Intermittent aeration control method based on multi-region and multi-working condition

Abstract

The application discloses a kind of intermittent aeration control methods based on multi-region multi-working condition, which comprises: collecting multi-source data and interacting with intelligent pipe network management system to identify water intake condition, and reserving buffer volume in advance under abnormal water intake condition;Construct a three-variable reduced-order water quality dynamics model, and use Lagrange accompanying state method to solve the feedforward theory aeration time of each biological reaction tank online;The feedback time compensation value of each biological reaction tank is calculated through multi-stage feedback mechanism, and then the aeration time of each biological reaction tank is obtained by adding the feedforward theory time;Select the maximum value of the aeration time of each biological reaction tank as the operation control input of total blower set, calculate the ratio of the aeration time of each biological reaction tank, and adjust the opening of the electric regulating valve on the corresponding aeration branch pipe;Combined with load state, switch between zero, half and full power to execute.The application can reduce aeration energy consumption to the greatest extent and improve system impact resistance under the premise of guaranteeing stable and standard effluent water quality.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control technology for wastewater treatment, and specifically relates to an intermittent aeration control method based on multiple regions and multiple operating conditions. Background Technology

[0002] In modern wastewater treatment processes, intermittent aeration technology is widely used in AAO (anaerobic-anoxic-aerobic) processes because it can achieve alternating aerobic, anoxic, and anaerobic states within a physical space over time. Especially in the process of achieving short-cut nitrification and denitrification in AAO, the precise control of dissolved oxygen (DO) concentration and the dynamic allocation of aeration time are extremely demanding. Reasonable intermittent aeration can not only significantly improve nitrogen and phosphorus removal efficiency but also significantly reduce the system's carbon source dosage and blower power consumption.

[0003] However, existing wastewater treatment plants still face many technical bottlenecks in intermittent aeration control:

[0004] First, existing intelligent control technologies mostly rely on complex activated sludge models (such as the ASM series), which have many state variables and are extremely difficult to calibrate, making it difficult to achieve online solutions in actual controllers with control cycles of hours or even minutes. On the other hand, black-box models based on pure data-driven approaches (such as neural networks) often suffer from poor generalization ability and are prone to failure when faced with drastic fluctuations in influent load.

[0005] Second, actual wastewater treatment plants often contain multiple parallel biological reaction zones that share the same set of aeration blowers. Traditional control methods often use equal distribution of air volume or fixed valve opening, ignoring the differences in actual oxygen demand caused by uneven influent distribution and sludge concentration in each zone, resulting in some zones being over-aerated while others are under-aerated.

[0006] Third, there is a lack of a comprehensive shock mitigation strategy that integrates plant and network operations. When encountering abnormal influent conditions such as heavy rain or concentrated discharge from the pipeline network, traditional single-point level feedback control suffers from severe lag, often leading to frequent start-ups and shutdowns of booster pumps, severe hydraulic shocks to the biological treatment tank, and consequently disrupting the fragile balance of short-cut nitrification and other biological reactions.

[0007] Therefore, there is an urgent need for an intermittent aeration control method that can minimize aeration energy consumption and improve the system's shock resistance while ensuring stable effluent quality. Summary of the Invention

[0008] In view of the above-mentioned shortcomings of the existing technology, the purpose of this invention is to provide an intermittent aeration control method based on multiple regions and multiple operating conditions. This method can minimize aeration energy consumption and improve the system's shock resistance while ensuring that the effluent water quality meets the standards.

[0009] The technical solution of this invention is implemented as follows:

[0010] A method for controlling intermittent aeration based on multiple regions and operating conditions, specifically including the following steps:

[0011] S1: Real-time acquisition of multi-source heterogeneous data from the wastewater treatment plant's influent system, as well as water quality data from the biological reactor and effluent outlet; and interaction with the intelligent pipeline network management system to identify influent conditions, which are classified as normal or abnormal influent conditions, and the required reserved buffer volume of the influent tank is calculated under abnormal influent conditions. Increase the pumping volume and lower the inlet water level in advance;

[0012] S2: Construct a three-variable water quality kinetic model for reducing nitrogen levels, including nitrate nitrogen, ammonia nitrogen, and dissolved oxygen. Input the influent water quality and quantity characteristics into the model, and use the Lagrange adjoint state method and gradient descent algorithm to iteratively optimize the aeration start and stop times, outputting the feedforward theoretical aeration time for each bioreactor. ;

[0013] S3: Calculate the feedback time compensation value for each bioreactor through a multi-level feedback mechanism. Then, the aeration time for each biological reactor is obtained by adding it to the feedforward theoretical time. ;

[0014] S4: Based on the aeration time of each bioreactor, select the maximum value as the operating control input of the main blower unit; at the same time, calculate the ratio of the aeration time of each bioreactor and adjust the opening of the electric regulating valve on the corresponding aeration branch pipe; combined with the load status, switch between zero power, half power and full power variable power states to achieve coordinated optimization control of intermittent aeration of each bioreactor.

[0015] Furthermore, in step S1, under abnormal water intake conditions, the sludge level in the secondary sedimentation tank is monitored in real time. When the sludge level in the secondary sedimentation tank reaches or exceeds the preset sludge level warning value, the water delivery rate of the booster pump is adjusted so that the water delivery rate of the booster pump is [value missing]. ,in It is to increase the target set water delivery volume of the booster pump. It is the real-time water inflow rate.

[0016] Furthermore, in step S1, the system interacts with the intelligent pipeline network management system to obtain the pipeline network catchment flow sequence within the future prediction time domain. when At that time, the water intake condition is the normal water intake condition; when At that time, the water intake condition is an abnormal water intake condition; when it is an abnormal water intake condition, the required reserved buffer volume of the water intake booster pump station is calculated according to the following formula. :

[0017]

[0018] Where t is the current time; To predict the duration of hydraulic impact, h; for The inflow rate of the pipeline network at any time, in m 3 / h; To increase the maximum allowable safe water delivery flow rate of the pump set, m 3 / h.

