Intelligent grading identification method for potential infiltration risk area of drainage pipe network

Abstract

The invention provides an intelligent grading identification method for a potential infiltration risk area of a drainage pipe network, and relates to the technical field of risk identification of the drainage pipe network. Establishing spatial relation mapping between triangular grid units of the surface water-underground water coupling model and pipe sections of the drainage pipe network model, calculating hourly liquid level difference time sequence data by utilizing underground water level time sequence data and corresponding pipe section liquid level time sequence data which are obtained through simulation calculation, and constructing risk indexes according to the hourly liquid level difference time sequence data; and then constructing an uncertainty scene sample set and respectively inputting the two models to obtain corresponding simulation calculation results, calculating to obtain a risk index corresponding to each pipe section, and judging a grading result of a potential infiltration risk area according to the risk indexes. Simulation calculation and intelligent diagnosis analysis are fused, and efficient identification of the potential infiltration risk area of the drainage pipe network is achieved on the premise that large-scale sensor deployment is not needed.

Description

Technical Field

[0001] The present invention relates to the technical field of drainage network risk identification, and in particular to a method for intelligently grading and identifying potential infiltration risk areas in a drainage network. Background Art

[0002] The drainage network primarily consists of rainwater pipes, sewage pipes, and their ancillary facilities. Rainwater pipes collect and transport rainwater, while sewage pipes collect and transport urban sewage. As a crucial component of urban infrastructure, the operational status of the drainage network is directly related to the quality of the urban environment. However, due to initial design issues, aging from prolonged use, and the influence of external environmental factors, groundwater infiltration caused by damage to the drainage network has become a bottleneck restricting its normal operation, seriously affecting its operational efficiency and increasing the burden on sewage treatment. This is particularly true in areas with abundant rainfall and high groundwater levels, where clean water encroaching on sewage pipes is a prominent problem.

[0003] To address groundwater infiltration caused by damaged drainage pipe networks, intelligent diagnostic methods for identifying potential infiltration risk areas within drainage networks are being developed. This not only helps identify risk areas but also provides a basis for subsequent risk warnings, operation and maintenance scheduling, and preventive remediation measures. This approach is crucial for ensuring the sustainable development of urban drainage systems. However, identifying potential infiltration risk areas within drainage networks currently faces numerous challenges. Drainage pipe networks are typically buried underground, complex in structure, large in scale, and highly coupled with other underground infrastructure, resulting in an intertwined spatial layout. Traditional inspection methods, such as CCTV endoscopic inspection and the deployment of liquid level sensors, while feasible in some areas, are difficult to achieve comprehensive regional coverage due to high deployment costs and labor requirements. Furthermore, inspection results are limited by pipe accessibility, diameter, and siltation conditions, making it difficult to achieve comprehensive regional coverage. Furthermore, drainage pipe networks exhibit significant time-varying and uncertainties. Even sections currently in good condition may face infiltration risks in the short term. Consequently, single-time inspections alone cannot accurately reflect the system's true risk level and cannot capture future risks. Furthermore, pipeline structural defects are inherently hidden, and even if local inspections reveal no anomalies, potential infiltration risks cannot be ruled out, making it difficult to achieve forward-looking and continuous risk identification using traditional methods. These dynamic evolutionary characteristics significantly increase the technical challenges of identifying high-risk areas and reinforce the urgency of building a system capable of dynamic simulation and intelligent identification.

[0004] Existing technologies primarily focus on upgrading "front-end" detection equipment, such as improving sensor performance and video recognition technology. However, comprehensive analysis of "back-end" data and diagnostic research based on mechanism-based deduction remain insufficient. This is especially true given the sparse nature of monitoring points and limited coverage. Relying solely on monitoring data is difficult to support the identification of potential infiltration risks across a large area, lacking systematicity and intelligence. Summary of the Invention

[0005] Purpose of the invention: The purpose of the present invention is to provide an intelligent hierarchical identification method for potential infiltration risk areas in drainage networks based on the collaborative deduction of a surface water-groundwater coupling model and a drainage network model.

[0006] Technical solution: A method for intelligently grading and identifying potential infiltration risk areas in a drainage network, comprising the following steps:

[0007] S1. Obtain basic data of the target area, construct and calibrate the initial surface water-groundwater coupling model and the initial drainage network model, and obtain the target surface water-groundwater coupling model and the target drainage network model;

[0008] S2. Using GIS spatial analysis tools, establish the correspondence or proximity relationship between each triangular grid unit of the target surface water-groundwater coupling model and each pipe segment of the target drainage network model based on spatial location;

[0009] S3. Close the groundwater simulation module in the target drainage network model, input meteorological data and typical sewage discharge time series characteristic data from the basic data to drive the target drainage network model to obtain the liquid level time series data for each pipe segment in the target area; use the same meteorological data to drive the target surface water-groundwater coupling model in parallel to obtain the groundwater level time series data for the triangular grid cells corresponding to each pipe segment;

[0010] S4. Utilizing the correspondence or proximity relationship between each triangular grid unit and each pipe segment, the groundwater level time series data of each triangular grid unit and the liquid level time series data of the corresponding pipe segment are used to calculate the liquid level difference data obtained by subtracting the liquid level of the corresponding pipe segment from the groundwater level of each triangular grid unit. Hourly liquid level difference time series data are constructed, and based on this, a risk index for each pipe segment is constructed.

[0011] S5. Set a number of key control factors and construct an uncertainty scenario sample set by perturbing these key control factors;

[0012] S6. Input the uncertainty scenario sample set into the target surface water-groundwater coupling model and the target drainage network model for simulation respectively to obtain the time series results of groundwater level and pipe segment liquid level under each scenario; use this result to calculate the risk index in S4 to obtain the risk index of each pipe segment under all uncertainty scenarios, and determine the classification results of potential infiltration risk areas based on the risk index of each pipe segment.

[0013] Specifically, in step S1, the basic data include basic data of the initial surface water-groundwater coupling model and basic data of the initial drainage network model; the basic data of the initial surface water-groundwater coupling model include elevation data, study scope, network distribution scope, underlying surface attribute data, meteorological data and calibration data; the basic data of the initial drainage network model include typical sewage discharge time series characteristic data of different land use types in the study area, hydraulic structure data, drainage zoning data, network data, demographic data, elevation data, underlying surface attribute data, sub-catchment area division data, meteorological data and calibration data.

[0014] Specifically, in step S1, the construction of the initial surface water-groundwater coupling model is based on the Intelliway-SSIM software. The steps of constructing and calibrating the initial surface water-groundwater coupling model are as follows:

[0015] Based on the elevation data, research scope, and pipe network distribution, the target area is divided into triangular grid cells in Intelliway-SSIM.

[0016] Based on the elevation data and underlying surface attribute data, each triangular grid unit is assigned corresponding attribute parameters;

[0017] Input meteorological data to drive Intelliway-SSIM to obtain simulation results of groundwater level data;

[0018] The calibration data is the historical monitoring data of groundwater levels. The groundwater level simulation results are compared with the historical monitoring data. If the error between the two exceeds the set threshold, the model parameters of Intelliway-SSIM are adjusted, or the grid is re-divided and the simulation is re-performed until the error does not exceed the set threshold.

[0019] Specifically, in step S1, the construction of the initial drainage network model is based on the Intelliway-SWMM software. The steps for constructing and calibrating the initial drainage network model are as follows:

[0020] Based on the elevation data and underlying surface attribute data, the corresponding attribute value is assigned to each sub-catchment, and the rainfall time series of the nearest meteorological station corresponding to each sub-catchment is extracted according to the geographical scope of each sub-catchment.

