Urban pipeline repair monitoring method and system based on remote sensing
By building a pipeline network database and joint attention network, combined with remote sensing technology, the interpretation difficulty of remote sensing technology in urban pipeline monitoring is solved, and detailed monitoring and early warning of pipeline structure and faults are achieved, and data acquisition accuracy and interpretation are improved. .
Patent Information
- Application Number
- CN202411188287.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-28
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-08-28
AI Technical Summary
In the prior art, remote sensing technology is difficult to interpret and analyze in urban pipeline repair monitoring, resulting in limited interpretability of remote sensing images and the inability to provide detailed pipeline structure and fault abnormal locations.
By building a pipeline database and a joint attention network, combining remote sensing interference factors and data characteristics, a pipeline detection module with a dislocation twin network structure is designed for preprocessing and identification, and the active and passive remote sensing detection dimensions are used to perform dimension independent detection and mutual detection. Verification, perform binary processing based on the knowledge graph of the pipeline hierarchy, locate abnormal pipeline segments and generate visual warnings.
It improves the accuracy and accuracy of remote sensing data collection, realizes intelligent early warning and monitoring of urban pipelines, can promptly detect pipeline problems and provide detailed repair support.
Smart Images

Figure CN118691982B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of pipeline remote sensing information processing, and specifically to a remote sensing-based urban pipeline repair monitoring method and system. Background Art
[0002] Remote sensing technology, especially satellite remote sensing, provides a means to obtain ground information from high altitude for urban pipeline repair monitoring. Through remote sensing images, the distribution and direction of pipelines and changes in the surrounding environment can be observed, providing important data support for pipeline repair. In addition, remote sensing technology can also monitor the operation status of pipelines in real time, discover potential safety hazards in a timely manner, and provide strong guarantees for the safe operation of urban pipelines.
[0003] However, the application of remote sensing technology in urban pipeline repair monitoring is not always smooth. Due to the complexity of the urban environment, the interpretation and analysis of remote sensing images often require the use of professional technologies and tools. At the same time, due to the diversity and complexity of urban pipeline systems, different pipeline types and materials respond differently to remote sensing images, which increases the difficulty of remote sensing image interpretation.
[0004] In summary, due to the limitations of remote sensing technology and environmental influences, the existing technology has limited interpretability of remote sensing images and is unable to provide detailed pipeline structures and abnormal fault locations. Summary of the invention
[0005] Based on this, it is necessary to provide a remote sensing-based urban pipeline repair monitoring method and system to address the above-mentioned technical problems, which can achieve the technical effect of intelligent early warning monitoring of urban pipelines by optimizing the precision and accuracy of remote sensing data acquisition.
[0006] In a first aspect, a remote sensing-based urban pipeline repair monitoring method is provided, the method comprising: constructing a pipeline network database of a target area, the pipeline network database comprising pipeline network distribution data and a pipeline network hierarchical knowledge graph; combining remote sensing interference factors and remote sensing data characteristics to determine a joint attention network, the joint attention network comprising a preprocessing dimension and a remote sensing recognition dimension; combining the pipeline network database and the joint attention network to supervise the training of a pipeline detection module, the pipeline detection module being a dislocated twin network structure, and the pipeline detection module being communicatively connected to an intelligent early warning platform; determining a remote sensing detection dimension, returning remote sensing detection data to the pipeline detection module, performing dimension independent detection and mutual verification, and determining a detection distribution result, the remote sensing detection dimension comprising active and passive, the active type using natural electromagnetic wave radiation as a detection source, and the passive type using detection echo as a detection source; based on the pipeline network hierarchical knowledge graph, reading a standardized threshold, binarizing the detection distribution result, and locating an abnormal pipeline section, wherein the standardized threshold corresponds to a pipeline type one by one; based on the abnormal pipeline section, generating a pipeline repair single column, and performing a visual early warning on an intelligent early warning platform.
[0007] In the second aspect, a remote sensing-based urban pipeline repair monitoring system is provided, the system comprising: a pipeline database construction module, the pipeline database construction module is used to construct a pipeline database of a target area, the pipeline database comprises pipeline distribution data and a pipeline hierarchical knowledge graph; a joint attention network determination module, the joint attention network determination module is used to combine remote sensing interference factors with remote sensing data characteristics to determine a joint attention network, the joint attention network comprises a preprocessing dimension and a remote sensing recognition dimension; a pipeline detection module, the pipeline detection module is used to combine the pipeline database with the joint attention network, supervise the training of the pipeline detection module, the pipeline detection module is a dislocated twin network structure, and the pipeline detection module is communicatively connected to an intelligent early warning platform; a remote sensing detection dimension determination module Block, the remote sensing detection dimension determination module is used to determine the remote sensing detection dimension, return the remote sensing detection data to the pipeline detection module, perform independent dimension detection and mutual verification, and determine the detection distribution result. The remote sensing detection dimension includes active and passive types. The active type uses natural electromagnetic wave radiation as the detection source, and the passive type uses detection echo as the detection source; a binarization processing module, the binarization processing module is used to read the standardized threshold based on the pipeline network hierarchical knowledge graph, perform binarization processing on the detection distribution result, and locate the abnormal pipeline section, wherein the standardized threshold corresponds to the pipeline type one by one; a visual warning module, the visual warning module is used to generate a pipeline repair list based on the abnormal pipeline section, and perform visual warning on the intelligent warning platform.
