Urban scale drainage pipe network overflow risk diagnosis method and device, computer readable storage medium and electronic equipment

Abstract

The invention discloses an urban scale drainage pipe network overflow risk diagnosis method and device, a computer readable storage medium and electronic equipment, and belongs to the field of municipal engineering and urban flood control and drainage. The method aims at solving the problems that existing data dependence is high, precision is low, and large-scale calculation efficiency is poor. The method comprises the following steps: firstly, selecting a representative area, constructing a basic information base through fine geophysical prospecting and GIS integration, and correcting data through verification rules such as isolated point detection; calibrating a drainage simulation model based on sensor monitoring data, simulating the liquid level of the inspection well under rainfall in different recurrence periods, and grading risks; secondly, extracting pipe point features, and constructing a statistical relationship between hydraulic features and overflow risks through an SVM (Support Vector Machine) algorithm; and finally, applying the method to a global inspection well, predicting a risk level and generating a visual distribution diagram. According to the method, the advantages of a mechanism and a data model are fused, dynamic updating is supported, global monitoring is not needed, 100,000 + pipe section topology verification can be completed within 2 hours, the overflow risk detection rate exceeds 80%, and an accurate basis is provided for urban waterlogging prevention operation and maintenance.

Description

Technical Field

[0001] This invention relates to the field of municipal engineering and urban flood control and drainage technology, specifically to a method, diagnostic device, computer-readable storage medium, and electronic device for diagnosing the risk of overflow in urban-scale drainage networks. Background Technology

[0002] In recent years, the global climate change situation has become increasingly severe, with a significant increase in the frequency and intensity of extreme weather events. Frequent extreme weather events such as torrential rains and typhoons have ravaged cities, exacerbating urban flooding problems, whose severity is growing daily. Urban flooding not only paralyzes urban traffic, disrupting residents' daily travel and normal lives, but also causes serious damage to urban infrastructure, such as roads and bridges. More seriously, urban flooding poses a significant threat to the lives and property of urban residents, and the casualties and economic losses caused by urban flooding each year are heartbreaking.

[0003] In this severe situation, effectively addressing urban flooding risks and accurately identifying high-risk areas or locations has become a crucial basis for the layout and operation of urban flood control and drainage projects. Only by accurately identifying these high-risk areas and locations can we strengthen the construction and renovation of flood control and drainage facilities in a targeted manner during urban planning and construction, rationally plan key facilities such as drainage networks and pumping stations, and improve the city's flood control and drainage capacity. Simultaneously, in daily flood control operation and maintenance, this risk information allows for the development of emergency plans in advance, the rational allocation of resources, and the timely implementation of effective preventative measures, minimizing the impact of flooding disasters on the city and its residents.

[0004] However, assessing the risk of overflows in urban drainage networks is an extremely complex process, involving numerous factors and presenting significant technical challenges. Currently, while some technologies have attempted to address the problem of diagnosing overflow risks in urban drainage networks, these existing technologies all have considerable limitations.

[0005] For example, CN117909787A discloses a method for intelligent diagnosis of transportation bottlenecks and manhole overflows in drainage pipe networks. This method, based on open channel flow formulas and pressurized flow formulas, combined with pipe manhole parameters and measured flow and level data, deduce the transportation bottleneck and overflow risk of pipe sections under different operating conditions. However, in practical applications, this method requires the deployment of numerous monitoring devices when applied to large-scale pipe networks. This not only increases the initial investment cost of the project but also places enormous pressure on subsequent equipment maintenance and management. Moreover, due to the extremely complex topological relationships of urban drainage pipe networks, it is difficult to accurately determine overflow risk using only open channel flow formulas or pressurized flow formulas without considering the upstream and downstream level relationships. Because the water flow in the pipe network is interconnected, changes in upstream and downstream levels affect each other; ignoring this relationship can lead to misjudgments of the water flow state, thus affecting the overflow risk assessment results.

[0006] For example, CN106382471B discloses a diagnostic and assessment method for urban drainage pipe networks considering key nodes. This method proposes constructing a hydraulic model of the urban drainage pipe network based on water level and flow monitoring data, as well as relevant drainage pipe network data, to simulate and assess the operational status of the urban pipe network. However, this method requires high data accuracy; even a small data error can lead to significant deviations in the hydraulic model simulation. In practice, data collection for urban drainage pipe networks is often affected by various factors, such as the accuracy limitations of monitoring equipment, errors in data transmission, and human error in recording, making it difficult to guarantee the complete accuracy of the collected data. Once data errors occur, the accuracy of liquid level changes for individual manholes becomes extremely low, making it unreliable as a basis for judging the operational status of the pipe network, thus affecting the accurate assessment of overflow risks.

[0007] Furthermore, another example is the anomaly analysis method and device for stormwater and sewage drainage networks based on edge computing gateways disclosed in CN116205087B. This method collects real-time liquid level and flow velocity data from stormwater and sewage drainage network diversion wells and outlets, performs preprocessing, and then performs anomaly analysis. However, this method also relies on monitoring equipment. When conducting risk diagnosis on a large urban scale, a large number of monitoring devices need to be deployed to obtain comprehensive data, which undoubtedly leads to excessively high costs. For many cities, such high costs are unbearable, therefore this method is difficult to widely apply on a large urban scale.

[0008] In summary, existing technologies for diagnosing overflow risks in urban drainage networks generally suffer from problems such as high data requirements, insufficient accuracy, large computational load, or excessive cost. Especially in large-scale networks, simultaneously meeting the needs of low data requirements, high accuracy, and low computational load has become a pressing technical challenge. Existing methods either require deploying numerous monitoring devices to achieve a certain level of accuracy, leading to high costs; or have excessively high data accuracy requirements that are difficult to obtain in practical applications, thus compromising the accuracy of the assessment results; or require processing large amounts of data during the calculation process, resulting in enormous computational load and low efficiency. These problems severely restrict the effective implementation of urban drainage network overflow risk diagnosis and fail to meet the actual needs of urban flood control and drainage engineering layout and flood prevention operation and maintenance. Therefore, developing a new method for diagnosing overflow risks in urban-scale drainage networks has significant practical significance and application value. Summary of the Invention

[0009] The technical problem to be solved by this invention is to provide a method, diagnostic device, computer-readable storage medium, and electronic device for diagnosing the overflow risk of urban drainage pipe networks. This invention addresses the problems of high data requirements, insufficient accuracy, large computational load, or excessive cost in existing technologies for diagnosing the overflow risk of urban drainage pipe networks. This invention aims to overcome the shortcomings of existing technologies and provide an overflow risk diagnosis method that integrates the advantages of mechanistic models and data models, taking into account low cost, high accuracy, and efficient computation, so as to achieve accurate classification and visualization of overflow risk of inspection wells across the entire area.

