Cascade model-based blocking prediction control method for effluent ammonia nitrogen in sewage treatment

Abstract

The invention discloses a cascade model-based sewage treatment effluent ammonia nitrogen blocking prediction control method, which only requires the installation of ammonia nitrogen instruments at the water inlet and outlet positions of a multi-stage AO biochemical reaction tank, and does not need the installation of the ammonia nitrogen instruments in the intermediate link, thereby saving the cost. The ammonia nitrogen set values of the outlets of the first stage and the middle stage are dynamically adjusted according to the inlet water ammonia nitrogen measured value, so that each stage of AO biochemical reaction tank can work in the effective control range, and the final outlet water ammonia nitrogen is ensured to meet the environmental protection requirement; aiming at the fact that no ammonia nitrogen instruments exist in effluent of the first stage and the middle stage, an extended state observer is adopted to estimate effluent ammonia nitrogen measured values of the first stage and the middle stage, and predictive control over AO of each stage is achieved; according to the method, the multi-stage AO characteristics of the AO sewage treatment process are strongly associated with modeling and control of effluent ammonia nitrogen, and multi-point water inlet and internal and external reflux are fully considered, so that the interpretability of a model and a control scheme is improved, understanding and field debugging are facilitated, and the implementation efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of sewage treatment, and in particular to a method for predicting and controlling ammonia nitrogen in sewage treatment effluent based on a cascade model. Background Art

[0002] In modern wastewater treatment processes, outlet ammonia nitrogen concentration is a key indicator of water quality. Ammonia nitrogen primarily originates from domestic sewage, industrial wastewater, and agricultural runoff. To reduce dosage, a multi-stage AO anoxic-aerobic process is often used to achieve biological nitrification and denitrification, thereby achieving biological denitrification. Because ammonia nitrogen concentration sensors are relatively expensive, they are often installed only at the inlet and outlet of a multi-stage AO system, with no sensors present in the intermediate aerobic and anaerobic tanks. In practice, the complex composition of wastewater, the large fluctuations in influent load, and the nonlinear dynamic characteristics of the biological reaction process greatly increase the control difficulty of multi-stage AO processes. Furthermore, the multiple inlet and return points, and the fluctuating inlet sewage flow rates of multi-stage AO tanks all pose challenges to controlling outlet ammonia nitrogen. Traditional PID control methods not only make it difficult to adjust the PID parameters, but also hinder dynamic optimization of the DO (Dissolved Oxygen) setpoints for the anaerobic and aerobic tanks of the multi-stage AO system. In addition, since it is difficult to conduct system identification experiments under actual sewage treatment conditions, data that can be used for system identification can only be found under normal operating conditions, which brings difficulties to the relationship between ammonia nitrogen concentration and dissolved oxygen concentration.

[0003] In actual operation, insufficient aeration results in excessively high effluent ammonia and nitrogen concentrations, failing to meet the required concentration. Excessive aeration, however, results in excessive dissolved oxygen concentrations. While these concentrations can meet the required concentration, they can easily lead to over-aeration, increasing blower energy consumption and impacting other indicators such as phosphorus removal. The effluent ammonia and nitrogen concentration is affected by numerous factors, making the complex and complex mechanism model difficult to control. Summary of the Invention

[0004] The purpose of the present invention is to overcome the shortcomings of the existing technology and propose a card-edge predictive control method for ammonia nitrogen in sewage treatment effluent based on a cascade model. A cascade model of dissolved oxygen concentration and effluent ammonia nitrogen concentration is established in a data-driven manner. It is helpful to dynamically allocate the ammonia nitrogen concentration target values ​​of each level when the number of ammonia nitrogen concentration sensors is small, and give the set values ​​of the DO control loop at each level, thereby realizing full-process ammonia nitrogen control and precise aeration, ensuring aeration quality and reducing fan energy consumption.

[0005] The purpose of the present invention is achieved through the following technical scheme: a card edge predictive control method for ammonia nitrogen in sewage treatment effluent based on a cascade model, wherein in a multi-stage AO biochemical reaction tank, an ammonia nitrogen meter is installed at the water inlet of the first-stage AO biochemical reaction tank and the water outlet of the last-stage AO biochemical reaction tank, the ammonia nitrogen measurement value of the inlet of the first-stage AO biochemical reaction tank is obtained and the effluent ammonia nitrogen value of the last-stage AO biochemical reaction tank is set, and at the same time, a dissolved oxygen DO measuring instrument and a dissolved oxygen DO concentration control loop are set in each stage of the AO biochemical reaction tank, a cascade model of dissolved oxygen DO concentration and effluent ammonia nitrogen concentration is established, and the effluent ammonia nitrogen concentration of the first-stage AO biochemical reaction tank and the intermediate-stage AO biochemical reaction tank is dynamically adjusted according to the expanded state observer, and the set value of the dissolved oxygen DO concentration control loop in each stage of the AO biochemical reaction tank is further calculated, thereby realizing full-process ammonia nitrogen control and precise aeration.

[0006] Furthermore, the method comprises the following steps:

[0007] S1. Establish a multi-stage AO ammonia nitrogen-DO concentration data-driven cascade static model to calculate the effluent ammonia nitrogen set values ​​of the first-stage AO biochemical reaction tank and the intermediate-stage AO biochemical reaction tank;

[0008] S2. Identify the parameters of each level model and establish a cascade model from the first-stage AO to the intermediate-stage AO and from the intermediate-stage AO to the final-stage AO. Based on the multi-point water inflow, internal reflux, and external reflux operation mechanism of the AO process, introduce the soft measurement value of graded equivalent ammonia nitrogen in each stage of the AO biochemical reaction pool;

[0009] S3. Establish an expanded state observer for each AO stage with dissolved oxygen DO as input and effluent ammonia nitrogen concentration as output, obtain estimated ammonia nitrogen concentrations at the first-stage AO outlet and the intermediate-stage AO outlet, and estimate the ammonia nitrogen concentrations of the intermediate-stage AO when there is no ammonia nitrogen sensor in the intermediate-stage AO. The ammonia nitrogen concentrations are used to control the ammonia nitrogen concentrations of the respective AO stages.

[0010] S4. Based on the set value of the ammonia nitrogen concentration of the AO outlet of each stage and the estimated ammonia nitrogen concentration at the inlet and outlet of each stage, a controller and its control algorithm are designed through a cascade model to update the set value of the dissolved oxygen (DO) concentration of each stage in real time. Each stage of AO then adjusts the aeration air volume according to the difference between the set value and the measured value of the DO concentration of this stage to achieve precise aeration.

