A heat supply network leakage tracing monitoring method and system based on thermal infrared images

Abstract

The application discloses a heat supply pipe network leakage tracing monitoring method and system based on thermal infrared images, and belongs to the technical field of pipeline safety monitoring. The method comprises the following steps: receiving a ground thermal radiation signal collected by a thermal infrared device, and converting the thermal radiation signal into a visual thermal image; preprocessing the visual thermal image to obtain a preprocessed thermal infrared image; performing semantic segmentation on the preprocessed thermal infrared image by using a full convolution network, extracting a temperature abnormal area, and extracting temperature values of each pixel point pixel by pixel; identifying a suspected leakage point based on the temperature values, and correcting the position in combination with pipeline GIS data, buried depth and wind direction parameters; constructing an exponential decay mathematical model, and calculating a diffusion radius and a leakage intensity; and performing risk grade determination based on a temperature difference value, a leakage expansion intensity and pipeline risk assessment. The method realizes non-contact accurate tracing and hierarchical early warning of heat supply pipe network leakage, and solves the problems of poor positioning accuracy and great influence of the environment.

Description

Technical Field

[0001] This application belongs to the field of pipeline safety monitoring technology, and in particular relates to a method and system for tracing and monitoring leaks in heating pipelines based on thermal infrared images. Background Technology

[0002] my country's urban centralized heating system has developed rapidly, but pipeline leaks caused by factors such as aging pipelines, corrosion, and external construction are frequent. Leaks in heating pipelines not only waste large amounts of water and heat energy, but can also lead to road collapses, burns, and other safety accidents, posing a serious threat to urban public safety.

[0003] Traditional pipeline inspection methods rely heavily on manual inspections, listening poles, or correlators—contact-based methods that suffer from low efficiency, poor location accuracy, and the need for road excavation for verification. In recent years, infrared thermal imaging technology has been introduced into pipeline inspection, utilizing surface temperature anomalies caused by leaking hot water to identify leak points. However, existing infrared detection technologies often have the following limitations: first, they are mostly based on mobile platforms (such as drones), resulting in limited coverage and constraints on battery life; second, they lack dedicated image preprocessing algorithms for heating pipelines, making them susceptible to environmental interference that can lead to missed or false detections; and third, they can only identify areas of thermal anomaly, lacking the ability to accurately trace and locate leaks based on physical diffusion models, making it difficult to infer the specific parameters and risk level of the leak.

[0004] Therefore, there is an urgent need for a method for monitoring leaks in heating pipe networks that can achieve non-contact, wide-range, and high-precision tracing. Summary of the Invention

[0005] To address the aforementioned issues, this application provides a method and system for tracing and monitoring leaks in heating pipe networks based on thermal infrared images, which enables non-contact, wide-range, and high-precision tracing of leaks in heating pipe networks.

[0006] This application provides a method for tracing and monitoring leaks in heating pipe networks based on thermal infrared images, including: It receives surface thermal radiation signals collected by thermal infrared equipment and converts the thermal radiation signals into visual thermal images. The wavelength of the thermal radiation signals is greater than or equal to 8 μm and less than or equal to 14 μm. Image preprocessing is performed on the visualized thermal image. A fully convolutional network is used to perform semantic segmentation on the preprocessed thermal infrared image to obtain the segmentation mask of the temperature anomaly region, and the temperature value of each pixel is extracted pixel by pixel. Within the temperature anomaly area defined by the segmentation mask, suspected leak points are identified based on the temperature values ​​of pixels. The suspected leak points are then corrected by multi-parameter coupling by combining parameters such as pipeline direction, burial depth, and wind direction to obtain the final leak point coordinates. Centered on the coordinates of the final leak point, multiple profile lines are selected radially, and temperature-distance data of continuous temperature point sequences on each profile line are extracted; the diffusion boundary and leak area are determined based on the temperature-distance data; the temperature value corresponding to the diffusion boundary is substituted into a pre-constructed mathematical model of the exponential decay of surface temperature with distance to obtain the diffusion radius; the leakage expansion intensity is calculated based on the diffusion radius and the difference between the leak point temperature and the background temperature. The risk level of thermal leakage is determined based on the temperature difference between the leak area and the normal pipeline temperature, the intensity of leakage expansion, and the obtained pipeline risk assessment results.

