Urban-flooding-oriented optimization method for real-time control rules for river network

Abstract

An urban-flooding-oriented optimization method for real-time control rules for a river network. The method comprises: S1, collecting data of a river network; S2, constructing an urban hydrological-hydrodynamic coupling model; S3, on the basis of a rainstorm intensity formula, simulating and calculating flooding processes in different design rainstorm scenarios, and comprehensively identifying drainage outlets within a study area that are susceptible to backwater effects from water levels in the river network; S4, using a clustering algorithm to perform spatial clustering on the backwater-affected drainage outlets, so as to define regulation and control zones, further performing feature clustering on the backwater-affected drainage outlets on the basis of spatial clustering, and outputting a cluster center with the greatest influence and using same as a river network regulation and control point; S5, using the river network regulation and control point as a regulation and control reference object, and formulating optimized gate and pump control rules on the basis of warning water level data and ecological water level data of river channels; and S6, inputting the optimized gate and pump control rules into the urban hydrological-hydrodynamic coupling model, using a rainstorm event to evaluate the performance of the method, and quantitatively analyzing the response of flooding to the gate and pump control rules. The method enables the regulation and control of water levels in a river network during a rainstorm, thereby mitigating the risk of flooding.

Description

A method for optimizing real-time river network control rules for urban flooding Technical Field

[0001] This invention relates to the field of water resources management and flood control technology, and in particular to a method for optimizing real-time river network control rules for urban flooding. Background Technology

[0002] River networks, as a crucial component of urban drainage systems, undertake multiple tasks, including flood control, drainage, and water resource regulation. Reasonable real-time control rules for water projects are essential for the efficient operation of river networks, preventing river overflows and urban flooding. Especially during periods of heavy rainfall and tidal backwater, they can effectively lower inland river levels and mitigate flooding. Traditional real-time control technologies for water projects often use upstream and downstream sections as reference points for regulation, resulting in a lack of responsiveness from the urban river network to real-time control rules. While maintaining target section water levels within a safe range, they cannot guarantee that urban drainage outlets will not be affected by backwater from the river network. Therefore, traditional real-time control technologies for water projects struggle to achieve precise regulation of urban river networks, hindering their scientific management. Summary of the Invention

[0003] The purpose of this invention is to overcome the above-mentioned shortcomings and provide a method for optimizing real-time control rules for urban river networks in the context of urban flooding. This method solves the problem that traditional real-time control technology for water conservancy projects is difficult to achieve precise regulation of urban river networks, which is not conducive to the scientific management of urban river networks. It regulates the water level of river networks during rainstorms and thus alleviates the risk of flooding.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for optimizing real-time river network control rules for urban flooding, comprising the following steps:

[0005] S1, collect river network data;

[0006] S2. Based on detailed spatial division and different surface runoff characteristics, an urban hydrological model is constructed. A river network hydrodynamic model is constructed considering the hydraulic conditions of pumping stations, gates, weirs, reservoirs and lakes. A pipe network hydrodynamic model is constructed considering the hydraulic conditions of manholes and outlet facilities. The above models are coupled through the process of water exchange at cross-sectional nodes to construct an urban hydrological and hydrodynamic coupled model.

[0007] S3, based on the rainstorm intensity formula, simulates and calculates the flood process under different design rainstorm scenarios, and comprehensively identifies drainage outlets in the study area that are susceptible to backwater from the river network.

[0008] S4. A clustering algorithm is used to spatially cluster and divide the top drainage outlet into control zones. Based on the spatial clustering, the top drainage outlet is further clustered by feature, and the cluster center with the greatest influence is output as the river network control point.

[0009] S5 uses river network control points as the control reference object and compiles and optimizes gate and pump control rules based on river warning water level and ecological water level data;

[0010] S6 will input the optimized gate pump control rules into the urban hydrological and hydrodynamic coupling model, and use the performance of the rainstorm event assessment method to quantitatively analyze the response of floods to the gate pump control rules.

[0011] Preferably, the urban hydrological and hydrodynamic coupling model in S2 is constructed based on Infoworks ICM software, and the specific construction process is as follows:

[0012] S21, the runoff generation calculation for impermeable surfaces of buildings and roads uses the fixed runoff coefficient model in InfoWorks ICM, as shown in the following formula: R n =CR P ;

[0013] Where R n Net rainfall, mm; C is runoff coefficient; P is return period, years; R P The rainfall amount for the corresponding return period is in mm;

[0014] S22, The calculation of surface runoff generation in permeable green spaces uses the Horton infiltration model, as shown in the following formula: f t =f c +(f0-f c )e -kt ;

[0015] Where f t f is the infiltration rate at time t, in mm / h; c The stable infiltration rate is measured in mm / h; f0 is the initial infiltration rate, measured in mm / h; k is the reduction coefficient, with dimensions in h. -1 t represents the infiltration time, in hours.