[0011] Furthermore, the step S1 includes:

[0012] The multi-stage AO process adopts a three-stage AO process in its implementation. The sewage to be treated passes through the anaerobic tank, the first-stage anoxic tank, the first-stage aerobic tank, the facultative zone, the second-stage anoxic tank, the second-stage aerobic tank, the third-stage anoxic tank, the third-stage aerobic tank and the secondary sedimentation tank in sequence. The second-stage anoxic tank is the intermediate-stage anoxic tank, the second-stage aerobic tank is the intermediate-stage aerobic tank, the third-stage anoxic tank is the final-stage anoxic tank, and the third-stage aerobic tank is the final-stage aerobic tank, forming a three-stage AO treatment of the sewage; wherein, the anaerobic tank, the facultative zone and the third-stage anoxic tank are all provided with water inlets, an ammonia nitrogen meter is installed at the water inlet of the first-stage anoxic tank and the water outlet of the third-stage aerobic tank, a dissolved oxygen DO measuring meter is installed in each aerobic tank, and the blower controls the aeration of the aerobic tank through the aeration pipe and valve to form a dissolved oxygen DO concentration control loop;

[0013] The ammonia nitrogen-DO concentration data-driven cascade static model of multi-stage AO is as shown in formula (1):

[0014]

[0015] Among them, the static model refers to the relationship between ammonia nitrogen concentration and dissolved oxygen DO when the concentrations of ammonia nitrogen and dissolved oxygen DO are stable. y0 is the ammonia nitrogen measurement value at the first-stage AO inlet; y1 is the ammonia nitrogen measurement value at the first-stage AO outlet; y2 is the ammonia nitrogen measurement value at the second-stage AO outlet; y3 is the ammonia nitrogen measurement value at the third-stage AO outlet, that is, the outlet ammonia nitrogen target for the final ammonia nitrogen control of the entire AO process; δ1, δ2 and δ3 are the values ​​of ammonia nitrogen reduction in each stage of AO pool respectively; β1, β2 and β3 are the distribution ratio coefficients of the front-end sewage inflow in each stage of AO respectively, and β1+β2+β3=1;

[0016] The set value of ammonia nitrogen at the outlet of the first-stage AO biochemical reaction pool y 1SV As shown in formula (2a):

[0017]

[0018] The set value of ammonia nitrogen at the outlet of the second-stage AO biochemical reaction pool y 2SV As shown in formula (2b):

[0019]

[0020] Among them, y 3SV It is the target of ammonia nitrogen at the process outlet; is the amount of ammonia nitrogen in the three-stage drop, taking the same value, and can be deduced from (1) make w1>1 and w2>1 are the weight coefficients of the first two levels of ammonia nitrogen reduction.

[0021] Further, the step S2 includes:

[0022] The models of the AO biochemical reaction pools at each level are considered to have the same structure, but the parameters of the models at each level are different. The parameters are obtained by identifying the influent ammonia nitrogen and chemical oxygen demand (COD) when they are in the stable data segment. It is assumed that the influent and effluent flow rates of each AO biochemical reaction pool are the same, and the water in the AO biochemical reaction pool is fully mixed, that is, the concentration in the pool is equal to the effluent concentration. According to the multi-point water inlet, internal reflux and external reflux operation mechanism of the AO process, the soft measurement value of the graded equivalent ammonia nitrogen is introduced as the predicted value of y1. and the predicted value of y2

[0023] The ammonia nitrogen concentration change rate equations of the three AO biochemical reaction tanks are formulas (3a), (3b) and (3c), respectively:

[0024]

[0025]

[0026] Among them, T s is the sampling period, T s For 30 minutes; V1, V2, V3 represent the volume of water in each level of aerobic pool, which is calculated based on the liquid level and the geometric dimensions of the AO biochemical reaction pool. in Indicates the water inlet flow rate, Q out Indicates the effluent flow rate of the third aerobic tank; K NH is the ammonia nitrogen half-saturation constant, with an initial value of 1; K OA is the half-saturation constant of autotrophic bacteria for oxygen, with an initial value of 0.4, and iterative identification is performed based on the initial value; k1, k2, k3 are process gain coefficients, which need to be identified; d 10 d 20 and d 30 is the delay from multiple entry points to the first, second, and third stages, d 21 is the delay from the first stage water discharge to the second stage water discharge, d 32 It is the time delay from the second stage water outlet to the third stage water outlet. The time delay parameter is calculated based on the flow rate and the geometric dimensions of the AO biochemical reaction tank.

[0027] Input DO excitation signal to each aerobic pool step by step, measure the inlet ammonia nitrogen, aerobic pool dissolved oxygen DO and outlet ammonia nitrogen values ​​when the DO excitation signal is input to each level, so as to obtain the model parameters of DO and ammonia nitrogen at each level; input DO excitation signal to the aerobic pool step by step: DO amplitude changes up and down by 1 mg / L, the period T of the third aerobic pool is p1 The cycle of the second-stage aerobic pool and the first-stage aerobic pool is T p2The time period is 360 minutes. During parameter identification, when DO stimulation is performed on the third-stage aerobic pool, the DO of the second-stage aerobic pool remains unchanged. If the first-stage aerobic pool is far away from the third-stage aerobic pool, the DO of the first-stage aerobic pool may change during DO stimulation. During parameter identification, when DO stimulation is performed on the third-stage aerobic pool, the DO of the second-stage aerobic pool remains unchanged, while the DO of the first-stage aerobic pool may change. When DO stimulation is performed on the second-stage aerobic pool, the DO of the first-stage aerobic pool remains unchanged, while the DO of the third-stage aerobic pool is increased by 1 mg / L and remains unchanged. When DO stimulation is performed on the first-stage aerobic pool, the DO of the second and third-stage aerobic pools are increased by 0.5 mg / L to ensure that the output of ammonia nitrogen meets the target requirements during the identification process.

[0028] The parameters to be identified are The parameter model of the third-stage aerobic pool is formula (4),

[0029]

[0030] Where N3 is the number of samples in the third-level test, and the optimization index of the third-level aerobic pool is formula (5)

[0031]

[0032] The parameters in formula (5) are obtained through the optimization algorithm. Then, the ammonia nitrogen measurement data output by the third-stage aerobic pool is used to infer the estimated value of ammonia nitrogen output by the second-stage aerobic pool. After the third-level model parameter estimation is completed, the first-level model and the second-level model use the same K OA and K NH , the first-level model and the second-level model only estimate the single parameters k1 and k2. When the second-level model parameters are identified, the output ammonia nitrogen measurement data of the third-level aerobic pool is used to infer the estimated value of the output ammonia nitrogen of the second-level aerobic pool. As shown in formula (6):

[0033]

[0034] When performing parameter identification of the second-level model, the second-level measurement data needs to be moved forward as a whole. 32 T S Time, as shown in formula (7):

[0035]

[0036] And the objective function of the second level is as follows:

[0037]

[0038] Where N2 is the number of samples in the third level test. According to the structure of formula (11), the same identification method as the third level is used to identify the number of samples k before the current moment. 32 The sampling data of the sampling time is identified According to the same method as the second-stage model parameter identification, the output ammonia nitrogen of the first-stage aerobic pool is estimated by estimating the output ammonia nitrogen of the second-stage aerobic pool, and then the first-stage model parameters are obtained by identifying the single parameter.