[0010] To achieve the above technical objectives, the present invention adopts the following technical solution: A method, diagnostic device, computer-readable storage medium, and electronic device for diagnosing the risk of overflow in urban-scale drainage pipe networks are provided, as detailed below: (I) A method for diagnosing the risk of overflow in urban-scale drainage pipe networks The method includes the following steps: Step 1: Establish a basic information database for the urban drainage system Representative areas within the city were selected, and basic information was obtained using refined geophysical exploration techniques for the pipeline network. This basic information included the spatial topology of the pipeline network (coordinates and connections of pipes and manholes), pipeline and manhole attribute information (pipeline top and bottom elevations, slope, pipe diameter, material, manhole bottom elevation, and surface elevation), surface elevation, land use type, soil permeability coefficient, and catchment zone boundaries. The above data was integrated using GIS technology, and error diagnosis and correction were performed using validation rules, including isolated point detection, pipe diameter conflict detection, and flow direction correction. Batch processing was implemented using PyQGIS scripts. Isolated point detection: Traverse all pipe sections in the area If a certain section of pipe The starting point and the end point If there are no overlapping pipe points and the pipes are not located at the beginning or end of the drainage system, an alarm will be triggered. Points that are spatially adjacent and have similar pipe diameters will be selected as suggested connection points, which will be manually verified and confirmed. Pipe diameter conflict detection: when pipe section Pipe diameter Larger than downstream pipe section Pipe diameter An alarm is triggered, prompting a recalculation of the inner diameter of other upstream and downstream pipes. ; Flow correction: when and When the spaces overlap, determine the upstream pipe section. and If the flow direction is reversed, an alarm is triggered and the start and end points of the pipe segment are redefined.

[0011] Finally, QGIS software was used to integrate and correct the collected data (pipeline topology, elevation model, soil type, land use), and the QGIS plugin was used to extract the attributes of pipe segments (pipe diameter, slope, material), nodes (manhole elevation, coordinates), and sub-catchment areas (area, impermeability, slope), and convert them into drainage simulation input files in .inp format to establish a basic information database for the urban drainage system.

[0012] Step 2: Calibration and optimization of the drainage simulation model Level gauges and flow meters are installed at key drainage nodes, including the junctions of main and secondary pipelines, upstream and downstream of inverted siphons, and high-level pipe sections under sunny conditions. Level and flow data with a time resolution of no more than 5 minutes are collected. The drainage simulation model is calibrated and optimized using rainy day level and flow processes as the basis for calibration, combined with historical rainfall data and long-term level data. This involves adjusting the infiltration rate of sub-catchment areas and the Manning coefficient of the main pipeline to modify the simulated peak flow rate, correcting the level change process using the local loss coefficient of branch pipes, evaluating the adjustment effect using efficiency coefficients and RMSE indicators, and verifying the effect through rainfall events not involved in the calibration.

[0013] The drainage simulation model uses the Horton infiltration model for runoff generation in the sub-catchment area and the nonlinear reservoir method for runoff generation. Horton's infiltration formula is: (1); In the formula, For time Infiltration rate (mm / h) at that time. The soil constant infiltration rate (mm / h) The initial soil infiltration rate (mm / h) is given. The infiltration rate decay coefficient over time ( ), (Rainfall duration in hours); initial parameter values ​​are: , , .

[0014] The confluence model is solved by simultaneously solving the continuity equation and the Manning equation. The calculation formula is as follows: (2); (3); In the formula, The water volume of the reservoir within the catchment area ( ), The area of ​​the catchment area ( ), The rainfall rate is expressed in mm / h. The depth is in meters (m). Runoff ( ), The characteristic width (m) of the sub-catchment area. The Manning coefficient for the sub-catchment area is 0.013~0.017 for concrete pipes and 0.009~0.011 for HDPE pipes. The average slope (%) of the sub-catchment area. The surface water depth (m) in the catchment area. The maximum water depth (m) in the sub-catchment area.

[0015] Step 3: Risk Level Classification of Overflow A rainfall scenario with a 1-50 year return period was synthesized by combining local rainfall intensity formulas with Chicago rainfall patterns, prioritizing a 24-hour 20-year return period rainfall scenario as the core simulation scenario. The local rainfall intensity formula is as follows: (4); In the formula, Rainfall intensity ( ), The return period is in years.

[0016] Using the high-precision drainage network mechanism model obtained in step 2, the rise in manhole liquid level under rainfall with different return periods was simulated, and the overflow risk level was classified according to the relationship between liquid level and surface elevation and pipe top elevation: High risk: The highest liquid level at the pipe point during the simulation. ( (The corresponding surface elevation of the pipe point). Medium risk: Liquid level at pipe points ( (The elevation of the top of the pipe corresponding to the pipe point). Low risk: Pipeline level .

[0017] Further analysis of the difference between the surface elevation and the pipe top elevation in medium-risk liquid levels can be conducted by combining experience or quantile ratios to refine the risk categories.

[0018] Step 4: Construct the statistical relationship between hydraulic characteristics and overflow risk Pipeline feature information was extracted using QGIS, including catchment area, upstream pipeline slope, upstream pipeline bottom elevation, upstream pipeline diameter, upstream pipeline roughness, downstream pipeline bottom elevation, downstream pipeline slope, downstream pipeline diameter, downstream pipeline roughness, well bottom elevation, surface elevation, and rainfall return period. A statistical relationship between hydraulic features and overflow risk was constructed based on the SVM algorithm. The independent variable of this statistical relationship was the aforementioned pipeline feature information, and the dependent variable was a numerical overflow risk level (high risk = 2, medium risk = 1, low risk = 0).

[0019] The training formula for the SVM algorithm is: (5); In the formula, For the first Feature information vector of each tube point; Risk level; This is the weight vector; For bias; This is a soft-margin slack variable, representing the degree to which classification errors are allowed; For matrix The transpose of the matrix; This is the error penalty coefficient; For input vectors An image that is implicitly mapped to a high-dimensional feature space; It indicates that it is subject to restrictions.

[0020] Step 5: Global Overflow Risk Prediction and Visualization Obtain the large-scale urban manhole attribute database generated from the urban pipeline network survey. Connect the statistical relationships trained in step 4 to this database. Use the SVM algorithm to simulate and predict the overflow risk of the entire pipeline segment. The prediction formula for the SVM algorithm is: (6); In the formula, The predicted value of spillover risk for the target area. This is the feature information vector of all pipe points in the target pipe segment. Let be the support vector coefficients obtained during training; To fit the kernel function; This is a symbolic function that outputs the risk category.

[0021] Finally, using QGIS, a spatial distribution map of urban overflow risk levels was drawn based on the number, coordinates, and overflow risk level of each manhole.

[0022] (II) A device for diagnosing the risk of overflow in urban-scale drainage pipe networks This is an apparatus for implementing the aforementioned method for diagnosing the overflow risk of urban-scale drainage networks, the apparatus comprising: Basic Information Acquisition Module: This module is used to acquire basic information such as surface, pipeline, and manhole data of representative areas through detailed geophysical exploration of the pipeline network. After integration using GIS technology and correction using verification rules such as isolated point detection, pipe diameter conflict detection, and flow direction correction, a basic information database of the urban drainage system is established, and a drainage simulation input file in .inp format is generated. Model calibration module: Used to monitor the liquid level and flow parameters of key drainage nodes through sensors, and to calibrate and optimize the drainage simulation model by combining historical data. The drainage simulation model includes the Horton infiltration formula and the runoff calculation formula mentioned above. Risk classification module: Used to simulate rainfall data with different return periods using the above local rainfall intensity formula, simulate the changes in manhole liquid level through a high-precision drainage network mechanism model, and classify the overflow risk level according to preset standards; Statistical Relationship Construction Module: Used to extract pipe point feature information and construct statistical relationships between hydraulic features and overflow risk based on the above SVM training formula; Global prediction module: Used to predict the overflow risk of large-scale manholes in the city using the above SVM prediction formula, and generate a spatial distribution map of the city's overflow risk level.

[0023] (iii) A computer-readable storage medium The computer-readable storage medium stores a computer program that, when executed by a processor, implements all the steps of the aforementioned method for diagnosing the overflow risk of urban-scale drainage networks.

