A monitoring station layout and model online calibration method and system for urban waterlogging early warning

Abstract

This invention relates to the field of urban flood control technology and discloses a method for the deployment of monitoring stations and online calibration of models for urban flood early warning. The method includes the following steps: S1, acquiring drainage network data, underlying surface data, and field survey data of the target area, and constructing a flood process model based on these data; S2, constructing an observation simulation simulator based on the flood process model and the measurement errors of the monitoring instruments, to generate simulated observation data containing errors; S3, acquiring the rainstorm intensity formula and design rainfall pattern of the target area, and generating design rainfall datasets with different return periods. This invention, by constructing an observation simulation simulator and a parameter calibration model, can systematically analyze the quantitative relationship between the number of stations, instrument errors, and forecast effectiveness, providing a scientific basis for monitoring network planning, avoiding blind station deployment, and improving investment efficiency.

Description

Technical Field

[0001] This invention relates to the field of urban flood control technology, and in particular to a method and system for the deployment of monitoring stations and online calibration of models for urban flood early warning. Background Technology

[0002] Urban flooding is one of the major challenges facing the safe operation of cities today. Against the backdrop of global climate change, extreme rainfall events are becoming more frequent, and the insufficient pressure-bearing capacity of urban drainage systems leads to frequent flooding disasters, seriously affecting people's lives and property and social and economic stability. Therefore, developing a high-precision, high-timeliness urban flooding early warning system is of significant practical importance.

[0003] Currently, urban flooding early warning mainly relies on numerical simulation methods, such as stormwater management models (SWMM), which predict flooding risk by simulating the rainfall-runoff-pipeline confluence process. However, these models typically involve a large number of parameters (such as surface roughness, infiltration parameters, and pipe roughness), and these parameters have strong uncertainties and spatial variability, leading to significant deviations between simulation results and actual conditions, directly affecting the accuracy and reliability of early warnings.

[0004] Existing parameter calibration methods mainly include manual trial-and-error and automatic optimization methods based on observation data. Manual trial-and-error relies on experience, is time-consuming and inefficient, and struggles to guarantee calibration results. While optimization methods based on observation data (such as data assimilation and Bayesian inversion) have a clear mathematical foundation, their effectiveness heavily depends on the quantity and quality of the observation data. Currently, most cities in my country have a severe shortage of urban drainage network monitoring stations, resulting in sparse and unevenly distributed observation data, which restricts the application of such methods in practical engineering projects.

[0005] In recent years, the government has promoted the construction of new urban infrastructure and encouraged the deployment of monitoring equipment in drainage networks, providing new data support for model parameter optimization. However, a systematic methodology for scientifically deploying monitoring stations with limited investment to maximize model forecasting capabilities has yet to be established. Furthermore, how to utilize time-series observation data to achieve dynamic online calibration of model parameters is a key technical problem that urgently needs to be solved in current urban flooding early warning systems.

[0006] Therefore, there is an urgent need for a method for the deployment of monitoring stations and online calibration of models for urban flooding early warning. This method should be able to optimize the design of the monitoring network while taking into account factors such as investment constraints, instrument errors, and station distribution, and continuously improve the accuracy of model forecasts based on real-time observation data. Summary of the Invention

[0007] To address the technical problems mentioned in the background section, this invention provides a method and system for the deployment of monitoring stations and online calibration of models for urban flooding early warning.

[0008] This invention is achieved using the following technical solution: a method for deploying monitoring stations and calibrating models online for urban flooding early warning, comprising the following steps:

[0009] S1. Obtain drainage network data, underlying surface data and field survey data of the target area, and construct an urban flooding process model based on these data.

[0010] S2. Based on the waterlogging process model and the measurement errors of the monitoring instruments, an observation simulation simulator is constructed to generate simulation observation data containing errors;

[0011] S3. Obtain the rainstorm intensity formula and design rainfall pattern for the target area, and generate design rainfall datasets with different return periods;

[0012] S4. Determine the observation indicators and the number of target stations, and construct a parameter calibration model based on the ensemble Kalman filter algorithm;

[0013] S5. Using the observation simulation simulator and the parameter calibration model, analyze the quantitative relationship between the number of observation points, measurement error and the improvement of urban flood forecasting effect, and select the target number of monitoring stations from the candidate station set based on the quantitative relationship.

