An integrated space-air-ground monitoring method for leakage of thermal pipelines

Abstract

The present invention relates to the technical field of thermal systems, and particularly to a space-air-ground integrated monitoring method for heat pipeline leakage, including: collecting multi-source data of heat pipelines in a monitoring area; inputting the multi-source data into a preset conversion model to obtain simulated surface temperature data in the monitoring area, wherein the conversion model is constructed based on multi-source data of local heat pipelines combined with a generative adversarial network; performing anomaly detection on the simulated surface temperature data based on a preset temperature threshold to obtain anomaly temperature values and anomaly locations, thereby completing the monitoring of heat pipelines in the monitoring area. By combining remote sensing technology, optical fiber technology, and sensor technology, the present invention monitors large-scale heat pipelines from four levels: space, air, surface, and underground, realizing all-day, all-weather, and high-precision monitoring of heat pipeline leakage, and having a certain resistance to extreme weather.

Description

Technical Field

[0001] The present invention relates to the technical field of thermal systems, and particularly to an integrated space-air-ground monitoring method for leakage of thermal pipelines. Background Art

[0002] Thermal pipelines are mainly divided into two types: overhead and buried. The monitoring methods mainly rely on single means, such as using fiber optic sensors and tracer gases to monitor buried thermal pipelines; using unmanned aerial vehicle (UAV) thermal infrared images to monitor overhead thermal pipelines, etc. These methods have high monitoring costs, limited monitoring ranges, and single monitoring targets, and cannot achieve large-scale synchronous observation, and cannot meet the requirements of efficient, continuous, and accurate monitoring of thermal pipeline leakage. Especially under extreme weather conditions, the single means fails, which brings great uncertainty to the monitoring of pipeline leakage.

[0003] Therefore, there is an urgent need for an integrated space-air-ground monitoring method for leakage of thermal pipelines. Summary of the Invention

[0004] The purpose of the present invention is to provide an integrated space-air-ground monitoring method for leakage of thermal pipelines. By combining remote sensing technology, fiber optic technology, and sensor technology, large-scale thermal pipelines are monitored from four levels: space, air, surface, and underground, so as to achieve all-day, all-weather, and high-precision monitoring of thermal pipeline leakage and have a certain resistance to extreme weather.

[0005] To achieve the above purpose, the present invention provides the following solution:

[0006] An integrated space-air-ground monitoring method for leakage of thermal pipelines, comprising:

[0007] Collecting multi-source data of thermal pipelines in a monitoring area;

[0008] Inputting the multi-source data into a preset conversion model to obtain simulated surface temperature data in the monitoring area, wherein the conversion model is constructed based on multi-source data of local thermal pipelines combined with a generative adversarial network;

[0009] Performing anomaly detection on the simulated surface temperature data based on a preset temperature threshold to obtain abnormal temperature values and abnormal positions, and completing the monitoring of thermal pipelines in the monitoring area.

[0010] Optionally, the multi-source data includes: thermal infrared remote sensing images, multi-spectral high-resolution images, UAV thermal infrared images, UAV orthophotos, overhead thermal pipeline data, fiber optic sensing data, buried thermal pipeline layout data, and temperature sensor data.

[0011] Optionally, constructing the conversion model based on multi-source data of local thermal pipelines combined with a generative adversarial network includes:

[0012] Identify based on the multi-spectral high-resolution image and the UAV orthophoto image to obtain the overhead thermal pipelines;

[0013] Locate based on the buried thermal pipeline layout data to obtain the buried thermal pipelines;

[0014] Fuse the thermal infrared remote sensing image and the UAV thermal infrared image to obtain a fused image;

[0015] Adopt the single-window algorithm to perform land surface temperature inversion on the overhead thermal pipelines, buried thermal pipelines, and fused image to obtain local land surface temperature inversion data;

[0016] Associate the fiber optic sensing data and temperature sensor data with the local land surface temperature inversion data within the generative adversarial network to construct the conversion model.

[0017] Optionally, identifying the overhead thermal pipelines based on the multi-spectral high-resolution image and the UAV orthophoto image includes:

[0018] Input the multi-spectral high-resolution image and the UAV orthophoto image into a preset identification model to obtain the identification result of the overhead thermal pipelines, where the identification model adopts the DeepLab V3+ model of semantic segmentation technology.

