic tunnel attributes (such as tunnel name, mileage, length, single / double track type, and construction status) as well as measurement parameters (including the number of measuring lines, the height of each measuring line, and the length of the radar cross-section) and judgment rules (such as crack identification threshold and alarm level standards) to provide contextual basis for subsequent detection and analysis. The construction status mentioned above can include operation and construction in progress. The search information function supports quick location of target data by multiple dimensions such as tunnel name, cross-section mileage, defect number, measuring line number, or measurement date. The analysis information display function focuses on trend judgment. On the one hand, it retrieves visible light images or radar waveform sequences of the same location over many years and compares the evolution of crack width, expansion of seepage area, and development of voided areas in chronological order. On the other hand, it combines the strain gauge data deployed around the crack to compare the current stress value with historical time series data to generate a stress-time change curve to help determine whether the crack is in an accelerated deterioration stage. Building upon this foundation, the system presents real-time individual detection results from visible light, laser, and radar, as well as fused multi-source integrated detection results, through the "Current Detection Analysis and Result Display" module. The "Historical Data Query and Comparison" module automatically retrieves historical data based on user-defined search criteria and overlays it with the current results visually or compares them numerically. Finally, the "Export Detection Report" function automatically generates a structured detection report containing defect distribution, evolution trends, risk ratings, and maintenance recommendations, based on the user-specified query scope and format (e.g., PDF or Excel). All these functions are completed through a unified human-computer interaction interface, achieving a one-stop operation from parameter setting, data retrieval, intelligent analysis to report output.

[0032] Specifically, laser grayscale images are output data from laser cameras, quantifying the differences in brightness on the tunnel surface using different grayscale values ​​(0-255). The grayscale contrast between defective areas (such as cracks and water seepage) and normal areas is significant, allowing for precise characterization of the boundaries, extent, and morphology of defects. Visible light images are output data from high-definition cameras, providing high-resolution color or black-and-white images that visually reproduce the true condition of the tunnel surface, clearly presenting obvious surface defects, pollution, damage, and other details. Radar waveform data is output data from radar detection equipment, reflecting parameters such as electromagnetic wave propagation time and reflection intensity in waveform form. Different internal defects (such as voids and looseness) will form characteristic waveforms, serving as the core basis for identifying hidden (internal) defects in tunnels. This data can be used to invert hidden defects such as internal voids, looseness, and abnormal reinforcement.

[0033] In some embodiments, after the integrated tunnel defect detection system is activated, laser cameras, high-definition cameras, and radar detection equipment work synchronously and in parallel. As the detection vehicle carrying the integrated tunnel defect detection system travels along the tunnel, the laser cameras, high-definition cameras, and radar detection equipment collect full-coverage data of the tunnel detection area. The laser camera can scan the tunnel surface by emitting a laser beam, receive the reflected light signal, and convert it into grayscale values ​​to generate a laser grayscale image that accurately characterizes the contours of surface defects. The high-definition camera can capture the tunnel surface scene at a preset angle and focal length, and output a high-resolution visible light image that restores the true condition. The radar detection equipment is matched with a corresponding frequency antenna according to the detection area, emits electromagnetic waves to penetrate the tunnel lining, receives the reflected signals from the internal structure, and converts them into radar waveform data that reflects the characteristics of hidden defects. The three types of data (laser grayscale image, visible light image, and radar waveform data) are transmitted in real time to the computer (integrated detection subsystem) of the integrated tunnel defect detection system, providing comprehensive and complementary raw data for subsequent defect identification, cross-validation, and fusion analysis, ensuring that the detection covers all dimensions of defects on the tunnel surface and inside.

[0034] Step 102: Input the laser grayscale image into the first tunnel surface defect recognition model to generate the corresponding first tunnel defect detection result, and input the visible light image into the second tunnel surface defect recognition model to generate the corresponding second tunnel defect detection result.

[0035] Specifically, the first tunnel surface defect recognition model is a machine learning / deep learning model trained based on laser grayscale image features, such as a convolutional neural network (CNN) or YOLO series model. During offline training, laser grayscale images labeled with surface defects such as cracks and seepage are used as training samples to optimize the model parameters and adapt them to the defect recognition scenario of laser grayscale images. The first tunnel defect detection result is the output data of the first tunnel surface defect recognition model based on grayscale image features, including information such as whether defects exist on the tunnel surface, defect type (cracks / seepage), defect location coordinates, and defect level (classified by size / severity). The second tunnel surface defect recognition model is a dedicated recognition model trained based on visible light image features. During offline training, visible light images labeled with surface defects are used as training samples to adapt to defect feature extraction under visible light, thus complementing the training data and applicable scenarios of the first tunnel surface defect recognition model. The second tunnel defect detection result is the output data of the second tunnel surface defect identification model based on visible light features. The data format is consistent with the first tunnel defect detection result, including information such as whether there are defects on the tunnel surface, defect type (cracks / water seepage), defect location coordinates, and defect level (classified by size / severity).

[0036] In some embodiments, after receiving two types of raw data—laser grayscale images and visible light images—a dual-model parallel inference mechanism is initiated to ensure detection efficiency. The first tunnel surface defect identification model is a machine learning model (such as a convolutional neural network or YOLOv7) specifically trained based on laser grayscale image features. During offline training, its parameters have been optimized using a large number of laser grayscale image samples labeled with surface defects such as cracks and seepage. This allows it to accurately capture features such as grayscale abrupt changes and contour continuity in defect areas within the laser grayscale image. After the acquired laser grayscale image is input into the first tunnel surface defect identification model, the model automatically determines whether surface defects exist in the laser grayscale image through feature extraction and defect feature library matching processes. It then outputs a first tunnel defect detection result containing information on whether a defect exists on the tunnel surface, the defect type (crack / seepage), the defect location coordinates, and the defect level (classified by size / severity). Meanwhile, the second tunnel surface defect recognition model is a dedicated recognition model adapted to visible light images. Its training samples are high-resolution visible light images labeled with surface defects, and the model structure is optimized for features such as color differences and texture details in visible light images. Visible light images captured by a high-definition camera are input into the second tunnel surface defect recognition model. By analyzing the visual differences between defects and normal areas in the visible light image, the model completes defect recognition and outputs a second tunnel defect detection result in a format consistent with the first detection result. By adapting the feature advantages of two models to two types of images respectively, accurate preliminary extraction of surface defects is achieved. This leverages the sensitivity of laser grayscale images to subtle defects while utilizing the intuitive features of visible light images to assist in recognition. This provides high-quality single-source detection results to support subsequent cross-validation and improve the reliability of defect recognition. Parallel processing also helps improve detection efficiency.

[0037] Step 103: Perform consistency matching between the first tunnel defect detection results and the second tunnel defect detection results, and determine the surface defect fusion detection results of the tunnel when it is determined that the first tunnel defect detection results and the second tunnel defect detection results belong to the same tunnel defect.