[0039] According to the second-stage model, the estimated value of ammonia nitrogen at the outlet of the first-stage aerobic tank is obtained as shown in formula (9):

[0040]

[0041] From the sampling moment of the output of the third stage, we need to move forward (d 32 +d 21 )T S At this moment, the first-level model identification is formula (10):

[0042]

[0043] And the objective function of the first level is as follows:

[0044]

[0045] Further, the step S3 includes:

[0046] In real-time control, the ammonia nitrogen concentration of the intermediate AO biochemical reaction pool without an ammonia nitrogen sensor is estimated by the extended state observer; according to the measured value y0(kd 32 -d 21 ) and y3(k), and the observed value z3(k) of y3(k) is obtained by the extended state observer, and the observed value z1(kd of the effluent ammonia nitrogen of the first-stage aerobic pool is obtained. 32 -d 21 ), kd 32 The observed value of ammonia nitrogen in the effluent of the second-stage aerobic pool at the time z2(kd 32 ); On this basis, through the model recursion formula from kd 32 -d 21 Moment Get the k-time From kd 32 Moment At time k Finally based on and The real-time process control of the intermediate-stage AO biochemical reaction pool is realized; the extended state observer of the three-stage AO biochemical reaction pool is shown in formulas (12a), (12b) and (12c):

[0047]

[0048] Where l1, l2 and l3 are the gains of the first-level, second-level and third-level observers respectively, and their values ​​are selected to ensure the stable convergence of formulas (13a) and (13b);

[0049] The first-level soft sensor recursive formula (13a) is:

[0050]

[0051] Where i = -d 32 -d 21 ,-d 32 -d 21 +1,-d 32 -d 21 +2,...,-1;

[0052] The second-level soft sensor recursive formula (13b) is:

[0053]

[0054] Where i = -d 32 ,-d 32 +1,sd 32 +2,...,-1.

[0055] Further, the step S4 includes:

[0056] The controller and its control algorithm are designed through the predictive control of the cascade model, and the objective function and constraint conditions are set for each level of aerobic pool; the cascade model predictive control law x is obtained by optimization and solution. MPC (k); take the control law x MPC The first element in (k) represents the dissolved oxygen set value at that moment, and each T s Update the primary dissolved oxygen set value. At this time, the dissolved oxygen concentration of each aerobic tank is controlled by adjusting the air volume of the blower and the valve opening to track the given dissolved oxygen set value, thereby achieving energy-saving control of the effluent ammonia nitrogen concentration in the constrained sewage treatment process;

[0057] The objective function of the first-stage aerobic pool is formula (14a):

[0058]

[0059] The objective function of the second-stage aerobic pool is formula (14b):

[0060]

[0061] The objective function of the third-stage aerobic pool is formula (14c):

[0062]

[0063] Among them, N p is the prediction time domain, N c is the control time domain, the first-level recursive prediction model is formula (13a), and the second-level recursive prediction model is formula (13b). is the predicted value of ammonia nitrogen in the third-stage effluent, and the recursive prediction model is formula (15):

[0064]

[0065] Constrain the range of dissolved oxygen variation and the value of effluent ammonia nitrogen, as shown in formula (16):

[0066]

[0067]

[0068] Among them, DO min Is the lower limit of DO setting value, DO max is the upper limit of DO setting value, NH 4max It is the upper limit of the set value of effluent ammonia nitrogen.

[0069] A non-transitory computer-readable medium storing instructions, characterized in that when the instructions are executed by a processor, the steps of the card-edge predictive control method for ammonia nitrogen in sewage treatment effluent based on the above-mentioned cascade model are executed.

[0070] A computing device includes a processor and a memory for storing a program executable by the processor, characterized in that when the processor executes the program stored in the memory, it implements the above-mentioned card-edge predictive control method for ammonia nitrogen in sewage treatment effluent based on the cascade model.

[0071] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0072] 1. The present invention strongly links the multi-stage AO characteristics of the AO sewage treatment process with the modeling and control of effluent ammonia nitrogen, and fully considers multiple water inflow points and internal and external recirculation, which increases the interpretability of the model and control scheme, facilitates engineers' understanding and on-site debugging, and improves implementation efficiency.

[0073] 2. The set value of ammonia nitrogen in the effluent of the intermediate stage adopts a dynamic adjustment method, which fully considers the fluctuation of ammonia nitrogen in the first stage input, realizes the indirect feedforward control function of ammonia nitrogen control, and improves the response time of the ammonia nitrogen controller. The dynamic adjustment of the set value of ammonia nitrogen in the effluent of the first two stages also gives full play to the biological denitrification performance of each stage and reduces the input of chemicals.

[0074] 3. The present invention adopts an expanded state observer to realize the estimation of ammonia nitrogen in the first two stages of effluent, and uses a soft measurement recursive method to obtain real-time estimated values ​​of the first and second stages, thereby reducing expensive ammonia nitrogen measuring instruments and lowering the cost of project implementation. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 This is a schematic diagram of the multi-stage AO sewage treatment process of the present invention.

[0076] Figure 2 This is the logic block diagram of ammonia nitrogen control for multi-stage AO wastewater treatment.

[0077] Figure 3 Schematic diagram of parameter identification of the cascade model of multi-stage AO wastewater treatment.

[0078] Figure 4 This is the DO excitation waveform in the parameter identification of the cascade model.

[0079] Figure 5 Flowchart of the method for estimating ammonia nitrogen concentration in the intermediate tank of multi-stage AO wastewater treatment.

[0080] Figure 6 Block diagram of the DO setpoint model predictive control for a multi-stage AO wastewater treatment plant. DETAILED DESCRIPTION

[0081] The present invention will be further described below with reference to specific embodiments.

[0082] Example 1

[0083] The cascade model-based predictive control method for ammonia nitrogen in wastewater treatment effluent provided in this embodiment requires only the installation of ammonia nitrogen meters at the inlet and outlet of a multi-stage AO wastewater treatment process. Dynamically adjust the ammonia nitrogen setpoints for the first and intermediate stages based on the influent ammonia nitrogen value, ensuring that each stage of the reaction tank operates within an effective control range and meets effluent standards. To address the lack of instrumentation in the intermediate stages, ammonia nitrogen control for the entire cascade wastewater treatment system is achieved through the use of process modeling, observer estimation of intermediate tank variables, and predictive control of each AO tank.