[0008] The above-mentioned remote sensing-based urban pipeline repair monitoring method and system solve the technical problems in the prior art that the remote sensing image has limited interpretability and cannot provide detailed pipeline structure and abnormal fault locations due to the limitations of remote sensing technology and environmental impacts. By optimizing the precision and accuracy of remote sensing data acquisition, the technical effect of intelligent early warning monitoring of urban pipelines is achieved.
[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] Figure 1 A schematic diagram of a flow chart of a remote sensing-based urban pipeline repair monitoring method in one embodiment;
[0011] Figure 2 A schematic diagram of a flow chart of abnormal pipeline section integration in a remote sensing-based urban pipeline repair monitoring method in one embodiment;
[0012] Figure 3 The structure block diagram of a remote sensing-based urban pipeline repair monitoring system in one embodiment.
[0013] Explanation of the reference numerals: pipeline network database construction module 11, joint attention network determination module 12, pipeline detection module 13, remote sensing detection dimension determination module 14, binarization processing module 15, visual warning module 16. DETAILED DESCRIPTION
[0014] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0015] like Figure 1 As shown, the present application provides a remote sensing-based urban pipeline repair monitoring method, the method comprising:
[0016] Constructing a pipe network database for the target area, wherein the pipe network database includes pipe network distribution data and a pipe network hierarchical knowledge graph;
[0017] Remote sensing is a non-contact, long-distance detection technology that collects electromagnetic waves or other physical field information from the earth's surface and other objects through sensors, and processes and analyzes this information to identify the nature and motion state of objects. Remote sensing-based technologies have been widely used in many fields, including environmental monitoring, resource surveys, urban planning, disaster warning, etc. Urban pipelines are an important part of urban infrastructure, including power lines, telecommunications lines, industrial production pipelines and other types. These pipelines and lines together constitute the city's infrastructure network, providing necessary support and guarantee for the normal operation of the city and the lives of residents. This application provides a remote sensing-based urban pipeline repair monitoring method, which can quickly discover possible problems with pipelines, such as leakage, breakage, etc., and provide timely data support for repair work. Remote sensing technology can provide high-resolution image data, show the status of urban pipelines in detail, so that pipeline problems can be more accurately identified, and provide a more accurate basis for repair work. Remote sensing technology can achieve real-time monitoring and continuous observation and evaluation of urban pipelines. This helps to timely discover changes and problems in pipelines, and take appropriate measures to deal with them to avoid further deterioration of the problem. Through the above methods, the urban pipeline repair monitoring method based on remote sensing can provide timely, accurate and comprehensive pipeline information, providing strong support for the repair work.
[0018] Determine the atlas layer features, wherein the atlas layer features at least include pipeline type, application features, fault feature-standardization threshold, and fault evolution features;
[0019] Based on the characteristics of the atlas layer, the basic information of the pipeline network in the target area is retrieved and effective characteristics are screened, hierarchically integrated and intra-layer and inter-layer associations are performed to determine the initial atlas structure;
[0020] The initial graph structure is traversed, and graph nodes are traversed to perform coreference resolution and distribution adjustment to determine the pipe network hierarchical knowledge graph.
[0021] Graph layer features refer to the core attributes or features used to describe and distinguish different pipelines and their related information. These features are usually defined according to actual application requirements and the characteristics of the pipeline system, so as to facilitate the construction of a hierarchical knowledge graph and subsequent query, analysis and reasoning. At least pipeline type, application features, fault features-standardized thresholds. Pipeline type is to classify pipelines according to the material, purpose, diameter and other attributes of the pipeline, such as water supply pipelines, drainage pipelines, gas pipelines, etc.; application features include the geographical location of the pipeline, the region to which it belongs, the water supply or drainage system to which it belongs, etc.; fault features-standardized thresholds: define the characteristics and corresponding thresholds of various faults, such as leakage, rupture, blockage, etc., for subsequent monitoring and identification; fault evolution features, describing the trend and law of fault evolution over time, such as the increase in leakage volume, the expansion of the fault area, etc. Extract the basic information of the pipeline network from the existing urban pipeline management system, such as the location, length, diameter, material, etc. of the pipeline, and screen out the effective features related to the construction of the pipeline network knowledge graph according to the determined graph layer features. The extracted basic information of the pipeline network is organized according to a hierarchical structure, such as stratification according to pipeline type, geographical location, etc. In the same level, the association relationship between pipelines, such as connection relationship, adjacent relationship, etc., is established. At different levels, the association relationship between pipeline type, application characteristics and fault characteristics is established to facilitate subsequent analysis and reasoning. The integrated initial graph structure is traversed to check each graph node and relationship. During the traversal process, multiple graph nodes pointing to the same entity or concept are found, that is, the co-reference phenomenon. It is resolved by algorithm or manual method to ensure the consistency and accuracy of the graph. According to the traversal results and actual needs, the nodes and relationships in the graph are distributed and adjusted to optimize the structure and performance of the graph. After completing the above steps, a clearly structured and hierarchical pipeline network hierarchical knowledge graph is obtained, which not only contains the basic information of the pipeline network, but also establishes the association relationship between pipeline type, application characteristics and fault characteristics, providing strong support for subsequent pipeline network monitoring, fault warning and decision support.