[0038] Optionally, the steps for consistency matching between the first tunnel defect detection results and the second tunnel defect detection results include: Extract the first defect category and the first defect location coordinates from the first tunnel defect detection results, and extract the second defect category and the second defect location coordinates from the second tunnel defect detection results; Determine whether the first defect category and the second defect category are the same; If they are the same, map the coordinates of the first defect location and the coordinates of the second defect location to a unified three-dimensional coordinate system to obtain the coordinates of the third defect location of the first tunnel defect and the coordinates of the fourth defect location of the second tunnel defect. Calculate the distance difference between the coordinates of the third defect location and the coordinates of the fourth defect location. If the distance difference is less than a preset distance threshold, determine that the first tunnel defect detection result and the second tunnel defect detection result belong to the same tunnel defect.

[0039] Specifically, the first and second defect categories refer to the defect types identified in the first and second tunnel defect detection results, such as cracks and water seepage. The first and second defect location coordinates refer to the location data (e.g., pixel coordinates) of the defects marked in the first and second detection results within their respective original image coordinate systems. The unified three-dimensional coordinate system can be a standardized spatial coordinate system based on tunnel mileage, height, and lateral width, used to unify the location reference for data collected by different devices. The third and fourth defect location coordinates are the standardized location data of the first and second defect location coordinates in the unified three-dimensional coordinate system after coordinate transformation. The preset distance threshold is a pre-set distance critical value based on detection accuracy requirements, used to determine whether the locations of two tunnel defects overlap (e.g., 5cm, adjustable as needed).

[0040] For example, the first defect category (such as cracks, seepage, etc.) and the corresponding first defect location coordinates (original pixel coordinates of the laser grayscale image) are accurately extracted from the first tunnel defect detection results. Simultaneously, the second defect category and second defect location coordinates (original pixel coordinates of the visible light image) are extracted from the second tunnel defect detection results. This clarifies the core feature parameters of the two types of detection results, laying the foundation for subsequent comparison. Next, a consistency judgment is made between the first and second defect categories. If the first and second defect categories are different (e.g., one is a crack and the other is seepage), it can be directly determined that the two types of detection results (the first tunnel defect detection results and the second tunnel defect detection results) do not point to the same tunnel defect, terminating the subsequent matching process. If the first and second defect categories are consistent, the location coordinate consistency verification stage begins. Subsequently, a coordinate transformation mechanism is initiated, mapping the coordinates of the first and second defects to a pre-established unified three-dimensional coordinate system for the tunnel. During the transformation, the differences between the coordinate systems of different devices are eliminated by considering the installation positions, shooting angles, and tunnel mileage calibration parameters of the laser and high-definition cameras, resulting in standardized coordinates for the third defect (corresponding to the first defect) and the fourth defect (corresponding to the second defect). Then, the difference in the straight-line distance between the third and fourth defect coordinates in three-dimensional space is calculated using a spatial distance calculation formula. This difference is compared with a preset distance threshold (set according to detection accuracy requirements, such as 5cm, which can be adjusted as needed). If the distance difference is less than the preset threshold, it indicates that the tunnel defects marked by the two types of detection results (the first tunnel defect detection result and the second tunnel defect detection result) highly overlap in space, and the first and second tunnel defect detection results are determined to belong to the same actual tunnel defect. If the distance difference is greater than or equal to the preset threshold, the tunnel defects marked by the first and second tunnel defect detection results are determined to be different tunnel defects.

[0041] Optionally, when determining that the first tunnel defect detection result and the second tunnel defect detection result belong to the same tunnel defect, the steps for determining the surface defect fusion detection result of the tunnel include: If the first tunnel defect detection result and the second tunnel defect detection result are determined to belong to the same tunnel defect, either the first tunnel defect detection result or the second tunnel defect detection result shall be determined as the surface defect fusion detection result of the tunnel.

[0042] Specifically, the surface defect fusion detection result is the final detection result selected from the first and second tunnel defect detection results after consistency matching verification. It is used to characterize the actual defects on the tunnel surface. Its data format is consistent with the single detection result, including core information such as defect category, defect location coordinates, defect level, and defect size. It will be used for subsequent spatial matching with internal defects and generation of comprehensive reports, which is a manifestation of the results of multi-source collaboration.

[0043] For example, after consistency matching determines that the first tunnel defect detection result and the second tunnel defect detection result point to the same tunnel surface defect, the surface defect fusion detection result generation process is initiated. Since the two types of detection results have been confirmed to be from the same source through dual verification of category and location, and both can accurately characterize the core information of the tunnel defect (category, location, level, etc.), either one (the first tunnel defect detection result and the second tunnel defect detection result) can be directly selected as the final surface defect fusion detection result. In addition, the selection can be flexibly chosen based on the needs of the detection scenario. For example, the first tunnel defect detection result corresponding to the laser grayscale image can be selected first to accurately reflect the subtle defect features, or the second tunnel defect detection result corresponding to the visible light image can be selected to restore the intuitive state of the defect. Under the premise of ensuring the reliability of the results, the fusion process is simplified, the detection efficiency is improved, and unified and high-quality surface defect data is formed, laying a solid foundation for subsequent spatial correlation matching with internal defect detection results and generation of multi-source fusion defect detection results.

[0044] Optionally, after the step of matching the spatial location consistency between the first tunnel defect detection results and the second tunnel defect detection results, the method further includes: If it is determined that the first tunnel defect detection result and the second tunnel defect detection result do not belong to the same tunnel defect, a laser defect detection report of the tunnel is generated based on the first tunnel defect detection result, which includes the first defect category, the third defect location coordinates, and the first defect level of the first tunnel defect. Based on the second tunnel defect detection results, an image defect detection report of the tunnel is generated, which includes the second defect category, the fourth defect location coordinates, and the second defect level.

[0045] Specifically, the laser defect detection report is a specialized report generated solely based on the first tunnel defect detection results, focusing on surface defects identified by the laser camera that do not match the visible light detection results. The image defect detection report is a specialized report generated solely based on the second tunnel defect detection results, focusing on surface defects identified by the high-definition camera that do not match the laser detection results. The first / second defect levels refer to the levels classified according to the size, extent, and severity of the defects in the first and second tunnel defect detection results, such as minor, moderate, and severe.

[0046] For example, after completing the spatial location consistency matching of the first and second tunnel defect detection results, if it is determined that the two (the first and second tunnel defect detection results) do not belong to the same tunnel defect, a parallel generation process for dual-report is initiated. First, core effective information can be extracted from the first tunnel defect detection result, including the first defect category (such as minor cracks, local spalling, etc.), the third defect location coordinates after transformation to a unified three-dimensional coordinate system (precisely locating the spatial position of the defect), and the first defect level (such as minor, moderate, severe) classified according to preset standards. Following a preset standardized report template, the first defect category, third defect location coordinates, and first defect level of the first tunnel defect are structurally integrated to generate a laser defect detection report that independently identifies defects using a focused laser camera. Simultaneously, key data from the second tunnel defect detection result is extracted, covering the second defect category, the fourth defect location coordinates under a unified coordinate system, and the second defect level. Using the same report format specifications, an image defect detection report containing only defects independently identified by a high-definition camera is generated. By generating specialized categorized reports, the identification results of different testing equipment can be clearly distinguished, providing tunnel maintenance personnel with targeted defect tracing basis, facilitating the subsequent accurate formulation of maintenance plans, and further improving the practicality and operational guidance of the testing results.

