Underground electric power tunnel inspection system and method based on fixed digital optical cable

Abstract

The application relates to the field of electric power inspection, in particular to an underground electric power tunnel inspection system and method based on a fixed digital optical cable, which comprises an inspection platform module, an optical cable communication module, an image calibration module, an information fusion module and a tunnel monitoring module; the inspection platform module is used for constructing a mobile inspection platform; the optical cable communication module is used for providing energy and a communication link for the inspection platform; the image calibration module is used for reconstructing a tunnel model; the information fusion module is used for fusing sensor data and generating a physical model; and the tunnel monitoring module is used for monitoring abnormal phenomena in a tunnel. The application can solve the problems of data conflicts, noises and redundancies caused in the inspection process, improve sensor data quality and data fusion precision, reduce data redundancy, reduce the risk of electric leakage in the inspection process, improve the automation and intelligent level of underground electric power tunnel operation and maintenance, and has high precision and strong robustness.

Description

Technical Field

[0001] This invention relates to the field of power line inspection, specifically to an underground power tunnel inspection system and method based on fixed digital optical cables. Background Technology

[0002] Underground power tunnels are specialized underground structures used for the centralized laying of high-voltage power cables. With an internal radius typically between 2 and 3 meters, they save land space and can accommodate large-capacity power transmission demands, making them a crucial power supply method for large urban power grids. As the undergrounding of transmission lines accelerates and the mileage of power tunnels continues to increase, problems such as erosion of the internal buffer layer, wall settlement, wall cracks, foreign object intrusion, and waterlogging are becoming increasingly prominent, necessitating regular inspections to mitigate tunnel safety risks.

[0003] Due to the complex and variable environment inside power tunnels, manual inspection is inefficient and prone to human error, necessitating the use of sensors for data collection. However, the accuracy of sensor data is easily affected by external environmental factors, leading to false alarms or missed alarms. Furthermore, sensor data is large in volume, making communication difficult. Current communication and power supply methods in power tunnels have limitations. Commonly used leaky cables have high power supply requirements, limited coverage, and the risk of leakage. Information transmission also suffers from poor real-time performance and insufficient accuracy.

[0004] In addition, fixed sensors require extensive deployment and are costly, while mobile sensors face difficulties in data processing, making it hard to extract fault information from large amounts of sensor data. Fault location is time-consuming and inaccurate. The intelligent inspection process faces problems such as large differences in heterogeneous data, difficulty in aligning heterogeneous data, and low fusion accuracy. A reliable communication and power supply technology is needed to support this process in order to detect potential risks in tunnels in a timely manner. Summary of the Invention

[0005] The purpose of this invention is to provide an underground power tunnel inspection system and method based on fixed digital optical cables to solve the problems mentioned in the background art.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an underground power tunnel inspection system based on fixed digital optical cable, comprising: an inspection platform module, an optical cable communication module, an image calibration module, an information fusion module, and a tunnel monitoring module; The inspection platform module is used to mount tunnel sensors, including laser light sources, fiber optic clamps, radar, encoders and color cameras, on a track-mounted platform. It provides power to the three-dimensional electric translation stage and performs motion control through a guide transmission wheel system and servo motor. The collimation adjustment motor adjusts the position and incident angle of the optical fiber to construct a mobile inspection platform. The optical fiber communication module is used to connect the serial port of the inspection platform and the main power station at the tunnel entrance using a fixed digital optical fiber cable, providing energy and communication links for the inspection platform. The communication optical fiber connector is equipped with a positioning sensing element to store the location information of the optical fiber line into the GIS database. The serial port is configured using VISA to control the optical fiber transmission switch of the inspection platform, and the collected signals are time-division multiplexed to control the movement of the inspection platform along the underground power tunnel. The image calibration module is used to acquire visible images and radar images inside the tunnel, use the energy gradient function to focus the images, perform geometric calibration and feature detection on the images using platform navigation data, superimpose overlapping pixels in overlapping areas of adjacent images, stitch together all tunnel images to obtain a panoramic view of the tunnel, compare the ground plane of the radar coordinate system with the actual ground plane, solve the pitch angle, roll angle, heading angle and translation displacement matrix of the radar coordinate system relative to the camera coordinate system, project the radar data onto a two-dimensional plane, use radar depth information to assist the camera inverse perspective transformation, fuse radar signals and camera images, and introduce ground height information to reconstruct the tunnel model; The information fusion module is used to calculate the spatiotemporal information distribution density of the event stream, reduce the information scale based on density sorting, remove redundant events, perform consistency detection on the data of each sensor based on the spatiotemporal neighborhood association of event clusters, fit the degree of conflict between sensor signals by Pearson correlation coefficient to obtain data credibility, correct the data credibility using information entropy, perform data fusion based on Dempster rule to obtain fused data matrix, use residual network to calibrate matrix parameters when the sensor shakes, and put the fused data into the tunnel model to obtain multi-source physical model; The tunnel monitoring module uses the pose value of the inspection platform IMU as the initial value for inter-frame registration to obtain the tunnel surface point cloud in the model, generate a binary mask, perform point cloud registration and sensor calibration using the ICP algorithm, detect parameter space peaks using the connected component labeling algorithm, locate the wall by point cloud traversal and peak fitting, segment the wall and ground model, extract OTDR curve data, train an SVM multi-classifier to classify the curve data, and monitor the appearance, water accumulation, settlement and foreign object intrusion phenomena inside the tunnel.