[0019] Furthermore, the state variable vector of the reduced-order water quality dynamics model The system dynamic equations are as follows:

[0020]

[0021]

[0022]

[0023] in, The values ​​represent the rates of change of nitrate nitrogen, ammonia nitrogen, and dissolved oxygen concentrations over time, in mg / (L·h). The dilution factor is the ratio of influent flow rate to the volume of the biological reactor, h. -1 ; The influent ammonia nitrogen concentration is in mg / L. The yield of heterotrophic bacteria is dimensionless. Nitrogen content in biomass, dimensionless; and These represent the maximum specific rates of nitrification by autotrophic bacteria and denitrification by heterotrophic bacteria, respectively, d -1 ; This is a dimensionless parameter for aeration control. h is the oxygen transfer coefficient. -1 ; Dissolved oxygen saturation concentration, mg / L; Monod activation coefficient of nitrate nitrogen; Monod activation coefficient of ammonia nitrogen; Monod activation coefficient of dissolved oxygen; This represents the inhibition coefficient of dissolved oxygen.

[0024] Furthermore, and Based on the concentration of suspended solids in the mixture and the water temperature in the biological reactor Dynamic adaptive correction is performed, and the correction formula is as follows:

[0025]

[0026] in, , and The constant coefficients are dimensionless and obtained through online parameter identification using the Gauss-Newton method based on historical operating data.

[0027] Furthermore, in step S2, the constrained optimization problem with minimizing total aeration energy consumption as the optimization objective is defined as: the objective function for minimizing total aeration energy consumption. The constraint is the weighted average ammonia nitrogen concentration in the effluent. Below the set threshold :

[0028]

[0029] Where T is the control cycle duration, in hours (h); Let be the ammonia nitrogen concentration in the bioreactor at time t, in mg / L; Let be the aeration control state parameter at time t, with a value range of [0, 1] and dimensionless.

[0030] Furthermore, in step S3, the multi-level feedback mechanism specifically involves: performing primary feedback time compensation based on the ammonia nitrogen and nitrate nitrogen data of the effluent from each biological reactor; and performing secondary feedback time compensation based on the total nitrogen data of the effluent.

[0031] Furthermore, the first-level feedback time compensation value is calculated using a discrete incremental PID control algorithm, and the calculation formula is as follows:

[0032]

[0033] in, Let be the feedback time compensation value for the k-th sampling period, min; , , These are the proportional coefficient, integral coefficient, and derivative coefficient of the feedback controller, respectively, and are dimensionless. The ammonia nitrogen concentration deviation of the current control step. , This represents the measured ammonia nitrogen value in the effluent from the biological reactor. Set the target value for ammonia nitrogen in mg / L; and These represent the ammonia nitrogen concentration deviations for the previous and two previous sampling periods, respectively, in mg / L.

[0034] Furthermore, in step S3, if the nitrate nitrogen index at the outlet of the bioreactor reaches the first warning value and the ammonia nitrogen at the outlet of the bioreactor is lower than the set safety lower limit, the target aeration time of the corresponding bioreactor is forcibly corrected to zero; if the nitrate nitrogen index at the outlet of the bioreactor reaches the first warning value and the ammonia nitrogen at the outlet of the bioreactor is higher than the set risk upper limit, the target aeration time is guaranteed first, and an alarm for artificial carbon source supplementation is triggered.

[0035] Furthermore, in step S4, the control state parameters in the variable power state are... The range of values ​​is expanded to When the continuous non-aeration time exceeds the set high load threshold, or the feedforward-judged weighted average effluent ammonia nitrogen concentration approaches... And when the displacement control range fails to effectively lower the vehicle, switch to... Full-power aeration; when the previous continuous aeration time is less than the set low-load threshold, and the total aeration time of the system is sufficient, switch to full-power aeration; Half-power aeration is used to reduce energy consumption; at the same time, the shortest continuous running time and the shortest continuous shutdown time of the blower are set.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1. This invention effectively solves the problems of computing power bottleneck, uneven distribution, and poor shock resistance in the traditional intermittent aeration process of sewage treatment by deeply integrating and coordinating the reduced-order water quality dynamics model, plant-network linkage feedforward-feedback composite control, multi-regional dynamic air volume allocation, and variable power execution strategy. Under the premise of ensuring stable effluent water quality compliance, it minimizes aeration energy consumption and improves the system's shock resistance.

[0038] 2. This invention breaks through the physical boundaries of the factory area, interacts in real time with the intelligent pipeline network management system, obtains the predicted water flow sequence of the pipeline network in advance, and calculates the reserved buffer volume before the arrival of high flow. The system proactively empties the influent tank and lowers its level in advance to absorb instantaneous hydraulic shocks. Simultaneously, it monitors the sludge level in the secondary sedimentation tank in real time to adjust the water delivery rate, preventing large amounts of sludge from being flushed to the effluent end due to high flow rates, thus protecting the system's microbial concentration. This transforms the traditional passive response into an active feedforward approach. Furthermore, the sludge level in the secondary sedimentation tank acts as a hard safety constraint, effectively mitigating hydraulic load fluctuations and reducing the risk of effluent exceeding standards under extreme influent conditions by more than 85%, ensuring the continuous and stable operation of the biological process.

[0039] 3. This invention creatively utilizes three core variables (nitrate nitrogen, ammonia nitrogen, and dissolved oxygen) that dominate the short-cut nitrification-denitrification process to construct a simplified three-variable reduced-order water quality kinetic model. This model retains the Monod nonlinear kinetic mechanism of substrate degradation while exponentially reducing the computational dimensionality. Furthermore, it dynamically and adaptively corrects the biochemical reaction rate by incorporating water temperature and mixed liquor suspended solids concentration within the bioreactor, aiming to minimize total aeration energy consumption while ensuring the effluent ammonia nitrogen concentration remains below a set threshold. This enables the controller to perform online rolling optimization within millisecond-level operation cycles using the Lagrange adjoint state method and gradient descent method, thereby simultaneously improving control accuracy and computational efficiency.

