Integrated dispatching optimization method for pipe network and pump station of sewage plant

Abstract

The invention provides a sewage plant pipe network pump station integrated scheduling optimization method, which integrates sewage plant water inlet flow, water quality concentration, drainage flow of each pump station and district rainfall historical data, combines time rules, and uses a fusion prediction model to predict drainage flow of each catchment district and sewage plant water inlet load in the future; taking the all-day predicted total water inlet flow as a reference, combining flow fluctuation and equipment efficient operation constraints, accounting the aeration air volume based on the predicted load, adjusting the predicted water inlet flow of the sewage plant by using an optimization algorithm, and outputting the optimized water inlet flow and the flow adjustment amount; calculating the pipe network regulation and storage space of each catchment area and the hydraulic time lag from each area to the sewage plant, constructing a linear programming model to decompose the flow regulation amount to each catchment area, and generating the target flow per hour in the future of each area; and according to the target flow of each area and the optimized water inlet flow, collaborative scheduling is carried out on pump stations, water plant lift pumps and aeration fans in the areas, and dynamic optimization and safety regulation and control are carried out in combination with liquid level feedback.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment plant scheduling, and in particular to an integrated scheduling and optimization method for wastewater treatment plant pipeline network and pumping stations. Background Technology

[0002] Currently, urban drainage scheduling mainly adopts a hybrid model of "manual regulation + automatic control." Specifically, under normal operating conditions, the pipe network and pumping stations automatically start and stop equipment based on preset liquid level thresholds. Drainage begins when the liquid level reaches the upper limit and stops when it drops to the lower limit, thus controlling the pipe network liquid level within a reasonable range, with the core objective of preventing pipe network overflow. During the rainy season or flood season, the scheduling method shifts to manual intervention, adjusting pumping station liquid level setpoints based on experience or conducting on-site pump start-up and shutdown to minimize the risk of urban flooding and pipe network overflow. Sewage treatment plant scheduling is even more passive, only starting and stopping pumps or adjusting operating frequency based on the influent forebay liquid level. The existing urban drainage system scheduling model has significant limitations, specifically the following shortcomings: 1. Lack of proactive scheduling, passively responding to load fluctuations The operation of the wastewater treatment plant-pipeline network-pumping station relies entirely on real-time data response and human experience, without incorporating drainage load prediction for the catchment area. For example, predictable flow fluctuations such as morning peak drainage from residential areas and periodic industrial wastewater discharge often lead to sudden increases in pipe network liquid levels (risk of overflow) or sudden changes in the influent load of the wastewater treatment plant, forcing frequent start-ups and shutdowns of booster pumps and aeration fans or causing them to deviate from their high-efficiency range, thus increasing energy consumption.

[0003] 2. Lack of system coordination and imbalance of constraints. Core objectives such as pipeline overflow prevention, efficient operation of pumping stations, and stable load of sewage treatment plants are still under independent control and lack a global coordination mechanism: pipeline booster pumping stations may blindly increase drainage intensity to prevent overflow, leading to a surge in instantaneous load on sewage treatment plants; while sewage treatment plants restrict water intake to maintain stable operation, which may cause the pipeline liquid level to exceed the standard, forming a vicious cycle of "paying attention to one thing but losing attention to another", resulting in a significant decrease in the overall system efficiency.

[0004] 3. The operation of the fan is disconnected from the inlet water load, resulting in artificially high energy consumption. In the aeration system of a wastewater treatment plant, the operating load of the blowers is directly related to the influent load. However, under the current operation and scheduling mode, the operating status of the blowers and the influent load are often significantly disconnected: when the influent load is in a certain range, the system often falls into the predicament of multiple blowers operating in parallel but all in an inefficient state. This leads to inefficient operation of the blowers, which not only causes meaningless high energy consumption, but may also affect the stability of biological treatment due to the existence of blower air volume blind spots and aeration mismatch.

[0005] While existing technologies offer optimization solutions for certain aspects, they still have shortcomings. Existing technology CN114491893B discloses a wastewater lifting optimization scheduling method based on the characteristic curves of combined pumps. Although it optimizes the energy consumption within the pumping station, it does not consider the biochemical carrying capacity of downstream wastewater treatment plants, easily causing influent load shocks. Existing technology CN114671523B discloses a wastewater treatment energy-saving method based on predictive control, focusing on MPC control within the wastewater treatment plant, lacking utilization of upstream pipeline storage capacity, and failing to reduce peak flow at the source. Existing technology CN119047724B discloses a scheduling decision-making method and system suitable for integrated management of plants, networks, and rivers. Although it involves joint scheduling of plants, networks, and rivers, its focus is on macro-level water environment quality assurance and flood control decisions, lacking refined energy consumption control at the micro-equipment level. Furthermore, when utilizing pipeline storage, it does not fully consider the risk of sediment deposition due to low flow velocities, and also ignores the hysteresis response characteristics of biochemical reactions relative to hydraulic transport, making it difficult to achieve deep energy saving and stable operation of the entire system under multiple physical constraints. Summary of the Invention

[0006] The main objective of this invention is to provide an integrated scheduling and optimization method for sewage treatment plant pipelines and pumping stations. The technical problem solved by this application is that the existing drainage system scheduling lacks forward-looking coordination of plant and station scheduling, resulting in high energy consumption and weak resistance to load shocks. Furthermore, the existing optimization schemes do not fully consider the process constraints of pipeline deposition risk and the delayed response of biochemical systems.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: an integrated scheduling and optimization method for sewage treatment plant pipeline pumping stations, comprising the following steps: S1. Integrate historical data on sewage treatment plant influent flow, water quality concentration, drainage flow of each pumping station, and rainfall in the area, and combine them with time rules to use a fusion prediction model to predict the future drainage flow and sewage treatment plant influent load of each catchment area. S2. Based on the constant total influent flow rate predicted throughout the day, and combined with flow fluctuations and equipment efficiency constraints, the aeration volume is calculated based on the predicted load. An optimization algorithm is used to adjust the predicted influent flow rate of the wastewater treatment plant, and the optimized influent flow rate and flow adjustment amount are output. S3. Calculate the storage space of the pipe network in each catchment area and the hydraulic time delay from each area to the sewage treatment plant. Construct a linear programming model to decompose the flow adjustment amount to each catchment area and generate the target flow for each area in the future per hour. S4. Based on the target flow rate and optimized influent flow rate of each area, coordinate the scheduling of pumping stations, water plant booster pumps and aeration blowers in the area, and perform dynamic optimization and safety control in conjunction with liquid level feedback.

[0008] In the preferred scheme, in step S1, the fusion prediction model is established using an LSTM long short-term memory network combined with an XGBoost gradient boosting tree. Specific steps include: We collected hourly drainage flow data for each catchment area over the past year, influent flow, COD and total nitrogen data for wastewater treatment plants over the past year, and hourly rainfall data from local weather stations. The collected data is divided into training set, validation set and test set, and the model is trained according to a preset ratio; Input features include historical load data of water plants, time features, regional features, and meteorological features. The time features include hours, weekdays, weekends, holiday identifiers, and months. The model uses an LSTM network to process historical flow, COD, and total nitrogen time-series data, combined with XGBoost to process real-time rainfall and holiday scheduling signals. The model is iteratively trained until it converges, and outputs the hourly influent load of the wastewater treatment plant and the drainage flow of each area for the next 24 hours.

[0009] In the preferred scheme, in step S2, the predicted influent flow rate of the wastewater treatment plant is optimized and adjusted based on the premise that the total predicted influent flow rate remains unchanged throughout the day, the fluctuation of the influent flow rate of the wastewater treatment plant is ≤20%, and the wastewater treatment plant's influent lift pump and aeration blower operate efficiently. The aeration volume G(t) is calculated based on the influent COD and total nitrogen load predicted by the LSTM-XGBoost model, combined with the reaction kinetics of the activated sludge process. The formula is as follows: ; In the formula, Let be the inflow rate at time t. Let t be the COD removal amount. Let represent the total nitrogen removal at time t, a represent the carbon oxidation coefficient, b represent the nitrification coefficient, Cs20 represent saturated DO at 20℃, CSW represent saturated DO, and CL represent the remaining DO. and These represent the wastewater oxygen transfer coefficient and salinity correction coefficient, respectively, where T is the water temperature and EA represents the oxygen transfer efficiency. The flow optimization solution uses the Particle Swarm Optimization (PSO) algorithm to calculate the hourly predicted flow. Adjustments are made, and the constraints are as follows: The fan operates in the high-efficiency range: ; Traffic volatility: ; The booster pump operates in the high-efficiency range: ; The particle swarm optimization algorithm (PSO) is used to solve the constrained objective function for hourly predicted flow. Adjustments were made to optimize the influent flow rate. and flow adjustment amount ; In the formula, , These are the lower and upper limits of the high-efficiency operating range for wind turbines, respectively. This represents the average daily inflow rate. , These are the lower and upper limits of the high-efficiency operating range of the booster pump, respectively.

[0010] In the preferred scheme, in step S2, the Particle Swarm Optimization (PSO) algorithm employs an adaptive chaotic mutation strategy based on Logistic mapping to avoid the algorithm getting trapped in local optima. The steps are as follows: S21. Initialize the particle swarm by generating the initial position sequence of particles using a Logistic chaotic mapping, as shown in the formula: ; In the formula, For the k-th generation of chaotic variables, As a control parameter, a value of 4 will be used to generate... Mapping to solution space This was used as the initial population; S22. In the iterative optimization process, an inertial weight that decays non-linearly with the number of iterations is introduced. The formula is: ; In the formula, iter is the current iteration number. The maximum number of iterations, and These are the maximum and minimum inertia weights, respectively; S23. Perform a Cauchy mutation operation on the globally optimal particle gbest generated in each iteration to produce a new mutated position gbest': ; In the formula, These are standard Cauchy distribution random numbers. The mutation operator is used; if the fitness function value corresponding to gbest' is better than gbest, then gbest' is used to replace gbest, otherwise the original value is retained until the iteration termination condition is met.

