A GIS-based urban lifeline pipeline risk identification method

Abstract

The application discloses a kind of city lifeline pipeline risk identification method based on GIS, comprising: S1: collecting underground water original data, metal mineral original data, pipeline data, engineering latitude and longitude coordinates, population distribution map data and pre-processing;S2: the feature of underground environment is extracted by convolution layer and the pipeline graph model after pre-processing is updated, obtain the updated pipeline graph model, then construct the improved double gate control graph neural network, extract the fusion feature of all pipelines;S3: the construction influence feature of pipeline, the population risk feature of pipeline is extracted;Then combine the fusion feature of all pipelines, extract comprehensive risk feature, and convert it into risk level;S4: calculate pipeline risk graph model, and mark display in GIS platform.The present application can solve the problem that the traditional method is inefficient, it is difficult to combine multi-source heterogeneous data, and then the problem that the accurate risk identification of city lifeline pipeline cannot be realized.

Description

Technical Field

[0001] This invention relates to the field of pipeline risk identification technology, and in particular to a GIS-based method for identifying risks in urban lifeline pipelines. Background Technology

[0002] Urban lifeline pipelines are primarily buried deep underground in complex and variable environments. They are subject to long-term corrosion from groundwater, chemicals, and stray currents, and their inherent weaknesses exist due to variations in pipeline materials, manufacturing standards, and operational loads. Furthermore, external factors such as road construction, excavation, and geological subsidence further exacerbate the risk of damage to the pipeline structure. Moreover, the actual consequences of the same pipeline risk can differ significantly across different urban areas due to variations in topography, population density, traffic load, and other environmental conditions. This multi-source and dynamic nature of risk makes it difficult to achieve comprehensive and accurate risk perception and systematic management using traditional methods.

[0003] Traditionally, risk identification for this type of pipeline has relied primarily on periodic manual inspections, qualitative judgments based on historical experience, and simple statistical analysis. Manual inspections are not only inefficient and costly, but also heavily dependent on the personal experience of inspectors, making it difficult to uncover hidden potential risks and impossible to achieve real-time monitoring of large-scale pipeline networks. Experience-based methods are highly subjective, lack quantitative evidence, and struggle to form standardized, reusable assessment systems. Furthermore, when processing survey data using traditional statistical models, only superficial analysis of single risk factors is often possible. For comprehensive risks arising from the interplay of complex factors such as pipeline material, corrosion status, third-party activities, and dynamic environments, effective modeling and fusion analysis capabilities are lacking, leading to biased assessment results and delayed early warnings. In recent years, although some studies have attempted to use deep learning technology to automatically learn risk characteristics, existing methods usually view various risk factors in isolation and have failed to construct a unified analytical framework that can organically couple multi-source information such as the pipeline's own physical properties, external erosion environment, third-party interference, and surface dynamics. The models suffer from drawbacks such as fragmentation, poor adaptability, and weak interpretability, and cannot accurately quantify the real impact of risks in different urban contexts. Therefore, there are still significant deficiencies in the comprehensiveness and accuracy of the assessment. Summary of the Invention

[0004] In view of this, the present invention aims to provide a GIS-based method for identifying risks in urban lifeline pipelines, in order to solve the problems of low inspection efficiency and difficulty in combining multi-source heterogeneous data in traditional methods, thus making it impossible to achieve accurate risk identification of urban lifeline pipelines.

[0005] A GIS-based method for identifying risks in urban lifeline pipelines includes:

[0006] S1: Collect raw groundwater data, raw metal mineral data, pipeline data, engineering latitude and longitude coordinates, and population distribution map data, and preprocess them to obtain preprocessed groundwater data, preprocessed mineral data, preprocessed pipeline diagram model, preprocessed engineering latitude and longitude coordinate data, and preprocessed population distribution map data.

[0007] S2: Based on the preprocessed groundwater data and preprocessed mineral data, the underground environment features are extracted through a convolutional layer, and then the associated features of the underground environment are extracted and the preprocessed pipeline graph model is updated to obtain the updated pipeline graph model. Then, an improved dual-gated graph neural network is constructed to extract the fusion features of all pipelines.

[0008] The improved dual-gated graph neural network first processes the pipeline material data and laying depth data in the pipeline data through a fully connected layer based on the updated pipeline graph model to obtain pipeline attribute features. At the same time, it calculates the environmental feature similarity matrix and attribute matching matrix, and then obtains the edge attention matrix through the Softmax function. Then, based on the underground environment association features and pipeline attribute features, it calculates the environment sensitivity gate and attribute adaptation gate respectively through the Sigmoid function. Finally, it combines the edge attention matrix, environment sensitivity gate and attribute adaptation gate to calculate the fused features of all pipelines.

[0009] S3: Based on the preprocessed engineering latitude and longitude coordinate data, the preprocessed population distribution map data, and the updated pipeline diagram model, extract the construction impact characteristics and population risk characteristics of the pipeline respectively; then combine the fusion characteristics of all pipelines to extract the comprehensive risk characteristics and convert them into risk levels;

[0010] S4: Calculate the pipeline risk map model based on the risk level and display it in the GIS platform.