[0084] In the actual sewage treatment process, in order to take into account the removal efficiency of carbon, nitrogen and phosphorus while controlling the construction cost, a three-stage AO process is usually used. Figure 1As shown in the figure, taking three-stage AO as an example, a schematic diagram of the three-stage AO sewage treatment process is given. The entire process includes the sewage to be treated entering the anaerobic tank, the first-stage anoxic tank, the first-stage aerobic tank, the facultative zone, the second-stage anoxic tank (i.e., the intermediate-stage anoxic tank), the second-stage aerobic tank (i.e., the intermediate-stage aerobic tank), the third-stage anoxic tank (i.e., the final-stage anoxic tank, which is the last stage of the multi-stage system, and the third stage when there are only three stages), the third-stage aerobic tank (i.e., the final-stage aerobic tank), and finally entering the secondary sedimentation tank to become AO-treated sewage. After subsequent disinfection and other processes, it meets the discharge standards and is discharged into the river. This multi-stage AO process features multiple water inflow points, with the anaerobic tank, facultative zone, and third-stage anoxic tank each receiving approximately 40% of the water flow, the facultative zone approximately 30%, and the third-stage anoxic tank approximately 30%. Sludge water from the secondary sedimentation tank is externally refluxed to the anaerobic tank, while the second-stage aerobic tank and the third-stage aerobic tank have internal refluxes to the facultative zone. The multi-point water inflow, external reflux, and internal reflux utilize fixed flow rates. In the actual process, ammonia and nitrogen meters are installed at the inlet of the first-stage anoxic tank and at the outlet of the third-stage aerobic tank. A dissolved oxygen (DO) meter is installed in each aerobic tank. A blower controls aeration in the aerobic tanks through aeration pipes and valves, thereby controlling DO concentration.

[0085] Figure 2 This is the logic block diagram of ammonia nitrogen control for multi-stage AO wastewater treatment. Each of the three stages has an AO control system. The AO control system is a cascade control system. The outer loop is the ammonia nitrogen control loop, and the inner loop is the DO control loop. The DO control loop and the AO process are drawn in a block diagram, and the valve adjustment is realized in the DO control loop. Figure 2 The three cascade control systems correspond to Figure 1 The three-stage AO sewage treatment process and cascade control scheme are easy for on-site engineers to understand and operate, overcoming the problem of poor interpretability of using only a single control system for the entire process.

[0086] To achieve Figure 2 For the three-stage AO ammonia nitrogen control, the ammonia nitrogen set value of each stage must be given first. The ammonia nitrogen set value of each stage is determined by Figure 2 The algorithm module of the three-level outlet ammonia nitrogen setting value is implemented. 0PV is the ammonia nitrogen measurement value at the first stage inlet; 1PV and y 1SV is the measured value and set value of ammonia nitrogen at the first stage AO outlet; 2PV and y 2SV is the measured value and set value of ammonia nitrogen at the second stage AO outlet; 3PV and y 3SV is the measured value and set value of ammonia nitrogen at the outlet of the third-stage AO, where y 3SVIt is the outlet ammonia nitrogen target of the final ammonia nitrogen control of the entire OA process; let the values ​​of the three-stage ammonia nitrogen reduction be δ1, δ2 and δ3 respectively; the distribution ratio coefficients of the front-end sewage inflow in the three stages are β1, β2 and β3, and β1+β2+β3=1. In a stable sewage treatment process, the flow rate of the outflow and inflow of each pool at the same time is approximately equal, and the static relationship of the cascade ammonia nitrogen concentration can be approximately obtained, which is equivalent to the end of the dynamic control process. The ammonia nitrogen measurement value output by each stage is approximately the set value.

[0087]

[0088] From (1),

[0089] y 0PV -y 3SV =β1δ1+(β1+β2)δ2+δ3. (2)

[0090] If the amount of ammonia nitrogen dropped in the three levels is the same, that is, Then there is

[0091]

[0092] Since the third stage AO biochemical reaction pool has the fastest response time to the output ammonia nitrogen, in order to ensure that the third stage is always in the rapid response stage and the outlet ammonia nitrogen concentration is below the target value, the ammonia nitrogen concentration drop of the first two stages is greater than the average ammonia nitrogen drop concentration.

[0093]

[0094] Among them, w1>1 and w2>1 are the weight coefficients of the first two levels of ammonia nitrogen reduction. Take w1=1.2 and w2=1.1, then

[0095]

[0096] and

[0097]

[0098] Among them, y 0PV is the measured value of the inlet ammonia nitrogen concentration, y 3SV The process outlet ammonia nitrogen target is two known process variables. Figure 2 Given the three-stage outlet ammonia nitrogen set value algorithm module, the first-stage and second-stage outlet ammonia nitrogen set values ​​are calculated according to formula (5) and formula (6).

[0099] Figure 2In the multi-stage AO wastewater treatment system shown, ammonia nitrogen control is not installed at the first and second stage outlets, so it is estimated from ammonia nitrogen meters at the process inlet and outlet. To estimate ammonia nitrogen in the effluent from the first and second stage AO biochemical reactors, an interpretable data-driven cascade model that meets process requirements is required.

[0100] Assume that y0, y1, y2 and y3 are the ammonia nitrogen measurement values ​​at the first-stage water inlet, first-stage water outlet, second-stage water outlet and third-stage water outlet, respectively, which are numerically equal to y 0PV ,y 1PV ,y 2PV and y 3PV , used to simplify the formula. x 1PV and x 1SV , x 2PV and x 2SV , x 3PV and x 3SV These are the measured and set values ​​of dissolved oxygen (DO) in the first, second, and third aerobic tanks, respectively. The measured values ​​of dissolved oxygen (DO) are abbreviated as x1, x2, and x3. Since the three-stage sewage treatment uses the same process, the three-stage AO process uses the same model, but with different model parameters. The ammonia nitrogen concentration change rate equations of the three AO biochemical reaction tanks are:

[0101]

[0102] Among them, T s is the sampling period, in the ammonia nitrogen control system, T s Take 30 minutes; V1, V2, V3 represent the volume of water in each level of aerobic pool, which can be calculated based on the liquid level and the geometric dimensions of the AO pool. in Indicates the water inlet flow rate, Q out represents the effluent flow rate of the third aerobic tank; K NH is the ammonia nitrogen half-saturation constant, the initial value is 1, K OA is the half-saturation constant of autotrophic bacteria for oxygen, with an initial value of 0.4, and iterative identification is performed based on the initial value; k1, k2, and k3 are process gain coefficients, which need to be identified. 10 d 20 and d 30 is the delay from multiple entry points to the first, second, and third stages, d 21 is the delay from the first stage water discharge to the second stage water discharge, d 32 It is the delay from the second stage water outlet to the third stage water outlet. The delay parameter can be calculated based on the flow rate and the geometric dimensions of the AO biochemical reaction tank.