[0011] In the preferred scheme, step S3, the method for calculating the storage capacity of the pipeline network in each area includes: Key parameter acquisition: Obtain pipeline network parameters such as pipe diameter, pipe length, and pipe top elevation in the area, as well as the cross-sectional area of ​​the water collection tank and the parameters of the pumping station in the high-efficiency head range of the booster pump; A geometric model of the pipe network is constructed, and the cross-sectional area data points of the pipe network and the collection tank at different liquid levels are calculated based on the obtained parameters. The model is then fitted to a continuous function A(L). ; In the formula, A(L) is the cross-sectional area of ​​the water surface when the liquid level is L; The adjustable storage capacity V(L) of the pipeline network is calculated by integration: ; In the formula, Lmin and Lmax represent the lower and upper limits of the liquid level range for efficient operation of the booster pump in the pumping station.

[0012] In the preferred scheme, the specific process of decomposing the wastewater treatment plant flow adjustment to each catchment area and compensating for hydraulic time lag in step S3 is as follows: Time delay coefficient calculation: The hydraulic time delay Ti from each area to the sewage treatment plant is determined by measuring the time difference between the start-up and shutdown of the pumping station in the area and the change in the influent. Decomposition Model Construction: Based on the premise that the sum of the flow adjustment amounts in each zone equals the total adjustment amount of the wastewater treatment plant, and constrained by the flow rates in each zone not exceeding the adjustable range, a linear programming model is constructed: ; In the formula, This represents the total adjustment of the wastewater treatment plant at time t. Let Ti be the adjustment amount for the i-th area, and Ti be the hydraulic time delay from the i-th area to the wastewater treatment plant; The constraints are: Traffic flow adjustment for the area: ,in For the i-th area, the flow rate adjustable range is determined based on the pipeline storage capacity; The booster pump operates in the high-efficiency range: ; Decomposition results: Target traffic flow per hour for each area over the next 24 hours: .

[0013] In the preferred scheme, the linear programming model in step S3 also includes a minimum scour constraint to prevent low-velocity deposition in the pipeline network. The construction process includes: Define the pipeline deposition risk index When the target traffic of the area This makes the flow velocity inside the pipe Below the minimum self-cleaning flow rate Triggered at time; Add pulse flushing logic to the constraints: If a certain period of time is predicted to Inside, Duration exceeds threshold Then force setting The lower limit constraint for flow rate at any given time is: ; In the formula, To effectively flush the flow rate, The diameter of the main pipeline in the area; When solving a linear programming model, the above-mentioned deposition constraints must be satisfied simultaneously. This can be achieved by reducing... If regulating the water storage leads to a violation of sedimentation constraints, then the previous time step will be used first. The pipeline network storage space is used for flow compensation to ensure that no hardening or deposition occurs in the pipeline network.

[0014] In the preferred scheme, step S4, the coordinated scheduling includes: Regional pump station control: The controller uses a PID algorithm to control the target flow rate of the area. Real-time collection of pump station flow rate to set value. Adjust the operating frequency of the water pump; Water plant booster pump control: The controller uses a PID algorithm to optimize the influent flow rate. Real-time collection of influent flow rate to set value. Adjust the operating frequency of the water pump; Aeration blower control: The controller uses an intelligent algorithm to set the aeration volume calculated based on the influent load as the baseline value. It also corrects the set value by combining the dissolved oxygen control target of the biological tank with the actual feedback deviation, and adjusts the blower operating frequency intermittently by taking into account the hydraulic retention time.

[0015] In the preferred embodiment, in step S4, the PID algorithm of the area pump station controller adopts fuzzy self-tuning PID control based on the variable universe of discourse of prediction error. The specific steps are as follows: S41. Define the error of the PID input. and its rate of change ; S42. Introduce a scaling factor based on the prediction results of S1. The interval of the fuzzy domain is dynamically adjusted based on the predicted trend of traffic changes. ; In the formula, This is the trend gain coefficient; at this point, the quantization factor of the fuzzy input is... and Revised to , ; S43. Obtain the PID parameter correction amount by querying the fuzzy inference rule table. And calculate the final control parameters: ; ; ; This allows the pump station control to increase or decrease its sensitivity in advance before drastic fluctuations in flow are predicted.

[0016] In the preferred scheme, in step S4, the coordinated scheduling of the aeration blower and the water plant booster pump incorporates a biological hysteresis response constraint. The method is as follows: Calculate the oxygen transfer response time constant of a biochemical system This constant is related to the current water temperature T and the sludge concentration MLSS; Set the maximum rate of change limit for the flow rate of the water plant booster pump. This limitation is not a fixed value, but is derived in reverse from the wind turbine's response capability: ; In the formula, For the target dissolved oxygen, To maintain the minimum dissolved oxygen for microbial activity, The partial derivative sensitivity of oxygen consumption rate with respect to influent flow rate; When S2 outputs the optimized influent flow rate Required rate of change exceeds At that time, the controller forcibly limits the frequency adjustment step size of the booster pump to within. Within the corresponding frequency range, the unexecuted flow difference is fed back to step S3 and redistributed to each area's pipeline network for regulation and storage.

[0017] In the preferred embodiment, step S4 further includes a dynamic optimization strategy: Real-time monitoring of the liquid level H(t) in the forebay of the wastewater treatment plant; If H(t) continues to rise and exceeds the preset warning level Hmax, reduce the target flow rate of each area according to the adjustable space ratio; If H(t) continues to decrease and falls below the warning level Hmin, increase the target flow rate for each zone to ensure that the forebay level remains stable within the range of [Hmin, Hmax].

[0018] In the preferred embodiment, step S4 also includes a security control strategy: When there is pipe network blockage, a surge in water volume in the area due to heavy rain, or abnormal operation of sewage treatment plant equipment, an overflow risk assessment shall be conducted. If the water level in the area's collection tank is... Greater than or equal to overflow level Or the wastewater treatment plant forebay level H(t) is greater than or equal to the overflow level. This is classified as an overflow risk. The pump station in the high-risk area switches to overflow control mode, prioritizing the prevention of overflow with the set value of safe liquid level. At this time, it is allowed to temporarily exceed the limit of the high-efficiency operation range. After the abnormal working condition is resolved and confirmed, the system switches back to dynamic optimization mode.

[0019] This invention provides an integrated scheduling and optimization method for sewage treatment plant pipeline pump stations. By constructing a closed-loop "prediction-optimization-scheduling" system, this invention utilizes a fusion prediction model to transform passive response into proactive prediction, dynamically utilizing the pipeline network's storage capacity to smooth influent fluctuations, significantly improving the system's energy efficiency and operational stability. This invention employs an improved chaotic mutated particle swarm optimization algorithm, effectively avoiding local optima problems in multi-constraint optimization processes; it creatively introduces pulsed flushing constraints based on minimum self-cleaning flow velocity, solving the problem of sediment compaction and deposition easily caused by pipeline network storage, ensuring smooth pipeline flow; simultaneously, it establishes a biological hysteresis response constraint mechanism to prevent excessively rapid increases in hydraulic load from causing biochemical system collapse, combined with fuzzy PID control based on variable universe of discourse for prediction errors, achieving precise execution and rapid response from macro-level scheduling strategies to micro-level equipment control. Attached Figure Description

[0020] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a flowchart of the integrated scheduling and optimization method for sewage treatment plant pipelines and pumping stations according to the present invention; Figure 2 This is a block diagram of the urban drainage system of the present invention; Figure 3 This is a diagram of the interface of the load prediction and optimization control workbench system of the present invention. Detailed Implementation

[0021] Example 1 like Figure 1-3 As shown, an integrated scheduling and optimization method for sewage treatment plant pipeline pump stations includes the following steps: S1. Integrate historical data on sewage treatment plant influent flow, water quality concentration, drainage flow of each pumping station, and rainfall in the area, and combine them with time rules to use a fusion prediction model to predict the future drainage flow and sewage treatment plant influent load of each catchment area. S2. Based on the constant total influent flow rate predicted throughout the day, and combined with flow fluctuations and equipment efficiency constraints, the aeration volume is calculated based on the predicted load. An optimization algorithm is used to adjust the predicted influent flow rate of the wastewater treatment plant, and the optimized influent flow rate and flow adjustment amount are output. S3. Calculate the storage space of the pipe network in each catchment area and the hydraulic time delay from each area to the sewage treatment plant. Construct a linear programming model to decompose the flow adjustment amount to each catchment area and generate the target flow for each area in the future per hour. S4. Based on the target flow rate and optimized influent flow rate of each area, coordinate the scheduling of pumping stations, water plant booster pumps and aeration blowers in the area, and perform dynamic optimization and safety control in conjunction with liquid level feedback.

[0022] Step S1 integrates multi-dimensional historical data and real-time rules, using a fusion prediction model to predict future operating conditions. This step solves the problem of traditional scheduling lagging behind actual load fluctuations, providing a forward-looking data foundation for subsequent optimization. Step S2 establishes an optimization principle based on a constant total daily inflow. This effectively utilizes the network's storage capacity to smooth peak flows and fill valleys, while combining this with constraints on the equipment's efficient operating range to ensure the theoretical energy efficiency of the scheduling scheme. Step S3 introduces spatial and temporal dimensions, namely the physical storage space of the network and the time lag characteristics of hydraulic transport. Through linear programming, the total demand of the wastewater treatment plant is accurately decomposed into various areas, solving the problem of spatiotemporal asynchrony when multiple sources converge. Step S4 is the execution and feedback stage. Through coordinated scheduling and dynamic control, it ensures the effective implementation of the theoretical scheme in the actual physical system and possesses robustness against errors.