[0011] Furthermore, step S1 also includes:

[0012] S11: Collect raw groundwater data and raw metal mineral data from urban underground areas, with the data type being raster data; the raw groundwater data includes groundwater level and groundwater flow velocity; the raw metal mineral data includes metal mineral types and metal mineral content; outliers are removed from the raw groundwater data and raw metal mineral data using the three-standard-deviation principle, and missing values ​​are filled in using the mean method to obtain cleaned groundwater data and cleaned mineral data.

[0013] S12: Collect pipeline data for urban lifeline pipelines, including pipeline material data and burial depth data. The data type of the pipeline data is graph structure data, with pipeline connection points as graph nodes, pipelines as edges, and pipeline material data and burial depth data as edge attributes. The pipeline material data includes pipeline identification codes corresponding to PVC, steel, and cast iron materials. The burial depth data includes the vertical depth information of each pipeline section buried underground.

[0014] S13: Collect the latitude and longitude coordinates of various construction projects within the city. The data type is point data. Duplicate coordinates are removed to obtain cleaned latitude and longitude coordinate data.

[0015] S14: Collect population distribution map data for different administrative regions of the city. The data type is raster data. The population distribution map data includes the population density of each administrative region of the city.

[0016] S15: Unify the geographic coordinate benchmarks for the cleaned groundwater data, cleaned mineral data, pipeline data, cleaned engineering latitude and longitude coordinate data, and population distribution map data to obtain preprocessed groundwater data, preprocessed mineral data, preprocessed pipeline diagram model, preprocessed engineering latitude and longitude coordinate data, and preprocessed population distribution map data.

[0017] Furthermore, step S2 also includes:

[0018] S21: Based on the preprocessed groundwater data and preprocessed mineral data, subsurface environmental features are extracted using a convolutional layer. The calculation method is as follows:

[0019]

[0020] in, Characteristics of the underground environment It is a convolutional layer. For splicing operations, This is the pre-processed groundwater data. This is the preprocessed mineral data;

[0021] S22: Based on the start and end coordinates of each pipeline in the preprocessed pipeline graph model, determine the grid range covered by each pipeline, extract the average feature value of the underground environment features within the grid range as the underground environment association feature of the corresponding pipeline, and incorporate it into the edge attribute to obtain the updated pipeline graph model.

[0022] S23: Based on the updated pipeline diagram model, construct an improved dual-gated graph neural network to extract pipeline structural features;

[0023] Furthermore, step S23 also includes:

[0024] S231: Based on the updated pipeline graph model, calculate the pipeline attribute features, the edge attention matrix integrating the environment and attributes, and calculate the following:

[0025]

[0026]

[0027] ;

[0028]

[0029] in, Let be the edge attention matrix. For the Softmax function, This is an environmental feature similarity matrix. For balance coefficient, This is the attribute matching degree matrix. Let be the element in the i-th row and j-th column of the environmental feature similarity matrix, where i is the index of the first node and j is the index of the second node. The underground environment features associated with the pipeline connecting nodes i and j. This represents the average of the underground environmental characteristics associated with all pipelines. To obtain the L2 norm, The pipe attribute characteristics of the pipe connecting nodes i and j. It is a fully connected layer. These are the pipe material data and burial depth data for the pipes connecting nodes i and j, respectively. Let be the element in the i-th row and j-th column of the attribute matching degree matrix. To take the absolute value, This represents the average value of the pipe attribute characteristics for all pipes.

[0030] S232: Based on underground environment characteristics, pipeline attribute characteristics, and burial depth data, calculate the environmental sensitivity threshold and attribute adaptation threshold. The calculation method is as follows:

[0031]

[0032]

[0033] in, As an environmentally sensitive door, For the Sigmoid function, This is an environmentally sensitive weight matrix. It is a vector composed of the associated underground environmental features of all pipelines. To obtain the minimum value, This is the environment-sensitive bias vector. For attribute adaptation doors, To adapt the weight matrix to the attributes, It is a vector formed by concatenating the pipe attribute features of all pipes. A vector composed of stitched data of the burial depth of all pipelines. Adapt bias vectors to properties;

[0034] S233: Calculate the fusion features of all pipelines based on the edge attention matrix, environment-sensitive gate, and attribute adaptation gate. The calculation method is as follows:

[0035]

[0036] in, For the fusion characteristics of all pipes, For Hadama accumulation, This is the edge attribute matrix. This is the burial depth weight matrix. This is the pipeline attribute weight matrix.

[0037] It should be further explained that the risk characteristics of urban lifeline pipelines are jointly determined by factors such as the underground environment (groundwater, minerals) and the pipeline's own properties (materials, burial depth). Moreover, the two do not act in isolation, but rather have a complex interactive influence. The erosion intensity of the underground environment will have different effects due to the corrosion resistance of the pipeline material and the burial depth. At the same time, the risk correlation between different pipelines is also affected by their environmental similarity and property matching degree.

