Tunnel leakage detection method and device based on DTS optical fiber and robot

Abstract

The invention discloses a tunnel leakage detection method and device based on a DTS optical fiber and a robot, and belongs to the technical field of underground engineering structure safety monitoring, and the method comprises the steps: laying a distributed temperature sensing optical fiber network at the back of a tunnel lining for long-term static temperature monitoring; identifying a leakage potential area based on the temperature anomaly judgment condition; dispatching the track inspection robot to carry out infrared thermal image and visible light dynamic detailed survey on the potential area; the temperature field, the infrared thermal image, the visible light image and the point cloud data are subjected to conjoint analysis through the multi-source fusion recognition module, the leakage type, area and grade are output, integrated detection of static general survey, dynamic detailed survey and fusion recognition is achieved, and the method has the advantages of being comprehensive in coverage, high in efficiency, accurate in positioning and accurate in recognition.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of underground engineering structure safety monitoring and intelligent detection, and particularly relates to a tunnel leakage detection method and device based on DTS optical fiber and robots. BACKGROUND

[0002] With the rapid development of urban rail transit and underground space engineering, a large number of shield tunnels have entered the long-term operation stage. During the long-term operation, affected by factors such as underground water pressure, structure aging, segment joint deformation and construction disturbance, leakage is prone to occur at the tunnel lining and segment joint positions. Leakage not only affects the durability of the structure, but also may cause safety hazards such as reinforcement corrosion, lining falling off and secondary lining cracking, which poses a serious threat to the safety and service life of the tunnel in operation.

[0003] The prior art discloses a tunnel detection robot and a detection method (Chinese invention: CN105082180A). The technical solution adopts a crawler-type mobile robot to travel at the bottom of the tunnel, and detects through the superimposed ultrasonic sensor, infrared sensor, multi-parameter gas sensor, temperature sensor and infrared thermal imager and other sensing devices to obtain gas information, temperature information and internal structure information in the tunnel. The scheme has certain practical value in the field of tunnel detection, but still has the following technical problems: First, the detection coverage is limited. CN105082180A adopts a method of traveling at the bottom of the tunnel by a crawler-type robot, and the detection angle mainly covers the bottom and lower side wall of the tunnel, which is difficult to comprehensively detect the top and high side wall of the tunnel. However, leakage of shield tunnels often occurs at the segment joint, especially in the top and high side wall areas, which are more prone to leakage due to water pressure. The bottom traveling method of the crawler-type robot cannot effectively cover these high-risk areas.

[0004] Second, there is a lack of continuous monitoring capability along the entire line. CN105082180A adopts a single detection method of robot inspection, and each detection needs to manually dispatch the robot into the tunnel, which has a long detection period and low efficiency. For a tunnel in operation for tens of kilometers, it is difficult to realize continuous online monitoring along the entire line, and it is impossible to timely find weak abnormal signals in the early stage of leakage.

[0005] Third, there is a lack of a cooperative mechanism for static general survey and dynamic detailed survey. The detection method of CN105082180A is to collect information by the robot along the detection route in sections, and lacks the ability to quickly screen the leakage risk areas along the entire line. In actual operation, leakage often occurs locally, and if detailed detection is performed in sections along the entire line, it will not only waste time but also consume resources, and it is difficult to complete efficient detection within the limited maintenance window period.

[0006] Fourth, the data fusion and intelligent recognition capabilities are insufficient. Although CN105082180A uses multiple sensors, the data from each sensor is collected and analyzed independently, lacking technical means for multi-source data fusion. Existing technologies struggle to jointly analyze temperature field data, infrared thermograms, visible light images, and point cloud geometric features, resulting in low leakage identification accuracy and a tendency to produce missed or false detections.

[0007] Therefore, there is an urgent need for a comprehensive detection technology that combines static fiber optic surveys with dynamic robotic surveys to achieve full-coverage monitoring, efficient screening, precise location, and intelligent identification of leaks in shield tunnels. Summary of the Invention

[0008] The purpose of this invention is to overcome the problems of limited detection coverage, lack of continuous monitoring capability along the entire line, lack of coordination mechanism between static survey and dynamic detailed survey, and insufficient data fusion and intelligent identification capability in the existing technology, and to provide a tunnel leakage detection method and device based on DTS optical fiber and robot.