[0023] In the preferred scheme, in step S1, the fusion prediction model is established using an LSTM long short-term memory network combined with an XGBoost gradient boosting tree. Specific steps include: We collected hourly drainage flow data for each catchment area over the past year, influent flow, COD and total nitrogen data for wastewater treatment plants over the past year, and hourly rainfall data from local weather stations. The collected data is divided into training set, validation set and test set, and the model is trained according to a preset ratio; Input features include historical load data of water plants, time features, regional features, and meteorological features. The time features include hours, weekdays, weekends, holiday identifiers, and months. The model uses an LSTM network to process historical flow, COD, and total nitrogen time-series data, combined with XGBoost to process real-time rainfall and holiday scheduling signals. The model is iteratively trained until it converges, and outputs the hourly influent load of the wastewater treatment plant and the drainage flow of each area for the next 24 hours.

[0024] This model employs an LSTM (Long Short-Term Memory) network combined with an XGBoost (Gradient Boosting Tree) system because urban wastewater discharge exhibits significant time-series characteristics. The LSTM network, through its internal gating mechanism, effectively captures and memorizes long-term daily and weekly variations, making it suitable for processing continuous time-series data such as flow rate, COD, and total nitrogen. Meanwhile, the XGBoost algorithm, as an ensemble learning method, demonstrates higher sensitivity and accuracy when handling abrupt or discrete features such as rainfall and holiday signals. The combination of these two methods allows the model to maintain the stability of daily forecasts while rapidly responding to load fluctuations caused by heavy rains or holidays, thus significantly improving forecast accuracy and providing reliable input for precise scheduling.

[0025] In the preferred scheme, in step S2, the predicted influent flow rate of the wastewater treatment plant is optimized and adjusted based on the premise that the total predicted influent flow rate remains unchanged throughout the day, the fluctuation of the influent flow rate of the wastewater treatment plant is ≤20%, and the wastewater treatment plant's influent lift pump and aeration blower operate efficiently. The aeration volume G(t) is calculated based on the influent COD and total nitrogen load predicted by the LSTM-XGBoost model, combined with the reaction kinetics of the activated sludge process. The formula is as follows: ; In the formula, Let be the inflow rate at time t. Let t be the COD removal amount. Let represent the total nitrogen removal at time t, a represent the carbon oxidation coefficient, b represent the nitrification coefficient, Cs20 represent saturated DO at 20℃, CSW represent saturated DO, and CL represent the remaining DO. and These represent the wastewater oxygen transfer coefficient and salinity correction coefficient, respectively, where T is the water temperature and EA represents the oxygen transfer efficiency. The flow optimization solution uses the Particle Swarm Optimization (PSO) algorithm to calculate the hourly predicted flow. Adjustments are made, and the constraints are as follows: The fan operates in the high-efficiency range: ; Traffic volatility: ; The booster pump operates in the high-efficiency range: ; The particle swarm optimization algorithm (PSO) is used to solve the constrained objective function for hourly predicted flow. Adjustments were made to optimize the influent flow rate. and flow adjustment amount ; In the formula, , These represent the lower and upper limits of the high-efficiency operating range of the wind turbine; Qavg is the average daily inflow rate. , These are the lower and upper limits of the high-efficiency operating range of the booster pump, respectively.

[0026] Among them, aeration air volume The accounting formula is derived from the reaction kinetics of activated sludge, and the formula contains... This represents the inflow rate at time t. This represents the amount of COD removed at time t. Represents the total nitrogen removal at time t, coefficient The coefficient is the carbon oxidation coefficient. These are the nitrification coefficients, and these two coefficients quantify the theoretical oxygen demand required to remove a unit of organic matter and ammonia nitrogen, respectively. In the denominator of the formula, Cs20 is the saturated dissolved oxygen at 20 degrees Celsius, CSW is the saturated dissolved oxygen at the actual temperature, and CL is the residual dissolved oxygen in the biological treatment tank. The oxygen transfer coefficient of wastewater. This is the salinity correction factor. The current water temperature is EA, and EA represents the oxygen transfer efficiency.

[0027] This formula comprehensively considers the physical laws governing the influence of temperature and water quality on the oxygen demand of biochemical reactions and oxygen transfer efficiency, ensuring the scientific nature of the airflow setting. Based on this, the blower and booster pump are set to operate in the high-efficiency zone. and and limit traffic volatility This is to force the equipment to avoid inefficient operating areas while meeting biochemical requirements, and to prevent drastic fluctuations in influent load from impacting the biochemical system, thereby achieving dual optimization of energy consumption and process stability.

[0028] In the optimization and adjustment process of step S2, this method establishes the principle of keeping the total predicted influent flow rate constant throughout the day as the baseline, while setting strict operational constraints. These constraints require that the fluctuation rate of the wastewater treatment plant's influent flow rate must be controlled within 20%, and that the wastewater treatment plant's influent lift pumps and aeration blowers must always operate within their high-efficiency range. Based on these conditions, the system performs refined optimization and adjustment of the wastewater treatment plant's predicted influent flow rate. Among these adjustments, the aeration air volume... The calculation process deeply integrates the influent chemical oxygen demand (COD) and total nitrogen load data predicted by the Long Short-Term Memory (LSTM) network and the gradient boosting tree model, and performs calculations based on the reaction kinetics principles of the activated sludge process. The specific mathematical formula is expressed as follows: ; In this formula, each variable has a clear physical meaning. This represents the inflow rate at time t. This represents the amount of chemical oxygen demand removed at time t. This represents the total nitrogen removal at time t. (Coefficient) The carbon oxidation coefficient is used to quantify the oxygen demand for removing organic carbon; the coefficient... The nitrification coefficient is used to quantify the oxygen demand for nitrification of ammonia nitrogen. Cs20 represents the saturated dissolved oxygen concentration at 20 degrees Celsius, CSW represents the saturated dissolved oxygen concentration at the actual water temperature, and CL represents the residual dissolved oxygen concentration in the biological treatment tank. and These represent the wastewater oxygen transfer coefficient and salinity correction coefficient, respectively, used to correct for oxygenation capacity under standard conditions. Represents the actual water temperature, and EA indicates oxygen transfer efficiency.

[0029] To obtain the optimal traffic scheduling scheme, this step uses the particle swarm optimization algorithm to solve the constrained objective function mentioned above, predicting the hourly traffic flow for the next 24 hours. Iterative adjustments are made. During the solution process, the algorithm must strictly satisfy three core constraints. The first is the fan operating efficiency constraint, which requires the calculated aeration air volume to... It must be located within the high-efficiency operating range of the wind turbine. Inside, among which This represents the lower limit of the high-efficiency operating range of the wind turbine. This represents the upper limit of the high-efficiency operating range of the wind turbine. Secondly, there is the constraint of flow rate fluctuation, requiring the absolute value of the rate of change of the influent flow rate. It must be less than or equal to 20%, where Qavg represents the average daily inflow rate. Finally, there is the booster pump operating efficiency constraint, requiring the booster pump to operate at a certain flow rate. It must be located within its efficient operating range. Inside, among which To increase the lower limit of the pump's high-efficiency operating range, To increase the upper limit of the pump's high-efficiency operating range, the system ultimately outputs an optimized influent flow rate curve through optimization calculations that satisfy the aforementioned constraints. and the corresponding flow adjustment amount This adjustment is based on the optimized influent flow rate. Subtract the original predicted flow The conclusion is as follows.

[0030] In the preferred scheme, in step S2, the Particle Swarm Optimization (PSO) algorithm employs an adaptive chaotic mutation strategy based on Logistic mapping to avoid the algorithm getting trapped in local optima. The steps are as follows: S21. Initialize the particle swarm by generating the initial position sequence of particles using a Logistic chaotic mapping, as shown in the formula: ; In the formula, For the k-th generation of chaotic variables, As a control parameter, a value of 4 will be used to generate... Mapping to solution space This was used as the initial population; S22. In the iterative optimization process, an inertial weight that decays non-linearly with the number of iterations is introduced. The formula is: ; In the formula, iter is the current iteration number. The maximum number of iterations, and These are the maximum and minimum inertia weights, respectively; S23. Perform a Cauchy mutation operation on the globally optimal particle gbest generated in each iteration to produce a new mutated position gbest': ; In the formula, These are standard Cauchy distribution random numbers. The mutation operator is used; if the fitness function value corresponding to gbest' is better than gbest, then gbest' is used to replace gbest, otherwise the original value is retained until the iteration termination condition is met.

[0031] The Particle Swarm Optimization (PSO) algorithm has been improved in detail to address the problem of premature convergence in the standard algorithm when searching for complex multimodal functions.

[0032] In step S21, the initial population is generated using the Logistic chaotic mapping, as shown in the formula. Logistic mapping was used in the control parameters The completely chaotic property of the value 4 allows the initial particles to traverse the solution space uniformly and randomly. This avoids the search blind spot caused by the aggregation of initial solutions.

[0033] The arcsine function inertia weight introduced in step S22 It utilizes the slope change characteristics of the arcsine function at both ends of the interval, making the weight larger in the early stage of the iteration to maintain global search capability, and rapidly decreasing in the later stage of the iteration to focus on local development. This is more in line with the optimization law than the traditional linear decreasing strategy.

[0034] Cauchy mutation operation in step S23 It utilizes the characteristic that the Cauchy distribution has a longer tail than the Gaussian distribution. When the algorithm gets stuck in a local optimum, it can generate a larger step size perturbation to help the particles jump out of the local trap, thereby greatly improving the algorithm's ability to find the global optimum.