[0014] S6. Based on the correlation of observation data between stations, perform iterative calculations to optimize the location distribution of the selected monitoring stations and determine the final monitoring station deployment scheme.

[0015] S7. Based on the final monitoring station deployment scheme, assimilate the time-series observation data to dynamically optimize the parameters of the urban flooding process model and achieve online model calibration.

[0016] Furthermore, in step S1, the construction of the urban flooding process model specifically involves: verifying and correcting topological errors in the drainage network data, and constructing a stormwater management model based on the corrected data.

[0017] Furthermore, in step S2, the construction of the observation simulation simulator specifically involves: constructing it based on the relationship between simulated observations, model output results, observation operators, and observation errors, where the relationship is expressed as follows: ,in, For simulated observations, The output of the waterlogging process model is as follows: For the observation operator, The observation error term characterizes the measurement error of the monitoring instrument.

[0018] Furthermore, in step S3, generating the design rainfall dataset specifically includes: calculating the rainfall intensity under a set return period based on the local rainfall intensity formula for the target area; allocating the total rainfall over time according to the Chicago rainfall pattern to generate a design rainfall process line that meets the input requirements of the urban flooding process model.

[0019] Furthermore, in step S4, the construction of the parameter calibration model specifically includes: constructing a set containing multiple model parameter samples, and updating the model parameter samples based on the deviation between the observed values ​​and the simulated observation set using a localized set Kalman filter algorithm.

[0020] Furthermore, the number of samples in the set is estimated and determined based on the effective degrees of freedom of the model parameters.

[0021] Furthermore, in step S5, the analysis of quantitative relationships specifically includes: constructing multiple random disturbance trajectories in the model parameter space, applying single-factor incremental disturbances to each parameter, calculating the disturbance effect statistics of each parameter based on the changes in the model output results before and after the disturbance, and analyzing the quantitative relationship between the number of observation points, measurement error, and the improvement of urban flooding forecasting effect based on the disturbance effect statistics; wherein, the urban flooding forecasting effect is evaluated using the Nash efficiency coefficient.

[0022] Furthermore, in step S5, the candidate site set consists of all rainwater inspection wells within the target area.

[0023] Furthermore, step S6 specifically involves: under the preset constraint of the number of stations, with the optimization objective of minimizing the correlation between observation data between stations, using a genetic algorithm to iteratively search the spatial distribution of monitoring stations to obtain the optimal monitoring station deployment scheme.

[0024] This invention also proposes a monitoring station deployment and online model calibration system for urban flooding early warning, used to implement the aforementioned online calibration method, including:

[0025] The model building module is used to acquire data from the target area and build a model of the urban flooding process.

[0026] The simulator construction module is used to construct an observation simulation simulator based on the aforementioned waterlogging process model and measurement errors;

[0027] The rainfall generation module is used to generate design rainfall datasets with different return periods;

[0028] The calibration model building module is used to build parameter calibration models based on the ensemble Kalman filter algorithm;

[0029] The site number filtering module is used to analyze the quantitative relationship between the number of sites, errors, and effects, and to filter the target number of sites.

[0030] The location optimization module is used to perform iterative optimization calculations on the location distribution of monitoring stations to determine the final deployment scheme;

[0031] An online calibration module is used to assimilate time-series observation data to dynamically optimize model parameters.

[0032] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0033] This invention, by constructing an observation simulation simulator and a parameter calibration model, can systematically analyze the quantitative relationship between the number of monitoring stations, instrument errors, and forecast effects, providing a scientific basis for monitoring network planning, avoiding blind station deployment, and improving investment efficiency.

[0034] This invention is based on correlation analysis of observation data between stations and uses optimization methods such as genetic algorithms to iteratively optimize the station locations, thereby maximizing information representativeness under the constraint of a limited number of stations and enhancing the overall coverage and information complementarity of the monitoring network.

[0035] This invention is based on the ensemble Kalman filter algorithm, which can assimilate time-series observation data, update model parameters in real time, adapt to changes in rainfall processes and pipeline conditions, and continuously improve the accuracy and timeliness of urban flooding early warning.

[0036] The method of this invention fully considers the real-world factors such as investment constraints, instrument errors, and data gaps in actual engineering projects. It adopts robust algorithms such as localized ensemble Kalman filtering and is suitable for optimizing parameters of high-dimensional, nonlinear, and highly uncertain urban flooding models. It has high computational efficiency and engineering feasibility.