[0009] To achieve the above objectives, this invention provides a tunnel leakage detection method based on DTS optical fiber and a robot, comprising the following steps: S1: Static survey stage: Distributed temperature sensing optical fibers are laid out along the circumferential and longitudinal directions behind the lining of the shield tunnel to construct a distributed temperature sensing optical fiber network. The spatial mileage coordinates corresponding to each optical fiber measuring point are recorded. The distributed temperature sensing optical fiber network is continuously collected over a long period of time through signal demodulation and data analysis units to obtain the spatiotemporal distribution data of the temperature field of the tunnel lining and surrounding rock.

[0010] S2: Leakage Potential Zone Identification Stage: The signal demodulation and data analysis unit compares the collected real-time temperature values ​​with the initial baseline temperature. When a persistent low temperature anomaly is detected in a certain area, the unit performs time-series analysis on the spatiotemporal distribution data of the temperature field based on the preset temperature anomaly judgment conditions to determine the leakage potential zone and output its mileage coordinates.

[0011] S3: Dynamic detailed inspection stage: Based on the mileage coordinates of the potential leakage area, the track inspection robot is dispatched to enter the target section along the wall-attached track. The track inspection robot uses the onboard infrared thermal imager and high-definition camera to dynamically detect the suspected leakage area, and simultaneously collects infrared thermal image data, high-definition visible light images and point cloud geometric data, and obtains robot pose data through the automatic navigation module.

[0012] S4: Multi-source fusion identification stage: The multi-source fusion identification module performs spatiotemporal registration and joint analysis on temperature field spatiotemporal distribution data, infrared thermal imaging data, high-definition visible light images and point cloud geometric data. Based on a deep learning model, it extracts the thermal and morphological features of the leakage area, outputs the leakage type, leakage area and leakage level, and uploads the identification results to the decision platform to generate a leakage risk classification report.

[0013] The present invention also provides a tunnel leakage detection device based on DTS optical fiber and robot, including: a distributed temperature sensing optical fiber network, a signal demodulation and data analysis unit, a track inspection robot and a multi-source fusion identification module.

[0014] Compared with the prior art, the present invention has the following beneficial effects: First, it offers comprehensive coverage. By deploying distributed temperature sensing optical fibers along the circumferential and longitudinal directions behind the tunnel lining, a continuous temperature monitoring network covering the entire tunnel can be constructed, enabling temperature field monitoring of the entire circumference and length of the tunnel without any blind spots.

[0015] Second, the detection efficiency is high. A two-stage detection mode combining static general survey and dynamic detailed survey is adopted. First, a distributed temperature sensing fiber optic network is used to quickly screen potential leakage areas along the entire line. Then, track inspection robots are dispatched to conduct detailed inspections only on potential leakage areas, avoiding detailed inspections of the entire line section by section, which significantly improves detection efficiency.

[0016] Third, it has high positioning accuracy. The track inspection robot travels along the wall-attached track, which can approach the surface of the tunnel lining for close-range inspection. Combined with odometers and attitude sensors, it achieves centimeter-level spatial positioning and forms a precise correspondence with the coordinates of the measuring points in the distributed temperature sensing fiber optic network.

[0017] Fourth, it boasts high accuracy. By employing a multi-source data fusion and deep learning-based identification method, it comprehensively analyzes temperature field data, infrared thermograms, visible light images, and point cloud geometric features, thereby improving the accuracy and robustness of leak identification. Attached Figure Description

[0018] Figure 1 This is a flowchart of the method of the present invention.

[0019] Figure 2 This is an architecture diagram of the system of the present invention.

[0020] Figure 3 This is a schematic diagram of the DTS fiber optic deployment according to the present invention.

[0021] Figure 4 This is a cross-sectional view of the wall-mounted track of the present invention. Detailed Implementation

[0022] Please refer to the attached document. Figures 1-4The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the embodiments described are for illustrative purposes only and are not intended to limit the scope of the invention.

[0023] This invention provides a tunnel leakage detection method based on DTS optical fiber and robots, such as... Figure 1 As shown, this method combines the functions of wide-area general survey and local detailed survey, and comprehensively utilizes distributed optical fiber sensing technology, time series statistical analysis, intelligent robot inspection and multi-source data fusion algorithm to achieve early warning, automatic identification and visual assessment of tunnel leakage.