[0040] 4. Due to unmodeled dynamics and environmental noise disturbances in actual wastewater treatment processes, this invention further constructs a multi-level feedback mechanism including water quality deviation and water quantity feedforward: First-level feedback time compensation is performed using ammonia nitrogen and nitrate nitrogen data from the biological reactor effluent, and second-level feedback time compensation is performed using total nitrogen data from the effluent. Simultaneously, a safety hard constraint logic is constructed: when nitrate nitrogen reaches the warning value and ammonia nitrogen is below the safety lower limit, the aeration time is forcibly reset to zero to prevent over-aeration from damaging the short-range nitrifying bacteria community; when nitrate nitrogen reaches the warning value and ammonia nitrogen is above the risk upper limit, a carbon source artificial compensation alarm is triggered to avoid denitrification collapse. This combination of feedforward theoretical values ​​and feedback compensation values ​​forms a feedforward + feedback composite control mechanism, which can significantly improve the system's robustness to unmodeled disturbances and environmental noise.

[0041] 5. This invention compares the final target aeration times of multiple independent bioreactors and selects the maximum value as the operating control input duration of the main blower unit; simultaneously, it calculates the ratio of the target aeration times of each bioreactor and adaptively adjusts the opening distribution ratio of the electric regulating valves on the corresponding aeration branch pipes. During execution, combined with the current system load status, at zero power ( ), half power ( ) and full power ( The system switches between variable power states, effectively eliminating localized over- and under-aeration. The introduction of half-power allows the blower to match the actual oxygen consumption rate during low-load periods, reducing aeration energy consumption. Simultaneously, embedded absolute safety hard constraints of minimum operating and minimum downtime effectively shield against high-frequency start-stop abrupt commands caused by probe water quality fluctuations or signal interference. The daily average start-stop frequency of large centrifugal blowers and booster pumps decreases by over 60%, significantly reducing motor thermal stress fatigue and mechanical bearing wear, thereby significantly reducing the overall maintenance cost of all plant equipment. Attached Figure Description

[0042] Figure 1 - Control logic flowchart of the present invention.

[0043] Figure 2 - Example: Diagram showing the effect of intermittent aeration control of the parallel south and north biological pools in a wastewater treatment plant on a single day.

[0044] Figure 3 - Example: Intermittent nighttime aeration effect diagram of a wastewater treatment plant for one week.

[0045] Figure 4 - Example: Comparison of automatic and manual control effects in a wastewater treatment plant. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0047] A method for intermittent aeration control based on multiple regions and operating conditions, see [link to relevant documentation]. Figure 1 Specifically, it includes the following steps:

[0048] S1: Multi-source data acquisition and plant-network linkage condition identification:

[0049] The system collects multi-source heterogeneous data from the wastewater treatment plant's influent system in real time (including influent flow rate, influent chemical oxygen demand, influent ammonia nitrogen, etc.), as well as water quality data from the biological reactor and effluent (wastewater effluent) (including mixed liquor suspended solids concentration, temperature, dissolved oxygen, effluent ammonia nitrogen, and nitrate nitrogen, etc.). The influent CNC drive system interacts with the intelligent pipeline network management system to obtain the pipeline network catchment flow sequence for the future predicted time domain. .

[0050] The system's water intake conditions are divided into normal water intake conditions and abnormal water intake conditions. The hydraulic impact limit of the pipe network is defined as follows: ( When the predicted flow meets the continuous Each sampling period When the condition is determined to be normal water intake, an energy-saving pumping strategy to smooth liquid level fluctuations is adopted. This involves using the deviation between the current liquid level and the target liquid level in the intake tank, as well as the rate of liquid level change, as control inputs to dynamically and smoothly adjust the operating frequency of the booster pump. This minimizes the start-up and shutdown frequency of the booster pump while ensuring the liquid level remains within a safe range. When this occurs, it is determined to be an abnormal water inflow condition. Under abnormal water inflow conditions, the system calculates in advance the required reserve buffer volume of the inflow tank. The calculation formula is as follows:

[0051]

[0052] in, The current moment; The variable is the integral variable, representing the instantaneous moment in the prediction time domain. ; To predict the duration of hydraulic impact, h; for The inflow rate of the pipeline network at any time, in m 3 / h; To increase the maximum allowable safe water delivery flow rate of the pump set, m 3 / h. System based The operating frequency of the booster pump is forcibly corrected at the current moment, and the liquid level in the inlet pool is lowered in advance to absorb the instantaneous hydraulic impact.

[0053] Simultaneously, the sludge level in the secondary sedimentation tank is introduced as a hard safety constraint. If the sludge level in the secondary sedimentation tank is greater than or equal to the sludge level warning value, the pumping water volume will be adjusted accordingly. This prevents a large amount of sludge from the biological treatment tank from being washed to the effluent end due to high flow rate impact, thereby protecting the system's microbial concentration. It is to increase the target set water delivery volume of the booster pump. It is the real-time water inflow rate.