[0024] (iv) An electronic device The electronic device includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the computer program, it implements all the steps of the above-mentioned method for diagnosing the overflow risk of urban-scale drainage pipe networks.

[0025] The present invention provides a method, diagnostic device, computer-readable storage medium, and electronic device for diagnosing the overflow risk of urban-scale drainage pipe networks, which has the following beneficial effects: 1. This invention effectively solves the problem of efficient and accurate diagnosis of overflow risks in urban-scale drainage pipe networks, breaks through the limitations of existing technologies, achieves accurate identification and quantitative assessment of overflow risks, overcomes the problems of stringent data requirements, difficulty in guaranteeing accuracy, large amount of calculation, and high cost when facing large-scale complex pipe networks, ensures the scientific layout and efficient operation and maintenance of urban flood control and drainage projects, and safeguards urban safety.

[0026] 2. This invention brings significant technical and economic benefits, providing an economical and efficient solution for urban drainage system management.

[0027] 3. This invention integrates the advantages of mechanistic models and data models, significantly reducing equipment investment and cost expenditure for monitoring and evaluating liquid level and flow in the entire pipeline network, while significantly improving diagnostic accuracy and computational efficiency, reducing resource consumption, and enhancing diagnostic effectiveness.

[0028] 4. The actual test cases of this invention show that it can efficiently complete the topology verification of a system with more than 100,000 pipe segments (time < 2 hours), and the overflow risk detection rate is as high as 80% or more, which fully verifies the effectiveness and practicality of the solution and provides a reliable basis for practical application.

[0029] 5. This invention has the characteristics of long-term practicality. It uses an artificial intelligence data model to quantify the risk of overflow and supports a dynamic update mechanism to ensure that the model continuously adapts to the dynamic changes of the urban drainage network and maintains long-term effectiveness and accuracy.

[0030] 6. This invention combines the advantages of mechanism and data model. It decouples the core cause of overflow through a high-precision mechanism model of small area, reduces the deployment cost of full-area monitoring equipment, reduces the analysis bias caused by pipeline data exploration errors, and improves the accuracy of analysis.

[0031] 7. This invention constructs statistical relationships based on the SVM algorithm, breaking through the efficiency bottleneck of traditional pure mechanistic models in large-scale calculations, avoiding the resource consumption of global fine-grained modeling, and improving computational efficiency.

[0032] 8. This invention provides accurate and efficient risk classification. Based on limited rainfall data and mechanistic model simulation, it achieves overflow risk classification for a large number of inspection wells by analyzing the differences in the rise of liquid levels. This solves the problems of limited urban rainstorm frequency and difficulty in observing and statistically analyzing actual overflow well points, providing reliable support for prioritizing flood control and maintenance during the flood season. 9. This invention addresses the problems of existing technologies in large-scale urban drainage networks, such as high data requirements, insufficient accuracy, large computational load, or excessive cost, thereby improving the management technology level of urban drainage systems.

[0033] 10. This invention effectively overcomes the difficulty of efficiently and accurately diagnosing the risk points of overflow in urban-scale drainage networks, improves the layout of urban flood control and drainage projects and the effectiveness of flood control operation and maintenance, and enhances the city's ability to cope with flood disasters.

[0034] 11. The actual test cases of this invention show that it can complete the topology verification of large-scale pipe segment systems and has a high overflow risk detection rate, providing an efficient and reliable method for risk diagnosis of large-scale pipe networks in urban drainage systems.

[0035] 12. This invention reduces the amount of equipment and cost required for monitoring and evaluating the liquid level and flow rate of the entire pipe network, improves diagnostic accuracy and calculation efficiency, and makes urban drainage system management more economical and efficient.

[0036] 13. This invention has long-term practicality and can adapt to the dynamic changes in urban drainage pipe networks caused by urban development and renovation, and continuously provide accurate and effective support for urban drainage system management.

[0037] 14. By integrating the advantages of the model and adopting artificial intelligence algorithms, this invention improves the accuracy and comprehensiveness of diagnosis while reducing equipment costs and computing resource consumption, providing a better solution for urban drainage system management.

[0038] 15. This invention supports a dynamic update mechanism. When pipeline network is modified or new monitoring data is added, the prediction model can be quickly iterated through incremental learning, avoiding the huge workload of rebuilding the analysis mechanism model, saving time and resources, and improving operation and maintenance efficiency.

[0039] 16. This invention provides a comprehensive, efficient, accurate and long-term practical diagnostic method for urban drainage system management, which helps to improve the overall flood control and drainage capacity of the city and ensure the normal operation of the city and the safety of residents' lives. Attached Figure Description

[0040] The present invention will be further described below with reference to the accompanying drawings: Figure 1 This is a schematic diagram illustrating the establishment of high-precision drainage mechanism models for cities in embodiments 5 to 8 of the present invention; Figure 2 This is a schematic diagram of the calibration results for typical sites at some locations in Embodiments 5 to 8 of the present invention; Figure 3 Examples 5 to 8 of this invention demonstrate the simulated overflow and liquid level rise in inspection wells. Detailed Implementation

[0041] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments: Example 1 This embodiment provides a method for diagnosing the risk of overflow in urban-scale drainage networks. Taking a 6km² independent urban drainage area in City L as a representative region, the specific implementation process of the method for diagnosing the risk of overflow in urban-scale drainage networks is described in detail: Step 1: Establish a basic information database for the urban drainage system Pipeline geophysical exploration technology was used to obtain the spatial topology of the pipeline network in this area (coordinates and connections of pipelines and manholes), pipeline and manhole attribute information (top and bottom elevations of pipelines, slope, pipe diameter, material, bottom elevation of manholes, and surface elevation), surface elevation, land use type, soil permeability coefficient, and catchment zone boundaries. The above data was integrated using GIS (Geographic Information System) technology, and batch processing and validation rules were implemented using PyQGIS scripts. Isolated point detection: Traverse all pipe sections in the area If the pipe section The starting point and the end point If there are no overlapping pipe points and the pipes are not located at the beginning or end of the drainage system, an alarm is triggered and nearby connection points with similar pipe diameters are recommended, subject to manual verification and confirmation. Pipe diameter conflict detection: when pipe section Pipe diameter Larger than downstream pipe section Pipe diameter An alarm is triggered, prompting a recalculation of the inner diameter of other upstream and downstream pipes. ; Flow correction: when and When the spaces overlap, determine the upstream pipe section. and If the flow direction is reversed, an alarm is triggered and the start and end points of the pipe segment are redefined.

[0042] By extracting pipe segment (diameter, slope, material), node (manhole elevation, coordinates), and sub-catchment (area, impermeability, slope) attributes using QGIS software and plugins, the integrated and corrected data is converted into a .inp format drainage simulation input file to establish a basic information database for the urban drainage system.

[0043] Step 2: Calibration and optimization of the drainage simulation model Level gauges and flow meters were installed at key points in the area, such as the intersection of main and secondary pipelines, upstream and downstream of inverted siphons, and high-level pipe sections under sunny conditions, to collect level and flow data with a time resolution of 5 minutes. Combining historical rainfall data and long-term level data, and using the rainy-day level and flow process as the calibration basis, the infiltration rate of the sub-catchment area and the Manning coefficient of the main pipeline were adjusted (0.015 for concrete pipes and 0.010 for HDPE pipes). The level change process was corrected using the local loss coefficient of the branch pipes. The adjustment effect was evaluated using the efficiency coefficient and RMSE (Root Mean Square Error) indicators, and the accuracy of the model was verified by selecting rainfall events not involved in the calibration.