[0035] In the preferred scheme, step S3, the method for calculating the storage capacity of the pipeline network in each area includes: Key parameter acquisition: Obtain pipeline network parameters such as pipe diameter, pipe length, and pipe top elevation in the area, as well as the cross-sectional area of ​​the water collection tank and the parameters of the pumping station in the high-efficiency head range of the booster pump; A geometric model of the pipe network is constructed, and the cross-sectional area data points of the pipe network and the collection tank at different liquid levels are calculated based on the obtained parameters. The model is then fitted to a continuous function A(L). ; In the formula, A(L) is the cross-sectional area of ​​the water surface when the liquid level is L; The adjustable storage capacity V(L) of the pipeline network is calculated by integration: ; In the formula, Lmin and Lmax represent the lower and upper limits of the liquid level range for efficient operation of the booster pump in the pumping station.

[0036] This method constructs a geometric model of the pipeline network and fits the discrete relationship between liquid level and cross-sectional area into a continuous polynomial function. ,in Indicates liquid level as The water surface cross-sectional area at time t, pk is the polynomial coefficient. This fitting method facilitates subsequent integration calculations.

[0037] Then, through the definite integral formula Calculate the adjustable storage capacity of the pipeline network within the efficient operating liquid level range [Lmin, Lmax] of the pump station. This calculation process transforms the complex geometries of the pipeline network into mathematically calculable volume values, accurately quantifying the "reservoir" capacity of the network and providing clear physical boundary constraints for subsequent flow scheduling.

[0038] In the preferred scheme, the specific process of decomposing the wastewater treatment plant flow adjustment to each catchment area and compensating for hydraulic time lag in step S3 is as follows: Time delay coefficient calculation: The hydraulic time delay Ti from each area to the sewage treatment plant is determined by measuring the time difference between the start-up and shutdown of the pumping station in the area and the change in the influent. Decomposition Model Construction: Based on the premise that the sum of the flow adjustment amounts in each zone equals the total adjustment amount of the wastewater treatment plant, and constrained by the flow rates in each zone not exceeding the adjustable range, a linear programming model is constructed: ; In the formula, This represents the total adjustment of the wastewater treatment plant at time t. Let Ti be the adjustment amount for the i-th area, and Ti be the hydraulic time delay from the i-th area to the wastewater treatment plant; The constraints are: Traffic flow adjustment for the area: ,in For the i-th area, the flow rate adjustable range is determined based on the pipeline storage capacity; The booster pump operates in the high-efficiency range: ; Decomposition results: Target traffic flow per hour for each area over the next 24 hours: .

[0039] Because the distances of each zone to the wastewater treatment plant differ, the arrival times of the water flow vary, i.e., there is a hydraulic time lag Ti. If this time lag is not considered, the synchronous scheduling of each zone will lead to a time-discrepancy superposition of the wastewater inflow to the treatment plant. Therefore, a decomposition model is needed. It was clarified that the sewage treatment plant is Real-time traffic adjustment needs In reality, it is determined by each district. Adjustments to be executed in advance The model is formed by superimposing the data after transportation. This model ensures that the scheduling actions of each area are precisely aligned on the time axis, so that the influent flow rate of the wastewater treatment plant can strictly follow the optimized target curve.

[0040] First, to ensure the precise execution of dispatch commands on the timeline, accurate hydraulic time delay parameters must be obtained. These parameters are not calculated based on theoretical flow rates but are determined through on-site measurements. Specifically, by monitoring the start-up and shutdown times of pumping stations in each area and the moments when significant changes occur in the influent flow rate of the downstream wastewater treatment plant, the difference between the two is calculated to obtain the hydraulic time delay Ti from each area to the wastewater treatment plant. This measured data reflects the actual hydraulic transmission characteristics under a specific pipeline network topology, eliminating errors introduced by theoretical calculations and providing a time reference for subsequent timing alignment.

[0041] Based on this, a linear programming model for flow decomposition is constructed. This model uses the summation effect of flow adjustments in each catchment area equaling the total adjustment at the wastewater treatment plant as an equality constraint. Its mathematical expression is as follows: This formula profoundly reveals the spatiotemporal coupling relationship of multi-source confluence systems, where... This represents the total influent flow adjustment required by the wastewater treatment plant at the current time t, which is the global objective calculated in step S2. Represents the total number of watershed areas; This represents the flow adjustment required for the i-th catchment area; Ti is the hydraulic time delay for the i-th catchment area. The physical meaning of this formula is that the wastewater treatment plant... The changes in traffic received at any given time are actually determined by each area at its own advance time. The adjustment actions performed are transported through the pipeline network, and then... The flow rates are generated by the simultaneous arrival of pumps at the wastewater treatment plant. Through this time-series backward calculation and matching, the model ensures that pump stations located in different geographical locations can operate in coordination, enabling the wastewater treatment plant's inlet to accurately obtain the expected flow rate curve.

[0042] To ensure the physical feasibility and security of the scheduling scheme, the linear programming model is strictly constrained by two key boundary conditions during the solution process. The first is the constraint of the pipeline network's storage capacity, namely the regional flow adjustment. It must be within the adjustable flow range Within. This interval is determined by the pipeline storage capacity calculated in the preceding step S3, where... and These represent the minimum and maximum flow regulation capacity that the i-th zone can provide based on the pipeline geometry at the current liquid level. This constraint prevents pipeline overflow due to overuse of pipeline storage or pipeline depletion due to over-extraction. Secondly, there is the constraint of equipment performance, namely the adjusted instantaneous flow rate. It must be within the high-efficiency operating range of the booster pump. Inside, among which and The lower and upper limits of the high-efficiency zone of the booster pump in the i-th region are defined. This constraint ensures that even during the dynamic process of flow regulation, the pump equipment still operates under low-energy-consumption and high-efficiency conditions, avoiding sacrificing equipment lifespan or increasing additional energy consumption to meet regulation requirements.

[0043] Finally, by solving the constrained linear programming model described above, the system obtains the target flow instructions for each area at each time step within the next 24 hours, and the calculation formula is as follows: In the formula This is the final target setpoint issued to the controller of the i-th area pump station, which is determined by the original predicted base flow rate of that area. The flow adjustment amount calculated after optimization allocation This result is achieved through a combination of factors. It retains the response to baseline load trends while incorporating system-level collaborative optimization and adjustments, thus realizing a precise transformation from macro-level energy-saving goals to micro-level equipment execution instructions.

[0044] First, to ensure the precise execution of dispatch commands on the timeline, accurate hydraulic time delay parameters must be obtained. These parameters are not calculated based on theoretical flow rates but are determined through on-site measurements. Specifically, by monitoring the start-up and shutdown times of pumping stations in each area and the moments when significant changes occur in the influent flow rate of the downstream wastewater treatment plant, the difference between the two is calculated to obtain the hydraulic time delay Ti from each area to the wastewater treatment plant. This measured data reflects the actual hydraulic transmission characteristics under a specific pipeline network topology, eliminating errors introduced by theoretical calculations and providing a time reference for subsequent timing alignment.

[0045] Based on this, a linear programming model for flow decomposition is constructed. This model uses the summation effect of flow adjustments in each catchment area equaling the total adjustment at the wastewater treatment plant as an equality constraint. Its mathematical expression is as follows: This formula profoundly reveals the spatiotemporal coupling relationship of multi-source confluence systems, where... This represents the total influent flow adjustment required by the wastewater treatment plant at the current time t, which is the global objective calculated in step S2. Represents the total number of watershed areas; This represents the flow adjustment required for the i-th catchment area; Ti is the hydraulic time delay for the i-th catchment area. The physical meaning of this formula is that the wastewater treatment plant... The changes in traffic received at any given time are actually determined by each area at its own advance time. The adjustment actions performed are transported through the pipeline network, and then... The flow rates are generated by the simultaneous arrival of pumps at the wastewater treatment plant. Through this time-series backward calculation and matching, the model ensures that pump stations located in different geographical locations can operate in coordination, enabling the wastewater treatment plant's inlet to accurately obtain the expected flow rate curve.

[0046] To ensure the physical feasibility and security of the scheduling scheme, the linear programming model is strictly constrained by two key boundary conditions during the solution process. The first is the constraint of the pipeline network's storage capacity, namely the regional flow adjustment. It must be within the adjustable flow range Within. This interval is determined by the pipeline storage capacity calculated in the preceding step S3, where... and These represent the minimum and maximum flow regulation capacity that the i-th zone can provide based on the pipeline geometry at the current liquid level. This constraint prevents pipeline overflow due to overuse of pipeline storage or pipeline depletion due to over-extraction. Secondly, there is the constraint of equipment performance, namely the adjusted instantaneous flow rate. It must be within the high-efficiency operating range of the booster pump. Inside, among which and The lower and upper limits of the high-efficiency zone of the booster pump in the i-th region are defined. This constraint ensures that even during the dynamic process of flow regulation, the pump equipment still operates under low-energy-consumption and high-efficiency conditions, avoiding sacrificing equipment lifespan or increasing additional energy consumption to meet regulation requirements.

[0047] Finally, by solving the constrained linear programming model described above, the system obtains the target flow instructions for each area at each time step within the next 24 hours, and the calculation formula is as follows: In the formula This is the final target setpoint issued to the controller of the i-th area pump station, which is determined by the original predicted base flow rate of that area. The flow adjustment amount calculated after optimization allocation This result is achieved through a combination of factors. It retains the response to baseline load trends while incorporating system-level collaborative optimization and adjustments, thus realizing a precise transformation from macro-level energy-saving goals to micro-level equipment execution instructions.

[0048] In the preferred scheme, the linear programming model in step S3 also includes a minimum scour constraint to prevent low-velocity deposition in the pipeline network. The construction process includes: Define the pipeline deposition risk index When the target traffic of the area This makes the flow velocity inside the pipe Below the minimum self-cleaning flow rate Triggered at time; Add pulse flushing logic to the constraints: If a certain period of time is predicted to Inside, Duration exceeds threshold Then force setting The lower limit constraint for flow rate at any given time is: ; In the formula, To effectively flush the flow rate, The diameter of the main pipeline in the area; When solving a linear programming model, the above-mentioned deposition constraints must be satisfied simultaneously. This can be achieved by reducing... If regulating the water storage leads to a violation of sedimentation constraints, then the previous time step will be used first. The pipeline network storage space is used for flow compensation to ensure that no hardening or deposition occurs in the pipeline network.