[0047] Optionally, after the step of matching the spatial location consistency between the first tunnel defect detection results and the second tunnel defect detection results, the method further includes: If it is determined that the defect detection results of the first tunnel and the defect detection results of the second tunnel do not belong to the same tunnel defect, determine whether strain gauge detection was performed. If none is found, based on the first tunnel defect detection results, a laser defect detection report is generated for the tunnel containing the first defect category, the third defect location coordinates, and the first defect level. Similarly, based on the second tunnel defect detection results, an image defect detection report is generated for the tunnel containing the second defect category, the fourth defect location coordinates, and the second defect level. If so, obtain stress change data in the area where the tunnel defect is located, and determine whether the stress in the area where the tunnel defect is located has deteriorated based on the stress change data; If there is stress deterioration in the area where the tunnel defect is located, a corresponding first defect alarm detection report is generated by combining the first tunnel defect detection results and stress change data, and a corresponding second defect alarm detection report is generated by combining the second tunnel defect detection results and stress change data.

[0048] Step 104: Input the radar waveform data into the tunnel internal defect identification model to generate the corresponding tunnel internal defect detection results.

[0049] Specifically, the tunnel internal defect identification model is a machine learning or deep learning model, such as U-Net or SegNet, specifically trained based on radar waveform data features. During the training phase, radar waveforms labeled with internal defect types are used as samples to optimize model parameters, thereby accurately extracting defect features from the waveform data. This tunnel internal defect identification model is a defect detection model specifically designed to identify internal defects invisible to the tunnel surface. The tunnel internal defect detection result is the output data of the tunnel internal defect identification model based on radar waveform data, which can include core information such as the category of the tunnel internal defect (void, incomplete compaction, insufficient thickness, etc.), the location coordinates of the tunnel internal defect (spatial location in a unified three-dimensional coordinate system), the size range, and the level of the tunnel internal defect.

[0050] In some embodiments, after receiving radar waveform data transmitted by radar detection equipment, a preset tunnel internal defect identification model is activated. This tunnel internal defect identification model is a deep learning model trained specifically on radar waveform features. During the training phase, parameters have been optimized using a large number of labeled radar waveform samples with internal defects such as voids, incomplete compaction, and insufficient thickness. This model can accurately capture waveform abrupt changes, intensity anomalies, and other features corresponding to different defects. The preprocessed radar waveform data is input into the tunnel internal defect identification model. Through feature extraction and defect feature library matching, the model automatically analyzes signal changes in the waveform data caused by differences in the medium, accurately determines whether defects exist inside the tunnel, further identifies the defect type, calculates its spatial location coordinates, size range, and severity level in a unified three-dimensional coordinate system, and finally integrates the above core information to generate standardized and structured tunnel internal defect detection results. This provides accurate internal defect data support for subsequent spatial correlation matching with surface defect fusion results.

[0051] Step 105: Perform spatial location consistency matching between the tunnel internal defect detection results and the surface defect fusion results, and when it is determined that there are overlapping defect areas in the tunnel corresponding to the tunnel internal defect detection results and the surface defect fusion detection results, fuse them to generate the multi-source fusion defect detection results of the tunnel.

[0052] Optionally, the step of matching the spatial location consistency between the tunnel interior defect detection results and the surface defect fusion results includes: Extract the coordinates of the fifth defect location from the tunnel internal defect detection results, and extract the coordinates of the sixth defect location from the tunnel surface defect fusion results; The coordinates of the fifth and sixth defects are mapped to a unified three-dimensional coordinate system to obtain the corresponding coordinates of the seventh and eighth defects. Calculate the second distance difference between the coordinates of the seventh defect location and the coordinates of the eighth defect location. If the second distance difference is less than the second preset distance threshold, it is determined that there is an overlapping defect area between the tunnel defects corresponding to the tunnel internal defect detection results and the surface defect fusion detection results.

[0053] Optionally, when it is determined that there are overlapping defect areas between the tunnel defects corresponding to the internal defect detection results and the surface defect fusion detection results, the steps for fusing and generating multi-source fusion defect detection results for the tunnel include: If it is determined that there is an overlapping defect area between the tunnel defects corresponding to the tunnel internal defect detection results and the surface defect fusion detection results, both the tunnel internal defect detection results and the surface defect fusion detection results are determined as multi-source fusion defect detection results.

[0054] Specifically, overlapping defect areas refer to areas where surface defects (such as cracks) and internal defects (such as voids) intersect in spatial location within a unified three-dimensional coordinate system, or where the distance is less than the engineering-allowed correlation threshold (e.g., the projection range of a surface crack overlaps with the internal void in the depth direction, or the lateral offset is less than 0.2 m). This indicates a possible causal relationship between the two (e.g., voids lead to stress concentration and trigger cracking). Multi-source fusion defect detection results integrate the core information from internal tunnel defect detection results with surface defect fusion results, forming a comprehensive detection result that simultaneously includes surface defects, internal defects, and their spatial correlation. This result can be used for risk assessment, maintenance decisions, and trend analysis.

[0055] For example, when the system determines that the detection results of internal tunnel defects (such as voids or looseness) and the fusion detection results of surface defects (such as cracks or water seepage) confirmed by laser and visible light fusion have overlapping defect areas in a unified three-dimensional engineering coordinate system (i.e., the spatial distance between the two is less than a preset correlation threshold, for example, the lateral offset is ≤0.2 m and the depth direction covers the lining thickness range), a multi-source fusion generation operation is performed. Specifically, the system integrates the structured information of the internal and surface defects (including their respective categories, three-dimensional coordinates, geometric parameters, confidence levels, and defect levels) into a joint record, which is jointly marked as the multi-source fusion defect detection result. This record clearly reflects the coupling relationship of "surface appearance - internal cause" (e.g., "the longitudinal crack in the arch corresponds to voids 0.3 m below"), and can automatically increase the risk level. When internal and surface defects are determined to be physically related in space (i.e., "overlapping defect areas"), they are included as a whole defect entity in the comprehensive inspection results, rather than being recorded in isolation. This avoids treating intrinsically related surface and internal defects separately and ensures that high-risk complex defects are not underestimated. At the same time, it provides structured input for subsequent deterioration trend analysis (such as in conjunction with stress monitoring), significantly improving the scientific nature, completeness, and engineering guidance value of tunnel health status assessment.

[0056] Optionally, after the step of matching the spatial location consistency of the tunnel interior defect detection results and the surface defect fusion results, the method further includes: If it is determined that there are no overlapping defect areas between the tunnel defects corresponding to the internal defect detection results and the surface defect fusion detection results, a joint surface defect detection report is generated based on the surface defect fusion detection results, and an internal defect detection report is generated based on the internal defect detection results.

[0057] Specifically, the tunnel internal defect detection report is a specialized report generated based on the tunnel internal defect detection results when there is no overlap between internal and surface defects, focusing on complete information about hidden defects inside the tunnel. The surface defect joint detection report is a specialized report generated based on the surface defect fusion detection results when there is no overlap between internal and surface defects, focusing on reliable defect information on the tunnel surface after cross-validation.