[0037] This invention forms a closed-loop technical process from data preparation, model building, site optimization to online calibration, supporting full-cycle management from planning to operation, and helping to improve the intelligence and resilience of urban drainage systems. Attached Figure Description

[0038] Figure 1 This is a flowchart of the online calibration method proposed in this embodiment of the invention;

[0039] Figure 2 This is a diagram showing the result of a local SWMM model constructed using the collected data in an embodiment of the present invention.

[0040] Figure 3 This is a graph showing the relationship between the calculated observation error, the number of observation points, and the improvement effect of urban flooding early warning in an embodiment of the present invention.

[0041] Figure 4 This is a correlation diagram between inspection well information in an embodiment of the present invention;

[0042] Figure 5This is a diagram of the selected monitoring site scheme in an embodiment of the present invention;

[0043] Figure 6 This is the assimilation process for monitoring sites in an embodiment of the present invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0045] This application provides a method for the deployment of monitoring stations and online model calibration for urban flooding early warning. Its core innovation lies in constructing a complete technical closed loop of "optimized planning first, then precise operation." This method quantitatively correlates limited monitoring investment with the improvement of model forecasting capabilities, realizing full-process management from scientific station deployment to dynamic calibration.

[0046] Its core working framework and main steps can be clearly summarized into the following three progressive stages:

[0047] Phase 1: Digital Simulation and Scientific Planning Phase.

[0048] Before the construction of physical sites, this stage involves optimizing the site deployment plan within the digital twin environment to address the core planning issues of "deployment quantity and spatial location".

[0049] 1. Building the foundation for digital simulation:

[0050] Step S1: Integrate underlying surface data such as pipe network and topography of the target area to construct a high-fidelity urban flooding process mechanism model (such as the SWMM model) as the basic simulation platform for all subsequent analyses.

[0051] Step S2: Construct an observation simulation simulator. This simulator, based on the aforementioned mechanistic model and superimposed with the measurement error characteristics of the selected monitoring equipment, is used to generate a simulated data sequence containing errors that approximates future real-world observations. It establishes a mapping relationship between the mechanistic model output and the simulated observation data.

[0052] Step S3: Generate a dataset of design rainfall scenarios covering different intensities (recurrence periods) and patterns. These rainfall scenarios constitute a series of simulation test conditions that drive the model and test different station placement schemes.

[0053] 2. Establish a framework for parameter optimization and performance evaluation:

[0054] Step S4: Construct a parameter calibration model based on ensemble Kalman filtering (EnKF). This algorithm constitutes a core optimizer capable of automatically and efficiently inverting and calibrating uncertain parameters of the model using (simulated or real) observation data.

[0055] 3. Collaboratively optimize the number and spatial layout of sites:

[0056] Step S5 (Determining the Optimal Number of Sites): Utilizing the established basic simulation platform (S1), a series of experimental conditions (S3), an observation simulator (S2), and a parameter optimizer (S4), conduct systematic simulation experiments. Analyze the quantitative mapping relationship between the number of sites, instrument measurement errors, and model prediction accuracy improvement indicators (such as the improvement in the Nash efficiency coefficient NSE). Figure 3 As shown in the figure, an "investment-benefit" curve is formed. Therefore, the optimal number of target sites can be scientifically determined based on investment constraints and expected performance goals.

[0057] Step S6 (Determining the Optimal Spatial Location): After determining the number of stations, the optimization objective is to "maximize the representativeness of spatial information and minimize the redundancy of observation data between stations." Based on simulation data, the correlation coefficient matrix of observation sequences between all candidate locations is calculated (e.g., ...). Figure 4 The optimal site layout scheme is determined through iterative search using optimization methods such as genetic algorithms. Figure 5 Steps S5 and S6 together achieve collaborative global optimization of the monitoring network's "size" and "layout" under given constraints.

[0058] Phase Two: Physical Construction and Deployment Phase

[0059] This stage is the project implementation phase, which involves installing and networking monitoring equipment in the actual urban drainage network according to the final deployment plan optimized in the first stage, forming a physical monitoring system.

[0060] Phase 3: Online Operation and Dynamic Calibration Phase

[0061] This stage begins after the monitoring system is built and put into operation, and then enters the operational and continuous improvement phase.