[0016] S23, simulating urban surface runoff, utilizes the nonlinear reservoir model in the stormwater and flood management module, and is solved using a combination of the continuity equation and Manning's formula, as shown below:

[0017] Manning's formula:

[0018] Where Q is the flow rate, m 3 / s; W is the overflow width of the sub-catchment area, m; n is the Manning coefficient; d is the water depth, m; d p The maximum water storage depth is given by S in meters; S is the average slope of the sub-catchment area.

[0019] Continuity equation:

[0020] Where V is the surface water volume, in m 3 t is time, s; A1 is the surface area, m.2 i* represents net rainfall, mm / s;

[0021] S24, using the hydrodynamic module in Infoworks ICM, calculates the river network confluence by using dynamic waves to fully solve the Saint-Venant equations, as shown in the following equation:

[0022] Continuity equation:

[0023] Momentum equation:

[0024] Where A2 is the cross-sectional area, in m 2 Q represents flow rate, m 3 / s; g is the acceleration due to gravity, m / s² 2 θ is the angle between the pipeline centerline and the horizontal line, in degrees; K is the water conveyance rate; h is the water depth, in meters; t is the time, in seconds; S0 is the slope of the canal bottom.

[0025] S25 uses a cross-sectional node water exchange method to couple surface runoff with river network flow.

[0026] Preferably, in step S25, the formula for coupling surface runoff and river network flow using cross-sectional node water exchange is as follows:

[0027] in It is the flow rate at the i-th river network cross-section at time t. It is the flow rate at the i-th river network cross-section at time t-1. It is the surface runoff corresponding to the i-th river network cross-section at time t.

[0028] Preferably, the formula for the rainstorm intensity in S3 is:

[0029] Where q is the average rainfall intensity (mm / min), P is the return period (a), and t is the rainfall duration (min); the Chicago rainfall pattern is used to calculate the rainfall process, with the design peak rainfall coefficient r = 0.39, the rainfall process time interval set to 1 min, and the rainfall duration set to 120 min.

[0030] Preferably, the AGNES algorithm is used for spatial clustering of the top drain outlet in step S4. This is a bottom-up agglomerative hierarchical clustering algorithm, and the calculation process is as follows:

[0031] S41, Initialization: Each sample point forms its own cluster; if the population has m samples, then there are m clusters in the initial state;

[0032] S42, Calculate the distance matrix: Initially, use the selected distance metric, choosing Euclidean distance to calculate the distance matrix between all clusters;

[0033] S43, Merge the nearest clusters: Find the two closest clusters, merge them into a new cluster, and update the distance matrix to reflect the distances between the new cluster and other clusters; the merging process is based on the chosen linking method, selecting Ward's method to minimize the sum of squares within the merged cluster:

[0034] Where C i and C j For two clusters to be merged; |C i | and |C j | are clusters C i and C j The number of data points contained; μ i and μ j They are cluster C i and C j The mean vector; d(μ) i ,μ j ) represents the distance between the mean vectors of the two clusters; d ward (C i C j ) represents the "merging cost" or "distance" between two clusters calculated based on the Ward method;

[0035] S44, Repeat step S43 until all sample points are merged into one cluster or the predetermined number of clusters is reached;

[0036] S45, Generate a hierarchical tree: Use a tree diagram to represent the clustering process, merging one node from the corresponding tree each time.

[0037] Preferably, in step S42, the Euclidean distance between two n-dimensional vectors is calculated as follows:

[0038] Where d is the Euclidean distance; x 1i and x 2i They are any two n-dimensional vectors.

[0039] Preferably, the K-Medoids algorithm is used for feature clustering of the top drain outlet in S4. This is a partitioning clustering method that directly divides data points into different clusters. The calculation process is as follows:

[0040] S46, Initialization: Randomly select k points from the total m sample points as the initial cluster centers;

[0041] S47, Cluster Assignment: Assign the remaining mk points to the clusters containing their nearest cluster centers according to the minimum distance principle to achieve initial clustering; the distance between each sample point is calculated using Euclidean distance.

[0042] S48, Update cluster centers: Calculate the distance from each point in the cluster to all other points except the cluster center, take the point with the minimum distance as the new cluster center, and recalculate the clustering results;

[0043] S49. Repeat processes S47 and S48 until all cluster centers no longer change or the set maximum number of iterations has been reached, to obtain the final clustering result.

[0044] Preferably, the basis for writing the real-time control rules for water projects in S5 is the warning water level and ecological water level data of the river, as follows: Z(i) min =H(i) 生态水位 Z(i) max =H(i) 警戒水位 B(j) min =H(j) 警戒水位 B(j) max =H(j) 警戒水位 +Δh(j);

[0045] Where Z(i) max Z(i) represents the upper limit water level controlled by the i-th gate. min B(j) is the lower limit water level controlled by the i-th gate; max B(j) represents the upper limit water level regulated by the j-th pumping station. min H(i) is the lower limit water level regulated by the j-th pumping station. 警戒水位 H(i) 生态水位 H(j) represents the warning water level and ecological water level at the control reference section of the i-th gate. 警戒水位 Δh(j) is the warning water level of the control reference section of the j-th pumping station; Δh(j) is the control threshold of the control reference section of the j-th pumping station.