[0022] Combining remote sensing interference factors with remote sensing data characteristics, determining a joint attention network, wherein the joint attention network includes a preprocessing dimension and a remote sensing recognition dimension;
[0023] Remote sensing interference factors refer to the fact that remote sensing data may be affected by various interference factors during the acquisition process, such as atmospheric interference, surface coverage, lighting conditions, sensor noise, etc., which may lead to a decline in data quality and affect subsequent analysis and identification; Remote sensing data characteristics: Remote sensing data has many characteristics, such as multi-source, macroscopic, periodic, comprehensive and quantitative. These characteristics give remote sensing data unique advantages in identifying objects and monitoring environmental changes; The key to the joint attention network lies in how to effectively combine the preprocessing dimension and the remote sensing recognition dimension. One possible approach is to use the attention mechanism to dynamically adjust the weights between the two dimensions. Specifically, the weight distribution between the two dimensions can be dynamically adjusted according to the needs of the current task and the characteristics of the remote sensing data to ensure that the network can better process the remote sensing data and extract useful information. Preprocessing Dimension In the joint attention network, the preprocessing dimension can be responsible for processing these preprocessing steps. This dimension can contain multiple sub-networks or modules, each of which is responsible for a specific preprocessing task. These sub-networks or modules can work in parallel or serially to ensure that the data is fully processed before being passed to the remote sensing recognition dimension. Remote sensing recognition dimension design,In the joint attention network, the remote sensing recognition dimension can be responsible for processing these recognition tasks.,This dimension can contain multiple sub-networks or modules, each of which is responsible for a specific recognition task.,These sub-networks or modules can share some common feature extraction layers to make full use of the characteristics of remote sensing data.,Through the above method, building a joint attention network to process remote sensing data requires full consideration of remote sensing interference factors and the characteristics of remote sensing data, and designing a suitable network structure and training strategy to ensure that the network can effectively process remote sensing data and extract useful information.
[0024] The remote sensing interference factors at least include the surface coverage interference dimension, the field disturbance interference dimension and the pipeline self-influence dimension;
[0025] Traversing the remote sensing interference factors and the remote sensing data characteristics, determining the remote sensing source data defects, and obtaining the first attention element of the preprocessing dimension;
[0026] Interactive pipeline monitoring needs to obtain the second attention factor of the remote sensing recognition dimension;
[0027] Based on the first attention element and the second attention element, the joint attention network is determined.
[0028] Remote sensing interference factors are key factors affecting the quality and recognition accuracy of remote sensing data. The surface coverage interference dimension, such as vegetation coverage, building occlusion, water reflection, etc., may cover or confuse pipeline information; the field disturbance interference dimension, for example, the distributed magnetic field, interferes with remote sensing transmission; the pipeline self-influence dimension includes pipeline burial depth, small pipelines, etc., and insufficient resolution of remote sensing detection; by traversing and analyzing the characteristics of remote sensing interference factors and remote sensing data, potential defects or problems in remote sensing source data can be identified, including noise, missing information, insufficient resolution, insufficient contrast, etc. Based on the analysis of remote sensing source data defects, the first attention factor of the preprocessing dimension can be determined, which can guide the preprocessing steps to minimize the impact of interference factors on data quality. Interacting with the specific needs of pipeline monitoring to determine the second attention factor of the remote sensing recognition dimension can guide the network on how to accurately extract and identify pipeline information from preprocessed data, such as feature extraction, target detection, etc.; Based on the first attention factor and the second attention factor, we can design the structure of the joint attention network. The network should be able to process information from both the preprocessing dimension and the remote sensing recognition dimension at the same time, and establish an effective connection between the two dimensions. For example, the preprocessing module preprocesses the remote sensing source data according to the first attention factor to improve the data quality; the feature extraction module extracts pipeline-related features from the preprocessed data; the joint attention mechanism dynamically adjusts the weights between the two dimensions through the attention mechanism to ensure that the network can focus on the most important information; the recognition and output module recognizes and processes the extracted features according to the second attention factor and outputs the final monitoring results. Through the above methods, the joint attention network can better cope with the influence of remote sensing interference factors and improve the accuracy and efficiency of urban pipeline monitoring.
[0029] Combine the pipeline network database with the joint attention network to supervise the training of a pipeline detection module, wherein the pipeline detection module is a dislocated twin network structure, and the pipeline detection module is communicatively connected with the intelligent early warning platform;
[0030] A pipeline detection module based on a pipeline network database and a joint attention network is designed. The module adopts a dislocated twin network structure and is connected to the intelligent early warning platform. The basic information, historical monitoring data, fault records, etc. of the pipeline are obtained from the existing pipeline network database. Remote sensing data related to the pipeline area, such as satellite images, drone aerial images, etc., are collected, and the remote sensing data is annotated to mark the location, status and other information of the pipeline for supervised training. A joint attention network is designed, which includes a preprocessing dimension and a remote sensing recognition dimension, as well as a joint attention mechanism for connecting the two dimensions. The annotated remote sensing data and the corresponding pipeline network database information are used as training data. The annotated data is used to supervise the training of the joint attention network to ensure that the network can accurately extract the feature information of the pipeline from the remote sensing data and identify the status of the pipeline. Twin networks are usually used to compare the similarity or difference of two inputs. In pipeline detection, one remote sensing image can be used as a reference and the other as an image to be detected. In order to capture the position change of the pipeline in the image, a dislocated structure is introduced. By performing small operations such as translation, rotation or scaling on the reference image and the image to be detected, the possible position changes of the pipeline in the image are simulated, and the features of the two images are extracted using the joint attention network. The similarity of these features is compared to determine whether the pipeline is misaligned or faulty. Through the above method, combined with the pipeline network database and the joint attention network, a pipeline detection module based on the misaligned twin network structure is supervised and trained, and connected to the intelligent early warning platform to achieve real-time monitoring and early warning of the urban pipeline system.
