Real-time scheduling method and system for sewer network based on meta-modeling

Abstract

The application relates to a sewer network real-time scheduling method and system based on meta-modeling. The application constructs a dynamic change model of a sewer network by acquiring spatial structure information of the sewer network, simultaneously trains a graph neural network prediction model by using historical pump station operation data, realizes accurate prediction of the sewer network under real-time water inflow, and generates a scheduling strategy of each pump of the pump station based on comprehensive real-time state data and the prediction result, so that the scheduling decision can truly reflect the dynamic operation state of the sewer network under variable working conditions, the forward-looking and accuracy of the scheduling are effectively improved, sewerage efficiency reduction and overflow risk caused by state cognition lag are reduced, and the stability and safety of the overall sewerage system operation are improved.

Description

Technical Field

[0001] This application relates to the field of drainage technology, and in particular to a method and system for real-time scheduling of drainage networks based on meta-modeling. Background Technology

[0002] The efficient operation of urban drainage systems is crucial for ensuring public safety and environmental health. With the increasing frequency of extreme weather events due to climate change, urban drainage systems face increasingly severe challenges, especially during heavy rains. Overloading of drainage networks can easily lead to overflows and other problems, seriously affecting the normal operation of cities. Therefore, improving the scheduling efficiency of drainage systems and optimizing network operation strategies have become urgent issues to be addressed.

[0003] As a crucial infrastructure for urban water environment management, the operational status of drainage networks directly impacts urban drainage efficiency and environmental safety. Traditional drainage network scheduling relies on static planning and empirical models, which cannot respond in real-time to dynamically changing factors such as rainfall and flow rates, often leading to network instability or inefficiency. This is particularly true in large-scale urban drainage systems, where the complex connections between drainage pipes and significant fluctuations in network load make it difficult for existing scheduling systems to make accurate decisions under varying operating conditions. Summary of the Invention

[0004] Based on this, a real-time scheduling method and system for drainage pipe networks based on meta-modeling is provided to solve the technical problems of large-scale urban drainage systems failing to accurately reflect the operating status of drainage pipe networks under variable operating conditions, resulting in poor stability, low drainage efficiency, and overflow.

[0005] On the one hand, a real-time scheduling method for drainage pipe networks based on meta-modeling is provided, the method comprising: Obtain spatial structure information of the drainage pipe network, and construct a dynamic change model of the drainage pipe network based on the spatial structure information of the drainage pipe network; Historical pump station operation data of the drainage network is obtained, including forebay liquid level, outflow, pump start and stop status and rainfall data during the same period. A graph neural network is trained based on the historical pump station operation data to form a graph neural network prediction model. The current rainfall and the liquid level in the forebay of the drainage network are acquired in real time at the minute level as real-time status data. The real-time status data is then input into the graph neural network prediction model to obtain the inflow rate of each pumping station in the future time period. The decision variables, optimization objectives, and constraints for controlling the drainage network are determined. Based on the dynamic change model of the drainage network, a pump scheduling strategy for each pump station is generated according to the decision variables, optimization objectives, constraints, real-time status data, and future water flow of each pump station. The generated pump scheduling strategies for each pump in the pumping station are updated in real time and displayed on the system operation display page for users to view and select.

[0006] In one embodiment, the spatial structure information includes the topology of main pipes, branch pipes, nodes, pumping stations, and ancillary structures.

[0007] In one embodiment, constructing a dynamic change model of the drainage network based on the spatial structure information of the drainage network includes: The spatial structure information of the drainage network is analyzed to determine point features, line features, polygon features, coordinate system and attribute fields. A spatial data element model is constructed based on the coordinates of the point features, line features and polygon features in the coordinate system and the attribute fields. Obtain the data block text file of the drainage pipe network, parse the data block text file to determine the nodes, pipes, pumping stations and upstream and downstream connection relationships, take the nodes, pipes and pumping stations as basic elements, form the pipe network topology structure through the upstream and downstream connection relationships, and describe the function of the pipe network topology structure with hydraulic parameters and operating parameters to construct the engineering structure meta-model; Determine the element mapping between the spatial data meta-model and the engineering structure meta-model; The system obtains user settings for basic elements and structural attributes of the pipeline network based on operating conditions, adjusts the drainage system structure in the spatial data element model and the engineering structure element model according to the settings, reads the spatial structure information file of the drainage pipeline network, identifies all elements according to the spatial data element model, and converts them into data block text files according to preset mapping rules to obtain the modified pipeline network mechanism model. The aforementioned pipeline mechanism model constructs a dynamic change model of the drainage pipeline network based on the liquid level in the pump station forebay under different rainfall amounts.

[0008] In one embodiment, determining the element mapping between the spatial data meta-model and the engineering structure meta-model includes: By mapping the start point ID and end point ID of the point features in the spatial data meta-model to the upstream and downstream nodes of the pipeline in the engineering structure meta-model; The length, slope, and pipe diameter attributes of the line elements in the spatial data element model are mapped to the length, slope, and pipe diameter parameters of the pipes in the engineering structure element model. Based on the surface features and coordinate system of the spatial data element model, the elevation information of the point features is mapped to the node elevation parameters of the engineering structure element model.