[0024] Step S1: Static survey stage. In one embodiment of the present invention, the core task of the static survey stage is to establish a continuous temperature monitoring network covering the entire line and obtain the spatiotemporal distribution data of the temperature field of the tunnel lining and surrounding rock.

[0025] The distributed temperature sensing fiber optic network is deployed using a three-dimensional approach combining longitudinal backbone and circumferential monitoring. Specifically, two longitudinal backbone distributed temperature sensing fibers are laid behind the tunnel lining, located on the left and right sides of the tunnel cross-section, respectively, covering the entire longitudinal length of the section. Simultaneously, circumferential distributed temperature sensing fibers are laid at the top, side walls, and bottom of the tunnel, forming a circumferential monitoring zone. Preferably, the spacing between the circumferential monitoring zones is 50m to 100m to achieve effective coverage of the temperature field around the entire tunnel. The connection points between the longitudinal backbone fibers and the circumferential monitoring zones are connected to the signal demodulation and data analysis unit via optical cables.

[0026] The distributed temperature sensing fiber adopts a water pressure resistant and radiation-resistant coating structure, enabling it to operate stably for a long time in high humidity environments. Preferably, the distributed temperature sensing fiber uses a polyimide coating or a carbon / polyimide composite coating, with an operating temperature range of -40℃ to +85℃, a water pressure resistance of not less than 1MPa, and a service life of not less than 20 years.

[0027] The distributed temperature sensing fiber optic network uses the optical time-domain reflectometry (OTDR) principle for temperature acquisition. In this embodiment, the signal demodulation and data analysis unit includes a DTS distributed temperature sensing demodulator. This demodulator determines the temperature value at each location point on the fiber by emitting laser pulses into the distributed temperature sensing fiber and detecting the intensity ratio of the Stokes and anti-Stokes beams of the Raman scattered light in the backscattered light. Preferably, the spatial resolution is 1 m, the temporal resolution is no more than 5 min, and the temperature measurement accuracy is ±0.1℃.

[0028] After deployment, the system records the spatial mileage coordinates corresponding to each fiber optic measuring point. Preferably, the spatial mileage coordinates are encoded using tunnel mileage station numbers and ring numbers to establish a mapping relationship between the measuring point location and the tunnel structure.

[0029] The signal demodulation and data analysis unit performs long-term continuous temperature acquisition on the distributed temperature sensing fiber optic network. In this embodiment, the system conducts long-term static acquisition during the tunnel's operation period, forming multi-time-period temperature field time-series curves. Preferably, the acquisition period is more than 30 days to obtain sufficient background temperature data. After preliminary filtering and smoothing, the acquired temperature data establishes a benchmark steady-state temperature field model. This stage achieves a statistical characterization of the background temperature distribution and steady-state thermal environment along the entire tunnel, providing a comparative baseline for subsequent identification of potential leakage areas.

[0030] Step S2: Leakage potential zone identification stage. In one embodiment of the present invention, the core task of the leakage potential zone identification stage is to identify leakage potential zones based on the spatiotemporal distribution data of the temperature field through time series analysis methods.

[0031] The signal demodulation and data analysis unit compares the collected real-time temperature values ​​with the initial baseline temperature. When a persistent low-temperature anomaly is detected in a certain area, that is, when the temperature is consistently lower than that of the surrounding concrete, it is preliminarily determined that there may be a localized cooling phenomenon caused by groundwater infiltration in that area.

[0032] To accurately identify potential leakage zones and eliminate the influence of ambient temperature disturbances, this invention employs an adaptive statistical method to calculate temperature anomaly detection thresholds. Specifically, the preset temperature anomaly detection conditions include three aspects: the temperature anomaly gradient exceeds a first preset threshold, the temperature abrupt change rate exceeds a second preset threshold, and the steady-state deviation relative to the baseline temperature exceeds a third preset threshold.

[0033] The first preset threshold, i.e., the temperature anomaly gradient threshold, is calculated using a time window sliding algorithm and an adaptive statistical method. The temperature anomaly gradient adaptive determination algorithm proposed in this invention is as follows: Suppose a certain measuring point is at time t. Temperature value The length of the sliding time window is Calculate the temperature standard deviation within a sliding time window over hours. and average rate of change The formula for calculating the standard deviation of temperature is: , in, The standard deviation of temperature within the sliding time window. This represents the number of sampling points within the sliding time window. For the first Temperature values ​​at each sampling point This represents the average temperature value within the sliding time window.