[0058] For example, after matching the spatial location consistency between the tunnel interior defect detection results and the surface defect fusion detection results, if it is determined that the defects corresponding to the two (tunnel interior defect detection results and surface defect fusion detection results) do not overlap (i.e., the spatial distance exceeds a preset association threshold, and there is no physical or mechanical coupling relationship), the dual-report generation process is initiated. Specifically, core information from the surface defect fusion detection results is extracted, including defect category, location coordinates in a unified three-dimensional coordinate system, and defect level, and integrated into a joint surface defect detection report according to a standardized template. Simultaneously, key data such as internal defect category, defect location coordinates, size range, and defect level from the tunnel interior defect detection results are extracted, and a tunnel interior defect detection report is generated in the same format. Both reports (tunnel interior defect detection report and surface defect joint detection report) are organized by mileage interval, support sorting by defect level, and can be accompanied by original images, radar profiles, or three-dimensional point cloud models as supporting evidence. By classifying, archiving, and independently outputting defects that do not form a surface-to-interface coupling relationship, it is ensured that even if internal defects and surface defects have no spatial correlation, their respective effective detection results can still be fully recorded, clearly presented, and used for subsequent maintenance decisions. This provides targeted inspection basis for tunnel maintenance and improves the accuracy of operation and maintenance decisions.

[0059] Based on the comprehensive tunnel defect detection method provided in this application, a laser grayscale image of the tunnel is acquired by a laser camera, a visible light image of the tunnel is acquired by a high-definition camera, and radar waveform data of the tunnel is acquired by a radar detection device, thereby realizing multi-source data acquisition covering the surface and interior. By fully leveraging the sensing advantages of different modalities, laser grayscale images are input into a first tunnel surface defect recognition model to generate corresponding first tunnel defect detection results, and visible light images are input into a second tunnel surface defect recognition model to generate corresponding second tunnel defect detection results. Next, the first and second tunnel defect detection results are matched for consistency. When it is determined that the first and second tunnel defect detection results belong to the same tunnel defect, the surface defect fusion detection result is determined, and radar waveform data is input into a tunnel internal defect recognition model to generate corresponding tunnel internal defect detection results. Spatial location consistency matching is performed between the tunnel internal defect detection results and the surface defect fusion results to associate surface and internal defects. When it is determined that the tunnel defects corresponding to the tunnel internal defect detection results and the surface defect fusion detection results have overlapping defect areas, a complete and highly reliable multi-source fusion defect detection result for the tunnel is generated. This upgrades the originally isolated and fragile single-point judgment to a robust, interpretable, and highly confident comprehensive defect detection, thereby solving the technical problem of low recognition reliability caused by isolated multi-source tunnel defect detection data in existing technologies. This improves the overall reliability of tunnel defect identification and provides a highly reliable basis for risk assessment and maintenance priority ranking.

[0060] In some embodiments, after determining that the tunnel defects corresponding to the internal tunnel defect detection results and the surface defect fusion detection results have overlapping defect regions, the method further includes: The system acquires real-time stress data from wireless stress sensors deployed in overlapping defect areas and retrieves historical stress data, historical visible light image data, and historical radar waveform data of overlapping defect areas from historical databases. By combining real-time stress data, historical stress data, historical visible light image data, and historical radar waveform data, a stress degradation judgment result is generated.

[0061] Specifically, wireless stress sensors are wireless sensing devices deployed in key parts of the tunnel structure (such as overlapping defect areas). They can collect and transmit stress change data of the tunnel structure in real time to monitor the structural stress state of the defect area. Real-time stress data consists of the stress values ​​and trends of the overlapping defect area output by the wireless stress sensor at the current detection moment. Historical stress data is stored in a historical database, containing stress monitoring data of the overlapping defect area over past detection periods, including stress values ​​at different time points and long-term variation patterns. Historical visible light image data is stored in the historical database; these are visible light images of the overlapping defect area captured by a high-definition camera during past detection periods, which can be used to trace the historical development of surface defects. Historical radar waveform data is stored in the historical database; these are radar waveforms of the overlapping defect area captured by radar detection equipment during past detection periods, which can be used to trace the historical development of internal defects. The stress deterioration judgment result is a comprehensive analysis of real-time and historical multi-source data, determining whether the stress in the overlapping defect area continues to deteriorate, the rate of deterioration, and the corresponding structural safety risk level.

[0062] As an example, after determining the existence of an overlapping defect area, a multi-source data acquisition and retrieval process is initiated. Real-time stress data output by the wireless stress sensor in the overlapping defect area is acquired through the wireless communication subsystem. Simultaneously, historical stress data, historical visible light image data, and historical radar waveform data for the overlapping defect area are accurately extracted from the historical database. Subsequently, a time-series comparative analysis of the real-time and historical stress data is performed to analyze stress change trends. The evolution of surface defects is traced using historical visible light images, and the development pattern of internal defects is reconstructed based on historical radar waveforms. Finally, a pre-set stress degradation assessment algorithm is used to integrate multi-dimensional data to determine whether the stress in the overlapping defect area exhibits continuous increases, abrupt changes, or other degradation characteristics. Ultimately, a stress degradation judgment result is generated, including the stress degradation trend, rate, and corresponding safety risk level.

[0063] In some embodiments, after the step of generating the multi-source fusion defect detection results of the tunnel, the method further includes: When the stress degradation judgment result indicates that degradation exists, a comprehensive degradation alarm initial detection report is generated based on the multi-source fusion defect detection results, which includes defect location coordinates, defect category, defect level and degradation trend warning information. A 3D point cloud model of the tunnel output by a laser camera is obtained, and the tunnel defects in the 3D point cloud model are labeled with the defect location coordinates, defect category and defect level in the multi-source fusion defect detection results of the tunnel, and a 3D visualization defect model is generated. Embed the 3D visualized defect model into the initial detection report of the comprehensive degradation alarm, and output the comprehensive degradation alarm detection report.

[0064] Specifically, the 3D point cloud model is a 3D spatial data model of the tunnel acquired by laser cameras, which can accurately reconstruct the three-dimensional shape and spatial relationship of the tunnel structure. The 3D visualization defect model is based on the 3D point cloud model, using multi-source fusion defect detection results to mark the location, type, and risk level of defects (e.g., highlighting a Level III void area in red), achieving a visually intuitive spatial representation of defects. It can intuitively show the location, category, and level of defects in the tunnel's three-dimensional structure. The comprehensive deterioration alarm detection report is the final alarm report embedded in the 3D visualization defect model, combining data integrity with intuitive visualization.

[0065] As an example, when a multi-source fusion defect detection result is generated and the stress degradation judgment result indicates the presence of degradation, core data such as defect location coordinates, surface and internal defect categories, and defect levels are extracted based on this multi-source fusion defect detection result. Combined with early warning information such as stress degradation trends (e.g., "annual stress growth rate reaches 12%, crack width is expected to exceed the limit within 6 months") and risk levels, a comprehensive degradation alarm initial detection report is generated according to a preset template. Subsequently, a 3D point cloud model of the tunnel, synchronously acquired and reconstructed by a laser camera, is invoked. Using the defect location coordinates from the multi-source fusion defect detection result, visual markers are overlaid on the corresponding areas of the 3D point cloud model. For example, red semi-transparent blocks are used to indicate the void area, yellow line segments are used to mark the crack direction, and defect category and level labels are attached, thereby generating a 3D visualized defect model. Finally, this 3D visualized defect model is embedded into the comprehensive degradation alarm initial detection report in the form of interactive graphics (such as WebGL or PDF embedded in a 3D view), forming the final comprehensive degradation alarm detection report.

