An intelligent dosing method based on multi-point precise dosing distribution box and gradient dosing algorithm

Abstract

The application provides an intelligent dosing method based on a multi-point dosing precision distribution tank and a gradient dosing algorithm, and belongs to the technical field of sewage treatment. The method applies a matched intelligent dosing optimization system, first collects real-time measured values of ammonia nitrogen, nitrate nitrogen concentration and inflow flow rate, sets initial flow distribution ratios of each dosing point through the multi-point dosing precision distribution tank, constructs a dosing dose prediction model based on the gradient dosing algorithm, solves model output dosing dose prediction values D', finally performs dynamic regulation and control operation according to the prediction values D' and process unit effluent treatment effect feedback, generates a regulation and control signal and transmits the signal to a PLC, and performs linkage regulation and control on the distribution tank electric water stop clamp and the dosing pump. The application solves the problems of uneven distribution, reagent waste, high hardware dependency and difficult algorithm landing of the conventional dosing system through mechanical distribution and algorithm regulation and control cooperation, can significantly reduce reagent consumption and improve effluent water quality stability.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment technology, and in particular to an intelligent dosing method based on a multi-point precise dosing distribution box and a gradient dosing algorithm. Background Technology

[0002] Chemical consumption costs in wastewater treatment typically account for 25%-40% of total operating costs, with carbon source and phosphorus removal chemicals representing a significant portion. As a core component of the nitrogen and phosphorus removal process in wastewater treatment, the dosing system provides the necessary carbon source and reaction conditions for microbial metabolism, playing a crucial role in removing nitrogen and phosphorus pollutants from water. Insufficient dosing will lead to poor nitrogen and phosphorus removal efficiency and effluent exceeding standards; excessive dosing not only wastes chemicals and increases operating costs but may also disrupt the microbial community balance within the biological treatment tank, affecting the efficiency of subsequent treatment processes.

[0003] In existing wastewater treatment dosing technologies, multi-point dosing systems often employ a one-pump-one-point configuration, resulting in high equipment procurement and maintenance costs. Manual dosing devices suffer from uneven flow distribution and the inability to achieve automatic gradient adjustment. Existing intelligent dosing methods rely heavily on complex instrument configurations, requiring simultaneous collection of multiple water quality parameters such as DO, ORP, MLSS, and COD, leading to high hardware investment and complex operation and maintenance. Existing dosing control algorithms are mostly complex ASM mechanism models or big data machine learning models, which are difficult to model, require high on-site debugging, and are difficult to implement in small and medium-sized wastewater treatment plants. Furthermore, existing control methods can only adjust the total dosage of the dosing pumps, failing to achieve coordinated closed-loop control of the total dosing amount and the distribution ratio at multiple points. This results in low reagent utilization efficiency, poor adaptability to different process units, and an inability to adapt to dynamic fluctuations in influent water quality and flow rate. Summary of the Invention

[0004] The purpose of this invention is to overcome the above-mentioned defects in the prior art and provide an intelligent dosing method based on a multi-point precise dosing distribution box and a gradient dosing algorithm. Through the deep synergy of mechanical precise distribution and gradient algorithm control, the method achieves precise distribution and efficient dosing of the agent, and systematically solves the core problems of uneven distribution, agent waste, high hardware dependence, difficulty in algorithm implementation, and poor process adaptability in conventional dosing systems.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A smart dosing method based on a multi-point precise dosing distribution box and a gradient dosing algorithm is applied to a supporting smart dosing optimization system. The system includes a multi-point precise dosing distribution box, dosing point components adapted to multiple process units, and a smart control module. The smart control module includes a PLC, a water quality monitoring unit connected to the PLC, an algorithm processing unit, and an execution unit. The execution unit is connected to the electric stop clamp of the multi-point precise dosing distribution box and the dosing pump to realize automatic linkage control of flow distribution ratio and total dosing amount.