[0021] Input meteorological data and typical sewage discharge time series characteristic data to drive Intelliway-SWMM to run and obtain simulation results of flow, liquid level and water quality in each pipe section;

[0022] The calibration data are the historical monitoring data of the flow, liquid level and water quality of each pipe section. The simulation results of the flow, liquid level and water quality of each pipe section are compared with the historical monitoring data of the flow, liquid level and water quality of each pipe section. If the error between the two exceeds the set threshold, the model parameters of Intelliway-SWMM are adjusted and the simulation is repeated until the error does not exceed the set threshold.

[0023] Specifically, in step S4, the steps for constructing the risk index of each pipe section are as follows:

[0024] Based on the hourly liquid level difference time series data, a liquid level difference threshold is set to determine the presence of potential infiltration risk, and a risk index for the pipe section is constructed. The risk index includes the average water level difference, the proportion of time exceeding the standard, and the cumulative high head.

[0025] The average water level difference calculation formula is:

[0026]

[0027] Where: is the average water level difference in meters, and n is the The total number of moments in the simulation period, t represents the tth moment in the simulation period, T is the total number of moments in the simulation period, is the liquid level difference at the tth moment, in meters, The liquid level difference threshold for judging the potential risk of infiltration is in meters. is the groundwater level at the tth moment, in meters, is the liquid level of the pipe section at the tth moment, in meters;

[0028] The calculation formula for the time ratio exceeding the standard is:

[0029]

[0030]

[0031] Where: is the time ratio exceeding the standard, is the indicator function;

[0032] The calculation formula for cumulative super-high head is:

[0033]

[0034]

[0035] Where: is the cumulative super-high head, in meters, is the time step in hours.

[0036] Specifically, step S5 includes the following sub-steps:

[0037] S51. Select target surface water-groundwater coupling model parameters, rainfall boundary factors, and sewage boundary factors as key control factors. Target surface water-groundwater coupling model parameters include permeability, porosity, recharge rate, and discharge coefficient. Rainfall boundary factors include rainstorm return period, rainfall duration, and intensity-duration-frequency curve parameters. Sewage boundary factors include sewage discharge intensity per unit population and diurnal variation of sewage discharge.

[0038] S52. Setting corresponding disturbance modes for different key control factors;

[0039] S53. Set disturbance intervals for different key control factors based on historical monitoring data or expert knowledge;

[0040] S54. Apply perturbations to the key control factors to generate perturbation samples of the key control factors, perform physical rationality checks and parameter constraint relationship checks on the samples, and eliminate unqualified samples to obtain an initial perturbation sample set for each key control factor;

[0041] S55. Sample and combine the initial disturbance sample sets of each key control factor to construct an uncertainty scenario sample set.

[0042] Specifically, step S52 includes:

[0043] The distribution fitting method is used to determine the perturbation interval and probability distribution form of the target surface water-groundwater coupling model parameters;

[0044] The statistical characteristics of the rainfall boundary factor are inferred based on the historical meteorological data of the target area, and statistical modeling is carried out in combination with the historical rainstorm measured data in the target area. The rainfall process is empirically disturbed according to the rainstorm classification standard.

[0045] The pollution intensity per unit population is disturbed by normal distribution or uniform distribution, and a number of representative daily variation factor sequences are randomly selected from the typical daily variation curve sample set for the daily variation law of pollution discharge.

[0046] Specifically, step S55 includes:

[0047] S551. Performing Monte Carlo high-density sampling on the target surface water-groundwater coupling model parameter initial perturbation sample set to obtain an initial parameter sample set, performing normalization preprocessing on the initial parameter samples, clustering the initial parameter sample set using a clustering algorithm, extracting a number of samples of each category from the clustering results, performing denormalization on the samples, and constructing the target surface water-groundwater coupling model parameter perturbation sample set using the denormalized samples;

[0048] S552. Using the Latin hypercube sampling method on the initial perturbation sample set of the rainfall boundary factor, a perturbation sample set of the rainfall boundary factor is constructed;

[0049] S553. For the initial perturbation sample set of the discharge boundary factor, obtain the daily total sewage volume time series data based on the per capita sewage intensity. Combine several groups of representative daily variation factor sequences with the corresponding daily total sewage volume data, generate the daily sewage discharge time series by hourly multiplication, and construct the perturbation sample set of the discharge boundary factor.

[0050] S554. The obtained target surface water-groundwater coupling model parameter perturbation sample set, rainfall boundary factor perturbation sample set, and sewage boundary factor initial perturbation sample set are combined to obtain an uncertainty scenario sample set. The sample expression in the uncertainty scenario sample set is:

[0051]

[0052] Where: is a sample in the uncertainty scenario sample set, is the i-th sample in the parameter perturbation sample set of the target surface water-groundwater coupling model, is the jth sample in the rainfall boundary factor perturbation sample set, is the kth sample in the pollution boundary factor perturbation sample set.

[0053] Specifically, in step S6, the steps for determining the classification results of the potential infiltration risk areas according to the risk indicators of each pipe section are as follows: normalizing the risk indicators and setting a comprehensive risk indicator. The calculation formula of the comprehensive risk indicator is:

[0054]

[0055] Where: is the comprehensive risk index of the i-th pipeline section under all uncertainty scenarios, 、 and To set the weight, is the normalized average water level difference of the i-th pipe section under all uncertainty scenarios, is the normalized excess time ratio of the i-th pipe section under all uncertainty scenarios, is the normalized cumulative high water head of the i-th pipe segment under all uncertainty scenarios;

[0056] According to the comprehensive risk index results, the quantile method is used in combination with the threshold value set based on expert experience to determine and divide the potential infiltration risk level of each pipe section.

[0057] Specifically, the steps for determining and classifying the potential infiltration risk level of each pipe section using the quantile method combined with thresholds set based on expert experience are as follows:

[0058] Collect the comprehensive risk indicators of all pipeline sections under all uncertainty scenarios to form a risk indicator data set. Use the quantile method to divide the data set into several quantile intervals, and map each quantile interval to a predetermined risk level.

[0059] Adjust the upper and lower thresholds of the quantile interval based on expert experience;

[0060] Using the adjusted quantile interval, the comprehensive risk index of each pipe section is mapped to the corresponding risk level to obtain the potential infiltration risk level classification results of all pipe sections.

[0061] Beneficial effects: Compared with the prior art, the present invention has the following significant effects:

[0062] 1. The present invention integrates multi-source data under the framework of causal reasoning, uses the surface water-groundwater coupling model and the drainage network model to build a unified diagnostic framework, conducts systematic simulation of the pipeline network in the target area and identifies potential infiltration risks, simulates the dynamic changes of regional groundwater through the surface water-groundwater coupling model, and combines the liquid level difference and risk index calculation to realize intelligent identification of potential groundwater infiltration risk areas and risk level determination, which makes up for the limitation of traditional data measurement methods that are difficult to achieve full spatial coverage, improves the comprehensiveness and systematicness of potential infiltration risk identification, and at the same time greatly reduces monitoring costs and improves identification efficiency compared with traditional manual sampling and detection methods and monitoring methods that require the deployment of a large number of equipment.

[0063] 2. This invention has designed a three-dimensional risk indicator system centered around the following: average head difference exceeding threshold (R1), proportion of time exceeding threshold (R2), and cumulative high head (R3). This system quantitatively describes the potential groundwater infiltration risk based on infiltration intensity, duration, and cumulative effect. It can simultaneously capture both short-term sudden seepage and long-term delayed infiltration. Compared to single indicators or empirical methods, this three-dimensional risk indicator system is more comprehensive and sensitive, effectively reducing the risk of missed and misjudgment.

[0064] 3. The present invention constructs an uncertainty scenario sample set by setting and perturbation of key control factors, and uses the uncertainty scenario sample set to simulate the dynamic changes of groundwater status in time and space, breaking through the traditional single working condition assumption, and can accurately reflect the diverse response characteristics of potential infiltration risks under different conditions, and can improve the generalization ability and applicability of potential infiltration risk assessment results.