[0019] Optionally, the identification model includes: an encoder and a decoder. The encoder includes a deep convolutional neural network, a spatial pyramid pooling module, and a first convolutional layer. The decoder includes a second convolutional layer, a splicing layer, a third convolutional layer, and an upsampling layer;

[0020] Pass the multi-spectral high-resolution image and the UAV orthophoto image through the deep convolutional neural network to extract shallow features; pass the shallow features through the spatial pyramid pooling module and the first convolutional layer in sequence to extract deep features;

[0021] Pass the shallow features through the second convolutional layer and then input them into the splicing layer to splice with the deep features to obtain the spliced features;

[0022] Input the spliced features into the third convolutional layer and the upsampling layer to restore the spliced features to the original size of the image to obtain the identification result.

[0023] Optionally, fusing the thermal infrared remote sensing image and the UAV thermal infrared image to obtain a fused image includes:

[0024] Register the thermal infrared remote sensing image and the UAV thermal infrared image, and calculate the eigenvalues and their corresponding eigenvectors of the thermal infrared remote sensing image and the UAV thermal infrared image respectively through the principal component transformation matrix;

[0025] Sort the eigenvalues from largest to smallest, and calculate the principal component components based on the sorted eigenvalues and their corresponding eigenvectors;

[0026] Perform histogram matching on the thermal infrared remote sensing image and the UAV thermal infrared image based on the principal component components, and replace the principal component components of the thermal infrared remote sensing image with the principal component components of the UAV thermal infrared image to obtain the fused image.

[0027] Optionally, associate the fiber optic sensing data and the temperature sensor data with the local surface temperature inversion data within the generative adversarial network. Constructing the conversion model includes:

[0028] Establish implicit mapping relationships between the fiber optic sensing data and the temperature sensor data and the local surface temperature inversion data respectively within the generator, and obtain local surface temperature simulation data through the implicit mapping relationships;

[0029] Within the discriminator, distinguish the local surface temperature simulation data from the local surface temperature inversion data by calculating the loss value, and optimize the model by reducing the loss value through iterative operations until the model converges to complete the construction of the conversion model.

[0030] Optionally, perform anomaly detection on the surface temperature simulation data based on a preset temperature threshold to obtain the abnormal temperature value and the abnormal location, including:

[0031] Preset the temperature threshold;

[0032] Perform time series analysis and change detection on the surface temperature simulation data. If the surface temperature simulation data exceeds the temperature threshold, it is an abnormal temperature value;

[0033] Based on the overhead and buried thermal pipelines, lock the geographical location of the abnormal temperature value, and judge the abnormal level by the temperature difference between the abnormal temperature value and the surface temperature simulation data before and after it.

[0034] The beneficial effects of the present invention are:

[0035] The proposed integrated space-air-ground monitoring method for thermal pipeline leakage combines remote sensing technology, fiber optic technology, and sensor technology to monitor large-scale thermal pipelines from four levels: space, air, surface, and underground, realizing all-day, all-weather, and high-precision monitoring of thermal pipeline leakage, having a certain resistance to extreme weather; even if one means fails, the monitoring task can still be completed, with a wide monitoring range and strong continuity, suitable for continuous leakage monitoring tasks of large-scale thermal pipelines. Description of the Drawings

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0037] Figure 1 Flowchart of an integrated space-air-ground monitoring method for heat pipeline leakage in an embodiment of the present invention;

[0038] Figure 2 Schematic diagram of the recognition model structure in an embodiment of the present invention;

[0039] Figure 3 Schematic diagram of the conversion model structure in an embodiment of the present invention. Detailed implementation manners

[0040] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0041] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the drawings and specific implementation manners.

[0042] This embodiment provides an integrated space-air-ground monitoring method for heat pipeline leakage, including:

[0043] Collect multi-source data of heat pipelines in the monitoring area;

[0044] Input the multi-source data into a preset conversion model to obtain the simulated surface temperature data in the monitoring area, where the conversion model is constructed based on the multi-source data of local heat pipelines combined with a generative adversarial network;

[0045] Perform anomaly detection on the simulated surface temperature data based on a preset temperature threshold to obtain the abnormal temperature value and abnormal position, and complete the monitoring of heat pipelines in the monitoring area.

[0046] Further, use drones, remote sensing technology, sensors, etc. to collect multi-source data of heat pipelines in the monitoring area. The multi-source data includes: thermal infrared remote sensing images, multi-spectral high-resolution images, drone thermal infrared images, drone orthophotos, aerial heat pipeline data, fiber optic sensing data, buried heat pipeline layout data, and temperature sensor data.