[0031] The pipeline detection module includes active branches and passive branches, and the branch training uses the pipeline network database and the joint attention network as the underlying architecture to distinguish sample data as training samples;
[0032] The branch structure includes:
[0033] Based on the joint attention network, build a first attention layer and a second attention layer;
[0034] A preprocessing layer is placed after the first attention layer, and a remote sensing recognition layer is placed after the second attention layer, and hierarchical full connections are performed to generate a pre-block;
[0035] The preceding block and the following block are combined to form a branch structure, wherein the following block is a data storage block.
[0036] The pipeline detection module includes active branches and passive branches. These two branches are trained with the pipeline database and joint attention network as the underlying architecture, and use the distinguishing sample data as training samples. Based on the joint attention network, the first attention layer is constructed, which is mainly responsible for capturing important image features and spatial information in the preprocessing stage for subsequent processing. After the first attention layer, the second attention layer is constructed. This layer is mainly responsible for further focusing and strengthening the features related to pipeline detection in the remote sensing recognition stage. The preprocessing layer is placed after the first attention layer. This layer is responsible for performing a series of image preprocessing operations, such as denoising, enhancement, and correction, to improve the quality of remote sensing images and make them more suitable for subsequent pipeline detection tasks. The remote sensing recognition layer is placed after the second attention layer. This layer uses the features extracted from the attention layer and combines the information in the pipeline database to perform pipeline detection tasks, such as target detection and classification. Through the detection data of remote sensing images, radar detection signal spectra, etc., spectral feature recognition, signal feature recognition, etc. are performed to determine the detection of pipeline location, geometric features, gas leakage, etc. The preprocessing layer and the remote sensing recognition layer are hierarchically fully connected to ensure smooth information transfer between layers and allow the network to automatically learn and optimize feature representation during training. The front block consists of the preprocessing layer, the first attention layer and the remote sensing recognition layer for active branches or only the preprocessing layer and the first attention layer for passive branches. This block is mainly responsible for extracting and preparing features related to pipeline detection from remote sensing images. The post-block, as a data storage block, is used to store and manage data related to pipeline detection tasks, such as pipeline network database information, training samples, model parameters, etc. This block does not directly participate in the feature extraction and recognition process, but provides necessary data support and storage functions for the entire branch structure. During the training process, the distinguishing sample data is used as the training sample. These sample data should contain remote sensing images of normal pipelines and faulty pipelines, as well as the corresponding annotation information. Through supervised learning, the active branch and the passive branch are trained so that they can accurately detect the location and status of the pipeline from the remote sensing image. By combining the pipeline network database and the joint attention network, and designing a branch structure including the attention layer, preprocessing layer, remote sensing recognition layer, and pre- and post-blocks, an efficient pipeline detection module can be constructed, which can achieve real-time monitoring and early warning of urban pipeline systems and improve the efficiency and safety of pipeline management.
[0037] The active branch and the passive branch are configured in reverse parallel, and a lateral interaction channel between the rear blocks is established to generate the pipeline detection module;
[0038] The execution modes of the pipeline detection module include:
[0039] Setting a preset detection cycle, performing active and passive synchronous remote sensing detection based on the preset detection cycle, and executing branch parallel analysis and post-block mutual verification;
[0040] Based on the time zone within the cycle, passive remote sensing inspection and independent analysis based on passive branches are performed.
[0041] The active branch and the passive branch are configured in a reverse parallel manner. This means that the two branches will process the remote sensing data at the same time, but the processing flow may be different to adapt to different detection needs; a lateral interaction channel is established between the post-blocks of the active branch and the passive branch. This channel allows the two branches to exchange information, share data or verify results when necessary to improve the accuracy and reliability of detection. The execution mode of the pipeline detection module includes active and passive synchronous remote sensing detection, setting a preset detection cycle, such as performing remote sensing detection once a day, week or month, or can be set according to the law of fault occurrence, etc., to ensure the accuracy of detection in parallel; for the time zone within the cycle, inspection can be carried out based on passive inspection, with limited accuracy but meeting the requirements of daily inspection. During the detection cycle, the active branch and the passive branch simultaneously acquire and process the remote sensing data, and the two branches execute their own analysis processes respectively, and compare and verify the results through the lateral interaction channel of the post-block. The active branch and the passive branch are executed in parallel in their respective analysis processes without interfering with each other. Each branch uses different algorithms to analyze the remote sensing data to extract the status information of the pipeline and detect potential problems. After the two branches complete the analysis, their post-blocks will compare and verify the results through the lateral interaction channel. If the results of the two branches are consistent, the confidence of the detection can be improved; if there are differences, further verification or processing is required.