[0046] Preferably, the real-time control rules for water engineering in the urban hydrological and hydrodynamic coupling model input in S6 are expressed as follows:

[0047] S61, the gate control rule is: If NODE A <Z(i) min ; THEN ORIFICE A=0; If NODE A>Z(i) max THEN ORIFICE A = 1.0;

[0048] Where NODE A is the control reference section ID of the gate; 0.0 indicates that the gate is closed, and 1.0 indicates that the gate is fully open;

[0049] S62, Pumping Station Control Rules: If NODE A <B(j) min ; THEN PUMP A STATUS=OFF; If NODE A>B(j)max ; THEN PUMP A STATUS=ON;

[0050] Where NODE A is the control reference section ID of the pumping station; OFF indicates that the pumping station is closed, and ON indicates that the pumping station is open;

[0051] S63, using the performance of the rainstorm event assessment method, quantitatively analyzes the indicators of the response of floods to the gate pump control rules, including the number of overflows from rainwater wells and the water level process at key river sections.

[0052] Preferably, in S1, the collected river network data includes: the river network topology, cross-sectional data, underlying surface, gate and pumping station data required for constructing the river network system. The gate and pumping station data includes design parameters and distribution locations, measured rainfall data, and measured river network water level and flow data corresponding to the measured rainfall data.

[0053] Beneficial effects of this invention:

[0054] 1. This invention addresses the problem of relatively slow response of urban internal river networks to gate and pump control rules. By constructing an urban hydrological and hydrodynamic model, it identifies backwater outlets in different design storm areas. Based on a clustering algorithm, it clusters the characteristics of the backwater outlets to determine river network control points. These control points are then used as new control reference objects to optimize the real-time control rules of the river network. This ensures that urban internal drainage outlets are not affected by backwater from the river network water level, improves the response speed of the internal river network to real-time control rules, and enhances the efficiency and accuracy of existing river network water engineering control. It ensures flood control safety while also considering ecological and environmental needs, and possesses good real-time performance and applicability.

[0055] 2. This invention solves the problem that traditional real-time control technology for water conservancy projects is difficult to achieve precise regulation of urban river networks, which is not conducive to the scientific management of urban river networks. It can regulate the water level of river networks during rainstorms and thus alleviate the risk of flooding. Attached Figure Description

[0056] Figure 1 is a schematic diagram of the study area in Cangshan District, Fuzhou City;

[0057] Figure 2 is a statistical chart of the number of top support drainage outlets;

[0058] Figure 3 shows the distribution of river network control points;

[0059] Figure 4 shows the rainstorm and tidal process of Typhoon Soudelor;

[0060] Figure 5 is a comparison of the water head process at the river network control point between the original gate pump rules and the optimized gate pump rules.

[0061] Figure 6 shows a comparison of simulated and observed water level processes for samples from rainwater wells and river cross-sections. Detailed Implementation

[0062] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0063] Example 1:

[0064] The area of ​​a certain region in Cangshan District, Fuzhou City, Fujian Province is 39.17 km². 2 Rainfall is concentrated during the main flood season from April to October. The region has a rich river network with 13 waterways totaling 47.69 km in length. Surrounded by tributaries of the Minjiang River on its north and south sides, it is a bidirectional tidal region, where river levels are easily affected by rising tides. Fuzhou City has been hit by typhoons multiple times, including but not limited to "Haitang," "Longwang," and "Megi," leading to frequent flooding in the study area due to typhoon weather and heavy rainfall.

[0065] A method for optimizing real-time river network control rules for urban flooding, with the following steps for inputting data:

[0066] S1: Collect river network data, including the river network topology required for constructing the river network system, data on 175 river segments, data on 7 sluice gates (including the Changpu Sluice Gate, Shengli Sluice Gate, and Jiangbian Sluice Gate located on the north side, all with a gate bottom elevation of 2m, and the Yangqi Control Gate, Jimuyu Sluice Gate, Wufeng Sluice Gate, and Yixu Sluice Gate located on the south side, with gate bottom elevations of 2m, 1.8m, 2m, and 1.5m respectively), and 4 pumping stations (including the Changpu Pumping Station and Shengli Pumping Station located on the north side, with a scale of 6m respectively). 3 / s、16m 3 / s, and the Yangqi Pumping Station and Yixu Pumping Station located to the south, both with a scale of 8m 3 The measured total duration of the typhoon and rainstorm event "Soudelor" was 3360 minutes, with a rainfall of 408 mm. This step S1 was completed by computer.