[0007] Furthermore, image preprocessing includes adaptive weighted bilateral filtering with temperature gradient weighting factors; adaptive weighted bilateral filtering with temperature gradient weighting factors includes: Calculate the temperature gradient of neighboring pixels, where the temperature gradient is used to characterize the degree of temperature change in the region; For leakage feature regions where the temperature gradient is greater than a preset threshold, the edge preservation weight of the bilateral filter is increased; for background regions where the temperature gradient is less than a preset threshold, the smoothing and noise reduction weight of the bilateral filter is increased. The thermal image is filtered based on the increased weights to remove noise while preserving details of the leak edges.

[0008] Furthermore, image preprocessing also includes: multi-scale super-resolution reconstruction with embedded temperature feature attention modules; Multi-scale super-resolution reconstruction with embedded temperature feature attention module includes: Construct a generative adversarial network (GAN), which includes a generator and a discriminator. A temperature feature attention module is embedded in the generator, which is used to capture temperature difference features in the image and to weight and enhance the key leakage features contained therein. A generative adversarial network (GAN) is trained, and the trained GAN is used to complete the details of the thermal image and repair the blurred features of the leak point.

[0009] Furthermore, image preprocessing also includes: dynamic temperature calibration normalization based on blackbody radiation theory; Dynamic temperature calibration normalization based on blackbody radiation theory includes: The normal area with stable medium temperature on the pipeline surface was selected as the standard temperature measurement point, and the area around the pipeline without thermal anomalies was selected as the environmental reference sampling point. Collect the actual absolute temperature and infrared image grayscale values ​​of standard temperature measurement points and environmental reference sampling points under different ambient temperature conditions; Based on the quantitative relationship between blackbody radiance and temperature, a dynamic calibration function is constructed; The grayscale values ​​of the visualized thermal image are corrected point by point using a dynamic calibration function, and the relative grayscale values ​​are converted into absolute temperature values.

[0010] Furthermore, within the temperature anomaly area defined by the segmentation mask, suspected leak points are identified based on the temperature values ​​of individual pixels. Multi-parameter coupling correction is then applied to these suspected leak points, incorporating parameters such as pipeline direction, burial depth, and wind direction, to obtain the final leak point coordinates, including: Calculate the temperature gradient vector for each pixel based on its temperature value. Threshold segmentation is performed in the temperature anomaly region defined by the segmentation mask to filter out abnormal image regions where the temperature difference exceeds the safety baseline; For the segmented temperature anomaly area, the temperature gradient vector calculated pixel by pixel is called. Based on the characteristic that the gradient vector uniformly points to the temperature extreme point of the area and represents the direction of heat diffusion, local extreme value detection is carried out. The intersection point of all gradient vectors is traced back in reverse to lock the core position of the temperature extreme value of the area. Finally, the coordinates of the center of the extreme value are extracted as the suspected leakage point. Obtain pipeline route and burial depth data to determine whether the suspected leak point is located within the preset error zone of the pipeline centerline; if it is located outside the error zone, the suspected leak point is excluded. If it is located within the error zone, the corresponding horizontal deviation correction amount is obtained based on the burial depth data, and the burial depth is corrected for the coordinates of the suspected leak point. By acquiring wind direction data and adjusting the coordinates of suspected leak points after depth correction based on the direction in which surface winds push hot air, the final leak point coordinates are obtained.

[0011] Furthermore, a mathematical model for the exponential decay of surface temperature with distance is constructed as follows: Constructing an exponential decay mathematical model:

[0012] in, To the leak point Horizontal distance, Distance from the leak point The surface temperature at that location The ambient background temperature, This is the difference between the temperature at the leak point and the background temperature. This is the temperature decay coefficient; The least squares method was used to fit the temperature-distance data, and the model parameters were solved. , and .

[0013] Furthermore, based on the temperature difference between the leak area and the normal pipeline temperature data, the intensity of leak expansion, and the obtained pipeline risk assessment results, the risk level of the thermal leak is determined, specifically including: Obtain the temperature difference between the leak area and the normal pipeline area, the expansion rate of the leak area over time, and the inherent risk assessment parameters of the pipeline; A four-level risk warning and judgment model is established, which includes low risk, medium risk, high risk, and extremely high risk. Based on the temperature difference, leakage expansion intensity, and pipeline risk assessment parameters, a corresponding four-level risk warning judgment model is matched to output the risk level.