[0007] Furthermore, the inspection platform module includes: a sensor unit and a motion control unit; The sensor unit is used to deploy multi-source sensors on the translation platform. The radar performs 3D modeling of the tunnel, the camera performs visual inspection, the encoder provides platform position coordinates, and the laser source and fiber optic clamp constitute a distributed fiber optic sensing front end to monitor vibration and temperature. The motion control unit is used to coordinate the electric translation stage, guide transmission wheel system and collimation adjustment motor with a multi-axis motion controller to control the platform to move along the road surface inside the tunnel. At the same time, RFID markers are preset on the tunnel wall to calibrate the position when the platform passes by.

[0008] Furthermore, the optical fiber communication module includes: a main station unit and a signal processing unit; The main station unit is used to establish data processing and power supply equipment at the tunnel entrance. It uses Ethernet composite optical cable to connect the inspection platform and embeds magnetic induction elements in the connectors of the composite optical cable for real-time position calibration. The signal processing unit is used to receive the inspection task order input by the user through the VISA standard configuration platform main control serial port, send remote instructions to the inspection platform, plan the inspection path and speed, call the task-related sensors, and dynamically allocate the transmission bandwidth according to the task priority.

[0009] Furthermore, the image calibration module includes: a panoramic tunnel unit, a radar calibration unit, and a digital reconstruction unit; The panoramic tunnel unit is used to control the camera to perform isochronous triggering of shooting sampling according to the encoder pulse signal. Using the initial pose provided by the platform navigation data, the sampled image is bound with the platform pose and position coordinates. Image fusion technology is used to stitch the image together, and bundle adjustment is used to optimize the global image to obtain a panoramic image of the tunnel. The radar calibration unit is used to calibrate the extrinsic parameter matrix from the radar coordinate system to the camera coordinate system using a three-dimensional calibration board, perform outlier removal, statistical filtering, and axial channel filtering on the radar point cloud, and project the non-ground point cloud onto a two-dimensional profile plane perpendicular to the tunnel axis to form point cloud slices. The digital reconstruction unit is used to input radar depth point cloud into the inverse perspective transformation model to generate a ground depth map, and to generate a 3D geometric model of the tunnel including tunnel segments, civil structures and cable equipment through the SLAM 3D reconstruction algorithm.

[0010] Furthermore, the information fusion module includes: an event monitoring unit, a data repair unit, and a physical integration unit; The event monitoring unit is used to convert continuous sensor data, including vibration, temperature and image feature changes, into discrete spatiotemporal event streams, calculate the spatiotemporal density distribution of the event streams, and when the global distribution density is higher than a threshold, cluster spatiotemporally adjacent events to form cross-modal event clusters. The data repair unit is used to calculate the Pearson correlation coefficient between signals in the cross-modal event cluster, apply the Dempster combination rule, fuse all sensor data, and output a fused data matrix. The elements in the fused data matrix represent the probability of various anomalies occurring at the current location. The physical integration unit is used to register the point cloud of consecutive frames with the pose data output by the IMU as the initial value, and to compensate for vibration error by using the iterative nearest point algorithm. It also aligns and fuses the data matrix with the geometric model, and integrates the physical parameters of the event location into the three-dimensional geometric model.

[0011] Furthermore, the tunnel monitoring module includes: a wall positioning unit and an anomaly matching unit; The wall positioning unit is used to project the point cloud onto the transverse profile according to the three-dimensional geometric model, identify the position and normal direction of the two walls by connecting region marking and peak fitting, and segment the model into semantic components based on the detected walls by region growing. The anomaly matching unit is used to obtain the OTDR phase curve from the semantic component, extract the features of the event, including: attenuation value, reflection peak width, time domain period and frequency domain period, train a multi-classifier using historical data, and output the anomaly event type.