[0067] S2: Based on detailed spatial division and different surface runoff characteristics, an urban hydrological model is constructed. A river network hydrodynamic model is constructed considering the hydraulic conditions of facilities such as pumping stations, sluice gates, weirs, reservoirs, and lakes. A pipe network hydrodynamic model is constructed considering the hydraulic conditions of facilities such as inspection wells and outlets. These models are then coupled using Infoworks ICM software to construct a coupled urban hydrological and hydrodynamic model through the process of water exchange at cross-sectional nodes. The regional hydrological model consists of multiple sub-catchments, each with different parameter settings. Therefore, the following only shows the parameter value range or illustrates a specific sub-catchment and river cross-section. This step S2 is completed by computer. The specific construction process is as follows:

[0068] The calculation process of the hydrological model is explained using the sub-catchment area Y3501540059 as an example.

[0069] S21. The runoff generation calculation for impermeable surfaces such as buildings and roads uses the fixed runoff coefficient model in InfoWorks ICM, as shown in the following formula: R n =CR P

[0070] Where R n Net rainfall, mm; P is the return period, years, which is referred to below as 9 return periods: 1a, 2a, 3a, 5a, 10a, 20a, 30a, 50a, and 100a; C is the runoff coefficient, empirically taken as 0.95; R P The value represents the rainfall amount for the corresponding return period, in mm; the net rainfall over 2 hours for different return periods is shown in Table 1.

[0071] Table 1. Net rainfall over 2 hours at different rainfall return periods.

[0072] S22, The calculation of surface runoff generation in permeable green spaces uses the Horton infiltration model, as shown in the following formula: f t =f c +(f0-f c )e -kt

[0073] Where f t f is the infiltration rate at time t, in mm / h; c To stabilize the infiltration rate, the empirical range is 2.5–5.3 mm / h; the initial infiltration rate f0 is empirically ranged from 55–76 mm / h; k is the reduction coefficient, with dimensions in h. -1 Based on experience, the value is taken as 3.79–5.29; t is the infiltration time, in hours. Taking a 2-hour, 20-year return period design rainfall as an example, the infiltration rate at t = 1 hour is 2.5 + (76 - 2.5)e -3.8×1 = 4.18 mm / h.

[0074] S23 simulates urban surface runoff by calling the nonlinear reservoir model in the stormwater and flood management module, and solves it using a combination of the continuity equation and Manning's formula.

[0075] Manning's formula:

[0076] Where Q is the outflow rate, m 3 / s; W is the overflow width of the sub-catchment area, with a value of 34.6m; n is the Manning coefficient of the surface, with a value of 0.03; d is the water depth, with an initial value of 0.05m; d p The maximum water storage depth is 5m; S is the average slope of the sub-catchment area, with a value of 0.003.

[0077] Continuity equation:

[0078] Where V is the surface water volume, in m 3 t represents time, in seconds; A1 represents the land surface area, with a value of 36670 m². 2 i* represents net rainfall, and the net rainfall for a 2-hour, 20-year return period is 0.0118 mm / s.

[0079] Substituting the relevant parameter values ​​of the sub-catchment Y3501540059 into the equation, the flow rate Q is calculated. Using the Manning formula to dynamically update Q, the change in surface water volume over time, i.e., runoff, can be obtained. The flow process of this sub-catchment is shown in Table 2.

[0080] Table 2. Flow process of the secondary catchment area Y3501540059 under 2-hour 20-year return period design rainfall.

[0081] The calculation process of the hydrodynamic model is explained using the yjh-wmh-2_us.1 section of the river segment yqh-wmh-2_us.1 as an example.

[0082] S24 uses the hydrodynamic module in Infoworks ICM to calculate the river network confluence by using dynamic waves to fully solve the Saint-Venant equations.

[0083] Continuity equation:

[0084] Momentum equation:

[0085] Where Q is the flow rate, m 3 / s; A2 is the cross-sectional area, with a value of 76.824m². 2 g is the acceleration due to gravity, 9.8 m / s². 2 θ is the angle between the centerline of the pipeline and the horizontal line, with a value of 0 degrees; K is the water conveyance rate, with a value of 10; S0 is the slope of the canal bottom, with a value of 0.002.

[0086] S25 uses cross-sectional node water exchange to couple surface runoff and river network flow, as shown in the following formula:

[0087] in It is the flow rate at the i-th river network cross-section at time t. It is the flow rate at the i-th river network cross-section at time t-1. It is the surface runoff corresponding to the i-th river network cross-section at time t.

[0088] The process of substituting the relevant parameter values ​​to obtain the flow rate of the river cross section yjh-wmh-2_4 is shown in Table 3.