[0042] Determine the remote sensing detection dimension, transmit the remote sensing detection data back to the pipeline detection module, perform independent dimension detection and mutual verification, and determine the detection distribution result. The remote sensing detection dimension includes active and passive types. The active type uses natural electromagnetic wave radiation as the detection source, and the passive type uses detection echo as the detection source;
[0043] Determine the dimension of remote sensing detection. Active remote sensing detection uses natural electromagnetic wave radiation as the detection source, such as radar, laser radar, etc. The active remote sensing system transmits electromagnetic wave signals, then receives the signals reflected by the target, and obtains the target information by analyzing these signals. It is suitable for situations where it is necessary to penetrate the surface cover, such as vegetation and soil to detect pipelines. Passive remote sensing detection uses detection echoes as the detection source, usually referring to the use of sunlight, thermal radiation, etc. radiated or reflected by the target itself. For example, passive, such as infrared remote sensing, performs remote sensing detection based on its own electromagnetic wave radiation to determine information such as thermal radiation characteristics. Passive remote sensing systems obtain target information by receiving electromagnetic waves radiated or reflected by the target itself, which is suitable for detecting pipelines on the surface, such as identifying the location and status of pipelines through satellite images or drone aerial images; active, such as radar remote sensing, analyzes by emitting radar detection waves based on the received echo reflection signals; preferably, active and passive detection and analysis can be performed at the same frequency, or only based on one of them. Through the above method, the remote sensing detection dimension can be determined, and the remote sensing detection data can be effectively transmitted back to the pipeline detection module for dimension independent detection and mutual verification, so as to accurately determine the detection distribution result.
[0044] Based on the pipeline network hierarchical knowledge graph, a standardized threshold is read, the detection distribution result is binarized, and the abnormal pipeline section is located, wherein the standardized threshold corresponds to the pipeline type one by one;
[0045] Accessing the network hierarchy knowledge graph contains detailed information on various pipeline types, including pipeline type, location, material, diameter, etc. In the knowledge graph, it is necessary to find the standardized threshold corresponding to the specific pipeline type, which is set based on historical data, expert knowledge or relevant standards to determine whether the pipeline status is normal. Prepare the detection distribution results, such as continuous numerical values or categorical data, for binarization. According to the standardized threshold read from the knowledge graph, a threshold is set for each pipeline type to determine whether the pipeline status is abnormal. Binarize the detection distribution results using the set threshold. Specifically, if the detection value of a pipeline section exceeds the corresponding threshold, it is marked as abnormal, such as 1 for abnormal and 0 for normal. After binarization, we can get a result set with abnormal pipeline sections marked, and use GIS systems, maps or other visualization tools to mark the abnormal pipeline sections on the map. This helps managers intuitively understand which pipelines have problems. Through the above method, the standardized threshold is read based on the network hierarchy knowledge graph, the detection distribution results are binarized, and the abnormal pipeline sections are accurately located. This helps improve the efficiency and safety of pipeline management.
[0046] like Figure 2As shown, based on the pipeline type, the detection distribution results are clustered and divided to determine multiple cluster areas;
[0047] Mapping the standardized threshold value to the plurality of clustering areas, taking the positive and negative difference of the threshold value as a reference, performing binary identification on each clustering area, and determining a binary distribution, wherein the abnormal direction is taken as a negative judgment standard, the negative difference is binarized as 1, and the positive difference is binarized as 0;
[0048] Identify the binary distribution, select box 1 to identify the partition, and integrate it as the abnormal pipeline segment.
[0049] According to the pipeline type information in the pipeline detection data, the detection distribution results are divided into different categories or subsets. In the data set of each pipeline type, the clustering algorithm is applied to cluster the detection data and group similar pipeline sections together for subsequent processing. The standardized thresholds corresponding to each pipeline type are read from the pipeline network hierarchical knowledge graph, and each cluster area is mapped to the standardized threshold of the corresponding pipeline type. Ensure that each cluster area has a threshold corresponding to it. For each pipeline section in each cluster area, the difference between its detection value and the corresponding threshold is calculated, and each pipeline section is binarized according to the positive and negative direction of the difference (i.e., whether the detection value exceeds the threshold). The abnormal direction is used as the negative judgment standard, that is, if the detection value exceeds the threshold (negative difference), it is marked as 1 (abnormal); if the detection value does not exceed the threshold (positive difference), it is marked as 0 (normal). In the binarized results, all pipeline sections marked as 1 (abnormal) are identified, and the identified abnormal pipeline sections are framed and integrated to form one or more continuous or discrete abnormal areas. These abnormal areas are the final abnormal pipeline sections. Through the above method, the detection distribution results can be effectively clustered, binarized and abnormal pipeline sections can be identified, providing strong support for subsequent pipeline maintenance and management.
[0050] Based on the abnormal pipeline section, a pipeline repair list is generated, and a visual warning is performed on the intelligent warning platform.