[0009] In one embodiment, determining the decision variables, optimization objectives, and constraints for controlling the drainage network, and generating a pump scheduling strategy for each pumping station based on the dynamic change model of the drainage network according to the decision variables, the optimization objectives, the constraints, the real-time status data, and the future inflow rate of each pumping station, includes: The start and stop liquid levels of each pump at each pumping station within the control cycle are used as decision variables. If the liquid level in the forebay exceeds the pump start level, one pump is added. If the liquid level in the forebay is lower than the pump stop level, one pump is shut down. The start and stop liquid level strategy of each pump at the pumping station remains unchanged within each control cycle. The pumping station is allowed to change the operating status of each pump according to the start and stop liquid level strategy within each control cycle. The start and stop liquid level strategy of each pump at the pumping station is adjusted at the boundary time of the next cycle. Optimization objectives are set for both dry and rainy weather conditions, and objective functions are set based on these optimization objectives. Constraints are set based on hydraulic constraints, water quality constraints, and operational safety constraints; Based on the dynamic change model of the drainage network, the non-dominated sorting genetic algorithm II is used to solve the optimal scheduling strategy according to the decision variables, the optimization objective, the constraints, the real-time status data and the inflow of water to each pumping station in future time periods. The optimal scheduling strategy is used to generate scheduling strategies for each pump in the pumping station.

[0010] In one embodiment, the step of setting optimization objectives based on two different working conditions—dry day and rainy day—and setting an objective function based on the optimization objectives includes: The optimization objectives for dry weather are set as minimizing the number of pump station start-ups and shutdowns and maintaining a stable influent flow rate at the wastewater treatment plant. The objective function corresponding to the optimization objectives for dry weather is: , ;in, This refers to the number of times the i-pump is turned on and off; It is the instantaneous flow rate delivered from the pipeline network to the sewage treatment plant at time t; It is the average inflow rate within the control cycle; The optimization objectives for rainy days are set as minimizing the discharge from the branch pumping station into the river and minimizing the number of pump starts. The objective function corresponding to the optimization objectives for rainy days is: , , , ;in, It is the time step of SWMM reporting; It is the instantaneous flow rate discharged directly from the pipeline to the water body at time t; , It is a first-order difference; , It is the height of the peak flow rate; , It is an indicator function; it returns 1 if the condition is met, and 0 otherwise. It is the peak detection threshold.

[0011] In one embodiment, the dynamic change model based on the drainage network employs a non-dominated sorting genetic algorithm II to solve for the optimal scheduling strategy based on the decision variables, the optimization objective, the constraints, the real-time status data, and the future inflow rate of each pumping station, including: An initial population is constructed using a non-dominated sorting genetic algorithm II based on the decision variables, the optimization objective, the constraints, the real-time status data, and the future inflow rate of each pumping station. Subpopulations are then generated based on the initial population. Determine whether the initial population has generated a first generation of child populations. If so, merge the parent and child populations to generate a new parent population. Otherwise, perform non-dominated sorting on the initial population and then perform selection, crossover, and mutation processing to update the initial population. Determine whether a new parent population has been generated. If so, perform selection, crossover, and mutation on the new parent population. Otherwise, perform non-dominated sorting on the new parent population and calculate the crowding degree. Determine whether each individual is selected into the next generation of subpopulation based on the crowding degree of each individual. Determine if the number is less than the maximum generation. If it is, increment the generation by one. Otherwise, terminate the subpopulation and output the current parent population.

[0012] In one embodiment, performing the non-dominated sort includes: The population is divided into different levels by non-dominated sorting. The first level contains all solutions that are not dominated by other individuals. As the non-dominated sorting proceeds, the second level contains all solutions that are dominated by the solutions in the first level, and so on. For each solution and , and For the corresponding objective function, the dominance relation is defined as follows: and Then it is called Dominate .

[0013] On the other hand, a real-time scheduling system for drainage pipe networks based on meta-modeling is provided, the system comprising: The pipeline dynamic modeling module is used to acquire the spatial structure information of the drainage pipeline network and construct a dynamic change model of the drainage pipeline network based on the spatial structure information of the drainage pipeline network. The prediction model training module is used to acquire historical pump station operation data of the drainage network. The historical pump station operation data includes forebay liquid level, outflow rate, pump start and stop status and rainfall data during the same period. The graph neural network is trained based on the historical pump station operation data to form a graph neural network prediction model. The real-time status prediction module is used to acquire the minute-level current rainfall and pump station forebay liquid level of the drainage network as real-time status data, and input the real-time status data into the graph neural network prediction model to obtain the inflow rate of each pump station in the future period. The optimization scheduling generation module is used to determine the decision variables, optimization objectives, and constraints for controlling the drainage network. Based on the dynamic change model of the drainage network, it generates scheduling strategies for each pump at each pumping station according to the decision variables, optimization objectives, constraints, real-time status data, and future inflow rates for each pumping station. The scheduling strategy display module is used to update and display the generated scheduling strategies for each pump in the pumping station in real time on the system operation display page for users to view and select.