[0089] Table 3. Flow process at river cross section yjh-wmh-2_4

[0090] S3: Model simulations were used to calculate flood events under nine scenarios with design return periods of 1 year, 2 years, 3 years, 5 years, 10 years, 20 years, 30 years, 50 years, and 100 years. This step, S3, was performed computer-generated to comprehensively identify drainage outlets within the study area susceptible to backwater effects from the river network. Statistical data is shown in Figure 2, and the storm intensity formula used is as follows:

[0091] Where q is the average rainfall intensity (mm / min), P is the return period (years), and t is the rainfall duration (set to 120 min). The Chicago rainfall pattern was used to calculate the rainfall process, with a design peak rainfall coefficient r = 0.39 and a rainfall process time interval set to 1 min. The rainfall intensities calculated for P = 1a, 2a, 3a, 5a, 10a, 20a, 30a, 50a, and 100a are shown in Table 4, with only the first 15 minutes displayed.

[0092] Table 4. Rainfall Intensity Table for Designed Rainfall Scenario

[0093] S4: A clustering algorithm was used to spatially cluster and divide the backwater outlets into control zones. Based on the spatial clustering, feature clustering was performed on the backwater outlets again, and the cluster center with the greatest influence was output as the river network control point. There are 198 outlets in the entire study area. Under 9 design rainstorm scenarios, a total of 105 outlets were affected by backwater from the river network, and there were 78 river cross sections that generated backwater. This step S4 was completed by computer.

[0094] The spatial clustering used is the AGNES algorithm, which is a bottom-up agglomerative hierarchical clustering algorithm. The calculation process is as follows:

[0095] S41, Initialization: Each sample point forms its own cluster. With a total of 78 samples, there are 78 clusters initially.

[0096] S42, Calculate the distance matrix: Initially, a selected distance metric is used; this paper chooses Euclidean distance to calculate the distance matrix between all clusters. The method for calculating the Euclidean distance between two n-dimensional vectors is as follows:

[0097] Where d is the Euclidean distance; x 1i and x 2i Let x be any two n-dimensional vectors. Taking points 04047_2630 and 04001_2490 as an example, x... 1i and x 2i The spatial coordinates of a point are represented by two two-dimensional vectors, x and y. 1i The coordinates are (431781.408, 2880105.636), x 2iGiven the coordinates (433116.133, 2878371.127), the distance between them is:

[0098] S43, Merge the nearest clusters: Find the two closest clusters, merge them into a new cluster, and update the distance matrix to reflect the distances between the new cluster and other clusters. The merging process is based on the chosen linking method; this paper chooses Ward's method to minimize the sum of squares within the merged clusters.

[0099] Where C i and C j For two clusters to be merged; |C i | and |C j | are clusters C i and C j The number of data points contained; μ i and μ j They are cluster C i and C j The mean vector; d(μ) i ,μ j d is the Euclidean distance between the mean vectors of the two clusters. ward (C i C j The distance (μ) is the "merging cost" or "distance" between two clusters calculated using the Ward method. Given two clusters C1 and C2 to be merged, with 15 and 8 samples respectively, and corresponding mean vectors μ... i = (4,6), μ j = (7,2), Euclidean distance The "merging cost" between two clusters

[0100] S44, repeat step S43 until all sample points are merged into one cluster or the predetermined number of clusters is reached. Spatial clustering is used to divide the control zones, and the final clustering result is 3, divided into North, Southwest, and Southeast regions;

[0101] The K-Medoids algorithm is used for feature clustering. Clustering is performed based on the spatial clustering results. The calculation process is illustrated using the spatial clusters of the northern region as an example:

[0102] S46, Initialization: Randomly select 9 points from the 39 sample points in the northern area as the initial cluster centers;

[0103] S47, Cluster Assignment: Assign the remaining 30 points to the cluster containing their nearest cluster center according to the minimum distance principle, thus achieving initial clustering. The distance between each sample point is calculated using Euclidean distance, with the same formula as in S42.

[0104] S48, Update cluster centers: Calculate the distance from each point in the cluster to all other points except the cluster center, take the point with the minimum distance as the new cluster center, and recalculate the clustering results;

[0105] S49. Repeat processes S47 and S48 until all cluster centers no longer change or the set maximum number of iterations has been reached. The final clustering result is 4. Output the cluster center with the greatest influence as the river network control point. The corresponding river section ID is 04047_2630. Using the same method, the river network control point IDs for the southwest and southeast regions are 04053_1500 and 04001_2490, respectively, as shown in Figure 3.