[0054] S2: Reduced-order feedforward prediction and Lagrange adjoint state optimization solution:

[0055] Based on the current influent operating conditions and collected data, the influent water quality and quantity characteristics are input into a pre-constructed reduced-order water quality dynamics model based on ordinary differential equations. To overcome the drawback of complex state variables in traditional activated sludge models, this invention constructs a three-variable reduced-order model, with the three variables including nitrate nitrogen concentration. ammonia nitrogen concentration and dissolved oxygen concentration Its state variable vector The system dynamics equations are defined as follows:

[0056] Nitrate nitrogen equation:

[0057]

[0058] Ammonia nitrogen equation:

[0059]

[0060] Dissolved oxygen equation:

[0061]

[0062] in, The values ​​represent the rates of change of nitrate nitrogen, ammonia nitrogen, and dissolved oxygen concentrations over time, in mg / (L·h). The dilution factor is the ratio of influent flow rate to the volume of the biological reactor, h. -1 ; The influent ammonia nitrogen concentration is in mg / L. The yield of heterotrophic bacteria is dimensionless. Nitrogen content in biomass, dimensionless; and These represent the maximum specific rates of nitrification by autotrophic bacteria and denitrification by heterotrophic bacteria, respectively, d -1 ; This is a dimensionless parameter for aeration control. h is the oxygen transfer coefficient. -1 ; Dissolved oxygen saturation concentration, mg / L; For the corresponding substrate Monod activation coefficients for (concentrations of nitrate nitrogen, ammonia nitrogen, and dissolved oxygen) The inhibition coefficient for the corresponding substrate is calculated accordingly. and The substrate S represents the dissolved oxygen concentration, and its corresponding half-saturation constant is given. (mg / L) and inhibition constant (mg / L) is determined by the biochemical parameters set by the system. That is... Monod activation coefficient of nitrate nitrogen; Monod activation coefficient of ammonia nitrogen; Monod activation coefficient of dissolved oxygen; The constants 2.86 and 4.57 in the kinetic equation are the stoichiometric oxygen equivalent conversion factors fixed in the standard activated sludge model. middle Specific parameters representing heterotrophic bacteria. middle Specific parameters representing autotrophic bacteria.

[0063] Combined with the water temperature in the biological reactor and the concentration of suspended solids in the mixture right and Perform dynamic adaptive correction:

[0064]

[0065] in, and The constant coefficients are dimensionless and obtained through online parameter identification using the Gauss-Newton method based on historical operating data.

[0066] Based on this model, a constrained optimization control problem is constructed with the objective of minimizing total aeration energy consumption: The constraint is that the weighted average effluent ammonia nitrogen concentration is lower than a set threshold. :

[0067]

[0068] Where T is the control cycle duration, in hours (h); Let be the ammonia nitrogen concentration in the bioreactor at time t, in mg / L; Let be the aeration control state parameter at time t, with a value range of [0, 1], and dimensionless; The ammonia nitrogen threshold, in mg / L, set for regulations or processes.

[0069] To solve this nonlinear constrained optimization problem, a Lagrangian function is constructed and an adjoint state vector is introduced. , Hamiltonian functions are constructed to correspond to the adjoint states of the three water quality state variables: nitrate nitrogen, ammonia nitrogen, and dissolved oxygen. The gradient descent algorithm is used to determine the start time of intermittent aeration. and closing time nodes Perform iterative optimization: After iterative solution, the required feedforward theoretical aeration time for each independent bioreactor in the future period is output. .

[0070] S3: Multi-level parallel feedback compensation mechanism:

[0071] Due to unmodeled dynamics and environmental noise disturbances in actual wastewater treatment processes, the feedforward theoretical model inevitably contains biases. A multi-level feedback mechanism incorporating water quality bias and water quantity feedforward is constructed to correct the theoretical aeration time.

[0072] Level 1 Feedback: Calculate the actual measured value of ammonia nitrogen in the effluent from the biological reactor at the current moment. Compared with the target set value deviation The feedback time compensation component is calculated using a discrete incremental PID control algorithm.

[0073]

[0074] in, Let be the feedback time compensation value for the k-th sampling period, min; , , These are the proportional coefficient, integral coefficient, and derivative coefficient of the feedback controller, respectively, and are dimensionless. The ammonia nitrogen concentration deviation of the current control step. , This represents the measured ammonia nitrogen value in the effluent from the biological reactor. Set the target value for ammonia nitrogen in mg / L; and These represent the ammonia nitrogen concentration deviations for the previous and two previous sampling periods, respectively, in mg / L.

[0075] Secondary feedback and hard safety constraints: If the nitrate nitrogen level at the outlet of the biological reactor reaches the set first warning value. Furthermore, the ammonia nitrogen at the outlet of the biological reactor is below the set safety lower limit. (Indicating a serious risk of over-aeration), in mg / L, triggers a safety hard constraint, forcibly correcting the feedback compensation amount to make the final aeration time zero; if the nitrate nitrogen index reaches the first warning value and the ammonia nitrogen is higher than the set risk limit (indicating a serious carbon source deficiency leading to denitrification obstruction), the system maintains the target aeration time and triggers an alarm signal for "linked carbon source artificial compensation addition" to the external system.

[0076] The multi-level feedback time compensation value Compared with theoretical aeration time By superimposing these values, the final target aeration time for each independent bioreactor can be obtained. .

[0077] S4: Multi-zone coordinated aeration and variable power operation:

[0078] For a system comprising n (n≥2) parallel bioreactors, obtain their final target aeration time for each. The operating control input duration for the main blower unit is set to... To overcome the physical limitations of unbalanced air supply across multiple zones in a single-pipeline network, the ratio of target aeration times for each independent bioreactor is calculated, and the opening degree of the electric regulating valves on the corresponding aeration branch pipes is adaptively adjusted. The opening ratio of the electric regulating valves on the branch pipes of each bioreactor is as follows: Dynamically adjusted to:

[0079]

[0080] Where i is 1 or 2 When the final target aeration time for some bioreactors has been met while that for others has not, the system automatically reduces or closes the branch valves of the bioreactors that have met the target, while the main blower continues to aerate the bioreactors that have not met the target.

[0081] At the variable power execution level, the control state parameters The range of values ​​is expanded to Switching based on system load status:

[0082] When the previous continuous non-aeration time exceeds the set high load threshold This indicates a significant accumulation of oxygen-consuming substrates within the system, necessitating a switch to [a specific method / mechanism]. Full-power, full-frequency aeration.