[0103] In equations (7a), (7b) and (7c), k1, k2 and k3 need to be identified. Figure 3 The system parameter identification process shown in Figure 4 It is the DO excitation signal used in identification. Figure 4 The amplitude of the excitation signal DO varies by 1 mg / L. The period of the third level is T P3 The period of the second and first stages is T P2 and T P1 The time is 360 minutes. When identifying parameters, when the third level is performing DO stimulation, the second level keeps the DO unchanged, and the first level DO can be changed. When the second level is performing DO stimulation, the first level keeps the DO unchanged, and the third level DO is increased by 1mg / L and remains unchanged. When performing the first level DO stimulation, the second and third levels increase the DO by 0.5mg / L to ensure that the output ammonia nitrogen meets the target requirements during the identification process. Figure 4 During the three-level excitation process, the flow rate, liquid level, influent ammonia nitrogen, aerobic tank DO and effluent ammonia nitrogen of the process flow are measured. Figure 3 , the parameters of models (7a), (7b) and (7c) are identified in three stages.

[0104] Figure 3 This is the flow chart of the system parameter identification subroutine. In the initialization of the main program, the initial identification state IdentState = 0, that is, no system parameter identification is performed; when system parameter identification is required, let IdentState = 3, and start a system parameter identification. When IdentState is not 0, the main program periodically calls Figure 3 System parameter identification subroutine.

[0105] Figure 3In Step 31, the value of IdentState is determined. If IdentState is 3, the program proceeds to Step 311 for the third-level system parameter identification. If IdentState is not 3, the program proceeds to Step 32. In Step 32, the program determines whether IdentState is 2. If so, the program proceeds to Step 321 for the second-level system parameter identification. If IdentState is not 2, the program proceeds to Step 33. In Step 33, the program determines whether IdentState is 1. If so, the program proceeds to Step 331 for the first-level system parameter identification. If IdentState is not 1, the program returns from the identification subroutine to the main program. After each level of identification is completed, the program also returns to the main program. In Step 311, the third-level system parameter identification is performed. DO3 performs three complete square wave changes, completes data acquisition and alignment, and identifies the model time parameters and process gain. After completion, IdentState = 2, and the program prepares for the second-level parameter identification. In Step 321, the second-level system parameter identification is performed. DO2 undergoes a complete square wave variation, completes data collection and alignment, and identifies the model's process gain. Once completed, IdentState is set to 1, preparing for the next first-level parameter identification. In Step 331, the first-level system parameter identification is completed. DO1 undergoes a complete square wave variation, completes data collection and alignment, and identifies the process gain. Once completed, IdentState is set to 0, concluding the parameter identification for the three-level AO pool.

[0106] For the third-stage aerobic pool, when the influent ammonia nitrogen is relatively stable and the DO concentration of the second-stage aerobic pool is basically unchanged, the corresponding y2 is basically unchanged and y0 can be measured. Therefore, by changing the DO concentration of the third-stage aerobic pool and measuring the value of the effluent ammonia nitrogen, the parameter K of the model (7c) can be identified. NH , K OA and k3.

[0107]

[0108] Identify the parameter K of model (7c) NH , K OA and k3 adopts the objective function as

[0109]

[0110] Where N3 is the number of samples in the third level test. According to the structural characteristics of formula (8), Relative to k3, K OA and K NH is a monotonic function, and k3, K OA and K NHThe sensitivity of K is gradually reduced, and these parameters are identified in an iterative manner. OA and K NH The initial value is and Through the sample data and the objective function (24), the first estimate of k3 can be obtained by the least squares method: The second step iteration, and According to the same sample data, K can be obtained by the least square method with the objective function as formula (24): OA The estimated value is The third step iteration, and By using the same sample data and the least square method with the objective function as formula (9), we can get K NH The estimated value is The third step iteration, and Using the same sample data and the least square method with the objective function as formula (24), the estimated value of k3 can be obtained as So far and is the third-level model parameter. This method makes full use of the process initial parameters and The robustness of the identification parameters is improved, the identification algorithm requires less computing power, and can run in real time.

[0111] After the third-level model parameter estimation is completed, the first-level and second-level models use the same K OA and K NH , then the first and second levels only need to estimate single parameters k1 and k2, reducing the requirements for the variation range of sample data. After the first and second levels are partially mixed by the third level, the variation range is reduced. When conducting the second level identification test, the estimated value of ammonia nitrogen output by the third level is inferred from the ammonia nitrogen measurement data output by the second level.

[0112]

[0113] When performing the second-level parameter identification, the second-level measurement data needs to be moved forward as a whole. 32 T S Time, there

[0114]

[0115] and the second-level objective function

[0116]

[0117] Where N2 is the number of samples in the third-level test. According to the structure of formula (11), the same identification method as the third level is used to identify the number of samples before the current moment k. 32 The sampling data can be identified According to the same method as the second-level identification, the output ammonia nitrogen estimation of the first level is obtained by the output ammonia nitrogen estimation of the second level, and then the first-level model parameters are obtained by the identification of single parameters.

[0118] According to the second-stage model, the estimated value of ammonia nitrogen at the outlet of the first-stage aerobic pool is obtained.

[0119]

[0120] From the sampling moment of the output of the third stage, we need to move forward (d 32 +d 21 )T S At this moment, the first-level identification model is

[0121]

[0122] And the first-level objective function

[0123]

[0124] After completing the three-stage model identification, in real-time online control, the outlet ammonia nitrogen values ​​of the first and second stages can be estimated through the observer through the three-stage model and the input ammonia nitrogen, the output ammonia nitrogen of the third stage and the dissolved oxygen DO value of each stage.

[0125] Reference method for estimating ammonia nitrogen concentration in the intermediate pool Figure 5 , the specific steps are as follows:

[0126] Step 51, calculate the observed value of ammonia nitrogen in the effluent of the third-level aerobic pool z3(k). Let the current time be k. For the third-level aerobic pool, since the effluent ammonia nitrogen can be directly measured, the observed value of ammonia nitrogen in the effluent of the third aerobic pool at time k is z3(k) for the measured value at time k-1.