[0106] S5: Using the river network control points as the reference sections for control, and based on the warning water level and ecological water level data of the river channel, optimize the gate and pump control rules; this step S5 is completed by computer. The optimized gate and pump control rules are as follows: Z(i) min =H(i) 生态水位 Z(i) max =H(i) 警戒水位 B(j) min =H(j) 警戒水位 B(j) max =H(j) 警戒水位 +Δh(j)

[0107] Where Z(i) max Z(i) represents the upper limit water level controlled by the i-th gate. min B(j) is the lower limit water level controlled by the i-th gate; max B(j) represents the upper limit water level regulated by the j-th pumping station. min H(i) is the lower limit water level regulated by the j-th pumping station. 警戒水位 H(i) 生态水位 H(j) represents the warning water level and ecological water level at the control reference section of the i-th gate. 警戒水位 Δh(j) is the warning water level of the control reference section of the j-th pumping station; Δh(j) is the control threshold of the control reference section of the j-th pumping station, which is empirically taken as 0.3m.

[0108] The ecological and warning water levels at river section 04047-2630 serve as control references for the Changpu Sluice Gate, Shengli Sluice Gate, Jiangbian Sluice Gate, Changpu Pumping Station, and Shengli Pumping Station in the northern area. The ecological and warning water levels at river section 04053-1500 serve as control references for the Yangqi Control Gate, Jimuyu Sluice Gate, and Yangqi Pumping Station in the southwestern area. The ecological and warning water levels at river section 04001-2490 serve as control references for the Wufeng Sluice Gate, Yixu Sluice Gate, and Yixu Pumping Station in the southeastern area. Warning and ecological water level data for the control reference sections provided by relevant units are shown in Table 5, and optimized sluice gate and pump control rules are shown in Table 6.

[0109] Table 5. Ecological water level and warning water level corresponding to the reference cross-sections for regulation.

[0110] Table 6 Optimized Gate Pump Control Rules

[0111] S6: Input the optimized gate pump control rules into the urban hydrological and hydrodynamic coupling model, and use the performance evaluation method based on rainstorm events to quantitatively analyze the response of flooding to the gate pump control rules. This step S6 is completed by computer.

[0112] The real-time control rules for water engineering in the input urban hydrological and hydrodynamic coupling model are expressed as follows:

[0113] S61, the gate control rule is: If NODE A <Z(i) min THEN ORIFICE A=0 If NODE A>Z(i) max THEN ORIFICE A = 1.0

[0114] Where NODE A is the gate's control reference section ID; 0.0 indicates the gate is closed, and 1.0 indicates the gate is fully open.

[0115] S62, Pumping Station Control Rules: If NODE A <B(j) min THEN PUMP A STATUS=OFF If NODE A>B(j) max THEN PUMP A STATUS=ON

[0116] Where NODE A is the control reference section ID of the pumping station; OFF indicates that the pumping station is closed, and ON indicates that the pumping station is open;

[0117] S63, using the performance of the rainstorm event assessment method, quantitatively analyzes the indicators of the response of floods to the gate pump control rules, including the number of overflows from rainwater wells and the water level process at key river sections;

[0118] Considering the existing gate pump control rules (Table 7) (denoted as Scenario A), a city hydrological and hydrodynamic coupling model is used to simulate the flooding process of the "Soudelor" typhoon and rainstorm event (Figure 4). The inundation depth of rainwater wells refers to the difference between the water level of the rainwater well and the ground elevation where the rainwater well is located. Under the existing gate pump control rules, the number of rainwater well overflows is 2203, the overflow ratio is 43.15%, and the highest water heads at the three river network control points are 6.475m, 7.193m, and 5.935m, respectively.

[0119] Table 7 Existing Gate Pump Control Rules

[0120] Furthermore, considering the optimized gate pump control rules (Table 3) (Scenario B), a coupled urban hydrological and hydrodynamic model was used to simulate the flooding process of the "Soudelor" typhoon and rainstorm event. Compared with Scenario A, in Scenario B, the number of overflowing rainwater wells was 2125, the overflow rate was 41.63%, and the number of overflows decreased by 1.52%, indicating an improvement in drainage system overflow. The highest water heads at the three river network control points were 6.367m, 6.726m, and 5.866m, respectively, representing reductions of 1.67%, 6.49%, and 1.16%, as shown in Figure 5.

[0121] This demonstrates that optimized gate and pump control rules during heavy rains played a crucial role in flood control, significantly reducing river head and minimizing overflows from drainage systems around the river network, thus mitigating flood disasters.

[0122] Example 2:

[0123] The urban hydrological and hydrodynamic coupled model was calibrated and validated using two rainstorm events: from 14:00 on March 26, 2022 to 4:00 on March 27, 2022, and from 00:00 to 12:00 on April 27, 2022, respectively. Event 1 was used to calibrate the model, and Event 2 was used to validate it. The rainfall amounts for Event 1 and Event 2 were 68.3 mm and 126.0 mm, respectively. Three rainwater wells (M1, M2, M3) and three river cross-sections (S1, S2, S3) were randomly selected from the observation points as samples. The mean head NSE of the samples under Event 1 and Event 2 were higher than 0.85 and 0.79, respectively, indicating good model simulation performance. The calibrated model parameter values ​​are shown in Table 8. Figure 6 compares the simulated and observed head processes of rainwater well samples and river cross-section samples during the rainfall of Event 1. As shown in the figure, the simulated head process of some rainwater well samples exhibits certain fluctuations, such as rainwater well 1, mainly manifested in an earlier peak head and larger head variation. However, the errors between the simulated and measured head at all three nodes are within the allowable range. The simulated head process of the river cross-section samples is basically consistent with the measured head process, and the overall simulation accuracy is higher than that of the rainwater well samples. The simulated head process of the selected nodes and cross-sections largely matches the measured head process, indicating that the model can largely reflect the actual situation and can be used for river network regulation calculations and analysis. The above process was completed by computer.