[0044] The runoff generation in the sub-catchment area was modeled using the Horton infiltration model, with initial parameters set as follows: , , The formula is: (1); The confluence model is solved by simultaneously solving the continuity equation and the Manning equation, as shown in the following formula: (2); (3); Step 3: Risk Level Classification of Overflow By combining the rainfall intensity formula for City L with the Chicago rainfall pattern, a rainfall scenario with a return period of 1-50 years was synthesized. The core simulation focused on a 24-hour rainfall event with a 20-year return period. The rainfall intensity formula is as follows: (4); Using a high-precision drainage network mechanism model, the changes in manhole liquid levels under rainfall with different return periods were simulated, and the risk levels were classified according to the following criteria: High risk: ( The highest liquid level at the pipe point. (Surface elevation); Medium risk: ( (Elevation of the top of the pipe); Low risk: .

[0045] By combining the quantile ratio, the risk level is further refined to form a complete risk classification result.

[0046] Step 4: Construct statistical relationships Pipeline feature information was extracted using QGIS, including catchment area, upstream and downstream pipeline slope, upstream and downstream pipeline bottom elevation, upstream and downstream pipeline diameter, upstream and downstream pipeline roughness, well bottom elevation, surface elevation, and rainfall return period. Risk levels were quantified (high risk = 2, medium risk = 1, low risk = 0). Using feature information as independent variables and quantified risk level as dependent variables, an SVM algorithm was used to train statistical relationships. The training formula is as follows: (5); Step 5: Global Overflow Risk Prediction and Visualization Obtain the comprehensive manhole attribute database from the city's urban pipeline network survey in City L, and interface the trained SVM model with this database using the prediction formula: (6); The overflow risk of all inspection wells in the region under different return periods of rainfall was calculated. Using QGIS, a spatial distribution map of the overflow risk level under a 24-hour rainfall event with a 20-year return period was drawn.

[0047] Actual test results show that this embodiment can complete the topology verification of a 100,000+ pipe section system in City L within 2 hours, with an overflow risk detection rate of 82%.

[0048] Example 2 In another preferred embodiment, based on Embodiment 1 above, this embodiment provides a city-scale drainage network overflow risk diagnosis device, used in the city-scale drainage network overflow risk diagnosis method described in Embodiment 1, applied to the overflow risk diagnosis scenario in City L. The device includes: Basic information acquisition module: Configure pipeline fine geophysical exploration equipment and GIS processing unit to acquire basic data such as pipeline topology, manhole attributes, and surface elevation of a representative 6km² area in L City. Perform isolated point detection, pipe diameter conflict detection, and flow direction correction through the built-in PyQGIS verification script, generate drainage simulation input file in .inp format, and establish a basic information database; Model calibration module: Connects to sensors such as level gauges and flow meters to collect level and flow data at key nodes with a 5-minute resolution. It has built-in Horton infiltration model and nonlinear reservoir confluence model, and adjusts model parameters (infiltration rate, Manning coefficient, etc.) by combining historical data. The model is calibrated and verified through efficiency coefficient and RMSE index. Risk classification module: Stores the rainfall intensity formula for L city and the Chicago rainfall pattern generation algorithm, synthesizes rainfall data with a return period of 1-50 years, calls up the calibrated mechanism model to simulate liquid level changes, and outputs the risk level (high, medium, low) and detailed classification results of the pipeline according to preset rules; Statistical Relationship Construction Module: Integrates QGIS feature extraction tool and SVM (Support Vector Machine) algorithm training unit to extract 12 feature information of pipeline points, quantify risk level and train statistical relationship model between hydraulic features and overflow risk; Global Prediction Module: Connects to the L City's comprehensive manhole attribute database, uses a trained SVM model to predict the global overflow risk, and links with QGIS to draw a spatial distribution map of risk levels, supporting visualization and export.

[0049] Example 3 In another preferred embodiment, based on Embodiment 1 above, this embodiment provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements all the steps of the urban-scale drainage network overflow risk diagnosis method described in Embodiment 1. The storage medium is a USB flash drive storing the computer program, which, when executed by a processor, implements the following steps: Receive geophysical exploration data of pipeline networks from a representative 6km² area in City L, integrate the data using GIS and correct the data using PyQGIS verification rules, and generate an input file in .inp format. By reading key node liquid level and flow data collected by sensors and combining them with historical rainfall data, the drainage simulation model, which includes the Horton infiltration formula and the runoff calculation formula, is calibrated and optimized. Using the rainfall intensity formula for City L, rainfall scenarios with different return periods are generated to simulate changes in the liquid level of inspection wells and classify the risk level of overflow. Extract pipe point feature information and train the statistical relationship between hydraulic features and overflow risk using the SVM algorithm; Load the attribute data of all inspection wells in L city, use the trained SVM model to predict the risk of overflow in the whole area, and generate and output the spatial distribution map of risk level through QGIS.

[0050] This USB drive can be connected to a municipal engineering monitoring terminal to quickly deploy and execute the aforementioned urban-scale drainage network overflow risk diagnosis method.

[0051] Example 4 In another preferred embodiment, based on Embodiment 1, this embodiment provides an electronic device for diagnosing the overflow risk of urban-scale drainage pipe networks in City L. The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements all the steps of the urban-scale drainage pipe network overflow risk diagnosis method described in Embodiment 1. Specifically, the device includes: Storage: A 1TB solid-state drive is used to store L City pipeline survey data, sensor monitoring data, rainfall intensity formula parameters, SVM model training data and computer programs; Processor: Intel Core i7-13700K, 3.4GHz, used to execute computer programs in memory, specifically implementing: data integration and verification, drainage simulation model calibration, rainfall scenario generation, risk level classification, SVM model training and global prediction; Input interfaces: Supports USB 3.2 and Ethernet interfaces to connect to sensors, geophysical equipment, and pipeline database servers to receive raw data; Output interfaces: Supports HDMI and Ethernet interfaces, connecting to a monitor to display the spatial distribution map of risk levels, or transmitting diagnostic results to the municipal operation and maintenance platform; Power module: Uses 220V AC input to provide stable power supply.

[0052] When the electronic device is running, the processor executes a computer program that can complete the topology verification of more than 100,000 pipe sections in L City within 2 hours, with an overflow risk detection rate of over 80%, providing accurate data support for urban flood control and maintenance.

[0053] Example 5 In another preferred embodiment, based on Embodiment 1, this embodiment provides a method for diagnosing the risk of overflow in urban-scale drainage networks. Taking an 8km² independent drainage area in the core urban area of ​​City M as the research object, it combines... Figures 1 to 3 Using the data in Table 1, the document details the implementation process of the diagnostic methods, highlighting the results of comprehensive risk visualization. Step 1: Establish a basic information database for the urban drainage system High-precision pipeline geophysical exploration technology was employed to obtain basic information for this area, including typical well points such as 41WS5331, 47WS0398, and 48WS0078 listed in Table 1. This information encompasses core parameters (corresponding to fields in Table 1) such as catchment area, upstream and downstream pipeline slope, pipe bottom elevation, pipe diameter, roughness coefficient, well bottom elevation, and surface elevation. Simultaneously, spatial topology of the pipeline network, land use type, soil permeability coefficient, and catchment zone boundaries were collected. After integrating the data using GIS technology, validation rules were executed using a PyQGIS script. Table 1 Summary of Location Information

[0054] Isolated spot detection: For the pipe segment associated with well point 48WS0078 in Table 1, detect its starting point. and the end point After confirming that there are no overlapping pipe points and that they are not the start or end point of the system, a connection point with a matching pipe diameter is recommended, which is then manually reviewed and corrected. Pipe diameter conflict detection: It was found that the pipe diameter of the upstream pipe section (600mm) of well point 47WS0398 was larger than that of the downstream pipe section (400mm). After triggering the alarm, the reasonable pipe diameter of the downstream pipe section was recalculated to be 500mm. Flow direction correction: Corrected 3 flow direction errors caused by overlapping pipe segment endpoints.