[0034] The formula for calculating the average rate of change is: , in, This represents the average rate of temperature change within the sliding time window.

[0035] The first preset threshold is the temperature anomaly gradient threshold. The calculation formula is: , in, This is the temperature anomaly gradient threshold. The standard deviation weighting coefficient is used. The rate of change weighting coefficient. Preferably, The value range is from 0.6 to 0.8. The value range is from 0.2 to 0.4.

[0036] The second preset threshold is the temperature mutation rate threshold. The method for determining the threshold is as follows: based on statistical analysis of historical data, the 95th percentile of the temperature change rate under normal operating conditions is selected as a reference value, and multiplied by a safety factor to determine the threshold. Preferably, the safety factor ranges from 1.5 to 2.0.

[0037] The third preset threshold is the steady-state deviation threshold. The method for determining the steady-state deviation is as follows: calculate the deviation of the temperature value at each measuring point relative to the baseline temperature. When the deviation continuously exceeds a preset threshold and the duration exceeds 24 hours, it is determined to be an abnormal steady-state deviation. Preferably, the steady-state deviation threshold ranges from 0.5℃ to 2.0℃.

[0038] The comprehensive algorithm for determining leakage potential zones proposed in this invention is as follows: Suppose a certain measuring point is at time t. The temperature anomaly gradient is Temperature mutation rate The steady-state deviation is Comprehensive abnormal score The calculation formula is: , in, To calculate the overall abnormal score, , , The weighting coefficients and , For temperature anomaly gradient, Temperature mutation rate, This represents the steady-state deviation. When the overall anomaly score... Exceeding the preset comprehensive threshold When this occurs, the area where the measuring point is located is determined to be a potential leakage zone. Preferably, a comprehensive threshold is used. The value is 1.0.

[0039] After identifying potential leakage areas, the signal demodulation and data analysis unit automatically outputs their mileage coordinates, ring number, and anomaly intensity level. This step effectively identifies early-stage, subtle thermal anomalies in leakage, reducing the missed detection rate of traditional manual inspections.

[0040] Step S3: Dynamic detailed inspection stage. In one embodiment of the present invention, the core task of the dynamic detailed inspection stage is to schedule a track inspection robot to conduct fine inspection of the potential leakage areas identified in the static general survey stage.

[0041] Based on the mileage coordinates of the potential leakage zone, the signal demodulation and data analysis unit sends a scheduling command to the track inspection robot, specifying the start and end mileage of the target section. After receiving the command, the track inspection robot enters the target section along the wall-attached track.

[0042] like Figure 4 As shown, the wall-mounted track is fixed to the inner wall of the tunnel using expansion bolts, preferably to the upper part of the tunnel top or sidewall. The structure of the wall-mounted track includes: expansion bolt interfaces connecting to the tunnel structure, screw interfaces connecting to the wall-mounted track, screw clips that mate with the screw interfaces, transmission serrations, and signal cable grooves. The track inspection robot adapts to tunnel cross-sections of different diameters and slopes via a telescopic guide rail module, preferably adapting to shield tunnels with diameters ranging from 5m to 15m.

[0043] The track inspection robot is equipped with multiple sensing devices, including an infrared thermal imager, a high-definition camera, a laser scanner, and an automatic navigation module. The infrared thermal imager preferably has a resolution of 640×512 pixels, a temperature measurement range of -20℃ to +150℃, and a temperature resolution of no less than 0.05℃, capable of revealing subtle temperature unevenness on the lining surface. The high-definition camera preferably has a resolution of 4K, capable of capturing visual signs such as watermarks, salt efflorescence, cracks, and biological deposits. The laser scanner is used to acquire point cloud geometric data of the tunnel interior wall, preferably using a laser scanner with a point cloud density of no less than 1000 points / m².

[0044] The automatic navigation module includes an odometer and an attitude sensor for high-precision pose positioning. Preferably, the positioning accuracy is no less than ±5cm. The robot records the distance traveled along the track using the odometer and obtains the robot's attitude angles using the attitude sensor, thereby determining the robot's precise position and attitude in the tunnel coordinate system.