[0038] This invention constructs an improved dual-gated graph neural network, achieving multiple effects through the synergistic effect of its components: First, when calculating the edge attention matrix for fusing environment and attributes, environmental feature similarity reflects the degree of correlation between different pipelines in terms of underground environmental impact, while attribute matching reflects the degree of correlation between pipelines in terms of material and burial depth. Combining these two factors and processing them with the Softmax function ensures that the topological correlation weights between pipelines simultaneously consider the dual influence of environment and attributes, providing a more realistic correlation basis for subsequent feature fusion. Second, the environmental sensitivity gate, calculated based on underground environmental correlation features, can dynamically adjust the contribution weight of environmental factors to pipeline features, making the environmental impact of pipelines with stronger environmental erosion more prominent in the fused features, while attributes... The adaptation gate combines pipeline attribute characteristics with burial depth data to dynamically adjust the contribution weight of the pipeline's own risk resistance (material and burial depth), making pipelines with strong corrosion resistance or reasonable burial depth more prominent in the fusion features, thus achieving differentiated control over environmental and attribute impacts. Finally, the calculation of the fusion features combines the pipeline association information brought by the edge attention matrix and the dynamic weights of the two gates with environmental and attribute features respectively through Hadamard product, and then integrates them through addition, so that the final fusion features not only include the environmental and attribute features of individual pipelines, but also incorporate the risk association information between pipelines. At the same time, the gating mechanism ensures a reasonable proportion of environmental and attribute impacts in different scenarios, effectively achieving accurate extraction of pipeline risk features.

[0039] Existing technologies for addressing such problems typically employ a single feature (e.g., based solely on pipeline attributes) to construct a graph neural network, or fuse environmental and attribute features through simple weighting. Their attention mechanisms often rely solely on a single dimension of the environment or attribute to calculate pipeline association weights, and their gating mechanisms are mostly single-gating, failing to distinguish the different roles of the environment and attributes. In contrast, this invention captures the risk associations between pipelines more comprehensively by fusing an edge attention matrix of the environment and attributes. By using a dual-gating mechanism to dynamically adjust environmental influences and attribute resistance respectively, the feature fusion becomes more scenario-specific, avoiding feature bias caused by single features or simple weighting, and more accurately reflecting the actual risk characteristics of the pipeline.

[0040] Furthermore, step S3 also includes:

[0041] Based on the preprocessed engineering latitude and longitude coordinate data and the updated pipeline diagram model, the straight-line distance between the k-th pipeline and each construction point is calculated using an exponential function and then spliced ​​together. After processing with a fully connected layer, the construction influence characteristics of the k-th pipeline are obtained. Then, the construction influence characteristics of all pipelines are spliced ​​together and processed with a fully connected layer to obtain the construction influence characteristics of all pipelines.

[0042] Furthermore, step S3 also includes:

[0043] Based on the preprocessed population distribution map data and the updated pipeline graph model, the weighted sum of the population distribution grid within a 50-meter radius of the pipeline midpoint is calculated using the Gaussian kernel function. Combined with the city's average population density, the population risk characteristics of the k-th pipeline are obtained. Then, the population risk characteristics of all pipelines are spliced ​​together and processed through a fully connected layer to obtain the population risk characteristics of all pipelines.

[0044] Furthermore, step S3 also includes:

[0045] S31: Based on the preprocessed engineering latitude and longitude coordinate data and the updated pipeline diagram model, extract the construction impact characteristics of the pipeline. The calculation method is as follows:

[0046]

[0047]

[0048] in, Let the construction impact characteristics of the k-th pipeline be defined. It is a fully connected layer. To concatenate the elements within the parentheses from s=1 to s=Num, This represents the total number of construction sites. For engineering construction index, It is an exponential function. The construction attenuation coefficient is the coefficient of performance. Let k be the straight-line distance between the k-th pipeline and the s-th construction site, where k is the pipeline index. The coordinates of the midpoint of the k-th pipe in the updated pipe diagram model are: This refers to the preprocessed latitude and longitude coordinates of the s-th construction site. Let K represent the construction impact characteristics of all pipelines, where K is the total number of pipelines. To concatenate the elements within the parentheses from k=1 to k=K consecutively;

[0049] S32: Based on the preprocessed population distribution map data and the updated pipeline diagram model, extract the population risk characteristics of the pipeline. The calculation method is as follows:

[0050]

[0051]

[0052] in, The population risk characteristics of the k-th pipeline, The Gaussian kernel function is calculated by summing the weighted values ​​of the population distribution grid within a 50-meter radius of the pipe's midpoint. This is the preprocessed population distribution map data. Average population density in cities Population risk characteristics for all pipelines;

[0053] S33: Based on the fusion characteristics, construction impact characteristics, and population risk characteristics of all pipelines, a comprehensive risk characteristic is extracted through a scene attention mechanism. The calculation method is as follows:

[0054]

[0055]

[0056] in, For scene attention features, For the Softmax function, For splicing operations, To comprehensively assess risk characteristics, This is the first element in the scene attention weight vector. For the fusion characteristics of all pipes, This is the second element in the scene attention weight vector. This is the third element in the scene attention weight vector;

[0057] S34: Based on the comprehensive risk characteristics, it is converted into the risk level of all pipelines, calculated as follows:

[0058]

[0059] in, Risk levels for all pipelines are categorized as low, lower, medium, higher, and high. This is the maximum value index function.

[0060] It should be further explained that the comprehensive risk of urban lifeline pipelines is the result of the combined effects of three factors: the integration characteristics of the pipeline itself with the underground environment, the damage risk from third-party construction, and the severity of the consequences in densely populated areas. Moreover, the weight of these three factors is not fixed. In areas with intensive construction, the importance of construction impact characteristics increases significantly; in densely populated areas, the contribution of population risk characteristics is more prominent; while in areas with severe environmental erosion or weak pipeline properties, integration characteristics dominate.