[0055] The pipe segment, node, and sub-catchment attributes of each well point in Table 1 were extracted using the QGIS plugin, converted into .inp format input files, and a basic information database containing more than 1,200 inspection wells (including the points listed in Table 1) was established.

[0056] Step 2: Calibration and optimization of the drainage simulation model Level gauges and flow meters were installed at 20 key nodes in the area (including the nodes surrounding the three typical well points in Table 1) to collect level and flow data at a 5-minute resolution. Based on measured data from three typical rainstorms, the drainage simulation model was calibrated. The flow generation module uses the Horton infiltration model, with initial parameters set to... , , The formula is: (1); The flow module solves the problem by combining the continuity equation and the Manning equation. It then adjusts the Manning coefficient (0.014 for concrete pipes and 0.010 for HDPE pipes) by taking into account parameters such as the catchment area and slope of each well point in Table 1.

[0057] Calibration results as follows Figure 1 , Figure 2 As shown: the simulated values ​​at points WS1, WS5, WS6, and WS9 have a high degree of fit with the actual monitored values, and the liquid level change trend is consistent. The efficiency coefficient reaches 0.85, and the RMSE is 0.03m, verifying the model accuracy (corresponding to...). Figure 2 "Results of typical field tests at some locations"

[0058] Step 3: Risk Level Classification of Overflow The local rainfall intensity formula for City M was combined with the Chicago rainfall pattern to synthesize a 20-year return period 24-hour rainfall scenario. A calibrated mechanistic model was then used to simulate the liquid level rise process at each well point. For example... Figure 3 As shown in the "Simulated Overflow and Liquid Level Rise of Inspection Wells", the liquid level at well point 48WS0078 reached its peak 120 minutes after rainfall and remained above the surface elevation. \ The well was classified as high-risk (corresponding to risk level "2" in Table 1); the fluid level at well 47WS0398 was higher than the top elevation of the pipe. But below the surface elevation ( Well 41WS5331, whose liquid level consistently remained below the pipe top elevation, was classified as medium risk (corresponding to risk level "1" in Table 1); well 41WS5331, with its liquid level consistently below the pipe top elevation, was classified as low risk (corresponding to risk level "0" in Table 1). Simultaneously, the model outputs the liquid level rise curves for each well point (e.g., ...). Figure 3 The liquid level change trend at different risk points (indicating the timing and intensity of spills) visually presents the timing and intensity of spills.

[0059] Step 4: Construct statistical relationships Using the characteristic information of over 1200 well points in Table 1 as independent variables (12 parameters including catchment area, upstream and downstream pipeline gradient, pipe bottom elevation, pipe diameter, roughness coefficient, well bottom elevation, and surface elevation), and the quantified risk level (high=2, medium=1, low=0) as the dependent variable, the SVM algorithm was used to train the statistical relationship. The training formula is as follows: (5); Among them, the feature data of three typical well points in Table 1 are used as core training samples to improve the model's adaptability to points with different risk levels.

[0060] Step 5: Global Overflow Risk Prediction and Visualization Obtain survey data (including similar feature fields from Table 1) for 32,000 manholes across City M. Connect the trained SVM model to the city-wide database and use the prediction formula: (6); The risk of overflow from manholes across the entire area was calculated. The final result was a spatial distribution map as shown in "Overflow Risk Prediction Based on Measured Data Simulation". The map clearly delineates high, medium and low overflow risk areas (corresponding to the three risk levels in Table 1). The high-risk area is concentrated around well point 48WS0078 in the old city area, the medium-risk area is distributed along the secondary roads (including well point 47WS0398), and the low-risk area is mainly in the newly built urban area (including well point 41WS5331). This data closely matches the actual aging of the pipeline network and the terrain conditions.

[0061] Actual test results show that this embodiment can complete the topology verification of over 150,000 pipe segments across the entire area within 1.8 hours, with an overflow risk detection rate of 85%. Figure 3 The simulated liquid level curve deviates from the actual monitoring data after rainfall by less than 4%, and the risk distribution is highly consistent with the areas of annual flooding that have occurred in M ​​City in the past.

[0062] Example 6 In another preferred embodiment, based on embodiments 2 and 5 above, this embodiment provides a city-scale drainage network overflow risk diagnosis device, used in the city-scale drainage network overflow risk diagnosis method described in embodiment 5, adapted to the diagnosis scenario of the aforementioned city M, and combined with... Figures 1 to 3 The data in Table 1 specifically includes: Basic Information Acquisition Module: Configured with a geophysical data receiving unit and a GIS processing module, this module specifically reads data from the catchment area, pipeline parameters, elevation, and other fields of each well point in Table 1. It uses a built-in verification script to correct isolated points and pipe diameter conflicts at well points such as 48WS0078, generating an input file in .inp format containing all fields from Table 1 and establishing associations. Figure 1 A database of basic data for mechanistic models; Model calibration module: Connects to the sensor data interface, receives measured data on liquid level and flow rate at key nodes, and incorporates the Horton infiltration model and confluence model for reference. Figure 1 , Figure 2 The model automatically adjusts the model parameters based on the fitting relationship between the actual monitored value and the simulated value, and outputs the efficiency coefficient and RMSE index to ensure that the simulation accuracy of points such as WS1 and WS6 meets the standards. Risk grading module: Integrates a rainfall scenario generation unit, synthesizes rainfall data with different return periods, and calls upon the output of the calibrated model. Figure 3 The well level rise curve shown, combined with the difference between "surface elevation and pipe top elevation" in Table 1, is classified into high, medium and low risk standards, supporting the detailed classification of medium risk. Statistical Relationship Construction Module: Built-in Table 1 field parsing tool, automatically extracts 12 management point feature information, quantifies the risk level into 0, 1, 2 (corresponding to the risk level in Table 1), trains the mapping relationship between features and risks through SVM algorithm, and generates a model file adapted for global prediction; The overall prediction module connects to the M City's comprehensive manhole attribute database, loads the trained SVM model, calculates the overall risk level, and, in conjunction with QGIS, generates spatial distribution maps of "high-medium-low overflow risk zones" and "overflow risk of M City's main drainage network under a 20-year return period 24h rainfall". It supports risk zone boundary labeling and pop-up display of core point information (including manhole number, risk level, peak liquid level, and estimated overflow duration). It also supports querying the risk level and liquid level simulation curve of a single point by well number in Table 1. Figure 3 It can simultaneously export vector format files and Table 1 extended version (including the risk level of all well points).

