Urban underground space emergency drainage control method under flood disaster

Abstract

The invention discloses an urban underground space emergency drainage control method under flood disasters, and the method comprises the steps: collecting multi-dimensional real-time data through a multi-source heterogeneous digital twin drainage intelligent management and control platform, and inputting a 1D-2D dynamic bidirectional coupling model to simulate a dynamic coupling relation between pipe network water flow and regional ponding; acquiring water flow pressure distribution and accumulated water diffusion trend data; starting a spatio-temporal feature driven drainage scheduling algorithm to extract spatio-temporal correlation features and construct a decision variable set, and screening variables in combination with emergency drainage control parameters; and inputting the screened variables into a model for feedback verification, if not, adjusting algorithm parameters, and if yes, sending a control instruction to execution equipment through the platform. The method can improve accumulated water and water flow state prediction precision, improve scheduling scheme adaptability and reliability, accelerate emergency drainage response speed, comprehensively improve underground space flood response capability, and guarantee underground space operation safety.

Description

Technical Field

[0001] This invention relates to the field of urban underground space drainage technology, and in particular to an emergency drainage control method for urban underground spaces under flood disasters. Background Technology

[0002] With the acceleration of urbanization, the development scale of urban underground spaces, such as underground shopping malls, subway stations, and underground utility tunnels, is constantly expanding. These spaces are located in low-lying areas and are highly enclosed, making them extremely prone to water accumulation during floods. Once the water accumulation exceeds the drainage system's capacity, it can not only paralyze the underground space's functions but also damage electromechanical equipment, threaten personnel safety, and cause huge economic losses and social impacts. Currently, urban underground space drainage control relies heavily on traditional fixed-mode scheduling, which is insufficient to cope with the complex and ever-changing water flow conditions and water spread during floods. There is an urgent need for a technical method that can accurately simulate water flow coupling relationships, dynamically generate scheduling plans, and manage them in real time to improve the timeliness and effectiveness of emergency drainage in underground spaces and ensure their operational safety during floods.

[0003] Existing technologies for emergency drainage control in urban underground spaces have two significant drawbacks: First, existing drainage simulation models mostly employ single-dimensional pipe network simulation or regional water accumulation simulation, failing to achieve dynamic bidirectional coupling between one-dimensional pipe network water flow and two-dimensional regional water accumulation distribution. This results in simulation results that cannot accurately reflect the mutual influence between the two, making it difficult to accurately predict water accumulation diffusion trends and changes in pipe network water pressure, thus affecting the scientific validity of drainage scheme formulation. Second, existing drainage scheduling algorithms lack in-depth analysis of the spatiotemporal characteristics of water flow and water accumulation diffusion processes. They cannot adaptively adjust scheduling strategies based on the dynamic characteristics of water flow and water accumulation at different times and locations. Furthermore, they fail to fully incorporate multiple key parameters for emergency drainage control in underground spaces for constraint screening during the scheduling process, resulting in insufficient adaptability and reliability of the scheduling scheme, making it difficult to meet the needs of rapid and efficient drainage under flood disasters. Summary of the Invention

[0004] In order to overcome the shortcomings and deficiencies of existing technologies, this invention provides an emergency drainage control method for urban underground spaces under flood disasters.

[0005] The technical solution adopted in this invention is an emergency drainage control method for urban underground space under flood disasters, comprising the following steps: S1, collecting multi-dimensional real-time data related to the drainage system in urban underground space through a multi-source heterogeneous digital twin drainage intelligent management and control platform. The multi-dimensional real-time data covers the water depth in various areas of the underground space, the water flow velocity in the drainage network, the operating status of the drainage pump group, and the surrounding surface runoff; S2, inputting the multi-dimensional real-time data collected in S1 into a 1D-2D dynamic bidirectional coupling model, using the model to simulate the dynamic coupling relationship between the 1D network water flow movement and the 2D regional water distribution in the underground space drainage system, and obtaining data on the water pressure distribution in the network and the regional water diffusion trend at different times; S3, based on the water pressure distribution in the network and the regional water diffusion trend data obtained in S2, starting a spatiotemporal feature-driven drainage scheduling algorithm. This algorithm extracts the spatiotemporal correlation features in the process of water flow movement and water diffusion, and constructs a system based on the spatiotemporal correlation features. S4, during the operation of the spatiotemporal feature-driven drainage scheduling algorithm, the variables in the drainage scheduling decision variable set are constrained and screened in combination with the preset parameters of emergency drainage control in urban underground space. The preset parameters include the maximum head of the drainage pump group, the maximum pressure value of the pipe network, and the threshold corresponding to the target time for water receding. S5, the drainage scheduling decision variables screened in S4 are input into the 1D-2D dynamic bidirectional coupling model for feedback verification to determine whether the water flow pressure in the pipe network is within a safe range and whether the regional water diffusion is effectively suppressed under the drainage scheme corresponding to the decision variable. If not satisfied, the algorithm returns to S3 to readjust the parameters of the spatiotemporal feature-driven drainage scheduling algorithm. S6, when the feedback verification result in S5 meets the preset conditions, a control command is sent to the emergency drainage execution equipment in urban underground space through the multi-source heterogeneous digital twin drainage intelligent management and control platform. The control command includes the number of drainage pump groups to start and stop, the valve opening adjustment range, and the emergency drainage channel switching command.

[0006] Furthermore, in S2, the 1D-2D dynamic two-way coupled model performs the coupling calculation of 1D pipe network water flow motion and 2D regional water accumulation distribution through the following formula: Among them, Q 1D-2D (x, t) represents the water exchange rate at position x after 1D-2D coupling at time t, α is the influence coefficient of the 1D pipe network flow on the coupling process, and Q 1D (x, t) represents the 1D water flow rate at position x in the pipe network at time t, h 2D (x, t) represents the water depth of the 2D region at position x at time t. Let v be the average water depth in the 2D region, β be the influence coefficient of the water accumulation in the 2D region on the coupling process, and v 2D (x, t) represents the flow velocity of water in the 2D region at position x at time t, and A 1D(x, t) represents the 1D pipe network flow area at position x at time t.

[0007] Furthermore, in S3, when the spatiotemporal feature-driven drainage scheduling algorithm extracts spatiotemporal correlation features, the following formula is used to calculate the spatiotemporal correlation degree:

[0008]

[0009] Where R(x1, x2, t1, t2) represents the spatiotemporal correlation between the positions of x1 and x2, and between the times of t1 and t2, and Q(x1, t2) represents the spatiotemporal correlation between the positions of x1 and x2, and between the times of t1 and t2. i () represents the water flow rate in the pipe network at position x1 at time ti. Let h(x2, t) be the average flow rate of the pipe network at location x1. i (x) represents the water depth at position x2 at time ti. denoted as x2, where x is the average water depth, n is the number of data samples, γ is the time decay coefficient, and δ is the spatial decay coefficient.

[0010] Furthermore, in S4, when constraining and screening drainage scheduling decision variables by combining urban underground space emergency drainage control parameters, the following formula is used: When G(P1, P2, P3) ≤ θ, the drainage scheduling decision variable is retained, where G(P1, P2, P3) is the comprehensive constraint value of the decision variable, P1 is the actual head of the drainage pump set, and P... 1max P1 is the maximum head of the drainage pump set, P2 is the actual pressure value of the pipeline network, and P... 2max P3 represents the maximum pressure bearing capacity of the pipeline network, and P4 represents the actual time for the accumulated water to recede. 3max The thresholds corresponding to the target duration of water receding are μ1, μ2, and μ3, which are weighting coefficients, and θ is the constraint threshold.

[0011] Furthermore, in S5, when the 1D-2D dynamic two-way coupled model performs feedback verification of the drainage scheme, the verification index is calculated using the following formula: Where V is the validation index value, λ1 and λ2 are weight coefficients, and max(P) pipe (x, t) represents the maximum water flow pressure at various locations in the pipe network at time t, P pipemax The maximum pressure bearing capacity of the pipeline network is max(h) area (x, t) represents the maximum water depth at each location in the region at time t, and h represents the maximum water depth at each location in the region. areamax The maximum allowable water depth in the area is defined as V. When V ≤ 1, the feedback verification result is determined to meet the preset conditions.