[0014] This application also provides a heating network leak tracing and monitoring system based on high-point thermal infrared images, including: The data receiving module is used to receive the surface thermal radiation signal collected by the thermal infrared device and convert the thermal radiation signal into a visual thermal image. The waveband corresponding to the thermal radiation signal is greater than or equal to 8um and less than or equal to 14um. The preprocessing module is used to preprocess the thermal image obtained by converting thermal radiation signals to obtain a preprocessed thermal infrared image. The feature extraction module is used to perform semantic segmentation on the preprocessed thermal infrared image using a fully convolutional network, obtain the segmentation mask of the temperature anomaly region, and extract the temperature value of each pixel pixel by pixel. The source tracing and location module is used to identify suspected leak points based on the temperature values ​​of pixels within the temperature anomaly area defined by the segmentation mask, and to perform multi-parameter coupling correction on the suspected leak points by combining pipeline direction, burial depth and wind direction parameters to obtain the final leak point coordinates. The model calculation module is used to select multiple profile lines radially centered on the final leak point coordinates, extract temperature-distance data of continuous temperature point sequences on each profile line; determine the diffusion boundary and leak area based on the temperature-distance data; substitute the temperature value corresponding to the diffusion boundary into a pre-constructed mathematical model of exponential decay of surface temperature with distance to obtain the diffusion radius; and calculate the leak expansion intensity based on the diffusion radius and the difference between the leak point temperature and the background temperature. The risk assessment module is used to determine the risk level of thermal leakage based on the temperature difference between the temperature value of the leak area and the normal pipeline temperature data value, the intensity of leakage expansion, and the obtained pipeline risk assessment results.

[0015] This application also provides an electronic device, including: at least one processor and at least one memory; The memory is used to store one or more program instructions; The processor is used to execute one or more program instructions to perform the methods described above.

[0016] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.

[0017] Compared with the prior art, this application has the following advantages: Image preprocessing algorithms effectively improved image quality and feature clarity in complex environments; combined with physical diffusion models and multi-parameter correction, precise source tracing and location of leaks were achieved; non-contact, large-scale monitoring was realized, enabling leak detection without excavation; and a risk level determination mechanism based on multi-dimensional features was established, providing a scientific basis for operation and maintenance decisions and significantly improving the safety operation and maintenance level of heating pipe networks.

[0018] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

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

[0020] Figure 1 A flowchart of a method for tracing and monitoring leaks in heating pipe networks based on thermal infrared images, provided in an embodiment of the present invention; Figure 2 A schematic diagram of a heating pipeline leakage tracing and monitoring system based on thermal infrared images provided in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] like Figure 1 As shown in the figure, this embodiment provides a method for tracing and monitoring leaks in heating pipe networks based on thermal infrared images, which mainly includes the following steps: Step 1: Thermal infrared image acquisition.

[0023] Thermal infrared equipment is deployed at high points (such as tall buildings and towers) to specifically collect surface thermal radiation signals in the 8-14μm band. This band experiences low atmospheric attenuation, effectively capturing surface temperature anomalies caused by leaks in buried pipelines. The system then converts the collected signals into visualized thermal images.

[0024] Thermal infrared images can intuitively map the temperature distribution information of scene targets. Each pixel contains corresponding temperature value features. Compared with visible light images, they can effectively capture temperature differences and hidden abnormal states that cannot be detected by the naked eye, providing core data support for abnormal area detection.

[0025] Step 2: Image Preprocessing. To improve image quality, perform the following preprocessing operations: (1) Adaptive weighted bilateral filtering with temperature gradient weight factor; the adaptive weighted bilateral filtering with temperature gradient weight factor includes: calculating the temperature gradient of neighboring pixels, wherein the temperature gradient is used to characterize the degree of temperature change in the region; increasing the edge preservation weight of the bilateral filter for leakage feature regions with temperature gradient greater than a preset threshold; increasing the smoothing and noise reduction weight of the bilateral filter for background regions with temperature gradient less than a preset threshold; and filtering the thermal image according to the increased weight to retain leakage edge details while filtering out noise.

[0026] It should be noted that, based on physics, when buried heating pipelines are operating normally, the pipeline and the surrounding soil form a stable temperature field through heat conduction, with a uniform temperature distribution and no significant gradient change. When a pipeline leaks, the high-temperature medium (such as hot water or steam) breaks through the pipe wall and diffuses into the surrounding soil from the leak point. The heat gradually decreases with the diffusion distance, creating a "temperature attenuation gradient" on the ground surface, with the temperature gradually decreasing from directly above the leak point outwards. The closer to the leak point, the higher the surface temperature. As the horizontal distance increases, the temperature decreases systematically. This gradient characteristic becomes the key basis for source tracing analysis; therefore, the solution in this application requires calculation of the temperature gradient.