[0045] The track-based inspection robot moves slowly along the wall-attached track, employing a combined detection mode of stationary stopping and dynamic scanning. In stationary stopping mode, the robot stops at the core location of the potential leakage zone to acquire high-resolution infrared thermal images and high-definition visible light images. In dynamic scanning mode, the robot moves along the track at a constant speed, simultaneously acquiring infrared thermal image data, high-definition visible light images, and point cloud geometric data. Preferably, the dynamic scanning speed is 0.1 m / s to 0.5 m / s.

[0046] The robot pose data acquired by the automatic navigation module forms a one-to-one correspondence with the coordinates of the measurement points in the distributed temperature sensing fiber optic network, thereby realizing a three-dimensional mapping of light, heat, and vision, and providing a spatial registration basis for subsequent multi-source data fusion.

[0047] Step S4: Multi-source fusion identification stage. In one embodiment of the present invention, the core task of the multi-source fusion identification stage is to perform spatiotemporal registration and joint analysis of multi-source heterogeneous data to achieve intelligent identification of leakage areas.

[0048] The multi-source fusion identification module receives temperature field spatiotemporal distribution data from a distributed temperature sensing fiber optic network, infrared thermal image data from an orbital inspection robot, high-definition visible light images, point cloud geometric data, and robot pose data.

[0049] First, the multi-source fusion identification module performs spatiotemporal registration on the multi-source data. The spatiotemporal registration algorithm for multi-source data proposed in this invention is as follows: A unified three-dimensional coordinate system for the tunnel is established based on the robot's pose data. Let the robot at time [time value missing]... The position is ,in Let these be the robot's position coordinates in the tunnel coordinate system. For the robot's Euler angles.

[0050] Pixel coordinates of infrared thermal image data to tunnel three-dimensional coordinates The mapping formula is: , in, Let the rotation matrix be determined by Euler angles. This is the intrinsic parameter matrix of the infrared thermal imager. The depth value corresponding to a pixel is provided by the point cloud geometry data. The coordinates of the pixel in the tunnel's three-dimensional coordinate system.

[0051] Through the above coordinate transformation, infrared thermal imaging data, high-definition visible light images, and point cloud geometric data are uniformly mapped to the tunnel's three-dimensional coordinate system, forming spatial alignment with the measurement point coordinates of the distributed temperature sensing fiber optic network.

[0052] Secondly, the multi-source fusion identification module extracts multimodal features of the leakage area. The weighted fusion algorithm for leakage features proposed in this invention is as follows: Extracting temperature feature vectors from spatiotemporal distribution data of temperature field This includes features such as local temperature values, temperature gradients, and temperature change trends. Thermal feature vectors are extracted from infrared thermal imaging data. This includes features such as hotspot area, hotspot shape, and hotspot boundary gradient. Visual feature vectors are extracted from high-resolution visible light images. Features include watermark color, salt precipitation texture, and crack orientation. Geometric feature vectors are extracted from point cloud geometric data. Features include surface roughness, depth of unevenness, and deformation.

[0053] Fusion feature vectors The calculation formula is: , in, To fuse feature vectors, , , , The weighting coefficients for each mode are as follows: , This indicates a feature concatenation operation. Preferably, The value range is from 0.2 to 0.3. The value range is from 0.3 to 0.4. The value range is from 0.2 to 0.3. The value range is from 0.1 to 0.2.

[0054] Finally, the multi-source fusion recognition module classifies and identifies the fused features based on a deep learning model. Preferably, the DeepLabV3+ semantic segmentation network is used to extract the boundary and area of ​​the leakage region from the infrared thermal image data. The backbone network of the DeepLabV3+ network adopts the Xception structure, the ASPP module uses dilated convolutions with dilation rates of 6, 12, and 18 to achieve multi-scale feature extraction, and the decoder module fuses high-level semantic features with low-level detail features.

[0055] The deep learning model outputs the leakage type, leakage area, and leakage level. Leakage types include three categories: seepage, wet patch, and gushing. Seepage manifests as point-like or linear seepage with small temperature anomalies and an area typically less than 0.1 m². Wet patch manifests as a surface-like wetted area with moderate temperature anomalies and an area typically between 0.1 m² and 1.0 m². Gushing manifests as a large-area flow or gushing of water with significant temperature anomalies and an area typically greater than 1.0 m².

[0056] Leakage is classified into three levels: Level I, Level II, and Level III. Level I indicates minor leakage, with a small leakage volume that does not currently affect the safe operation of the tunnel; repair is recommended during the next routine maintenance. Level II indicates moderate leakage, with a moderate leakage volume that may affect the durability of the tunnel structure; repair is recommended within one month. Level III indicates severe leakage, with a large leakage volume that may threaten the safe operation of the tunnel; immediate emergency measures and prompt repair are recommended.