[0014] The aforementioned real-time scheduling method and system for drainage networks based on meta-modeling constructs a dynamic change model of the drainage network by acquiring its spatial structure information. Simultaneously, it trains a graph neural network prediction model using historical pump station operation data to accurately predict the real-time inflow of water into the drainage network. Based on this, it integrates real-time status data and prediction results to generate scheduling strategies for each pump in the pumping stations. This ensures that scheduling decisions truly reflect the dynamic operating status of the drainage network under varying conditions, effectively improving the foresight and accuracy of scheduling, reducing drainage efficiency decline and overflow risks caused by delayed state perception, and enhancing the overall stability and safety of the drainage system. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating a real-time scheduling method for drainage pipe networks based on meta-modeling in one embodiment of this application. Figure 2 This is a structural block diagram of a real-time scheduling system for drainage pipe networks based on meta-modeling in one embodiment of this application. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0018] In one embodiment, such as Figure 1 As shown, a real-time scheduling method for drainage pipe networks based on meta-modeling is provided, including the following steps: Step S1: Obtain the spatial structure information of the drainage pipe network, and construct a dynamic change model of the drainage pipe network based on the spatial structure information of the drainage pipe network; Step S2: Obtain historical pump station operation data of the drainage network. The historical pump station operation data includes forebay liquid level, outflow rate, pump start-up and shut-off status and rainfall data during the same period. Train a graph neural network to form a graph neural network prediction model based on the historical pump station operation data. Step S3: Real-time acquisition of minute-level current rainfall and pump station forebay liquid level of the drainage network as real-time status data, and input of the real-time status data into the graph neural network prediction model to obtain the inflow rate of each pump station in the future period. Step S4: Determine the decision variables, optimization objectives, and constraints for controlling the drainage network. Based on the dynamic change model of the drainage network, generate a pump scheduling strategy for each pump station according to the decision variables, optimization objectives, constraints, real-time status data, and future water flow of each pump station. Step S5: The generated pump scheduling strategies for each pump in the pumping station are updated in real time and displayed on the system operation display page for users to view and select.

[0019] Specifically, by acquiring spatial structure information of the drainage network, a dynamic change model of the drainage network is constructed. At the same time, a graph neural network prediction model is trained using historical pump station operation data to accurately predict the real-time inflow of water into the drainage network. Based on this, the scheduling strategy for each pump in the pump station is generated by integrating real-time status data and prediction results. This enables the scheduling decision to truly reflect the dynamic operating status of the drainage network under changing conditions, effectively improving the foresight and accuracy of scheduling, reducing the decline in drainage efficiency and overflow risk caused by lag in status perception, and improving the stability and safety of the overall drainage system operation.

[0020] In this embodiment, the spatial structure information includes the topology of the main pipe, branch pipe, nodes, pumping stations, and ancillary structures.

[0021] In this embodiment, constructing a dynamic change model of the drainage network based on the spatial structure information of the drainage network includes: The spatial structure information of the drainage network is analyzed to determine point features, line features, polygon features, coordinate system and attribute fields. A spatial data element model is constructed based on the coordinates of the point features, line features and polygon features in the coordinate system and the attribute fields. Obtain the data block text file of the drainage pipe network, parse the data block text file to determine the nodes, pipes, pumping stations and upstream and downstream connection relationships, take the nodes, pipes and pumping stations as basic elements, form the pipe network topology structure through the upstream and downstream connection relationships, and describe the function of the pipe network topology structure with hydraulic parameters and operating parameters to construct the engineering structure meta-model; Determine the element mapping between the spatial data meta-model and the engineering structure meta-model; The system obtains user settings for basic elements and structural attributes of the pipeline network based on operating conditions, adjusts the drainage system structure in the spatial data element model and the engineering structure element model according to the settings, reads the spatial structure information file of the drainage pipeline network, identifies all elements according to the spatial data element model, and converts them into data block text files according to preset mapping rules to obtain the modified pipeline network mechanism model. The aforementioned pipeline mechanism model constructs a dynamic change model of the drainage pipeline network based on the liquid level in the pump station forebay under different rainfall amounts.

[0022] Specifically, by constructing spatial data meta-models and engineering structure meta-models separately, and establishing element mapping relationships between the two, the spatial structure information and engineering mechanism information of the drainage network can be uniformly expressed and collaboratively updated. At the same time, it supports flexible adjustment of the network structure based on user configuration according to operating conditions, ultimately forming a network mechanism model that is highly consistent with the actual drainage system. Based on this, a dynamic change model is constructed, thereby significantly improving the adaptability of the drainage network model to changes in the real engineering structure and operating conditions, enabling subsequent scheduling calculations to be based on a more accurate and updatable system model.