[0062] Step S7: The time-series observation data stream collected in real time by the monitoring network is continuously input into and drives the EnKF parameter optimization framework built in the first stage. This framework performs a continuous "forecast-update" assimilation loop, dynamically and automatically correcting the key states and parameters of the urban flooding model using real-time observations, keeping the model state synchronized with the real drainage system, thereby achieving online, adaptive calibration and continuous performance optimization of the urban flooding early warning model (process as follows). Figure 6 (As shown).

[0063] In summary, the method proposed in this application involves: first, constructing a complete simulation evaluation environment in digital space, including mechanistic models, error characteristics, and optimization algorithms; second, quantitatively optimizing and outputting the optimal monitoring network deployment scheme through systematic numerical experiments; and finally, materializing the optimized scheme and processing real data streams using the same assimilation algorithm framework to achieve online dynamic updates of the model. This process forms a complete technical closed loop from "simulation design and virtual evaluation" to "physical construction and deployment" and then to "data-driven and online optimization," providing a systematic solution for the scientific planning and efficient operation of urban flood monitoring and early warning systems.

[0064] In more detail, Figure 1 This is an overall flowchart of the monitoring site deployment and online assimilation method in one embodiment of the present invention. Figure 1 As shown, and in combination Figures 2 to 6 The method for deploying monitoring stations and calibrating models online for urban flooding early warning provided in this embodiment mainly includes the following steps:

[0065] S1. Obtain drainage network data, underlying surface data, and field survey data for the target area, and construct an urban flooding process model based on these data.

[0066] First, clearly define the target area (e.g., a drainage subsystem in City L) and the equipment investment budget. Collect multi-source data for this area, specifically including: drainage network data (such as pipe location, burial depth, pipe diameter, location and elevation of storm drains and outlets, usually derived from DWG format design drawings, which need to be converted to GIS formats such as Shapefile), underlying surface data (such as land use types, including buildings, roads, green spaces, and water bodies), and field survey data of the pipeline network (such as pipeline flow direction and slope confirmation data, as well as closed-circuit television (CCTV) and endoscopic (QV) inspection data of underground pipelines).

[0067] When constructing a flooding process model, it is necessary to verify and correct topological errors in the original data (such as conflicting pipe flow directions, missing pipe segments, or closed loops). Using Python's SWMMIO library, the processed data is automatically converted into an input file (INP format) for a Storm Water Management Model (SWMM), thereby establishing a flooding process model for the study area. An example of the model construction results is shown below. Figure 2 As shown.

[0068] Table 1: Rainwater Well (Node) Data Table

[0069] Node Name Elevation Maximum well depth Initial water level height 35YS0509 37.21m 5.77m 0 35YS0511 37.03m 5.64m 0 35YS0514 36.68m 5.84m 0

[0070] Table 2: Pipeline (Pipe Segment) Data Table

[0071] Pipeline section number Import node Export node length Pipe bottom burial depth shape Pipe diameter Pipe Material YSGD051 35YS0509 35YS0511 31m 32.58m cylinder 1200 mm concrete pipe YSGD052 35YS0511 35YS0514 27m 32.46m cylinder 1200 mm concrete pipe

[0072] S2. Based on the waterlogging process model and the measurement errors of the monitoring instruments, an observation simulation simulator is constructed to generate simulated observation data containing errors.

[0073] The observation simulation simulator aims to simulate observational data that may be acquired by real monitoring stations in the future, including instrumental errors. Its core is to combine the output of the mechanistic model with measurement errors. The specific construction method is as follows:

[0074] Extracting the SWMM model from step S1 with given parameters The original output results (Such as water levels in each inspection well and flow rates in pipelines). Simultaneously, the measurement error of the selected monitoring instruments (such as level gauges and flow meters) should be determined based on their technical specifications.

[0075] The observation simulation simulator is built based on the following mathematical model:

[0076]

[0077] in, These are simulated observations; This is the original output of the SWMM model; This is the observation operator, used to extract the physical quantities corresponding to the future monitoring station locations from the full-field output of the model. Here, it is the simulation result directly extracted from the model, such as liquid level height and flow rate plus error. This is the observation error term, used to characterize the instrument's measurement error. Multiple sets can be set to evaluate different investment options. Scenarios, such as setting the water level observation error to 1% or 5% and the flow rate observation error to 10% or 15%, are used to simulate data under different instrument accuracies.

