Rainwater and sewage intelligent diversion method and system under smart water affairs

Abstract

This invention discloses a method and system for intelligent separation of rainwater and sewage under smart water management. The method includes the following steps: S1: Constructing a multi-source sensing dataset for rainwater and sewage separation. This invention systematically solves the core pain points of traditional rainwater and sewage separation technology through multi-dimensional sensing and all-element decision-making, spatiotemporal collaborative control, and closed-loop adaptive response mechanisms. By integrating water quality, hydrology, and plant load data, it overcomes the limitations of single-sensoring, accurately identifies runoff pollution characteristics, and dynamically matches sewage treatment plant capacity, avoiding the risks of accidental interception and overflow. Based on the pipeline topology and water flow migration time, it constructs an upstream and downstream collaborative scheduling strategy, realizing the transformation from single-point control to regional spatiotemporal linkage, effectively eliminating systemic conflicts. Through an integrated "source-network-plant" load balancing model and millisecond-level closed-loop emergency response, it prevents impact loads in real time and automatically responds to equipment failures or water quality anomalies, significantly improving system resilience and emergency response capabilities.

Description

Technical Field

[0001] This invention relates to the field of smart water management technology, and in particular to a method and system for intelligent separation of rainwater and sewage under smart water management. Background Technology

[0002] With the acceleration of urbanization and the advancement of ecological civilization construction, the problems of rainwater overflow pollution (CSO) and low influent concentration in sewage treatment plants are becoming increasingly prominent in areas with combined sewer systems or incomplete separation systems. Traditional rainwater and sewage separation control mainly relies on level sensors or simple rain gauges for start-stop control, lacking comprehensive analysis of water quality, water quantity and pipeline topology. During rainfall, how to accurately identify the degree of runoff pollution and dynamically optimize the pipeline flow direction has become the key to achieving "clean water into rivers and sewage into treatment plants".

[0003] While existing rainwater and sewage separation systems have achieved a certain degree of automation, they still have the following shortcomings in practical applications:

[0004] 1. Limited perception dimension: Relying solely on liquid level data ignores the rate of change in rainfall intensity, sudden changes in water quality (COD / SS) in the pipe network, and the real-time load capacity of downstream sewage treatment plants. This can easily lead to the accidental interception of clean rainwater during light rain or overflow due to delayed judgment during peak rainfall.

[0005] 2. The control logic is rigid and lacks spatiotemporal coordination. Traditional PLC control adopts the "single point, threshold method" and lacks coordination consideration of pipeline topology, water flow migration time and upstream and downstream nodes. When the upstream valve is opened to divert the flow, the downstream pipeline is prone to backflow or overflow due to the instantaneous surge in flow.

[0006] 3. Delayed emergency response: In the face of sudden changes in water quality or equipment failure, there is a lack of automatic emergency decision-making capabilities based on real-time situation, making it difficult to achieve a millisecond-level closed loop of "monitoring-decision-execution".

[0007] Based on the above, this application proposes a method and system for intelligent separation of rainwater and sewage under smart water management. Summary of the Invention

[0008] Based on the technical problems existing in the background technology, this invention proposes a method and system for intelligent separation of rainwater and sewage under smart water management.

[0009] The intelligent rainwater and sewage separation method under smart water management proposed in this invention includes the following steps:

[0010] S1: Construct a multi-source sensing dataset for rainwater and sewage separation: Collect hydrological and water quality data, meteorological monitoring data, and sewage treatment plant operation load data of key nodes in the drainage network to construct a basic data resource pool;

[0011] S2: Runoff pollution and overflow risk identification based on multi-dimensional features: fusion analysis of multi-source data to identify water quality mutation points and overflow risk points in the rainfall runoff process, and generate diversion control trigger signals;

[0012] S3: Coordinated scheduling decision based on pipeline topology and water flow migration time: Establish a pipeline hydraulic transmission model, calculate water flow migration time, coordinate the action sequence and opening degree of upstream and downstream valves, and generate the optimal diversion scheme;

[0013] S4: Construct an integrated load balancing model of "source-network-plant": connect with the influent load and treatment capacity of the sewage treatment plant in real time, dynamically adjust the total interception of the pipeline network, and achieve system-level supply and demand balance;

[0014] S5: Dynamic optimization of control strategy based on real-time feedback: The current control strategy is matched for similarity using a historical diversion case library, and the control model parameters are corrected online based on the real-time execution effect;

[0015] S6: Achieve closed-loop execution and emergency response for diversion control: Send control commands to the actuator, monitor valve status and pipeline response in real time, and automatically switch emergency bypass strategies for abnormal operating conditions.

[0016] Preferably, in S1, the hydrological and water quality data include pipeline liquid level, flow rate, flow velocity, and COD / SS concentration; the meteorological monitoring data includes rainfall intensity and cumulative rainfall; and the sewage treatment plant operating load data includes influent flow rate, dissolved oxygen concentration in the biological treatment tank, and excess sludge discharge.

[0017] Preferably, the specific logical steps of S2 are as follows:

[0018] S201: Multi-source data preprocessing and feature extraction: The raw data collected in S1 is denoised and standardized, and the following core feature parameters are extracted:

[0019] Rainfall characteristics: Current rainfall intensity Rainfall intensity change rate ;

[0020] Hydrological characteristics of the pipeline network: Pipeline fill rate (Liquid level / pipe diameter), rate of change in filling degree ;

[0021] Water quality characteristics: Chemical oxygen demand (COD) concentration Suspended solids concentration ;

[0022] S202: Calculation of Runoff Pollution Index: To distinguish between clean rainwater and heavily polluted initial rainwater, a runoff pollution index is introduced, which combines water quality concentration and water volume growth rate. The formula used is as follows:

[0023] ;

[0024] in The current runoff pollution index. This is the historical maximum concentration value (normalized). This represents the maximum rate of change of flow velocity in the pipeline (normalized). Weighting coefficients (usually) (This can be adjusted according to the regional water quality characteristics).

[0025] S203: Identifying abrupt changes in water quality: The sliding window algorithm is used to detect whether a step change has occurred in water quality. The judgment logic formula used is as follows:

[0026] ;

[0027] in To represent COD or SS sequences, To adjust the size of the sliding window (e.g., comparing the first 5 minutes with the first 10-5 minutes). This is the mutation threshold coefficient. The standard deviation is the historical data; if a sudden change is triggered, a high-priority shunt signal is generated directly.