[0006] The core steps of the method include: firstly, collecting real-time measured values ​​of influent ammonia nitrogen concentration, influent nitrate nitrogen concentration, and influent flow rate; secondly, setting the initial flow distribution ratio of each dosing point based on the number of dosing points and reagent requirements of the wastewater treatment process using a multi-point dosing precision distribution box; thirdly, constructing a dosing dosage prediction model based on a gradient dosing algorithm, and completing model training and parameter calibration; fourthly, solving the constructed dosing dosage prediction model to output the dosing dosage prediction value D'; and finally, performing dynamic control calculations based on the dosing dosage prediction value D' and feedback data on the effluent treatment effect of the process unit, generating control signals and transmitting them to a PLC, which then controls the electric stop clamp and dosing pump of the multi-point dosing precision distribution box in a coordinated manner.

[0007] Furthermore, the dosing points cover the dosing points corresponding to the anaerobic tank, anoxic tank, and denitrification filter. The multi-point dosing precision distribution box, through the combination of water distribution tanks and water-stop components, realizes the flexible setting and dynamic adjustment of the flow distribution ratio of each dosing point.

[0008] Furthermore, the construction of the drug dosage prediction model consists of three core steps: the first step is to select the model input and output variables, divide the training set and the test set to form the sample dataset of the gradient dosing algorithm model; the second step is to perform normalization preprocessing on the training set data of the sample dataset to eliminate the influence of data units; the third step is to determine the core parameters of the gradient dosing algorithm to complete the construction of the drug dosage prediction model. The formula for normalization preprocessing is: D = (Dmax - Dmin) × (C_measured - Cmin) / (Cmax - Cmin); In the formula, D is the dosage, in L / h; Dmax is the historical maximum dosage, and Dmin is the historical minimum dosage; C is the real-time measured value of influent ammonia nitrogen concentration or influent nitrate nitrogen concentration, in mg / L; Cmax is the historical maximum value of the corresponding water quality parameter, and Cmin is the historical minimum value of the corresponding water quality parameter.

[0009] Furthermore, when constructing the sample dataset, the influent ammonia nitrogen concentration, influent nitrate nitrogen concentration, influent flow rate and corresponding dosing data for 6 consecutive months were selected as input variables, and the optimized dosing values ​​for the subsequent 3 months of the corresponding period were selected as output variables, and so on to form a complete sample dataset; the sample data of the first 12 months were divided into the training set, and the sample data of the last 6 months were divided into the test set.

[0010] Furthermore, the normalization preprocessing uses a dedicated calculation formula to standardize the input and output data; the core formula of the gradient dosing algorithm calculates the dosage adjustment value: △D=K×Q×(C 当前 -C 控制 The core parameters include the carbon-to-nitrogen ratio K, the dosage adjustment coefficient, Q (influent flow rate), and C. 当前 To measure nitrate concentration, C 控制 The target values ​​for nitrate and nitrogen control are defined, with the initial value of the carbon-nitrogen ratio coefficient K ranging from 3 to 4. The dosage adjustment coefficients include an increment coefficient of 1.0 and a reduction coefficient of 0.7.

[0011] Furthermore, during model solving, the dataset is set as Ci, Qi, Di, i = 1, 2, 3…n, where Di is the expected value of the dosage, Ci is the input vector composed of influent ammonia nitrogen concentration and influent nitrate nitrogen concentration, and Qi is the input vector composed of influent flow rate. The dosage adjustment value is calculated through the core formula of the gradient dosing algorithm, and the nitrate nitrogen control target value C is set simultaneously according to the preset rules. 控制 A standardized program was developed in the MATLAB 2024a environment. After training the model using the training set data, the test set data and the real-time collected input data were input into the prediction model to calculate the predicted drug dosage value D'.

[0012] Furthermore, during dynamic control, the electric water-stop clamps of the multi-point precision dosing distribution box are adjusted first to optimize the flow distribution ratio of each dosing point. If the distribution ratio adjustment reaches its limit and still cannot meet the effluent water quality requirements, the output flow of the dosing pump is then adjusted, and the flow distribution ratio of each dosing point is recalculated and set simultaneously. The dynamic correction rules and dosing protection mechanism of the carbon-nitrogen ratio coefficient K are also matched simultaneously to ensure the stable operation of the system.