[0023] In this embodiment, determining the element mapping between the spatial data meta-model and the engineering structure meta-model includes: By mapping the start point ID and end point ID of the point features in the spatial data meta-model to the upstream and downstream nodes of the pipeline in the engineering structure meta-model; The length, slope, and pipe diameter attributes of the line elements in the spatial data element model are mapped to the length, slope, and pipe diameter parameters of the pipes in the engineering structure element model. Based on the surface features and coordinate system of the spatial data element model, the elevation information of the point features is mapped to the node elevation parameters of the engineering structure element model.

[0024] Specifically, by accurately mapping the key attributes of point, line, and surface elements in the spatial data element model to nodes, pipes, and their hydraulic parameters in the engineering structure element model, a consistent expression between the spatial geometric features of the drainage network and the hydraulic calculation parameters is achieved. This avoids model distortion caused by inconsistent data sources or parameter conversion errors, thereby improving the accuracy of the network mechanism model in hydraulic calculation and operation simulation, and providing reliable basic support for dynamic change models and scheduling optimization.

[0025] In this embodiment, determining the decision variables, optimization objectives, and constraints for controlling the drainage network, and generating pump scheduling strategies for each pumping station based on the dynamic change model of the drainage network according to the decision variables, optimization objectives, constraints, real-time status data, and future inflow rates for each pumping station, includes: The start and stop liquid levels of each pump at each pumping station within the control cycle are used as decision variables. If the liquid level in the forebay exceeds the pump start level, one pump is added. If the liquid level in the forebay is lower than the pump stop level, one pump is shut down. The start and stop liquid level strategy of each pump at the pumping station remains unchanged within each control cycle. The pumping station is allowed to change the operating status of each pump according to the start and stop liquid level strategy within each control cycle. The start and stop liquid level strategy of each pump at the pumping station is adjusted at the boundary time of the next cycle. Optimization objectives are set for both dry and rainy weather conditions, and objective functions are set based on these optimization objectives. Constraints are set based on hydraulic constraints, water quality constraints, and operational safety constraints; Based on the dynamic change model of the drainage network, the non-dominated sorting genetic algorithm II is used to solve the optimal scheduling strategy according to the decision variables, the optimization objective, the constraints, the real-time status data and the inflow of water to each pumping station in future time periods. The optimal scheduling strategy is used to generate scheduling strategies for each pump in the pumping station.

[0026] This method involves using the start-stop liquid level of each pump in the pumping station as a decision variable and maintaining a stable start-stop strategy within the control cycle. Optimization objectives and constraints are set according to dry and rainy weather conditions, and the optimal scheduling strategy is solved using a non-dominated sorting genetic algorithm II, combined with a dynamic change model of the drainage network and predicted inflow. This enables the pumping station scheduling to achieve multi-objective coordinated optimization while meeting hydraulic, water quality, and operational safety constraints. This effectively reduces the frequency of pumping station start-stop operations, balances system load, reduces overflow risk, and improves the overall operational performance of the drainage system under different operating conditions.

[0027] In this embodiment, the step of setting optimization objectives based on two working conditions—dry day and rainy day—and setting an objective function based on the optimization objectives includes: The optimization objectives for dry weather are set as minimizing the number of pump station start-ups and shutdowns and maintaining a stable influent flow rate at the wastewater treatment plant. The objective function corresponding to the optimization objectives for dry weather is: , ;in, This refers to the number of times the i-pump is turned on and off; It is the instantaneous flow rate transported from the pipeline network to the sewage treatment plant at time t. ); It is the average inflow rate during the control cycle. ).

[0028] In this embodiment, the step of setting optimization objectives based on two working conditions—dry day and rainy day—and setting an objective function based on the optimization objectives includes: The optimization objectives for rainy days are set as minimizing the discharge from the branch pumping station into the river and minimizing the number of pump starts. The objective function corresponding to the optimization objectives for rainy days is: , , , ;in, It is the time step of SWMM reporting; It is the instantaneous flow rate discharged directly from the pipeline to the water body at time t. ); , It is a first-order difference ( ); , It is the peak flow height ( ); , It is an indicator function; it returns 1 if the condition is met, and 0 otherwise. It is the peak detection threshold.

[0029] Specifically, by setting objective functions for dry weather conditions to minimize the number of pump station start-ups and shutdowns and stabilize the influent flow of the sewage treatment plant, and for rainy weather conditions to minimize the discharge of branch pump stations into the river and minimize the number of pump start-ups, the scheduling optimization can be differentiated according to the actual needs of different operating scenarios. This reduces equipment wear and ensures the stable operation of the sewage treatment system under dry weather conditions, and effectively suppresses flow spikes and direct discharge into the river under rainy weather conditions, thereby enhancing the drainage system's ability to cope with sudden rainfall.