[0127]

[0128] Among them, z2 is the observation value of the observer of ammonia nitrogen at the second-stage outlet, z3 is the predicted value of ammonia nitrogen at the third-stage outlet, and l3 is the gain of the error compensation term of the third-stage observer.

[0129] Step 52, calculate the kd 32 -d 21 The observed value of ammonia nitrogen in the effluent of the first-stage aerobic pool at the sampling time is z1(kd 32 -d 21 ) and kd32 The observed value of ammonia nitrogen in the effluent of the second-stage aerobic pool at the time z2(kd 32 ).

[0130] Using the measured value y3(k) of the effluent ammonia nitrogen from the third-stage AO biochemical reaction pool and the observed value z3(k) of the effluent ammonia nitrogen from the third-stage AO biochemical reaction pool, the observed kd 32 -d 21 The observed value of ammonia nitrogen in the effluent of the first-stage aerobic pool at time z1(kd 32 -d 21 )for

[0131]

[0132] Among them, z1 is the observation value of the observer of ammonia nitrogen at the first-stage outlet, and l1 is the gain of the error compensation term of the second-stage observer.

[0133] For the second-stage aerobic pool, observe kd 32 The observed value of ammonia nitrogen in the effluent of the second aerobic pool at time z2(kd 32 )for

[0134]

[0135] Where l1 is the gain of the error compensation term of the first-level observer.

[0136] Step 53: Forecast the ammonia nitrogen value of the effluent from the first-stage aerobic pool at the current moment Using the form of model (7a), kd 32 -d 21 Substitute the relevant variables at time k and deduce kd 32 -d 21 +1 to time k Forecast value of ammonia nitrogen in the effluent of the second-stage aerobic pool at the current moment Using the form of model (7b), kd 32 Substitute the relevant variables at time k and deduce k s -d 32 +1 to time k Used as the initial value for model predictive control.

[0137] For the current moment Using the form of model (7a), kd 32 -d 21 Substitute the relevant variables at time k and deduce kd 32 -d 21 +1 to time k Used as the initial value of model predictive control, the recursive formula is

[0138]

[0139] Where i = -d 32 -d 21 ,-d 32 -d 21 +1,-d 32 -d 21 +2,...,-1.

[0140] Step 54, for the current moment Using the form of model (22b), kd 32 Substitute the relevant variables at time k and deduce k s -d 32 +1 to time k Used as the initial value of model predictive control, the recursive formula is

[0141]

[0142] Where i = -d 32 ,-d 32 +1,sd 32 +2,...,-1.

[0143] Step 55, perform model predictive control at time k+1.

[0144] Model predictive control block diagram reference for DO setpoints in multi-stage AO wastewater treatment Figure 6 As shown, first, objective functions and constraints are set for each level of aerobic pools; taking the objective function, the goal is to minimize the sum of the dissolved oxygen value and the set value of ammonia nitrogen at each level outlet and the target ammonia nitrogen.

[0145] For the first-stage aerobic pool, set the objective function

[0146]

[0147] Among them, N p is the prediction time domain, N c is the control time domain, and the recursive prediction model of soft measurement is formula (17).

[0148] For the second-stage aerobic pool, the prediction model is:

[0149]

[0150] The recursive prediction model of soft measurement is formula (18).

[0151] For the third-level aerobic pool, set the objective function

[0152]

[0153] in, is the predicted value of ammonia nitrogen in the third-stage effluent, and the recursive prediction model is

[0154]

[0155] To constrain the range of dissolved oxygen and the value of effluent ammonia nitrogen,

[0156]

[0157] Among them, DO min Is the lower limit of DO setting value, DO max is the upper limit of DO setting value, NH 4max It is the upper limit of the set value of effluent ammonia nitrogen. min and DO max The purpose is to make the solver output conform to the constraints of the actual working conditions, setting NH 4max The purpose is to enable the solver to solve the DO set value that can make the DO concentration smaller and meet the effluent index according to different ammonia nitrogen effluent index requirements. From the performance index objective function of formula (19) and formula (21), it can be seen that the DO set value obtained is to make the ammonia nitrogen less than the set value but as close to the set value as possible, to achieve lower limit card edge control, which can both meet the control target and minimize energy consumption. After rolling optimization, the DO set values ​​x at each level are obtained MPC (k) = [x 1SV (k),x 2SV (k),x 3SV (k)] T ; Use the optimization algorithm fmincon function to solve the constrained optimization problem composed of the dissolved oxygen value, objective function and constraint conditions at each level at each moment, and obtain the model predictive control law:

[0158]

[0159] Here, arg min represents the decision variable that is used to calculate the minimum value of the objective function.

[0160] Figure 6 After rolling optimization, take the control law x MPC The first element in (k) represents the dissolved oxygen set value at that moment, and each T s Update the dissolved oxygen set value once. At this time, the dissolved oxygen concentration in the aerobic tank is controlled to track the given dissolved oxygen set value by adjusting the air volume of the blower and the valve opening, thereby achieving energy-saving control of the effluent ammonia nitrogen concentration in the constrained urban sewage treatment process.

[0161] Example 2

[0162] This embodiment discloses a non-transitory computer-readable medium storing instructions. When the instructions are executed by a processor, the steps of the card-edge predictive control method for ammonia nitrogen in sewage treatment effluent based on a cascade model described in Example 1 are performed.

[0163] The non-transitory computer-readable medium in this embodiment can be a disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), a USB flash drive, a mobile hard disk, or other media.

[0164] Example 3

[0165] This embodiment discloses a computing device, including a processor and a memory for storing a program executable by the processor. When the processor executes the program stored in the memory, the card-edge predictive control method for ammonia nitrogen in sewage treatment effluent based on the cascade model described in Example 1 is implemented.

[0166] The computing device described in this embodiment may be a desktop computer, a laptop computer, a smart phone, a PDA handheld terminal, a tablet computer, a programmable logic controller (PLC), or other terminal devices with a processor function.

[0167] The embodiments described above are only preferred embodiments of the present invention and are not intended to limit the scope of implementation of the present invention. Therefore, any changes made based on the shape and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A predictive control method for ammonia nitrogen in sewage treatment effluent based on a cascade model, characterized by: The method comprises the following steps: in a multi-stage AO biochemical reaction tank, ammonia nitrogen meters are installed at the water inlet of the first-stage AO biochemical reaction tank and the water outlet of the final-stage AO biochemical reaction tank, ammonia nitrogen measurement value of the inlet water of the first-stage AO biochemical reaction tank is obtained, and an effluent ammonia nitrogen value of the final-stage AO biochemical reaction tank is set; at the same time, a dissolved oxygen (DO) measurement meter and a dissolved oxygen (DO) concentration control loop are set in each stage of the AO biochemical reaction tank, a cascade model of dissolved oxygen (DO) concentration and effluent ammonia nitrogen concentration is established, the effluent ammonia nitrogen concentrations of the first-stage AO biochemical reaction tank and the intermediate-stage AO biochemical reaction tank are dynamically adjusted according to an expanded state observer, and the set values ​​of the dissolved oxygen (DO) concentration control loops in each stage of the AO biochemical reaction tank are further calculated, thereby realizing full-process ammonia nitrogen control and precise aeration.