[0028] S204: Predicting Pipeline Overflow Risk: Based on current pipe filling levels and rainfall intensity, predicting future... The prediction model formula used for the pipeline network status after 15 minutes (e.g., 15 minutes) is as follows:

[0029] ;

[0030] in To predict future liquid levels;

[0031] The overflow risk probability formula used is as follows:

[0032] ;

[0033] in The overflow critical level (e.g., 85% of the pipe diameter). The probability of overflow risk (between 0 and 1);

[0034] S204: Generate shunt control trigger signal: Based on the above calculation results, generate the final control command tag. The command tag generated by the comprehensive judgment logic is as follows:

[0035] ;

[0036] in For the preset threshold, To open the valve, sewage / initial rainwater is diverted to the wastewater treatment plant. To maintain the status quo or to open the storm drains to allow clean rainwater to flow directly into the river, To force all overflow outlets to open and prevent backflow into the pipeline.

[0037] Preferably, the specific logical steps of S3 are as follows:

[0038] S301: Constructing the pipeline network topology connection matrix: Abstracting the physical pipeline network into a directed graph. The topological connection relationship between nodes and pipe segments is established, and the locations of key control nodes such as valves and pump stations are identified. The topological adjacency matrix used is as follows:

[0039] ;

[0040] in For elements of the adjacency matrix, For a set of nodes (manholes, confluence points). For the set of edges (pipeline);

[0041] S302: Calculate water flow migration time: Based on the current pipe segment velocity and length, calculate the water flow from the upstream control point. Migrate to downstream confluence point Required time lag The migration time calculation formula used is as follows:

[0042] ;

[0043] in To be from upstream Downstream Total migration time The set of paths between two points. For the first The length of the pipe section For the first The average flow velocity in the section of the pipe;

[0044] The real-time flow rate estimation formula used is as follows:

[0045] ;

[0046] in For pipe roughness coefficient, For hydraulic radius, Hydraulic gradient;

[0047] S303: Establishing a Spatiotemporal Co-constraint Model: To ensure the safe operation of the pipeline network, valve actions must satisfy hydraulic balance and time series constraints. The mass conservation constraint formula during valve use is as follows:

[0048] ;

[0049] in These represent the inflow and outflow flows to the nodes, respectively. This represents the water storage capacity at the node (0 for pressure wells, and varying for regulating reservoirs).

[0050] The timing coordination constraint formula used is as follows:

[0051] ;

[0052] in The timing of downstream valve actuation. The timing of the upstream valve's operation. For safety redundancy time (to prevent water hammer effect);

[0053] S304: Generating the Optimal Flow Diversion Scheme: With the objectives of minimizing the total system overflow and minimizing pump station energy consumption, the optimal opening degree of each valve is determined. The multi-objective optimization function formula used is as follows:

[0054] ;

[0055] in The total objective function value, for Overflow at any moment for Pump station energy consumption at any time , These are the weighting coefficients;

[0056] The valve opening-flow rate relationship formula used is as follows:

[0057] ;

[0058] in The valve flow coefficient, This represents the relative valve opening (0-1). The pressure difference across the valve;

[0059] S305: Output Cooperative Scheduling Instructions: Converts the optimized valve opening sequence into specific PLC control instructions and sends them to the execution layer. The instruction generation logic used is as follows:

[0060] ;

[0061] in Each valve has a unique serial number. The optimal opening degree calculated for S304. This is the precise execution timestamp calculated based on S302.

[0062] Preferably, the specific logical steps of S4 are as follows:

[0063] S401: Real-time acquisition of wastewater treatment plant load status: Obtain the following key parameters from the wastewater treatment plant's DCS system:

[0064] Inlet flow rate ;

[0065] Influent COD load ;

[0066] Remaining processing capacity ;in To enhance the design processing capacity, Current load factor;

[0067] Based on the real-time influent flow data collected by the wastewater treatment plant's DCS system and the COD concentration value fed back by the online water quality monitoring instrument, the total influent organic pollution load at the current moment is quantified by multiplication calculation; at the same time, combined with the maximum treatment capacity parameters of the process design, the remaining acceptance capacity of the treatment system at the current moment is dynamically calculated, providing boundary constraints for subsequent pipeline network scheduling.

[0068] S402: Calculation of the dynamic threshold for total interception in the pipeline network: Based on the current remaining treatment capacity of the wastewater treatment plant and the target effluent water quality standard, the maximum allowable interception flow of the pipeline network system under the premise of ensuring safe operation at the plant end is calculated in reverse; a safety factor is introduced to cope with the risk of water quality fluctuations, ensuring that the total amount of pollutants entering the treatment plant under extreme operating conditions does not exceed the tolerance limit of the biological system. The formula used is as follows:

[0069] ;

[0070] in The target concentration for intercepted water quality (e.g., 400 mg / L). For safety factors (usually 0.85-0.95);

[0071] S403: The maximum allowable flow rate is compared in real time to construct an absolute error constraint function; by minimizing the deviation between the two, a real-time dynamic balance relationship between the pipeline network's transmission capacity and the plant's processing capacity is established to prevent system collapse or overflow risks caused by overload operation. The formula used is as follows:

[0072] ;

[0073] To set the allowable deviation threshold, take... ×0.05, if not satisfied, trigger dynamic adjustment of S404;

[0074] S404: Dynamic Adjustment of Total Catchment Allocation: The proportional allocation method is used to reset the opening of each choke point. The formula used is as follows:

[0075] ;

[0076] in For the first The relative opening degree (0-1) of each shut-off valve.

[0077] when Decrease the opening when it is open, and increase it when it is closed.

[0078] Preferably, the specific logical steps of S5 are as follows:

[0079] S501: Construct a historical case feature vector library: Store historical diversion cases as feature vectors.

[0080] ;

[0081] in For historical case indexing, Output from step S2;

[0082] Taking into account both the energy consumption cost of pipeline operation and the efficiency of pollutant reduction, a weighted multi-objective optimization function is constructed. By setting weight coefficients to adjust the priority of economic efficiency and environmental protection, the aim is to find the optimal scheduling strategy solution set that can meet emission standards and minimize the energy consumption of pump station operation.