[0083] When the last continuous aeration time is less than the set low load threshold And the current feedback ammonia nitrogen level is close to the threshold. When the system has a safety margin greater than a preset percentage, it determines that the current operating condition is a low water inflow or low load condition, and controls the blower frequency converter to switch to... Half-power, half-frequency aeration.

[0084] In addition, the underlying control system superimposes hard constraint rules for equipment start-stop protection: setting the minimum continuous operating time of the blower. and shortest continuous downtime This is to forcibly block illegal start / stop commands caused by high-frequency measurement noise from sensors.

[0085] Example

[0086] I. Implementation Environment and Overall System Hardware Architecture

[0087] This embodiment takes a typical large-scale wastewater treatment plant that adopts the AAO (anaerobic-anoxic-aerobic) process and is committed to achieving short-cut nitrification and denitrification as an example.

[0088] The plant's biological system is divided into a south biological tank and a north biological tank, both sharing a main aeration centrifugal blower system. The main blower duct branches into two distribution branches at the front end of the biological tank, each equipped with an independent air flow meter and an electrically adjustable air volume valve. Traditional control methods struggle to cope with drastic fluctuations in influent load, and uneven water distribution between the south and north biological tanks often results in over-aeration on one side and under-aeration on the other. This not only disrupts the fragile short-range nitrifying bacteria community environment but also leads to significant energy waste.

[0089] This invention proposes an intermittent aeration control method based on multiple regions and operating conditions, which relies on a "data-driven multi-region optimization controller" deployed on an edge computing gateway or plant-level cloud platform. This controller connects downwards to the underlying PLC system via industrial Ethernet, reading sensor data and issuing control commands; upwards, it interacts with upstream platforms such as the Three Gorges Smart Pipeline Management System in real time, achieving plant-network linkage.

[0090] The core reduced-order dynamics model and the symbols and definitions of system control-related variables involved in this control system are shown in Table 1.

[0091]

[0092] II. Detailed Implementation Steps of the Control Method

[0093] Step S1: Multi-source data acquisition and plant-network linkage condition identification

[0094] The controller uses a set period (e.g.) (min) Real-time acquisition of multi-source heterogeneous data from the system. Through integration with the smart pipeline network, future predictive time-domain data (such as...) is obtained. Pipeline runoff sequence within hours .

[0095] Operating condition identification and plant-network linkage drain logic:

[0096] Set the hydraulic impact safety limit of the pipeline network to .

[0097] (1) Normal water intake condition: When the continuous water intake condition is met within the prediction time domain Each sampling period At the same time, a conventional fuzzy energy-saving pumping strategy is implemented, which smoothly adjusts the frequency of the booster pump based on the water level and change rate in the inlet pool, prioritizing the reduction of the number of times the booster pump is started and stopped.

[0098] (2) Abnormal water inflow condition: When the predicted sequence contains (For example, if a sudden surge in water volume is caused by heavy rain), the system calculates in advance the required buffer volume for the inlet booster pump station. The calculation formula is as follows:

[0099]

[0100] in, The current moment; The variable is the integral variable, representing the instantaneous moment in the prediction time domain. ; To predict the duration of hydraulic impact, h; for The inflow rate of the pipeline network at any time, in m 3 / h; To increase the maximum allowable safe water delivery flow rate of the pump set, m 3 / h. System based The operating frequency of the booster pump is forcibly corrected at the current moment, and the liquid level in the inlet pool is lowered in advance to absorb the instantaneous hydraulic impact.

[0101] At the same time, the sludge level in the secondary sedimentation tank is introduced. As a hard constraint for safety: if If the mud level warning value is reached, a strong intervention strategy will be triggered, adjusting the target water delivery volume to equal... This prevents a large amount of sludge from being washed to the effluent end due to high flow rate, thereby protecting the system's microbial concentration.

[0102] Step S2: Reduced-order feedforward prediction and Lagrange adjoint state optimization solution. Discarding the complex and unsolvable ASM activated sludge model, a three-variable reduced-order water quality kinetic model including nitrate nitrogen, ammonia nitrogen, and dissolved oxygen is constructed. Define the state vector. Its nonlinear ordinary differential dynamic equations are constructed as follows:

[0103]

[0104]

[0105]

[0106] in, The values ​​represent the rates of change of nitrate nitrogen, ammonia nitrogen, and dissolved oxygen concentrations over time, in mg / (L·h). The dilution factor is the ratio of influent flow rate to the volume of the biological tank, h. -1 ; The influent ammonia nitrogen concentration is in mg / L. The yield of heterotrophic bacteria is dimensionless. Nitrogen content in biomass, dimensionless; and These represent the maximum specific rates of nitrification by autotrophic bacteria and denitrification by heterotrophic bacteria, respectively, d -1 ; This is a dimensionless parameter for aeration control. h is the oxygen transfer coefficient. -1 ; The dissolved oxygen saturation concentration is expressed in mg / L. Monod activation coefficient of nitrate nitrogen; Monod activation coefficient of ammonia nitrogen; Monod activation coefficient of dissolved oxygen; This represents the inhibition coefficient of dissolved oxygen.

[0107] Monod activation coefficients corresponding to nitrate nitrogen, ammonia nitrogen, and dissolved oxygen concentrations With the inhibition coefficient They are respectively:

[0108]

[0109]

[0110] in, The value is the half-saturation constant of the corresponding substrate, in mg / L; is the inhibition constant of the corresponding substrate, in mg / L.

[0111] Because microbial activity is significantly affected by temperature and sludge concentration during actual operation, the system uses the least squares method to identify historical operating data online, combined with water quality issues within the biological tank. and the concentration of suspended solids in the mixture Adaptive correction to the maximum specific reaction rate:

[0112]

[0113] in, and The constant coefficients are dimensionless and obtained through online parameter identification using the Gauss-Newton method based on historical operating data.