[0013] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention achieves precise gradient distribution and intelligent linkage control of a single pump to multiple process unit dosing points through deep collaboration between a multi-point dosing precision distribution box and a gradient dosing algorithm. It replaces the traditional one-pump-one-point dosing mode, significantly reduces equipment investment and operation and maintenance costs, and fundamentally solves the core defects of conventional dosing systems such as uneven flow distribution and poor process adaptability.

[0014] 2. This invention only requires the collection of three core parameters—influent ammonia nitrogen concentration, influent nitrate nitrogen concentration, and influent flow rate—to achieve precise dosing control. It breaks through the strong dependence of existing intelligent dosing systems on multiple complex online monitoring instruments, reduces hardware procurement and maintenance costs, and is particularly suitable for the upgrading and transformation of small and medium-sized sewage treatment plants, making it highly scalable.

[0015] 3. The gradient dosing algorithm prediction model constructed in this invention has clear physical meaning, few core parameters, and is convenient for on-site debugging. It does not require complex mechanism modeling and massive data training, and solves the industry pain points of existing intelligent dosing algorithms, such as difficulty in modeling, implementation, and poor generalization. It can be quickly adapted to wastewater treatment scenarios of different scales and processes.

[0016] 4. This invention pioneered a priority control logic that first optimizes the flow distribution ratio at the dosing point and then adjusts the total dosage of the dosing pump. Combined with dynamic correction rules for the carbon-nitrogen ratio and a dosing protection mechanism, it forms a complete closed-loop control system. This system can maximize the utilization efficiency of the reagents in each process unit. While ensuring that the effluent quality meets the standards, it can reduce the consumption of reagents such as carbon sources by 15%-30%, significantly reducing the operating costs of wastewater treatment plants. At the same time, it avoids the impact of excessive reagent dosing on the microbial community of the biological system, and significantly improves the stability of process operation. Attached Figure Description

[0017] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a schematic diagram of the control system of the intelligent dosing optimization method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the precise drug dosing dispensing box structure according to an embodiment of the present invention; Figure 3 This is a graph showing the fitting results between the predicted and actual drug dosage values ​​in an embodiment of the present invention. Figure 4 This is a control logic diagram of the carbon source dosing system in an embodiment of the present invention.

[0018] In the diagram, the components are: 1. Multi-point precise dispensing box; 2. Electric water-stop clamp; 3. Dosing pump; 4. PLC; 5. Dosing point assembly; 6. Inlet pipe; 7. Water-stop plate; 8. Diversion pipe; 9. Water distribution tank; and 10. Outlet pipe. Detailed Implementation

[0019] The specific embodiments of the present invention will be further described in detail with reference to the accompanying drawings.