[0083] S502: Current operating conditions are similar to historical cases:

[0084] Calculate the current operating condition The cosine similarity to the historical database is calculated using the following formula:

[0085] ;

[0086] Select the case with the highest similarity ( >0.85);

[0087] Based on the hydraulic model of the pipeline network and the physical characteristics of the equipment, hard constraints are set for the upper and lower limits of liquid level, the start and stop frequency of pump stations, and the flow velocity in the pipeline; the physical constraints in the actual project are transformed into the feasible domain boundary of the mathematical programming problem to ensure that the generated optimized scheduling scheme is feasible in the actual project.

[0088] S503: Control parameters based on execution deviation correction:

[0089] If the actual overflow Compared with S403 predicted value Deviation exceeds limit:

[0090] ;

[0091] Then, the key threshold in S2 is dynamically adjusted:

[0092] ;

[0093] in To adjust the step size (usually 0.05).

[0094] The sign function of the deviation direction (positive deviation increases) (negative deviation decreases).

[0095] S504: Online Update Case Library: Saves the current operating conditions and corrected control parameters into the historical library. The formula used is as follows:

[0096] ;

[0097] Triggering condition: When Add new cases as required.

[0098] Preferably, the specific logical steps of S6 are as follows:

[0099] S601: Control command safety verification: Verify the feasibility of valve operation before issuing the command. The verification formula used is as follows:

[0100] ;

[0101] in This is the maximum single-step opening change (usually 0.1), used to prevent water hammer caused by sudden valve opening and closing;

[0102] An improved intelligent optimization algorithm is used to find the optimal solution for the constructed load balancing model. By continuously updating the decision variables, the algorithm gradually approaches the global optimum. During each iteration, the satisfaction of the constraints is checked in real time, and invalid solutions are eliminated until the objective function converges to the preset accuracy range.

[0103] S602: Real-time monitoring of execution deviation: Calculates the error between the command and the actual opening degree using the following formula:

[0104] ;

[0105] If connected N cycles (e.g., 0.15), triggers S603;

[0106] S603: Automatic Emergency Switching Strategy for Abnormal Operating Conditions: Automatically switches to the preset emergency bypass strategy when an abnormal operating condition is detected, such as a valve malfunction or a pipeline liquid level exceeding a safe threshold. Forcefully open the nearest overflow outlet to meet the requirements. , Overflow outlet cross-sectional area, The difference in water head between upstream and downstream;

[0107] S604: Closed-loop feedback termination condition: The current control cycle ends when the following conditions are met simultaneously:

[0108] ;

[0109] in The overflow high-risk threshold defined for S205 is used to trigger a new round of monitoring in S2;

[0110] The optimal scheduling instructions obtained from the solution are substituted into historical operating data for backtracking verification. The degree of agreement between the actual operating curve and the simulated prediction curve is compared. The relative error is calculated to evaluate the accuracy of the model. Finally, a standardized control instruction sequence containing the optimal interception flow rate of each node and the operating frequency of the pump station is output.

[0111] This invention also proposes a smart rainwater and sewage diversion system under smart water management, including a multi-source sensing and fusion modeling module, a runoff pollution and overflow risk identification module, a pipeline topology collaborative scheduling decision module, a "source-network-plant" load balancing module, a control strategy adaptive optimization module, a diversion control closed-loop execution module, and a visualization display and situation simulation module.

[0112] The multi-source sensing and fusion modeling module is used to collect, synchronize and standardize drainage network liquid level, flow rate, water quality COD / SS data, meteorological rainfall data and sewage treatment plant operation load data in real time, and build a unified data resource pool. At the same time, it generates a pipeline network operation status profile with spatiotemporal labels to support subsequent pollution identification and topology analysis.

[0113] The runoff pollution and overflow risk identification module is used to perform multi-dimensional feature extraction and fusion analysis on standardized multi-source data, identify water quality mutation points and pipeline overflow risk points in rainfall runoff, output diversion control trigger signals and risk level labels, and simultaneously generate pollution-related area information to provide input constraints for collaborative scheduling.

[0114] The pipeline topology collaborative scheduling decision module is used to construct a valve opening optimization model based on the GIS pipeline topology, water flow migration time model and upstream and downstream node status, so as to realize the transformation from single threshold control to multi-node collaborative control that considers time and space migration, and generate the optimal diversion scheme that takes into account both sewage interception efficiency and downstream carrying capacity.

[0115] The “source-network-plant” load balancing module is used to connect the influent load and remaining treatment capacity of the sewage treatment plant in real time, establish a dynamic linkage feedback mechanism between the pipeline interception flow and the sewage treatment capacity, prevent the sewage treatment plant from being paralyzed due to the impact load during the peak of the rainstorm, and prevent clean rainwater from crowding out the sewage treatment resources during the light rain period.

[0116] The adaptive optimization module of the control strategy is used to perform similarity matching between the feature vector of the current rainfall event and the historical diversion case library, use case reasoning technology to dynamically correct the control model parameters, and realize online learning and continuous adaptive optimization of the model based on the real-time execution effect.

[0117] The diversion control closed-loop execution module is used to send scheduling instructions to the field PLC execution terminal and monitor valve action feedback and pipeline water level changes in real time. It automatically switches the emergency bypass strategy for abnormal working conditions such as valve jamming, pipeline rupture or sudden high concentration pollution, and realizes millisecond-level closed-loop control from monitoring to execution.

[0118] The visualization and situation simulation module is used to visualize the pipeline liquid level, water flow direction, valve status, pollution diffusion range and emergency response process on the GIS map in a time and space, so as to realize dynamic situation monitoring and simulation of the entire process of rainwater and sewage separation.

[0119] Compared with existing technologies, the beneficial effects of this invention are:

[0120] 1. By constructing a multi-source sensing dataset that includes water quality (COD / SS), hydrology (rainfall, flow rate), and plant load, this technology overcomes the limitations of traditional technologies that rely solely on liquid level sensors. It identifies runoff pollution peaks by utilizing the rate of change in rainfall intensity and the characteristics of sudden changes in water quality, and dynamically adjusts interception strategies in conjunction with the real-time treatment capacity margin of wastewater treatment plants. This effectively avoids the waste of resources caused by clean rainwater being mistakenly intercepted by wastewater treatment plants during light rain, as well as overflow pollution caused by relying solely on liquid level during peak rainfall. It significantly improves the scientific nature and accuracy of diversion decisions, forming a multi-dimensional sensing and all-element decision-making system, and solves the problem of misjudgment caused by a single sensing dimension.