[0065] 4. The present invention constructs a comprehensive risk index based on the design of multiple risk indicators. For the results of the comprehensive risk index, a fusion grading mechanism of "initial quantile demarcation - expert threshold correction - event alignment verification" is constructed, taking into account the distribution characteristics of the scoring data and the reasonable judgment of actual experience. By using the quantile method to define the initial risk level boundaries, and then combining expert knowledge to make targeted adjustments to the key quantile points, a multi-level classification system with interpretability and generalizability is constructed. The present invention breaks through the traditional grading method's reliance on a single data or experience, improves the adaptability and credibility of the grading results in different regions and working conditions, and provides a more scientific basis for the identification and graded management of drainage pipe section risks.

[0066] 5. This method primarily relies on two types of fundamental monitoring data: pipe level and groundwater level. It is practically applicable even in the absence of a network-wide flow meter deployment, making it suitable for existing, aging pipe networks, complex hybrid systems, and newly constructed drainage areas. Furthermore, the identification of potential infiltration risk areas can complement existing technologies such as CCTV monitoring, smoke testing, and leakage measurement, creating a multi-faceted diagnostic system.

[0067] 6. Identification results are quantitatively graded and spatially located using pipe segments as basic units, providing excellent mapping capabilities and can be used to guide target management tasks such as annual treatment plan formulation, performance assessment, and overflow control. Supporting a closed-loop management logic of "monitoring-diagnosis-treatment-review," this system provides a scientific basis and technical support for the refined management and operation and maintenance of urban drainage networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is a flow chart of the method of the present invention.

[0069] Figure 2 It is a schematic diagram of the spatial mapping relationship between the ISSIM model and the ISWMM model units of the present invention.

[0070] Figure 3 It is a schematic diagram of the spatial distribution of local groundwater levels in the target area of ​​the present invention.

[0071] Figure 4 It is a time series curve diagram of groundwater depth and rainfall in the target area of ​​the present invention.

[0072] Figure 5 It is a schematic diagram of the hierarchical identification results of local potential infiltration risk areas in the target area of ​​the present invention. DETAILED DESCRIPTION

[0073] A preferred embodiment of the present invention is further described below with reference to the accompanying drawings.

[0074] See also Figure 1As shown, the present invention provides a method for intelligently grading and identifying potential infiltration risk areas in a drainage network, comprising the following steps:

[0075] S1. Obtain basic data of the target area, construct an initial surface water-groundwater coupling model and an initial drainage network model using the basic data, drive the initial surface water-groundwater coupling model and the initial drainage network model using the basic data to obtain initial simulation results, calibrate the initial surface water-groundwater coupling model using calibration data in the basic data and the initial simulation results to obtain a target surface water-groundwater coupling model, and calibrate the initial drainage network model using calibration data in the basic data and the initial simulation results to obtain a target drainage network model;

[0076] The initial surface water-groundwater coupling model construction and calibration process is as follows:

[0077] The initial surface water-groundwater coupling model adopted in the present invention is constructed based on the inteliway-SSIM software (hereinafter referred to as ISSIM). ISSIM is a three-dimensional surface water-groundwater coupling simulation software developed by Beijing Intelliway Environmental Technology Co., Ltd. The model can accurately depict the response of surface and groundwater processes under different conditions and the combined influence of human activities by fully coupling surface hydrological and groundwater processes. The model includes the following functions: one-dimensional river channel hydrodynamic process simulation, two-dimensional surface water process simulation, three-dimensional groundwater process simulation, two-way interactive simulation of surface water and groundwater, interactive simulation of river channel and surrounding surface and groundwater, open boundary interactive simulation and reservoir simulation. The reservoir simulation function module is not used in the present invention.

[0078] The basic data used to construct ISSIM include elevation data (DEM), research scope, pipeline network distribution range, underlying surface attribute data, meteorological data and calibration data; the underlying surface attribute data include soil type, vegetation distribution, geological structure and land use type; the meteorological data include rainfall time series data, temperature time series data, relative humidity time series data, wind speed time series data, and radiation time series data; the calibration data is the historical monitoring data of groundwater level.

[0079] Based on the elevation data, research scope and pipe network distribution range, the research scope of the target area is divided into triangular grid units in ISSIM; the triangular grid units are assigned corresponding attributes based on the elevation data and underlying surface attribute data; the meteorological data is input into ISSIM and driven by ISSIM, which outputs the initial simulation results, namely the simulation results of the groundwater level data. The simulation results of the groundwater level data are compared with the historical groundwater level monitoring data. If the error between the two exceeds the set threshold, the model parameters of ISSIM are adjusted, or the grid is re-divided and the simulation is re-run until the error does not exceed the set threshold.

[0080] The initial drainage network model construction and calibration process is as follows:

[0081] The initial drainage network model used in this paper is built using Intelliway-SWMM (ISWMM), an urban hydrology and stormwater drainage system simulation software developed by Beijing Intelliway Environmental Technology Co., Ltd. ISWMM, a readily available computing platform, can simulate flexible hydraulic processes and is used to simulate runoff and external inflows through structures such as pipes, channels, water storage / treatment facilities, and diversion structures. The model includes the following functions: one-dimensional network hydrodynamic process simulation, one-dimensional surface runoff generation and convergence simulation, pollutant accumulation and scour simulation, rainfall-induced inflow and infiltration simulation, snow accumulation simulation, groundwater simulation, and external groundwater interaction. The present invention utilizes a one-dimensional network hydrodynamic process simulation module.

[0082] The basic data used to construct the ISWMM include typical sewage discharge time series characteristic data, hydraulic structure data, drainage zoning data, demographic data, elevation data, underlying surface attribute data, sub-catchment area division data, meteorological data and calibration data for different land use types (residential land, commercial land, industrial land, public service land) in the study area; the typical sewage discharge time series characteristic data for different land use types in the study area are preferentially obtained through the monitoring data of drainage household water meters or sewage treatment plants to obtain the hourly sewage discharge time series characteristics of typical land use types; in the absence of measured data, it can be estimated through relevant literature or existing statistical results under similar land use types. Hydraulic structure data includes the layout, structural parameters, and operational logic data of facilities such as pump stations, gates, storage tanks, overflows, and diverters; drainage zoning data includes drainage zoning, catchment area, and control unit division data; pipe network data includes pipe shape, pipe drainage type, pipe size, pipe type, node type, node invert elevation, node depth, and the topological connection relationship of the pipe network system; underlying surface attribute data includes soil type, vegetation distribution, geological structure, and land use type; sub-catchment area division data is a standard vectorized shapfile format file obtained by sub-catchment area division based on elevation data and drainage zoning data; meteorological data includes rainfall time series data, temperature time series data, and evaporation time series data; calibration data includes historical monitoring data of flow, liquid level, and water quality at monitoring points in the pipe network system.

[0083] In the present invention, the data input process of ISWMM is as follows: among the input meteorological data, the rainfall time series data comes from the monitoring data of the rain gauge station. In the present invention, the latitude and longitude coordinates of the rain gauge station need to be input, and the sub-catchment area and the rain gauge station are matched by using the latitude and longitude coordinates. The matching method can adopt the nearest distance method or the inverse distance weighted method, so as to associate each sub-catchment area with the most reasonable rainfall data source, and realize the accurate matching of rainfall time series data and sub-catchment area; the model boundary data is obtained through the typical sewage discharge time series characteristic data, drainage zoning data, demographic data and pipeline network data of different land use types (residential land, commercial land, industrial land, public service land) in the study area, and finally generate the pipeline network sewage inflow boundary data file for input.

[0084] The specific steps for generating the sewage inflow boundary data file of the pipe network are as follows:

[0085] The land use layer is spatially overlaid with the drainage zoning data to count the population of each type of land use (such as residential, commercial, industrial, and public service land) in each drainage area.