[0051] Integrate the relevant information of the abnormal pipeline section into a repair list, including but not limited to: the specific location of the abnormal section, pipeline type, abnormality type, degree of abnormality, possible repair suggestions, etc. Prioritize the abnormal sections in the repair list according to the degree and urgency of the abnormality. Ensure that serious or urgent abnormalities are handled first. Transmit the data of the repair list to the intelligent early warning platform through an appropriate interface or API. Ensure that the data can be transmitted to the early warning platform in real time and accurately. On the intelligent early warning platform, use GIS or map technology to display the abnormal pipeline section on the map. Different colors or icons can be used to represent different abnormality types and priorities. While displaying the abnormal sections on the map, warning information related to these sections can also be displayed, such as abnormal descriptions, repair suggestions, etc. This information can help maintenance personnel understand the abnormal situation more quickly and make corresponding treatments. Through the above method, the pipeline repair list generated based on the abnormal pipeline section can be visually warned on the intelligent early warning platform, providing strong support for pipeline maintenance and management.
[0052] Establishing a unified information platform, wherein the unified information platform is connected to multiple regional ports;
[0053] Based on the unified information platform, shared records are read, and the pipeline repair list is verified and repair plan decision-making is assisted.
[0054] Design a unified information platform architecture that can accommodate a large amount of data, support high concurrent access, and is easy to expand and maintain. Establish a high-performance, high-reliability database system to store pipeline repair lists, shared records, and other related information. Set up multiple interfaces or APIs on the unified information platform to connect ports in various regions. Ensure that each region can exchange data with the unified information platform in real time and securely. Synchronize the shared records of each region to the database of the unified information platform regularly or in real time. Verify the shared records synchronized to the unified information platform to ensure the accuracy and completeness of the data. Establish appropriate data indexes to quickly retrieve and query shared records. Verify the pipeline repair list based on shared records and other related information. Check whether the information in the repair list is consistent with the actual situation, such as pipeline location, abnormality type, abnormality degree, etc. Use the data analysis tools on the unified information platform to conduct in-depth analysis of the repair list to provide auxiliary decision support for the repair plan. For example, the difficulty of repair, the time and resources required, etc. can be predicted based on historical data. Optimize the repair plan based on the best practices, expert advice, and other related information in the shared records. Ensure that the repair plan is both in line with the actual situation and efficient and feasible. The verification results and auxiliary decision-making suggestions are fed back to the relevant regions or departments so that more accurate repair plans and solutions can be formulated based on this information. Through the above method, a unified information platform is established, and shared records are read based on the platform to conduct verification and auxiliary decision-making of pipeline repair plans. This will help improve the efficiency and accuracy of pipeline maintenance and provide a more reliable basis for related decisions.
[0055] like Figure 3 As shown, the embodiment of the present application includes a remote sensing-based urban pipeline repair monitoring system, the system comprising:
[0056] A pipe network database construction module 11, the pipe network database construction module 11 is used to construct a pipe network database of a target area, the pipe network database includes pipe network distribution data and a pipe network hierarchical knowledge graph;
[0057] A joint attention network determination module 12, the joint attention network determination module 12 is used to determine a joint attention network by combining remote sensing interference factors and remote sensing data characteristics, the joint attention network including a preprocessing dimension and a remote sensing recognition dimension;
[0058] A pipeline detection module 13, which is used to combine the pipeline network database with the joint attention network to supervise the training of a pipeline detection module. The pipeline detection module is a dislocated twin network structure, and the pipeline detection module is communicatively connected with the intelligent early warning platform;
[0059] A remote sensing detection dimension determination module 14, which is used to determine the remote sensing detection dimension, transmit remote sensing detection data back to the pipeline detection module, perform independent dimension detection and mutual verification, and determine the detection distribution result. The remote sensing detection dimension includes active and passive types. The active type uses natural electromagnetic wave radiation as the detection source, and the passive type uses detection echo as the detection source;
[0060] A binarization processing module 15, the binarization processing module 15 is used to read a standardized threshold based on the pipe network hierarchical knowledge graph, perform binarization processing on the detection distribution result, and locate the abnormal pipe section, wherein the standardized threshold corresponds to the pipe type one by one;
[0061] The visual warning module 16 is used to generate a pipeline repair list based on the abnormal pipeline section and perform a visual warning on the intelligent warning platform.
[0062] Furthermore, the embodiment of the present application also includes:
[0063] A graph layer feature determination module, the graph layer feature determination module is used to determine graph layer features, the graph layer features at least include pipeline type, application features, fault feature-standardization threshold, fault evolution features;
[0064] An initial atlas structure determination module, which is used to retrieve basic information of the pipeline network and screen effective features of the target area based on the atlas layer features, hierarchically integrate and perform intra-layer association and inter-layer association to determine the initial atlas structure;
[0065] A pipeline network hierarchical knowledge graph determination module is used to traverse the initial graph structure, traverse the graph nodes to perform co-reference resolution and distribution adjustment, and determine the pipeline network hierarchical knowledge graph.