[0012] Furthermore, in S6, before sending control commands to emergency drainage execution equipment, the multi-source heterogeneous digital twin drainage intelligent management and control platform optimizes the accuracy of command output using the following formula: Among them, C opt For the optimized control command parameters, C init Here, η is the initial control command parameter, and Q is the command optimization coefficient. sim (x, t) represents the water flow rate at position x at time t, simulated by the 1D-2D dynamic two-way coupled model, and Q is the flow rate at position x. act (x, t) represents the actual water flow rate at position x at time t.

[0013] Further, S3 includes the following sub-steps: S31, from the water flow pressure distribution and regional water diffusion trend data output from S2, extract the water flow rate, water flow velocity, water depth, and water diffusion direction data of each monitoring point within each time interval, and classify and store these data according to time series and spatial location to form a spatiotemporal data matrix; S32, extract spatiotemporal features from the data in the spatiotemporal data matrix, divide it into different spatiotemporal segments using a sliding window algorithm, and calculate the rate of change of water flow rate, the rate of change of water depth, and the direction of water flow and water diffusion within each spatiotemporal segment. Consistency coefficient, constructing spatiotemporal feature vectors; S33, inputting the spatiotemporal feature vectors into the feature analysis module of the spatiotemporal feature-driven drainage scheduling algorithm. This module uses a clustering algorithm to group the spatiotemporal feature vectors and determine the feature patterns under different spatiotemporal scenarios, providing a basis for the subsequent construction of a set of drainage scheduling decision variables; S34, based on the determined feature patterns and combined with the structural parameters of the urban underground space drainage system, including pipe diameter, number of drainage pump sets and location of emergency channels, constructing a set of drainage scheduling decision variables including pump set operating parameters, valve adjustment parameters and channel switching parameters.

[0014] Further, S4 includes the following sub-steps: S41, retrieving preset parameters from the urban underground space emergency drainage control parameter database. These preset parameters include, in addition to the maximum head of the drainage pump set, the maximum pressure value of the pipe network, and the threshold corresponding to the target duration of water receding, the minimum flow area of ​​the drainage pipe network, the maximum flow rate of the emergency drainage channel, and the continuous operating time limit of the pump set; S42, comparing each decision variable in the drainage scheduling decision variable set with the retrieved preset parameters one by one to determine the degree of deviation between the actual value of each parameter corresponding to each decision variable and the preset parameters; S43, setting constraint weights according to the degree of deviation, assigning higher constraint weights to decision variables corresponding to parameters with larger deviations and lower constraint weights to decision variables corresponding to parameters with smaller deviations; S44, comprehensively scoring each decision variable based on the constraint weights, and selecting decision variables whose comprehensive scores meet the preset requirements to form a subset of filtered drainage scheduling decision variables.

[0015] Further, S5 includes the following sub-steps: S51, converting each decision variable in the subset of drainage scheduling decision variables selected in S4 into an input parameter format recognizable by the 1D-2D dynamic bidirectional coupling model, including converting pump group operating parameters into pump group flow coefficients in the model and valve adjustment parameters into pipeline impedance coefficients in the model; S52, inputting the converted input parameters into the 1D-2D dynamic bidirectional coupling model, starting the model to perform simulation calculations, and obtaining the simulation calculation output data of water flow pressure at various locations in the pipeline network, water accumulation depth at various locations in the region, and water receding rate; S53, comparing the simulation output data with preset safety standards, which include the upper limit of safety for pipeline water flow pressure, the upper limit of safety for regional water accumulation depth, and the lower limit of water receding rate; S54, if the simulation output data meets the preset safety standards, then the drainage scheme corresponding to the decision variable is deemed feasible; if not, then the parameter type and deviation value that are not met are recorded, and the process returns to S3, adjusting the feature weight coefficients in the spatiotemporal feature-driven drainage scheduling algorithm according to the deviation value.

[0016] Beneficial Effects: This invention proposes an emergency drainage control method for urban underground spaces during floods. It utilizes a multi-source heterogeneous digital twin intelligent drainage management platform to accurately collect multi-dimensional real-time data. Combined with a 1D-2D dynamic bidirectional coupling model, it achieves dynamic coupling simulation of one-dimensional pipe network flow and two-dimensional regional water accumulation, completely overcoming the limitations of single-dimensional simulation and significantly improving the accuracy of water accumulation diffusion trends and pipe network flow pressure prediction, providing a scientific basis for drainage scheme formulation. Based on spatiotemporal characteristics, the drainage scheduling algorithm deeply mines the spatiotemporal correlation characteristics of water flow and water accumulation. Combined with emergency drainage control parameters for underground spaces, it constrains and filters scheduling decision variables. Simultaneously, a feedback verification mechanism continuously optimizes the scheduling scheme, effectively solving the problems of existing algorithms' inability to adaptively adjust strategies and insufficient reliability, significantly improving the adaptability and rationality of the scheduling scheme. Finally, the platform sends precise control commands to the execution equipment, improving the overall response speed and execution efficiency of emergency drainage in underground spaces, effectively ensuring the operational safety of underground spaces during floods, minimizing economic losses and social impact, and comprehensively enhancing the city's underground space's ability to cope with floods. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method steps of the present invention;

[0018] Figure 2 This is a diagram showing the unit composition for implementing the method of the present invention. Detailed Implementation

[0019] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] like Figure 1 As shown, the emergency drainage control method for urban underground spaces under flood disasters includes the following steps:

[0021] S1, collects multi-dimensional real-time data related to the drainage system in the urban underground space through a multi-source heterogeneous digital twin drainage intelligent management and control platform. The multi-dimensional real-time data covers the water depth in various areas of the underground space, the water flow velocity in the drainage network, the operating status of the drainage pump group and the surrounding surface runoff.

[0022] Specifically, step S1 forms the data foundation of the entire emergency drainage control method, providing comprehensive, real-time, and accurate raw data support for subsequent model calculations and scheduling decisions, avoiding deviations in subsequent simulations and decisions due to missing or delayed data. This step, through a multi-source heterogeneous digital twin intelligent drainage management platform, breaks through the limitations of traditional single and dispersed data collection, achieving centralized collection of multi-dimensional data on urban underground space drainage systems. This data covers key information related to drainage efficiency, system status, and external influences. The completeness and real-time nature of this data directly determine the reliability of subsequent 1D-2D dynamic bidirectional coupled model simulation results, as well as the scientific validity of the spatiotemporal characteristics-driven drainage scheduling algorithm decisions, which is a prerequisite for ensuring the effective operation of the entire emergency drainage control process. In the specific implementation process, the multi-source heterogeneous digital twin intelligent drainage management platform completes data collection through a sensor network deployed in different areas of the underground space. The data collected includes: water depth data collected via level sensors spaced 5 meters apart, with a sampling frequency of once per minute to ensure real-time monitoring of water level changes in each area; water flow velocity data collected via flow velocity sensors installed at key nodes of the drainage network, with a sampling frequency of once every 30 seconds, covering the main pipe network with diameters ranging from 300 mm to 1200 mm; drainage pump unit operating status data, including parameters such as pump current, voltage, and operating time, transmitted in real time through the pump unit control system, with a data update interval of 10 seconds; and surrounding surface runoff data collected collaboratively via surface rain gauges and flow monitoring stations, with a sampling frequency of once per minute. All collected data is uploaded in real time via the platform's data transmission module, with transmission delays controlled within 5 seconds to ensure data timeliness meets the requirements of subsequent steps.

[0023] S2 inputs the multi-dimensional real-time data collected by S1 into the 1D-2D dynamic bidirectional coupling model. The model is used to simulate the dynamic coupling relationship between the 1D pipe network water flow movement and the 2D regional water accumulation distribution in the underground space drainage system, and to obtain data on the water flow pressure distribution and regional water accumulation diffusion trend in the pipe network at different times.