[0078] S3. Obtain the storm intensity formula and design rainfall pattern for the target area, and generate design rainfall datasets with different return periods. The design rainfall datasets include 9 scenarios with return periods of 1, 2, 3, 5, 10, 20, 30, 50, and 100 years, lasting [60, 90, 120, 180, 240] minutes; the peak rainfall coefficient can be adjusted between 0.3 and 0.5 according to regional characteristics.

[0079] Obtain the local heavy rainfall intensity formula published by the meteorological department of the target area. Taking City L as an example, the formula is as follows:

[0080] Formula for long-duration rainstorm intensity:

[0081]

[0082] Formula for short-duration rainstorm intensity:

[0083]

[0084] in, The rainfall recurrence interval (in years) is the annual return period. The duration of rainfall (in minutes).

[0085] When generating the design rainfall dataset, a series of return periods (e.g., 0.5 years, 1 year, 3 years, 5 years, 10 years, etc.) and total durations are set according to the evaluation requirements. The total design rainfall for each scenario is calculated using the above formula. Subsequently, the Chicago rainfall pattern (unimodal) is used to allocate the total rainfall over time; in this embodiment, the peak rainfall coefficient is set to 0.414. Finally, an external rainfall time series file that meets the input requirements of the SWMM model is generated, constituting the design rainfall dataset used to drive the model and evaluate its effectiveness.

[0086] Taking a return period of 2 years, a precipitation history of 90 minutes, and a peak rainfall coefficient of 0.414 as an example, the following calculations were performed:

[0087] The rainfall intensity q = 98.89 L / (s·hm²), the total rainfall H = 53.29 mm, the design average intensity = 0.592 mm / min, and the design total rainfall H = 53.29 mm. A total of 45 sets of data were generated.

[0088] S4. Determine the observation indicators and the number of target stations, and construct a parameter calibration model based on the ensemble Kalman filter algorithm.

[0089] Based on the current status of the study area, observation indicators (water level / flow rate), and investment budget, the number of target monitoring stations to be constructed is estimated. Based on this, a parameter calibration model based on localized ensemble Kalman filtering is constructed. Geometric and spatial characteristic parameters include: area (calculated from geometry), catchment width (calculated from geometry), slope (0.1–10%), impermeability (0–100%), presence of sinkholes in impermeable areas (0–100%), sinkhole depth in impermeable areas (1–10 mm), sinkhole depth in permeable areas (3–2 mm), and the maximum infiltration rate (25–100 mm / h), minimum infiltration rate (2–15 mm / h), and attenuation coefficient (2–7) from the Horton model, as well as the Manning roughness of the pipeline (0.011–0.030).

[0090] 1. Collection generation: Constructing collections containing... This is a set of sample parameters for an SWMM model. The parameters include key uncertainties such as the Manning roughness coefficient and infiltration parameters. Each sample value is generated by random perturbation within its prior range. It is generally set to 0.2-3 times the prior value.

[0091] 2. Determining the number of sets: The number of sets Based on the effective degrees of freedom of the model parameters Estimation is performed. According to the theory of high-dimensional covariance sample estimation, it is generally required that... . The prior distribution of parameters can be estimated using principal component analysis (PCA), which is generally equal to the dimension of a single set of parameters.

[0092] 3. Core model update algorithm:

[0093] Forecasting steps: Run the SWMM model corresponding to each parameter sample to obtain the state (including model variables and parameters) forecast set. .

[0094] Covariance Calculation and Localization: Calculating the Forecast Error Covariance Matrix :

[0095]

[0096] To suppress spurious correlations at long distances, a distance function (such as the Gaspari-Cohn function) is used to generate a localized weight matrix. The localized covariance is obtained through the Schur product (element-by-element multiplication). :

[0097]

[0098] Analysis and update steps: When actual observation data is obtained Then, calculate the Kalman gain matrix. And update the state and parameter set:

[0099]

[0100]

[0101] in, The observation error covariance matrix, For observation operators.

[0102] S5. Using the observation simulation simulator and the parameter calibration model, analyze the quantitative relationship between the number of observation points, measurement error and the improvement of urban flood forecasting effect, and select the target number of monitoring stations from the candidate station set based on the quantitative relationship.

[0103] This step aims to quantitatively pre-evaluate the effectiveness of different site selection schemes before investment and construction.