[0030] In this embodiment, the dynamic change model based on the drainage network employs a non-dominated sorting genetic algorithm II to solve for the optimal scheduling strategy based on the decision variables, the optimization objective, the constraints, the real-time status data, and the future inflow of water to each pumping station, including: An initial population is constructed using a non-dominated sorting genetic algorithm II based on the decision variables, the optimization objective, the constraints, the real-time status data, and the future inflow rate of each pumping station. Subpopulations are then generated based on the initial population. Determine whether the initial population has generated a first generation of child populations. If so, merge the parent and child populations to generate a new parent population. Otherwise, perform non-dominated sorting on the initial population and then perform selection, crossover, and mutation processing to update the initial population. Determine whether a new parent population has been generated. If so, perform selection, crossover, and mutation on the new parent population. Otherwise, perform non-dominated sorting on the new parent population and calculate the crowding degree. Determine whether each individual is selected into the next generation of subpopulation based on the crowding degree of each individual. Determine if the number is less than the maximum generation. If it is, increment the generation by one. Otherwise, terminate the subpopulation and output the current parent population.

[0031] In this study, a non-dominated sorting genetic algorithm II was introduced to solve the scheduling strategy. By using the methods of initializing the population, merging parent and offspring, non-dominated sorting, crowding calculation, and iterative reproduction, the optimal solution set was gradually approached under multiple objectives and constraints. This enabled the scheduling optimization process to maintain good global search capability and solution set diversity under complex nonlinear conditions, thereby obtaining a more stable pump station scheduling strategy that meets the actual engineering needs and improving the reliability and feasibility of real-time scheduling results.

[0032] In this embodiment, the non-dominated sorting includes: The population is divided into different levels by non-dominated sorting. The first level contains all solutions that are not dominated by other individuals. As the non-dominated sorting proceeds, the second level contains all solutions that are dominated by the solutions in the first level, and so on. For each solution and , and For the corresponding objective function, the dominance relation is defined as follows: and Then it is called Dominate .

[0033] In this study, by clarifying the rules for determining the dominance relationship in non-dominated sorting and performing hierarchical sorting of individuals in the population based on multiple objective functions, the algorithm can accurately distinguish the superiority or inferiority of different scheduling strategies under multi-objective conditions during the optimization process. This ensures that the algorithm takes into account the balance between various optimization objectives when selecting and retaining excellent solutions, avoids a single objective dominating the scheduling results, and improves the overall performance of the final scheduling strategy in terms of drainage efficiency, operational stability, and security.

[0034] In the aforementioned real-time scheduling method for drainage networks based on meta-modeling, a dynamic change model of the drainage network is constructed by acquiring the spatial structure information of the network. Simultaneously, a graph neural network prediction model is trained using historical pump station operation data to accurately predict the real-time inflow of water into the drainage network. Based on this, scheduling strategies for each pump in the pump station are generated by integrating real-time status data and prediction results. This enables scheduling decisions to truly reflect the dynamic operating status of the drainage network under varying conditions, effectively improving the foresight and accuracy of scheduling, reducing the decline in drainage efficiency and overflow risk caused by lag in state perception, and enhancing the stability and safety of the overall drainage system operation.

[0035] In one embodiment, such as Figure 2 As shown, a real-time scheduling system 10 for drainage pipe network based on meta-modeling is provided, including: a pipe network dynamic modeling module 1, a prediction model training module 2, a real-time status prediction module 3, an optimized scheduling generation module 4, and a scheduling strategy display module 5.

[0036] The pipeline dynamic modeling module 1 is used to obtain the spatial structure information of the drainage pipeline network and construct a dynamic change model of the drainage pipeline network based on the spatial structure information of the drainage pipeline network.

[0037] The prediction model training module 2 is used to obtain historical pump station operation data of the drainage network. The historical pump station operation data includes forebay liquid level, outflow, pump start-up and shut-off status and concurrent rainfall data. Based on the historical pump station operation data, a graph neural network is trained to form a graph neural network prediction model.

[0038] The real-time status prediction module 3 is used to acquire the minute-level current rainfall and pump station forebay level of the drainage network as real-time status data, and input the real-time status data into the graph neural network prediction model to obtain the inflow rate of each pump station in the future period.

[0039] The optimization scheduling generation module 4 is used to determine the decision variables, optimization objectives, and constraints for controlling the drainage network. Based on the dynamic change model of the drainage network, it generates a scheduling strategy for each pump at each pumping station according to the decision variables, optimization objectives, constraints, real-time status data, and the inflow rate of each pumping station in future time periods.

[0040] The scheduling strategy display module 5 is used to update the generated scheduling strategies for each pump in the pumping station in real time and display them on the system operation display page for users to view and select.

[0041] In this embodiment, the spatial structure information includes the topology of the main pipe, branch pipe, nodes, pumping stations, and ancillary structures.