2. The card edge prediction control method for ammonia nitrogen in sewage treatment effluent based on the cascade model according to claim 1 is characterized in that: The following steps are involved: S1. Establish a multi-stage AO ammonia nitrogen-DO concentration data-driven cascade static model to calculate the effluent ammonia nitrogen set values ​​of the first-stage AO biochemical reaction tank and the intermediate-stage AO biochemical reaction tank; S2. Identify the parameters of each level model and establish a cascade model from the first-stage AO to the intermediate-stage AO and from the intermediate-stage AO to the final-stage AO. Based on the multi-point water inflow, internal reflux, and external reflux operation mechanism of the AO process, introduce the soft measurement value of graded equivalent ammonia nitrogen in each stage of the AO biochemical reaction pool; S3. Establish an expanded state observer for each AO stage with dissolved oxygen DO as input and effluent ammonia nitrogen concentration as output, obtain estimated ammonia nitrogen concentrations at the first-stage AO outlet and the intermediate-stage AO outlet, and estimate the ammonia nitrogen concentrations of the intermediate-stage AO when there is no ammonia nitrogen sensor in the intermediate-stage AO. The ammonia nitrogen concentrations are used to control the ammonia nitrogen concentrations of the respective AO stages. S4. Based on the set value of the ammonia nitrogen concentration of the AO outlet of each stage and the estimated ammonia nitrogen concentration at the inlet and outlet of each stage, a controller and its control algorithm are designed through a cascade model to update the set value of the dissolved oxygen (DO) concentration of each stage in real time. Each stage of AO then adjusts the aeration air volume according to the difference between the set value and the measured value of the DO concentration of this stage to achieve precise aeration.

3. The card edge prediction control method for ammonia nitrogen in sewage treatment effluent based on the cascade model according to claim 2 is characterized in that: The step S1 comprises: The multi-stage AO process adopts a three-stage AO process in its implementation. The sewage to be treated passes through the anaerobic tank, the first-stage anoxic tank, the first-stage aerobic tank, the facultative zone, the second-stage anoxic tank, the second-stage aerobic tank, the third-stage anoxic tank, the third-stage aerobic tank and the secondary sedimentation tank in sequence. The second-stage anoxic tank is the intermediate-stage anoxic tank, the second-stage aerobic tank is the intermediate-stage aerobic tank, the third-stage anoxic tank is the final-stage anoxic tank, and the third-stage aerobic tank is the final-stage aerobic tank, forming a three-stage AO treatment of the sewage; wherein, the anaerobic tank, the facultative zone and the third-stage anoxic tank are all provided with water inlets, an ammonia nitrogen meter is installed at the water inlet of the first-stage anoxic tank and the water outlet of the third-stage aerobic tank, a dissolved oxygen DO measuring meter is installed in each aerobic tank, and the blower controls the aeration of the aerobic tank through the aeration pipe and valve to form a dissolved oxygen DO concentration control loop; The ammonia nitrogen-DO concentration data-driven cascade static model of multi-stage AO is as shown in formula (1): Among them, the static model refers to the relationship between ammonia nitrogen concentration and dissolved oxygen DO when the concentrations of ammonia nitrogen and dissolved oxygen DO are stable. y0 is the ammonia nitrogen measurement value at the first-stage AO inlet; y1 is the ammonia nitrogen measurement value at the first-stage AO outlet; y2 is the ammonia nitrogen measurement value at the second-stage AO outlet; y3 is the ammonia nitrogen measurement value at the third-stage AO outlet, that is, the outlet ammonia nitrogen target for the final ammonia nitrogen control of the entire AO process; δ1, δ2 and δ3 are the values ​​of ammonia nitrogen reduction in each stage of AO pool respectively; β1, β2 and β3 are the distribution ratio coefficients of the front-end sewage inflow in each stage of AO respectively, and β1+β2+β3=1; The set value of ammonia nitrogen at the outlet of the first-stage AO biochemical reaction pool y 1SV As shown in formula (2a): The set value of ammonia nitrogen at the outlet of the second-stage AO biochemical reaction pool y 2SV As shown in formula (2b): Among them, y 3SV It is the target of ammonia nitrogen export from AO process; is the amount of ammonia nitrogen in the three-stage drop, taking the same value, and can be deduced from (1) make w1>1 and w2>1 are the weight coefficients of the first two levels of ammonia nitrogen reduction.