[0086] Combined with the typical sewage discharge time series characteristic curves of various types of land use, the sewage inflow intensity of each drainage area is calculated on an hourly scale.

[0087] The sewage generation per unit population is further refined through demographic data, and the total sewage discharge of each drainage area at different times is calculated.

[0088] Establish a spatial mapping relationship between the drainage area and the actual pipe network nodes, determine the specific node location corresponding to the sewage inflow, and distribute the calculated sewage water volume and water quality load to the corresponding nodes in the pipe network system to ensure the reasonable distribution of sewage inflow and water quality.

[0089] The processed node sewage inflow distribution information is organized into a pipe network sewage inflow boundary data file that meets the requirements of the ISWMM platform, realizing accurate control of the spatiotemporal dynamic inflow process in the drainage system.

[0090] Based on elevation data and underlying surface attribute data, corresponding attribute values ​​are assigned to each sub-catchment, and the corresponding rainfall time series of the nearest meteorological station is extracted according to the geographical scope of each sub-catchment. Specifically, the latitude and longitude coordinates of the sub-catchment and the rain gauge are matched to achieve the allocation and spatial link between the sub-catchment and rainfall time series data. After the drainage network model is completed, the meteorological data and sewage discharge time series characteristic data are used to drive the ISWMM operation. The model response is then tested to verify its rationality and ensure that the ISWMM can accurately express and operate normally based on the existing data. The ISWMM is driven by the above input data and outputs initial simulation results, which are the simulation results of the flow rate, liquid level and water quality of each pipe section. The simulation results of the flow rate, liquid level and water quality of each pipe section are compared with the historical monitoring data of the flow rate, liquid level and water quality of each pipe section. If the error between the two exceeds the set threshold (such as 30%), the model parameters of the ISWMM are adjusted and the simulation (including hydraulic parameters and water quality parameters) is repeated until the error does not exceed the set threshold.

[0091] After completing the calibration of the ISSIM and ISWMM, the target surface water-groundwater coupling model and target drainage network model are obtained, respectively. By inputting the basic data, the target surface water-groundwater coupling model outputs groundwater level time series data for each triangular grid cell within the target area. The target drainage network model outputs flow rate time series data, water quality time series data, and liquid level time series data for each pipe segment within the target area. In this invention, flow rate time series data and water quality time series data are only used for ISWMM calibration.

[0092] S2. Use the spatial join function of the GIS spatial analysis tool to establish a one-to-one spatial relationship mapping between each triangular grid unit of the target surface water-groundwater coupling model and each pipe segment of the target drainage network model.

[0093] Use GIS spatial analysis tools to establish the spatial correspondence between each triangular grid unit of the target surface water-groundwater coupling model and each pipe segment of the target drainage network model. Please refer to Figure 2 As shown, Figure 2 Each small triangle in the figure represents a triangular grid unit, and each line segment with an arrow represents a pipe segment. With the help of the above spatial correspondence, the drainage pipeline can be located in the groundwater simulation area where it is located, realizing a one-to-one mapping of the simulation results of the above two models in space.

[0094] S3. Use the basic data to drive the target surface water-groundwater coupling model and the target drainage network model to obtain the groundwater level time series data of each triangular grid unit and the liquid level time series data of each pipe section.

[0095] To understand the potential driving force of groundwater infiltration into the drainage network, we first need to establish a baseline operating state for the network under no-infiltration conditions. We then compare the groundwater level time series data for each triangular grid cell within the target area with the corresponding pipe level time series data. The specific steps are as follows:

[0096] In the target drainage network model, the groundwater simulation module is turned off, and then meteorological data and typical sewage discharge time series characteristic data are used to drive the operation of the target drainage network model, obtain the liquid level time series data of each pipe section in the target area, and obtain the reference working condition without infiltration.

[0097] Under the same boundary conditions, meteorological data are used to drive the target surface water-groundwater coupling model in parallel to obtain the groundwater level time series data of the triangular grid cells corresponding to each pipe segment.

[0098] S4. Using the spatial relationship mapping established in step S2, calculate the hourly liquid level difference time series data of the groundwater level and the pipe segment liquid level through the groundwater level time series data of each triangular grid unit and the liquid level time series data of the corresponding pipe segment, and use the hourly liquid level difference time series data to construct the risk index of the pipe segment.

[0099] Since the spatial correspondence between each triangular grid unit of the target surface water-groundwater coupling model and each pipe section of the target drainage network model has been established, based on the established spatial mapping relationship, the groundwater level and pipe section liquid level data of the triangular grid units corresponding to each pipe section are selected, compared at the same time step, and the hourly liquid level difference time series is calculated.

[0100] Based on the hourly liquid level difference time series data, the liquid level difference threshold for judging the potential infiltration risk is set, and risk indicators are constructed. The risk indicators include the average water level difference, the proportion of time exceeding the standard, and the cumulative high head.

[0101] The liquid level difference threshold for determining potential infiltration risk can be set or dynamically calibrated based on on-site monitoring data, historical case data backtracking, or engineering experience:

[0102] Engineering experience: Commonly used value range: 0.05m–0.20m; reference can be made to literature or existing experience in urban drainage system risk assessment.

[0103] Historical case data backtracking: Backtrack existing historical infiltration cases or monitoring data; find the liquid level difference data at the corresponding time and space nodes, take the average value or a conservative value to set the liquid level difference threshold for judging the potential infiltration risk.

[0104] On-site monitoring data: Select typical well points to lay groundwater monitoring wells and conduct pipe network liquid level monitoring; measure the liquid level difference when infiltration occurs, and use it as the initial value of the liquid level difference threshold to determine the existence of potential infiltration risks; and optimize and iteratively update this initial threshold value as the risk identification system operates.

[0105] The average water level difference is used to measure the average degree to which the groundwater level is generally above the potential infiltration threshold of the pipeline network. Only the positive values ​​of the liquid level difference are averaged. This indicator is used to reflect the intensity of the pipeline being driven by groundwater over a long period of time. The calculation formula is:

[0106]

[0107]

[0108] Where: is the average water level difference in meters, and n is the The total number of moments in the simulation period, t represents the tth moment in the simulation period, T is the total number of moments in the simulation period, is the liquid level difference at the tth moment, in meters, The liquid level difference threshold for judging the potential risk of infiltration is in meters. is the groundwater level at the tth moment, in meters, is the liquid level of the pipe section at the tth moment, in meters;

[0109] The exceeding time ratio is used to measure the proportion of the duration of groundwater level difference exceeding the threshold value for judging the potential infiltration risk in the total simulation period. The calculation formula is:

[0110]

[0111]

[0112] Where: is the time ratio exceeding the standard, is an indicator function that reflects the persistence and frequency of the potential infiltration driving force. The exceeding state at each time point t is converted into a binary value of 1 or 0 for easy statistics. The exceeding time ratio Reflects the proportion of time that the groundwater level is higher than the pipe network for a long time, and the value range is [0,1]. It means that the groundwater has never been higher than the pipe network; It means that the groundwater is always higher than the pipe network; The closer it is to 1, the more continuous and frequent the potential infiltration driving force of groundwater into the pipeline network is.

[0113] Cumulative high head estimates the potential scale of infiltration by accumulating all high head differences that exceed the level difference threshold for determining potential infiltration risk. This indicator quantifies the cumulative head pressure of groundwater exceeding the pipe network and can serve as a reliable proxy parameter for the potential scale of groundwater infiltration (volume or total driving force), providing a scientific basis for identifying potential infiltration risk areas. The calculation formula is:

[0114]

[0115]

[0116] Where: is the cumulative super-high head, in meters, is the time step in hours.

[0117] Accumulated super high head The larger the value, the higher the potential risk of groundwater infiltration into the pipe network, which may correspond to a larger total infiltration amount or a more serious area.