[0047] Further, the construction of the conversion model based on the multi-source data of local thermal pipelines combined with a generative adversarial network includes:

[0048] Identify based on the multi-spectral high-resolution image and the UAV orthophoto image to obtain the overhead thermal pipelines;

[0049] Locate based on the buried thermal pipeline layout data to obtain the buried thermal pipelines;

[0050] Fuse the thermal infrared remote sensing image and the UAV thermal infrared image to obtain the fused image;

[0051] Adopt the single-window algorithm to perform surface temperature inversion on the overhead thermal pipelines, buried thermal pipelines, and the fused image to obtain the local surface temperature inversion data;

[0052] Associate the fiber optic sensing data and the temperature sensor data with the local surface temperature inversion data within the generative adversarial network to construct the conversion model.

[0053] Further, identify based on the multi-spectral high-resolution image and the UAV orthophoto image to obtain the overhead thermal pipelines, including:

[0054] Input the multi-spectral high-resolution image and the UAV orthophoto image into a preset recognition model to obtain the recognition result of the overhead thermal pipelines. Among them, the recognition model uses the DeepLab V3+ model of semantic segmentation technology;

[0055] The recognition model includes: an encoder and a decoder. The encoder includes a deep convolutional neural network, a spatial pyramid pooling module, and a first convolutional layer. The decoder includes a second convolutional layer, a splicing layer, a third convolutional layer, and an upsampling layer;

[0056] Pass the multi-spectral high-resolution image and the UAV orthophoto image through the deep convolutional neural network to extract shallow features; pass the shallow features through the spatial pyramid pooling module and the first convolutional layer in sequence to extract deep features;

[0057] Input the shallow features into the splicing layer after passing through the second convolutional layer to splice with the deep features to obtain the spliced features;

[0058] Input the spliced features into the third convolutional layer and the upsampling layer to restore the spliced features to the original image size to obtain the recognition result.

[0059] Further, fuse the thermal infrared remote sensing image and the UAV thermal infrared image to obtain the fused image, including:

[0060] Register the thermal infrared remote sensing image and the UAV thermal infrared image, and calculate the eigenvalues and their corresponding eigenvectors of the thermal infrared remote sensing image and the UAV thermal infrared image respectively through the principal component transformation matrix;

[0061] Sort the eigenvalues from largest to smallest, and calculate the principal component components based on the sorted eigenvalues and their corresponding eigenvectors;

[0062] Perform histogram matching on the thermal infrared remote sensing image and the UAV thermal infrared image based on the principal component components, and replace the principal component components of the thermal infrared remote sensing image with the principal component components of the UAV thermal infrared image to obtain a fused image.

[0063] Further, associate the fiber optic sensing data and the temperature sensor data with the local surface temperature inversion data within a generative adversarial network, and construct a conversion model including:

[0064] Establish implicit mapping relationships between the fiber optic sensing data and the temperature sensor data and the local surface temperature inversion data within the generator respectively, and obtain local surface temperature simulation data through the implicit mapping relationships;

[0065] Within the discriminator, distinguish the local surface temperature simulation data from the local surface temperature inversion data by calculating the loss value, and optimize the model by iterative operations to reduce the loss value until the model converges, completing the construction of the conversion model.

[0066] Further, perform anomaly detection on the surface temperature simulation data based on a preset temperature threshold, and obtain the abnormal temperature value and the abnormal location, including:

[0067] The preset temperature threshold;

[0068] Perform time series analysis and change detection on the surface temperature simulation data. If the surface temperature simulation data exceeds the temperature threshold, it is an abnormal temperature value;

[0069] Based on the overhead and buried heat pipelines, lock the geographical location of the abnormal temperature value, and judge the abnormal level by the temperature difference between the abnormal temperature value and the surrounding surface temperature simulation data.

[0070] The following combines Figure 1 Further describe a space-air-ground integrated monitoring method for heat pipeline leakage provided in this embodiment:

[0071] S1. Collect multi-source space-air-ground data of heat pipelines in the monitoring area;

[0072] Collect space data such as thermal infrared remote sensing images and multi-spectral high-resolution images of heat pipelines in the monitoring area; space data such as UAV thermal infrared images and UAV orthophotos; surface data such as overhead heat pipeline data and fiber optic sensing data; underground data such as buried heat pipeline layout data and temperature sensor data.