[0024] Specifically, step S2 is the core step in realizing the dynamic simulation of the drainage system. Through a 1D-2D dynamic bidirectional coupling model, the multi-dimensional real-time data collected in step S1 is transformed into key information reflecting the actual operating status of the drainage system, solving the problem that traditional single-dimensional simulation cannot accurately reflect the interaction between the pipe network and regional water accumulation. This step, by simulating the dynamic coupling relationship between 1D pipe network water flow and 2D regional water accumulation distribution, can accurately present the changing patterns of water pressure within the pipe network at different times, as well as the diffusion direction and speed of regional water accumulation. This provides a direct basis for the subsequent spatiotemporal feature-driven drainage scheduling algorithm to extract effective features and construct reasonable decision variables, serving as a crucial bridge connecting data acquisition and scheduling decisions. In specific implementation, the multi-dimensional real-time data collected in step S1 is first imported into the corresponding input ports of the 1D-2D dynamic bidirectional coupling model according to data type. Specifically, water depth data is imported into the 2D regional water accumulation simulation module of the model, water flow velocity data is imported into the 1D pipe network water flow simulation module, and pump operation status data and surface runoff data are imported as boundary condition parameters of the model. During model execution, the time step is set to 1 minute, and the simulation duration is dynamically adjusted according to the duration of the flood disaster, with a minimum simulation duration of 30 minutes and a maximum of 24 hours. During the simulation, the model outputs intermediate results every 5 minutes, including water pressure values ​​at each node within the pipe network, water flow rates in each pipe section, and water depth values ​​per 10 square meters of grid space in the 2D region. After the simulation, a complete spatiotemporal variation curve is generated, with the accuracy of water pressure distribution data controlled within ±0.02 MPa and the accuracy of regional water diffusion trend data controlled within ±0.05 meters, ensuring that the output data meets the requirements of subsequent steps.

[0025] S3. Based on the water flow pressure distribution and regional water accumulation diffusion trend data obtained in S2, the spatiotemporal feature-driven drainage scheduling algorithm is started. This algorithm extracts the spatiotemporal correlation features in the process of water flow movement and water accumulation diffusion, and constructs a set of drainage scheduling decision variables based on the spatiotemporal correlation features.

[0026] Specifically, step S3 is the core step in constructing a scientific drainage scheduling scheme. Based on the simulation data obtained in step S2, the dynamic patterns of water flow and water accumulation are explored through spatiotemporal feature-driven drainage scheduling algorithms, forming a targeted set of drainage scheduling decision variables. This avoids the problem of rigid decision-making caused by the lack of spatiotemporal feature analysis in traditional scheduling algorithms. By extracting the spatiotemporal correlation features in the water flow movement and water accumulation diffusion process, this step can accurately identify the key needs of the drainage system under different spatiotemporal scenarios. Then, combined with the structural characteristics of the underground space drainage system, decision variables are constructed, providing rich and practical alternative solutions for subsequent constraint screening. This is a key link to ensure the adaptability and scientific nature of the drainage scheduling scheme. In the specific implementation process, firstly, the pipe network water flow pressure distribution data and regional water accumulation diffusion trend data for the past hour are extracted from the results output in step S2. These are divided into 12 time segments, each lasting 5 minutes, in chronological order. At the same time, the underground space is divided into 20 regional units according to spatial location, with each regional unit having an area not exceeding 500 square meters. Next, the spatiotemporal feature-driven drainage scheduling algorithm extracts features from the data of each spatiotemporal unit, calculating the rate of change of water pressure (pressure change per unit time), the rate of water spread (change in water coverage area per unit time), and the correlation coefficient between the two. During the extraction process, the data smoothing window is set to three time segments to ensure feature stability. Subsequently, combined with the structural parameters of the underground space drainage system, including the distribution locations of 25 drainage pump sets, the control nodes of 18 pipeline valves, and the passage capacity of 8 emergency drainage channels, a set of decision variables is constructed. The decision variables include the number of pump sets started and stopped (range 1-25), the valve opening adjustment range (range 0-100%), and the emergency channel switching status (open / closed). Each variable corresponds to 3-5 alternative values, forming an initial set containing 120-150 decision variables.

[0027] S4. During the operation of the spatiotemporal feature-driven drainage scheduling algorithm, the variables in the drainage scheduling decision variable set are constrained and screened in combination with the preset parameters of emergency drainage control in urban underground space. The preset parameters include the maximum head of the drainage pump group, the maximum pressure value of the pipeline network, and the threshold corresponding to the target time of water receding.

[0028] Specifically, step S4 is a crucial step in ensuring the safety and feasibility of the drainage scheduling plan. By combining the preset parameters of emergency drainage control in urban underground spaces, the set of decision variables constructed in step S3 is constrained and screened, eliminating variables that do not meet safety standards and actual operational capabilities. This prevents equipment damage or drainage failure due to decision variables exceeding the system's carrying capacity. This step, through clear parameter constraints, ensures that the screened decision variables meet drainage needs without exceeding the safety limits of the drainage system. It provides alternative solutions that meet actual operating conditions for subsequent feedback verification and serves as a vital link between decision variable construction and plan verification. In specific implementation, preset constraint parameters are first retrieved from the urban underground space emergency drainage control parameter database, including the maximum head of the drainage pump set (set to 15 meters), the maximum pressure of the pipe network (set to 0.8 MPa), the threshold corresponding to the target time for water receding (set to 2 hours), the minimum flow area of ​​the drainage pipe network (set to 0.07 square meters), and the maximum flow rate of the emergency drainage channel (set to 50 cubic meters per hour). Next, each variable in the decision variable set constructed in step S3 is compared with the above parameters one by one. For example, for the variable of pump start-up and shutdown quantity, the actual pump head under the corresponding operating state is calculated. If the actual head exceeds 15 meters, it is marked as unqualified. For the variable of valve opening adjustment range, the actual flow area of ​​the pipeline network after adjustment is calculated. If it is less than 0.07 square meters, it is marked as unqualified. During the screening process, a weighted scoring mechanism is adopted, in which the weight of pump head and pipeline pressure value is set to 0.3, the weight of water receding time is set to 0.2, and the weight of flow area and channel flow rate is set to 0.1 each. Decision variables with a comprehensive score of less than 80 points are eliminated, and finally 30-40 qualified decision variables are retained to form the selected subset of decision variables.

[0029] S5. Input the drainage scheduling decision variables selected in S4 into the 1D-2D dynamic bidirectional coupling model for feedback verification. Determine whether the water pressure in the pipe network is within a safe range and whether the regional water accumulation and diffusion are effectively suppressed under the drainage scheme corresponding to the decision variable. If not satisfied, return to S3 to readjust the parameters of the spatiotemporal feature-driven drainage scheduling algorithm.

[0030] Specifically, step S5 is the verification step to ensure the effectiveness of the drainage scheduling plan. A 1D-2D dynamic bidirectional coupling model is used to verify the decision variables selected in step S4, determining whether the corresponding drainage plan can meet the requirements of pipe network safety and water accumulation control, thus avoiding problems in practical application due to design flaws. This step, through a closed-loop feedback mechanism, can promptly identify unreasonable aspects in the decision variables and guide a return to step S3 for adjustment, forming a "construction-screening-verification-optimization" cycle. This ensures that the final determined drainage plan has high reliability and applicability, making it a crucial checkpoint for ensuring the effectiveness of emergency drainage control. In the specific implementation process, each variable in the subset of decision variables selected in step S4 is first converted into an input format recognizable by the 1D-2D dynamic bidirectional coupling model. For example, the number of pump start / stop operations is converted into the number of pumps operating in the model; the valve opening adjustment range is converted into the pipe network impedance coefficient in the model (the impedance coefficient decreases by 8% for every 10% increase in opening); and the emergency channel switching status is converted into the channel flow coefficient in the model (the coefficient is 1 when open and 0 when closed). Next, the converted parameters are input into the model, and simulation verification is started. The simulation duration is set to 1 hour, the time step is kept at 1 minute, and verification data is output every 10 minutes, including the maximum water flow pressure at each node of the pipe network, the maximum water accumulation depth in each unit of the area, and the water receding rate. The verification criteria are set as follows: the maximum water flow pressure of the pipe network does not exceed 0.8 MPa, the maximum water accumulation depth in the area does not exceed 0.3 meters, and the water receding rate is not less than 0.005 meters / minute. If the simulation result corresponding to a certain decision variable does not meet any of the criteria, the parameter type and deviation value that are not met are recorded (such as pressure exceeding 0.05 MPa, water accumulation depth exceeding 0.03 meters), and the process returns to step S3. The deviation value is used as the basis for adjustment, and the corresponding feature weight in the spatiotemporal feature-driven drainage scheduling algorithm is increased by 20%, and the set of decision variables is reconstructed. If all criteria are met, the drainage scheme corresponding to the decision variable is determined to be feasible.