[0012] The method for inspecting underground power tunnels based on fixed digital optical cables includes the following steps: Step S1. Place a multi-source sensor, including a laser light source, fiber optic clamp, radar, encoder and color camera, on a track-mounted platform. The motor provides power and adjusts the position and incident angle of the fiber optic cable to build a mobile inspection platform. Step S2. Connect the serial port of the inspection platform and the main station at the tunnel entrance using a fixed digital optical cable. Set up a positioning sensor element at the optical cable connector to obtain the location information of the optical cable line. Use VISA to configure the serial port to control the optical cable transmission switch of the inspection platform, so that the inspection platform moves along the power tunnel. Step S3. Acquire visual images and radar images, stitch adjacent frames of visual images to obtain a panoramic view of the tunnel, project the radar image onto a two-dimensional plane, solve for the pitch angle, roll angle, heading angle and translation displacement matrix of the radar coordinate system, and fuse the radar signal and camera image to obtain the tunnel geometric model. Step S4. Monitor sensor events, remove redundant events based on the spatiotemporal information distribution density of the event stream, cluster spatiotemporally adjacent events to form cross-modal event clusters, calculate the correlation coefficient of signals between cross-modal event clusters, determine the data credibility, fuse credible data based on Dempster rules to obtain a fused data matrix, put the fused data matrix into the tunnel geometry model, and output a multi-source physical model. Step S5. Use the pose value of the inspection platform IMU as the initial value for inter-frame registration to obtain the tunnel surface point cloud in the multi-source physical model. Use the connected component labeling algorithm to detect the peak value in the parameter space, locate the wall by peak fitting, segment semantic components, extract the OTDR curve data of each semantic component, and train an SVM classifier to determine the type of abnormal event.

[0013] Furthermore, step S1 includes: Step S11. Deploy multi-source sensors on the translation platform. The radar performs 3D modeling of the tunnel, the camera performs visual inspection, the encoder provides the platform position coordinates, and the laser light source and fiber optic clamps constitute a distributed fiber optic sensing front end to monitor vibration and temperature. Step S12. A multi-axis motion controller is used to coordinate the electric translation stage, guide transmission wheel system and collimation adjustment motor to control the platform to move along the road surface inside the tunnel. At the same time, RFID markers are preset on the tunnel wall to calibrate the position when the platform passes by.

[0014] Furthermore, step S2 includes: Step S21. Establish data processing and power supply equipment at the tunnel entrance, connect the inspection platform with Ethernet composite optical cable, and embed magnetic induction elements in the connectors of the composite optical cable for real-time position calibration. Step S22. Using the VISA standard configuration platform main control serial port, receive the inspection task order input by the user, send remote instructions to the inspection platform, plan the inspection path and speed, call the task-related sensors, and dynamically allocate transmission bandwidth according to the task priority.

[0015] Furthermore, step S3 includes: Step S31. Control the camera to perform isochronous triggering of shooting sampling according to the encoder pulse signal, use the initial pose provided by the platform navigation data to bind the sampled image with the platform pose and position coordinates, use image fusion technology to stitch the image, and optimize the global image by bundle adjustment to obtain a panoramic image of the tunnel. Step S32. Use a 3D calibration board to calibrate the extrinsic parameter matrix from the radar coordinate system to the camera coordinate system. Perform outlier removal, statistical filtering, and axial channel filtering on the radar point cloud. Project the non-ground point cloud onto a 2D profile plane perpendicular to the tunnel axis to form point cloud slices. Input the radar depth point cloud into the inverse perspective transformation model to generate a ground depth map. Use the SLAM 3D reconstruction algorithm to generate a 3D geometric model of the tunnel containing tunnel segments, civil structures, and cable equipment.

[0016] Furthermore, step S4 includes: Step S41. Convert continuous sensor data, including vibration, temperature and image feature changes, into discrete spatiotemporal event streams, calculate the spatiotemporal density distribution of the event streams, and when the global distribution density is higher than a threshold, cluster spatiotemporally adjacent events to form cross-modal event clusters. Step S42. Calculate the Pearson correlation coefficient between each signal in the cross-modal event cluster, apply the Dempster combination rule to fuse all sensor data, and output a fused data matrix. The elements in the fused data matrix represent the probability of each type of anomaly occurring at the current location. Step S43. Using the pose data output by the IMU as the initial value, the iterative nearest point algorithm is used to register the point cloud of consecutive frames, compensate for vibration errors, align and fuse the data matrix and the geometric model, and integrate the physical parameters of the event location into the three-dimensional geometric model.

[0017] Furthermore, step S5 includes: Step S51. Based on the 3D geometric model, project the point cloud onto the transverse profile, identify the position and normal direction of the two walls by connecting region labeling and peak fitting, and segment the model into semantic components by region growing based on the detected walls. Step S52. Obtain the OTDR phase curve from the semantic component, extract the features of the event, including: attenuation value, reflection peak width, time domain period and frequency domain period, train a multi-classifier using historical data, and output the abnormal event type.

[0018] Compared with the prior art, the beneficial effects achieved by the present invention are: 1. This invention constructs a mobile inspection platform and uses a fixed digital optical cable to connect the serial port of the inspection platform and the main power station at the tunnel entrance, providing the inspection platform with energy and communication links. This solves the problems of long time consumption and low accuracy in manual fault location. It can carry out tunnel inspection quickly, accurately and efficiently. It adopts single-end power supply, which has the advantages of wide coverage and high safety, reduces the risk of leakage during the inspection process, and realizes high-power, high-efficiency, long-distance energy transmission and information conversion.