[0049] When using pipe networks to store water, the flow velocity within the pipes is often low, which can easily lead to the settling and caking of solid particles. Therefore, a sedimentation risk index was defined, and a minimum self-cleaning flow velocity was set. When the predicted flow rate remains below this threshold for more than [time period missing]... At that time, the system forcibly triggers the scouring constraint, requiring that at a subsequent moment... Increase traffic to ,in For an effective scouring flow velocity sufficient to remove sediment, The diameter is [not specified]. This intermittent, pulsed high-flow operation is similar to "hydraulic dredging," which, while ensuring the storage function, uses the shear force of the water flow itself to prevent pipe siltation, thus resolving the contradiction of solely pursuing storage while neglecting the safety of pipeline operation and maintenance.

[0050] In the preferred scheme, step S4, the coordinated scheduling includes: Regional pump station control: The controller uses a PID algorithm to control the target flow rate of the area. Real-time collection of pump station flow rate to set value. Adjust the operating frequency of the water pump; Water plant booster pump control: The controller uses a PID algorithm to optimize the influent flow rate. Real-time collection of influent flow rate to set value. Adjust the operating frequency of the water pump; Aeration blower control: The controller uses an intelligent algorithm to set the aeration volume calculated based on the influent load as the baseline value. It also corrects the set value by combining the dissolved oxygen control target of the biological tank with the actual feedback deviation, and adjusts the blower operating frequency intermittently by taking into account the hydraulic retention time.

[0051] Targeted control strategies were adopted for the pumping stations in the area, the booster pumps of the water plant, and the aeration blowers to ensure that the upper-level optimization commands could be accurately executed by the lower-level equipment.

[0052] First, for the pumping stations distributed across various catchment areas, their controllers incorporate a proportional-integral-derivative (PID) control algorithm. This controller processes the hourly target flow rate for each area obtained from the linear programming decomposition in step S3. As the system setpoint, the actual outflow rate of the pumping station is collected in real time using a flow meter. As a feedback value, the controller calculates the deviation between the setpoint and the feedback value in real time. The inverter's output frequency is dynamically adjusted based on this deviation. By changing the pump speed, the actual flow rate can quickly and without static error follow the target flow rate curve, ensuring that the pipeline regulation and storage tasks are strictly implemented in each area.

[0053] Secondly, for the regulation of the wastewater treatment plant's influent booster pump, the controller also employs a PID closed-loop control strategy. Here, the controller's setpoint is no longer a fixed value based on experience, but rather the optimized influent flow rate calculated and output by the particle swarm optimization algorithm in step S2. The system collects readings from the influent flow meter in real time. As a process variable, the operating frequency of the booster pump unit is adjusted by outputting a control signal through PID calculation. This stage is the central execution point of the entire system. Its precise control ensures that the influent flow rate of the wastewater treatment plant can strictly follow the preset smooth curve, effectively avoiding the load impact on the biological system caused by influent fluctuations, and also ensuring that the booster pump always operates within the preset high-efficiency energy consumption range.

[0054] Finally, for the aeration blower system, which accounts for the largest proportion of energy consumption, this invention adopts an intelligent composite control strategy that integrates feedforward and feedback. The controller first uses the theoretical aeration volume calculated in step S2 based on the predicted influent load. This serves as the baseline setpoint, forming the feedforward portion of the control. Based on this, the system combines online dissolved oxygen monitoring data from the biological treatment tank to calculate the deviation between the actual and target dissolved oxygen concentrations. This deviation is then used to correct the baseline setpoint, creating a closed-loop feedback. More importantly, this intelligent algorithm comprehensively considers the impact of hydraulic retention time, i.e., the time required for wastewater to flow from the inlet to the biological reaction zone. The controller uses this time parameter to compensate for phase lag in the blower's adjustment actions, ensuring that changes in air supply precisely match the arrival time of the pollutant concentration peak in the biological treatment tank on the timeline. This avoids energy waste due to premature air supply or water quality deterioration due to delayed air supply. Through this targeted hierarchical control strategy, this invention successfully establishes a link between upper-level data optimization and lower-level equipment execution, ensuring the effective implementation of energy-saving scheduling across the entire system.

[0055] In the preferred embodiment, in step S4, the PID algorithm of the area pump station controller adopts fuzzy self-tuning PID control based on the variable universe of discourse of prediction error. The specific steps are as follows: S41. Define the error of the PID input. and its rate of change ; S42. Introduce a scaling factor based on the prediction results of S1. The interval of the fuzzy domain is dynamically adjusted based on the predicted trend of traffic changes. ; In the formula, This is the trend gain coefficient; at this point, the quantization factor of the fuzzy input is... and Revised to , ; S43. Obtain the PID parameter correction amount by querying the fuzzy inference rule table. And calculate the final control parameters: ; ; ; This allows the pump station control to increase or decrease its sensitivity in advance before drastic fluctuations in flow are predicted.

[0056] An improved PID control strategy, namely fuzzy self-tuning PID control based on predictive error variable universe of discourse, is introduced. The core of this control method lies in using the future prediction information obtained in step S1 to proactively change the controller's internal parameters, thereby overcoming the lag defect of traditional PID control. First, in step S41, the system defines the input signal of the PID controller and calculates the control error at the current moment. This error is the area target flow generated in step S3. With real-time collected pump station flow The difference between them, and the rate of change of the error are calculated simultaneously. The two parameters, namely the rate at which the error changes over time, constitute the basic input of the fuzzy control logic, reflecting the degree and trend of the current control state deviating from the target.

[0057] Next, in step S42, in order to achieve a predictive response to flow fluctuations, a scaling factor based on the prediction results is introduced. The formula for calculating this factor involves the trend gain coefficient. Predicting the flow rate at the next moment Current moment predicted flow And the average daily inflow rate Qavg. (Formula) The physical meaning of this term lies in quantifying the drastic change in flow rate at future moments. When a significant change in flow rate is predicted at the next moment, the difference in the numerator increases, causing the scaling factor to... Significantly greater than 1. In this case, the system uses this factor to quantize the fuzzy input. and The correction is made, and the calculation formula is as follows: and In this way, the system dynamically changes the domain of fuzzy control, which is equivalent to automatically scaling the input sensitivity of the controller according to the future stability or intensity of the flow, thus realizing variable domain control.

[0058] Finally, in step S43, the system retrieves the baseline correction amount of the PID parameters based on the fuzzy inference rule table. , and However, these baseline corrections are not used directly, but are instead combined with the scaling factor again. Weighted calculations are performed to obtain the final real-time control parameters. Calculation formula. , as well as This indicates that the proportional, integral, and derivative coefficients ultimately acting on the water pump frequency converter are based on the initial set values. , , This is based on a dynamically adjusted value that has been amplified or reduced. When a sharp fluctuation in flow is predicted, it is used to adjust the value accordingly. When the flow rate increases, the correction amplitude of the PID parameters is amplified, enabling the controller to take stronger regulatory actions and anticipate upcoming load shocks. Conversely, when the flow is stable, the parameter correction amplitude decreases, ensuring control stability. This mechanism gives the pump station control system predictive capabilities, enabling proactive adjustments before fluctuations occur.

[0059] In the preferred scheme, in step S4, the coordinated scheduling of the aeration blower and the water plant booster pump incorporates a biological hysteresis response constraint. The method is as follows: Calculate the oxygen transfer response time constant of a biochemical system This constant is related to the current water temperature T and the sludge concentration MLSS; Set the maximum rate of change limit for the flow rate of the water plant booster pump. This limitation is not a fixed value, but is derived in reverse from the wind turbine's response capability: ; In the formula, For the target dissolved oxygen, To maintain the minimum dissolved oxygen for microbial activity, The partial derivative sensitivity of oxygen consumption rate with respect to influent flow rate; When S2 outputs the optimized influent flow rate Required rate of change exceeds At that time, the controller forcibly limits the frequency adjustment step size of the booster pump to within. Within the corresponding frequency range, the unexecuted flow difference is fed back to step S3 and redistributed to each area's pipeline network for regulation and storage.

[0060] Biological systems exhibit a significant time lag in oxygen utilization and microbial growth and reproduction, typically slower than the response speed of hydraulic systems. To prevent excessively rapid pumping from causing oxygen deficiency in the biological system due to insufficient reaction time, the oxygen transfer response time constant of the biological system was calculated. Based on this, the limit of the maximum rate of change of the booster pump flow was derived. The physical meaning of this formula is that the rate of change in flow rate must be constrained by the dissolved oxygen response rate, ensuring that the dissolved oxygen level remains at the minimum requirement for microbial activity as the flow rate increases. The above. This cross-physical field hydraulic-biochemical coupling constraint effectively avoids the risk of damaging the biochemical environment due to blindly pursuing hydraulic balance.

[0061] In the preferred embodiment, step S4 further includes a dynamic optimization strategy: Real-time monitoring of the liquid level H(t) in the forebay of the wastewater treatment plant; If H(t) continues to rise and exceeds the preset warning level Hmax, reduce the target flow rate of each area according to the adjustable space ratio; If H(t) continues to decrease and falls below the warning level Hmin, increase the target flow rate for each zone to ensure that the forebay level remains stable within the range of [Hmin, Hmax].