[0061] This invention constructs a comprehensive risk feature. First, the extraction of construction impact features uses an exponential function to process the distance between the pipeline and the construction point, reflecting the characteristic that the construction impact decays with spatial distance. Then, a fully connected layer integrates the impact of all construction points on a single pipeline and the construction impact of all pipelines, ensuring that the quantification of construction risk conforms to spatial patterns and has consistency. Second, the extraction of population risk features uses a Gaussian kernel function to focus on the local population distribution around the pipeline and combines it with the average urban population density, highlighting the direct impact of the population around the pipeline and using average density for overall calibration. Finally, a scene attention mechanism processes the splicing result of the fused features, construction impact features, and population risk features through a fully connected layer, and generates dynamic weights through a Softmax function, enabling the weight allocation to adapt to different scenarios—when the pipeline is in a construction zone, the weight of the construction impact features automatically increases; when the pipeline is in a densely populated area, the weight of the population risk features increases accordingly; when the pipeline is dominated by its own attributes or underground environment, the weight of the fused features is dominant. Finally, a weighted summation is used to obtain the comprehensive risk feature, ensuring that the comprehensive risk feature can accurately reflect the actual contribution of each factor under different scenarios.

[0062] Existing technologies, when dealing with such comprehensive risk characteristics, typically employ fixed weights to simply superimpose various characteristics or select only a subset for evaluation. They fail to dynamically adjust the importance of each characteristic based on the specific scenario of the pipeline's location (e.g., construction zones, densely populated areas). For example, some technologies weight construction impact, population distribution, and pipeline characteristics according to a preset ratio, leading to underestimation of risk in densely populated areas due to insufficient construction weights, or overestimation of risk in sparsely populated areas due to excessive population weights. In contrast, this invention dynamically generates weights through a scenario attention mechanism, enabling comprehensive risk characteristics to automatically adapt to the actual scenario of the pipeline's location. This avoids the limitations of fixed weights and significantly improves the scenario fit and accuracy of risk assessment.

[0063] In terms of integration and fusion, the key challenge of this invention lies in how to organically integrate three types of risk features with different properties (pipeline coupling characteristics with the environment, spatial distance attenuation characteristics, and local population distribution characteristics), preserving the unique information of each feature while avoiding interference between features. Its innovations are: First, the construction impact feature, through the combination of a distance attenuation function and a fully connected layer, achieves a precise transformation of construction risk from spatial patterns to feature vectors, overcoming the limitation that simple distance statistics cannot quantify the intensity of impact. Second, the population risk feature, through the combination of Gaussian kernel function focusing on local areas and average density calibration, balances the direct impact of the population around the pipeline with the overall urban population background, solving the bias problem of isolated use of local population data. Third, the scene attention mechanism, through dynamic weight allocation, achieves scene-adaptive fusion of the three types of features, overcoming the shortcomings of traditional static weighting in adapting to complex scenarios. This allows the comprehensive risk feature to fully cover the multi-dimensional impact of pipeline risks and accurately reflect the dominant factors in different scenarios, achieving deep collaborative fusion of multi-source risk information.

[0064] Furthermore, step S4 also includes:

[0065] S41: Add the risk level of each pipeline from the risk levels of all pipelines to the attribute list of the corresponding pipeline in the updated pipeline diagram model to obtain the pipeline risk diagram model;

[0066] S42: Integrate and load the pipeline risk map model according to the data format compatible with the GIS platform;

[0067] S43: Differentiated visual identifiers are used to display each pipeline, enabling visualization of the spatial distribution of risk levels of urban lifeline pipelines.

[0068] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0069] (1) In view of the problems of low inspection efficiency and difficulty in integrating and analyzing multi-source heterogeneous data in traditional pipeline risk identification, this invention first collects and preprocesses groundwater, metal minerals, pipelines, engineering latitude and longitude coordinates, and population distribution map data. Then, it extracts underground environmental features through convolutional layers, updates the pipeline map model in combination with pipeline data, and uses an improved dual-gated graph neural network to fuse underground environmental correlation features and pipeline attribute features to obtain fused features. Subsequently, it extracts construction impact features and population risk features, combines the fused features to generate comprehensive risk features and converts them into risk levels. Finally, it is displayed on the GIS platform. Through the collaborative fusion of multiple features, it achieves accurate and comprehensive risk identification and significantly improves the efficiency of urban lifeline pipeline risk management.

[0070] (2) In view of the problem that existing technologies rely on single features, simple weighting, or single-dimensional attention and single gating, which cannot distinguish the interaction between underground environment and pipeline attributes (material, burial depth), this invention innovatively constructs an improved dual-gated graph neural network. First, it integrates environmental feature similarity and attribute matching degree, and calculates the edge attention matrix through the Softmax function to fully capture pipeline risk association. Then, it dynamically adjusts the contribution weights of environment and attribute through environmental sensitivity gate and attribute adaptation gate. Finally, it combines the gating weights and features with the Hadamard product and integrates them additively to obtain fused features containing pipeline association information and scene adaptation, which effectively solves the feature bias problem of traditional methods and improves the accuracy of risk feature extraction.