[0118] The above-mentioned risk indicator settings have the advantages of being quantifiable, visual, and integrable. They can be directly embedded in the computing platform to achieve automatic identification and prioritization of potential infiltration risk areas in the drainage network, supporting subsequent risk warnings, operation and maintenance scheduling, or renovation decisions.

[0119] S5. Set key control factors and construct an uncertainty scenario sample set based on the disturbance of key control factors.

[0120] In order to characterize the sources of uncertainty in the potential groundwater infiltration risk, this paper constructs a representative scenario sample set based on the disturbance of control factors, and constructs a scenario uncertainty analysis framework for the potential groundwater infiltration risk as the driving input for model calculation.

[0121] S51. Select target surface water-groundwater coupling model parameters, rainfall boundary factors, and sewage boundary factors as key control factors. Target surface water-groundwater coupling model parameters include permeability, porosity, recharge rate, and discharge coefficient. Rainfall boundary factors include rainstorm return period, rainfall duration, and intensity-duration-frequency (IDF) curve parameters. Sewage boundary factors include sewage intensity per unit population and diurnal variation of sewage discharge.

[0122] S52. Set corresponding disturbance modes for different key control factors.

[0123] Target surface water-groundwater coupling model parameter perturbation method: For each parameter, if the parameter has engineering measurement or monitoring data in the study area, the distribution fitting method (such as normal distribution, uniform distribution, etc.) is preferably used to determine its disturbance interval and its probability distribution form; if the parameter has no direct data support, a reasonable disturbance range is set based on empirical data or literature research, usually using uniform distribution or triangular distribution modeling, and setting a relative fluctuation range of ±20%-±50%.

[0124] Rainfall boundary factor perturbation method: infer the statistical characteristics of the rainfall boundary factor based on the historical meteorological data of the target area (such as rainfall time series, rainstorm process data), conduct statistical modeling based on the historical rainstorm measured data of the target area, refer to the rainstorm measured process in the target area for the control parameters in the designed rainfall type, and empirically perturb the rainfall process according to the rainstorm classification standard, and set a reasonable offset range, such as ±10%-±30%.

[0125] Pollution boundary factor perturbation method: The pollution intensity per unit population is disturbed by normal distribution or uniform distribution, and the daily variation law of pollution discharge is randomly selected from the typical daily variation curve sample set to generate several representative daily variation factor sequences. The daily variation factor sequence refers to the time series variation ratio sequence of sewage flow or load at the unit day scale, which is usually extracted by normalization of historical monitoring data. It is randomly called according to the scenario in the simulation to achieve diversified expression of sewage boundary conditions.

[0126] S53. Set the disturbance range of different key control factors.

[0127] Target surface water-groundwater coupling model parameter disturbance range:

[0128] If there is support from measured or monitored data, the disturbance interval should be set based on the parameter statistics as follows:

[0129]

[0130]

[0131] Where: is a parameter the mean, median or typical value during the monitoring period; For this parameter The disturbance amplitude (for example, 20%, 30%, 50%) can be obtained through the following paths: the maximum and minimum values ​​of historical monitoring values; the statistical range of samples from similar areas; the results of sensitivity analysis in the literature and expert experience judgment.

[0132] Rainfall boundary factor disturbance interval:

[0133] Heavy rain return period (hereinafter referred to as T) disturbance: multiple representative return periods (such as 2 years, 5 years, 10 years, 20 years, and 50 years) are selected to form a discrete disturbance sample set.

[0134] Rainfall duration disturbance (hereinafter referred to as D): Select multiple representative durations as discrete samples. Common durations include 5 minutes, 15 minutes, 30 minutes, 1 hour, 3 hours, and 6 hours. Ensure that typical scenarios of short-term heavy rainfall and long-term slow rainfall are covered.

[0135] IDF parameter disturbance: The IDF empirical formula obtained by fitting the regional measured rainstorm data is as follows:

[0136]

[0137] Where: is the rainfall intensity under a certain rainstorm return period T and rainfall duration D; a, b, c, d are all fitting parameters.

[0138] Identify the fitting parameters a, b, c, d. Based on the fitting results of references or measured data, set the disturbance range for each fitting parameter:

[0139]

[0140]

[0141]

[0142]

[0143] Where: * represents the reference value of the fitting parameter (reference literature or measured data fitting result), Indicates the magnitude of the disturbance (typically ±10%-30%), determined from historical matching, regional studies, or empirical data.

[0144] Discharge boundary factor disturbance interval:

[0145] Perturbation of pollution intensity per unit population: Combined with the sewage treatment plant inflow concentration monitoring data or statistical water discharge coefficient, set normal distribution or uniform distribution, and construct a disturbance sample set within the range of ±30%;

[0146] Perturbation of Diurnal Emission Patterns: Multiple typical daily variation curves (e.g., weekday, weekend, holiday, and extreme peak patterns) are obtained, each corresponding to the emission ratio for a specific time period within a 24-hour period. During the simulation, a curve is randomly selected or specified based on the scenario as the hourly ratio input. Each curve sample can be derived from historical regional monitoring data or configured based on different types of living and / or emission behaviors to ensure uniformity (the sum of the 24-hour factors is 1) and physical rationality.

[0147] S54. Apply disturbance to the key control factors to generate disturbance samples of the key control factors, perform physical rationality check and parameter constraint relationship check on the disturbance samples of the key control factors, and obtain the initial disturbance sample set of different key control factors after eliminating unqualified samples.

[0148] Physical rationality check: exclude parameters with unreasonable physical meaning (such as negative permeability), verify each parameter item by item using the judgment rules within the set range, and judge and eliminate illegal values ​​row by row. The typical rules used in this invention are as follows:

[0149] Target surface water-groundwater coupling model parameters:

[0150] Permeability coefficient Porosity ;Replenishment rate , and cannot exceed a reasonable proportion of the total annual rainfall (such as 70%); the discharge coefficient .

[0151] Rainfall boundary factor:

[0152] Heavy rain recurrence period ;In IDF parameters: , , , ,

[0153] Discharge boundary factor:

[0154] The pollution intensity per unit population is greater than 0, the sum of the daily variation factors is equal to 1, and the hourly value is in the interval [0,1].

[0155] Parameter constraint relationship check: perform constraint judgment on related parameters, and create a Boolean expression filter check_param_constraints(sample) for each type of parameter constraint to form a set of filtering and control conditions. The typical rules used in this invention are as follows:

[0156] Target surface water-groundwater coupling model parameters:

[0157] Porosity relationship constraints:

[0158]

[0159] Where: is the total porosity, is the effective porosity.

[0160] Relationship between recharge rate and infiltration volume:

[0161]

[0162] Where: is the replenishment rate, The infiltration amount.

[0163] Interlayer water permeability constraint:

[0164]

[0165] Where: is the vertical hydraulic conductivity, is the horizontal hydraulic conductivity coefficient.

[0166] Hydraulic gradient constraint:

[0167]

[0168] Where: is the water head difference, is the distance difference.

[0169] Rainfall boundary factor:

[0170] The IDF parameter curve decreases monotonically (i.e., rainfall intensity decreases with increasing duration D):

[0171]

[0172] Discharge boundary factor:

[0173] The fluctuation range of sewage load in different periods should be controlled within a reasonable range (to avoid peaks that exceed actual capacity); the maximum fluctuation ratio after normalization of the daily variation curve shall not exceed a certain threshold (such as less than or equal to 3 times the mean).

[0174] S55. Sample and combine the initial disturbance sample sets of each key control factor to construct an uncertainty scenario sample set.

[0175] This step includes the following sub-steps:

[0176] S551. Perform Monte Carlo high-density sampling on the initial perturbation sample set of the target surface water-groundwater coupling model parameters to obtain an initial parameter sample set, perform normalization preprocessing on the initial parameter samples, and then use a clustering algorithm to cluster the initial parameter sample set. Extract several samples of each category from the clustering results, denormalize the samples, and use the denormalized samples to construct the target surface water-groundwater coupling model parameter perturbation sample set.