[0124] Table 8. Range of Model Parameter Values

[0125] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A method for optimizing real-time river network control rules for urban flooding, characterized in that: Includes the following steps: S1, collect river network data; S2. Based on detailed spatial division and different surface runoff characteristics, an urban hydrological model is constructed. A river network hydrodynamic model is constructed considering the hydraulic conditions of pumping stations, gates, weirs, reservoirs and lakes. A pipe network hydrodynamic model is constructed considering the hydraulic conditions of manholes and outlet facilities. The above models are coupled through the process of water exchange at cross-sectional nodes to construct an urban hydrological and hydrodynamic coupled model. S3, based on the rainstorm intensity formula, simulates and calculates the flood process under different design rainstorm scenarios, and comprehensively identifies drainage outlets in the study area that are susceptible to backwater from the river network. S4. A clustering algorithm is used to spatially cluster and divide the top drainage outlet into control zones. Based on the spatial clustering, the top drainage outlet is further clustered by feature, and the cluster center with the greatest influence is output as the river network control point. S5 uses river network control points as the control reference object and compiles and optimizes gate and pump control rules based on river warning water level and ecological water level data; S6 will input the optimized gate pump control rules into the urban hydrological and hydrodynamic coupling model, and use the performance of the rainstorm event assessment method to quantitatively analyze the response of floods to the gate pump control rules.

2. The method for optimizing real-time river network control rules for urban flooding as described in claim 1, characterized in that: The urban hydrological and hydrodynamic coupling model in S2 is built based on Infoworks ICM software. The specific construction process is as follows: S21. The runoff generation calculation for impermeable surfaces of buildings and roads uses the fixed runoff coefficient model in InfoWorks ICM, as shown in the following formula: R n =CR P ; Where R n Net rainfall, mm; C is runoff coefficient; P is return period, years; R P The rainfall amount for the corresponding return period is in mm; S22, The calculation of permeable surface runoff in green spaces adopts the Horton infiltration model, as shown in the following formula: f t =f c +(f0-f c )e -kt ; Where f t f is the infiltration rate at time t, in mm / h; c The stable infiltration rate is measured in mm / h; f0 is the initial infiltration rate, measured in mm / h; k is the reduction coefficient, with dimensions in h. -1 t represents the infiltration time, in hours. S23, simulating urban surface runoff, utilizes the nonlinear reservoir model in the stormwater and flood management module, and is solved using a combination of the continuity equation and Manning's formula, as shown below: Manning's formula: Where Q is the flow rate, m 3 / s; W is the overflow width of the sub-catchment area, m; n is the Manning coefficient; d is the water depth, m; d p The maximum water storage depth is given by S in meters; S is the average slope of the sub-catchment area. Continuity equation: Where V is the surface water volume, in m 3 t is time, s; A1 is the surface area, m. 2 i* represents net rainfall, mm / s; S24, using the hydrodynamic module in Infoworks ICM, calculates the river network confluence by using dynamic waves to fully solve the Saint-Venant equations, as shown in the following equation: Continuity equation: Momentum equation: Where A2 is the cross-sectional area, in m 2 Q represents flow rate, m 3 / s; g is the acceleration due to gravity, m / s² 2 θ is the angle between the pipeline centerline and the horizontal line, in degrees; K is the water conveyance rate; h is the water depth, in meters; t is the time, in seconds; S0 is the slope of the canal bottom. S25 uses a cross-sectional node water exchange method to couple surface runoff with river network flow.

3. The method for optimizing real-time river network control rules for urban flooding according to claim 2, characterized in that: In S25, the formula for coupling surface runoff and river network flow using cross-sectional node water exchange is as follows: in It is the flow rate at the i-th river network cross-section at time t. It is the flow rate at the i-th river network cross-section at time t-1. It is the surface runoff corresponding to the i-th river network cross-section at time t.

4. The method for optimizing real-time river network control rules for urban flooding as described in claim 1, characterized in that: The formula for the intensity of the rainstorm in S3 is: Where q is the average rainfall intensity, mm / min; P is the return period, a; t represents the rainfall duration in minutes; the rainfall process is calculated using the Chicago rainfall pattern, with a design peak rainfall coefficient of r = 0.39, a rainfall process time interval of 1 minute, and a rainfall duration of 120 minutes.