[0066] Furthermore, the embodiment of the present application also includes:
[0067] A remote sensing interference factor inclusion module, wherein the remote sensing interference factor inclusion module is used for the remote sensing interference factor to include at least a surface coverage interference dimension, a field disturbance interference dimension, and a pipeline self-influence dimension;
[0068] A remote sensing source data defect determination module, the remote sensing source data defect determination module is used to traverse the remote sensing interference factors and the remote sensing data characteristics, determine the remote sensing source data defects, and obtain the first attention element of the preprocessing dimension;
[0069] An attention factor acquisition module, the attention factor acquisition module is used to interact with pipeline monitoring requirements and acquire a second attention factor of the remote sensing recognition dimension;
[0070] A joint attention network determination module, wherein the joint attention network determination module is used to determine the joint attention network based on the first attention element and the second attention element.
[0071] Furthermore, the embodiment of the present application also includes:
[0072] A training sample module, wherein the training sample module is used for the pipeline detection module and includes active branches and passive branches, and branch training uses the pipeline network database and the joint attention network as the underlying architecture to distinguish sample data as training samples;
[0073] A branch structure module, wherein the branch structure module is used, and the branch structure includes:
[0074] An attention layer building module, wherein the attention layer building module is used to build a first attention layer and a second attention layer based on the joint attention network;
[0075] A pre-block generation module, the pre-block generation module is used to perform hierarchical full connection on the first attention layer and the second attention layer, and to generate a pre-block;
[0076] A branch structure forming module is used to combine the front block and the rear block to form a branch structure, wherein the rear block is a data storage block.
[0077] Furthermore, the embodiment of the present application also includes:
[0078] A reverse parallel configuration module, the reverse parallel configuration module is used to perform reverse parallel configuration on the active branch and the passive branch, and establish a lateral interaction channel between the post-blocks to generate the pipeline detection module;
[0079] An execution mode module, wherein the execution mode of the pipeline detection module includes:
[0080] An active and passive synchronous remote sensing detection module, which is used to set a preset detection period, perform active and passive synchronous remote sensing detection based on the preset detection period, and perform branch parallel analysis and post-block mutual verification;
[0081] An independent analysis module is used to perform passive remote sensing inspection and independent analysis based on passive branches based on time zones within a cycle.
[0082] Furthermore, the embodiment of the present application also includes:
[0083] A clustering area determination module, the clustering area determination module is used to cluster the detection distribution results based on pipeline types to determine multiple clustering areas;
[0084] A binary distribution determination module, which is used to map the standardized threshold value and the plurality of clustering areas, and to perform binary identification on each clustering area based on the positive and negative difference of the threshold value, and determine the binary distribution, wherein the abnormal direction is used as the negative judgment standard, the negative difference is binarized as 1, and the positive difference is binarized as 0;
[0085] An integration module is used to identify the binary distribution, select 1 to identify the partition, and integrate it as the abnormal pipeline segment.
[0086] Furthermore, the embodiment of the present application also includes:
[0087] A unified information platform establishment module, the unified information platform establishment module is used to establish a unified information platform, the unified information platform is connected to multiple regional ports;
[0088] A shared record reading module is used to read shared records based on the unified information platform, verify the pipeline repair list and assist in decision-making on the repair plan.
[0089] For the specific embodiment of the urban pipeline repair monitoring system based on remote sensing, please refer to the embodiment of the urban pipeline repair monitoring method based on remote sensing above, which will not be repeated here. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
[0090] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0091] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application.
Claims
1. A remote sensing-based urban pipeline repair monitoring method, characterized in that: The method comprises: Constructing a pipe network database for the target area, wherein the pipe network database includes pipe network distribution data and a pipe network hierarchical knowledge graph; Combining remote sensing interference factors with remote sensing data characteristics, determining a joint attention network, wherein the joint attention network includes a preprocessing dimension and a remote sensing recognition dimension; Combine the pipeline network database with the joint attention network to supervise the training of a pipeline detection module, wherein the pipeline detection module is a dislocated twin network structure, and the pipeline detection module is communicatively connected with the intelligent early warning platform; Determine the remote sensing detection dimension, transmit the remote sensing detection data back to the pipeline detection module, perform independent dimension detection and mutual verification, and determine the detection distribution result. The remote sensing detection dimension includes active and passive types. The active type uses natural electromagnetic wave radiation as the detection source, and the passive type uses detection echo as the detection source; Based on the pipeline network hierarchical knowledge graph, a standardized threshold is read, the detection distribution result is binarized, and the abnormal pipeline section is located, wherein the standardized threshold corresponds to the pipeline type one by one; Based on the abnormal pipeline section, a pipeline repair list is generated, and a visual warning is performed on the intelligent warning platform; The pipeline detection module includes active branches and passive branches, and the branch training uses the pipeline network database and the joint attention network as the underlying architecture to distinguish sample data as training samples; The branch structure includes: Based on the joint attention network, build a first attention layer and a second attention layer; A preprocessing layer is placed after the first attention layer, and a remote sensing recognition layer is placed after the second attention layer, and hierarchical full connections are performed to generate a pre-block; Combining the front block and the rear block to form a branch structure, wherein the rear block is a data storage block; The supervised training pipeline detection module comprises: The active branch and the passive branch are configured in reverse parallel, and a lateral interaction channel between the rear blocks is established to generate the pipeline detection module; The execution modes of the pipeline detection module include: Setting a preset detection cycle, performing active and passive synchronous remote sensing detection based on the preset detection cycle, and executing branch parallel analysis and post-block mutual verification; Based on the time zone within the cycle, passive remote sensing inspection and independent analysis based on passive branches are performed.