[0177] Monte Carlo high-density sampling:

[0178] In the set parameter perturbation space, each parameter is independently randomly sampled to form an initial high-dimensional parameter sample set , the sample size N is usually 500 to 5000; each sample point They are all vectors of dimension d (d represents the number of disturbance parameters), representing a complete set of model input parameter combinations.

[0179] Parameter space normalization:

[0180] After the parameter perturbation samples are generated, the original parameter data needs to be scaled uniformly. −6 ∼10 −4 m / s, while porosity is generally a dimensionless value of 0.2-0.5). If left unprocessed, certain high-dimensional parameters will dominate the distance metric in cluster analysis, affecting the accuracy of selecting representative samples. To improve the performance of clustering algorithms in high-dimensional parameter spaces, normalization preprocessing is required to bring all parameters into a uniform scale range, preventing the absolute dimension of a single parameter from dominating the clustering results.

[0181] Normalization method:

[0182] For parameters with known physical range and relatively uniform distribution, each dimension parameter x is normalized according to the following formula:

[0183]

[0184] Where: is the normalized parameter, are the parameters to be processed, is the minimum value of the parameter sample, is the maximum value of the parameter sample.

[0185] After processing, all parameters fall uniformly into the interval [0,1], which is conducive to the uniform weight processing of Euclidean distance; if there are out-of-bounds samples (such as sampling beyond the empirical interval), they can be uniformly cropped or extrapolated.

[0186] For parameters that obey a normal distribution or an approximately symmetrical distribution, the Z-score method is used for normalization:

[0187]

[0188] Where: is the sample mean, is the sample standard deviation.

[0189] After processing, the mean of all parameters is 0 and the standard deviation is 1, which is conducive to preserving relative variability. It is suitable for situations where there are outlier parameters or obvious distribution fluctuations.

[0190] The specific processing method of the target surface water-groundwater coupling model parameters is shown in Table 1 below.

[0191] Table 1 Parameter Type Treatment method illustrate Permeability coefficient, excretion coefficient Z-score method With order of magnitude fluctuations, avoid cluster dominance Porosity Normalization The physical range is known, suitable for scaling Infiltration rate Depends on the distribution When the distribution is abnormal, consider first performing Box-Cox transformation and then normalization

[0192] The processed sample set is stored in a uniform matrix format for subsequent clustering algorithm invocation. The normalization effect is verified by using the parameter principal component analysis (PCA) dimensionality reduction projection results to determine if there are any abnormal shifts or over-concentrations. If the parameters remain highly correlated after normalization, feature dimensionality reduction or covariance filtering is performed before clustering.

[0193] A clustering algorithm is used to spatially cluster the initial parameter sample set, compress redundant information, and extract representative parameter scenario cluster centers.

[0194] The available clustering methods are as follows:

[0195] K-Means clustering: suitable for scenarios with a large number of samples (>500), and easy to control the number of clusters K;

[0196] K-Medoids (PAM) clustering: more robust, the cluster center is the actual sample;

[0197] DBSCAN / hierarchical clustering: used to discover non-spherical clusters and is suitable for complex distribution scenarios;

[0198] The number of clusters K can be automatically optimized based on indicators such as the Silhouette Score and the Davies-Bouldin Index, or empirically set to 5%-10% of the total number of samples.

[0199] After clustering is completed, one or more representative samples are extracted from each cluster as subsequent simulation samples (such as selecting the cluster center, the most densely populated sample within the cluster, etc.) to construct the target surface water-groundwater coupling model parameter perturbation sample set. , K is the total number of samples.

[0200] S552. Use the Latin hypercube sampling method on the initial perturbation sample set of the rainfall boundary factor to construct the perturbation sample set of the rainfall boundary factor.

[0201] Based on the established perturbation methods and intervals for each parameter, the Latin Hypercube Sampling (LHS) method was used to construct a joint perturbation sample set in a high-dimensional space. The rainfall process was perturbed using a six-factor combination (return period T, rainfall duration D, and perturbation factors a, b, c, and d). Although correlated, the parameters were treated as independent dimensions during the uncertainty sampling process, forming a six-dimensional perturbation parameter space.

[0202] After determining the reasonable value range and distribution assumptions for the parameters T, a, b, c, d, and D (which can be uniform or normal distribution, based on historical measured data or design specifications), the six dimensions are divided into layers and a sample point is randomly selected from each layer. By reorganizing the sampling results of the six parameter dimensions, N groups of different perturbation parameter combinations are generated to form a high-dimensional sampling matrix:

[0203]

[0204] Each parameter combination The corresponding design rainfall intensity value is calculated by the IDF empirical formula:

[0205]

[0206] In order to construct a complete rainfall process curve (rainfall-time series) that can be used for model driving, Divide into several time periods and The rainfall in each period is allocated according to the assumed rainstorm type (such as positive skewness, moderate skewness, inverted skewness or bell-shaped distribution), and finally a complete rainfall boundary factor perturbation sample set is formed.

[0207] S553. For the initial perturbation sample set of the discharge boundary factor, the daily total sewage volume time series data is obtained according to the pollution intensity per unit population. Several groups of representative daily variation factor sequences are combined with the corresponding daily total sewage volume data. The daily sewage discharge time series is generated by hourly multiplication operations to construct the perturbation sample set of the discharge boundary factor.

[0208] Based on long-term historical observation data, monitoring results, and actual operating experience, the system extracts typical daily variation coefficient curves for sewage discharge. These curves cover typical operating conditions, including but not limited to: weekday discharge patterns, holiday discharge patterns, rainy and rainy day discharge patterns, and drought or low-discharge periods. Each typical curve, based on a 24-hour period, depicts the hourly distribution of sewage discharge, meeting coefficient normalization requirements (summing to 1 over 24 hours), reflecting the intraday variation in sewage discharge intensity.

[0209] To reflect the volatility of total sewage discharge, a typical daily variation coefficient curve is used to simulate the total sewage discharge volume for each day with random perturbations. The perturbation amplitude is usually set to ±20%, and uniform or normal distribution random sampling is used to ensure the rationality and diversity of sewage discharge changes.

[0210] N representative curves were selected from a library of typical daily variation coefficients, covering the main operating conditions. For each typical curve, M sets of different total sewage volume perturbation values ​​were generated to create a diverse set of total discharge volumes. Combining the typical daily variation coefficients with the corresponding perturbed total sewage volumes, N×M sets of sewage discharge pattern scenarios were constructed, yielding a sample set of perturbation boundary factors. Each sewage discharge pattern was derived by multiplying the typical daily coefficient by the total sewage volume for that day, ensuring the typicality of the daily sewage discharge ratio while also incorporating dynamic variations in sewage discharge volumes.

[0211] S554. The obtained target surface water-groundwater coupling model parameter perturbation sample set, rainfall boundary factor perturbation sample set, and sewage boundary factor initial perturbation sample set are combined to obtain an uncertainty scenario sample set. The sample expression in the uncertainty scenario sample set is:

[0212]

[0213] Where: is a sample in the uncertainty scenario sample set, is the i-th sample in the parameter perturbation sample set of the target surface water-groundwater coupling model, is the jth sample in the rainfall boundary factor perturbation sample set, is the kth sample in the pollution boundary factor perturbation sample set.

[0214] Samples in the uncertainty scenario sample set The input path is: and Input the target drainage network model and and Input target surface water-groundwater coupled model.

[0215] S6. Input the uncertainty scenario sample set into the target surface water-groundwater coupling model and the target drainage network model respectively to obtain the simulation calculation results of the groundwater level time series data and the pipe segment liquid level time series data. Input the simulation calculation results into step S4 to calculate the risk index of each pipe segment under all uncertainty scenarios, and determine the potential infiltration risk area based on the risk index.