[0066] Optionally, after the step of fusing the multi-source fusion defect detection results of the generated tunnel, the method further includes: When the stress degradation judgment result is no degradation, an initial comprehensive inspection report of tunnel defects is generated based on the multi-source fusion defect detection results, which includes the defect location coordinates, defect category, and defect level. The tunnel's 3D point cloud model output by the laser camera is obtained, and the tunnel defects in the 3D point cloud model are labeled with the defect location coordinates, defect category and defect level in the tunnel's multi-source fusion defect detection results, generating a 3D visualized defect model. Embed the 3D visualized defect model into the initial comprehensive inspection report of tunnel defects, and output the comprehensive inspection report of tunnel defects.

[0067] In some embodiments, the step of generating a stress degradation judgment result by integrating real-time stress data, historical stress data, historical visible light image data, and historical radar waveform data, and combining real-time stress data with historical stress data of overlapping defect areas, includes: Based on historical stress data, determine the time series data of stress changes in the overlapping defect area; Based on historical visible light image data, extract time-series data of crack width in overlapping defect areas; Based on historical radar waveform data, the internal defect time series data of the overlapping defect area are obtained by inversion. A first correlation curve is constructed based on stress change time series data and crack width time series data, and a second correlation curve is constructed based on stress change rate data and void or non-dense volume time series data. Based on the first and second correlation curves, a defect degradation prediction model is established. Real-time stress data and historical stress data are input into the defect deterioration prediction model, and the stress change rate prediction value of the overlapping defect area is predicted by the defect deterioration prediction model. If the predicted stress change rate exceeds the preset degradation threshold, the stress degradation judgment result is determined to be that degradation exists; otherwise, the stress degradation judgment result is determined to be that there is no degradation.

[0068] Specifically, the crack width time-series data is extracted from historical visible light image data, showing the numerical sequence of surface crack width changes over time in the overlapping defect area. The internal defect time-series data is obtained by inverting historical radar waveform data, showing the time-series data of internal cavity volume, non-compacted area, etc., within the overlapping defect area (including cavity / non-compacted volume time-series data). The first correlation curve is a fitting curve characterizing the correlation between stress change time-series data and crack width time-series data, reflecting the correlation between surface defects and stress changes. The second correlation curve is a fitting curve characterizing the correlation between stress change rate data (the rate of stress change over time) and cavity / non-compacted volume time-series data, reflecting the correlation between internal defects and stress change rate. The stress change rate prediction value is the output of the defect deterioration prediction model, predicting the stress change rate of the overlapping defect area over a future period. The preset deterioration threshold is a critical value for stress change rate preset according to tunnel structure safety standards, used to determine whether there is a risk of deterioration.

[0069] As an example, historical stress data accumulated from previous inspections of the overlapping defect area are retrieved from a historical database, sorted by timestamp, and noise filtered to generate time-series stress change data reflecting the long-term mechanical response of the structure. Simultaneously, based on historical visible light image data of the overlapping defect area, image registration and sub-pixel edge detection algorithms are used to automatically measure the width of the same crack location, constructing time-series crack width data. Furthermore, using historical radar waveform data of the overlapping defect area, combined with an electromagnetic wave propagation model and a deep learning inversion network, the volume or area of ​​the voided or non-dense regions is calculated successively, forming time-series data of internal defects. Based on this, the stress change time-series data and crack width time-series data are synchronized and their correlation modeled to fit a first correlation curve (such as a power function or exponential growth model) describing the stress-crack propagation relationship. Similarly, regression analysis is performed on the first derivative of the stress time series (i.e., stress change rate data) and the internal defect volume time series to establish a second correlation curve reflecting the coupling law of stress growth rate and internal degradation. Subsequently, using the first and second correlation curves as physical constraints, a defect degradation prediction model (which can employ temporal neural networks such as LSTM or GRU) is trained. In actual operation, the system inputs the currently acquired real-time stress data and recent historical stress data into the defect degradation prediction model. The model then outputs a predicted stress change rate for the overlapping defect area within the next short period (e.g., 7 or 30 days). If this predicted value exceeds a preset degradation threshold set according to engineering specifications or expert experience, the stress degradation is determined to be present, triggering a high-level warning; otherwise, it is determined to be without degradation, and routine monitoring continues.

[0070] Optionally, after determining that there are overlapping defect areas corresponding to the tunnel defects detected by the internal defect detection results and the surface defect fusion detection results, the method further includes: Acquire real-time stress data from wireless stress sensors deployed in overlapping defect areas, and retrieve historical stress data for overlapping defect areas from a historical database. By combining real-time stress data and historical stress data, a stress deterioration judgment result is generated.

[0071] Optionally, the steps for generating stress degradation judgment results by combining real-time stress data and historical stress data include: Based on real-time stress data and historical stress data, the stress change rate sequence is determined; Each stress change rate in the stress change rate sequence is compared with a preset degradation threshold. If the stress change rate continues to exceed the degradation threshold for a preset time window, the stress degradation judgment result is determined to be that degradation exists; otherwise, the stress degradation judgment result is determined to be that there is no degradation.

[0072] In some embodiments, the step of acquiring radar waveform data of a tunnel output by a radar detection device includes: The radar antenna controlling the radar detection equipment transmits electromagnetic wave signals to the tunnel lining and receives the echo signals reflected by the tunnel lining based on the electromagnetic wave signals. During the transmission of electromagnetic wave signals, the contact pressure between the radar antenna and the lining surface is monitored in real time by a pressure detection sensor. Based on the feedback signal of the contact pressure, the air pressure of the controllable airbag connected to the radar antenna is dynamically adjusted to maintain the contact pressure within a preset range. Radar waveform data is generated based on the echo signal.

[0073] Specifically, regarding the radar detection equipment provided in this application, in order to improve detection efficiency and perform parallel detection on multiple survey lines, the radar detection equipment adopts an octopus structure, combining air pressure and motor adjustment to optimize the radar detection equipment and support flexible configuration of height and direction adjustments, such as... Figure 5 This is a schematic diagram of the optimized radar detection equipment. Figure 5 The medium-pressure, controllable, and angle-adjustable radar antenna is connected to the radar detection platform via a telescopic support rod. The radar antenna is connected to the radar signal processing unit within the radar detection platform via a radar feed line. The telescopic rods numbered 5 and 6 on the left and right sides can be adjusted downwards for bottom detection.

[0074] A schematic diagram of a pressure-controlled and angle-adjustable radar antenna is shown below. Figure 6 As shown. Figure 6 The system includes three directional motors, each controlling the movement of the rotating platform, including rotation, X-axis tilt, and Y-axis tilt, thereby rotating the radar antenna and controlling its tilt in the X-axis and Y-axis directions. To ensure the radar antenna is firmly attached to the object being inspected, a pressure-controlled airbag applies pressure behind it. The air pressure in the airbag is supplied by a cylinder on the radar detection platform, inflated through an air tube. A pressure sensor between the airbag and the radar antenna determines whether the antenna is in close contact with the object's surface. If the pressure decreases, the sensor transmits a pressure adjustment command to the cylinder, which then inflates the airbag. A telescopic support rod, adjustable via a motor at its base, facilitates the detection of survey lines at different heights within the tunnel.