[0104] 1. Determine the candidate site set: Select all rainwater inspection wells within the target area as the candidate site set.

[0105] 2. Analyze the quantitative relationship:

[0106] Parameter perturbation: Multiple random perturbation trajectories are constructed within the SWMM key parameter value space. A single-factor incremental perturbation is then applied sequentially to each parameter. .

[0107] Simulation and Effect Calculation: For each disturbance, the observation simulation simulator built in step S2 is used to run the model before and after the disturbance, generating simulation observation data covering all candidate sites. The flow simulation results of the model at key sections (such as the system outlet) before and after the disturbance are calculated.

[0108] Effectiveness quantification: using the change in the Nash efficiency coefficient (NSE). As a "perturbation effect statistic", the NSE is calculated using the following formula:

[0109]

[0110] in, and They are respectively Simulated values ​​and (simulated) observed values ​​at time points. This represents the mean of the observed sequence.

[0111] Relationship establishment and screening: Considering the impact of disturbances in different parameters, the system analyzes the number of observation points and the level of instrument measurement error. ) and the potential for improving the overall forecasting effect of the system ( The quantitative relationship between the expected value and the three. Figure 3 As shown, an "investment-accuracy" curve can be plotted. Ultimately, based on the investment budget and expected accuracy target, the optimal number of target sites is determined from this quantitative relationship curve.

[0112] S6. Based on the correlation of observation data between stations, perform iterative calculations to optimize the location distribution of the selected monitoring stations and determine the final monitoring station deployment scheme.

[0113] After determining the number of sites, optimize their spatial distribution.

[0114] 1. Correlation Analysis: Based on the multiple design rainfall events generated in step S3, the model is run to simulate the time-series data (e.g., water level) of all candidate inspection wells. The Pearson correlation coefficient matrix of the observation sequences between all well pairs is calculated, such as... Figure 4 As shown.

[0115] 2. Optimization Solution: With the optimization objective of minimizing the average correlation of observation data among the selected stations, and under the constraint of a fixed number of stations, the problem is modeled as a combinatorial optimization problem. A genetic algorithm is used for iterative solution. Algorithm parameters (such as population size and crossover / mutation probability) are set according to the problem size. After iterative search, the optimal spatial information representation of the monitoring station deployment scheme is finally obtained, with an example of its location shown below. Figure 5 As shown. The genetic algorithm is set with a population size of 100, a maximum number of iterations of 200, a crossover probability of 0.8, and a mutation probability of 0.1; the objective function is the reciprocal of the average of the upper triangular elements of the Pearson correlation coefficient matrix between the selected stations; at the same time, the distance between any two stations is constrained to be no less than 500 meters.

[0116] S7. Based on the final monitoring station deployment scheme, assimilate the time-series observation data to dynamically optimize the parameters of the urban flooding process model and achieve online model calibration.

[0117] Based on the monitoring network identified and established in step S6, continuous online calibration of the model is implemented.

[0118] 1. The system periodically receives actual time-series observation data from each monitoring station. ).

[0119] 2. Input the observation data into the LEnKF parameter calibration model constructed in step S4.

[0120] 3. The model automatically executes a forecast-update cycle: first, a short-term forecast is made, and then the gain is calculated using the latest observation data according to the formula. And update the model parameter set .

[0121] 4. The updated mean of the parameter set is used as the current optimal parameter and fed back to the SWMM early warning model in real time. This process is repeated cyclically, thereby enabling dynamic, online, and adaptive calibration of model parameters as data accumulates, ensuring continuous improvement in the accuracy of urban flooding early warning. The process is as follows: Figure 6 As shown.

[0122] The present invention also provides a system for implementing the above method, the system comprising:

[0123] Model building module: Used to execute step S1, realizing data integration, topology verification and SWMM model building.

[0124] Simulator building module: Used to execute step S2 to build the observation simulation simulator.

[0125] Rainfall generation module: Used to execute step S3 to generate the designed rainfall dataset.

[0126] Calibration model construction module: used to execute step S4 to build and run the parameter calibration model based on LEnKF.

[0127] Site Quantity Filtering Module: Used to execute step S5, enabling quantitative analysis and filtering of site quantity and effectiveness.

[0128] Location optimization module: Used to execute step S6 to optimize the deployment of monitoring station locations.