[0027] Furthermore, to address the noise issues in thermal infrared images caused by changes in lighting and dust interference in outdoor environments, traditional bilateral filtering relies on both spatial and grayscale weights for noise reduction, providing basic edge protection capabilities, but suffers from the inherent limitation of fixed weights. Traditional bilateral filtering algorithms apply uniform filtering parameters to the entire image, failing to adaptively distinguish between leakage feature areas with abrupt temperature changes and background areas with stable temperatures. This often results in incomplete background noise reduction or overly smoothed target edges and loss of detail, making it unsuitable for the preprocessing needs of complex outdoor infrared images. Therefore, this step introduces a temperature gradient weight factor based on traditional bilateral filtering. By utilizing local pixel temperature gradients to characterize the degree of temperature change in a region, the filtering weights are adaptively adjusted, balancing noise reduction and feature preservation requirements from a mechanistic perspective, demonstrating good physical rationality and scene adaptability.

[0028] (2) Multi-scale super-resolution reconstruction: A generative adversarial network (GAN) framework is adopted, and a temperature feature attention module is embedded in the generator. This module can identify temperature difference features such as low-temperature blue spots and high-temperature red spots in the image, and perform weighted enhancement on the key leakage features contained therein, thereby repairing the blurring and lack of detail of small leakage points and improving the recognition of small targets.

[0029] Multi-scale super-resolution reconstruction: A generative adversarial network (GAN) framework is adopted, with the overall network consisting of a generator and a discriminator. The generator backbone is a multi-scale residual convolutional layer, internally embedding a temperature feature attention module; the discriminator adopts a multi-layer convolutional + fully connected structure to distinguish between real high-definition infrared thermal images and network-reconstructed images. The training dataset collects massive amounts of infrared images of on-site thermal pipelines, including samples of normal pipelines, low-temperature blue spots from minor leaks, high-temperature steam red spots, and soil environmental interference, and is divided into training, validation, and test sets in a 7:2:1 ratio. The training process is divided into two stages: the first stage pre-trains the generator separately to reduce basic reconstruction errors; the second stage alternates between the generator and discriminator in an adversarial iteration, synchronously updating the network weights. A composite loss function is adopted, fusing pixel mean square error loss, temperature feature perception loss, and adversarial loss to constrain image grayscale temperature value deviations and enhance the feature weights of leak areas. The temperature feature attention module can identify temperature difference features such as low-temperature blue spots and high-temperature red spots in images, and weight and enhance the key leakage features contained therein, thereby repairing the blurring and lack of detail of small leak points and improving the identification of small target leaks.

[0030] (3) Dynamic Temperature Calibration and Normalization: A normal area with stable medium temperature on the pipe surface is selected as the standard temperature measurement point, and an area around the pipe without thermal anomalies is selected as the environmental reference sampling point. The actual absolute temperature and infrared image grayscale values ​​of the standard temperature measurement point and the environmental reference sampling point are collected under different ambient temperature conditions. Based on the quantitative relationship between blackbody radiation brightness and temperature, a dynamic calibration function is constructed. The dynamic calibration function is used to correct the grayscale values ​​of the visualized thermal image point by point, converting the relative grayscale values ​​into absolute temperature values. This calibration function can convert grayscale values ​​affected by environmental interference into standard absolute temperature values ​​in real time, eliminating the influence of ambient temperature fluctuations on the measurement accuracy.