[0075] refer to Figure 7 and Figure 8 , Figure 7 This is a side view of a radar antenna with controllable air pressure and adjustable angle. Figure 8This is a top view of a pressure-controlled, angle-adjustable radar antenna. A motor controlling rotation is mounted on a rotating platform and connected to it via gears, used to control the radar antenna's rotation. The rotating platform is connected to the platform via a metal support column. The top of the support column is spherical, and a bowl-shaped metal component at the center of the rotating platform fits into the spherical head of the support column, facilitating the platform's rotation and tilting along the X or Y axes. A motor controlling X-axis tilt is mounted at the end of a telescopic support rod. The motor controls the rotation of the X-axis connecting rod, raising or lowering it to adjust the X-axis tilt angle of the rotating platform. Similarly, a motor controlling Y-axis tilt is mounted at the end of a telescopic support rod. This motor controls the rotation of the Y-axis connecting rod, raising or lowering it to adjust the Y-axis tilt angle of the rotating platform.

[0076] The radar antenna in the radar detection equipment provided in this application can be configured differently according to different needs. For example, if it is for detecting the tunnel arch, an airbag is used to control the pressure of the radar antenna, and the radar antenna is a 900MHz or 400MHz antenna. If it is for detecting the invert arch, a ground-based antenna is used, with an antenna frequency of 270MHz. Alternatively, the airbag can be simplified, and an air-coupled antenna can be used.

[0077] In some embodiments, the step of controlling the radar antenna of the radar detection equipment to transmit electromagnetic wave signals toward the tunnel lining includes: The operating frequency of the radar antenna is dynamically configured according to the type of the tunnel lining to be inspected. For example, if the type of the tunnel lining to be inspected is an arch, the operating frequency of the radar antenna is configured to be 900MHz and 400MHz. If the radar antenna is an inverted arch, the operating frequency of the radar antenna is configured to be 270MHz.

[0078] Specifically, the radar antenna is the core transmitting and receiving component of radar detection equipment, and its operating frequency directly determines the penetration depth, resolution, and detection effect of electromagnetic waves. The tunnel lining is the load-bearing and protective structure of the tunnel, mainly including the arch (the arched structure at the top of the tunnel), the invert (the arched structure at the bottom of the tunnel), and the sidewalls. The component type refers to the specific structural part of the tunnel lining; here, the key distinction is made between the arch and the invert, two critical load-bearing components.

[0079] As an example, to optimize radar detection performance for different structural components, the system dynamically configures the frequency based on the type of tunnel lining component when controlling the radar antenna to transmit electromagnetic wave signals to the tunnel lining. Specifically, the lining component type of the current test area can be automatically identified by combining odometer positioning with tunnel cross-section design information or real-time point cloud segmentation. If it is determined to be an arch (top arc section), because its lining is relatively thin (typically 0.3–0.5 m) and high-precision identification of surface voids, micro-cracks, and other defects is required, the radar antenna is configured to operate in dual-frequency mode at 900 MHz and 400 MHz—900 MHz provides millimeter-level resolution to capture detailed anomalies, and 400 MHz ensures full thickness coverage. If it is determined to be an inverted arch (bottom reverse arch section), because the inverted arch structure is relatively thick and susceptible to groundwater influence, requiring detection of deep non-compact layers or base erosion, the system switches to a low-frequency mode of 270 MHz to obtain an effective penetration depth of more than 1.5 meters. The above dynamic configuration is automatically completed by the vehicle-mounted control system.

[0080] In some embodiments, a three-dimensional point cloud model of the tunnel output by a laser camera is obtained, and the three-dimensional point cloud model is compared with a preset tunnel design model to generate corresponding over-excavation and under-excavation defect detection data. If the over- or under-excavation defect detection data indicates that the tunnel has an encroachment area, a special tunnel defect report for the construction period will be generated based on the over- or under-excavation defect detection data. Among them, the encroachment area refers to the tunnel area where there is over-excavation or under-excavation. The special tunnel defect report during the construction period includes the coordinates of the encroachment area and the deviation of the encroachment area.

[0081] Specifically, the tunnel design model is a standardized 3D model built based on tunnel construction design drawings, which may include theoretical parameters such as the tunnel lining design thickness, cross-sectional dimensions, and axis position. Over-excavation / under-excavation defect detection data is generated by comparing the 3D point cloud model with the design model, covering the deviation values ​​between the actual and design shapes of various tunnel areas, the extent of over-excavation / under-excavation areas, coordinates, and other information. Encroachment areas are areas where, after actual tunnel construction, there is over-excavation (actual excavation size greater than design size) or under-excavation (actual excavation size less than design size) with deviations exceeding the allowable range, potentially affecting lining construction or structural safety. The construction-phase dedicated tunnel defect report is a specialized defect report generated for the construction phase, focusing on over-excavation / under-excavation and encroachment issues, providing direct evidence for construction adjustments.

[0082] As an example, the comprehensive tunnel defect detection method provided in this application can also be used to detect defects in tunnels under construction. A 3D point cloud model of the tunnel, scanned by a laser camera, is acquired. This 3D point cloud model completely records the actual internal shape and spatial coordinate data of the tunnel during construction. Subsequently, a model comparison process is initiated, importing the 3D point cloud model and a preset tunnel design model into the same 3D coordinate system. Using point cloud registration and geometric comparison algorithms, the dimensional deviation between the actual shape and the design shape is calculated region by region. Based on the comparison results, over-excavation and under-excavation defect detection data are generated to clarify whether over-excavation or under-excavation exists in each region and the specific deviation values. Next, the over-excavation and under-excavation defect detection results are judged. If the detection data shows areas with deviations exceeding the allowable construction range, it is determined that the tunnel has encroachment areas. At this point, based on the over-excavation and under-excavation defect detection data, the precise 3D coordinates of the encroachment areas, the over-excavation / under-excavation type, and the corresponding encroachment area deviation can be extracted. Information is then integrated according to construction period control requirements to generate a dedicated tunnel defect report for the construction period. By leveraging the high precision of 3D laser scanning, tunnel over- or under-excavation defects during construction can be quickly and accurately identified, preventing potential structural safety hazards caused by these defects and ensuring that tunnel construction quality matches design standards.