[0019] 2. This invention generates a panoramic view of the tunnel by stitching together all tunnel images, calibrates matrix parameters when the sensor shakes, fuses radar signals and camera images, and incorporates ground height information to reconstruct the tunnel model. This solves the accuracy problem caused by the inability of radar to directly measure height, maximizes the value of inspection data, and improves the automation and intelligence level of underground power tunnel operation and maintenance. It has high accuracy and strong robustness.

[0020] 3. This invention reduces information scale by density sorting, removes redundant events, and incorporates sensor fusion data into the tunnel model to construct a multi-source physical model. It relies on the physical model to monitor fault phenomena inside the tunnel, solves the problems of data conflict, noise and redundancy during the inspection process, improves the quality of sensor data and the accuracy of data fusion, reduces data redundancy, and realizes optimized inspection of underground infrastructure. Attached Figure Description

[0021] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the underground power tunnel inspection system based on fixed digital optical cable according to the present invention. Figure 2 This is a schematic diagram illustrating the steps of the underground power tunnel inspection method based on fixed digital optical cables according to the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see Figures 1 to 2 The present invention provides a technical solution: an underground power tunnel inspection system based on fixed digital optical cable, comprising: an inspection platform module, an optical cable communication module, an image calibration module, an information fusion module, and a tunnel monitoring module; The inspection platform module is used to mount tunnel sensors, including laser light sources, fiber optic clamps, radar, encoders and color cameras, on a track-mounted platform. It provides power to the three-dimensional electric translation stage and performs motion control through a guide transmission wheel system and servo motor. The collimation adjustment motor adjusts the position and incident angle of the optical fiber to construct a mobile inspection platform. The inspection platform module includes: a sensor unit and a motion control unit; The sensor unit is used to deploy multi-source sensors on the translation platform. The radar performs 3D modeling of the tunnel, the camera performs visual inspection, the encoder provides platform position coordinates, and the laser source and fiber optic clamp constitute a distributed fiber optic sensing front end to monitor vibration and temperature. The motion control unit is used to coordinate the electric translation stage, guide transmission wheel system and collimation adjustment motor with a multi-axis motion controller to control the platform to move along the road surface inside the tunnel. At the same time, RFID markers are preset on the tunnel wall to calibrate the position when the platform passes by.

[0024] The optical fiber communication module is used to connect the serial port of the inspection platform and the main power station at the tunnel entrance using a fixed digital optical fiber cable, providing energy and communication links for the inspection platform. The communication optical fiber connector is equipped with a positioning sensing element to store the location information of the optical fiber line into the GIS database. The serial port is configured using VISA to control the optical fiber transmission switch of the inspection platform, and the collected signals are time-division multiplexed to control the movement of the inspection platform along the underground power tunnel. The optical fiber communication module includes: a main station unit and a signal processing unit; The main station unit is used to establish data processing and power supply equipment at the tunnel entrance. It uses Ethernet composite optical cable to connect the inspection platform and embeds magnetic induction elements in the connectors of the composite optical cable for real-time position calibration. The signal processing unit is used to receive the inspection task order input by the user through the VISA standard configuration platform main control serial port, send remote instructions to the inspection platform, plan the inspection path and speed, call the task-related sensors, and dynamically allocate the transmission bandwidth according to the task priority.

[0025] The image calibration module is used to acquire visible images and radar images inside the tunnel, use the energy gradient function to focus the images, perform geometric calibration and feature detection on the images using platform navigation data, superimpose overlapping pixels in overlapping areas of adjacent images, stitch together all tunnel images to obtain a panoramic view of the tunnel, compare the ground plane of the radar coordinate system with the actual ground plane, solve the pitch angle, roll angle, heading angle and translation displacement matrix of the radar coordinate system relative to the camera coordinate system, project the radar data onto a two-dimensional plane, use radar depth information to assist the camera inverse perspective transformation, fuse radar signals and camera images, and introduce ground height information to reconstruct the tunnel model; The image calibration module includes: a panoramic tunnel unit, a radar calibration unit, and a digital reconstruction unit; The panoramic tunnel unit is used to control the camera to perform isochronous triggering of shooting sampling according to the encoder pulse signal. Using the initial pose provided by the platform navigation data, the sampled image is bound with the platform pose and position coordinates. Image fusion technology is used to stitch the image together, and bundle adjustment is used to optimize the global image to obtain a panoramic image of the tunnel. The radar calibration unit is used to calibrate the extrinsic parameter matrix from the radar coordinate system to the camera coordinate system using a three-dimensional calibration board, perform outlier removal, statistical filtering, and axial channel filtering on the radar point cloud, and project the non-ground point cloud onto a two-dimensional profile plane perpendicular to the tunnel axis to form point cloud slices. The digital reconstruction unit is used to input radar depth point cloud into the inverse perspective transformation model to generate a ground depth map, and to generate a 3D geometric model of the tunnel including tunnel segments, civil structures and cable equipment through the SLAM 3D reconstruction algorithm.