[0071] (3) In view of the problem that existing technologies have fixed weights and insufficient adaptability in different scenarios such as engineering construction areas and high population density, this invention first processes the distance between the pipeline and the construction point through an exponential function to reflect the distance attenuation characteristics, and then integrates the construction impact features through a fully connected layer to ensure that the quantification conforms to spatial laws and is consistent; then, a Gaussian kernel is used to focus on the population around the pipeline, and the population risk features are obtained by combining the average density of the city, taking into account both local impact and overall calibration; finally, through the scene attention mechanism, the three types of features are spliced ​​together and then the dynamic weights are generated through a fully connected layer and a Softmax function, which automatically adapts to scenarios such as construction areas and densely populated areas, and weights the comprehensive risk features to significantly improve the scene fit and accuracy of risk assessment. Attached Figure Description

[0072] Figure 1 A flowchart illustrating a GIS-based urban lifeline pipeline risk identification method provided by the present invention;

[0073] Figure 2 This invention provides a schematic diagram of the risk levels of lifeline pipelines in a certain area of ​​a city. Detailed Implementation

[0074] The present invention will be further described below with reference to the accompanying drawings, but this is not intended to limit the present invention in any way. Any modifications or substitutions made based on the teachings of the present invention shall fall within the protection scope of the present invention.

[0075] Example 1: A GIS-based method for identifying risks in urban lifeline pipelines, such as... Figure 1 As shown, it includes the following steps:

[0076] S1: Collect raw groundwater data, raw metal and mineral data, pipeline data, engineering latitude and longitude coordinates, and population distribution map data, and preprocess them to obtain preprocessed groundwater data, preprocessed mineral data, preprocessed pipeline diagram model, preprocessed engineering latitude and longitude coordinate data, and preprocessed population distribution map data, including:

[0077] S11: Collect raw groundwater data and raw metal mineral data from urban underground areas, with the data type being raster data; the raw groundwater data includes groundwater level and groundwater flow velocity; the raw metal mineral data includes metal mineral types and metal mineral content; outliers are removed from the raw groundwater data and raw metal mineral data using the three-standard-deviation principle, and missing values ​​are filled in using the mean method to obtain cleaned groundwater data and cleaned mineral data.

[0078] S12: Collect pipeline data for urban lifeline pipelines, including pipeline material data and burial depth data. The data type of the pipeline data is graph structure data, with pipeline connection points as graph nodes, pipelines as edges, and pipeline material data and burial depth data as edge attributes. The pipeline material data includes pipeline identification codes corresponding to PVC, steel, and cast iron materials. The burial depth data includes the vertical depth information of each pipeline section buried underground.

[0079] S13: Collect the latitude and longitude coordinates of various construction projects within the city. The data type is point data. Duplicate coordinates are removed to obtain cleaned latitude and longitude coordinate data.

[0080] S14: Collect population distribution map data for different administrative regions of the city. The data type is raster data. The population distribution map data includes the population density of each administrative region of the city.

[0081] S15: Unify the geographic coordinate benchmarks for the cleaned groundwater data, cleaned mineral data, pipeline data, cleaned engineering latitude and longitude coordinate data, and population distribution map data to obtain preprocessed groundwater data, preprocessed mineral data, preprocessed pipeline diagram model, preprocessed engineering latitude and longitude coordinate data, and preprocessed population distribution map data.

[0082] S2: Based on the preprocessed groundwater and mineral data, subsurface environmental features are extracted using convolutional layers. Then, subsurface environmental correlation features are extracted and used to update the preprocessed pipeline graph model, resulting in an updated pipeline graph model. Finally, an improved dual-gated graph neural network is constructed to extract the fused features of all pipelines, including:

[0083] S21: Based on the preprocessed groundwater data and preprocessed mineral data, subsurface environmental features are extracted using a convolutional layer. The calculation method is as follows:

[0084]

[0085] in, Characteristics of the underground environment It is a convolutional layer. For splicing operations, This is the pre-processed groundwater data. This is the preprocessed mineral data;

[0086] S22: Based on the start and end coordinates of each pipeline in the preprocessed pipeline graph model, determine the grid range covered by each pipeline, extract the average feature value of the underground environment features within the grid range as the underground environment association feature of the corresponding pipeline, and incorporate it into the edge attribute to obtain the updated pipeline graph model.

[0087] S23: Based on the updated pipeline diagram model, construct an improved dual-gated graph neural network to extract pipeline structural features, including:

[0088] S231: Based on the updated pipeline graph model, calculate the pipeline attribute features, the edge attention matrix integrating the environment and attributes, and calculate the following:

[0089]

[0090]

[0091] ;

[0092]

[0093] in, Let be the edge attention matrix. For the Softmax function, This is an environmental feature similarity matrix. For balance coefficient, This is the attribute matching degree matrix. Let be the element in the i-th row and j-th column of the environmental feature similarity matrix, where i is the index of the first node and j is the index of the second node. The underground environment features associated with the pipeline connecting nodes i and j. This represents the average of the underground environmental characteristics associated with all pipelines. To obtain the L2 norm, The pipe attribute characteristics of the pipe connecting nodes i and j. It is a fully connected layer. These are the pipe material data and burial depth data for the pipes connecting nodes i and j, respectively. Let be the element in the i-th row and j-th column of the attribute matching degree matrix. To take the absolute value, This represents the average value of the pipe attribute characteristics for all pipes.