[0063] Example 7 In another preferred embodiment, based on embodiments 3 and 5 above, this embodiment provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements all the steps of the urban-scale drainage network overflow risk diagnosis method described in embodiment 5. The storage medium is a portable solid-state drive (2TB) storing the computer program and associated map data. When the computer program is executed by a processor, it implements the following steps: Step 1: Read the stored data from Table 1 "Symptoms of Partial Point Information" and the geophysical data of the 8km² area in City M. Use the GIS integration tool and PyQGIS verification script to correct data problems such as isolated points, pipe diameter conflicts, and flow direction errors, and generate a drainage simulation input file in .inp format. Step 2: Retrieve stored data Figure 1 , Figure 2 Corresponding measured liquid level / flow rate data, calibrating Horton infiltration model and confluence model parameters, outputting efficiency coefficients, RMSE index, and Figure 2The "actual monitored value - simulated value" fitting curve shown verifies the model accuracy; Step 3: Load the local rainfall intensity formula parameters, generate a 20-year return period 24-hour rainfall scenario, simulate the liquid level rise process at each well point, and output the results. Figure 3 Based on the liquid level change curves shown, and in conjunction with the risk level standards in Table 1, classify the risks as high, medium, and low. Step 4: Extract the 12 feature fields from Table 1, train the SVM statistical relationship model, store the model parameters and training logs, and ensure that the model achieves 100% accuracy in predicting the risk of typical locations in Table 1; Step 5: Read the attribute data of all inspection wells in M ​​City, use the trained model to predict the risk level, call QGIS to generate the overflow risk prediction and the spatial distribution map of the overflow risk level of the main drainage pipe network in M ​​City under the 20-year return period 24h rainfall, and simultaneously store the map in both PNG (Portable Network Graphics) and SVG (Scalable Vector Graphics) formats, as well as the risk level data table of the entire area (including all field extensions in Table 1).

[0064] This solid-state drive can be directly connected to municipal maintenance terminals or mobile workstations without the need for additional data import. It can start and execute the above diagnostic process within 10 minutes, making it suitable for both on-site inspection and back-end analysis scenarios.

[0065] Example 8 In another preferred embodiment, based on embodiments 4 and 5, this embodiment provides an electronic device for diagnosing the overflow risk of urban-scale drainage pipe networks in City M. The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements all the steps of the urban-scale drainage pipe network overflow risk diagnosis method described in embodiment 5. The device is hardware and software compatible. Figures 1 to 3 And the data processing requirements in Table 1: Storage: A 2TB NVMe solid-state drive is used to store the M City-wide pipeline network survey data and complete location information from Table 1. Figure 1 and Figure 2 Actual monitoring data, Figure 3 The system includes liquid level simulation parameters, intermediate risk prediction data, and visualization template files, ensuring data read latency is less than 1ms. Processor: AMD Ryzen 9 7900X, 4.7GHz, supports multi-threaded processing, quickly completes topology verification of 150,000+ pipe segments (time < 2 hours), and simultaneously performs SVM model training and global prediction, accurately matching the mapping relationship between feature data and risk levels in Table 1; Input interfaces: Supports USB 4.0 and RJ45 Gigabit Ethernet ports, and can directly import Excel format data from Table 1. Figures 1 to 3 The original monitoring files are compatible with real-time data transmission from geophysical equipment; Output interfaces: Equipped with an HDMI 2.1 interface and a 4K high-definition display, allowing for simultaneous display. Figure 2 The calibration fitting curve, Figure 3 The map displays the liquid level rise trend and the overall risk distribution heatmap, supporting zooming and panning to view details of high-risk concentrated areas in the old city. Clicking on any risk point in the map retrieves the corresponding feature data from Table 1. Figure 3 Liquid level simulation curve; it also supports printing out the spatial distribution map of "Risk of Overflow of Main Drainage Pipeline in City M under 24h Rainfall with a 20-Year Return Period" in print, as well as the extended version of Table 1 (including the risk level of all well points in the region) and the diagnostic report. Auxiliary module: Built-in lithium battery, 8-hour battery life, suitable for outdoor on-site diagnostic scenarios, can retrieve the risk level of points in Table 1 in real time. Figure 3 The corresponding simulation curves support on-site decision-making; equipped with a professional heat dissipation system, it ensures stable operation of the equipment without performance degradation during long-term high-load operations (such as the prediction of the whole-domain model of overflow risk based on actual measurement data and the rendering of the spatial distribution map of "overflow risk of the main drainage pipe network in M ​​City under 24h rainfall once in 20 years").

[0066] When the device is running, it can quickly process map data through the processor, and output a flood risk prediction based on measured data simulation and a spatial distribution map of "Flood Risk of Major Drainage Pipeline in City M under 20-Year Return Period 24h Rainfall" (Table 1). Figure 3 The liquid level simulation results are fully linked, providing integrated "data-model-visualization" support for flood control operation and maintenance in M ​​City. The priority recommendations for high-risk area renovation are highly consistent with the spatial distribution map of "the risk of overflow of the main drainage pipe network in M ​​City under a 20-year return period 24h rainfall". This provides intuitive and accurate technical support for flood control operation and maintenance in M ​​City. In practical applications, it has guided the renovation projects of three high-risk pipe networks in the old city area.

[0067] In the preferred embodiment, the basic information in step 1 includes the spatial topology of the pipeline network, pipeline and manhole attribute information, surface elevation, land use type, soil permeability coefficient, and catchment zone boundaries. The spatial topology of the pipeline network includes the coordinates and connection relationships between pipelines and manholes. The pipeline and manhole attribute information includes the top and bottom elevations of the pipelines, slope, pipe diameter, material, and the bottom elevation of the manholes. These settings ensure that the data comprehensively and accurately reflects the characteristics of the pipeline network system, providing a reliable basis for subsequent analysis and calculations. By clarifying the relationships between various elements, the water flow path and state can be accurately simulated, drainage capacity can be effectively assessed, and scientific support can be provided for pipeline network planning, renovation, and operation and maintenance management.

[0068] In the preferred embodiment, the verification rules used in step 1 for error diagnosis and correction include isolated point detection, pipe diameter conflict detection, and flow direction correction. The verification rules are implemented in batch processing using PyQGIS scripts. Isolated point detection specifically involves: traversing all pipe segments in the area; if the starting point of a certain pipe segment... and the end point If there are no overlapping pipe points and the pipe is not located at the start or end point of the drainage system, an alarm will be triggered and a connection point will be recommended; pipe diameter conflict detection specifically refers to: when the pipe segment Pipe diameter Larger than downstream pipe section Pipe diameter An alarm is triggered, and the downstream pipe diameter is recalculated; the flow direction correction is specifically as follows: when and An alarm is triggered when pipe sections overlap, and the start and end points of the pipe segments are redefined. These settings effectively improve the accuracy of drainage system data and reduce manual verification workload. After the verification rules are executed, a detailed report is generated, recording the location, type, and correction suggestions of any anomalies, facilitating quick problem identification by engineers. The system also supports custom verification parameters to adapt to different regional design specifications.

[0069] In the preferred embodiment, the drainage simulation input file in step 1 uses QGIS software to convert the collected data into .inp format. The collected data includes pipe network topology, elevation model, soil type, and land use. The QGIS plugin extracts pipe segment, node, and sub-catchment attributes. These settings ensure a consistent data format and accurate adaptation to the simulation software requirements. The converted .inp file can be efficiently imported into professional drainage simulation programs, providing reliable data support for subsequent accurate simulations of water flow direction and water accumulation, ensuring the smooth operation of the entire drainage simulation process.

[0070] In the preferred embodiment, the key nodes in step 2 include the junction of the main and secondary pipelines, the upstream and downstream of the inverted siphon, and the high-level pipe section under sunny conditions. The time resolution of the sensor data acquisition is no higher than 5 minutes. The calibration and optimization are based on the liquid level and flow process during rainy days. The simulation effect is optimized by adjusting the infiltration rate of the sub-catchment area, the Manning coefficient of the main pipe, and the local loss coefficient of the branch pipe. The efficiency coefficient and RMSE index are used for evaluation and verified by independent rainfall events. The above settings can ensure that the model accurately simulates the dynamic response of complex pipe network systems. Especially during short-duration heavy rainfall, the liquid level peak error is controlled within 5%, and the flow process line consistency reaches more than 0.9, providing reliable data support for urban flooding early warning.