[0026] The information fusion module is used to calculate the spatiotemporal information distribution density of the event stream, reduce the information scale based on density sorting, remove redundant events, perform consistency detection on the data of each sensor based on the spatiotemporal neighborhood association of event clusters, fit the degree of conflict between sensor signals by Pearson correlation coefficient to obtain data credibility, correct the data credibility using information entropy, perform data fusion based on Dempster rule to obtain fused data matrix, use residual network to calibrate matrix parameters when the sensor shakes, and put the fused data into the tunnel model to obtain multi-source physical model; The information fusion module includes: an event monitoring unit, a data repair unit, and a physical integration unit; The event monitoring unit is used to convert continuous sensor data, including vibration, temperature and image feature changes, into discrete spatiotemporal event streams, calculate the spatiotemporal density distribution of the event streams, and when the global distribution density is higher than a threshold, cluster spatiotemporally adjacent events to form cross-modal event clusters. The data repair unit is used to calculate the Pearson correlation coefficient between signals in the cross-modal event cluster, apply the Dempster combination rule, fuse all sensor data, and output a fused data matrix. The elements in the fused data matrix represent the probability of various anomalies occurring at the current location. The physical integration unit is used to register the point cloud of consecutive frames with the pose data output by the IMU as the initial value, and to compensate for vibration error by using the iterative nearest point algorithm. It also aligns and fuses the data matrix with the geometric model, and integrates the physical parameters of the event location into the three-dimensional geometric model.

[0027] The tunnel monitoring module uses the pose value of the inspection platform IMU as the initial value for inter-frame registration to obtain the tunnel surface point cloud in the model, generate a binary mask, perform point cloud registration and sensor calibration using the ICP algorithm, detect parameter space peaks using the connected component labeling algorithm, locate the wall by point cloud traversal and peak fitting, segment the wall and ground model, extract OTDR curve data, train an SVM multi-classifier to classify the curve data, and monitor the appearance, water accumulation, settlement and foreign object intrusion phenomena inside the tunnel.

[0028] The tunnel monitoring module includes: a wall positioning unit and an anomaly matching unit; The wall positioning unit is used to project the point cloud onto the transverse profile according to the three-dimensional geometric model, identify the position and normal direction of the two walls by connecting region marking and peak fitting, and segment the model into semantic components based on the detected walls by region growing. The anomaly matching unit is used to obtain the OTDR phase curve from the semantic component, extract the features of the event, including: attenuation value, reflection peak width, time domain period and frequency domain period, train a multi-classifier using historical data, and output the anomaly event type.

[0029] The method for inspecting underground power tunnels based on fixed digital optical cables includes the following steps: Step S1. Place a multi-source sensor, including a laser light source, fiber optic clamp, radar, encoder and color camera, on a track-mounted platform. The motor provides power and adjusts the position and incident angle of the fiber optic cable to build a mobile inspection platform. Step S1 includes: Step S11. Deploy multi-source sensors on the translation platform. The radar performs 3D modeling of the tunnel, the camera performs visual inspection, the encoder provides the platform position coordinates, and the laser light source and fiber optic clamps constitute a distributed fiber optic sensing front end to monitor vibration and temperature. Step S12. A multi-axis motion controller is used to coordinate the electric translation stage, guide transmission wheel system and collimation adjustment motor to control the platform to move along the road surface inside the tunnel. At the same time, RFID markers are preset on the tunnel wall to calibrate the position when the platform passes by.

[0030] Step S2. Connect the serial port of the inspection platform and the main station at the tunnel entrance using a fixed digital optical cable. Set up a positioning sensor element at the optical cable connector to obtain the location information of the optical cable line. Use VISA to configure the serial port to control the optical cable transmission switch of the inspection platform, so that the inspection platform moves along the power tunnel. Step S2 includes: Step S21. Establish data processing and power supply equipment at the tunnel entrance, connect the inspection platform with Ethernet composite optical cable, and embed magnetic induction elements in the connectors of the composite optical cable for real-time position calibration. Step S22. Using the VISA standard configuration platform main control serial port, receive the inspection task order input by the user, send remote instructions to the inspection platform, plan the inspection path and speed, call the task-related sensors, and dynamically allocate transmission bandwidth according to the task priority.

[0031] Step S3. Acquire visual images and radar images, stitch adjacent frames of visual images to obtain a panoramic view of the tunnel, project the radar image onto a two-dimensional plane, solve for the pitch angle, roll angle, heading angle and translation displacement matrix of the radar coordinate system, and fuse the radar signal and camera image to obtain the tunnel geometric model. Step S3 includes: Step S31. Control the camera to perform isochronous triggering of shooting sampling according to the encoder pulse signal, use the initial pose provided by the platform navigation data to bind the sampled image with the platform pose and position coordinates, use image fusion technology to stitch the image, and optimize the global image by bundle adjustment to obtain a panoramic image of the tunnel. Step S32. Use a 3D calibration board to calibrate the extrinsic parameter matrix from the radar coordinate system to the camera coordinate system. Perform outlier removal, statistical filtering, and axial channel filtering on the radar point cloud. Project the non-ground point cloud onto a 2D profile plane perpendicular to the tunnel axis to form point cloud slices. Input the radar depth point cloud into the inverse perspective transformation model to generate a ground depth map. Use the SLAM 3D reconstruction algorithm to generate a 3D geometric model of the tunnel containing tunnel segments, civil structures, and cable equipment.