[0094] S232: Based on underground environment characteristics, pipeline attribute characteristics, and burial depth data, calculate the environmental sensitivity threshold and attribute adaptation threshold. The calculation method is as follows:

[0095]

[0096]

[0097] in, As an environmentally sensitive door, For the Sigmoid function, This is an environmentally sensitive weight matrix. It is a vector composed of the associated underground environmental features of all pipelines. To obtain the minimum value, This is the environment-sensitive bias vector. For attribute adaptation doors, To adapt the weight matrix to the attributes, It is a vector formed by concatenating the pipe attribute features of all pipes. A vector composed of stitched data of the burial depth of all pipelines. Adapt bias vectors to properties;

[0098] S233: Calculate the fusion features of all pipelines based on the edge attention matrix, environment-sensitive gate, and attribute adaptation gate. The calculation method is as follows:

[0099]

[0100] in, For the fusion characteristics of all pipes, For Hadama accumulation, This is the edge attribute matrix. This is the burial depth weight matrix. This is the pipeline attribute weight matrix.

[0101] Specifically, for scenarios where the preprocessed groundwater data exhibits significant spatial heterogeneity (such as sudden changes in groundwater level in local areas or locally elevated mineral concentrations), this invention also provides a calculation method based on dilated convolution to enhance local feature extraction, replacing step S21. The calculation method is as follows:

[0102]

[0103] in, A convolutional layer with dilated convolutions. This represents the void ratio.

[0104] S3: Based on the preprocessed engineering latitude and longitude coordinate data, the preprocessed population distribution map data, and the updated pipeline map model, extract the construction impact characteristics and population risk characteristics of the pipeline respectively; then, combine the fused characteristics of all pipelines to extract the comprehensive risk characteristics and convert them into risk levels, including:

[0105] S31: Based on the preprocessed engineering latitude and longitude coordinate data and the updated pipeline diagram model, extract the construction impact characteristics of the pipeline. The calculation method is as follows:

[0106]

[0107]

[0108] in, Let the construction impact characteristics of the k-th pipeline be defined. It is a fully connected layer. To concatenate the elements within the parentheses from s=1 to s=Num, This represents the total number of construction sites. For engineering construction index, It is an exponential function. The construction attenuation coefficient is the coefficient of performance. Let k be the straight-line distance between the k-th pipeline and the s-th construction site, where k is the pipeline index. The coordinates of the midpoint of the k-th pipe in the updated pipe diagram model are: This refers to the preprocessed latitude and longitude coordinates of the s-th construction site. Let K represent the construction impact characteristics of all pipelines, where K is the total number of pipelines. To concatenate the elements within the parentheses from k=1 to k=K consecutively;

[0109] S32: Based on the preprocessed population distribution map data and the updated pipeline diagram model, extract the population risk characteristics of the pipeline. The calculation method is as follows:

[0110]

[0111]

[0112] in, The population risk characteristics of the k-th pipeline, The Gaussian kernel function is calculated by summing the weighted values ​​of the population distribution grid within a 50-meter radius of the pipe's midpoint. This is the preprocessed population distribution map data. Average population density in cities Population risk characteristics for all pipelines;

[0113] S33: Based on the fusion characteristics, construction impact characteristics, and population risk characteristics of all pipelines, a comprehensive risk characteristic is extracted through a scene attention mechanism. The calculation method is as follows:

[0114]

[0115]

[0116] in, For scene attention features, For the Softmax function, For splicing operations, To comprehensively assess risk characteristics, This is the first element in the scene attention weight vector. For the fusion characteristics of all pipes, This is the second element in the scene attention weight vector. This is the third element in the scene attention weight vector;

[0117] S34: Based on the comprehensive risk characteristics, it is converted into the risk level of all pipelines, calculated as follows:

[0118]

[0119] in, Risk levels for all pipelines are categorized as low, lower, medium, higher, and high. This is the maximum value index function.

[0120] For example, if a pipeline model for a certain area of ​​a city contains 5 core pipelines, and there are 3 construction sites after preprocessing, with an average population density of 882 people / square kilometer, the risk level of each pipeline is calculated using the method of this invention, as shown in the table below:

[0121]

[0122] Pipeline 1: The cast iron material has weak corrosion resistance and is buried at a shallow depth (1.2m). It is only 8m away from the construction site (strong construction impact) and the surrounding population density (1070 people / square kilometer) is high. The scene attention mechanism automatically increases the weight of construction impact and population risk characteristics, and the overall risk is judged as "high".

[0123] Pipeline 2: The PVC material has strong corrosion resistance, reasonable burial depth (3.5m), is far from the construction site (120m, weak construction impact), and the surrounding population density (573 people / square kilometer) is lower than the average. The scene attention mechanism focuses on the integration of features (environment + attributes), and is judged as "low" risk.

[0124] The remaining pipelines are assigned "higher", "medium" and "lower" risk levels based on factors such as material corrosion resistance, reasonable burial depth, construction distance, and population density, through dynamic weighting.