[0031] S6. When the feedback verification result in S5 meets the preset conditions, a control command is sent to the emergency drainage execution equipment in the urban underground space through the multi-source heterogeneous digital twin drainage intelligent management and control platform. The control command includes the number of drainage pump sets to start and stop, the valve opening adjustment range, and the emergency drainage channel switching command.

[0032] Specifically, step S6 is the execution phase of the entire emergency drainage control method. It transforms the drainage scheme verified in step S5 into actual control commands, which are then distributed to the execution equipment via a multi-source heterogeneous digital twin intelligent drainage management platform. This achieves precise control of the underground space emergency drainage system, avoiding the slow response and large errors of traditional manual operation. This step, through the platform's command generation and distribution functions, ensures that control commands are accurately and quickly transmitted to each execution device, enabling the drainage system to operate according to the optimal scheme and maximizing emergency drainage capacity. It is a crucial step in transforming the scheduling plan into actual drainage results and directly determines the final effectiveness of emergency drainage control. In the specific implementation process, the instruction generation module of the multi-source heterogeneous digital twin drainage intelligent management and control platform first generates control instructions based on the decision variables verified in step S5. The instructions include: drainage pump group start / stop instructions (specifying the specific pump group number, such as pumps 1-5 start, pumps 6-10 stop), valve opening adjustment instructions (specifying the valve number and the adjusted opening value, such as valve 3 adjusted to 70%, valve 8 adjusted to 50%), and emergency drainage channel switching instructions (specifying the channel number and switching status, such as channels 2 and 5 open, the rest remain closed). After the instructions are generated, they are sent out through the platform's dedicated communication module. The communication method uses industrial Ethernet, and the transmission rate is set to 100Mbps to ensure that the instruction transmission delay does not exceed 2 seconds. Meanwhile, the platform monitors the command issuance status in real time, receiving feedback signals from the executing devices every 5 seconds (including whether the device has received the command, whether it has started execution, and its current execution status). If a device is detected not to have received the command, it will immediately reissue the command, with a maximum of 3 reissues. If the device receives the command but does not execute it, the backup control channel (using 4G wireless communication) will be activated to reissue the command, ensuring that all executing devices start operating according to the command within 10 seconds, thus enabling the rapid implementation of the emergency drainage plan.

[0033] Preferably, in S2, the 1D-2D dynamic two-way coupling model performs the coupling calculation between 1D pipe network water flow motion and 2D regional water accumulation distribution through the following formula: Among them, Q 1D-2D (x, t) represents the water exchange rate at position x after 1D-2D coupling at time t, α is the influence coefficient of the 1D pipe network flow on the coupling process, and Q 1D (x, t) represents the 1D water flow rate at position x in the pipe network at time t, h 2D (x, t) represents the water depth of the 2D region at position x at time t. Let v be the average water depth in the 2D region, β be the influence coefficient of the water accumulation in the 2D region on the coupling process, and v 2D (x, t) represents the flow velocity of water in the 2D region at position x at time t, and A 1D (x, t) represents the 1D pipe network flow area at position x at time t.

[0034] Specifically, in step S2, the 1D-2D dynamic bidirectional coupling model realizes the coupling calculation between 1D pipe network water flow and 2D regional water accumulation. Through a specific operational relationship, it accurately quantifies the water exchange between the two, solving the problem that traditional models cannot accurately reflect the coupling effect. This operational relationship organically combines the water flow parameters of the 1D pipe network and the 2D region, improving the accuracy of water exchange calculation and ensuring the reliability of subsequent simulation results. In the implementation process, the values ​​of various parameters required for the calculation are first determined. The influence coefficient of 1D pipe network water flow on the coupling process is set to 0.6, and the influence coefficient of 2D regional water accumulation on the coupling process is set to 0.4. The average water accumulation depth of the 2D region is determined based on a combination of historical data and real-time acquired data, typically ranging from 0.2 to 0.5. Next, the flow rate within the 1D pipe network at a specific location at a specific time is obtained. This flow rate value is collected by a pipe network flow velocity sensor and calculated, ranging from 10 to 50. Simultaneously, the 2D regional water accumulation depth at that location is obtained, ranging from 0.1 to... The parameters are: a 1D-2D water flow velocity between 0.6 and 0.2 in the 2D region; and a 1D pipe network flow area, calculated based on the pipe diameter. The flow area is 0.07 square meters for a pipe diameter of 300 mm and 1.13 square meters for a pipe diameter of 1200 mm. Finally, these parameters are substituted into the calculation relationship to calculate the water exchange volume after 1D-2D coupling at that location at that moment. The accuracy of the calculation results is controlled within ±0.5, ensuring that it can accurately reflect the water flow interaction between the 1D pipe network and the 2D region, and providing accurate data support for the entire model simulation.

[0035] Preferably, in S3, when the spatiotemporal feature-driven drainage scheduling algorithm extracts spatiotemporal correlation features, the spatiotemporal correlation degree is calculated using the following formula:

[0036]

[0037] Where R(x1, x2, t1, t2) represents the spatiotemporal correlation between the positions of x1 and x2, and between the times of t1 and t2, and Q(x1, t2) represents the spatiotemporal correlation between the positions of x1 and x2, and between the times of t1 and t2. i () represents the water flow rate in the pipe network at position x1 at time ti. Let h(x2, t) be the average flow rate of the pipe network at location x1. i (x) represents the water depth at position x2 at time ti. denoted as x2, where x is the average water depth, n is the number of data samples, γ is the time decay coefficient, and δ is the spatial decay coefficient.

[0038] Specifically, in step S3, when the spatiotemporal feature-driven drainage scheduling algorithm extracts spatiotemporal correlation features, it calculates the spatiotemporal correlation degree. This is achieved by quantifying the correlation between water flow and water accumulation parameters at different locations and times, accurately identifying spatiotemporal feature patterns, providing a basis for the subsequent construction of a set of drainage scheduling decision variables, solving the problem that traditional algorithms have difficulty capturing spatiotemporal correlation features, and improving the algorithm's adaptability to dynamic spatiotemporal changes. During implementation, first determine the required parameter values ​​for the calculation. The number of data samplings is set to 60 times based on actual monitoring needs, the time attenuation coefficient is set to 0.02, and the spatial attenuation coefficient is set to 0.01. Then, collect 60 data points on the pipe network flow rate and water depth at two specific locations at different times. The pipe network flow rate ranges from 8 to 55 for each sampling, and the water depth ranges from 0.08 to 0.65 for each sampling. Next, calculate the average pipe network flow rate and average water depth at each location. The average flow rate is calculated by summing the 60 flow rate values ​​and dividing by 60, with the result ranging from 12 to 48. The average water depth is calculated similarly, with the result ranging from 0.15 to 0.5. Finally, based on the flow rate values ​​collected each time... The difference between average flow rate and the difference between water depth are calculated, and their products are summed. The sum of the squares of the flow rate difference and the sum of the squares of the water depth difference are also calculated. Then, the square root of the sum of the squares of the flow rate difference and the sum of the squares of the water depth difference are calculated. The sum of these products is divided by the product of these two square roots to obtain the basic correlation degree. Finally, an exponential term calculated from the time decay coefficient, spatial decay coefficient, time difference between two moments, and spatial difference between two locations is multiplied. The time difference ranges from 1 to 30, the spatial difference ranges from 5 to 50, and the exponential term ranges from 0.5 to 0.9. This yields the spatiotemporal correlation degree between the two locations and two moments, with a result ranging from 0.3 to 0.8, thus quantifying the spatiotemporal correlation characteristics.

[0039] Preferably, in S4, when constraining and screening drainage scheduling decision variables in conjunction with urban underground space emergency drainage control parameters, the following formula is used: When G(P1, p2, p3) ≤ θ, the drainage scheduling decision variable is retained, where G(P1, P2, p3) is the comprehensive constraint value of the decision variable, p1 is the actual head of the drainage pump set, and p 1max p1 is the maximum head of the drainage pump set, p2 is the actual pressure value of the pipeline network, and p... 2max P3 represents the maximum pressure capacity of the pipeline network, and P4 represents the actual time it takes for the accumulated water to recede. 3max The thresholds corresponding to the target duration of water receding are μ1, μ2, and μ3, which are weighting coefficients, and θ is the constraint threshold.