[0062] In the preferred embodiment, step S4 also includes a security control strategy: When there is pipe network blockage, a surge in water volume in the area due to heavy rain, or abnormal operation of sewage treatment plant equipment, an overflow risk assessment shall be conducted. If the water level in the area's collection tank is... Greater than or equal to overflow level Or the wastewater treatment plant forebay level H(t) is greater than or equal to the overflow level. This is classified as an overflow risk. The pump station in the high-risk area switches to overflow control mode, prioritizing the prevention of overflow with the set value of safe liquid level. At this time, it is allowed to temporarily exceed the limit of the high-efficiency operation range. After the abnormal working condition is resolved and confirmed, the system switches back to dynamic optimization mode. The dynamic optimization strategy, acting as a feedback correction mechanism, aims to address biases in model predictions and unforeseen disturbances during actual operation. The system monitors the real-time liquid level in the wastewater treatment plant's forebay. This physical quantity is used as a key indicator to measure the balance between supply and demand. When monitoring data shows the forebay level... When the influent flow rate shows a continuous upward trend and exceeds the preset warning level upper limit Hmax, it indicates that the current influent flow rate exceeds the treatment or lifting capacity of the wastewater treatment plant. The system immediately triggers a negative feedback adjustment mechanism, calculates the flow reduction share required for each catchment area based on the proportion of the remaining adjustable space in the pipe network, and issues instructions to reduce the target flow rate for each area, thereby curbing the rise in water level at the source. Conversely, when the forebay water level is monitored... When the water level continues to drop and falls below the preset warning level limit Hmin, it indicates that insufficient water intake may lead to pump cavitation or reduced operating efficiency. The system then instructs each zone to increase the target flow rate and use the water stored in the pipeline to supplement the water intake, ensuring that the forebay level is always kept stable within the optimal operating range [Hmin, Hmax].

[0063] The optimized solution further integrates safety control strategies, establishing a bottom-line logic for the system under extreme operating conditions: safety takes absolute priority when safety and energy conservation conflict. When the system encounters pipe network blockage, sudden heavy rainfall causing a surge in water volume in the area, or abnormal operation of critical equipment in the wastewater treatment plant, the dispatch system executes an overflow risk assessment procedure in real time. This procedure compares the liquid levels in the collection tanks of each area in real time. With the corresponding area overflow warning level At the same time, compare the liquid level in the forebay of the sewage treatment plant. With the factory area overflow warning level If any of the above liquid level monitoring values ​​is greater than or equal to its corresponding overflow warning threshold, the system determines that there is a substantial risk of overflow.

[0064] Upon determining an overflow risk, the system immediately switches the pump station control logic in the relevant area from the conventional "dynamic optimization mode" to "overflow control mode." In this mode, the pump station operation no longer prioritizes tracking flow targets or maintaining high-efficiency operating ranges, but instead focuses on maintaining a safe liquid level, prioritizing unobstructed drainage to prevent sewage overflow. To maximize drainage capacity, the control strategy allows pump operating parameters to temporarily exceed preset high-efficiency operating range limits, sacrificing some energy consumption for maximum hydraulic transport capacity. Only after staff confirm the abnormal condition has resolved and system monitoring data returns to normal will the scheduling strategy smoothly switch back from safety mode to dynamic optimization mode, gradually reverting to the energy-saving, conventional scheduling.

[0065] Example 2 Further explanation in conjunction with Example 1, such as Figure 1-3 As shown, taking a mixed-type urban drainage system (including 3 catchment areas: residential area A, commercial area B, and industrial area C; 2 pipeline booster pump stations P1 and P2; 1 water treatment plant booster pump station P3; and 1 sewage treatment plant with a daily treatment capacity of 80,000 tons) as an example, see details below. Figure 2 The implementation steps are as follows: 1) Determination of core system parameters Key system parameters were obtained through on-site measurements, equipment manuals, and design documents, providing input for the prediction and scheduling model. The key system parameters are shown in Table 1 below: Table 1: Key System Parameters

[0066] 2) Implementation and validation of prediction models Model training and deployment: Hourly drainage flow data for areas A, B, and C over the past year, influent flow, COD, and total nitrogen data for the sewage treatment plant over the past year, and hourly rainfall data from the local weather station were collected. An LSTM-XGBoost fusion prediction model was built using Python, with a training set, validation set, and test set ratio of 7:1.5:1.5. After 100 iterations, the model converged (test set flow prediction MAPE=8.2%, total nitrogen prediction MAPE=9.5%, COD prediction MAPE=8.7%).

[0067] Prediction Results Output: The model outputs the drainage flow rate and wastewater treatment plant influent load for each area 24 hours in advance. The prediction results are shown in Table 2 below: Table 2: 24-hour output of drainage flow and wastewater treatment plant influent load for each area

[0068] 2. Scheduling Model Calculation and Implementation 1) Optimization and adjustment of influent flow rate of wastewater treatment plant Calculate the aeration air volume: Using the activated sludge process reaction kinetics formula, and substituting the following parameters: carbon oxidation coefficient a = 0.6, nitrification coefficient b = 4.5, saturated DO at 20℃ = 8.5 mg / L, wastewater oxygen transfer coefficient correction α = 0.85, salinity correction β = 0.95, oxygen transfer efficiency EA = 25%, and water temperature T = 22℃, the required aeration air volume for the next 24 hours is calculated as shown in Table 3 below: Table 3: Calculation of Aeration Air Volume Required for the Next 24 Hours

[0069] Calculations show that the aeration air volume from 0:00 to 6:00 exceeds the total high-efficiency range of the blower. Based on the actual situation on site, it is necessary to start two blowers and operate them inefficiently at the lowest frequency. The inlet water flow needs to be optimized.

[0070] Set constraints: The particle swarm optimization (PSO) algorithm is used to solve the constrained objective function, and the hourly predicted flow rate Qin(t) is adjusted to output the optimized influent flow rate. and flow adjustment amount .

[0071] Traffic fluctuation constraints: ,in Therefore ; High-efficiency constraints of booster pumps: ; High-efficiency constraints of wind turbines:

[0072] S3: Flow optimization solution: The particle swarm optimization algorithm (PSO, 50 particles, 30 iterations) was used to optimize the predicted flow rate with the goal of "G(t) falling within the efficient range". The optimization results are shown in Table 4 below: Table 4: Optimization Results

[0073] 2) Calculation of water storage capacity in the catchment area The polynomial equation for the cross-sectional area of ​​the drainage pipe in area A, calculated through modeling and fitting, is as follows: ,

[0074] Adjusting at the hourly level (time step t=1h), the corresponding hourly flow rate adjustment capacity is approximately 3170m³ / h (i.e., through water storage / release in the pipeline network, the flow rate can be adjusted by ±1570m³ per hour).

[0075] The polynomial equation for the cross-sectional area of ​​the drainage pipe in area B, calculated through modeling and fitting, is as follows: ,

[0076] Adjusted on an hourly basis, the corresponding hourly flow rate adjustment capacity is approximately 711 m³ / h.

[0077] 3) Decomposition of Flow Adjustment Targets Determine hydraulic time delay Based on on-site measurements (the time difference between the start-up and shutdown of the pump station and the change in the influent of the sewage treatment plant), the time delay from area A to the sewage treatment plant is T_1=0.8h, and the time delay from area B to the sewage treatment plant is T_2=1.0h.

[0078] Constructing a decomposition model Taking the requirement to reduce the flow rate of the wastewater treatment plant by 375 m³ / h at 18:00 as an example, this needs to be done by area A at time 18-T_1 (i.e., area A adjusts at 17:12), and by area B at time 18-T_2 (i.e., area B adjusts at 17:00). The linear programming is as follows: ; Constraints: ; ; ; ; Solving the decomposition results Target traffic flow for each area over the next 24 hours per hour:

[0079] Control and Execution: The control strategy is embedded into the automatic control system of the wastewater treatment plant and pumping station, and adjustments are made based on real-time data.

[0080] Example 3 Further illustrating with Example 1, a specific implementation method for an integrated scheduling optimization method for sewage treatment plant pipeline pump stations includes the following steps. First, historical data on sewage treatment plant influent flow, water quality concentration, drainage flow of each pump station, and regional rainfall are integrated. Combined with time rules, a fusion prediction model is used to predict future drainage flow and sewage treatment plant influent load for each catchment area. In this step, the fusion prediction model is built using a long short-term memory network combined with a gradient boosting tree. Specifically, the model is built by collecting hourly drainage flow data for each catchment area over the past year, sewage treatment plant influent flow, chemical oxygen demand (COD), and total nitrogen data over the past year, as well as hourly rainfall data from local weather stations. The collected data is divided into training, validation, and test sets, and the model is trained according to a preset ratio. The model's input features explicitly include historical load data of the sewage treatment plant, time characteristics, regional characteristics, and meteorological characteristics. The historical load data for the wastewater treatment plant includes the influent flow rate, chemical oxygen demand (COD), and total nitrogen (TNO) data for the past year over 24 hours. Temporal characteristics include hourly data, weekday, weekend, and holiday markers, as well as month data. Regional characteristics include the influent flow rate of each catchment area for the past year over 24 hours. Meteorological characteristics include hourly cumulative rainfall and daily average temperature. During model operation, a long short-term memory (LSTM) network is used to process the historical flow rate, COD, and TNO time-series data. Simultaneously, a gradient boosting tree is used to process real-time rainfall and holiday scheduling signals. Iterative training continues until the model converges, ultimately outputting the hourly influent load of the wastewater treatment plant and the drainage flow rate of each catchment area for the next 24 hours.