[0073] S2. Input the multi-source data into a preset conversion model to obtain the surface temperature simulation data in the monitoring area;

[0074] Based on the identification of multi-spectral high-resolution images and UAV ortho-images, the overhead thermal pipelines are obtained. By inputting the multi-spectral high-resolution images and UAV ortho-images into a preset identification model, the identification results of the overhead thermal pipelines are obtained. Among them, the identification model uses the DeepLab V3+ model of semantic segmentation technology. As Figure 2 shown, the main structure of the model includes two parts: an encoder and a decoder. The encoder includes a deep convolutional neural network, a spatial pyramid pooling module, and a first convolutional layer. The decoder includes a second convolutional layer, a splicing layer, a third convolutional layer, and an upsampling layer. The multi-spectral high-resolution images and UAV ortho-images are passed through the deep convolutional neural network to extract shallow features. The shallow features are successively passed through the spatial pyramid pooling module and the first convolutional layer to extract deep features. The shallow features are input into the splicing layer after passing through the second convolutional layer and spliced with the deep features to obtain the spliced features. The spliced features are input into the third convolutional layer and the upsampling layer to restore the spliced features to the original size of the image and obtain the identification results.

[0075] Fuse the thermal infrared remote sensing image and the UAV thermal infrared image. To obtain the fused image, use the principal component analysis method to register the thermal infrared remote sensing image and the UAV thermal infrared image, and calculate the eigenvalues and their corresponding eigenvectors of the thermal infrared remote sensing image and the UAV thermal infrared image respectively through the principal component transformation matrix;

[0076] Sort the eigenvalues from largest to smallest, and calculate the principal component components based on the sorted eigenvalues and their corresponding eigenvectors. The calculation formula is as follows:

[0077]

[0078] In the formula, n is the total number of eigenvalues, x is the eigenvalue, is the eigenvector;

[0079] Based on the principal component components, perform histogram matching on the thermal infrared remote sensing image and the UAV thermal infrared image, and use the principal component components of the UAV thermal infrared image to replace the principal component components of the thermal infrared remote sensing image, and perform the inverse principal component transformation to obtain the fused image.

[0080] Associate the fiber optic sensing data and temperature sensor data with the local surface temperature inversion data within a generative adversarial network to construct a conversion model, including:

[0081] Using the local thermal pipeline fiber optic sensing data, temperature sensor data, and surface temperature inversion data as input data, establish a conversion model using a generative adversarial network to simulate the large-scale surface temperature. As Figure 3As shown in the figure, the conversion model includes two parts: a generator and a discriminator. The generator establishes an implicit mapping relationship between the fiber optic sensing data and the temperature sensor data and the surface temperature inversion data respectively. By inputting the fiber optic sensing data or the temperature sensing data, the surface temperature simulation data can be obtained. The discriminator distinguishes the surface temperature simulation data and the inversion data by calculating the loss value and optimizes the simulation data. Through iterative operations, the conversion model reduces the model loss value and narrows the gap between the surface temperature simulation data and the inversion data until the model converges, indicating the successful construction of the conversion model.

[0082] S3. Perform anomaly detection on the surface temperature simulation data based on a preset temperature threshold to obtain the abnormal temperature value and the abnormal location.

[0083] In this embodiment, the recognition result of the heat pipeline is used to detect the abnormal values of the temperature simulation data of the heat pipeline and its surrounding areas. If the value exceeds the specified threshold, a pipeline leakage warning is triggered, the range of the abnormal value is delimited, and the leakage level of the pipeline is judged by combining the temperature difference between the leakage point and the surrounding surface.

[0084] Anomaly value detection is carried out by performing time series analysis and change detection on the surface temperature simulation data, checking the surface temperature pixel by pixel, locking the abnormal temperature value near the heat pipeline, and obtaining the geographical coordinate information of the abnormal value.