[0125] S4: Calculate the pipeline risk map model based on the risk level and display it in the GIS platform, including:

[0126] S41: Add the risk level of each pipeline from the risk levels of all pipelines to the attribute list of the corresponding pipeline in the updated pipeline diagram model to obtain the pipeline risk diagram model;

[0127] S42: Integrate and load the pipeline risk map model according to the data format compatible with the GIS platform;

[0128] S43: Employing differentiated visual identifiers to display each pipeline, thereby visualizing the spatial distribution of risk levels for urban lifeline pipelines, such as... Figure 2 As shown.

[0129] It should be noted that the sequence numbers of the above embodiments of the present invention are merely for descriptive purposes and do not represent the superiority or inferiority of the embodiments. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, apparatus, article, or method. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0130] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0131] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A GIS-based method for identifying risks in urban lifeline pipelines, characterized in that, Includes the following steps: S1: Collect raw groundwater data, raw metal mineral data, pipeline data, engineering latitude and longitude coordinates, and population distribution map data, and preprocess them to obtain preprocessed groundwater data, preprocessed mineral data, preprocessed pipeline diagram model, preprocessed engineering latitude and longitude coordinate data, and preprocessed population distribution map data. S2: Based on the preprocessed groundwater data and preprocessed mineral data, the underground environment features are extracted through a convolutional layer, and then the associated features of the underground environment are extracted and the preprocessed pipeline graph model is updated to obtain the updated pipeline graph model. Then, an improved dual-gated graph neural network is constructed to extract the fusion features of all pipelines. The improved dual-gated graph neural network first processes the pipeline material data and laying depth data in the pipeline data through a fully connected layer based on the updated pipeline graph model to obtain pipeline attribute features. At the same time, it calculates the environmental feature similarity matrix and attribute matching degree matrix, and then obtains the edge attention matrix through the Softmax function. Then, based on the underground environment association features and pipeline attribute features, it calculates the environment sensitivity gate and attribute adaptation gate respectively through the Sigmoid function. Finally, it combines the edge attention matrix, environment sensitivity gate and attribute adaptation gate to calculate the fused features of all pipelines. S3: Based on the preprocessed engineering latitude and longitude coordinate data, the preprocessed population distribution map data, and the updated pipeline diagram model, extract the construction impact characteristics and population risk characteristics of the pipeline, respectively. Then, by combining the integrated characteristics of all pipelines, comprehensive risk characteristics are extracted and converted into risk levels; S4: Calculate the pipeline risk map model based on the risk level and display it in the GIS platform.

2. The GIS-based urban lifeline pipeline risk identification method according to claim 1, characterized in that, Step S1 includes: S11: Collect raw groundwater data and raw metal mineral data from urban underground areas, with the data type being raster data; the raw groundwater data includes groundwater level and groundwater flow velocity; the raw metal mineral data includes metal mineral types and metal mineral content; outliers are removed from the raw groundwater data and raw metal mineral data using the three-standard-deviation principle, and missing values ​​are filled in using the mean method to obtain cleaned groundwater data and cleaned mineral data. S12: Collect pipeline data for urban lifeline pipelines, including pipeline material data and burial depth data. The data type of the pipeline data is graph structure data, with pipeline connection points as graph nodes, pipelines as edges, and pipeline material data and burial depth data as edge attributes. The pipeline material data includes pipeline identification codes corresponding to PVC, steel, and cast iron materials. The burial depth data includes the vertical depth information of each pipeline section buried underground. S13: Collect the latitude and longitude coordinates of various construction projects within the city. The data type is point data. Duplicate coordinates are removed to obtain cleaned latitude and longitude coordinate data. S14: Collect population distribution map data for different administrative regions of the city. The data type is raster data. The population distribution map data includes the population density of each administrative region of the city. S15: Unify the geographic coordinate benchmarks for the cleaned groundwater data, cleaned mineral data, pipeline data, cleaned engineering latitude and longitude coordinate data, and population distribution map data to obtain preprocessed groundwater data, preprocessed mineral data, preprocessed pipeline diagram model, preprocessed engineering latitude and longitude coordinate data, and preprocessed population distribution map data.

3. The GIS-based urban lifeline pipeline risk identification method according to claim 1, characterized in that, Step S2 includes: S21: Based on the preprocessed groundwater data and preprocessed mineral data, subsurface environmental features are extracted using a convolutional layer. The calculation method is as follows: in, Characteristics of the underground environment It is a convolutional layer. For splicing operations, This is the pre-processed groundwater data. This is the preprocessed mineral data; S22: Based on the start and end coordinates of each pipeline in the preprocessed pipeline graph model, determine the grid range covered by each pipeline, extract the average feature value of the underground environment features within the grid range as the underground environment association feature of the corresponding pipeline, and incorporate it into the edge attribute to obtain the updated pipeline graph model. S23: Based on the updated pipeline diagram model, construct an improved dual-gated graph neural network to extract pipeline structural features.