[0121] 2. By introducing the concepts of pipeline topology and water flow migration time, an "upstream and downstream collaborative scheduling decision-making" mechanism was established. By calculating the water flow transmission time in the pipeline, the valve action sequence and opening degree of the upstream interception well and downstream nodes were coordinated in advance, avoiding downstream pipeline backflow or overflow caused by the superposition of instantaneous flow. The transformation from "single-point independent control" to "regional multi-node spatiotemporal collaborative control" achieved the global optimization of pipeline system operation and effectively solved the systemic contradiction of "pressing down the gourd and the ladle floats up".

[0122] 3. By constructing a millisecond-level closed-loop control system of "monitoring-decision-execution" and using an integrated "source-network-plant" load balancing model, the system matches the interception flow of the pipeline network with the treatment capacity of the sewage treatment plant in real time to prevent impact loads. At the same time, the system's built-in emergency response module can automatically switch bypass strategies or activate emergency storage in response to conditions such as valve jamming or sudden water quality anomalies. Compared with the lag in manual intervention, it achieves millisecond-level automatic response to emergencies, ensuring the operational safety and resilience of the drainage system under extreme weather or equipment failure, and improving emergency response and system resilience.

[0123] This invention systematically addresses the core pain points of traditional stormwater and sewage separation technologies through multi-dimensional perception and all-element decision-making, spatiotemporal collaborative control, and a closed-loop adaptive response mechanism. By integrating water quality, hydrology, and plant load data, it overcomes the limitations of single-sensorship, accurately identifies runoff pollution characteristics, and dynamically matches sewage treatment plant capacity, avoiding the risks of accidental interception and overflow. Based on the pipeline topology and water flow migration time, it constructs an upstream-downstream collaborative scheduling strategy, realizing the transformation from single-point control to regional spatiotemporal linkage, effectively eliminating systemic conflicts. Through an integrated "source-network-plant" load balancing model and millisecond-level closed-loop emergency response, it prevents impact loads in real time and automatically responds to equipment failures or water quality anomalies, significantly improving system resilience and emergency response capabilities. The innovation across the entire chain from perception and decision-making to execution forms a scientific, precise, collaborative, efficient, safe, and reliable intelligent separation system, providing a highly adaptable solution for urban drainage systems. Attached Figure Description

[0124] Figure 1 The flowchart is a process for the intelligent separation method of rainwater and sewage under smart water management proposed in this invention.

[0125] Figure 2 This is a system block diagram of the intelligent rainwater and sewage diversion system under the smart water management proposed in this invention. Detailed Implementation

[0126] The present invention will be further explained below with reference to specific embodiments.

[0127] Example

[0128] Reference Figure 1 This embodiment proposes a method for intelligent separation of rainwater and sewage under smart water management, including the following steps:

[0129] S1: Construct a multi-source sensing dataset for rainwater and sewage separation: Collect hydrological and water quality data, meteorological monitoring data, and sewage treatment plant operation load data of key nodes in the drainage network to construct a basic data resource pool;

[0130] Its hydrological and water quality data include pipe network liquid level, flow rate, flow velocity, and COD / SS concentration; meteorological monitoring data includes rainfall intensity and cumulative rainfall; and wastewater treatment plant operating load data includes influent flow rate, dissolved oxygen concentration in biological treatment tanks, and excess sludge discharge.

[0131] S2: Runoff pollution and overflow risk identification based on multi-dimensional features: fusion analysis of multi-source data to identify water quality mutation points and overflow risk points in the rainfall runoff process, and generate diversion control trigger signals;

[0132] The specific logical steps are as follows:

[0133] S201: Multi-source data preprocessing and feature extraction: The raw data collected in S1 is denoised and standardized, and the following core feature parameters are extracted:

[0134] Rainfall characteristics: Current rainfall intensity Rainfall intensity change rate ;

[0135] Hydrological characteristics of the pipeline network: Pipeline fill rate (Liquid level / pipe diameter), rate of change in filling degree ;

[0136] Water quality characteristics: Chemical oxygen demand (COD) concentration Suspended solids concentration ;

[0137] S202: Calculation of Runoff Pollution Index: To distinguish between clean rainwater and heavily polluted initial rainwater, a runoff pollution index is introduced, which combines water quality concentration and water volume growth rate. The formula used is as follows:

[0138] ;

[0139] in The current runoff pollution index. This is the historical maximum concentration value (normalized). This represents the maximum rate of change of flow velocity in the pipeline (normalized). Weighting coefficients (usually) (This can be adjusted according to the regional water quality characteristics).

[0140] S203: Identifying abrupt changes in water quality: The sliding window algorithm is used to detect whether a step change has occurred in water quality. The judgment logic formula used is as follows:

[0141] ;

[0142] in To represent COD or SS sequences, To adjust the size of the sliding window (e.g., comparing the first 5 minutes with the first 10-5 minutes). This is the mutation threshold coefficient. The standard deviation is the historical data; if a sudden change is triggered, a high-priority shunt signal is generated directly.

[0143] S204: Predicting Pipeline Overflow Risk: Based on current pipe filling levels and rainfall intensity, predicting future... The prediction model formula used for the pipeline network status after 15 minutes (e.g., 15 minutes) is as follows:

[0144] ;

[0145] in To predict future liquid levels;

[0146] The overflow risk probability formula used is as follows:

[0147] ;

[0148] in The overflow critical level (e.g., 85% of the pipe diameter). The probability of overflow risk (between 0 and 1);

[0149] S204: Generate shunt control trigger signal: Based on the above calculation results, generate the final control command tag. The command tag generated by the comprehensive judgment logic is as follows:

[0150] ;

[0151] in For the preset threshold, To open the valve, sewage / initial rainwater is diverted to the wastewater treatment plant. To maintain the status quo or to open the storm drains to allow clean rainwater to flow directly into the river, To force all overflow outlets to open and prevent backflow into the pipeline;

[0152] S3: Coordinated scheduling decision based on pipeline topology and water flow migration time: Establish a pipeline hydraulic transmission model, calculate water flow migration time, coordinate the action sequence and opening degree of upstream and downstream valves, and generate the optimal diversion scheme;

[0153] The specific logical steps are as follows:

[0154] S301: Constructing the pipeline network topology connection matrix: Abstracting the physical pipeline network into a directed graph. The topological connection relationship between nodes and pipe segments is established, and the locations of key control nodes such as valves and pump stations are identified. The topological adjacency matrix used is as follows:

[0155] ;

[0156] in For elements of the adjacency matrix, For a set of nodes (manholes, confluence points). For the set of edges (pipeline);

[0157] S302: Calculate water flow migration time: Based on the current pipe segment velocity and length, calculate the water flow from the upstream control point. Migrate to downstream confluence point Required time lag The migration time calculation formula used is as follows:

[0158] ;

[0159] in To be from upstream Downstream Total migration time The set of paths between two points. For the first The length of the pipe section For the first The average flow velocity in the section of the pipe;

[0160] The real-time flow rate estimation formula used is as follows:

[0161] ;

[0162] in For pipe roughness coefficient, For hydraulic radius, Hydraulic gradient;

[0163] S303: Establishing a Spatiotemporal Co-constraint Model: To ensure the safe operation of the pipeline network, valve actions must satisfy hydraulic balance and time series constraints. The mass conservation constraint formula during valve use is as follows:

[0164] ;

[0165] in These represent the inflow and outflow flows to the nodes, respectively. This represents the water storage capacity at the node (0 for pressure wells, and varying for regulating reservoirs).

[0166] The timing coordination constraint formula used is as follows:

[0167] ;

[0168] in The timing of downstream valve actuation. The timing of the upstream valve's operation. For safety redundancy time (to prevent water hammer effect);

[0169] S304: Generating the Optimal Flow Diversion Scheme: With the objectives of minimizing the total system overflow and minimizing pump station energy consumption, the optimal opening degree of each valve is determined. The multi-objective optimization function formula used is as follows:

[0170] ;

[0171] in The total objective function value, for Overflow at any moment for Pump station energy consumption at any time , These are the weighting coefficients;

[0172] The valve opening-flow rate relationship formula used is as follows:

[0173] ;

[0174] in The valve flow coefficient, This represents the relative valve opening (0-1). The pressure difference across the valve;

[0175] S305: Output Cooperative Scheduling Instructions: Converts the optimized valve opening sequence into specific PLC control instructions and sends them to the execution layer. The instruction generation logic used is as follows:

[0176] ;

[0177] in Each valve has a unique serial number. The optimal opening degree calculated for S304. This is the precise execution timestamp calculated based on S302;

[0178] S4: Construct an integrated load balancing model of "source-network-plant": connect with the influent load and treatment capacity of the sewage treatment plant in real time, dynamically adjust the total interception of the pipeline network, and achieve system-level supply and demand balance;

[0179] The specific logical steps are as follows:

[0180] S401: Real-time acquisition of wastewater treatment plant load status: Obtain the following key parameters from the wastewater treatment plant's DCS system:

[0181] Inlet flow rate ;

[0182] Influent COD load ;

[0183] Remaining processing capacity ;in To enhance the design processing capacity, Current load factor;

[0184] Based on the real-time influent flow data collected by the wastewater treatment plant's DCS system and the COD concentration value fed back by the online water quality monitoring instrument, the total influent organic pollution load at the current moment is quantified by multiplication calculation; at the same time, combined with the maximum treatment capacity parameters of the process design, the remaining acceptance capacity of the treatment system at the current moment is dynamically calculated, providing boundary constraints for subsequent pipeline network scheduling.

[0185] S402: Calculation of the dynamic threshold for total interception in the pipeline network: Based on the current remaining treatment capacity of the wastewater treatment plant and the target effluent water quality standard, the maximum allowable interception flow of the pipeline network system under the premise of ensuring safe operation at the plant end is calculated in reverse; a safety factor is introduced to cope with the risk of water quality fluctuations, ensuring that the total amount of pollutants entering the treatment plant under extreme operating conditions does not exceed the tolerance limit of the biological system. The formula used is as follows:

[0186] ;

[0187] in The target concentration for intercepted water quality (e.g., 400 mg / L). For safety factors (usually 0.85-0.95);

[0188] S403: The maximum allowable flow rate is compared in real time to construct an absolute error constraint function; by minimizing the deviation between the two, a real-time dynamic balance relationship between the pipeline network's transmission capacity and the plant's processing capacity is established to prevent system collapse or overflow risks caused by overload operation. The formula used is as follows:

[0189] ;

[0190] To set the allowable deviation threshold, take... ×0.05, if not satisfied, trigger dynamic adjustment of S404;

[0191] S404: Dynamic Adjustment of Total Catchment Allocation: The proportional allocation method is used to reset the opening of each choke point. The formula used is as follows:

[0192] ;

[0193] in For the first The relative opening degree (0-1) of each shut-off valve.

[0194] when Decrease the opening when it is open, and increase it when it is closed;

[0195] S5: Dynamic optimization of control strategy based on real-time feedback: The current control strategy is matched for similarity using a historical diversion case library, and the control model parameters are corrected online based on the real-time execution effect;

[0196] The specific logical steps are as follows:

[0197] S501: Construct a historical case feature vector library: Store historical diversion cases as feature vectors.

[0198] ;

[0199] in For historical case indexing, Output from step S2;

[0200] Taking into account both the energy consumption cost of pipeline operation and the efficiency of pollutant reduction, a weighted multi-objective optimization function is constructed. By setting weight coefficients to adjust the priority of economic efficiency and environmental protection, the aim is to find the optimal scheduling strategy solution set that can meet emission standards and minimize the energy consumption of pump station operation.

[0201] S502: Current operating conditions are similar to historical cases:

[0202] Calculate the current operating condition The cosine similarity to the historical database is calculated using the following formula:

[0203] ;

[0204] Select the case with the highest similarity ( >0.85);

[0205] Based on the hydraulic model of the pipeline network and the physical characteristics of the equipment, hard constraints are set for the upper and lower limits of liquid level, the start and stop frequency of pump stations, and the flow velocity in the pipeline; the physical constraints in the actual project are transformed into the feasible domain boundary of the mathematical programming problem to ensure that the generated optimized scheduling scheme is feasible in the actual project.

[0206] S503: Control parameters based on execution deviation correction:

[0207] If the actual overflow Compared with S403 predicted value Deviation exceeds limit:

[0208] ;

[0209] Then, the key threshold in S2 is dynamically adjusted:

[0210] ;

[0211] in To adjust the step size (usually 0.05).