[0071] In the preferred embodiment, the confluence model in step 2 is solved simultaneously using the continuity equation and the Manning equation. This setup accurately simulates the dynamic distribution of water flow in complex terrain, and by combining the Manning roughness coefficient to reflect the frictional characteristics of the channel bed, a stable flow state is obtained through iterative calculation. This model is particularly suitable for non-uniform flow scenarios, and its computational efficiency is improved by more than 30% compared to traditional methods.

[0072] In the preferred embodiment, the risk level classification standard in step 3 is as follows: high risk corresponds to... Medium risk corresponds to Low risk correspondence ,in This represents the highest liquid level at the pipe point during the simulation. The corresponding surface elevation of the pipe point. The system assigns the pipe top elevation to each pipe point; this setting allows for precise assessment of the likelihood of overflows at different pipe points. High risk corresponds to a liquid level above the ground elevation, medium risk corresponds to a liquid level between the pipe top and the ground elevation, and low risk corresponds to a liquid level below the pipe top elevation. This classification facilitates targeted prevention and control measures.

[0073] In the preferred embodiment, the rainfall intensity formula in step 3 adopts the local rainfall intensity formula, combined with the Chicago rainfall pattern to synthesize a 1-50 year return period rainfall scenario, prioritizing the use of a 24-hour 20-year return period rainfall. These settings ensure that the rainfall scenario simulation more closely reflects the actual local climate characteristics, improving model accuracy. Simultaneously, key parameters such as runoff coefficient and peak flow are calculated for different return period rainfall scenarios, providing a scientific basis for subsequent drainage system design.

[0074] In the preferred embodiment, the pipeline feature information in step 4 includes catchment area, upstream and downstream pipeline slope, upstream and downstream pipeline bottom elevation, upstream and downstream pipeline diameter, upstream and downstream pipeline roughness, well bottom elevation, surface elevation, and rainfall return period. The dependent variable of the statistical relationship is a numerical overflow risk level, with high risk defined as 2, medium risk as 1, and low risk as 0. These settings allow for the construction of a multiple regression model between pipeline features and overflow risk, with the weights of each feature determined through training on historical data. The model validation stage employs cross-validation to ensure a prediction accuracy of over 85%, ultimately forming a dynamic assessment system for pipeline overflow risk.

[0075] In the preferred embodiment, the large-scale urban inspection well attribute data in step 5 is derived from the urban pipeline network survey, and the risk level spatial distribution map is drawn using QGIS based on the inspection well number, coordinates, and risk level. The above settings can ensure that the inspection well attribute data is comprehensive and accurate, and the risk level spatial distribution map is intuitive and clear, providing a reliable basis for the risk assessment and management of the urban pipeline network, helping to identify potential risk points in a timely manner, and improving the safety and stability of the urban pipeline network operation.

[0076] In a preferred embodiment, the computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of any of the above-described methods for diagnosing the overflow risk of urban-scale drainage networks. This configuration enables the computer-readable storage medium to accurately execute the drainage network overflow risk diagnosis process with the help of a processor, quickly acquire various network data, and efficiently and accurately determine the overflow risk level through analysis using a preset algorithm model, providing a reliable basis for urban drainage management.

[0077] In the preferred embodiment, the system includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the computer program, it implements the steps of the urban-scale drainage network overflow risk diagnosis method described above. The above configuration enables the system to efficiently process massive amounts of network data, quickly locate potential overflow risk points through intelligent algorithm models, and dynamically update the risk level by combining real-time monitoring data, providing accurate decision support for the operation and maintenance of urban drainage networks.

[0078] In summary, the present invention provides a method, device, computer-readable storage medium, and electronic device for diagnosing overflow risks in urban-scale drainage networks, effectively solving a key problem urgently needing to be addressed in the field of urban drainage system management technology—the efficient and accurate diagnosis of overflow risks in urban-scale drainage networks. Under the existing technological framework, large-scale and complex urban drainage networks generally suffer from stringent data requirements, difficulty in guaranteeing accuracy, massive computational loads, and high costs. These problems restrict the scientific layout and efficient operation and maintenance of urban flood control and drainage projects, posing a potential threat to urban safety. The present invention successfully overcomes the limitations of existing technologies, achieving accurate identification and quantitative assessment of overflow risks.

[0079] This diagnostic method first meticulously investigates the urban pipe network and meteorological and hydrological conditions in small areas, constructs a high-precision drainage mechanism model for these small areas, reproduces the changes in manhole liquid levels under different rainfall intensities, and classifies the overflow risk of each manhole point. Then, based on the data model approach, it uses machine learning algorithms to construct the statistical relationship between the model and the parameters of the corresponding manhole and its upstream and downstream pipe sections. Combined with the parameters of the entire pipe network, it deduce the quantitative risk of overflow at the urban scale, breaking through the efficiency and accuracy bottlenecks of traditional pure mechanism models in large-scale calculations, and providing a basis for assessing urban overflow risk.

[0080] From a technical perspective, this approach integrates the advantages of mechanistic and data models to address the risk diagnosis of overflow in urban-scale drainage networks, breaking the limitations of traditional single-model applications. It employs an artificial intelligence data model to quantify overflow risks and supports a dynamic update mechanism, enabling rapid iterative prediction models through incremental learning based on network modifications or new monitoring data. This dynamic adaptation to network changes is unique. Based on limited rainfall data and mechanistic model simulations, it achieves overflow risk classification for a vast number of inspection wells by assessing differences in well level rises. This solves the problems of limited urban rainstorm frequency and difficulty in statistically analyzing actual overflow well points. The method is practical, feasible, and ingenious.

[0081] From an application perspective, it has successfully overcome the key challenge of efficient and accurate diagnosis of overflow risks in urban-scale drainage pipe networks, and has overcome many problems of existing technologies when dealing with large-scale and complex pipe networks. It has not only significantly reduced the equipment investment and cost of monitoring and evaluating the liquid level and flow of the entire pipe network, but also significantly improved the diagnostic accuracy and calculation efficiency, opening up new ideas for urban drainage system management. The constructed dynamic update mechanism avoids the huge workload of re-establishing the analysis mechanism model, can continuously adapt to the dynamic changes of urban drainage pipe networks, maintain long-term effectiveness and accuracy, and greatly improve the practicality and operation and maintenance efficiency of the model.

Claims

1. A method of diagnosing overflows risk in a sewer network at an urban scale, characterized in that, The method comprises the following steps: Step 1: establishing a basic information database of the urban drainage system, selecting a representative area in the city for fine pipe network exploration, obtaining the basic information of the area including the information of the ground, the pipeline and the inspection well, integrating the data through GIS technology and correcting errors to form a drainage simulation input file; Step 2: monitoring the liquid level and flow parameters of key nodes in the drainage system by using sensor technology, and calibrating and optimizing the drainage simulation model in combination with historical data, wherein the Horton infiltration model is used for runoff generation in the sub-catchment area, and the nonlinear reservoir method is used for the confluence model; Step 3: using the rain intensity formula to simulate rainfall data under different return periods, using the high-precision drainage pipe network mechanism model obtained in step 2 to simulate the inspection well liquid level rise, and classifying the risk grade according to the relationship between the liquid level and the ground elevation and the pipe top elevation; Step 4: extracting pipe point feature information and constructing a statistical relationship between hydraulic characteristics and overflow risk based on the SVM algorithm; Step 5: applying the statistical relationship to all inspection wells in the city on a large scale, predicting the global overflow risk by the SVM algorithm, determining the overflow grade of each well point and generating a risk grade spatial distribution map.