[0040] Specifically, in step S4, the decision variables for drainage scheduling are constrained and screened by combining the emergency drainage control parameters of urban underground space. By constructing a comprehensive constraint value calculation method, various control parameters are associated with decision variables, so as to achieve scientific screening of decision variables, eliminate variables that do not meet the requirements of safety and efficiency, improve the feasibility and safety of decision variables, and solve the problems of the single and one-sided nature of traditional screening methods. During implementation, the values ​​of various parameters were first determined. Regarding weighting coefficients, the weighting coefficient corresponding to the actual head of the drainage pump set was set to 0.4, the weighting coefficient corresponding to the actual pressure value of the pipe network was set to 0.3, the weighting coefficient corresponding to the actual water receding time was set to 0.3, and the constraint threshold was set to 0.8. Then, the actual parameter values ​​corresponding to the drainage scheduling decision variables were obtained. The actual head of the drainage pump set ranged from 5 to 15, with its maximum head set to 15. The actual pressure value of the pipe network ranged from 0.2 to 0.8, with its maximum pressure value set to 0.8. The actual water receding time ranged from 60 to 180, with the threshold corresponding to its target time set to 120. Next, the ratio of each actual parameter value to its corresponding maximum parameter value or threshold was calculated. The values ​​are calculated as follows: the ratio of actual pump head to actual pipe network pressure to actual water receding time ranges from 0.33 to 1; the ratio of actual pipe network pressure to actual water receding time ranges from 0.25 to 1; and the ratio of actual water receding time ranges from 0.5 to 1.5. Each ratio is then multiplied by its corresponding weighting coefficient to obtain a weighted product: the weighted product of pump head ranges from 0.13 to 0.4; the weighted product of pipe network pressure ranges from 0.08 to 0.3; and the weighted product of water receding time ranges from 0.15 to 0.45. Finally, the three weighted products are summed to obtain a comprehensive constraint value for the decision variable, ranging from 0.36 to 1.15. When the comprehensive constraint value is less than or equal to 0.8, the drainage scheduling decision variable is retained, thus achieving effective selection of decision variables.

[0041] Preferably, in S5, when the 1D-2D dynamic two-way coupled model performs feedback verification of the drainage scheme, the verification index is calculated using the following formula: Where V is the validation index value, λ1 and λ2 are weight coefficients, and max(P) pipe (x, t) represents the maximum water flow pressure at various locations in the pipe network at time t, P pipemax The maximum pressure bearing capacity of the pipeline network is max(h) area (x, t) represents the maximum water depth at each location in the region at time t, and h represents the maximum water depth at each location in the region. areamax The maximum allowable water depth in the area is defined as V. When V ≤ 1, the feedback verification result is determined to meet the preset conditions.

[0042] Specifically, in step S5, when the 1D-2D dynamic bidirectional coupling model performs feedback verification of the drainage scheme, it calculates verification indicators. By quantifying these indicators, it determines whether the drainage scheme meets the requirements for pipe network safety and regional water accumulation control, providing a clear basis for judging the feasibility of the scheme. This solves the problems of lacking quantitative standards and ambiguous judgments in traditional verification methods, improving the accuracy and objectivity of scheme verification. During implementation, the values ​​of various parameters are first set. Regarding weighting coefficients, the weighting coefficient corresponding to pipe network water flow pressure is set to 0.5, and the weighting coefficient corresponding to regional water accumulation depth is set to 0.5. Then, through model simulation calculations, the water flow pressure values ​​at various locations in the pipe network and the water accumulation depth values ​​at various locations in the region at specific times are obtained. The water flow pressure values ​​at various locations in the pipe network range from 0.1 to 0.9, from which the maximum water flow pressure value is extracted, ranging from 0.5 to 0.9. The water accumulation depth values ​​at various locations in the region range from 0.05 to 0.4, from which the maximum water accumulation depth value is extracted, ranging from 0.2 to 0.4. Simultaneously, the maximum pressure bearing capacity of the pipe network is set to 0.8, and the maximum allowable water accumulation depth in the region is set to 0.3. Next, the ratio of the maximum water flow pressure to the maximum pressure bearing capacity of the pipe network is calculated, ranging from 0.625 to 1.125, and the ratio of the maximum water accumulation depth to the maximum allowable water accumulation depth of the area is calculated, ranging from 0.67 to 1.33. These two ratios are then multiplied by their corresponding weighting coefficients to obtain two weighted ratios: the pipe network pressure weighted ratio ranging from 0.31 to 0.56, and the area water accumulation weighted ratio ranging from 0.33 to 0.67. Finally, the two weighted ratios are added together to obtain the verification index value, ranging from 0.64 to 1.23. When the verification index value is less than or equal to 1, the feedback verification result is deemed to meet the preset conditions, and the drainage scheme is feasible; otherwise, it needs to be readjusted.

[0043] Preferably, in S6, before the multi-source heterogeneous digital twin drainage intelligent management and control platform sends control commands to the emergency drainage execution equipment, the command output accuracy is optimized using the following formula: Among them, C opt For the optimized control command parameters, C init Here, η is the initial control command parameter, and Q is the command optimization coefficient. sim (x, t) represents the water flow rate at position x at time t, simulated by the 1D-2D dynamic two-way coupled model, and Q is the flow rate at position x. act (x, t) represents the actual water flow rate at position x at time t.

[0044] Specifically, in step S6, the multi-source heterogeneous digital twin drainage intelligent management and control platform optimizes the accuracy of command output before sending control commands to the emergency drainage execution equipment. By correcting the initial control command parameters, the deviation between the command and actual needs is reduced, improving the accuracy of the control commands and ensuring that the execution equipment operates according to the optimal parameters. This solves the problems of insufficient optimization and precision in traditional command output, and improves the execution effect of emergency drainage. During implementation, the parameter values ​​are first determined, with the command optimization coefficient set to 0.15 based on equipment characteristics and historical operating data. Then, the initial control command parameters are obtained, determined according to the drainage scheme, ranging from 20 to 80. Simultaneously, the water flow rate at a specific location at a specific time is simulated and calculated using a 1D-2D dynamic bidirectional coupling model, i.e., the simulated flow rate value, ranging from 12 to 52. The actual water flow rate at that location at that time is also collected by a flow sensor, i.e., the actual flow rate value, ranging from 10 to 50. Next, the difference between the simulated flow rate value and the actual flow rate value is calculated, ranging from -3 to 2, and then the difference is divided by... The actual flow rate is used to obtain the flow deviation rate, which ranges from -0.3 to 0.2. Then, the sum of 1 and the flow deviation rate multiplied by the instruction optimization coefficient is calculated, i.e., 1 + 0.15 × flow deviation rate, with the result ranging from 0.955 to 1.03. Finally, the initial control instruction parameters are multiplied by this result to obtain the optimized control instruction parameters, which range from 19.1 to 82.4. This optimizes the accuracy of the instruction output, ensuring that the control instructions sent to the emergency drainage execution equipment can more accurately match the actual drainage needs, improve equipment operating efficiency and drainage effect, and avoid problems such as low drainage efficiency or abnormal equipment operation caused by instruction deviation.

[0045] Preferably, S3 includes the following sub-steps: S31, extracting water flow rate, water flow velocity, water depth, and water diffusion direction data for each monitoring point within each time interval from the water pressure distribution and regional water diffusion trend data output from S2, and classifying and storing these data according to time series and spatial location to form a spatiotemporal data matrix; S32, extracting spatiotemporal features from the data in the spatiotemporal data matrix, dividing it into different spatiotemporal segments using a sliding window algorithm, and calculating the rate of change of water flow rate, the rate of change of water depth, and the direction of water flow and water diffusion within each spatiotemporal segment. Consistency coefficients are used to construct spatiotemporal feature vectors; S33, the spatiotemporal feature vectors are input into the feature analysis module of the spatiotemporal feature-driven drainage scheduling algorithm. This module uses a clustering algorithm to group the spatiotemporal feature vectors and determine the feature patterns under different spatiotemporal scenarios, providing a basis for the subsequent construction of a set of drainage scheduling decision variables; S34, based on the determined feature patterns and combined with the structural parameters of the urban underground space drainage system, including the pipe diameter, the number of drainage pump sets and the location of emergency channels, a set of drainage scheduling decision variables including pump set operating parameters, valve adjustment parameters and channel switching parameters is constructed.