[0212] The sign function of the deviation direction (positive deviation increases) (negative deviation decreases).

[0213] S504: Online Update Case Library: Saves the current operating conditions and corrected control parameters into the historical library. The formula used is as follows:

[0214] ;

[0215] Triggering condition: When Forced addition of new cases;

[0216] S6: Achieve closed-loop execution and emergency response for diversion control: Send control commands to the actuator, monitor valve status and pipeline response in real time, and automatically switch emergency bypass strategies for abnormal operating conditions;

[0217] The specific logical steps are as follows:

[0218] S601: Control command safety verification: Verify the feasibility of valve operation before issuing the command. The verification formula used is as follows:

[0219] ;

[0220] in This is the maximum single-step opening change (usually 0.1), used to prevent water hammer caused by sudden valve opening and closing;

[0221] An improved intelligent optimization algorithm is used to find the optimal solution for the constructed load balancing model. By continuously updating the decision variables, the algorithm gradually approaches the global optimum. During each iteration, the satisfaction of the constraints is checked in real time, and invalid solutions are eliminated until the objective function converges to the preset accuracy range.

[0222] S602: Real-time monitoring of execution deviation: Calculates the error between the command and the actual opening degree using the following formula:

[0223] ;

[0224] If connected N cycles (e.g., 0.15), triggers S603;

[0225] S603: Automatic Emergency Switching Strategy for Abnormal Operating Conditions: Automatically switches to the preset emergency bypass strategy when an abnormal operating condition is detected, such as a valve malfunction or a pipeline liquid level exceeding a safe threshold. Forcefully open the nearest overflow outlet to meet the requirements. , Overflow outlet cross-sectional area, The difference in water head between upstream and downstream;

[0226] S604: Closed-loop feedback termination condition: The current control cycle ends when the following conditions are met simultaneously:

[0227] ;

[0228] in The overflow high-risk threshold defined for S205 is used to trigger a new round of monitoring in S2;

[0229] The optimal scheduling instructions obtained from the solution are substituted into historical operating data for backtracking verification. The degree of agreement between the actual operating curve and the simulated prediction curve is compared. The relative error is calculated to evaluate the accuracy of the model. Finally, a standardized control instruction sequence containing the optimal interception flow rate of each node and the operating frequency of the pump station is output.

[0230] This embodiment systematically addresses the core pain points of traditional stormwater and sewage separation technologies through multi-dimensional perception and all-element decision-making, spatiotemporal collaborative control, and closed-loop adaptive response mechanisms. By integrating water quality, hydrology, and plant load data, it overcomes the limitations of single-point perception, accurately identifies runoff pollution characteristics, and dynamically matches sewage treatment plant capacity, avoiding the risks of accidental interception and overflow. Based on the pipeline topology and water flow migration time, it constructs an upstream-downstream collaborative scheduling strategy, realizing the transformation from single-point control to regional spatiotemporal linkage, effectively eliminating systemic conflicts. Through an integrated "source-network-plant" load balancing model and millisecond-level closed-loop emergency response, it prevents impact loads in real time and automatically responds to equipment failures or water quality anomalies, significantly improving system resilience and emergency response capabilities. The innovation across the entire chain from perception and decision-making to execution forms a scientific, precise, collaborative, efficient, safe, and reliable intelligent separation system, providing a highly adaptable solution for urban drainage systems.

[0231] Reference Figure 2 This embodiment also proposes a smart rainwater and sewage intelligent diversion system under smart water management, including a multi-source sensing and fusion modeling module, a runoff pollution and overflow risk identification module, a pipeline topology collaborative scheduling decision module, a "source-network-plant" load balancing module, a control strategy adaptive optimization module, a diversion control closed-loop execution module, and a visualization display and situation simulation module.

[0232] The multi-source sensing and fusion modeling module is used to collect, synchronize and standardize data on drainage network liquid level, flow rate, water quality COD / SS, meteorological rainfall data and sewage treatment plant operation load data in real time, and build a unified data resource pool. At the same time, it generates a pipeline network operation status profile with spatiotemporal labels to support subsequent pollution identification and topology analysis.

[0233] The runoff pollution and overflow risk identification module is used to extract and fuse standardized multi-source data in multiple dimensions, identify water quality mutation points and pipeline overflow risk points in rainfall runoff, output diversion control trigger signals and risk level labels, and simultaneously generate pollution-related area information to provide input constraints for collaborative scheduling.

[0234] The pipeline topology collaborative scheduling decision module is used to construct a valve opening optimization model based on the GIS pipeline topology, water flow migration time model and upstream and downstream node status, realizing the transformation from single threshold control to multi-node collaborative control that considers time and space migration, and generating the optimal diversion scheme that takes into account both sewage interception efficiency and downstream carrying capacity.

[0235] The "Source-Network-Plant" load balancing module is used to connect the influent load and remaining treatment capacity of the sewage treatment plant in real time, establish a dynamic linkage feedback mechanism between the pipeline interception flow and the sewage treatment capacity, prevent the sewage treatment plant from being paralyzed due to the impact load during the peak of the rainstorm, and prevent clean rainwater from crowding out the sewage treatment resources during the light rain period.

[0236] The adaptive optimization module for control strategy is used to perform similarity matching between the feature vector of the current rainfall event and the historical diversion case library, use case reasoning technology to dynamically correct the control model parameters, and realize online learning and continuous adaptive optimization of the model based on the real-time execution effect.

[0237] The diversion control closed-loop execution module is used to send dispatch instructions to the field PLC execution terminal and monitor valve action feedback and pipeline water level changes in real time. It automatically switches the emergency bypass strategy for abnormal working conditions such as valve jamming, pipeline rupture or sudden high concentration pollution, and realizes millisecond-level closed-loop control from monitoring to execution.

[0238] The visualization and situation simulation module is used to visualize the pipeline liquid level, water flow direction, valve status, pollution diffusion range and emergency response process on the GIS map in time and space, so as to realize dynamic situation monitoring and simulation of the entire process of rainwater and sewage separation.