[0042] In this embodiment, constructing a dynamic change model of the drainage network based on the spatial structure information of the drainage network includes: The spatial structure information of the drainage network is analyzed to determine point features, line features, polygon features, coordinate system and attribute fields. A spatial data element model is constructed based on the coordinates of the point features, line features and polygon features in the coordinate system and the attribute fields. Obtain the data block text file of the drainage pipe network, parse the data block text file to determine the nodes, pipes, pumping stations and upstream and downstream connection relationships, take the nodes, pipes and pumping stations as basic elements, form the pipe network topology structure through the upstream and downstream connection relationships, and describe the function of the pipe network topology structure with hydraulic parameters and operating parameters to construct the engineering structure meta-model; Determine the element mapping between the spatial data meta-model and the engineering structure meta-model; The system obtains user settings for basic elements and structural attributes of the pipeline network based on operating conditions, adjusts the drainage system structure in the spatial data element model and the engineering structure element model according to the settings, reads the spatial structure information file of the drainage pipeline network, identifies all elements according to the spatial data element model, and converts them into data block text files according to preset mapping rules to obtain the modified pipeline network mechanism model. The aforementioned pipeline mechanism model constructs a dynamic change model of the drainage pipeline network based on the liquid level in the pump station forebay under different rainfall amounts.

[0043] In this embodiment, determining the element mapping between the spatial data meta-model and the engineering structure meta-model includes: By mapping the start point ID and end point ID of the point features in the spatial data meta-model to the upstream and downstream nodes of the pipeline in the engineering structure meta-model; The length, slope, and pipe diameter attributes of the line elements in the spatial data element model are mapped to the length, slope, and pipe diameter parameters of the pipes in the engineering structure element model. Based on the surface features and coordinate system of the spatial data element model, the elevation information of the point features is mapped to the node elevation parameters of the engineering structure element model.

[0044] In this embodiment, determining the decision variables, optimization objectives, and constraints for controlling the drainage network, and generating pump scheduling strategies for each pumping station based on the dynamic change model of the drainage network according to the decision variables, optimization objectives, constraints, real-time status data, and future inflow rates for each pumping station, includes: The start and stop liquid levels of each pump at each pumping station within the control cycle are used as decision variables. If the liquid level in the forebay exceeds the pump start level, one pump is added. If the liquid level in the forebay is lower than the pump stop level, one pump is shut down. The start and stop liquid level strategy of each pump at the pumping station remains unchanged within each control cycle. The pumping station is allowed to change the operating status of each pump according to the start and stop liquid level strategy within each control cycle. The start and stop liquid level strategy of each pump at the pumping station is adjusted at the boundary time of the next cycle. Optimization objectives are set for both dry and rainy weather conditions, and objective functions are set based on these optimization objectives. Constraints are set based on hydraulic constraints, water quality constraints, and operational safety constraints; Based on the dynamic change model of the drainage network, the non-dominated sorting genetic algorithm II is used to solve the optimal scheduling strategy according to the decision variables, the optimization objective, the constraints, the real-time status data and the inflow of water to each pumping station in future time periods. The optimal scheduling strategy is used to generate scheduling strategies for each pump in the pumping station.

[0045] In this embodiment, the step of setting optimization objectives based on two working conditions—dry day and rainy day—and setting an objective function based on the optimization objectives includes: The optimization objectives for dry weather are set as minimizing the number of pump station start-ups and shutdowns and maintaining a stable influent flow rate at the wastewater treatment plant. The objective function corresponding to the optimization objectives for dry weather is: , ;in, This refers to the number of times the i-pump is turned on and off; It is the instantaneous flow rate transported from the pipeline network to the sewage treatment plant at time t. ); It is the average inflow rate during the control cycle. ); The optimization objectives for rainy days are set as minimizing the discharge from the branch pumping station into the river and minimizing the number of pump starts. The objective function corresponding to the optimization objectives for rainy days is: , , , ;in, It is the time step of SWMM reporting; It is the instantaneous flow rate discharged directly from the pipeline to the water body at time t. ); , It is a first-order difference ( ); , It is the peak flow height ( ); , It is an indicator function; it returns 1 if the condition is met, and 0 otherwise. It is the peak detection threshold.

[0046] In this embodiment, the dynamic change model based on the drainage network employs a non-dominated sorting genetic algorithm II to solve for the optimal scheduling strategy based on the decision variables, the optimization objective, the constraints, the real-time status data, and the future inflow of water to each pumping station, including: An initial population is constructed using a non-dominated sorting genetic algorithm II based on the decision variables, the optimization objective, the constraints, the real-time status data, and the future inflow rate of each pumping station. Subpopulations are then generated based on the initial population. Determine whether the initial population has generated a first generation of child populations. If so, merge the parent and child populations to generate a new parent population. Otherwise, perform non-dominated sorting on the initial population and then perform selection, crossover, and mutation processing to update the initial population. Determine whether a new parent population has been generated. If so, perform selection, crossover, and mutation on the new parent population. Otherwise, perform non-dominated sorting on the new parent population and calculate the crowding degree. Determine whether each individual is selected into the next generation of subpopulation based on the crowding degree of each individual. Determine if the number is less than the maximum generation. If it is, increment the generation by one. Otherwise, terminate the subpopulation and output the current parent population.