[0031] To effectively address the image grayscale drift caused by environmental temperature fluctuations, atmospheric radiation interference, and equipment imaging deviations during infrared thermal imaging detection, and to improve the accuracy of pipeline surface temperature detection, this step constructs a dynamic temperature-grayscale mapping correction model based on blackbody radiation theory to adaptively eliminate grayscale deviations under different environmental conditions. In infrared imaging detection systems, image grayscale values ​​are positively correlated with the radiation temperature of the target object. However, in complex field detection environments, interference factors such as environmental temperature, air convection, and soil thermal radiation can disrupt this inherent mapping relationship, leading to differentiated grayscale characteristics on pipeline surfaces at the same actual temperature, severely affecting the accuracy of pipeline leak temperature determination. To address this, this application employs a fixed-point sampling calibration method for data acquisition. On one hand, a normal area with stable medium temperature on the pipeline surface is selected as a standard temperature measurement point, accurately recording the actual absolute temperature and corresponding infrared image grayscale value of this area. On the other hand, soil and air areas around the pipeline without thermal anomalies are selected as environmental reference sampling points, simultaneously collecting background grayscale data under different environmental temperature conditions to quantify the grayscale deviation caused by environmental interference. Based on multiple sets of sampled data and the quantitative relationship between blackbody radiation brightness and temperature, a dynamic calibration function that can adapt to environmental changes in real time is constructed, overcoming the limitations of poor adaptability of traditional fixed mapping models. This dynamic calibration function performs point-by-point correction on the original grayscale image of the thermal infrared sensor, uniformly converting relative grayscale values ​​affected by environmental interference into standard absolute temperature values. This effectively eliminates the influence of interference factors such as ambient temperature fluctuations and background thermal radiation, ensuring the stability and accuracy of pipeline temperature detection data across all scenarios. This provides reliable temperature data support for the accurate determination of subsequent pipeline thermal anomalies and leakage faults.

[0032] Step 3: Use a convolutional neural network to extract and identify leakage features, and identify areas with abnormal temperatures.

[0033] The preprocessed image is input into a fully convolutional network (FCN). FCN eliminates fully connected layers, enabling pixel-level prediction of images of arbitrary sizes. The network outputs a semantic segmentation mask for temperature anomaly regions and extracts the temperature value of each pixel pixel-by-pixel. Based on the temperature difference between neighboring pixels, a temperature gradient vector is calculated for each pixel; this vector points to temperature extremes, representing the direction of heat diffusion.

[0034] In step 3, the temperature-sensitive characteristics of thermal infrared images are utilized, combined with the pixel-level precision segmentation advantage of fully convolutional networks, to build a high-precision semantic segmentation model, thereby achieving automated and refined identification and extraction of temperature anomaly regions in the image.

[0035] The fully convolutional network uses the original temperature features of thermal infrared images as its core input, fully mining the spatial texture features and temperature gradient features of the image. Through multi-layer convolution, pooling, and upsampling operations, it completes feature extraction and dimensionality restoration, accurately distinguishing between normal temperature regions and abnormal temperature regions, and outputting pixel-level semantic segmentation results that fully outline the contours and ranges of abnormal temperature regions. After completing the abnormal region segmentation, refined feature analysis is performed on the identified abnormal temperature regions, extracting the temperature values ​​of each pixel within the region. Based on the temperature difference between neighboring pixels, the temperature gradient vector corresponding to each pixel is accurately calculated. Strictly adhering to the temperature extremum guidance principle, the direction of the gradient vector is determined, ensuring it uniformly points to the maximum or minimum temperature values ​​within the region. This identification method can accurately characterize the temperature change trend and heat diffusion characteristics of abnormal regions, providing precise quantitative data support for subsequent temperature anomaly tracing and risk level determination.

[0036] Step 4: Perform source tracing and correction to determine if it is an anomaly.

[0037] (1) Suspected Leakage Point Identification: Through threshold segmentation and local maximum detection, the center of the region with a temperature higher than the background is extracted as a suspected leakage point. .

[0038] Specifically, threshold segmentation is performed on the temperature anomaly regions defined by the segmentation mask output in step 3 above to filter out abnormal image regions where the temperature difference exceeds the safety baseline. For the segmented temperature anomaly regions, the temperature gradient vector calculated pixel by pixel is called. Based on the characteristic that the gradient vector uniformly points to the temperature extreme point of the region and represents the direction of heat diffusion, local extreme value detection is performed. The intersection point is traced back along all gradient vectors to lock the core position of the temperature extreme value of the region. Finally, the coordinates of the center of this extreme value are extracted as the suspected leakage point where the temperature is significantly higher than the surrounding background. .

[0039] (2) GIS verification: The coordinates of the suspected leak point are compared with the pipeline direction and burial depth data. If the suspected leak point is within the error zone of the pipeline centerline, the suspected leak point is retained; if the suspected leak point is far from the pipeline direction, it is determined to be a temperature anomaly in a non-pipeline area and is directly excluded.

[0040] (3) Location Correction: Considering that the burial depth of the pipeline will cause a horizontal deviation between the center of the surface hot spot and the actual leak point, the location correction is based on the burial depth. Use the following correction formula to correct the coordinates.