[0081] Based on the above prediction results, assuming a constant total predicted influent flow rate throughout the day, and considering flow fluctuations and equipment efficiency constraints, the aeration volume is calculated based on the predicted load. An optimization algorithm is then used to adjust the predicted influent flow rate of the wastewater treatment plant, outputting the optimized influent flow rate and the adjusted flow rate. During this process, the aeration volume... The calculation is based on the influent chemical oxygen demand and total nitrogen load predicted by a long short-term memory network combined with a gradient boosting tree model, and is calculated in conjunction with the reaction kinetics of the activated sludge process. The formula is as follows: ; In the formula, Let be the inflow rate at time t. Let be the amount of chemical oxygen demand removed at time t. Let be the total nitrogen removal at time t, a represent the carbon oxidation coefficient, b represent the nitrification coefficient, Cs20 represent the saturated dissolved oxygen at 20 degrees Celsius, CSW represent the saturated dissolved oxygen, and CL represent the residual dissolved oxygen. and Here, represents the wastewater oxygen transfer coefficient and salinity correction coefficient, respectively; T is the water temperature; and EA represents the oxygen transfer efficiency. The flow rate optimization solution employs a particle swarm optimization algorithm to calculate the hourly predicted flow rate. Adjustments were made, and the constraints included the wind turbine operating in its high-efficiency zone, i.e. Located at the lower limit of the high-efficiency operating range of the wind turbine and upper limit Between; the absolute value of the ratio of the influent flow rate adjustment fluctuation to the daily average influent flow rate Qavg is less than or equal to 20%; the booster pump operates in the high-efficiency zone, i.e. Located at the lower limit of the high-efficiency operating range of the booster pump and upper limit between.

[0082] To avoid getting trapped in local optima, the particle swarm optimization algorithm employs an adaptive chaotic mutation strategy based on the Logistic map. First, the particle swarm is initialized by generating the initial position sequence of particles using the Logistic chaotic map, as shown in the formula: In the formula For the k-th generation of chaotic variables, The control parameter is set to 4, and the generated... This is mapped into the solution space as the initial population. Secondly, during the iterative optimization process, an inertial weight that decays non-linearly with the number of iterations is introduced. The formula is: ; In the formula, iter represents the current iteration number. The maximum number of iterations, and These are the maximum and minimum inertia weights, respectively. Finally, a Cauchy mutation operation is performed on the globally optimal particle gbest generated in each iteration to produce a new mutated position gbest', as shown in the formula. In the formula These are standard Cauchy distribution random numbers. This is the mutation operator. If the fitness function value corresponding to the new mutated position is better than that of the original globally optimal particle, then the particle is replaced; otherwise, the original value is retained until the iteration termination condition is met.

[0083] After flow optimization, the storage capacity of the pipe network in each catchment area and the hydraulic time delay from each area to the wastewater treatment plant are calculated. A linear programming model is constructed to decompose the flow adjustment to each catchment area, generating the target hourly flow for each area. When calculating the storage capacity of the pipe network in each area, the pipe diameter, pipe length, and pipe top elevation parameters, as well as the cross-sectional area of ​​the collection tank and the parameters of the pumping stations in the high-efficiency head range of the booster pumps, are first obtained. Then, a geometric model of the pipe network is constructed, and based on the obtained parameters, the cross-sectional area data points of the pipe network and collection tank at different liquid levels are calculated and fitted as a continuous function. In the formula, A(L) is the cross-sectional area of ​​the water surface when the liquid level is L. Then, the adjustable storage capacity V(L) of the pipeline network is calculated by integration, using the formula: In the formula, Lmin and Lmax represent the lower and upper limits of the liquid level range for efficient operation of the booster pump in the pumping station.

[0084] In the flow adjustment target decomposition stage, the hydraulic time delay Ti from each area to the wastewater treatment plant is first determined by measuring the time difference between the start-up and shutdown of the pumping stations in each area and the change in the influent. Then, a linear programming model is constructed with the premise that the sum of the flow adjustment amounts of each area equals the total adjustment amount of the wastewater treatment plant, and the constraint that the flow of each area does not exceed the adjustable range. ; In the formula This represents the total adjustment of the wastewater treatment plant at time t. Let this be the adjustment amount for the i-th region. Constraints include region adjustment flow. Located within the adjustable flow range of the i-th zone, determined by the pipeline storage capacity, and with the booster pump operating in its high-efficiency zone. The decomposition result is the target flow rate per hour for each zone over the next twenty-four hours. It is equal to the originally predicted flow. With adjustment amount The sum. Furthermore, the linear programming model also includes a minimum scour constraint to prevent low-velocity deposition in the pipe network. A pipe network deposition risk index is defined as the index that determines the deposition risk when the target flow rate in a region causes the pipe velocity to fall below the minimum self-cleaning velocity. Triggered by time. If the predicted flow rate is below the minimum self-cleaning flow rate for more than a threshold period, then the lower limit constraint on flow rate for subsequent time periods will be forcibly set. In the formula To effectively flush the flow rate, Let be the diameter of the main pipeline in the area. The linear programming model must simultaneously satisfy the above sedimentation constraints when solving the problem. If reducing the flow rate for storage leads to a violation of the sedimentation constraints, then the storage capacity of the pipeline network from the previous time step should be used first for flow compensation.

[0085] Based on the generated target flow rates for each area and the optimized influent flow rates, coordinated scheduling is implemented for the area pumping stations, water plant booster pumps, and aeration blowers. The area pumping station controller uses a PID algorithm, with the area target flow rate as the setpoint, to collect real-time pump flow data and adjust the pump operating frequency. The water plant booster pump controller also uses a PID algorithm, with the optimized influent flow rate as the setpoint, to collect real-time influent flow data and adjust the pump operating frequency. The aeration blower controller employs an intelligent algorithm, using the aeration volume calculated based on the influent load as the baseline setpoint, and corrects the setpoint based on the deviation between the dissolved oxygen control target in the biological treatment tank and the actual feedback, while also considering the hydraulic retention time to intermittently adjust the blower operating frequency.

[0086] For the area pump station controller, the PID algorithm specifically adopts fuzzy self-tuning PID control based on the variable universe of discourse of prediction error. First, the error of the PID input is defined. The difference between the target flow and the actual flow is calculated, along with its rate of change. Secondly, a scaling factor based on the prediction results is introduced. The interval of the fuzzy universe of discourse is dynamically adjusted based on the predicted flow trend, using the following formula: ; In the formula This is the trend gain coefficient. At this point, the quantization factor of the fuzzy input... and Corrected to the original value divided by Finally, the PID parameter correction is obtained by querying the fuzzy inference rule table, and the final control parameters are calculated, including the proportional coefficient. Integral coefficient and differential coefficients These are the base value plus the product of the scaling factor and the correction amount, which allows the pump station control to increase or decrease its control sensitivity in advance before drastic fluctuations in the predicted flow rate.

[0087] In the coordinated scheduling, biological hysteresis response constraints were also introduced for the aeration blowers and water plant booster pumps. The oxygen transfer response time constant of the biochemical system was calculated. This constant is related to the current water temperature and sludge concentration. The maximum rate of change limit for the water plant's booster pump flow rate is not a fixed value, but rather derived inversely from the blower's response capability, using the following formula: ; In the formula For the target dissolved oxygen, To maintain the minimum dissolved oxygen required for microbial activity, the denominator includes the partial derivative sensitivity of the oxygen consumption rate with respect to the influent flow rate. When the rate of change of the optimized influent flow rate requirement exceeds this limit, the controller forcibly limits the frequency adjustment step of the booster pump to within the allowable range and feeds back the unexecuted flow difference to the pipe network of each area for regulation and storage.

[0088] The system operation also includes dynamic optimization and safety control strategies. Real-time monitoring of the wastewater treatment plant forebay level H(t) is conducted. If the level continues to rise and exceeds the preset warning level upper limit, the target flow rate for each area is reduced proportionally within the adjustable range. If the level continues to fall and falls below the warning level lower limit, the target flow rate for each area is increased to ensure the forebay level remains stable within the preset range. Overflow risk assessment is performed when pipe network blockage, heavy rain causing a surge in water volume in a region, or abnormal equipment malfunctions at the wastewater treatment plant occur. If the level in the region's collection tank is greater than or equal to the region's overflow level, or the level in the wastewater treatment plant's forebay is greater than or equal to the plant's overflow level, an overflow risk is identified. In this case, the pumping station in the risk area switches to overflow control mode, prioritizing the safety level as the set value to prevent overflow, allowing temporary breaches of the high-efficiency operating range limit. Once the abnormal condition is resolved and confirmed, the system switches back to dynamic optimization mode.

[0089] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A method for integrated scheduling and optimization of sewage treatment plant pipeline network pumping stations, characterized by: Includes the following steps: S1. Integrate historical data on sewage treatment plant influent flow, water quality concentration, drainage flow of each pumping station, and rainfall in the area, and combine them with time rules to use a fusion prediction model to predict the future drainage flow and sewage treatment plant influent load of each catchment area. S2. Based on the constant total influent flow rate predicted throughout the day, and combined with flow fluctuations and equipment efficiency constraints, the aeration volume is calculated based on the predicted load. An optimization algorithm is used to adjust the predicted influent flow rate of the wastewater treatment plant, and the optimized influent flow rate and flow adjustment amount are output. S3. Calculate the storage space of the pipe network in each catchment area and the hydraulic time delay from each area to the sewage treatment plant. Construct a linear programming model to decompose the flow adjustment amount to each catchment area and generate the target flow for each area in the future per hour. S4. Based on the target flow rate and optimized influent flow rate of each area, coordinate the scheduling of pumping stations, water plant booster pumps and aeration blowers in the area, and perform dynamic optimization and safety control in conjunction with liquid level feedback.

2. The integrated scheduling and optimization method for sewage treatment plant pipeline pumping stations according to claim 1, characterized in that: In step S1, the fusion prediction model is built using an LSTM long short-term memory network combined with an XGBoost gradient boosting tree. Specific steps include: We collected hourly drainage flow data for each catchment area over the past year, influent flow, COD and total nitrogen data for wastewater treatment plants over the past year, and hourly rainfall data from local weather stations. The collected data is divided into training set, validation set and test set, and the model is trained according to a preset ratio; Input features include historical load data of water plants, time features, regional features, and meteorological features. The time features include hours, weekdays, weekends, holiday identifiers, and months. The model uses an LSTM network to process historical flow, COD, and total nitrogen time-series data, combined with XGBoost to process real-time rainfall and holiday scheduling signals. The model is iteratively trained until it converges, and outputs the hourly influent load of the wastewater treatment plant and the drainage flow of each area for the next 24 hours.