[0216] The target surface water-groundwater coupling model and the target drainage network model are driven by the uncertainty scenario sample set to obtain simulation results. The specific steps include:

[0217] The parameter combinations for the uncertainty scenario sample set were automatically imported into the above model to enable automated batch-driven model calculations. This process was driven by an automated simulation scheduling script (a Python batch processing tool) and executed in parallel on a high-performance computing cluster (a SLURM-based HPC platform). Using a unified format specification, each parameter was automatically converted to the model input format to ensure parameter consistency and accuracy. The model was then automatically executed in batches, completing simulations for all uncertainty scenarios in sequence and saving the corresponding simulation output results.

[0218] Simulation results were extracted from the outputs of the target surface water-groundwater coupling model and the target drainage network model, respectively. The former yielded groundwater level time series data, while the latter obtained pipeline level time series data. The extracted data from multiple uncertainty scenarios were converted to a unified format and normalized using the Z-score to ensure consistent spatial and temporal resolution, meeting the requirements of subsequent risk identification and analysis.

[0219] The three risk indicators for each pipe section under each uncertainty scenario are calculated. Through the integrated analysis of risk indicators under multiple scenarios, the identification and judgment of potential groundwater infiltration risk areas are supported.

[0220] The steps to determine potential infiltration risk areas based on risk indicators are as follows:

[0221] Normalize the risk indicators and set the comprehensive risk index. The calculation formula of the comprehensive risk index is:

[0222]

[0223] Where: is the comprehensive risk index of the i-th pipeline section under all uncertainty scenarios, 、 and To set the weight, equal weight distribution is adopted by default. is the normalized average water level difference of the i-th pipe section under all uncertainty scenarios, is the normalized excess time ratio of the i-th pipe section under all uncertainty scenarios, is the normalized cumulative high water head of the i-th pipe segment under all uncertainty scenarios.

[0224] Based on the comprehensive risk index results, the quantile method is used in combination with thresholds set based on expert experience to determine and classify the potential infiltration risk level of each pipe section. The steps are as follows:

[0225] Collect the comprehensive risk indicators of all pipe sections to form a risk indicator data set, and use the quantile method to divide the risk indicator data set into several quantile intervals (taking the quartile as an example, The first quartile, The second quartile, is the third quartile), and each quartile interval is mapped to a predetermined risk level:

[0226] : Corresponding to low risk level;

[0227] : Corresponding to medium and low risk levels;

[0228] : Corresponding to medium and high risk levels;

[0229] : Corresponding to high risk level.

[0230] After completing the basic grading, adjust the upper and lower thresholds of the quantile intervals obtained based on expert experience;

[0231] Using the adjusted quantile interval, the comprehensive risk index of each pipe section is mapped to the corresponding risk level to obtain the potential infiltration risk level classification results of all pipe sections.

[0232] The method for adjusting the upper and lower thresholds of the divided quantile interval is as follows:

[0233] Set empirical threshold range: Combine historical infiltration events, groundwater high water level statistics and measured infiltration intensity to set reference score cutoff values, such as:

[0234] : Corresponding to low risk level;

[0235] : Corresponding to medium risk level;

[0236] : Corresponding to high risk level.

[0237] Experts participate in fine-tuning: Organize experts to review the preliminary grade results and make appropriate fine-tuning to certain quantiles (such as The corresponding value was raised from 0.81 to 0.85), ensuring that the grading standards are more in line with regional management practices.

[0238] To strengthen the empirical basis of the classification standard, the scoring boundaries can also be reverse-calibrated through typical historical events: select typical groundwater infiltration events (such as overflows, sudden flow increases, and water level anomalies) and mark the affected pipe sections; extract the scores of the affected pipe sections; , compare it with the preset level threshold to identify whether the score value is consistent with the severity of the event; combine ROC curve, confusion matrix and other indicators to optimize the threshold setting and improve the accuracy and verifiability of graded discrimination.

[0239] To demonstrate the feasibility and effectiveness of the above method, a specific implementation scenario is described below. The present invention selects basic data for the entire year of 2023 in a target area, processes and calculates it using the above method, and obtains corresponding test results.

[0240] Please refer to Figure 3 As shown, Figure 3 This is a schematic diagram of the spatial distribution of groundwater levels in a local area at a certain time point within the target area. For the triangular grid cells divided in the target area, the redder the color, the higher the groundwater level, and the bluer the color, the lower the groundwater level.

[0241] Please refer to Figure 4 As shown, Figure 4 This is the calculated time series curve of groundwater depth and rainfall for the whole year of 2023. The groundwater depth is the vertical distance from the groundwater surface to the ground surface, which can intuitively reflect the changes in groundwater level. Figure 4 The blue curve is the time series curve of groundwater depth, and the red curve is the time series curve of rainfall. Figure 4 It can be seen intuitively that there are obvious differences in the response of groundwater level under different rainfall events, and it can respond to rainfall events in real time.

[0242] Please refer to Figure 5 As shown, Figure 5 The results of potential infiltration risk area classification within the target area calculated by step S6 are shown in the figure. The dots in the figure represent the corresponding pipe sections. Dark purple represents the pipe section corresponding to high risk level, light purple represents the pipe section corresponding to medium risk level, and white represents the pipe section corresponding to low risk level. Figure 5 It can be seen that the main pipe and some branch pipes face a high potential infiltration risk. At the same time, there is a large difference in potential infiltration risk level between two pipe sections with close geographical distances. This is related to the height of the pipeline network and the local height of the groundwater level, and is consistent with the actual measurement results.

Claims

1. A method for intelligently identifying potential infiltration risk areas in a drainage network, characterized by: The following steps are involved: S1. Obtain basic data of the target area, construct and calibrate the initial surface water-groundwater coupling model and the initial drainage network model, and obtain the target surface water-groundwater coupling model and the target drainage network model; S2. Using GIS spatial analysis tools, establish the correspondence or proximity relationship between each triangular grid unit of the target surface water-groundwater coupling model and each pipe segment of the target drainage network model based on spatial location; S3. Close the groundwater simulation module in the target drainage network model, input meteorological data and typical sewage discharge time series characteristic data in the basic data to drive the target drainage network model to operate, and obtain the liquid level time series data of each pipe section in the target area; The target surface water-groundwater coupling model is driven in parallel using the same meteorological data to obtain the groundwater level time series data of the triangular grid cells corresponding to each pipe segment. S4. Utilizing the correspondence or proximity relationship between each triangular grid unit and each pipe segment, the groundwater level time series data of each triangular grid unit and the liquid level time series data of the corresponding pipe segment are used to calculate the liquid level difference data obtained by subtracting the liquid level of the corresponding pipe segment from the groundwater level of each triangular grid unit. Hourly liquid level difference time series data are constructed, and based on this, a risk index for each pipe segment is constructed. S5. Set a number of key control factors and construct an uncertainty scenario sample set by perturbing these key control factors; S6. Input the uncertainty scenario sample set into the target surface water-groundwater coupling model and the target drainage network model for simulation respectively to obtain the time series results of groundwater level and pipe segment liquid level under each scenario; use this result to calculate the risk index in S4 to obtain the risk index of each pipe segment under all uncertainty scenarios, and determine the classification results of potential infiltration risk areas based on the risk index of each pipe segment.

2. The method for intelligently identifying potential infiltration risk areas in a drainage network according to claim 1, characterized in that: In step S1, the basic data include basic data of the initial surface water-groundwater coupling model and basic data of the initial drainage network model; the basic data of the initial surface water-groundwater coupling model include elevation data, study scope, network distribution scope, underlying surface attribute data, meteorological data and calibration data; the basic data of the initial drainage network model include typical sewage discharge time series characteristic data of different land use types in the study area, hydraulic structure data, drainage zoning data, network data, demographic data, elevation data, underlying surface attribute data, sub-catchment area division data, meteorological data and calibration data.