[0020] This embodiment applies to an AAO process wastewater treatment plant with a treatment capacity of 20,000 m³ / d, equipped with a three-stage treatment unit consisting of an anaerobic tank, an anoxic tank, and a denitrification deep-bed filter. Sodium acetate is used as the carbon source for dosing to enhance the nitrogen removal effect of the biological system. The intelligent dosing optimization system used in this embodiment includes, for example... Figure 2 As shown, the multi-point precision distribution box 1 is made of 304 stainless steel. An inlet pipe 6 is installed at the top, and several outlet water-stop plates 7 are installed on the inlet pipe 6. The water-stop plates 7 are made of corrosion-resistant PVC material, with a height adjustment range of 0-5cm. Ten horizontally level, equally spaced grooves are installed downstream of the water-stop plates 7, with the groove width adapted to the designed inlet flow rate. A diversion pipe 8 corresponding to the grooves is installed at the bottom. The diversion pipe 8 is equipped with an electric water-stop clamp 2, which is a normally closed, corrosion-resistant electric clamp valve with a response time ≤2s and a control accuracy of ±2%. By controlling the number and opening degree of the water-stop clamps, the different flow distribution ratios of the diversion pipe 8 can be achieved. For example, opening two water-stop clamps corresponding to two water distribution tanks 9 achieves a 1:1 distribution; opening six water-stop clamps corresponding to four water distribution tanks 9 (each tank connected to two diversion pipes 8) achieves a 2:1:2:1 distribution. The system includes a recessed water distribution tank 9 at the bottom of the diversion pipe 8, with four tanks 9 connected to an outlet pipe 10 at the bottom of each tank 9; a variable frequency diaphragm metering pump 3 with a flow rate adjustment range of 0-5000 L / h and a control accuracy of ±1%; a water quality monitoring unit including an online ammonia nitrogen analyzer, an online nitrate nitrogen analyzer, and an electromagnetic flowmeter. The online ammonia nitrogen analyzer uses a salicylic acid spectrophotometric method with a measurement range of 0-50 mg / L and an accuracy of ±2%FS. The online nitrate nitrogen analyzer uses an ultraviolet spectrophotometric method with a measurement range of 0-50 mg / L and an accuracy of ±2%FS. The electromagnetic flowmeter has a measurement range of 0-5000 m³ / h and an accuracy of ±0.5%FS. The PLC4 uses a Siemens S7-1200 series controller with a touch screen for easy switching between manual and automatic control modes.

[0021] This embodiment sets up 4 dosing points: the anaerobic tank dosing point, the anoxic tank front dosing point, the anoxic tank rear dosing point, and the denitrification filter inlet dosing point. The dosing components 5 at all dosing points are made of 304 stainless steel corrosion-resistant material. The depth of the dosing pipe at the anaerobic tank dosing point is 2 / 3 of the tank water depth. The anoxic tank dosing point has 4 anti-clogging nozzles evenly distributed along the length of the tank. The denitrification filter dosing point is set at 50cm from the filter inlet. All dosing components are equipped with a detachable cleaning structure, and the nozzle orifice diameter is 2mm to match the carbon source agent.

[0022] The specific implementation process of this embodiment is as follows, in conjunction with the appendix. Figure 1Explanation: First, real-time measured values ​​of influent ammonia nitrogen concentration and influent nitrate nitrogen concentration are collected using online ammonia nitrogen analyzers and online nitrate nitrogen analyzers. Real-time measured values ​​of influent flow rate are collected using electromagnetic flowmeters. All collected data are transmitted in real time to PLC4 and the algorithm processing unit. Based on the process denitrification requirements and hydraulic characteristics of the four dosing points, the initial flow distribution ratio of each dosing point is set to 1:4:3:2 using the water-stopping insert 7 and electric water-stopping clamp 2 of the multi-point dosing precision distribution box 1. This corresponds to a 10% allocation ratio for the anaerobic tank dosing point, a 40% allocation ratio for the front section of the anoxic tank dosing point, a 30% allocation ratio for the rear section of the anoxic tank dosing point, and a 20% allocation ratio for the denitrification filter dosing point.

[0023] Subsequently, a dosage prediction model based on the gradient dosing algorithm was constructed. The influent ammonia nitrogen concentration, nitrate nitrogen concentration, influent flow rate, and corresponding dosage data of the wastewater treatment plant for six consecutive months were used as input variables. The optimized dosage values ​​for the following three months were used as output variables, and so on, forming a complete sample dataset. The sample data from the first 12 months were divided into a training set, and the sample data from the last 6 months were divided into a test set. Normalization preprocessing was performed on the input and output data of the training set, using the corresponding normalization formula to complete data standardization, eliminating the influence of different parameters' dimensions and improving the model's prediction accuracy. The core parameters of the gradient dosing algorithm were determined: the initial carbon-to-nitrogen ratio coefficient K was set to 3.5, and the dosage adjustment coefficients were set to an increment coefficient of 1.0 and a reduction coefficient of 0.7, thus completing the construction of the dosage prediction model.