[0032] Step S4. Monitor sensor events, remove redundant events based on the spatiotemporal information distribution density of the event stream, cluster spatiotemporally adjacent events to form cross-modal event clusters, calculate the correlation coefficient of signals between cross-modal event clusters, determine the data credibility, fuse credible data based on Dempster rules to obtain a fused data matrix, put the fused data matrix into the tunnel geometry model, and output a multi-source physical model. Step S4 includes: Step S41. Convert continuous sensor data, including vibration, temperature and image feature changes, into discrete spatiotemporal event streams, calculate the spatiotemporal density distribution of the event streams, and when the global distribution density is higher than a threshold, cluster spatiotemporally adjacent events to form cross-modal event clusters. Step S42. Calculate the Pearson correlation coefficient between each signal in the cross-modal event cluster, apply the Dempster combination rule to fuse all sensor data, and output a fused data matrix. The elements in the fused data matrix represent the probability of each type of anomaly occurring at the current location. Step S43. Using the pose data output by the IMU as the initial value, the iterative nearest point algorithm is used to register the point cloud of consecutive frames, compensate for vibration errors, align and fuse the data matrix and the geometric model, and integrate the physical parameters of the event location into the three-dimensional geometric model.

[0033] Step S5. Use the pose value of the inspection platform IMU as the initial value for inter-frame registration to obtain the tunnel surface point cloud in the multi-source physical model. Use the connected component labeling algorithm to detect the peak value in the parameter space, locate the wall by peak fitting, segment semantic components, extract the OTDR curve data of each semantic component, and train an SVM classifier to determine the type of abnormal event.

[0034] Step S5 includes: Step S51. Based on the 3D geometric model, project the point cloud onto the transverse profile, identify the position and normal direction of the two walls by connecting region labeling and peak fitting, and segment the model into semantic components by region growing based on the detected walls. Step S52. Obtain the OTDR phase curve from the semantic component, extract the features of the event, including: attenuation value, reflection peak width, time domain period and frequency domain period, train a multi-classifier using historical data, and output the abnormal event type.

[0035] Example: Configure sensors, adjust the position of optical fibers and sensors, establish optical cable and energy communication links, control the mobile inspection platform, collect visual images and radar images, geometric calibration images, stitch images, generate a panoramic view of the tunnel, fuse radar data and camera images, reconstruct a 3D model of the tunnel, monitor sensor events, calculate the spatiotemporal information density of the event stream, remove redundant events, fuse sensor data, perform physical modeling of the tunnel surface, extract OTDR curve data, and output anomaly types.

[0036] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0037] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for inspecting underground power tunnels based on fixed digital optical cables, characterized in that, The method includes the following steps: Step S1. Place a multi-source sensor, including a laser light source, fiber optic clamp, radar, encoder and color camera, on a track-mounted platform. The motor provides power and adjusts the position and incident angle of the fiber optic cable to build a mobile inspection platform. Step S2. Connect the serial port of the inspection platform and the main station at the tunnel entrance using a fixed digital optical cable. Set up a positioning sensor element at the optical cable connector to obtain the location information of the optical cable line. Use VISA to configure the serial port to control the optical cable transmission switch of the inspection platform, so that the inspection platform moves along the power tunnel. Step S3. Acquire visual images and radar images, stitch adjacent frames of visual images to obtain a panoramic view of the tunnel, project the radar image onto a two-dimensional plane, solve for the pitch angle, roll angle, heading angle and translation displacement matrix of the radar coordinate system, and fuse the radar signal and camera image to obtain the tunnel geometric model. Step S4. Monitor sensor events, remove redundant events based on the spatiotemporal information distribution density of the event stream, cluster spatiotemporally adjacent events to form cross-modal event clusters, calculate the correlation coefficient of signals between cross-modal event clusters, determine the data credibility, fuse credible data based on Dempster rules to obtain a fused data matrix, put the fused data matrix into the tunnel geometry model, and output a multi-source physical model. Step S5. Use the pose value of the inspection platform IMU as the initial value for inter-frame registration to obtain the tunnel surface point cloud in the multi-source physical model. Use the connected component labeling algorithm to detect the peak value in the parameter space, locate the wall by peak fitting, segment semantic components, extract the OTDR curve data of each semantic component, and train an SVM classifier to determine the type of abnormal event.