2. The method according to claim 1, characterized in that The pipe network database includes a pipe network hierarchical knowledge graph, including: Determine the atlas layer features, wherein the atlas layer features at least include pipeline type, application features, fault feature-standardization threshold, and fault evolution features; Based on the characteristics of the atlas layer, the basic information of the pipeline network in the target area is retrieved and effective characteristics are screened, hierarchically integrated and intra-layer and inter-layer associations are performed to determine the initial atlas structure; The initial graph structure is traversed, and graph nodes are traversed to perform coreference resolution and distribution adjustment to determine the pipe network hierarchical knowledge graph.
3. The method according to claim 1, characterized in that The determining of the joint attention network comprises: The remote sensing interference factors at least include the surface coverage interference dimension, the field disturbance interference dimension and the pipeline self-influence dimension; Traversing the remote sensing interference factors and the remote sensing data characteristics, determining the remote sensing source data defects, and obtaining the first attention element of the preprocessing dimension; Interactive pipeline monitoring needs to obtain the second attention factor of the remote sensing recognition dimension; Based on the first attention element and the second attention element, the joint attention network is determined.
4. The method according to claim 1, characterized in that Binarization is performed on the detection distribution result to locate the abnormal pipeline section, including: Based on the pipeline type, clustering the detection distribution results to determine multiple cluster areas; Mapping the standardized threshold value to the plurality of clustering areas, taking the positive and negative difference of the threshold value as a reference, performing binary identification on each clustering area, and determining a binary distribution, wherein the abnormal direction is taken as the negative judgment standard, the negative difference is binarized as 1, and the positive difference is binarized as 0; Identify the binary distribution, select box 1 to identify the partition, and integrate it as the abnormal pipeline segment.
5. The method according to claim 1, characterized in that The method further comprises: Establishing a unified information platform, wherein the unified information platform is connected to multiple regional ports; Based on the unified information platform, shared records are read, and the pipeline repair list is verified and repair plan decision-making is assisted.
6. The urban pipeline repair monitoring system based on remote sensing is characterized by: The system comprises: A pipe network database construction module, which is used to construct a pipe network database of a target area, wherein the pipe network database includes pipe network distribution data and a pipe network hierarchical knowledge graph; A joint attention network determination module, the joint attention network determination module is used to determine a joint attention network by combining remote sensing interference factors and remote sensing data characteristics, the joint attention network including a preprocessing dimension and a remote sensing recognition dimension; A pipeline detection module, which is used to combine the pipeline network database with the joint attention network to supervise the training of the pipeline detection module. The pipeline detection module is a dislocated twin network structure, and the pipeline detection module is communicatively connected with the intelligent early warning platform; A remote sensing detection dimension determination module, which is used to determine the remote sensing detection dimension, transmit remote sensing detection data back to the pipeline detection module, perform independent dimension detection and mutual verification, and determine the detection distribution result. The remote sensing detection dimension includes active and passive types. The active type uses natural electromagnetic wave radiation as the detection source, and the passive type uses detection echo as the detection source; A binarization processing module, the binarization processing module is used to read a standardized threshold based on the pipe network hierarchical knowledge graph, perform binarization processing on the detection distribution result, and locate the abnormal pipe section, wherein the standardized threshold corresponds to the pipe type one by one; A visual warning module, which is used to generate a pipeline repair list based on the abnormal pipeline section and perform visual warning on the intelligent warning platform; A training sample module, wherein the training sample module is used for the pipeline detection module and includes active branches and passive branches, and branch training uses the pipeline network database and the joint attention network as the underlying architecture to distinguish sample data as training samples; A branch structure module, wherein the branch structure module is used, and the branch structure includes: An attention layer building module, wherein the attention layer building module is used to build a first attention layer and a second attention layer based on the joint attention network; A pre-block generation module, the pre-block generation module is used to perform hierarchical full connection on the first attention layer and the second attention layer, and to generate a pre-block; A branch structure forming module, the branch structure forming module is used to combine the front block and the rear block to form a branch structure, wherein the rear block is a data storage block; A reverse parallel configuration module, the reverse parallel configuration module is used to perform reverse parallel configuration on the active branch and the passive branch, and establish a lateral interaction channel between the post-blocks to generate the pipeline detection module; An execution mode module, wherein the execution mode of the pipeline detection module includes: An active and passive synchronous remote sensing detection module, which is used to set a preset detection period, perform active and passive synchronous remote sensing detection based on the preset detection period, and perform branch parallel analysis and post-block mutual verification; An independent analysis module is used to perform passive remote sensing inspection and independent analysis based on passive branches based on time zones within a cycle.
Citation Information
Patent Citations
UAV and network communication data acquisition based tour inspection and maintenance management platform for oil and gas pipelines
CN108304983A
Atmosphere environment ground monitoring station disposing networking method based on satellite remote sensing
CN109655583A
Land oil and gas pipeline defect diagnosis method based on hyperspectral remote sensing technology
CN112116566A
Remote sensing image change detection method and system based on twinborn attention fusion network
CN113469074A
Storage, oil and gas pipeline monitoring and early warning method, device and equipment
CN114458967A