[0114] Definition and solution of constrained optimization problems: The optimization objective is to minimize a single control cycle. The total aeration energy consumption (characterized by integral under controlled conditions) is also subject to regulatory constraints that the weighted average effluent ammonia nitrogen level must not exceed the standard.

[0115]

[0116]

[0117] To solve this nonlinear optimization problem, an adjoint state vector is introduced. Construct the Hamiltonian function:

[0118]

[0119] The differential equation of the adjoint state vector is:

[0120]

[0121] Let the first The start time of the second aeration is The closing time is Based on the Pontryagin maximum principle, the gradient descent algorithm is used to iteratively optimize the switching time (assuming...). (Learning rate):

[0122]

[0123]

[0124] After millisecond-level high-speed iterative solutions, the system outputs the theoretical feedforward aeration time required for both the south and north pools in the future. .

[0125] Step S3: Multi-level parallel feedback compensation mechanism Due to the unpredictable disturbances of environmental noise and influent composition, the feedforward theoretical calculation needs to be combined with a multi-level water quality feedback correction mechanism.

[0126] Taking the south-side biological tank as an example, calculate the deviation between the actual effluent ammonia nitrogen and the set value. Discrete incremental PID controller is used to obtain the first-level feedback time compensation. :

[0127]

[0128] Multi-level safety hard constraint rule logic: Hard constraint to prevent over-aeration:

[0129] If the nitrate nitrogen index at the outlet of the biological pond And the ammonia nitrogen at the outlet of the biological pond This indicates that the carbon source has been completely depleted and the system is severely over-aerated. Forced intervention in the system is necessary. (That is, the final aeration time is reset to zero, and the machine is immediately shut down and enters a hypoxic state).

[0130] Mechanism to prevent denitrification collapse: If the outlet nitrate nitrogen And ammonia nitrogen This indicates that a severe shortage of carbon source in the system is hindering denitrification and reducing ammonia nitrogen treatment capacity. The controller prioritizes ensuring the target aeration time and triggers a linkage command to the underlying PLC and the phosphorus removal dosing / carbon source addition system to manually compensate for the addition of an external carbon source.

[0131] The final target aeration time for Nanchi is: .

[0132] Step S4: Multi-zone coordinated aeration and variable power execution

[0133] 1. Dynamic gas volume allocation coordination:

[0134] The operating control input duration of the main blower is determined by the maximum demand of the parallel zone:

[0135]

[0136] To address the uneven gas distribution across multiple chambers in the main pipeline, the controller dynamically issues commands to adjust the opening ratios of the electric regulating valves on the south and north intake branch pipes. The target opening ratio for the south chamber valve is as follows: The calculation is as follows:

[0137]

[0138] North Pool Valve Opening Ratio As implementation progresses, if the cumulative aeration time in Beichi reaches T... 北 The system automatically shuts off the electric valve on the north side, and the main blower supplies 100% of the air volume to the independent aeration of the south pool, eliminating local over-aeration.

[0139] 2. Variable power adaptive execution strategy:

[0140] Expanding traditional start-stop control, introducing Variable frequency speed control:

[0141] (1) Full power and full frequency ( ): The previous continuous non-aeration time exceeded the high load threshold, and the substrate accumulation was serious. Oxygen was supplied at the upper limit of the rated power.

[0142] (2) Half-power half-frequency The previous continuous aeration time was shorter than the low-load threshold, and the current ammonia nitrogen level has a safety margin of more than 30% from the threshold. Switching to half-frequency operation precisely matches the low-load oxygen consumption rate, significantly reducing motor power consumption.

[0143] (3) Equipment physical protection constraints: Apply minimum operating time for the blower min and shortest downtime The min limit forcibly filters out destructive high-frequency reversal commands caused by probe drift.

[0144] III. Evaluation Index System and Simulation Experiment Result Analysis

[0145] To fully verify the control performance of the present invention, real influent data from a wastewater treatment plant for 14 consecutive days (including typical sunny day fluctuations and sudden rainstorm impact conditions) were imported based on a wastewater treatment benchmark simulation platform.

[0146] The following evaluation indicators were constructed for quantitative analysis:

[0147] (1) Aeration Energy (AE): Reflects the level of electrical energy consumption of the blower over a period of time.

[0148]

[0149] (2) Error integral indicators: including integral absolute error (IAE), integral squared error (ISE) and maximum absolute error (MAXE), used to evaluate the controller's tracking accuracy of ammonia nitrogen setpoint.

[0150]

[0151]

[0152]

[0153] The comparative experiment selected three strategies: 1. Traditional industrial PID feedback control; 2. Constant power nonlinear model predictive control (NMPC, which only supports 0 or 1 start-stop); 3. Reduced-order model variable power multi-region control of the present invention (hereinafter referred to as the method of the present invention).

[0154] 1. Comparison of control performance under normal water intake conditions (sunny day)

[0155] Under sunny conditions, the influent load exhibits typical diurnal periodic fluctuations. The focus of this study is on the controller's ability to maintain denitrification steady-state and its energy-saving effect. The results are shown in Table 2.

[0156]

[0157] As shown in Table 2, traditional PID control suffers from severe lag due to the lack of a predictive mechanism, resulting in ammonia nitrogen concentration fluctuations as high as 18.4%. The method of this invention, through accurate feedforward prediction and dynamic valve airflow allocation, reduces the average effluent ammonia nitrogen level to 0.51 mg / L. More importantly, with the support of variable power (u=0.5), energy waste during off-peak periods is avoided, and the average daily aeration energy consumption is reduced by approximately 33% compared to PID control. Furthermore, the frequency of blower start-up and shutdown is significantly reduced, effectively protecting the equipment.