3. The method for intelligently identifying potential infiltration risk areas in a drainage network according to claim 2, characterized in that: In step S1, the construction of the initial surface water-groundwater coupling model is based on the Intelliway-SSIM software. The steps of constructing and calibrating the initial surface water-groundwater coupling model are as follows: Based on the elevation data, research scope, and pipe network distribution, the target area is divided into triangular grid cells in Intelliway-SSIM. Based on the elevation data and underlying surface attribute data, each triangular grid unit is assigned corresponding attribute parameters; Input meteorological data to drive Intelliway-SSIM to obtain simulation results of groundwater level data; The calibration data is the historical monitoring data of the groundwater level. The groundwater level simulation results are compared with the historical monitoring data of the groundwater level. If the error between the two exceeds the set threshold, the model parameters of Intelliway-SSIM are adjusted, or the grid is re-divided and the simulation is re-simulated until the error does not exceed the set threshold.

4. The method for intelligently identifying potential infiltration risk areas in a drainage network according to claim 2, characterized in that: In step S1, the construction of the initial drainage network model is based on the Intelliway-SWMM software. The steps of constructing and calibrating the initial drainage network model are as follows: Based on the elevation data and underlying surface attribute data, the corresponding attribute value is assigned to each sub-catchment, and the rainfall time series of the nearest meteorological station corresponding to each sub-catchment is extracted according to the geographical scope of each sub-catchment. Input meteorological data and typical sewage discharge time series characteristic data to drive Intelliway-SWMM to run and obtain simulation results of flow, liquid level and water quality in each pipe section; The calibration data is the historical monitoring data of the flow rate, liquid level and water quality of each pipe section. The simulation results of the flow rate, liquid level and water quality of each pipe section are compared with the historical monitoring data of the flow rate, liquid level and water quality of each pipe section. If the error between the two exceeds the set threshold, the model parameters of Intelliway-SWMM are adjusted and the simulation is repeated until the error does not exceed the set threshold.

5. The method for intelligently identifying potential infiltration risk areas in a drainage network according to claim 1 is characterized by: In step S4, the steps of constructing the risk index of each pipe section are as follows: Based on the hourly liquid level difference time series data, a liquid level difference threshold is set to determine the presence of potential infiltration risk, and a risk index for the pipe section is constructed. The risk index includes the average water level difference, the proportion of time exceeding the standard, and the cumulative high head. The average water level difference calculation formula is: ， ， Where: is the average water level difference in meters, and n is the The total number of moments in the simulation period, t represents the tth moment in the simulation period, T is the total number of moments in the simulation period, is the liquid level difference at the tth moment, in meters, The liquid level difference threshold for judging the potential risk of infiltration is in meters. is the groundwater level at the tth moment, in meters, is the liquid level of the pipe section at the tth moment, in meters; The calculation formula for the time ratio exceeding the standard is: ， ， Where: is the time ratio exceeding the standard, is the indicator function; The calculation formula for cumulative super-high head is: ， ， Where: is the cumulative super-high head, in meters, is the time step in hours.

6. The intelligent hierarchical identification method for potential infiltration risk areas in a drainage network according to claim 1 is characterized in that: The step S5 includes the following sub-steps: S51. Select target surface water-groundwater coupling model parameters, rainfall boundary factors, and sewage boundary factors as key control factors. The target surface water-groundwater coupling model parameters include permeability, porosity, recharge rate, and discharge coefficient. The rainfall boundary factors include rainstorm return period, rainfall duration, and intensity-duration-frequency curve parameters. The sewage boundary factors include sewage discharge intensity per unit population and daily variation pattern of sewage discharge. S52. Setting corresponding disturbance modes for different key control factors; S53. Set disturbance intervals for different key control factors based on historical monitoring data or expert knowledge; S54. Apply perturbations to the key control factors to generate perturbation samples of the key control factors, perform physical rationality checks and parameter constraint relationship checks on the samples, and eliminate unqualified samples to obtain an initial perturbation sample set for each key control factor; S55. Sample and combine the initial disturbance sample sets of each key control factor to construct an uncertainty scenario sample set.

7. The method for intelligently identifying potential infiltration risk areas in a drainage network according to claim 6, characterized in that: The step S52 includes: The distribution fitting method is used to determine the perturbation interval and probability distribution form of the target surface water-groundwater coupling model parameters; The statistical characteristics of the rainfall boundary factor are inferred based on the historical meteorological data of the target area, and statistical modeling is carried out in combination with the historical rainstorm measured data in the target area. The rainfall process is empirically disturbed according to the rainstorm classification standard. The pollution intensity per unit population is disturbed by normal distribution or uniform distribution, and the diurnal variation pattern of pollution discharge is randomly selected from the sample set of typical daily variation curves to generate a number of representative daily variation factor sequences.

8. The method for intelligently identifying potential infiltration risk areas in a drainage network according to claim 6, characterized in that: The step S55 includes: S551. Performing Monte Carlo high-density sampling on the target surface water-groundwater coupling model parameter initial perturbation sample set to obtain an initial parameter sample set, performing normalization preprocessing on the initial parameter samples, clustering the initial parameter sample set using a clustering algorithm, extracting a number of samples of each category from the clustering results, performing denormalization on the samples, and constructing the target surface water-groundwater coupling model parameter perturbation sample set using the denormalized samples; S552. Using the Latin hypercube sampling method on the initial perturbation sample set of the rainfall boundary factor, a perturbation sample set of the rainfall boundary factor is constructed; S553. For the initial perturbation sample set of the discharge boundary factor, obtain the daily total sewage volume time series data based on the per capita sewage intensity. Combine several groups of representative daily variation factor sequences with the corresponding daily total sewage volume data, generate the daily sewage discharge time series by hourly multiplication, and construct the perturbation sample set of the discharge boundary factor. S554. The obtained target surface water-groundwater coupling model parameter perturbation sample set, rainfall boundary factor perturbation sample set, and sewage boundary factor initial perturbation sample set are combined to obtain an uncertainty scenario sample set. The sample expression in the uncertainty scenario sample set is: ， Where: is a sample in the uncertainty scenario sample set, is the i-th sample in the parameter perturbation sample set of the target surface water-groundwater coupling model, is the jth sample in the rainfall boundary factor perturbation sample set, is the kth sample in the pollution boundary factor perturbation sample set.

9. The intelligent hierarchical identification method for potential infiltration risk areas in a drainage network according to claim 5 is characterized in that: In step S6, the steps of determining the classification results of potential infiltration risk areas based on the risk indicators of each pipe section are as follows: normalizing the risk indicators and setting a comprehensive risk indicator. The calculation formula of the comprehensive risk indicator is: ， Where: is the comprehensive risk index of the i-th pipeline section under all uncertainty scenarios, 、 and To set the weight, is the normalized average water level difference of the i-th pipe section under all uncertainty scenarios, is the normalized excess time ratio of the i-th pipe section under all uncertainty scenarios, is the normalized cumulative high water head of the i-th pipe segment under all uncertainty scenarios; According to the comprehensive risk index results, the quantile method is used in combination with the threshold value set based on expert experience to determine and divide the potential infiltration risk level of each pipe section.

10. The intelligent hierarchical identification method for potential infiltration risk areas in a drainage network according to claim 9, characterized in that: The steps for determining and classifying the potential infiltration risk level of each pipe section using the quantile method combined with a threshold value set based on expert experience are as follows: Collect the comprehensive risk indicators of all pipeline sections under all uncertainty scenarios to form a risk indicator data set. Use the quantile method to divide the data set into several quantile intervals, and map each quantile interval to a predetermined risk level. Adjust the upper and lower thresholds of the quantile interval based on expert experience; Using the adjusted quantile interval, the comprehensive risk index of each pipe section is mapped to the corresponding risk level to obtain the potential infiltration risk level classification results of all pipe sections.

Images