[0239] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for intelligent separation of rainwater and sewage under smart water management, characterized in that, Includes the following steps: S1: Construct a multi-source sensing dataset for rainwater and sewage separation: Collect hydrological and water quality data, meteorological monitoring data, and sewage treatment plant operation load data of key nodes in the drainage network to construct a basic data resource pool; S2: Runoff pollution and overflow risk identification based on multi-dimensional features: fusion analysis of multi-source data to identify water quality mutation points and overflow risk points in the rainfall runoff process, and generate diversion control trigger signals; S3: Coordinated scheduling decision based on pipeline topology and water flow migration time: Establish a pipeline hydraulic transmission model, calculate water flow migration time, coordinate the action sequence and opening degree of upstream and downstream valves, and generate the optimal diversion scheme; S4: Construct an integrated load balancing model of "source-network-plant": connect with the influent load and treatment capacity of the sewage treatment plant in real time, dynamically adjust the total interception of the pipeline network, and achieve system-level supply and demand balance; S5: Dynamic optimization of control strategy based on real-time feedback: The current control strategy is matched for similarity using a historical diversion case library, and the control model parameters are corrected online based on the real-time execution effect; S6: Achieve closed-loop execution and emergency response for diversion control: Send control commands to the actuator, monitor valve status and pipeline response in real time, and automatically switch emergency bypass strategies for abnormal operating conditions.

2. The intelligent rainwater and sewage diversion method under smart water management according to claim 1, characterized in that, In S1, the hydrological and water quality data include pipeline liquid level, flow rate, flow velocity, and COD / SS concentration; the meteorological monitoring data includes rainfall intensity and cumulative rainfall; and the wastewater treatment plant operating load data includes influent flow rate, dissolved oxygen concentration in the biological treatment tank, and excess sludge discharge.

3. The intelligent rainwater and sewage diversion method under smart water management according to claim 2, characterized in that, The specific logical steps of S2 are as follows: S201: Multi-source data preprocessing and feature extraction: The raw data collected in S1 is denoised and standardized, and the following core feature parameters are extracted: Rainfall characteristics: Current rainfall intensity Rainfall intensity change rate ; Pipeline Hydrological characteristics: Pipeline fill level Rate of change in fullness ; Water quality characteristics: Chemical oxygen demand (COD) concentration Suspended solids concentration ; S202: Calculation of Runoff Pollution Index: To distinguish between clean rainwater and heavily polluted initial rainwater, a runoff pollution index is introduced, which combines water quality concentration and water volume growth rate. The formula used is as follows: ； in The current runoff pollution index. This is the highest concentration ever recorded. This represents the maximum rate of change of flow velocity in the pipeline. These are the weighting coefficients; S203: Identifying abrupt changes in water quality: The sliding window algorithm is used to detect whether a step change has occurred in water quality. The judgment logic formula used is as follows: ； in To represent COD or SS sequences, To adjust the sliding window size, This is the mutation threshold coefficient. The standard deviation is the historical data; if a sudden change is triggered, a high-priority shunt signal is generated directly. S204: Predicting Pipeline Overflow Risk: Based on current pipe filling levels and rainfall intensity, predicting future... The prediction model formula used for the pipeline network status after minutes is as follows: ； in To predict future liquid levels; The overflow risk probability formula used is as follows: ； in This is the critical overflow level. This represents the probability of overflow risk. S204: Generate shunt control trigger signal: Based on the above calculation results, generate the final control command tag. The command tag generated by the comprehensive judgment logic is as follows: ； in For the preset threshold, To open the valve, sewage / initial rainwater is diverted to the wastewater treatment plant. To maintain the status quo or to open the storm drains to allow clean rainwater to flow directly into the river, To force all overflow outlets to open and prevent backflow into the pipeline.

4. The intelligent rainwater and sewage separation method under smart water management according to claim 3, characterized in that, The specific logical steps of S3 are as follows: S301: Constructing the pipeline network topology connection matrix: Abstracting the physical pipeline network into a directed graph. The topological connection relationship between nodes and pipe segments is established, and the locations of key control nodes such as valves and pump stations are identified. The topological adjacency matrix used is as follows: ； in For adjacency matrix elements, For a set of nodes, Let it be the set of edges; S302: Calculate water flow migration time: Based on the current pipe segment velocity and length, calculate the water flow from the upstream control point. Migrate to downstream confluence point Required time lag The migration time calculation formula used is as follows: ； in To be from upstream Downstream Total migration time The set of paths between two points. For the first The length of the pipe section For the first The average flow velocity in the section of the pipe; The real-time flow rate estimation formula used is as follows: ； in For pipe roughness coefficient, For hydraulic radius, Hydraulic gradient; S303: Establishing a Spatiotemporal Co-constraint Model: To ensure the safe operation of the pipeline network, valve actions must satisfy hydraulic balance and time series constraints. The mass conservation constraint formula during valve use is as follows: ； in These represent the inflow and outflow flows to the nodes, respectively. The water storage capacity of the node; The timing coordination constraint formula used is as follows: ； in The timing of downstream valve actuation. The timing of the upstream valve's operation. For safety redundancy time; S304: Generating the Optimal Flow Diversion Scheme: With the objectives of minimizing the total system overflow and minimizing pump station energy consumption, the optimal opening degree of each valve is determined. The multi-objective optimization function formula used is as follows: ； in The total objective function value, for Overflow at any moment for Pump station energy consumption at any time , These are the weighting coefficients; The valve opening-flow rate relationship formula used is as follows: ； in For valve flow coefficient, This refers to the relative opening of the valve. The pressure difference across the valve; S305: Output Cooperative Scheduling Instructions: Converts the optimized valve opening sequence into specific PLC control instructions and sends them to the execution layer. The instruction generation logic used is as follows: ； in Each valve has a unique serial number. The optimal opening degree calculated for S304. This is the precise execution timestamp calculated based on S302.