2. The method for inspecting underground power tunnels based on fixed digital optical cables according to claim 1, characterized in that: Step S1 includes: Step S11. Deploy multi-source sensors on the translation platform. The radar performs 3D modeling of the tunnel, the camera performs visual inspection, the encoder provides the platform position coordinates, and the laser light source and fiber optic clamps constitute a distributed fiber optic sensing front end to monitor vibration and temperature. Step S12. A multi-axis motion controller is used to coordinate the electric translation stage, guide transmission wheel system and collimation adjustment motor to control the platform to move along the road surface inside the tunnel. At the same time, RFID markers are preset on the tunnel wall to calibrate the position when the platform passes through. Step S2 includes: Step S21. Establish data processing and power supply equipment at the tunnel entrance, connect the inspection platform with Ethernet composite optical cable, and embed magnetic induction elements in the connectors of the composite optical cable for real-time position calibration. Step S22. Using the VISA standard configuration platform main control serial port, receive the inspection task order input by the user, send remote instructions to the inspection platform, plan the inspection path and speed, call the task-related sensors, and dynamically allocate transmission bandwidth according to the task priority.

3. The method for inspecting underground power tunnels based on fixed digital optical cables according to claim 2, characterized in that: Step S3 includes: Step S31. Control the camera to perform isochronous triggering of shooting sampling according to the encoder pulse signal, use the initial pose provided by the platform navigation data to bind the sampled image with the platform pose and position coordinates, use image fusion technology to stitch the image, and optimize the global image by bundle adjustment to obtain a panoramic image of the tunnel. Step S32. Use a 3D calibration board to calibrate the extrinsic parameter matrix from the radar coordinate system to the camera coordinate system. Perform outlier removal, statistical filtering, and axial channel filtering on the radar point cloud. Project the non-ground point cloud onto a 2D profile plane perpendicular to the tunnel axis to form point cloud slices. Input the radar depth point cloud into the inverse perspective transformation model to generate a ground depth map. Use the SLAM 3D reconstruction algorithm to generate a 3D geometric model of the tunnel containing tunnel segments, civil structures, and cable equipment.

4. The method for inspecting underground power tunnels based on fixed digital optical cables according to claim 3, characterized in that: Step S4 includes: Step S41. Convert continuous sensor data, including vibration, temperature and image feature changes, into discrete spatiotemporal event streams, calculate the spatiotemporal density distribution of the event streams, and when the global distribution density is higher than a threshold, cluster spatiotemporally adjacent events to form cross-modal event clusters. Step S42. Calculate the Pearson correlation coefficient between each signal in the cross-modal event cluster, apply the Dempster combination rule to fuse all sensor data, and output a fused data matrix. The elements in the fused data matrix represent the probability of each type of anomaly occurring at the current location. Step S43. Using the pose data output by the IMU as the initial value, the iterative nearest point algorithm is used to register the point cloud of consecutive frames, compensate for vibration errors, align and fuse the data matrix and the geometric model, and integrate the physical parameters of the event location into the three-dimensional geometric model.

5. The method for inspecting underground power tunnels based on fixed digital optical cables according to claim 4, characterized in that: Step S5 includes: Step S51. Based on the 3D geometric model, project the point cloud onto the transverse profile, identify the position and normal direction of the two walls by connecting region labeling and peak fitting, and segment the model into semantic components by region growing based on the detected walls. Step S52. Obtain the OTDR phase curve from the semantic component, extract the features of the event, including: attenuation value, reflection peak width, time domain period and frequency domain period, train a multi-classifier using historical data, and output the abnormal event type.

6. An underground power tunnel inspection system based on fixed digital optical cables, characterized in that, The system includes the following modules: inspection platform module, optical fiber communication module, image calibration module, information fusion module, and tunnel monitoring module; The inspection platform module is used to mount tunnel sensors, including laser light sources, fiber optic clamps, radar, encoders and color cameras, on a track-mounted platform. It provides power to the three-dimensional electric translation stage and performs motion control through a guide transmission wheel system and servo motor. The collimation adjustment motor adjusts the position and incident angle of the optical fiber to construct a mobile inspection platform. The optical fiber communication module is used to connect the serial port of the inspection platform and the main power station at the tunnel entrance using a fixed digital optical fiber cable, providing energy and communication links for the inspection platform. The communication optical fiber connector is equipped with a positioning sensing element to store the location information of the optical fiber line into the GIS database. The serial port is configured using VISA to control the optical fiber transmission switch of the inspection platform, and the collected signals are time-division multiplexed to control the movement of the inspection platform along the underground power tunnel. The image calibration module is used to acquire visible images and radar images inside the tunnel, use the energy gradient function to focus the images, perform geometric calibration and feature detection on the images using platform navigation data, superimpose overlapping pixels in overlapping areas of adjacent images, stitch together all tunnel images to obtain a panoramic view of the tunnel, compare the ground plane of the radar coordinate system with the actual ground plane, solve the pitch angle, roll angle, heading angle and translation displacement matrix of the radar coordinate system relative to the camera coordinate system, project the radar data onto a two-dimensional plane, use radar depth information to assist the camera inverse perspective transformation, fuse radar signals and camera images, and introduce ground height information to reconstruct the tunnel model; The information fusion module is used to calculate the spatiotemporal information distribution density of the event stream, reduce the information scale based on density sorting, remove redundant events, perform consistency detection on the data of each sensor based on the spatiotemporal neighborhood association of event clusters, fit the degree of conflict between sensor signals by Pearson correlation coefficient to obtain data credibility, correct the data credibility using information entropy, perform data fusion based on Dempster rule to obtain fused data matrix, use residual network to calibrate matrix parameters when the sensor shakes, and put the fused data into the tunnel model to obtain multi-source physical model; The tunnel monitoring module uses the pose value of the inspection platform IMU as the initial value for inter-frame registration to obtain the tunnel surface point cloud in the model, generate a binary mask, perform point cloud registration and sensor calibration using the ICP algorithm, detect parameter space peaks using the connected component labeling algorithm, locate the wall by point cloud traversal and peak fitting, segment the wall and ground model, extract OTDR curve data, train an SVM multi-classifier to classify the curve data, and monitor the appearance, water accumulation, settlement and foreign object intrusion phenomena inside the tunnel.