4. The card edge predictive control method for ammonia nitrogen in sewage treatment effluent based on the cascade model according to claim 3 is characterized in that: The step S2 comprises: The models of the AO biochemical reaction pools at each level are considered to have the same structure, but the parameters of the models at each level are different. The parameters are obtained by identifying the influent ammonia nitrogen and chemical oxygen demand (COD) when they are in the stable data segment. It is assumed that the influent and effluent flow rates of each AO biochemical reaction pool are the same, and the water in the AO biochemical reaction pool is fully mixed, that is, the concentration in the pool is equal to the effluent concentration. According to the multi-point water inlet, internal reflux and external reflux operation mechanism of the AO process, the soft measurement value of the graded equivalent ammonia nitrogen is introduced as the predicted value of y1. and the predicted value of y2 The ammonia nitrogen concentration change rate equations of the three AO biochemical reaction tanks are formulas (3a), (3b) and (3c), respectively: Among them, T s is the sampling period, T s For 30 minutes; V1, V2, V3 represent the volume of water in each level of aerobic pool, which is calculated based on the liquid level and the geometric dimensions of the AO biochemical reaction pool. in Indicates the water inlet flow rate, Q out Indicates the effluent flow rate of the third aerobic tank; K NH is the ammonia nitrogen half-saturation constant, with an initial value of 1; K OA is the half-saturation constant of autotrophic bacteria for oxygen, with an initial value of 0.4, and iterative identification is performed based on the initial value; k1, k2, k3 are process gain coefficients, which need to be identified; d 10 d 20 and d 30 is the delay from multiple entry points to the first, second, and third stages, d 21 is the delay from the first stage water discharge to the second stage water discharge, d 32 is the time delay from the second stage effluent to the third stage effluent, and the time delay parameter is calculated based on the flow rate and the geometric dimensions of the AO biochemical reaction tank; x1, x2 and x3 are the measured values ​​of dissolved oxygen DO in the first, second and third stage aerobic tanks respectively; Input DO excitation signal to each aerobic pool step by step, measure the inlet ammonia nitrogen, aerobic pool dissolved oxygen DO and outlet ammonia nitrogen values ​​when the DO excitation signal is input to each level, so as to obtain the model parameters of DO and ammonia nitrogen at each level; input DO excitation signal to the aerobic pool step by step: DO amplitude changes up and down by 1 mg / L, the period T of the third aerobic pool is p1 The cycle of the second-stage aerobic pool and the first-stage aerobic pool is T p2 The time period is 360 minutes. During parameter identification, when DO stimulation is performed on the third-stage aerobic pool, the DO of the second-stage aerobic pool remains unchanged. If the first-stage aerobic pool is far away from the third-stage aerobic pool, the DO of the first-stage aerobic pool may change during DO stimulation. During parameter identification, when DO stimulation is performed on the third-stage aerobic pool, the DO of the second-stage aerobic pool remains unchanged, while the DO of the first-stage aerobic pool may change. When DO stimulation is performed on the second-stage aerobic pool, the DO of the first-stage aerobic pool remains unchanged, while the DO of the third-stage aerobic pool is increased by 1 mg / L and remains unchanged. When DO stimulation is performed on the first-stage aerobic pool, the DO of the second and third-stage aerobic pools are increased by 0.5 mg / L to ensure that the output of ammonia nitrogen meets the target requirements during the identification process. The parameters to be identified are The parameter model of the third-stage aerobic pool is formula (4), Where N3 is the number of samples in the third-level test, and the optimization index of the third-level aerobic pool is formula (5) The parameters in formula (5) are obtained through the optimization algorithm. Then, the ammonia nitrogen measurement data output by the third-stage aerobic pool is used to infer the estimated value of ammonia nitrogen output by the second-stage aerobic pool. After the third-level model parameter estimation is completed, the first-level model and the second-level model use the same K OA and K NH , the first-level model and the second-level model only estimate the single parameters k1 and k2. When the second-level model parameters are identified, the output ammonia nitrogen measurement data of the third-level aerobic pool is used to infer the estimated value of the output ammonia nitrogen of the second-level aerobic pool. As shown in formula (6): When performing parameter identification of the second-level model, the second-level measurement data needs to be moved forward as a whole. 32 T S Time, as shown in formula (7): And the objective function of the second level is as follows: Where N2 is the number of samples in the third level test. According to the structure of formula (11), the same identification method as the third level is used to identify the number of samples k before the current moment. 32 The sampling data of the sampling time is identified According to the same method as the second-stage model parameter identification, the output ammonia nitrogen of the first-stage aerobic pool is estimated by estimating the output ammonia nitrogen of the second-stage aerobic pool, and then the first-stage model parameters are obtained by identifying the single parameter. According to the second-stage model, the estimated value of ammonia nitrogen at the outlet of the first-stage aerobic tank is obtained as shown in formula (9): From the sampling moment of the output of the third stage, we need to move forward (d 32 +d 21 )T S At this moment, the first-level model identification is formula (10): And the objective function of the first level is as follows:

5. The card edge prediction control method for ammonia nitrogen in sewage treatment effluent based on the cascade model according to claim 4 is characterized in that: The step S3 comprises: In real-time control, the ammonia nitrogen concentration of the intermediate AO biochemical reaction pool without an ammonia nitrogen sensor is estimated by the extended state observer; according to the measured value y0(kd 32 -d 21 ) and y3(k), and the observed value z3(k) of y3(k) is obtained by the extended state observer, and the observed value z1(kd of the effluent ammonia nitrogen of the first-stage aerobic pool is obtained. 32 -d 21 ), kd 32 The observed value of ammonia nitrogen in the effluent of the second-stage aerobic pool at the time z2(kd 32 ); On this basis, through the model recursion formula from kd 32 -d 21 Moment Get the k-time From kd 32 Moment At time k Finally based on and The real-time process control of the intermediate-stage AO biochemical reaction pool is realized; the extended state observer of the three-stage AO biochemical reaction pool is shown in formulas (12a), (12b) and (12c): Where l1, l2 and l3 are the gains of the first-level, second-level and third-level observers respectively, and their values ​​are selected to ensure the stable convergence of formulas (13a) and (13b); The first-level soft sensor recursive formula (13a) is: where \(i = -d\) 32 -d 21 , -d 32 -d 21 +1, -d 32 -d 21 +2, ..., -1; The second-level soft sensor recursive formula (13b) is: Where i = -d 32 ,-d 32 +1,sd 32 +2,...,-1.

6. The card edge predictive control method for ammonia nitrogen in sewage treatment effluent based on the cascade model according to claim 5 is characterized in that: The step S4 comprises: The controller and its control algorithm are designed through the predictive control of the cascade model, and the objective function and constraint conditions are set for each level of aerobic pool; the cascade model predictive control law x is obtained by optimization and solution. MPC (k); take the control law x MPC The first element in (k) represents the dissolved oxygen set value at that moment, and each T s Update the primary dissolved oxygen set value. At this time, the dissolved oxygen concentration of each aerobic tank is controlled by adjusting the air volume of the blower and the valve opening to track the given dissolved oxygen set value, thereby achieving energy-saving control of the effluent ammonia nitrogen concentration in the constrained sewage treatment process; The objective function of the first-stage aerobic pool is formula (14a): The objective function of the second-stage aerobic pool is formula (14b): The objective function of the third-stage aerobic pool is formula (14c): Among them, N p is the prediction time domain, N c is the control time domain, the first-level recursive prediction model is formula (13a), and the second-level recursive prediction model is formula (13b). is the predicted value of ammonia nitrogen in the third-stage effluent, and the recursive prediction model is formula (15): Constrain the range of dissolved oxygen variation and the value of effluent ammonia nitrogen, as shown in formula (16): Among them, DO min Is the lower limit of DO setting value, DO max is the upper limit of DO setting value, NH 4max It is the upper limit of the set value of effluent ammonia nitrogen.

7. A non-transitory computer-readable medium storing instructions, characterized in that: When the instruction is executed by the processor, the steps of the card-edge predictive control method of ammonia nitrogen in sewage treatment effluent based on the cascade model according to any one of claims 1 to 6 are executed.

8. A computing device comprising a processor and a memory for storing a program executable by the processor, characterized in that: When the processor executes the program stored in the memory, it implements the card-edge predictive control method for ammonia nitrogen in sewage treatment effluent based on the cascade model as described in any one of claims 1-6.

Images