[0041] Correction formula:

[0042] in, , The horizontal deviation correction amount corresponding to the burial depth h is obtained by establishing a curve showing the relationship between burial depth and deviation through on-site calibration experiments.

[0043] Furthermore, surface winds can shift hot air around the surface, potentially causing extreme temperature points to shift towards the downwind side. Therefore, it's necessary to fine-tune the suspected leak location based on wind direction data. Finally, using temperature decay models along profiles in different directions, the center location of the leak point is calculated, and the average of these calculations is used as the final leak point coordinates.

[0044] Step 5: Extract the temperature difference between the leak area and the normal pipeline. Based on the temperature difference, use a diffusion model to determine the expansion intensity.

[0045] Multiple profile lines were selected radially from the final leak point to extract temperature-distance data. Based on the exponential decay law of temperature decay, a mathematical model of surface temperature variation with distance was established: Exponential decay model construction:

[0046] in: Distance from the leak point Surface temperature at that location; Background ambient temperature (surface temperature in the area unaffected by leakage); : The surface temperature rise directly above the leak point (the temperature difference between the leak point temperature and the background temperature). Temperature decay coefficient, which is related to soil thermal conductivity, medium leakage temperature, and pipeline burial depth; : and the leak point Horizontal distance.

[0047] Solving parameters using the least squares method , and The diffusion boundary is determined by setting a temperature change rate threshold, and the diffusion radius R is calculated by substituting it into the model. Furthermore, the leakage expansion intensity S is calculated, which is a comprehensive index of the leakage radius R and the temperature gradient. The larger the value of S, the higher the temperature rise and the smaller the diffusion radius of the leaking medium, and the higher the leakage intensity; conversely, the smaller the value of S, the lower the leakage intensity.

[0048] Step 6: Risk Level Determination. The system extracts the temperature difference between the leak area and the normal pipeline, the leakage expansion intensity S in the abnormal area, and the inherent risk assessment parameters of the pipeline. Based on these dimensions, a four-level risk warning determination model (blue, yellow, orange, red) is matched to output the current risk level of the leak.

[0049] Step 7: Push alerts through multiple channels to guide maintenance personnel to take appropriate actions, as shown in Table 1.

[0050] Table 1

[0051] The advantages of this application are as follows: 1) Safe and efficient: No need to stop heating, dig, or come into close contact with high-temperature pipes, completely avoiding safety hazards such as burns and gas explosions, and the detection efficiency is better than that of traditional manual inspection.

[0052] 2) Precise positioning: The high-temperature medium at the leak point will form a clear "thermal anomaly zone". The leak range and center point can be directly visualized through infrared images, with high positioning accuracy, avoiding blind excavation.

[0053] 3) All-weather operation: It is not affected by strong sunlight during the day, darkness at night, or severe weather such as rain, snow, and fog. Leaks can be identified simply by scanning with the equipment, making it especially suitable for continuous monitoring of complex urban pipe networks.

[0054] 4) Wide coverage: A single device can perform long-distance, wide-area scanning, quickly complete full coverage inspection of long-distance pipelines, and detect potential problems such as damage to pipeline insulation layers.

[0055] Based on the same inventive concept, this application also provides a heating pipe network leakage tracing and monitoring system based on high-point thermal infrared images, such as... Figure 2 As shown, it includes: The data receiving module is used to receive the surface thermal radiation signal collected by the thermal infrared device and convert the thermal radiation signal into a visual thermal image. The waveband corresponding to the thermal radiation signal is greater than or equal to 8 μm and less than or equal to 14 μm. The preprocessing module is used to perform image preprocessing on the thermal image obtained by converting the thermal radiation signal to obtain a preprocessed thermal infrared image. The feature extraction module is used to perform semantic segmentation on the preprocessed thermal infrared image using a fully convolutional network, obtain the segmentation mask of the temperature anomaly region, and extract the temperature value of each pixel pixel by pixel. The source tracing and positioning module is used to identify suspected leak points based on the temperature values ​​of pixels within the temperature anomaly area defined by the segmentation mask, and to perform multi-parameter coupling correction on the suspected leak points in combination with pipeline direction, burial depth and wind direction parameters to obtain the final leak point coordinates. The model calculation module is used to select multiple profile lines radially centered on the final leak point coordinates, extract temperature-distance data of continuous temperature point sequences on each profile line; determine the diffusion boundary and leak area based on the temperature-distance data; substitute the temperature value corresponding to the diffusion boundary into a pre-constructed mathematical model of exponential decay of surface temperature with distance to obtain the diffusion radius; and calculate the leak expansion intensity based on the diffusion radius and the difference between the leak point temperature and the background temperature. The risk assessment module is used to determine the risk level of thermal leakage based on the temperature difference between the temperature value of the leak area and the normal pipeline temperature data value, the intensity of leakage expansion, and the obtained pipeline risk assessment results.