3. The integrated scheduling and optimization method for sewage treatment plant pipeline pump stations according to claim 1, characterized in that: In step S2, based on the premise that the total predicted influent flow rate remains constant throughout the day, and with the constraints that the influent flow rate of the wastewater treatment plant fluctuates by ≤20% and that the influent lift pumps and aeration blowers of the wastewater treatment plant operate efficiently, the predicted influent flow rate of the water treatment plant is optimized and adjusted as follows: The aeration volume G(t) is calculated based on the influent COD and total nitrogen load predicted by the LSTM-XGBoost model, combined with the reaction kinetics of the activated sludge process. The formula is as follows: ； In the formula, Let be the inflow rate at time t. Let t be the COD removal amount. Let represent the total nitrogen removal at time t, a represent the carbon oxidation coefficient, b represent the nitrification coefficient, Cs20 represent saturated DO at 20℃, CSW represent saturated DO, and CL represent the remaining DO. and These represent the wastewater oxygen transfer coefficient and salinity correction coefficient, respectively, where T is the water temperature and EA represents the oxygen transfer efficiency. The flow optimization solution uses the Particle Swarm Optimization (PSO) algorithm to calculate the hourly predicted flow. Adjustments are made, and the constraints are as follows: The fan operates in the high-efficiency range: ; Traffic volatility: ; The booster pump operates in the high-efficiency range: ; The particle swarm optimization algorithm (PSO) is used to solve the constrained objective function for hourly predicted flow. Adjustments were made to optimize the influent flow rate. and flow adjustment amount ; In the formula, , These are the lower and upper limits of the high-efficiency operating range for wind turbines, respectively. This represents the average daily inflow rate. , These are the lower and upper limits of the high-efficiency operating range of the booster pump, respectively.

4. The integrated scheduling and optimization method for sewage treatment plant pipeline pumping stations according to claim 3, characterized in that: In step S2, the Particle Swarm Optimization (PSO) algorithm employs an adaptive chaotic mutation strategy based on the Logistic mapping to avoid getting trapped in local optima. The steps are as follows: S21. Initialize the particle swarm by generating the initial position sequence of particles using a Logistic chaotic mapping, as shown in the formula: ； In the formula, For the k-th generation of chaotic variables, As a control parameter, a value of 4 will be used to generate... Mapping to solution space This was used as the initial population; S22. In the iterative optimization process, an inertial weight that decays non-linearly with the number of iterations is introduced. The formula is: ； In the formula, iter represents the current iteration number. The maximum number of iterations, and These are the maximum and minimum inertia weights, respectively; S23. Perform a Cauchy mutation operation on the globally optimal particle gbest generated in each iteration to produce a new mutated position gbest': ； In the formula, These are standard Cauchy distribution random numbers. For mutation operators; If the fitness function value corresponding to gbest' is better than gbest, then replace gbest with gbest'; otherwise, retain the original value until the iteration termination condition is met.

5. The integrated scheduling and optimization method for sewage treatment plant pipeline pump stations according to claim 1, characterized in that: In step S3, the method for calculating the storage capacity of the pipeline network in each area includes: Key parameter acquisition: Obtain pipeline network parameters such as pipe diameter, pipe length, and pipe top elevation in the area, as well as the cross-sectional area of ​​the water collection tank and the parameters of the pumping station in the high-efficiency head range of the booster pump; A geometric model of the pipe network is constructed, and the cross-sectional area data points of the pipe network and the collection tank at different liquid levels are calculated based on the obtained parameters. The model is then fitted to a continuous function A(L). ； In the formula, A(L) is the cross-sectional area of ​​the water surface when the liquid level is L; The adjustable storage capacity V(L) of the pipeline network is calculated by integration: ； In the formula, Lmin and Lmax represent the lower and upper limits of the liquid level range for efficient operation of the booster pump in the pumping station.

6. The integrated scheduling and optimization method for sewage treatment plant pipeline pumping stations according to claim 5, characterized in that: In step S3, the specific process of decomposing the wastewater treatment plant flow adjustment to each catchment area and compensating for hydraulic time lag is as follows: Time delay coefficient calculation: The hydraulic time delay Ti from each area to the sewage treatment plant is determined by measuring the time difference between the start-up and shutdown of the pumping station in the area and the change in the influent. Decomposition Model Construction: Based on the premise that the sum of the flow adjustment amounts in each zone equals the total adjustment amount of the wastewater treatment plant, and constrained by the flow rates in each zone not exceeding the adjustable range, a linear programming model is constructed: ； In the formula, This represents the total adjustment of the wastewater treatment plant at time t. Let Ti be the adjustment amount for the i-th area, and Ti be the hydraulic time delay from the i-th area to the wastewater treatment plant; The constraints are: Traffic flow adjustment for the area: ,in For the i-th area, the flow rate adjustable range is determined based on the pipeline storage capacity; The booster pump operates in the high-efficiency range: ; Decomposition results: Target traffic flow per hour for each area over the next 24 hours: .

7. The integrated scheduling and optimization method for sewage treatment plant pipeline pumping stations according to claim 6, characterized in that: The linear programming model in step S3 also includes a minimum scour constraint to prevent low-velocity deposition in the pipe network. The construction process includes: Define the pipeline deposition risk index When the target traffic of the area This makes the flow velocity inside the pipe Below the minimum self-cleaning flow rate Triggered at time; Add pulse flushing logic to the constraints: If a certain period of time is predicted to Inside, Duration exceeds threshold Then force setting The lower limit constraint for flow rate at any given time is: ； In the formula, To effectively flush the flow rate, The diameter of the main pipeline in the area; When solving a linear programming model, the above-mentioned deposition constraints must be satisfied simultaneously. This can be achieved by reducing... If regulating the water storage leads to a violation of sedimentation constraints, then the previous time step will be used first. The pipeline network storage space is used for flow compensation to ensure that no hardening or deposition occurs in the pipeline network.

8. The integrated scheduling and optimization method for sewage treatment plant pipeline pumping stations according to claim 1, characterized in that: In step S4, the coordinated scheduling includes: Regional pump station control: The controller uses a PID algorithm to control the target flow rate of the area. Real-time collection of pump station flow rate to set value. Adjust the operating frequency of the water pump; Water plant booster pump control: The controller uses a PID algorithm to optimize the influent flow rate. Real-time collection of influent flow rate to set value. Adjust the operating frequency of the water pump; Aeration blower control: The controller uses an intelligent algorithm to set the aeration volume calculated based on the influent load as the baseline value. It also corrects the set value by combining the dissolved oxygen control target of the biological tank with the actual feedback deviation, and adjusts the blower operating frequency intermittently by taking into account the hydraulic retention time.

9. The integrated scheduling and optimization method for sewage treatment plant pipeline pump stations according to claim 8, characterized in that: In step S4, the PID algorithm of the area pump station controller adopts fuzzy self-tuning PID control based on the variable universe of discourse of prediction error. The specific steps are as follows: S41. Define the error of the PID input. and its rate of change ; S42. Introduce a scaling factor based on the prediction results of S1. The interval of the fuzzy domain is dynamically adjusted based on the predicted trend of traffic changes. ； In the formula, This is the trend gain coefficient; at this point, the quantization factor of the fuzzy input is... and Revised to , ; S43. Obtain the PID parameter correction amount by querying the fuzzy inference rule table. And calculate the final control parameters: ； ； ； This allows the pump station control to increase or decrease its sensitivity in advance before drastic fluctuations in flow are predicted.

10. The integrated scheduling and optimization method for sewage treatment plant pipeline pump stations according to claim 8, characterized in that: In step S4, the coordinated scheduling of the aeration blower and the water plant booster pump introduces a biological hysteresis response constraint, the method of which is as follows: Calculate the oxygen transfer response time constant of a biochemical system This constant is related to the current water temperature T and the sludge concentration MLSS; Set the maximum rate of change limit for the flow rate of the water plant booster pump. This limitation is not a fixed value, but is derived in reverse from the wind turbine's response capability: ； In the formula, For the target dissolved oxygen, To maintain the minimum dissolved oxygen for microbial activity, The partial derivative sensitivity of oxygen consumption rate with respect to influent flow rate; When S2 outputs the optimized influent flow rate Required rate of change exceeds At that time, the controller forcibly limits the frequency adjustment step size of the booster pump to within. Within the corresponding frequency range, the unexecuted flow difference is fed back to step S3 and redistributed to each area's pipeline network for regulation and storage.

11. The integrated scheduling and optimization method for sewage treatment plant pipeline pumping stations according to claim 1, characterized in that: Step S4 also includes a dynamic optimization strategy: Real-time monitoring of the liquid level H(t) in the forebay of the wastewater treatment plant; If H(t) continues to rise and exceeds the preset warning level Hmax, reduce the target flow rate of each area according to the adjustable space ratio; If H(t) continues to decrease and falls below the warning level Hmin, increase the target flow rate for each zone to ensure that the forebay level remains stable within the range of [Hmin, Hmax].

12. The integrated scheduling and optimization method for sewage treatment plant pipeline pump stations according to claim 1, characterized in that: Step S4 also includes security control strategies: When there is pipe network blockage, heavy rain causing a surge in water volume in the area, or abnormal operation of sewage treatment plant equipment, an overflow risk assessment is conducted. If the water level in the area's collection tank is... greater than or equal to overflow level Or the wastewater treatment plant forebay level H(t) is greater than or equal to the overflow level. This is classified as an overflow risk; The pump station in the high-risk area switches to overflow control mode, prioritizing the prevention of overflow with the set value of safe liquid level. At this time, it is allowed to temporarily exceed the limit of the high-efficiency operation range. After the abnormal working condition is resolved and confirmed, the system switches back to dynamic optimization mode.

Images