5. The method for optimizing real-time river network control rules for urban flooding as described in claim 1, characterized in that: In S4, the AGNES algorithm is used for spatial clustering of the top drain outlet. This is a bottom-up agglomerative hierarchical clustering algorithm, and the calculation process is as follows: S41, Initialization: Each sample point forms its own cluster; if the population has m samples, then there are m clusters in the initial state; S42, Calculate the distance matrix: Initially, use the selected distance metric, choosing Euclidean distance to calculate the distance matrix between all clusters; S43, Merge the nearest clusters: Find the two closest clusters, merge them into a new cluster, and update the distance matrix to reflect the distance between the new cluster and other clusters; The merging process is based on the chosen linking method, selecting Ward's method to minimize the sum of squares within the merged clusters: Where C i and C j For two clusters to be merged; |C i | and |C j | are clusters C i and C j The number of data points contained; μ i and μ j They are cluster C i and C j The mean vector; d(μ) i ,μ j ) represents the distance between the mean vectors of the two clusters; d ward (C i C j ) represents the "merging cost" or "distance" between two clusters calculated using the Ward method; S44, Repeat step S43 until all sample points are merged into one cluster or the predetermined number of clusters is reached; S45, Generate a hierarchical tree: Use a tree diagram to represent the clustering process, merging one node from the corresponding tree each time.

6. The method for optimizing real-time river network control rules for urban flooding as described in claim 5, characterized in that: In S42, the Euclidean distance between two n-dimensional vectors is calculated as follows: Where d is the Euclidean distance; x 1i and x 2i They are any two n-dimensional vectors.

7. The method for optimizing real-time river network control rules for urban flooding as described in claim 1, characterized in that: In step S4, the K-Medoids algorithm is used for feature clustering of the top drain outlet. This algorithm directly divides the data points into different clusters. The calculation process is as follows: S46, Initialization: Randomly select k points from the total m sample points as the initial cluster centers; S47, Cluster Assignment: Assign the remaining mk points to the clusters containing their nearest cluster centers according to the minimum distance principle to achieve initial clustering; the distance between each sample point is calculated using Euclidean distance. S48, Update cluster centers: Calculate the distance from each point in the cluster to all other points except the cluster center, take the point with the minimum distance as the new cluster center, and recalculate the clustering results; S49. Repeat processes S47 and S48 until all cluster centers no longer change or the set maximum number of iterations has been reached, to obtain the final clustering result.

8. The method for optimizing real-time river network control rules for urban flooding as described in claim 1, characterized in that: The basis for writing the real-time control rules for water projects in S5 is the warning water level and ecological water level data of the river, as follows: Z(i) min =H(i) 生态水位 Z(i) max =H(i) 警戒水位 B(j) min =H(j) 警戒水位 B(j) max =H(j) 警戒水位 +Δh(j); Where Z(i) max Z(i) represents the upper limit water level controlled by the i-th gate. min B(j) is the lower limit water level controlled by the i-th gate; max B(j) represents the upper limit water level regulated by the j-th pumping station. min H(i) is the lower limit water level regulated by the j-th pumping station. 警戒水位 H(i) 生态水位 H(j) represents the warning water level and ecological water level at the control reference section of the i-th gate. 警戒水位 Δh(j) is the warning water level of the control reference section of the j-th pumping station; Δh(j) is the control threshold of the control reference section of the j-th pumping station.

9. The method for optimizing real-time river network control rules for urban flooding as described in claim 1, characterized in that: The real-time control rules for water engineering in the urban hydrological and hydrodynamic coupling model input in S6 are expressed as follows: S61, the gate control rules are as follows: If NODE A<Z(i) min ; THEN ORIFICE A = 0; If NODE A>Z(i) max ; THEN ORIFICE A = 1.0; Where NODE A is the control reference section ID of the gate; 0.0 indicates that the gate is closed, and 1.0 indicates that the gate is fully open; S62, Pumping Station Control Rules: If NODE A<B(j) min ; THEN PUMP A STATUS = OFF; If NODE A>B(j) max ; THEN PUMP A STATUS = ON; Where NODE A is the control reference section ID of the pump station; OFF indicates that the pump station is closed, and ON indicates that the pump station is open; S63, using the performance of the rainstorm event assessment method, quantitatively analyzes the indicators of the response of floods to the gate pump control rules, including the number of overflows from rainwater wells and the water level process at key river sections.

10. The method for optimizing real-time river network control rules for urban flooding according to claim 1, characterized in that: In S1, the collected river network data includes: the river network topology, cross-sectional data, underlying surface, gate and pumping station data required to construct the river network system. The gate and pumping station data includes design parameters and distribution locations, measured rainfall data, and measured river network water level and flow data corresponding to the measured rainfall data.

Images