[0129] Online calibration module: used to perform step S7, realizing online assimilation and calibration of model parameters.

[0130] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0131] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0132] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and inventive constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0133] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0134] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0135] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0136] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for deploying monitoring stations and calibrating models online for urban flooding early warning, characterized in that, Includes the following steps: S1. Obtain drainage network data, underlying surface data and field survey data of the target area, and construct an urban flooding process model based on these data. S2. Based on the waterlogging process model and the measurement errors of the monitoring instruments, an observation simulation simulator is constructed to generate simulation observation data containing errors; S3. Obtain the rainstorm intensity formula and design rainfall pattern for the target area, and generate design rainfall datasets with different return periods; S4. Determine the observation indicators and the number of target stations, and construct a parameter calibration model based on the ensemble Kalman filter algorithm; S5. Using the observation simulation simulator and the parameter calibration model, analyze the quantitative relationship between the number of observation points, measurement error and the improvement of urban flood forecasting effect, and select the target number of monitoring stations from the candidate station set based on the quantitative relationship. S6. Based on the correlation of observation data between stations, perform iterative calculations to optimize the location distribution of the selected monitoring stations and determine the final monitoring station deployment scheme. S7. Based on the final monitoring station deployment scheme, assimilate the time-series observation data to dynamically optimize the parameters of the urban flooding process model and achieve online model calibration.

2. The method according to claim 1, characterized in that, In step S1, the construction of the urban flooding process model specifically involves: verifying and correcting topological errors in the drainage network data, and constructing a stormwater management model based on the corrected data.

3. The method according to claim 1, characterized in that, In step S2, constructing the observation simulation simulator specifically involves: constructing it based on the relationship between simulated observations, model output results, observation operators, and observation errors, where the relationship is expressed as follows: ,in, For simulated observations, The output of the waterlogging process model is as follows: For the observation operator, The observation error term characterizes the measurement error of the monitoring instrument.

4. The method according to claim 1, characterized in that, In step S3, generating the design rainfall dataset specifically includes: calculating the rainfall intensity under a set return period based on the local rainfall intensity formula for the target area; allocating the total rainfall over time according to the Chicago rainfall pattern to generate a design rainfall process line that meets the input requirements of the urban flooding process model.

5. The method according to claim 1, characterized in that, In step S4, the construction of the parameter calibration model specifically includes: constructing a set containing multiple model parameter samples, and updating the model parameter samples based on the deviation between the observed values ​​and the simulated observation set using a localized ensemble Kalman filter algorithm.

6. The method according to claim 5, characterized in that, The number of samples in the set is estimated and determined based on the effective degrees of freedom of the model parameters.

7. The method according to claim 1, characterized in that, In step S5, the quantitative analysis specifically includes: constructing multiple random disturbance trajectories in the model parameter space, applying single-factor incremental disturbances to each parameter, calculating the disturbance effect statistics of each parameter based on the changes in the model output results before and after the disturbance, and analyzing the quantitative relationship between the number of observation points, measurement error, and the improvement of urban flooding forecasting effect based on the disturbance effect statistics; wherein, the urban flooding forecasting effect is evaluated using the Nash efficiency coefficient.

8. The method according to claim 1 or 7, characterized in that, In step S5, the candidate site set consists of all rainwater inspection wells within the target area.

9. The method according to claim 1, characterized in that, Step S6 specifically involves: under the preset constraint of the number of stations, with the optimization objective of minimizing the correlation between observation data between stations, using a genetic algorithm to iteratively search the spatial distribution of monitoring stations to obtain the optimal monitoring station layout scheme.

10. A monitoring station deployment and online model calibration system for urban flooding early warning, used to implement the method described in any one of claims 1-9, characterized in that, include: The model building module is used to acquire data from the target area and build a model of the urban flooding process. The simulator construction module is used to construct an observation simulation simulator based on the aforementioned waterlogging process model and measurement errors; The rainfall generation module is used to generate design rainfall datasets with different return periods; The calibration model building module is used to build parameter calibration models based on the ensemble Kalman filter algorithm; The site number filtering module is used to analyze the quantitative relationship between the number of sites, errors, and effects, and to filter the target number of sites. The location optimization module is used to perform iterative optimization calculations on the location distribution of monitoring stations to determine the final deployment scheme; An online calibration module is used to assimilate time-series observation data to dynamically optimize model parameters.

Images