[0057] The multi-source fusion identification module uploads the identification results to the decision-making platform, generating a leakage risk classification report. This report includes the leakage level, location coordinates, leakage type, temporal evolution trend, and repair recommendations. Based on the leakage risk classification report and the tunnel structure model, the decision-making platform generates a 3D visualized leakage distribution map, intuitively displaying the spatial distribution and severity of leakage points. Furthermore, the platform can perform trend predictions based on historical data, providing maintenance units with repair priority ranking and dynamic maintenance decision-making suggestions.

[0058] Through the above steps, the present invention forms a closed-loop detection system of "static general survey - dynamic detailed survey - fusion identification - decision support", realizing intelligent identification, accurate positioning and full-process tracking of existing tunnel leakage risks.

[0059] To verify the effectiveness of the method of the present invention, an experimental verification was carried out in a cross-river shield tunnel. The tunnel is approximately 5km long and 15m in diameter, and has been in operation for more than 10 years, with multiple potential leakage hazards.

[0060] During the static survey phase, two longitudinally distributed temperature sensing optical fibers were deployed, forming a circumferential monitoring zone at the top, sides, and bottom of the tunnel. The distributed temperature sensing optical fibers were fixed to the tunnel segments with water-pressure-resistant sheaths, covering the entire section in length. The DTS demodulator had a sampling interval of 1m and an accuracy of ±0.1℃. After 30 days of continuous data collection, temperature anomalies were identified through time-series sliding window analysis. Experimental results showed that 12 potential leakage areas were identified during the static survey phase. Compared with manual inspection results, the accuracy rate reached 92.3%, and the missed detection rate was reduced to 3.8%.

[0061] During the dynamic detailed investigation phase, the wall-mounted rails were secured to the top of the shield tunnel using expansion bolts and clips, and a rail inspection robot was deployed. The robot was equipped with an infrared thermal imager (640×512 resolution), a high-definition camera (4K), and a laser scanner. The robot conducted detailed investigations of 12 potential leakage areas, collecting a total of 2400 infrared thermal images, 2400 high-definition visible light images, and 12 sets of point cloud data.

[0062] In the multi-source fusion identification stage, the DeepLabV3+ semantic segmentation network was used to segment and identify the leakage area. Experimental results show that the leakage detection accuracy reached 95.7%, which is 8.3 percentage points higher than the single infrared thermal imaging detection method and 12.1 percentage points higher than the single visible light detection method. The leakage localization accuracy is better than ±10cm, and the leakage classification accuracy reaches 91.2%.

[0063] In summary, the method of the present invention is superior to the prior art in terms of leakage detection coverage, detection efficiency, positioning accuracy and identification accuracy, and has good engineering application value.

[0064] like Figure 2 As shown, the present invention also provides a tunnel leakage detection device based on DTS optical fiber and robot, which is used to perform the detection method described in the above method embodiments.

[0065] The distributed temperature sensing fiber optic network 100 is used to deploy distributed temperature sensing fibers 101 in the circumferential and longitudinal directions behind the lining of the shield tunnel to obtain the spatiotemporal distribution data of the temperature field of the tunnel lining and surrounding rock.

[0066] like Figure 3 As shown, the distributed temperature sensing fiber optic network 100 includes a longitudinal trunk fiber and a circumferential monitoring fiber. The longitudinal trunk fiber is laid along the longitudinal direction of the tunnel, covering the entire length of the section. The circumferential monitoring fiber forms a circumferential monitoring band at the top, side walls, and bottom of the tunnel. The connection point between the longitudinal trunk fiber and the circumferential monitoring fiber is connected to the signal demodulation and data analysis unit 200 via an uplink optical cable 103.

[0067] The distributed temperature sensing fiber optic cable 101 adopts a water pressure resistant and radiation resistant coating structure, preferably a polyimide coating or a carbon / polyimide composite coating, which can operate stably for a long time in a high humidity environment.

[0068] The signal demodulation and data analysis unit 200 is used to demodulate the temperature signal collected by the distributed temperature sensing fiber optic network 100, compare the real-time temperature value with the initial baseline temperature, determine the leakage potential zone based on the preset temperature anomaly judgment conditions, and output its mileage coordinates.