[0024] The constructed dosing prediction model is solved, with the dataset defined as Ci, Qi, Di, i = 1, 2, 3…n, where Di is the expected dosing dose, Ci is the input vector composed of influent ammonia nitrogen concentration and influent nitrate nitrogen concentration, and Qi is the input vector composed of influent flow rate. The dosing adjustment value is calculated using the core formula of the gradient dosing algorithm, and the nitrate nitrogen control target value C is set simultaneously according to preset rules. 控制 A standardized program was developed in the MATLAB 2024a environment. After training the model using the training set data, the test set data and the real-time collected input data were input into the prediction model to calculate the predicted dosage value D'. In this embodiment, the goodness of fit R² of the model prediction result is ≥0.95, and the prediction accuracy fully meets the requirements of on-site control. The fitting results are shown in the attached figure. Figure 3 As shown.

[0025] After the model solution is completed, dynamic control calculations and execution are performed. Three user-defined variables are established: the predicted dosage D', the total nitrogen concentration in the effluent of the process unit, and the total phosphorus concentration in the effluent. Dynamic control calculations are performed based on the predicted dosage D', effluent feedback data, and the correction rules of the carbon-nitrogen ratio coefficient K. The deviation signal is used to generate control quantities according to the gradient dosing logic and transmitted to PLC4. During the control execution, the opening of the electric water stop clamp 2 of the multi-point dosing precision distribution box 1 is adjusted first to optimize the flow distribution ratio of each dosing point. In this embodiment, during operation, the opening of the electric water stop clamp 2 at the front of the anoxic tank is prioritized from the initial... The initial flow rate was adjusted from 30% to 45%. The opening of the electric water stop clamp 2 in the anoxic tank was adjusted from 25% to 35%. The electric water stop clamp 2 in the anaerobic tank and denitrification filter remained at its initial opening. After one week of continuous operation, the total nitrogen concentration in the effluent stabilized at around 6.8 mg / L, meeting the Class A discharge standard requirements. There was no need to adjust the flow rate of the dosing pump 3. The final stable flow rate distribution ratio was 1:4.5:3.5:1. If the effluent water quality requirements could not be met even after adjusting the distribution ratio to the limit, the output flow rate of the dosing pump 3 was adjusted again, and the flow rate distribution ratio of each dosing point was recalculated and set simultaneously.

[0026] During operation, a dynamic correction rule for the carbon-to-nitrogen ratio coefficient K and a dosing protection mechanism are implemented simultaneously. The total nitrogen concentration in the effluent is monitored in real time. When the total nitrogen concentration is >7.5 mg / L, the K value increases by 0.2; when the total nitrogen concentration is <5 mg / L, the K value decreases by 0.1. The correction range of the K value is always maintained between 3 and 4 to avoid system fluctuations caused by excessive parameter correction. A dual dosing protection mechanism is also set: when the predicted dosage D' is below 300 L / h, it is forcibly set to 0; when the influent nitrate nitrogen concentration is below 0.6 mg / L, the dosing is stopped to avoid cost waste caused by ineffective dosing. The carbon source dosing system control is as follows: Figure 4 As shown.

[0027] This embodiment was run continuously for 3 months, and the operating data was compared with that of the plant's original conventional dosing system. The comparison results are shown in Table 1. The conventional dosing system adopts a one-pump-one-point configuration, manual fixed ratio dosing, and PID single-loop control mode.

[0028] Table 1 Comparison of the operational effects of the method of the present invention and conventional dosing methods

[0029] The working principle of this invention is as follows: Through the mechanical structure design of the multi-point dosing precision distribution box 1, a single pump can accurately distribute the gradient flow rate to multiple dosing points, replacing the traditional one-pump-one-point configuration mode. This solves the problems of uneven reagent distribution and high equipment costs from a hardware perspective. By collecting only three core parameters of influent ammonia nitrogen, nitrate nitrogen, and flow rate, the system's dependence on complex monitoring instruments is greatly reduced, thus lowering hardware and maintenance costs. The constructed gradient dosing algorithm prediction model has clear physical meaning, few parameters, and is easy to debug. It does not require complex modeling and massive data training, and can be quickly implemented in the field. By first optimizing the dosing point distribution ratio and then adjusting the priority control logic of the total dosage, the reaction efficiency of the reagent in each process unit is maximized first, and then the total dosage is adjusted. Combined with the parameter dynamic correction mechanism based on effluent water quality feedback, a complete closed-loop control is formed. Under the premise of ensuring stable effluent water quality compliance, the reagent consumption is minimized, while avoiding the impact of excessive reagent dosing on the biological system, thus improving the system's operational stability.