[0083] refer to Figure 9 , Figure 9This is a flowchart illustrating a comprehensive method for detecting tunnel defects. The process is based on multi-source sensor collaboration and intelligent analysis to achieve comprehensive and accurate identification of tunnel structural defects. First, three methods—laser detection, visible light detection, and radar detection—are used to acquire the tunnel's 3D geometric data, laser grayscale images, visible light images, and radar waveform data, respectively. Laser detection uses laser 3D scanning to generate a 3D point cloud model and combines it with the laser grayscale image to identify surface defects such as cracks and water seepage. If over-excavation or under-excavation is found, the defect area is marked in the 3D point cloud model, and a corresponding defect detection report is output. Simultaneously, visible light detection uses a high-definition camera to acquire visible light images to determine the presence of cracks or water seepage and marks their location and severity. Radar detection identifies internal defects such as cavities and loose surfaces; if present, the defect area is marked. After completing a single-technology inspection, the results of laser and visible light inspections are first compared: it is determined whether the cracks / seepage identified by the two methods are in the same area. If they are different and strain gauge inspection is not required, each technology outputs its own defect inspection report. If strain gauge inspection is required, a report is output based on the strain data. If stress deterioration occurs around the crack, a deterioration alarm is also added. If the defects detected by laser and visible light are in the same area, radar inspection data is used to determine whether the defects identified by the three technologies are in the same area: if they are different, the crack / seepage and radar defects each output their own reports; if they are the same, it is further determined whether strain gauge inspection is required. If not, a combined defect inspection report using all three methods is output. If strain gauge inspection is required, stress data is collected simultaneously. If stress deterioration is detected, a comprehensive report with an additional deterioration alarm is output; otherwise, a standard comprehensive report is output. The entire process is based on the core logic of regional division + multi-technology cross-verification + data fusion + risk warning. It uses a three-dimensional point cloud map to uniformly display the defect location, which not only ensures the comprehensiveness and accuracy of defect detection, but also outputs deterioration alarms for high-risk areas, providing full-chain technical support for tunnel maintenance from defect identification to risk assessment.

[0084] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the comprehensive detection method for tunnel defects in this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0085] This application also provides a comprehensive detection device for tunnel defects, please refer to... Figure 10 The comprehensive detection device for tunnel defects includes: The data acquisition module 1010 is used to acquire the laser grayscale image of the tunnel output by the laser camera, the visible light image of the tunnel output by the high-definition camera, and the radar waveform data of the tunnel output by the radar detection device. The surface defect detection module 1020 is used to input a laser grayscale image into the first tunnel surface defect recognition model to generate the corresponding first tunnel defect detection result, and to input a visible light image into the second tunnel surface defect recognition model to generate the corresponding second tunnel defect detection result. The matching module 1030 is used to perform consistency matching between the first tunnel defect detection result and the second tunnel defect detection result, and when it is determined that the first tunnel defect detection result and the second tunnel defect detection result belong to the same tunnel defect, the surface defect fusion detection result of the tunnel is determined. The internal defect detection module 1040 is used to input radar waveform data into the tunnel internal defect identification model and generate corresponding tunnel internal defect detection results. The fusion module 1050 is used to perform spatial position consistency matching between the tunnel internal defect detection results and the surface defect fusion results, and when it is determined that there are overlapping defect areas in the tunnel corresponding to the tunnel internal defect detection results and the surface defect fusion detection results, it fuses and generates multi-source fusion defect detection results for the tunnel.

[0086] The tunnel defect comprehensive detection device provided in this application, employing the tunnel defect comprehensive detection method in the above embodiments, can solve the technical problem of low reliability in tunnel defect identification caused by isolated multi-source tunnel defect detection data in the prior art. Compared with the prior art, the beneficial effects of the tunnel defect comprehensive detection device provided in this application are the same as those of the tunnel defect comprehensive detection method provided in the above embodiments, and other technical features in the tunnel defect comprehensive detection device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0087] This application provides a comprehensive detection device for tunnel defects. The comprehensive detection device for tunnel defects includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the comprehensive detection method for tunnel defects in the first embodiment described above.

[0088] The following is for reference. Figure 11 The diagram illustrates a structural schematic of a comprehensive tunnel defect detection device suitable for implementing embodiments of this application. The comprehensive tunnel defect detection device in this application may include, but is not limited to, mobile terminals such as laptops, tablets (Portable Application Description, PADs), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 11The integrated detection equipment for tunnel defects shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0089] like Figure 11 As shown, the integrated tunnel defect detection device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 1002 or a program loaded from storage device 1003 into random access memory (RAM) 1004. The random access memory 1004 also stores various programs and data required for the operation of the integrated tunnel defect detection device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows the integrated tunnel defect detection equipment to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows an integrated tunnel defect detection equipment with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0090] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0091] The comprehensive tunnel defect detection device provided in this application, employing the comprehensive tunnel defect detection method described in the above embodiments, can solve the technical problem of low reliability in tunnel defect identification caused by isolated multi-source tunnel defect detection data in the prior art. Compared with the prior art, the beneficial effects of the comprehensive tunnel defect detection device provided in this application are the same as those of the comprehensive tunnel defect detection method provided in the above embodiments, and other technical features of this comprehensive tunnel defect detection device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0092] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0093] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0094] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the comprehensive detection method for tunnel defects in the above embodiments.

[0095] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0096] The aforementioned computer-readable storage medium may be included in the integrated detection equipment for tunnel defects; or it may exist independently and not be assembled into the integrated detection equipment for tunnel defects.

[0097] The aforementioned computer-readable storage medium carries one or more programs. When these programs are executed by the integrated tunnel defect detection device, the integrated tunnel defect detection device: acquires a laser grayscale image of the tunnel output by a laser camera, a visible light image of the tunnel output by a high-definition camera, and radar waveform data of the tunnel output by a radar detection device; inputs the laser grayscale image into a first tunnel surface defect recognition model to generate a corresponding first tunnel defect detection result, and inputs the visible light image into a second tunnel surface defect recognition model to generate a corresponding second tunnel defect detection result; performs consistency matching on the first and second tunnel defect detection results, and determines the surface defect fusion detection result of the tunnel when it is determined that the first and second tunnel defect detection results belong to the same tunnel defect; inputs the radar waveform data into a tunnel internal defect recognition model to generate a corresponding tunnel internal defect detection result; performs spatial position consistency matching on the tunnel internal defect detection result and the surface defect fusion result, and fuses the tunnel internal defect detection result and the surface defect fusion detection result to generate a multi-source fusion defect detection result of the tunnel when it is determined that the tunnel defects corresponding to the tunnel internal defect detection result and the surface defect fusion detection result have overlapping defect areas.

[0098] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0099] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0100] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0101] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the comprehensive tunnel defect detection method described above. This solves the technical problem of low reliability in tunnel defect identification caused by isolated multi-source tunnel defect detection data in the prior art. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the comprehensive tunnel defect detection method provided in the above embodiments, and will not be elaborated upon here.

[0102] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the comprehensive detection method for tunnel defects as described above.

[0103] The computer program product provided in this application can solve the technical problem of low reliability in tunnel defect identification caused by isolated multi-source tunnel defect detection data in the prior art. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the comprehensive tunnel defect detection method provided in the above embodiments, and will not be repeated here.