[0046] Specifically, the implementation of step S3 involves refining the process of extracting features and constructing decision variables for the spatiotemporal feature-driven drainage scheduling algorithm through four sub-steps. This transforms the abstract algorithm operation into a practical technical process, improving the accuracy of feature extraction and the rationality of decision variables, thus laying the foundation for subsequent scheduling scheme formulation. During implementation, S31 first extracts the network water flow pressure and regional water diffusion data every minute for the past hour from the output of S2. This data is then categorized and stored according to time series and spatial location. Spatially, the area is divided into 500-square-meter units, forming a spatiotemporal data matrix containing 60 time nodes and 20 spatial regions. S32 uses a sliding window algorithm with a window size of 3 minutes to divide the spatiotemporal segments, calculating the rate of change of water flow (range 0.2 to 1.5 per unit minute), the rate of change of water depth (range 0.01 to 0.05 per unit minute), and the consistency coefficient of water flow and water movement direction (range 0.3 to 0.9) within each segment. S33 constructs a spatiotemporal feature vector with dimension 180; S34 inputs the feature vector into the feature analysis module of the algorithm, and uses a clustering algorithm with a clustering number of 5 to group the features and determine the feature patterns of different spatiotemporal scenarios, such as "high flow and high diffusion" and "low flow and low diffusion"; S34 combines the structural parameters of the underground space drainage system, including the location of 25 drainage pump sets, 18 pipe network valve nodes, and the passage capacity of 8 emergency passages (maximum flow of 50 per passage), to construct a set of decision variables including the number of pump sets started and stopped (1 to 25), valve opening (0 to 100%), and passage switching status. Each variable is given 3 to 5 alternative values, and finally 120 to 150 initial decision variables are formed.

[0047] Preferably, S4 includes the following sub-steps: S41, retrieving preset parameters from the urban underground space emergency drainage control parameter database. These preset parameters include, in addition to the maximum head of the drainage pump set, the maximum pressure value of the pipe network, and the threshold corresponding to the target duration of water receding, the minimum flow area of ​​the drainage pipe network, the maximum flow rate of the emergency drainage channel, and the continuous operating time limit of the pump set; S42, comparing each decision variable in the drainage scheduling decision variable set with the retrieved preset parameters one by one to determine the degree of deviation between the actual value of each parameter corresponding to each decision variable and the preset parameters; S43, setting constraint weights according to the degree of deviation, assigning higher constraint weights to decision variables corresponding to parameters with larger deviations and lower constraint weights to decision variables corresponding to parameters with smaller deviations; S44, comprehensively scoring each decision variable based on the constraint weights, and selecting decision variables whose comprehensive scores meet the preset requirements to form a subset of filtered drainage scheduling decision variables.

[0048] Specifically, the constraint screening process in step S4 involves four sub-steps: parameter retrieval, comparison, weight setting, and scoring. It strictly eliminates unqualified decision variables based on emergency drainage control parameters, ensuring the safety and feasibility of decision variables and preventing subsequent solutions from exceeding the system's capacity. During implementation, S41 retrieves preset parameters from the parameter database. These include the maximum head of the drainage pump set (15), the maximum pressure of the pipe network (0.8), the target time threshold for water receding (120), the minimum flow area of ​​the drainage pipe network (0.07), the maximum flow rate of the emergency drainage channel (50), and the continuous operation time limit of the pump set (480 minutes). S42 compares each decision variable with the parameters one by one. For example, if the actual head corresponding to the number of pump starts / stops exceeds 15, it is marked; if the flow area of ​​the pipe network corresponding to the valve opening is less than 0.07, it is marked. 43. Set constraint weights according to the degree of deviation. The weight of the deviation between the pump head and the pipeline pressure value is set to 0.3, the weight of the deviation between the water receding time is set to 0.2, and the weight of the deviation between the flow area and the channel flow rate is set to 0.1 each. S44. Give a comprehensive score to each decision variable according to the weight. The full score is 100 points. The larger the deviation, the more points are deducted. For example, if the head exceeds 1, 20 points are deducted. Finally, select decision variables with a score of not less than 80 points to form a subset of 30 to 40 decision variables that meet the requirements, so as to ensure that all retained variables meet the system safety operation standards.

[0049] Preferably, S5 includes the following sub-steps: S51, converting each decision variable in the subset of drainage scheduling decision variables selected in S4 into an input parameter format recognizable by the 1D-2D dynamic bidirectional coupling model, including converting pump group operating parameters into pump group flow coefficients in the model and valve adjustment parameters into pipeline impedance coefficients in the model; S52, inputting the converted input parameters into the 1D-2D dynamic bidirectional coupling model, starting the model to perform simulation calculations, and obtaining the simulation calculation output data of water flow pressure at various locations in the pipeline network, water accumulation depth at various locations in the region, and water receding rate; S53, comparing the simulation output data with preset safety standards, the preset safety standards including the upper limit of pipeline water flow pressure safety, the upper limit of regional water accumulation depth safety, and the lower limit of water receding rate; S54, if the simulation output data meets the preset safety standards, then the drainage scheme corresponding to the decision variable is determined to be feasible; if not, then the parameter type and deviation value that are not met are recorded, and the process returns to S3 and the feature weight coefficients in the spatiotemporal feature-driven drainage scheduling algorithm are adjusted according to the deviation value.

[0050] Specifically, the feedback verification process in step S5 involves four sub-steps: decision variable transformation, model simulation, standard comparison, and parameter adjustment. Its core is to construct a closed-loop verification mechanism to improve the reliability of the drainage scheme and prevent actual drainage failure due to scheme defects. During implementation, S51 converts the decision variables selected in S4 into a model-recognizable format; the number of pump start / stop operations is converted into the pump flow coefficient in the model (0.04 for each pump); and the valve opening is converted into the network impedance coefficient (for every 10% increase in opening, the impedance coefficient decreases by 8%). S52 inputs the converted parameters into a 1D-2D dynamic bidirectional coupled model, with a simulation duration of 60 minutes and a time step of 1 minute. Every 10 minutes, the model outputs the water flow pressure at each node of the network (range 0.1 to 0.9), the water depth at each location in the area (range 0.05 to 0.4), and the water receding rate (range 0.05 to 0.4). S53 compares the simulated data with the preset safety standards, which include an upper limit of 0.8 for the safe pressure of the pipe network water flow, an upper limit of 0.3 for the safe depth of the area water accumulation, and a lower limit of 0.005 for the water receding rate; S54 if the simulated data meets the standards, the scheme is deemed feasible; if it does not meet the standards, such as pressure exceeding 0.8 or depth exceeding 0.3, the deviation value is recorded (e.g., pressure exceeding 0.05, depth exceeding 0.03), and the algorithm is returned to S3. The corresponding feature weights in the algorithm are increased by 20%, such as adjusting the weight of the "high pressure feature" from 0.2 to 0.24. The set of decision variables is reconstructed until the simulation results meet the standards.

[0051] The 1D-2D dynamic bidirectional coupling model is the core model for achieving accurate simulation of the drainage system in this invention. Specifically, it refers to a technical model that can simultaneously simulate the 1D drainage network flow movement and 2D regional water accumulation distribution in urban underground space, and realize the mutual influence and data interaction between the two. Its implementation involves: firstly, receiving multi-dimensional real-time data collected by a multi-source heterogeneous digital twin drainage intelligent management and control platform, including network water flow velocity and regional water accumulation depth, and importing the data into the 1D network simulation module and the 2D regional water accumulation simulation module respectively; setting a 1-minute time step during runtime, with the simulation duration adjusted according to flood conditions (30 minutes to 24 hours), and outputting intermediate results every 5 minutes; quantifying the water flow exchange between the 1D network and the 2D region through specific computational relationships, such as calculating the coupled water flow exchange by combining parameters such as network flow area and water flow velocity, and finally generating data such as network water flow pressure distribution and regional water accumulation diffusion trends. The role of this model is to break through the limitations of traditional single-dimensional simulation, accurately capture the dynamic coupling relationship between the network and water accumulation, and provide reliable data support for subsequent scheduling algorithms. Improving the simulation accuracy of drainage systems can prevent unreasonable scheduling plans due to simulation deviations, provide a scientific data basis for emergency drainage control, and ensure the accuracy and feasibility of the plan formulation.