[0158] 2. Comparison of impact resistance performance under abnormal water inflow conditions (heavy rain)

[0159] On day 8, heavy rainfall data was introduced, and the inflow rate surged by 300% within 2 hours. The focus was on examining the shock resistance robustness of the plant-grid coordinated drainage mechanism and multi-level feedback. The results are shown in Table 3.

[0160]

[0161] As shown in the table above, when faced with the impact of torrential rain, PID and conventional NMPC systems, which only focus on the water quality feedback within the plant, experience significant sludge flushing from the biological treatment tanks during large-volume inflows (high risk of sludge loss). This leads to a loss of short-range nitrifying bacteria, causing ammonia nitrogen levels to spike to 6.8 mg / L, resulting in system shutdown for up to 14.5 hours. In contrast, the method of this invention, through plant-network coordination, pre-calculates and proactively empties a buffer volume of 4500 cubic meters (…). The combination of sludge level hard constraint and flow peak was suppressed, and the peak effluent ammonia nitrogen was successfully controlled at 1.9 mg / L (within the safe limit of 2.0), demonstrating extremely strong shock resistance.

[0162] 3. Controller Dynamic Tracking Performance Evaluation

[0163] To quantify the algorithm's ability to finely track water quality targets under both sunny and rainy conditions, the error integral index is compared as shown in Table 4.

[0164]

[0165] Table 4 presents the error integral data, which visually reflects the steady-state and transient performance of the controller. Regardless of whether the operating conditions are sunny or rainy, the IAE (Integrated Absolute Error) and ISE (Implement Large Error Penalty) of the method described in this invention are at their absolute lowest levels. Particularly under rainy conditions, the MAXE (Maximum Error) of the method described in this invention is only 2.815, far lower than the 8.850 of PID controllers. This is attributed to the rapid response capability of the online rolling optimization of the reduced-order model in this invention, and the efficient compensation of prediction errors by multi-level closed-loop feedback.

[0166] Figure 2 This diagram illustrates the daily intermittent aeration control effect of the parallel south and north biological tanks in a wastewater treatment plant after being controlled according to the method of this invention. Figure 2 (a) Corresponding to the South Biological Pond, Figure 2 (b) Corresponding to the North biological pool, as can be seen from the figure, there are differences in the aeration start and stop time nodes and cumulative aeration time between the South and North biological pools. However, both can maintain a stable dissolved oxygen environment and ensure that the ammonia nitrogen in the effluent meets the standards. This shows the effectiveness of the present invention in achieving independent and precise control under the architecture of one blower driving multiple pools by dynamically adjusting the opening ratio of the branch pipe valves.

[0167] Figure 3 The figure shows the effect of intermittent nighttime aeration at a wastewater treatment plant for a week. As can be seen from the figure, the present invention can automatically identify the low period of influent load at night, actively reduce the aeration frequency and switch to half power operation mode. While maintaining stable compliance of effluent ammonia nitrogen for 7 consecutive days, it significantly reduces aeration energy consumption, demonstrating the capability of the present invention in all-time adaptive energy-saving control.

[0168] Figure 4 The graph shows a comparison of the effects of automatic and manual control in a wastewater treatment plant. The red rectangle in the graph represents the automatic control part, and the rest represents the manual control part. As can be seen from the graph, the automatic control is based on the reduced-order model prediction and online optimization of the Lagrange adjoint state. It can predict changes in influent load in advance and actively adjust the aeration strategy. The average ammonia nitrogen in the effluent is as low as 0.51 mg / L, with a fluctuation range of only 1.2%. The manual control relies on human experience for lagging adjustment. When faced with load fluctuations, it often adopts a conservative strategy of "over-aeration to meet the standard". As a result, the average ammonia nitrogen in the effluent is as high as 1.35 mg / L, with a fluctuation range of 18.4%, and the daily aeration energy consumption is about 33% higher.

[0169] In summary, this invention, through the deep integration of a reduced-order water quality dynamics model, a plant-network linkage anti-impact architecture, multi-regional dynamic air volume allocation, and a variable power execution mechanism, completely solves the problems of computational bottlenecks, uneven distribution, and poor impact resistance in traditional intermittent aeration processes for wastewater treatment. While ensuring the steady-state achievement of nitrification water quality standards in complex AAO processes, it achieves extreme energy saving and consumption reduction, greatly promoting the intelligent evolution of wastewater treatment control technology.

[0170] Finally, it should be noted that the above embodiments of the present invention are merely illustrative examples and not intended to limit the implementation of the invention. Those skilled in the art can make other variations and modifications based on the above description. It is impossible to exhaustively list all possible implementations here. All obvious variations or modifications derived from the technical solutions of this invention are still within the scope of protection of this invention.

Claims

1. A multi-zone and multi-condition based intermittent aeration control method, characterized in that, Specifically, the following steps are included: S1: Real-time acquisition of multi-source heterogeneous data from the wastewater treatment plant's influent system, as well as water quality data from the biological reactor and effluent outlet; and interaction with the intelligent pipeline network management system to identify influent conditions, which are classified as normal or abnormal influent conditions, and the required reserved buffer volume of the influent tank is calculated under abnormal influent conditions. Increase the pumping volume and lower the inlet water level in advance; S2: Construct a three-variable water quality kinetic model for reducing nitrogen levels, including nitrate nitrogen, ammonia nitrogen, and dissolved oxygen. Input the influent water quality and quantity characteristics into the model, and use the Lagrange adjoint state method and gradient descent algorithm to iteratively optimize the aeration start and stop times, outputting the feedforward theoretical aeration time for each bioreactor. ; S3: Calculate the feedback time compensation value for each bioreactor through a multi-level feedback mechanism. Then, the aeration time for each biological reactor is obtained by adding it to the feedforward theoretical time. ; S4: Based on the aeration time of each bioreactor, select the maximum value as the operating control input of the main blower unit; at the same time, calculate the ratio of the aeration time of each bioreactor and adjust the opening of the electric regulating valve on the corresponding aeration branch pipe; combined with the load status, switch between zero power, half power and full power variable power states to achieve coordinated optimization control of intermittent aeration of each bioreactor.