5. The intelligent rainwater and sewage separation method under smart water management according to claim 4, characterized in that, The specific logical steps of S4 are as follows: S401: Real-time acquisition of wastewater treatment plant load status: Obtain the following key parameters from the wastewater treatment plant's DCS system: Inlet flow rate ; Influent COD load ; Remaining processing capacity ;in To enhance the design processing capacity, Current load factor; Based on the influent flow data collected in real time by the wastewater treatment plant's DCS system and the COD concentration value fed back by the online water quality monitoring instrument, the total influent organic pollution load at the current moment is quantified by multiplication calculation; at the same time, combined with the maximum treatment capacity parameters of the process design, the remaining acceptance capacity of the treatment system at the current moment is dynamically calculated, providing boundary constraints for subsequent pipeline network scheduling. S402: Calculation of the dynamic threshold for total interception in the pipeline network: Based on the current remaining treatment capacity of the wastewater treatment plant and the target effluent water quality standard, the maximum allowable interception flow of the pipeline network system under the premise of ensuring safe operation at the plant end is calculated in reverse; a safety factor is introduced to cope with the risk of water quality fluctuations, ensuring that the total amount of pollutants entering the treatment plant under extreme operating conditions does not exceed the tolerance limit of the biological system. The formula used is as follows: ； in To intercept the target concentration of water quality, For safety factor; S403: Real-time comparison of the maximum allowable flow rate is used to construct an absolute error constraint function; by minimizing the deviation between the two, a real-time dynamic balance relationship between the pipeline's transmission capacity and the plant's processing capacity is established to prevent system collapse or overflow risks caused by overload operation. The formula used is as follows: ； To set the allowable deviation threshold, take... ×0.05, if not satisfied, trigger dynamic adjustment of S404; S404: Dynamic Adjustment of Total Catchment Allocation: The proportional allocation method is used to reset the opening of each choke point. The formula used is as follows: ； in For the first The relative opening of each shut-off valve when Decrease the opening when it is open, and increase it when it is closed.

6. The intelligent rainwater and sewage separation method under smart water management according to claim 5, characterized in that, The specific logical steps of S5 are as follows: S501: Construct a historical case feature vector library: Store historical diversion cases as feature vectors. ； in For historical case indexing, Output from step S2; S502: Current operating conditions are similar to historical cases: Calculate the current operating condition The cosine similarity to the historical database is calculated using the following formula: ； Select the case with the highest similarity >0.85; Based on the hydraulic model of the pipeline network and the physical characteristics of the equipment, hard constraints are set for the upper and lower limits of liquid level, the start and stop frequency of pump stations, and the flow velocity in the pipeline; the physical constraints in the actual project are transformed into the feasible domain boundary of the mathematical programming problem to ensure that the generated optimized scheduling scheme is feasible in the actual project. S503: Control parameters based on execution deviation correction: If the actual overflow Compared with S403 predicted value Deviation exceeds limit: ； Then the key threshold in S2 is dynamically adjusted: ； in To correct the step size, The sign function for the deviation direction; S504: Online Update Case Library: Saves the current operating conditions and corrected control parameters into the historical library. The formula used is as follows: ； Triggering condition: When Add new cases as required.

7. The intelligent rainwater and sewage separation method under smart water management according to claim 6, characterized in that, The specific logical steps of S6 are as follows: S601: Control command safety verification: Verify the feasibility of valve operation before issuing the command. The verification formula used is as follows: ； in This is the maximum change in valve opening in a single step, used to prevent water hammer caused by sudden valve opening and closing. S602: Real-time monitoring of execution deviation: Calculates the error between the command and the actual opening degree using the following formula: ； If connected N cycles This triggers S603; S603: Automatic Emergency Switching Strategy for Abnormal Operating Conditions: Automatically switches to the preset emergency bypass strategy when an abnormal operating condition is detected, such as a valve malfunction or a pipeline liquid level exceeding a safe threshold. Forcefully open the nearest overflow outlet to meet the requirements. , Overflow outlet cross-sectional area, The difference in water head between upstream and downstream; S604: Closed-loop feedback termination condition: The current control cycle ends when the following conditions are met simultaneously: ； in The overflow high-risk threshold defined for S205 is used to trigger a new round of monitoring in S2; The optimal scheduling instructions obtained from the solution are substituted into historical operating data for backtracking verification. The degree of agreement between the actual operating curve and the simulated prediction curve is compared. The relative error is calculated to evaluate the accuracy of the model. Finally, a standardized control instruction sequence containing the optimal interception flow rate of each node and the operating frequency of the pump station is output.

8. A smart rainwater and sewage separation system under smart water management, characterized in that, The method for implementing any one of claims 1-7 includes a multi-source sensing and fusion modeling module, a runoff pollution and overflow risk identification module, a pipeline topology collaborative scheduling decision module, a "source-network-plant" load balancing module, a control strategy adaptive optimization module, a diversion control closed-loop execution module, and a visualization and situation simulation module. The multi-source sensing and fusion modeling module is used to collect, synchronize and standardize drainage network liquid level, flow rate, water quality COD / SS data, meteorological rainfall data and sewage treatment plant operation load data in real time, and build a unified data resource pool. At the same time, it generates a pipeline network operation status profile with spatiotemporal labels to support subsequent pollution identification and topology analysis. The runoff pollution and overflow risk identification module is used to perform multi-dimensional feature extraction and fusion analysis on standardized multi-source data, identify water quality mutation points and pipeline overflow risk points in rainfall runoff, output diversion control trigger signals and risk level labels, and simultaneously generate pollution-related area information to provide input constraints for collaborative scheduling. The pipeline topology collaborative scheduling decision module is used to construct a valve opening optimization model based on the GIS pipeline topology, water flow migration time model and upstream and downstream node status, so as to realize the transformation from single threshold control to multi-node collaborative control that considers time and space migration, and generate the optimal diversion scheme that takes into account both sewage interception efficiency and downstream carrying capacity. The "source-network-plant" load balancing module is used to connect the influent load and remaining treatment capacity of the sewage treatment plant in real time, establish a dynamic linkage feedback mechanism between the pipeline interception flow and the sewage treatment capacity, prevent the sewage treatment plant from being paralyzed due to the impact load during the peak of the rainstorm, and prevent clean rainwater from crowding out the sewage treatment resources during the light rain period. The adaptive optimization module of the control strategy is used to perform similarity matching between the feature vector of the current rainfall event and the historical diversion case library, use case reasoning technology to dynamically correct the control model parameters, and realize online learning and continuous adaptive optimization of the model based on the real-time execution effect. The diversion control closed-loop execution module is used to send scheduling instructions to the field PLC execution terminal and monitor valve action feedback and pipeline water level changes in real time. It automatically switches the emergency bypass strategy for abnormal working conditions such as valve jamming, pipeline rupture or sudden high concentration pollution, and realizes millisecond-level closed-loop control from monitoring to execution. The visualization and situation simulation module is used to visualize the pipeline liquid level, water flow direction, valve status, pollution diffusion range and emergency response process on the GIS map in a time and space, so as to realize dynamic situation monitoring and simulation of the entire process of rainwater and sewage separation.

Images