[0052] The spatiotemporal feature-driven drainage scheduling algorithm is the core algorithm for constructing a scientific drainage scheduling scheme in this invention. Specifically, it refers to an algorithm that can extract the spatiotemporal correlation features of water flow and water accumulation diffusion processes, and generate and optimize drainage scheduling decision variables by combining emergency drainage parameters. Its implementation is as follows: First, extract the pipe network and water accumulation data for each minute within the past hour from the output data of the 1D-2D dynamic bidirectional coupling model, divide the spatial area into 500-square-meter regions, and form a spatiotemporal data matrix; use a 3-minute sliding window algorithm to divide the spatiotemporal segments, calculate the flow rate change rate, water depth change rate, and motion direction consistency coefficient, and construct a spatiotemporal feature vector; group the feature vectors using a clustering algorithm with a cluster size of 5 to determine the spatiotemporal scene pattern; construct a set of decision variables by combining drainage system structural parameters (such as the location of 25 pump sets and the flow rate of 8 emergency channels), and then filter them based on parameters such as the maximum pump head and the maximum pressure value of the pipe network. After model feedback verification, the parameters are optimized (e.g., feature weights are adjusted by 20%). The function of this algorithm is to accurately identify spatiotemporal dynamic features, generate scheduling decision variables that conform to the system's carrying capacity, and achieve adaptive adjustment of the scheduling scheme. This addresses the issues of traditional algorithms lacking spatiotemporal analysis and having rigid scheduling, improves the adaptability and reliability of scheduling schemes, provides a scientific basis for emergency drainage decisions, and ensures the efficient operation of drainage systems.

[0053] The multi-source heterogeneous digital twin intelligent drainage management platform is the core management carrier in this invention for data acquisition, processing, and command issuance. Specifically, it refers to a digital platform that can integrate data from multiple types of sensors, support model calculations and algorithm execution, and realize full-process management of emergency drainage. Its implementation is as follows: Through devices such as level sensors (5-meter intervals), flow velocity sensors (30-second sampling), and pump status monitors (10-second updates) deployed in underground spaces, multi-dimensional data such as water depth, pipe network flow velocity, and pump operating parameters are collected. After data cleaning and format conversion (transmission delay ≤ 5 seconds), the data is transmitted to a 1D-2D dynamic bidirectional coupling model. This supports feature extraction and decision variable screening for spatiotemporal feature-driven drainage scheduling algorithms, and receives model feedback verification results. Finally, control commands, including pump start / stop and valve opening adjustment, are sent to the executing devices via industrial Ethernet (100Mbps rate). Simultaneously, the command execution status is monitored; if the device does not respond, a backup 4G channel is activated to resend the command, ensuring device startup within 10 seconds. The platform's function is to achieve seamless data flow, collaborative operation of models and algorithms, and precise issuance of control commands, connecting all aspects of emergency drainage. It breaks through the limitations of traditional fragmented data and disjointed management, constructing an integrated emergency drainage control system to improve the response speed and execution efficiency of drainage control, and ensure operational safety in underground spaces during floods.

[0054] like Figure 2As shown, an emergency drainage control method for urban underground space under flood disasters is implemented through different units, including: a multi-dimensional data acquisition and transmission unit, a 1D-2D dynamic bidirectional coupling model calculation unit, a spatiotemporal feature-driven drainage scheduling algorithm processing unit, a drainage scheduling decision variable constraint screening unit, a drainage scheme feedback verification unit, and an emergency drainage control command generation and issuance unit. The multi-dimensional data acquisition and transmission unit executes operation S1, collecting real-time multi-dimensional data on water depth, water flow velocity, pump operation status, and surface runoff from sensors deployed at various monitoring points in the urban underground space. After data cleaning and format conversion, the data is transmitted to the 1D-2D dynamic bidirectional coupling model calculation unit. Upon receiving the data, the 1D-2D dynamic bidirectional coupling model calculation unit executes simulation calculation S2, analyzing the dynamic coupling relationship between 1D pipe network water flow movement and 2D regional water distribution. The system calculates and outputs data on the water pressure distribution and water accumulation diffusion trend in the pipe network. The spatiotemporal feature-driven drainage scheduling algorithm processing unit receives this output data, executes operation S3, extracts spatiotemporal correlation features, and constructs a set of drainage scheduling decision variables. The drainage scheduling decision variable constraint screening unit, combined with the preset parameters for emergency drainage control, executes constraint screening in S4 to obtain a subset of decision variables that meet the parameter requirements. The drainage scheme feedback verification unit inputs the subset into the 1D-2D dynamic bidirectional coupling model and executes feedback verification in S5. If the result does not meet the requirements, it sends a parameter adjustment signal to the spatiotemporal feature-driven drainage scheduling algorithm processing unit. If the result meets the requirements, it transmits a signal to the emergency drainage control command generation and distribution unit. After receiving the signal, the emergency drainage control command generation and distribution unit executes operation S6 to generate control commands including pump start / stop, valve adjustment, and channel switching, and distributes them to the emergency drainage execution equipment in the urban underground space.

[0055] This method for emergency drainage control in urban underground spaces during floods effectively overcomes the limitations of single-dimensional simulations, significantly improving data support and simulation accuracy. Through a multi-source heterogeneous digital twin intelligent drainage management platform, it comprehensively collects real-time data on underground water depth, flow velocity, and pump status, avoiding incomplete data collection. Combined with a 1D-2D dynamic bidirectional coupling model, it achieves dynamic coupling calculations of one-dimensional pipe network flow movement and two-dimensional regional water distribution, accurately capturing the mutual influence between the two. This significantly improves the accuracy of water diffusion trends and pipe network flow pressure prediction, providing a more scientific and reliable basis for subsequent drainage scheme development and completely solving the shortcoming of simulations failing to reflect complex flow coupling relationships.

[0056] In terms of drainage scheduling strategy optimization, this method effectively compensates for the shortcomings of existing algorithms in terms of adaptability and reliability, significantly improving the rationality and practicality of the scheduling scheme. By relying on spatiotemporal features to drive the drainage scheduling algorithm, it deeply mines the spatiotemporal correlation features of water flow and water accumulation at different times and locations, overcoming the limitations of algorithms lacking spatiotemporal dimension analysis. Simultaneously, by combining key parameters of emergency drainage control in underground spaces, it rigorously constrains and filters scheduling decision variables, and continuously adjusts and optimizes the scheme through a feedback verification mechanism. This ensures that the scheduling strategy can dynamically and adaptively adjust according to actual water flow and water accumulation, greatly improving the adaptability and reliability of the scheduling scheme and effectively solving the problem that existing algorithms cannot accurately match actual drainage needs.

[0057] In terms of emergency drainage execution efficiency, this method can comprehensively improve the flood response capabilities of underground spaces and effectively ensure the operational safety of underground spaces. Through a multi-source heterogeneous digital twin intelligent drainage management and control platform, the optimized drainage scheduling plan is transformed into precise control commands, which are quickly issued to emergency drainage execution equipment, including adjusting the number of pumps started and stopped, valve opening, and emergency channel switching, significantly improving the response speed and execution efficiency of emergency drainage. Compared with fixed-mode scheduling, it can more efficiently control the receding of accumulated water, minimize the threat of flood disasters to the safety of underground space equipment and personnel, significantly reduce economic losses and social impact, comprehensively enhance the overall ability of urban underground spaces to cope with flood disasters, and effectively overcome the shortcomings of slow response and low efficiency.