[0047] In this embodiment, the non-dominated sorting includes: The population is divided into different levels by non-dominated sorting. The first level contains all solutions that are not dominated by other individuals. As the non-dominated sorting proceeds, the second level contains all solutions that are dominated by the solutions in the first level, and so on. For each solution and , and For the corresponding objective function, the dominance relation is defined as follows: and Then it is called Dominate .

[0048] The aforementioned real-time scheduling system for drainage networks based on meta-modeling constructs a dynamic change model of the drainage network by acquiring its spatial structure information. Simultaneously, it trains a graph neural network prediction model using historical pump station operation data to accurately predict the real-time inflow of water into the drainage network. Based on this, it integrates real-time status data and prediction results to generate scheduling strategies for each pump at the pumping stations. This ensures that scheduling decisions accurately reflect the dynamic operating status of the drainage network under varying conditions, effectively improving the foresight and accuracy of scheduling, reducing drainage efficiency decline and overflow risks caused by delayed state perception, and enhancing the overall stability and safety of the drainage system.

[0049] Specific limitations regarding the meta-modeling-based real-time scheduling system for drainage networks can be found in the limitations of the meta-modeling-based real-time scheduling method for drainage networks described above, and will not be repeated here. Each module in the aforementioned meta-modeling-based real-time scheduling system for drainage networks can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0050] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0051] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A real-time scheduling method for drainage pipe networks based on meta-modeling, characterized in that, include: Obtain spatial structure information of the drainage pipe network, and construct a dynamic change model of the drainage pipe network based on the spatial structure information of the drainage pipe network; Historical pump station operation data of the drainage network is obtained, including forebay liquid level, outflow, pump start and stop status and rainfall data during the same period. A graph neural network is trained based on the historical pump station operation data to form a graph neural network prediction model. The current rainfall and the liquid level in the forebay of the drainage network are acquired in real time at the minute level as real-time status data. The real-time status data is then input into the graph neural network prediction model to obtain the inflow rate of each pumping station in the future time period. The decision variables, optimization objectives, and constraints for controlling the drainage network are determined. Based on the dynamic change model of the drainage network, a pump scheduling strategy for each pump station is generated according to the decision variables, optimization objectives, constraints, real-time status data, and future water flow of each pump station. The generated pump scheduling strategies for each pump in the pumping station are updated in real time and displayed on the system operation display page for users to view and select.

2. The real-time scheduling method for drainage pipe networks based on meta-modeling according to claim 1, characterized in that, The spatial structure information includes the topology of the main pipes, branch pipes, nodes, pumping stations, and ancillary structures.

3. The real-time scheduling method for drainage pipe networks based on meta-modeling according to claim 1, characterized in that, The step of constructing a dynamic change model of the drainage network based on the spatial structure information of the drainage network includes: The spatial structure information of the drainage network is analyzed to determine point features, line features, polygon features, coordinate system and attribute fields. A spatial data element model is constructed based on the coordinates of the point features, line features and polygon features in the coordinate system and the attribute fields. Obtain the data block text file of the drainage pipe network, parse the data block text file to determine the nodes, pipes, pumping stations and upstream and downstream connection relationships, take the nodes, pipes and pumping stations as basic elements, form the pipe network topology structure through the upstream and downstream connection relationships, and describe the function of the pipe network topology structure with hydraulic parameters and operating parameters to construct the engineering structure meta-model; Determine the element mapping between the spatial data meta-model and the engineering structure meta-model; The system obtains user settings for basic elements and structural attributes of the pipeline network based on operating conditions, adjusts the drainage system structure in the spatial data element model and the engineering structure element model according to the settings, reads the spatial structure information file of the drainage pipeline network, identifies all elements according to the spatial data element model, and converts them into data block text files according to preset mapping rules to obtain the modified pipeline network mechanism model. The aforementioned pipeline mechanism model constructs a dynamic change model of the drainage pipeline network based on the liquid level in the pump station forebay under different rainfall amounts.

4. The real-time scheduling method for drainage pipe networks based on meta-modeling according to claim 3, characterized in that, Determining the element mapping between the spatial data meta-model and the engineering structure meta-model includes: By mapping the start point ID and end point ID of the point features in the spatial data meta-model to the upstream and downstream nodes of the pipeline in the engineering structure meta-model; The length, slope, and pipe diameter attributes of the line elements in the spatial data element model are mapped to the length, slope, and pipe diameter parameters of the pipes in the engineering structure element model. Based on the surface features and coordinate system of the spatial data element model, the elevation information of the point features is mapped to the node elevation parameters of the engineering structure element model.