[0056] Based on the same inventive concept as the above disclosure, this disclosure also provides an electronic device. The electronic device of this disclosure includes at least one processor and at least one memory electrically connected to the processor. The memory is electrically connected to the processor, wherein the memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method described above.

[0057] It should be noted that the electrical connection between the above-mentioned units does not necessarily mean the connection between lines. The indirect connection method can be applied to the embodiments of this disclosure as long as it achieves the purpose of this disclosure.

[0058] Based on the same inventive concept, this disclosure also provides a computer storage medium storing a computer program, which, when executed by a processor, implements the above-described method.

[0059] Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for tracing and monitoring leaks in heating pipe networks based on thermal infrared images, characterized in that, include: Receive the surface thermal radiation signal collected by the thermal infrared device, and convert the thermal radiation signal into a visual thermal image. The waveband corresponding to the thermal radiation signal is greater than or equal to 8 μm and less than or equal to 14 μm. The visualized thermal image is preprocessed, and a fully convolutional network is used to perform semantic segmentation on the preprocessed thermal infrared image to obtain a segmentation mask for the temperature anomaly region, and the temperature value of each pixel is extracted pixel by pixel. Within the temperature anomaly area defined by the segmentation mask, suspected leak points are identified based on the temperature values ​​of pixels. The suspected leak points are then corrected using multi-parameter coupling by combining parameters such as pipeline direction, burial depth, and wind direction to obtain the final leak point coordinates. Centered on the coordinates of the final leak point, multiple profile lines are selected radially, and temperature-distance data of continuous temperature point sequences on each profile line are extracted; the diffusion boundary and leak area are determined based on the temperature-distance data; the temperature value corresponding to the diffusion boundary is substituted into a pre-constructed mathematical model of the exponential decay of surface temperature with distance to obtain the diffusion radius; the leakage expansion intensity is calculated based on the diffusion radius and the difference between the leak point temperature and the background temperature. The risk level of thermal leakage is determined based on the temperature difference between the leak area and the normal pipeline temperature, the intensity of leakage expansion, and the obtained pipeline risk assessment results.

2. The method according to claim 1, characterized in that, The image preprocessing includes adaptive weighted bilateral filtering with a temperature gradient weighting factor; the adaptive weighted bilateral filtering with a temperature gradient weighting factor includes: Calculate the temperature gradient of neighboring pixels, wherein the temperature gradient is used to characterize the degree of temperature change in the region; For leakage feature regions where the temperature gradient is greater than a preset threshold, the edge preservation weight of the bilateral filter is increased; for background regions where the temperature gradient is less than a preset threshold, the smoothing and noise reduction weight of the bilateral filter is increased. The thermal image is filtered based on the increased weights to remove noise while preserving details of the leak edges.

3. The method according to claim 2, characterized in that, The image preprocessing also includes: multi-scale super-resolution reconstruction with an embedded temperature feature attention module; The multi-scale super-resolution reconstruction with embedded temperature feature attention module includes: Construct a generative adversarial network, which includes a generator and a discriminator; A temperature feature attention module is embedded in the generator, wherein the temperature feature attention module is used to capture temperature difference features in the image and to weight and enhance the key leakage features contained therein; The generative adversarial network is trained, and the trained generative adversarial network is used to complete the details of the thermal image and repair the blurred features of the leakage point.

4. The method according to claim 3, characterized in that, The image preprocessing also includes: dynamic temperature calibration normalization based on blackbody radiation theory; The dynamic temperature calibration normalization based on blackbody radiation theory includes: The normal area with stable medium temperature on the pipeline surface was selected as the standard temperature measurement point, and the area around the pipeline without thermal anomalies was selected as the environmental reference sampling point. The actual absolute temperature and infrared image grayscale values ​​of the standard temperature measurement point and the environmental reference sampling point were collected under different ambient temperature conditions. Based on the quantitative relationship between blackbody radiance and temperature, a dynamic calibration function is constructed; The grayscale values ​​of the visualized thermal image are corrected point by point using the dynamic calibration function, and the relative grayscale values ​​are converted into absolute temperature values.