[0069] The signal demodulation and data analysis unit 200 includes a DTS distributed temperature sensor demodulator and a data processing server. The DTS demodulator is connected to the distributed temperature sensor fiber optic network 100 via an optical fiber cable 103 and uses the optical time domain reflectometry principle to acquire temperature data. The data processing server executes the temperature anomaly gradient adaptive determination algorithm and the leakage potential zone comprehensive determination algorithm described in the method embodiment to determine the leakage potential zone and output its mileage coordinates.

[0070] The track inspection robot body 300 includes a wall-mounted track 301, an infrared thermal imager 302, a high-definition camera 303, and an automatic navigation module 304, which is used to enter the target section according to the mileage coordinates of the leakage potential area and collect infrared thermal image data, high-definition visible light images, point cloud geometric data, and robot pose data.

[0071] like Figure 4 As shown, the wall-mounted track 301 is fixed to the tunnel structure via expansion bolt interface 1, screw interface 2 and screw clip 3 are used to connect track sections, transmission saw teeth 4 are used for robot drive, and signal cable trough 5 is used for laying communication cables. The track inspection robot adapts to tunnel cross-sections of different diameters and slopes via expansion bolts and telescopic guide rail modules.

[0072] The infrared thermal imager 302 preferably uses an uncooled infrared detector with a resolution of 640×512 and a temperature measurement range of -20℃ to +150℃. The high-definition camera 303 preferably uses a 4K resolution industrial camera. The automatic navigation module 304 includes an odometer and an attitude sensor to achieve high-precision pose positioning.

[0073] The multi-source fusion identification module 400 is used to perform spatiotemporal registration and joint analysis on temperature field spatiotemporal distribution data, infrared thermal imaging data, high-definition visible light images and point cloud geometric data. Based on a deep learning model, it outputs leakage type, leakage area and leakage level, and generates a leakage risk classification report.

[0074] The multi-source data spatiotemporal registration algorithm and leakage feature weighted fusion algorithm described in the implementation method embodiment of the multi-source fusion identification module 400 are used. The deep learning model preferably adopts the DeepLabV3+ semantic segmentation network, the backbone network adopts the Xception structure, and the ASPP module adopts dilated convolution with dilation rates of 6, 12, and 18.

[0075] The output of the multi-source fusion identification module 400 is uploaded to the decision-making platform to generate a leakage risk classification report and a three-dimensional visualization leakage distribution map.

[0076] In one embodiment of this device, the entire system is connected via an industrial Ethernet network. A distributed temperature sensing fiber optic network 100 and a DTS demodulator are deployed within the junction box. The track inspection robot 300 communicates with the data platform via a wireless local area network. A multi-source fusion identification module 400 runs on a ground server, and the data can be displayed in real time at the monitoring center. When the system detects a new leak, it automatically triggers a maintenance work order to guide maintenance personnel in targeted repairs.

[0077] The embodiments of the present invention are not limited to the specific embodiments described above. Those skilled in the art can make various equivalent changes or substitutions based on the technical solutions of the present invention, and all such changes or substitutions should be included within the protection scope of the present invention.

Claims

1. A tunnel leakage detection method based on DTS fiber optic cable and robot, characterized in that, Includes the following steps: S1: Static survey stage: Distributed temperature sensing optical fibers are laid out along the circumferential and longitudinal directions behind the lining of the shield tunnel to construct a distributed temperature sensing optical fiber network. The spatial mileage coordinates corresponding to each optical fiber measuring point are recorded. The distributed temperature sensing optical fiber network is continuously collected over a long period of time through the signal demodulation and data analysis unit to obtain the spatiotemporal distribution data of the temperature field of the tunnel lining and surrounding rock. S2: Leakage potential zone identification stage: The signal demodulation and data analysis unit compares the collected real-time temperature value with the initial baseline temperature. When a persistent low temperature anomaly is detected in a certain area, the unit performs time-series analysis on the temperature field spatiotemporal distribution data based on preset temperature anomaly judgment conditions to determine the leakage potential zone and output its mileage coordinates. S3: Dynamic detailed inspection stage: Based on the mileage coordinates of the potential leakage area, the track inspection robot is dispatched to enter the target section along the wall-attached track. The track inspection robot uses an onboard infrared thermal imager and a high-definition camera to dynamically detect the suspected leakage area, and simultaneously collects infrared thermal image data, high-definition visible light images and point cloud geometric data, and obtains robot pose data through the automatic navigation module. S4: Multi-source fusion identification stage: The multi-source fusion identification module performs spatiotemporal registration and joint analysis on the temperature field spatiotemporal distribution data, the infrared thermal image data, the high-definition visible light image and the point cloud geometric data. Based on the deep learning model, it extracts the thermal and morphological features of the leakage area, outputs the leakage type, leakage area and leakage level, and uploads the identification results to the decision platform to generate a leakage risk classification report.