7. The underground power tunnel inspection system based on fixed digital optical cable according to claim 6, characterized in that: The inspection platform module includes: a sensor unit and a motion control unit; The sensor unit is used to deploy multi-source sensors on the translation platform. The radar performs 3D modeling of the tunnel, the camera performs visual inspection, the encoder provides platform position coordinates, and the laser source and fiber optic clamp constitute a distributed fiber optic sensing front end to monitor vibration and temperature. The motion control unit is used to coordinate the electric translation stage, guide transmission wheel system and collimation adjustment motor with a multi-axis motion controller to control the platform to move along the road surface in the tunnel. At the same time, RFID markers are preset on the tunnel wall to calibrate the position when the platform passes by. The optical fiber communication module includes: a main station unit and a signal processing unit; The main station unit is used to establish data processing and power supply equipment at the tunnel entrance. It uses Ethernet composite optical cable to connect the inspection platform and embeds magnetic induction elements in the connectors of the composite optical cable for real-time position calibration. The signal processing unit is used to receive the inspection task order input by the user through the VISA standard configuration platform main control serial port, send remote instructions to the inspection platform, plan the inspection path and speed, call the task-related sensors, and dynamically allocate the transmission bandwidth according to the task priority.

8. The underground power tunnel inspection system based on fixed digital optical cable according to claim 7, characterized in that: The image calibration module includes: a panoramic tunnel unit, a radar calibration unit, and a digital reconstruction unit; The panoramic tunnel unit is used to control the camera to perform isochronous triggering of shooting sampling according to the encoder pulse signal. Using the initial pose provided by the platform navigation data, the sampled image is bound with the platform pose and position coordinates. Image fusion technology is used to stitch the image together, and bundle adjustment is used to optimize the global image to obtain a panoramic image of the tunnel. The radar calibration unit is used to calibrate the extrinsic parameter matrix from the radar coordinate system to the camera coordinate system using a three-dimensional calibration board, perform outlier removal, statistical filtering, and axial channel filtering on the radar point cloud, and project the non-ground point cloud onto a two-dimensional profile plane perpendicular to the tunnel axis to form point cloud slices. The digital reconstruction unit is used to input radar depth point cloud into the inverse perspective transformation model to generate a ground depth map, and to generate a 3D geometric model of the tunnel including tunnel segments, civil structures and cable equipment through the SLAM 3D reconstruction algorithm.

9. The underground power tunnel inspection system based on fixed digital optical cable according to claim 8, characterized in that: The information fusion module includes: an event monitoring unit, a data repair unit, and a physical integration unit; The event monitoring unit is used to convert continuous sensor data, including vibration, temperature and image feature changes, into discrete spatiotemporal event streams, calculate the spatiotemporal density distribution of the event streams, and when the global distribution density is higher than a threshold, cluster spatiotemporally adjacent events to form cross-modal event clusters. The data repair unit is used to calculate the Pearson correlation coefficient between signals in the cross-modal event cluster, apply the Dempster combination rule, fuse all sensor data, and output a fused data matrix. The elements in the fused data matrix represent the probability of various anomalies occurring at the current location. The physical integration unit is used to register the point cloud of consecutive frames with the pose data output by the IMU as the initial value, and to compensate for vibration error by using the iterative nearest point algorithm. It also aligns and fuses the data matrix with the geometric model, and integrates the physical parameters of the event location into the three-dimensional geometric model.

10. The underground power tunnel inspection system based on fixed digital optical cable according to claim 9, characterized in that: The tunnel monitoring module includes: a wall positioning unit and an anomaly matching unit; The wall positioning unit is used to project the point cloud onto the transverse profile according to the three-dimensional geometric model, identify the position and normal direction of the two walls by connecting region marking and peak fitting, and segment the model into semantic components based on the detected walls by region growing. The anomaly matching unit is used to obtain the OTDR phase curve from the semantic component, extract the features of the event, including: attenuation value, reflection peak width, time domain period and frequency domain period, train a multi-classifier using historical data, and output the anomaly event type.

Images