2. The intermittent aeration control method based on multiple regions and multiple operating conditions according to claim 1, characterized in that, In step S1, under abnormal water inflow conditions, the sludge level in the secondary sedimentation tank is monitored in real time. When the sludge level in the secondary sedimentation tank reaches or exceeds the preset sludge level warning value, the water delivery rate of the booster pump is adjusted so that the water delivery rate of the booster pump is [value missing]. ,in It is to increase the target set water delivery volume of the booster pump. It is the real-time water inflow rate.

3. The intermittent aeration control method based on multiple regions and multiple operating conditions according to claim 1, characterized in that, In step S1, the system interacts with the intelligent pipeline network management system to obtain the pipeline network catchment flow sequence within the future prediction time domain. when At that time, the water intake condition is the normal water intake condition; when At that time, the water intake condition is an abnormal water intake condition; When the water intake is under abnormal conditions, the required buffer volume for the water intake booster pump station should be calculated using the following formula. : Where t is the current time; To predict the duration of hydraulic impact, h; for The inflow rate of the pipeline network at any time, in m 3 / h; To increase the maximum allowable safe water delivery flow rate of the pump set, m 3 / h.

4. The intermittent aeration control method based on multiple regions and multiple operating conditions according to claim 1, characterized in that, The state variable vector of the reduced-order water quality dynamics model The system dynamic equations are as follows: in, The values ​​represent the rates of change of nitrate nitrogen, ammonia nitrogen, and dissolved oxygen concentrations over time, in mg / (L·h). The dilution factor is the ratio of influent flow rate to the volume of the biological reactor, h. -1 ; The influent ammonia nitrogen concentration is in mg / L. The yield of heterotrophic bacteria is dimensionless. Nitrogen content in biomass, dimensionless; and These represent the maximum specific rates of nitrification by autotrophic bacteria and denitrification by heterotrophic bacteria, respectively, d -1 ; This is a dimensionless parameter for aeration control. h is the oxygen transfer coefficient. -1 ; Dissolved oxygen saturation concentration, mg / L; Monod activation coefficient of nitrate nitrogen; Monod activation coefficient of ammonia nitrogen; Monod activation coefficient of dissolved oxygen; This represents the inhibition coefficient of dissolved oxygen.

5. The intermittent aeration control method based on multiple regions and multiple operating conditions according to claim 4, characterized in that, and Based on the concentration of suspended solids in the mixture and the water temperature in the biological reactor Dynamic adaptive correction is performed, and the correction formula is as follows: in, , and The constant coefficients are dimensionless and obtained through online parameter identification using the Gauss-Newton method based on historical operating data.

6. The intermittent aeration control method based on multiple regions and multiple operating conditions according to claim 1, characterized in that, In step S2, the constrained optimization problem with minimizing total aeration energy consumption as the optimization objective is defined as: the objective function for minimizing total aeration energy consumption. The constraint is the weighted average ammonia nitrogen concentration in the effluent. Below the set threshold : Where T is the control cycle duration, in hours (h); Let be the ammonia nitrogen concentration in the bioreactor at time t, in mg / L; Let be the aeration control state parameter at time t, with a value range of [0, 1] and dimensionless.

7. The intermittent aeration control method based on multiple regions and multiple operating conditions according to claim 1, characterized in that, In step S3, the multi-level feedback mechanism specifically involves: performing primary feedback time compensation based on the ammonia nitrogen and nitrate nitrogen data of the effluent from each biological reactor; and performing secondary feedback time compensation based on the total nitrogen data of the effluent.

8. The intermittent aeration control method based on multiple regions and multiple operating conditions according to claim 7, characterized in that, The first-level feedback time compensation value is calculated using a discrete incremental PID control algorithm, and the calculation formula is as follows: in, Let be the feedback time compensation value for the k-th sampling period, min; , , These are the proportional coefficient, integral coefficient, and derivative coefficient of the feedback controller, respectively, and are dimensionless. The ammonia nitrogen concentration deviation of the current control step. , This represents the measured ammonia nitrogen value in the effluent from the biological reactor. Set the target value for ammonia nitrogen in mg / L; and These represent the ammonia nitrogen concentration deviations for the previous and two previous sampling periods, respectively, in mg / L.

9. The intermittent aeration control method based on multiple regions and multiple operating conditions according to claim 7, characterized in that, In step S3, if the nitrate nitrogen index at the outlet of the bioreactor reaches the first warning value and the ammonia nitrogen at the outlet of the bioreactor is lower than the set safety lower limit, the target aeration time of the corresponding bioreactor will be forcibly corrected to zero; if the nitrate nitrogen index at the outlet of the bioreactor reaches the first warning value and the ammonia nitrogen at the outlet of the bioreactor is higher than the set risk upper limit, the target aeration time will be guaranteed first, and an alarm for artificial carbon source supplementation will be triggered.

10. The intermittent aeration control method based on multiple regions and multiple operating conditions according to claim 1, characterized in that, In step S4, the control state parameters in the variable power state are... The range of values ​​is expanded to When the continuous non-aeration time exceeds the set high load threshold, or the feedforward-judged weighted average effluent ammonia nitrogen concentration approaches... And when the displacement control range fails to effectively lower the vehicle, switch to... Full-power aeration; when the previous continuous aeration time is less than the set low-load threshold, and the total aeration time of the system is sufficient, switch to full-power aeration; Half-power aeration is used to reduce energy consumption; at the same time, the shortest continuous running time and the shortest continuous shutdown time of the blower are set.

Images