[0058] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," "link," and "fix" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0059] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A control method for urban underground space emergency drainage under flood disaster, characterized in that, Comprise the following steps: S1, through the multi-source heterogeneous digital twin drainage wisdom management and control platform, multi-dimensional real-time data related to the drainage system in the urban underground space is collected, the multi-dimensional real-time data covers the water depth of each region of underground space, the water flow velocity in the drainage pipe network, the running state of drainage pump group and the surrounding surface runoff;S2, the multi-dimensional real-time data collected in S1 is input into 1D-2D dynamic two-way coupling model, the dynamic coupling relationship between 1D pipe network water flow movement and 2D regional water distribution in underground space drainage system is simulated and calculated by using the model, and the water flow pressure distribution in pipe network and regional water diffusion trend data at different times are obtained;S3, based on the water flow pressure distribution in pipe network and regional water diffusion trend data obtained in S2, start the space-time feature driven drainage scheduling algorithm, the algorithm extracts the space-time correlation characteristics in the process of water flow movement and water diffusion, and constructs a set of drainage scheduling decision variables according to the space-time correlation characteristics;S4, in the process of running the space-time feature driven drainage scheduling algorithm, in combination with the preset parameters of the emergency drainage control of urban underground space, each variable in the set of drainage scheduling decision variables is constrained and selected, and the preset parameters include the maximum lift of drainage pump group, the maximum pressure value of pipe network and the threshold value corresponding to the target time length of water recession;S5, the drainage scheduling decision variables selected in S4 are input into 1D-2D dynamic two-way coupling model for feedback verification, to determine whether the pipe network water flow pressure is in the safe range and the regional water diffusion is effectively inhibited under the drainage scheme corresponding to the decision variable, if not, return to S3 to adjust the parameters of the space-time feature driven drainage scheduling algorithm;S6, when the feedback verification result in S5 meets the preset condition, send control instructions to the emergency drainage execution equipment in urban underground space through the multi-source heterogeneous digital twin drainage wisdom management and control platform, the control instructions include the start-stop number of drainage pump group, the adjustment range of valve opening degree and the emergency drainage channel switching instruction.

2. The urban underground space emergency drainage control method under flood disaster according to claim 1, characterized in that, In S2, the 1D-2D dynamic two-way coupling model is coupled with the 1D pipe network water flow movement and the 2D regional water distribution by the following formula: Wherein, Q 1D-2D (x, t) is the water flow exchange amount at the x position at the t time after the 1D-2D coupling, a is the influence coefficient of the 1D pipe network water flow on the coupling process, Q 1D (x, t) is the water flow at the x position in the 1D pipe network at the t time, h 2D (x, t) is the 2D regional water depth at the x position at the t time, is the 2D regional average water depth, b is the influence coefficient of the 2D regional water on the coupling process, v 2D (x, t) is the 2D regional water flow velocity at the x position at the t time, A 1D (x, t) is the 1D pipe network flow area at the x position at the t time.

3. The urban underground space emergency drainage control method under flood disaster according to claim 1, characterized in that, In S3, when the space-time feature driven drainage scheduling algorithm extracts the space-time correlation characteristics, the following formula is used to calculate the space-time correlation degree: wherein R(x1, x2, t1, t2) is the spatio-temporal correlation degree of x1 and x2 positions and t1 and t2 time, Q(x1, t i ) is the pipe network water flow at x1 position at t ) is the average pipe network water flow at x1 position, h(x2, t i ) is the water depth at x2 position at t ) is the average water depth at x2 position, n is the data sampling number, γ is the time attenuation coefficient, and δ is the space attenuation coefficient.

4. The urban underground space emergency drainage control method under flood disaster according to claim 1, characterized in that, In S4, when the urban underground space emergency drainage control parameters are combined to constrain and screen the drainage scheduling decision variables, the following formula is adopted: When G(P1, P2, P3)≤θ, the drainage scheduling decision variable is retained, wherein G(P1, P2, P3) is the comprehensive constraint value of the decision variable, P1 is the actual lift of the drainage pump set, P 1max is the maximum lift of the drainage pump set, P2 is the actual pressure-bearing value of the pipe network, P 2max is the maximum pressure-bearing value of the pipe network, P3 is the actual water accumulation recession time, P 3max is the threshold value corresponding to the water accumulation recession target time, μ1, μ2, μ3 are weight coefficients, and θ is a constraint threshold value.

5. The urban underground space emergency drainage control method under flood disaster according to claim 1, characterized in that, In S5, the 1D-2D dynamic two-way coupling model adopts the following formula to calculate the verification index when feedback verification is performed on the drainage scheme: Wherein, V is the verification index value, λ1 and λ2 are weight coefficients, max(P pipe (x, t)) is the maximum water flow pressure of each position in the pipe network at time t, P pipemax is the maximum pressure bearing value of the pipe network, max(h area (x, t)) is the maximum water depth of each position in the region at time t, h areamax is the maximum water depth allowed in the region, and when V≤1, it is determined that the feedback verification result meets the preset condition.

6. The urban underground space emergency drainage control method under flood disaster according to claim 1, characterized in that, In S6, before the multi-source heterogeneous digital twin drainage intelligent management and control platform sends the control instruction to the emergency drainage execution equipment, the instruction output accuracy is optimized through the following formula: Wherein, C opt is the optimized control instruction parameter, C init is the initial control instruction parameter, η is the instruction optimization coefficient, Q sim (x, t) is the water flow at x position at t time simulated by the 1D-2D dynamic two-way coupling model, Q act (x, t) is the actual water flow at x position at t time.

7. The urban underground space emergency drainage control method under flood disaster according to claim 1, characterized in that, S3 comprises the following steps: S31, extracting the water flow rate, water flow velocity, water depth and water spreading direction data of each monitoring point in each time interval from the pipe network water flow pressure distribution and regional water accumulation spreading tendency data output from S2, classifying and storing these data according to time sequence and spatial position to form a space-time data matrix; S32, extracting the space-time characteristics of the data in the space-time data matrix, dividing different space-time segments through a sliding window algorithm, calculating the water flow rate change rate, water depth change rate and consistency coefficient of water flow and water movement direction in each space-time segment, and constructing a space-time characteristic vector; S33, inputting the space-time characteristic vector into the feature analysis module of the space-time characteristic driven drainage scheduling algorithm, grouping the space-time characteristic vector through a clustering algorithm to determine the feature mode under different space-time scenarios, and providing a basis for subsequent construction of a drainage scheduling decision variable set; S34, based on the determined feature mode, combining the structural parameters of the urban underground space drainage system, including pipe network diameter, number of drainage pump groups and emergency passage location, and constructing a drainage scheduling decision variable set containing pump group operation parameters, valve adjustment parameters and passage switching parameters. 8.The method of claim 1, wherein, S4 comprises the following steps: S41, calling preset parameters from the urban underground space emergency drainage control parameter database, wherein the preset parameters include the minimum flow area of the drainage pipe network, the maximum traffic flow of the emergency drainage passage and the continuous operation time limit of the pump group in addition to the threshold values corresponding to the maximum lift of the drainage pump group, the maximum pressure bearing value of the pipe network and the water accumulation recession target time length; S42, comparing each decision variable in the drainage scheduling decision variable set with the called preset parameters one by one to determine the deviation degree of the actual value of each parameter corresponding to each decision variable from the preset parameters; S43, setting a constraint weight according to the deviation degree, giving a higher constraint weight to the decision variable corresponding to the parameter with larger deviation and a lower constraint weight to the decision variable corresponding to the parameter with smaller deviation; S44, comprehensively scoring each decision variable according to the constraint weight, screening out the decision variables with comprehensive scores meeting the preset requirements to form a screened drainage scheduling decision variable subset. 9.The method of claim 1, wherein, S5 comprises the following sub-steps: S51, converting each decision variable in the subset of drainage scheduling decision variables screened in S4 into a format of input parameters recognizable by the 1D-2D dynamic bidirectional coupling model, including converting pump set operation parameters into pump set flow coefficients in the model, and converting valve adjustment parameters into pipe network impedance coefficients in the model; S52, inputting the converted input parameters into the 1D-2D dynamic bidirectional coupling model, starting the model to perform simulation operation, and obtaining pipe network water flow pressure at each position, regional water depth at each position and water depth recession rate data output by the simulation operation; S53, comparing the simulation output data with preset safety standards, wherein the preset safety standards include a pipe network water flow pressure safety upper limit, a regional water depth safety upper limit and a water depth recession rate lower limit; S54, if the simulation output data meet the preset safety standards, determining that the drainage scheme corresponding to the decision variable is feasible; if not, recording the parameter type and deviation value that do not meet the standards, returning to S3 and adjusting the feature weight coefficient in the time-space feature driven drainage scheduling algorithm according to the deviation value.

Images