5. The real-time scheduling method for drainage pipe networks based on meta-modeling according to claim 1, characterized in that, The process of determining the decision variables, optimization objectives, and constraints used to control the drainage network, and generating pump scheduling strategies for each pumping station based on the dynamic change model of the drainage network according to the decision variables, optimization objectives, constraints, real-time status data, and future inflow rates for each pumping station, includes: The start and stop liquid levels of each pump at each pumping station within the control cycle are used as decision variables. If the liquid level in the forebay exceeds the pump start level, one pump is added. If the liquid level in the forebay is lower than the pump stop level, one pump is shut down. The start and stop liquid level strategy of each pump at the pumping station remains unchanged within each control cycle. The pumping station is allowed to change the operating status of each pump according to the start and stop liquid level strategy within each control cycle. The start and stop liquid level strategy of each pump at the pumping station is adjusted at the boundary time of the next cycle. Optimization objectives are set for both dry and rainy weather conditions, and objective functions are set based on these optimization objectives. Constraints are set based on hydraulic constraints, water quality constraints, and operational safety constraints; Based on the dynamic change model of the drainage network, the non-dominated sorting genetic algorithm II is used to solve the optimal scheduling strategy according to the decision variables, the optimization objective, the constraints, the real-time status data and the inflow of water to each pumping station in future time periods. The optimal scheduling strategy is used to generate scheduling strategies for each pump in the pumping station.

6. The real-time scheduling method for drainage pipe networks based on meta-modeling according to claim 5, characterized in that, The optimization objectives are set separately for dry and rainy weather conditions, and the objective function is set according to the optimization objectives, including: The optimization objectives for dry weather are set as minimizing the number of pump station start-ups and shutdowns and maintaining a stable influent flow rate at the wastewater treatment plant. The objective function corresponding to the optimization objectives for dry weather is: , ;in, This refers to the number of times the i-pump is turned on and off; It is the instantaneous flow rate delivered from the pipeline network to the sewage treatment plant at time t; It is the average inflow rate within the control cycle.

7. The real-time scheduling method for drainage pipe networks based on meta-modeling according to claim 5, characterized in that, The optimization objectives are set separately for dry and rainy weather conditions, and the objective function is set according to the optimization objectives, including: The optimization objectives for rainy days are set as minimizing the discharge from the branch pumping station into the river and minimizing the number of pump starts. The objective function corresponding to the optimization objectives for rainy days is: , , , ;in, It is the time step of SWMM reporting; It is the instantaneous flow rate discharged directly from the pipeline to the water body at time t; , It is a first-order difference; , It is the height of the peak flow rate; , It is an indicator function; it returns 1 if the condition is met, and 0 otherwise. It is the peak detection threshold.

8. The real-time scheduling method for drainage pipe networks based on meta-modeling according to claim 1, characterized in that, The dynamic change model based on the drainage network employs a non-dominated sorting genetic algorithm II to solve for the optimal scheduling strategy based on the decision variables, the optimization objective, the constraints, the real-time status data, and the future inflow rate of each pumping station, including: An initial population is constructed using a non-dominated sorting genetic algorithm II based on the decision variables, the optimization objective, the constraints, the real-time status data, and the future inflow rate of each pumping station. Subpopulations are then generated based on the initial population. Determine whether the initial population has generated a first generation of child populations. If so, merge the parent and child populations to generate a new parent population. Otherwise, perform non-dominated sorting on the initial population and then perform selection, crossover, and mutation processing to update the initial population. Determine whether a new parent population has been generated. If so, perform selection, crossover, and mutation on the new parent population. Otherwise, perform non-dominated sorting on the new parent population and calculate the crowding degree. Determine whether each individual is selected into the next generation of subpopulation based on the crowding degree of each individual. Determine if the number is less than the maximum generation. If it is, increment the generation by one. Otherwise, terminate the subpopulation and output the current parent population.

9. The real-time scheduling method for drainage pipe networks based on meta-modeling according to claim 8, characterized in that, The process of performing non-dominated sorting includes: The population is divided into different levels by non-dominated sorting. The first level contains all solutions that are not dominated by other individuals. As the non-dominated sorting proceeds, the second level contains all solutions that are dominated by the solutions in the first level, and so on. For each solution and , and For the corresponding objective function, the dominance relation is defined as follows: and Then it is called Dominate .

10. A real-time scheduling system for drainage pipe networks based on meta-modeling, characterized in that, The system includes: The pipeline dynamic modeling module is used to acquire the spatial structure information of the drainage pipeline network and construct a dynamic change model of the drainage pipeline network based on the spatial structure information of the drainage pipeline network. The prediction model training module is used to acquire historical pump station operation data of the drainage network. The historical pump station operation data includes forebay liquid level, outflow rate, pump start and stop status and rainfall data during the same period. The graph neural network is trained based on the historical pump station operation data to form a graph neural network prediction model. The real-time status prediction module is used to acquire the minute-level current rainfall and pump station forebay liquid level of the drainage network as real-time status data, and input the real-time status data into the graph neural network prediction model to obtain the inflow rate of each pump station in the future period. The optimization scheduling generation module is used to determine the decision variables, optimization objectives, and constraints for controlling the drainage network. Based on the dynamic change model of the drainage network, it generates scheduling strategies for each pump at each pumping station according to the decision variables, optimization objectives, constraints, real-time status data, and future inflow rates for each pumping station. The scheduling strategy display module is used to update and display the generated scheduling strategies for each pump in the pumping station in real time on the system operation display page for users to view and select.

Images