4. The GIS-based urban lifeline pipeline risk identification method according to claim 3, characterized in that, Step S23 includes: S231: Based on the updated pipeline graph model, calculate the pipeline attribute features, the edge attention matrix integrating the environment and attributes, and calculate the following: ； in, Let be the edge attention matrix. For the Softmax function, This is an environmental feature similarity matrix. For balance coefficient, This is the attribute matching degree matrix. Let be the element in the i-th row and j-th column of the environmental feature similarity matrix, where i is the index of the first node and j is the index of the second node. The underground environment features associated with the pipeline connecting nodes i and j. This represents the average of the underground environmental characteristics associated with all pipelines. To obtain the L2 norm, The pipe attribute characteristics of the pipe connecting nodes i and j. It is a fully connected layer. These are the pipe material data and burial depth data for the pipes connecting nodes i and j, respectively. Let be the element in the i-th row and j-th column of the attribute matching degree matrix. To take the absolute value, This represents the average value of the pipe attribute characteristics for all pipes. S232: Based on underground environment characteristics, pipeline attribute characteristics, and burial depth data, calculate the environmental sensitivity threshold and attribute adaptation threshold. The calculation method is as follows: in, As an environmentally sensitive door, For the Sigmoid function, This is an environmentally sensitive weight matrix. It is a vector composed of the associated underground environmental features of all pipelines. To obtain the minimum value, This is the environment-sensitive bias vector. For attribute adaptation doors, To adapt the weight matrix to the attributes, It is a vector formed by concatenating the pipe attribute features of all pipes. A vector composed of stitched data of the burial depth of all pipelines. Adapt the bias vector to the attribute; S233: Calculate the fusion features of all pipelines based on the edge attention matrix, environment-sensitive gate, and attribute adaptation gate. The calculation method is as follows: in, For the fusion characteristics of all pipes, For Hadama accumulation, This is the edge attribute matrix. This is the burial depth weight matrix. This is the pipeline attribute weight matrix.

5. The GIS-based urban lifeline pipeline risk identification method according to claim 3, characterized in that, Step S3 includes: Based on the preprocessed engineering latitude and longitude coordinate data and the updated pipeline diagram model, the straight-line distance between the k-th pipeline and each construction point is calculated using an exponential function and then spliced ​​together. After processing with a fully connected layer, the construction influence characteristics of the k-th pipeline are obtained. Then, the construction influence characteristics of all pipelines are spliced ​​together and processed with a fully connected layer to obtain the construction influence characteristics of all pipelines.

6. The GIS-based urban lifeline pipeline risk identification method according to claim 3, characterized in that, Step S3 includes: Based on the preprocessed population distribution map data and the updated pipeline graph model, the weighted sum of the population distribution grid within a 50-meter radius of the pipeline midpoint is calculated using the Gaussian kernel function. Combined with the city's average population density, the population risk characteristics of the k-th pipeline are obtained. Then, the population risk characteristics of all pipelines are spliced ​​together and processed through a fully connected layer to obtain the population risk characteristics of all pipelines.

7. The GIS-based urban lifeline pipeline risk identification method according to claim 3, characterized in that, Step S3 includes: S31: Based on the preprocessed engineering latitude and longitude coordinate data and the updated pipeline diagram model, extract the construction impact characteristics of the pipeline. The calculation method is as follows: in, Let the construction impact characteristics of the k-th pipeline be... It is a fully connected layer. To concatenate the elements within the parentheses from s=1 to s=Num, This represents the total number of construction sites. For engineering construction index, It is an exponential function. The construction attenuation coefficient is the coefficient of performance. Let k be the straight-line distance between the k-th pipeline and the s-th construction site, where k is the pipeline index. The coordinates of the midpoint of the k-th pipe in the updated pipe diagram model are: This refers to the preprocessed latitude and longitude coordinates of the s-th construction site. Let K represent the construction impact characteristics of all pipelines, where K is the total number of pipelines. To concatenate the elements within the parentheses from k=1 to k=K consecutively; S32: Based on the preprocessed population distribution map data and the updated pipeline diagram model, extract the population risk characteristics of the pipeline. The calculation method is as follows: in, The population risk characteristics of the k-th pipeline, The Gaussian kernel function is calculated by summing the weighted values ​​of the population distribution grid within a 50-meter radius of the pipe's midpoint. This is the preprocessed population distribution map data. Average population density in cities Population risk characteristics for all pipelines; S33: Based on the fusion characteristics, construction impact characteristics, and population risk characteristics of all pipelines, a comprehensive risk characteristic is extracted through a scene attention mechanism. The calculation method is as follows: ； in, For scene attention features, For the Softmax function, For splicing operations, To comprehensively assess risk characteristics, This is the first element in the scene attention weight vector. For the fusion characteristics of all pipes, This is the second element in the scene attention weight vector. This is the third element in the scene attention weight vector; S34: Based on the comprehensive risk characteristics, it is converted into the risk level of all pipelines, calculated as follows: in, Risk levels for all pipelines are categorized as low, lower, medium, higher, and high. This is the maximum value index function.

8. The GIS-based urban lifeline pipeline risk identification method according to claim 7, characterized in that, Step S4 includes: S41: Add the risk level of each pipeline from the risk levels of all pipelines to the attribute list of the corresponding pipeline in the updated pipeline diagram model to obtain the pipeline risk diagram model; S42: Integrate and load the pipeline risk map model according to the data format compatible with the GIS platform; S43: Differentiated visual identifiers are used to display each pipeline, enabling visualization of the spatial distribution of risk levels of urban lifeline pipelines.

Images