5. The method according to any one of claims 2-4, characterized in that, Within the temperature anomaly area defined by the segmented mask, suspected leak points are identified based on the temperature values ​​of individual pixels. These suspected leak points are then combined with parameters such as pipeline direction, burial depth, and wind direction for multi-parameter coupling correction to obtain the final leak point coordinates, including: Calculate the temperature gradient vector for each pixel based on its temperature value. Threshold segmentation is performed in the temperature anomaly region defined by the segmentation mask to filter out abnormal image regions where the temperature difference exceeds the safety baseline; For the segmented temperature anomaly area, the temperature gradient vector calculated pixel by pixel is called. Based on the characteristic that the gradient vector uniformly points to the temperature extreme point of the area and represents the direction of heat diffusion, local extreme value detection is carried out. The intersection point of all gradient vectors is traced back in reverse to lock the core position of the temperature extreme value of the area. Finally, the coordinates of the center of the extreme value are extracted as the suspected leakage point. Obtain pipeline route and burial depth data, and determine whether the suspected leak point is located within a preset error zone of the pipeline centerline; if it is located outside the error zone, then the suspected leak point is excluded. If it is located within the error band, the corresponding horizontal deviation correction amount is obtained based on the burial depth data, and the coordinates of the suspected leak point are corrected for burial depth. By acquiring wind direction data and adjusting the coordinates of suspected leak points after depth correction based on the direction in which surface winds push hot air, the final leak point coordinates are obtained.

6. The method according to claim 5, characterized in that, A mathematical model for the exponential decay of surface temperature with distance is constructed as follows: Constructing an exponential decay mathematical model: in, To the leak point Horizontal distance, Distance from the leak point The surface temperature at that location The ambient background temperature, This is the difference between the temperature at the leak point and the background temperature. This is the temperature decay coefficient; The temperature-distance data were fitted using the least squares method to solve for the model parameters. , and .

7. The method according to claim 1 or 6, characterized in that, Based on the temperature difference between the leak area and the normal pipeline temperature, the intensity of the leak expansion, and the obtained pipeline risk assessment results, the risk level of the thermal leak is determined, specifically including: Obtain the temperature difference between the leak area and the normal pipeline area, the expansion rate of the leak area over time, and the inherent risk assessment parameters of the pipeline; A four-level risk warning and judgment model is established, which includes low risk, medium risk, high risk and extremely high risk. Based on the temperature difference, leakage expansion intensity, and pipeline risk assessment parameters, a corresponding four-level risk warning judgment model is matched to output the risk level.

8. A leak tracing and monitoring system for heating pipe networks based on high-point thermal infrared images, characterized in that, include: The data receiving module is used to receive the surface thermal radiation signal collected by the thermal infrared device and convert the thermal radiation signal into a visual thermal image. The waveband corresponding to the thermal radiation signal is greater than or equal to 8 μm and less than or equal to 14 μm. The preprocessing module is used to perform image preprocessing on the thermal image obtained by converting the thermal radiation signal to obtain a preprocessed thermal infrared image. The feature extraction module is used to perform semantic segmentation on the preprocessed thermal infrared image using a fully convolutional network, obtain the segmentation mask of the temperature anomaly region, and extract the temperature value of each pixel pixel by pixel. The source tracing and positioning module is used to identify suspected leak points based on the temperature values ​​of pixels within the temperature anomaly area defined by the segmentation mask, and to perform multi-parameter coupling correction on the suspected leak points in combination with pipeline direction, burial depth and wind direction parameters to obtain the final leak point coordinates. The model calculation module is used to select multiple profile lines radially centered on the final leak point coordinates, extract temperature-distance data of continuous temperature point sequences on each profile line; determine the diffusion boundary and leak area based on the temperature-distance data; substitute the temperature value corresponding to the diffusion boundary into a pre-constructed mathematical model of exponential decay of surface temperature with distance to obtain the diffusion radius; and calculate the leak expansion intensity based on the diffusion radius and the difference between the leak point temperature and the background temperature. The risk assessment module is used to determine the risk level of thermal leakage based on the temperature difference between the temperature value of the leak area and the normal pipeline temperature data value, the intensity of leakage expansion, and the obtained pipeline risk assessment results.

9. An electronic device, characterized in that, include: At least one processor and at least one memory; The memory is used to store one or more program instructions; The processor is configured to execute the one or more program instructions to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 7.

Images