[0085] The above embodiments are only descriptions of the preferred embodiments of the present invention and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for monitoring thermal pipeline leakage in an integrated air-ground-space manner, characterized in that: include: Collect multi-source data of thermal pipelines in the monitoring area; Inputting the multi-source data into a preset conversion model to obtain surface temperature simulation data in the monitoring area, wherein the conversion model is constructed based on multi-source data of local thermal pipelines combined with a generative adversarial network; Performing anomaly detection on the surface temperature simulation data based on a preset temperature threshold, obtaining an abnormal temperature value and an abnormal position, and completing monitoring of the thermal pipeline in the monitoring area; The multi-source data includes: thermal infrared remote sensing images, multi-spectral high-resolution images, drone thermal infrared images, drone orthophotos, overhead thermal pipeline data, optical fiber sensor data, buried thermal pipeline layout data, and temperature sensor data; The conversion model is constructed based on the multi-source data of the local thermal pipeline combined with a generative adversarial network, including: Based on the multispectral high-resolution image and the drone orthophoto, the overhead thermal pipeline is identified; Perform positioning based on the buried thermal pipeline layout data to obtain the buried thermal pipeline; fusing the thermal infrared remote sensing image with the unmanned aerial vehicle thermal infrared image to obtain a fused image; Using a single window algorithm to invert the surface temperature of the overhead thermal pipeline, the buried thermal pipeline, and the fused image, obtaining local surface temperature inversion data; The optical fiber sensing data and the temperature sensor data are associated with the local surface temperature inversion data in the generative adversarial network to construct the conversion model.

2. The air-ground integrated monitoring method for thermal pipeline leakage according to claim 1 is characterized in that: Identifying overhead thermal pipelines based on the multispectral high-resolution image and the drone orthophoto includes: The multispectral high-resolution image and the drone orthophoto are input into a preset recognition model to obtain the recognition result of the overhead thermal pipeline, wherein the recognition model adopts the DeepLab V3+ model of semantic segmentation technology.

3. The air-ground integrated monitoring method for thermal pipeline leakage according to claim 2 is characterized in that: The recognition model includes: an encoder and a decoder, the encoder includes a deep convolutional neural network, a spatial pyramid pooling module, and a first convolutional layer, and the decoder includes a second convolutional layer, a splicing layer, a third convolutional layer, and an upsampling layer; The multispectral high-resolution image and the drone orthophoto are passed through the deep convolutional neural network to extract shallow features; the shallow features are passed through the spatial pyramid pooling module and the first convolutional layer in sequence to extract deep features; The shallow features are input into the splicing layer after passing through the second convolutional layer to be spliced ​​with the deep features to obtain the spliced ​​features; The spliced ​​features are input into the third convolutional layer and the upsampling layer, the spliced ​​features are restored to the original size of the image, and the recognition result is obtained.

4. The air-ground integrated monitoring method for thermal pipeline leakage according to claim 1 is characterized in that: The thermal infrared remote sensing image is fused with the UAV thermal infrared image to obtain a fused image, including: The thermal infrared remote sensing image and the UAV thermal infrared image are registered, and the eigenvalues ​​and corresponding eigenvectors of the thermal infrared remote sensing image and the UAV thermal infrared image are calculated respectively through the principal component transformation matrix; Sorting the eigenvalues ​​from large to small, and calculating the principal component based on the sorted eigenvalues ​​and their corresponding eigenvectors; The thermal infrared remote sensing image and the UAV thermal infrared image are histogram matched based on the principal component components, and the principal component components of the UAV thermal infrared image are used to replace the principal component components of the thermal infrared remote sensing image to obtain the fused image.

5. The air-ground integrated monitoring method for thermal pipeline leakage according to claim 1 is characterized in that: Associating the optical fiber sensing data and the temperature sensor data with the local surface temperature inversion data in the generative adversarial network to construct the conversion model includes: In the generator, an implicit mapping relationship is established between the optical fiber sensing data and the temperature sensor data and the local surface temperature inversion data, and local surface temperature simulation data is obtained through the implicit mapping relationship; The local surface temperature simulation data and the local surface temperature inversion data are distinguished in the discriminator by calculating the loss value, and the model is optimized by reducing the loss value through iterative operation until the model converges, thereby completing the construction of the conversion model.

6. The air-ground integrated monitoring method for thermal pipeline leakage according to claim 1 is characterized in that: Performing anomaly detection on the surface temperature simulation data based on a preset temperature threshold to obtain an abnormal temperature value and an abnormal position includes: Presetting the temperature threshold; Performing time series analysis and change detection on the surface temperature simulation data, if the surface temperature simulation data exceeds the temperature threshold, it is an abnormal temperature value; Based on the overhead thermal pipelines and the buried thermal pipelines, the geographical location of the abnormal temperature value is locked, and the abnormal level is determined by the temperature difference between the abnormal temperature value and the surface temperature simulation data before and after the abnormal temperature value.

Images