[0104] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A comprehensive detection method for tunnel defects, characterized in that, The comprehensive detection method for tunnel defects includes: Acquire the laser grayscale image of the tunnel output by the laser camera, the visible light image of the tunnel output by the high-definition camera, and the radar waveform data of the tunnel output by the radar detection device. The laser grayscale image is input into the first tunnel surface defect recognition model to generate the corresponding first tunnel defect detection result, and the visible light image is input into the second tunnel surface defect recognition model to generate the corresponding second tunnel defect detection result. The consistency of the first tunnel defect detection result and the second tunnel defect detection result is matched, and when it is determined that the first tunnel defect detection result and the second tunnel defect detection result belong to the same tunnel defect, the surface defect fusion detection result of the tunnel is determined; The radar waveform data is input into the tunnel internal defect identification model to generate the corresponding tunnel internal defect detection results. Spatial position consistency matching is performed on the tunnel internal defect detection results and the surface defect fusion results. When it is determined that there are overlapping defect areas in the tunnel defects corresponding to the tunnel internal defect detection results and the surface defect fusion detection results, the multi-source fusion defect detection results of the tunnel are fused to generate the tunnel.

2. The comprehensive detection method for tunnel defects as described in claim 1, characterized in that, The step of matching the first tunnel defect detection result with the second tunnel defect detection result includes: Extract the first defect category and the first defect location coordinates of the first tunnel defect from the first tunnel defect detection result, and extract the second defect category and the second defect location coordinates of the second tunnel defect from the second tunnel defect detection result; Determine whether the first defect category and the second defect category are the same; If they are the same, the coordinates of the first defect location and the coordinates of the second defect location are mapped to a unified three-dimensional coordinate system to obtain the coordinates of the third defect location of the first tunnel defect and the coordinates of the fourth defect location of the second tunnel defect. Calculate the distance difference between the coordinates of the third defect location and the coordinates of the fourth defect location. If the distance difference is less than a preset distance threshold, determine that the first tunnel defect detection result and the second tunnel defect detection result belong to the same tunnel defect.

3. The comprehensive detection method for tunnel defects as described in claim 2, characterized in that, The step of determining the surface defect fusion detection result of the tunnel when the first tunnel defect detection result and the second tunnel defect detection result belong to the same tunnel defect includes: If it is determined that the first tunnel defect detection result and the second tunnel defect detection result belong to the same tunnel defect, either the first tunnel defect detection result or the second tunnel defect detection result shall be determined as the surface defect fusion detection result of the tunnel. After the step of matching the spatial location consistency between the first tunnel defect detection result and the second tunnel defect detection result, the method further includes: If it is determined that the first tunnel defect detection result and the second tunnel defect detection result do not belong to the same tunnel defect, a laser defect detection report of the tunnel is generated based on the first tunnel defect detection result, which includes the first defect category, the third defect location coordinates and the first defect level of the first tunnel defect. Based on the second tunnel defect detection results, an image defect detection report of the tunnel is generated, which includes the second defect category, the fourth defect location coordinates, and the second defect level.

4. The comprehensive detection method for tunnel defects as described in claim 1, characterized in that, The step of fusing and generating a multi-source fusion defect detection result for the tunnel when it is determined that there is an overlapping defect region between the tunnel internal defect detection result and the surface defect fusion detection result includes: If it is determined that there is an overlapping defect area between the tunnel defects corresponding to the tunnel interior defect detection result and the surface defect fusion detection result, both the tunnel interior defect detection result and the surface defect fusion detection result are determined as the multi-source fusion defect detection result; After the step of matching the spatial location consistency of the tunnel interior defect detection results and the surface defect fusion results, the method further includes: If it is determined that the tunnel defects corresponding to the internal tunnel defect detection results and the surface defect fusion detection results do not overlap in the defect area, a joint surface defect detection report of the tunnel is generated based on the surface defect fusion detection results, and an internal tunnel defect detection report of the tunnel is generated based on the internal tunnel defect detection results.

5. The comprehensive detection method for tunnel defects as described in claim 1, characterized in that, After determining that the tunnel defects corresponding to the internal defect detection results and the surface defect fusion detection results overlap in a defect region, the method further includes: The system acquires real-time stress data from wireless stress sensors deployed in the overlapping defect area, and retrieves historical stress data, historical visible light image data, and historical radar waveform data of the overlapping defect area from the historical database. By combining the real-time stress data, the historical stress data, the historical visible light image data, and the historical radar waveform data, a stress degradation judgment result is generated; After the step of generating the multi-source fusion defect detection results of the tunnel, the method further includes: When the stress degradation judgment result indicates that degradation exists, a comprehensive degradation alarm initial detection report is generated based on the multi-source fusion defect detection result, which includes defect location coordinates, defect category, defect level and degradation trend warning information. A three-dimensional point cloud model of the tunnel output by the laser camera is obtained, and the tunnel defects in the three-dimensional point cloud model are labeled with the defect location coordinates, defect category and defect level in the multi-source fusion defect detection results of the tunnel, so as to generate a three-dimensional visualization defect model. The three-dimensional visualization defect model is embedded into the initial detection report of the comprehensive degradation alarm, and the comprehensive degradation alarm detection report is output.

6. The comprehensive detection method for tunnel defects as described in claim 5, characterized in that, The step of generating a stress degradation judgment result by combining the real-time stress data, the historical stress data, the historical visible light image data, and the historical radar waveform data, and combining the real-time stress data and the historical stress data of the overlapping defect area, includes: Based on the historical stress data, the stress change time series data of the overlapping defect region is determined; Based on the historical visible light image data, extract the time-series data of the crack width in the overlapping defect region; Based on the historical radar waveform data, the internal defect time series data of the overlapping defect region is obtained by inversion. Based on the stress change time series data and the crack width time series data, a first correlation curve is constructed, and based on the stress change rate data and the void or non-dense volume time series data, a second correlation curve is constructed. Based on the first correlation curve and the second correlation curve, a defect degradation prediction model is established. The real-time stress data and the historical stress data are input into the defect deterioration prediction model, and the stress change rate prediction value of the overlapping defect area is predicted by the defect deterioration prediction model. If the predicted stress change rate exceeds the preset degradation threshold, the stress degradation judgment result is determined to be that degradation exists; otherwise, the stress degradation judgment result is determined to be that there is no degradation.

7. The comprehensive detection method for tunnel defects as described in claim 1, characterized in that, The steps for obtaining the radar waveform data of the tunnel output by the radar detection equipment include: The radar antenna of the radar detection device is controlled to transmit electromagnetic wave signals to the tunnel lining and receive the echo signals reflected by the tunnel lining based on the electromagnetic wave signals. During the transmission of the electromagnetic wave signal, the contact pressure between the radar antenna and the lining surface is monitored in real time by a pressure detection sensor, and the air pressure of the controllable airbag connected to the radar antenna is dynamically adjusted according to the feedback signal of the contact pressure to maintain the contact pressure within a preset range. The radar waveform data is generated based on the echo signal.

8. The method as described in claim 7, characterized in that, The step of controlling the radar antenna of the radar detection equipment to transmit electromagnetic wave signals toward the tunnel lining includes: The operating frequency of the radar antenna is dynamically configured according to the type of the tunnel lining to be detected; wherein, if the type of the tunnel lining to be detected is an arch, the operating frequency of the radar antenna is configured to be 900MHz and 400MHz, and if the radar antenna is an inverted arch, the operating frequency of the radar antenna is configured to be 270MHz.

9. A comprehensive detection device for tunnel defects, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the integrated detection method for tunnel defects as described in any one of claims 1 to 8.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the comprehensive detection method for tunnel defects as described in any one of claims 1 to 8.

Images