2. The method according to claim 1, characterized in that, In step S1, the distributed temperature sensing fiber optic network uses the optical time domain reflectance principle to collect temperature data, with a spatial resolution of 1m, a temporal resolution of no more than 5min, and a temperature measurement accuracy of ±0.1℃.

3. The method according to claim 1, characterized in that, In step S1, two main optical fibers are laid along the longitudinal direction of the tunnel for distributed temperature sensing, and circumferential optical fibers are laid at the top, side walls and bottom of the tunnel to form a circumferential monitoring band. The connection point between the main optical fibers and the circumferential monitoring band is connected to the signal demodulation and data analysis unit through an optical cable.

4. The method according to claim 1, characterized in that, In step S2, the preset temperature anomaly determination conditions include: the temperature anomaly gradient exceeds a first preset threshold, the temperature mutation rate exceeds a second preset threshold, and the steady-state deviation relative to the baseline temperature exceeds a third preset threshold.

5. The method according to claim 4, characterized in that, The first preset threshold is calculated using an adaptive statistical method, which includes: calculating the temperature standard deviation and average rate of change within a sliding time window, and using the weighted sum of the temperature standard deviation multiplied by a first coefficient and the average rate of change multiplied by a second coefficient as the first preset threshold to eliminate the influence of environmental temperature disturbances.

6. The method according to claim 1, characterized in that, In step S3, the track inspection robot performs high-precision pose positioning using an odometer and attitude sensor, with a positioning accuracy of no less than ±5cm, so that the infrared thermal image data and the point cloud geometric data form a one-to-one correspondence with the coordinates of the measurement points in the distributed temperature sensing fiber optic network.

7. The method according to claim 1, characterized in that, In step S4, the deep learning model uses DeepLabV3+ semantic segmentation network to extract the boundary and area of ​​the leakage area from the infrared thermal image data. The leakage types include seepage type, wet spot type and gushing water type, and the leakage levels include Level I, Level II and Level III.

8. The method according to claim 1, characterized in that, In step S4, when the multi-source fusion recognition module performs spatiotemporal registration on the multi-source data, it uses the robot pose data as a reference to uniformly map the infrared thermal image data, the high-definition visible light image, and the point cloud geometric data to the tunnel three-dimensional coordinate system.

9. The method according to claim 1, characterized in that, The leakage risk classification report includes leakage level, location coordinates, leakage type, time evolution trend and repair suggestions. The decision-making platform generates a three-dimensional visualized leakage distribution map based on the leakage risk classification report.

10. A tunnel leakage detection device based on DTS fiber optics and robots, used to implement the method described in any one of claims 1-9, characterized in that, include: A distributed temperature sensing fiber optic network is used to deploy distributed temperature sensing fibers in the circumferential and longitudinal directions behind the lining of a shield tunnel to obtain spatiotemporal distribution data of the temperature field of the tunnel lining and surrounding rock. The signal demodulation and data analysis unit is used to demodulate the temperature signal collected by the distributed temperature sensing fiber optic network, compare the real-time temperature value with the initial baseline temperature, determine the leakage potential zone based on the preset temperature anomaly judgment conditions, and output its mileage coordinates. The track inspection robot includes a wall-attached track, an infrared thermal imager, a high-definition camera, and an automatic navigation module. It is used to enter the target section according to the mileage coordinates of the leakage potential zone and collect infrared thermal image data, high-definition visible light images, point cloud geometric data, and robot pose data. The multi-source fusion identification module is used to perform spatiotemporal registration and joint analysis on the temperature field spatiotemporal distribution data, the infrared thermal image data, the high-definition visible light image and the point cloud geometric data, and output the leakage type, leakage area and leakage level based on the deep learning model, and generate a leakage risk classification report.

Images