2. The urban-scale sewer network overflows risk diagnosis method according to claim 1, characterized in that: The basic information in step 1 includes the spatial topological relationship of the pipe network, the attribute information of the pipe network and the inspection well, the ground elevation, the land use type, the soil permeability coefficient and the catchment boundary, wherein the spatial topological relationship of the pipe network includes the coordinates and connection relationship of the pipeline and the inspection well, and the attribute information of the pipe network and the inspection well includes the top and bottom elevations of the pipeline, the slope, the pipe diameter, the material and the bottom elevation of the inspection well.

3. The urban-scale sewer network overflows risk diagnosis method of claim 1, wherein, The error diagnosis and correction in step 1 uses verification rules including isolated point detection, pipe diameter conflict detection, and flow direction correction. These verification rules are implemented in batch processing using PyQGIS scripts. Isolated point detection specifically involves: traversing all pipe segments within the area; if the starting point of a certain pipe segment... and the end point If there are no overlapping pipe points and the pipe is not located at the start or end point of the drainage system, an alarm will be triggered and a connection point will be recommended; pipe diameter conflict detection specifically refers to: when the pipe segment Pipe diameter Larger than downstream pipe section Pipe diameter An alarm is triggered, and the downstream pipe diameter is recalculated; the flow direction correction is specifically as follows: when and An alarm is triggered when the spaces overlap, and the start and end points of the pipe segment are redefined.

4. The urban-scale sewer network overflows risk diagnosis method of claim 1, wherein: The drainage simulation input file in step 1 is converted into the.inp format by QGIS software, and the collected data includes the pipe network topology, the elevation model, the soil type and the land use, and the pipe segment, node and sub-catchment area attributes are extracted by the QGIS plug-in.

5. The urban-scale sewer network overflows risk diagnosis method of claim 1, wherein: The key nodes in step 2 include the junction of the main and secondary trunk pipes, the upstream and downstream of the inverted siphon and the high liquid level pipe segment under sunny weather conditions, and the time resolution of the sensor collected data is not higher than 5 minutes; the calibration and optimization are based on the rain day liquid level and flow process, the simulation effect is optimized by adjusting the infiltration rate of the sub-catchment area, the Manning coefficient of the main pipe and the local loss coefficient of the branch pipe, and the efficiency coefficient and RMSE index are used for evaluation and verification by independent rainfall events.

6. The urban-scale sewer network overflows risk diagnosis method of claim 1, wherein, The Horton infiltration formula in step 2 is as follows: （1）； wherein is the time is the infiltration rate at time t, is the constant infiltration rate of the soil, is the initial infiltration rate of the soil, is the decay coefficient of the infiltration rate with time, is the duration of the rainfall.

7. The urban-scale sewer network overflows risk diagnosis method of claim 1, wherein, The confluence model in step 2 is solved by the continuity equation and the Manning equation, and the calculation formula is as follows: （2）； （3）； wherein, is the water volume in the catchment area, is the area of the catchment area, is the rainfall rate, is the water depth, is the runoff flow, is the characteristic width of the sub-catchment area, is the Manning coefficient of the sub-catchment area, is the average slope of the sub-catchment area, is the surface water depth of the sub-catchment area, is the maximum depression storage depth of the sub-catchment area.

8. The urban-scale sewer network overflows risk diagnosis method of claim 1, wherein, The risk level classification criteria in step 3 is: high risk corresponds to , medium risk corresponds to , and low risk corresponds to , wherein is the highest liquid level of the pipe point in the simulation process, is the corresponding ground elevation of the pipe point, is the corresponding pipe top elevation of the pipe point.

9. The urban-scale sewer network overflows risk diagnosis method of claim 1, wherein, The rain intensity formula in step 3 uses the local rain intensity formula, and combines the Chicago rain type to synthesize 1-50 year return period rainfall scenarios; the local rain intensity formula is as follows: （4）； wherein is the rainfall intensity, is the return period.

10. The urban-scale sewer network overflows risk diagnostic method of claim 1, wherein, The pipe point feature information in step 4 includes the catchment area, the upstream and downstream pipe slope, the upstream and downstream pipe bottom elevation, the upstream and downstream pipe diameter, the upstream and downstream pipe roughness, the well bottom elevation, the ground elevation and the rainfall return period; the dependent variable of the statistical relationship is the numerical overflow risk grade, the high risk is defined as 2, the medium risk is defined as 1, and the low risk is defined as 0.

11. The urban-scale sewer network overflows risk diagnosis method of claim 1 or 10, wherein, The training formula of the SVM algorithm in step 4 is as follows: （5）； In the formula, For the first Feature information vector of each tube point; Risk level; This is the weight vector; For bias; These are soft-interval relaxation variables; For matrix The transpose of the matrix; This is the error penalty coefficient; For input vectors An image that is implicitly mapped to a high-dimensional feature space; It indicates that it is subject to restrictions.

12. The urban-scale sewer network overflows risk diagnostic method of claim 1, wherein, The prediction formula of the SVM algorithm in step 5 is as follows: （6）； In the formula, is the target area risk prediction value, is the feature information vector of all pipe points of the target pipe section, is the support vector coefficient obtained by training; is the fitting kernel function; is the symbol function, outputting the risk category.

13. The urban-scale sewer network overflows risk diagnosis method of claim 1, wherein, The city large-scale inspection well attribute data in step 5 is derived from city pipe network census, and the risk level spatial distribution map is drawn by QGIS based on the inspection well number, coordinates and risk level.

14. An apparatus for diagnosing overflows in a municipal-scale sewer network, the apparatus being configured to implement a method for diagnosing overflows in a municipal-scale sewer network according to any one of claims 1 to 13, wherein the apparatus comprises: a processor; and a memory storing instructions that, when executed by the processor, cause the apparatus to perform the method. Comprise: The basic information acquisition module is used for acquiring representative area basic information through pipe network fine geophysical exploration, and a city drainage system basic information database is established after GIS integration and error diagnosis correction; The model calibration module is used for calibrating and optimizing the drainage simulation model by monitoring key node parameters through sensors and combining historical data, and the drainage simulation model is Horton infiltration formula and confluence calculation formula; The risk grading module is used for simulating different return period rainfall data by rain intensity formula, simulating liquid level change by high-precision drainage pipe network mechanism model, and classifying overflow risk level according to standards; The statistical relationship construction module is used for extracting pipe point feature information, and constructing the statistical relationship between hydraulic characteristics and overflow risk based on SVM training formula; The global prediction module is used for predicting the overflow risk of city large-scale inspection wells by SVM prediction formula, and generating an overflow risk level spatial distribution map.

15. A computer-readable storage medium, characterized in that: The computer readable storage medium stores a computer program, and the computer program is executed by the processor to realize the steps of the city scale drainage pipe network overflow risk diagnosis method in any one of claims 1-13.

16. An electronic device, comprising: Comprise a memory, a processor and a computer program stored on the memory and executable on the processor, and the processor executes the computer program to realize the steps of the city scale drainage pipe network overflow risk diagnosis method in any one of claims 1-13.

Images