[0030] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A smart dosing method based on a multi-point precise dosing box and a gradient dosing algorithm, characterized in that, The method is applied to a supporting intelligent dosing optimization system. The intelligent dosing optimization system includes a multi-point precise dosing distribution box, dosing point components adaptable to multiple process units, and an intelligent control module. The intelligent control module includes a PLC, a water quality monitoring unit connected to the PLC, an algorithm processing unit, and an execution unit. The execution unit is connected to the electric water-stop clamp of the multi-point precise dosing distribution box and the dosing pump. The method includes the following steps: Step 1: Collect real-time measured values ​​of influent ammonia nitrogen concentration, influent nitrate nitrogen concentration, and influent flow rate. Based on the number of dosing points and reagent requirements of the wastewater treatment process, set the initial flow distribution ratio of each dosing point through a multi-point dosing precision distribution box. Step 2: Construct a drug dosage prediction model based on the gradient dosing algorithm; Step 3: Solve the dosing dosage prediction model constructed in step S2 and output the predicted dosing dosage value D'. Step 4: Based on the predicted dosage D' and feedback data on the effluent treatment effect of the process unit, dynamic control calculations are performed to generate control signals and transmit them to the PLC. The PLC then controls the electric stop clamps and dosing pumps of the multi-point precision dosing distribution box in a coordinated manner.

2. The intelligent dosing method based on a multi-point precise dosing box and a gradient dosing algorithm as described in claim 1, characterized in that, In step S1, the dosing points include dosing points corresponding to the anaerobic tank, anoxic tank, and denitrification filter. The multi-point dosing precision distribution box is adjusted by combining the water distribution tank and the water-stopping component to realize the setting and adjustment of the flow distribution ratio of each dosing point.

3. The intelligent dosing method based on a multi-point precise dosing box and gradient dosing algorithm as described in claim 1, characterized in that, In step S2, the specific steps for constructing the drug dosage prediction model include: Step S21: Select the model input and output variables, divide the training set and test set to form the sample dataset of the gradient application algorithm model; Step S22: Normalize the training set data of the sample dataset. Step S23: Determine the core parameters of the gradient dosing algorithm and complete the construction of the drug dosage prediction model.

4. The intelligent dosing method based on a multi-point precise dosing box and a gradient dosing algorithm as described in claim 3, characterized in that, In step S21, the influent ammonia nitrogen concentration, influent nitrate nitrogen concentration, influent flow rate and corresponding dosing data for 6 consecutive months are selected as input variables, and the optimized dosing values ​​for the following 3 months of the corresponding period are selected as output variables, and so on to form a sample dataset; the sample data of the first 12 months are divided into a training set, and the sample data of the last 6 months are divided into a test set.

5. The intelligent dosing method based on a multi-point precise dosing box and a gradient dosing algorithm as described in claim 3, characterized in that, In step S22, the calculation formula for normalization preprocessing is: D=(Dmax-Dmin)×(C 实测 -Cmin) / (Cmax-Cmin); In the formula, D is the dosage, in L / h; Dmax is the historical maximum dosage, and Dmin is the historical minimum dosage; C 实测 The influent ammonia nitrogen concentration or nitrate nitrogen concentration is the real-time measured value, in mg / L; Cmax is the historical maximum value of the corresponding water quality parameter, and Cmin is the historical minimum value of the corresponding water quality parameter.

6. The intelligent dosing method based on a multi-point precise dosing box and a gradient dosing algorithm as described in claim 3, characterized in that, In step S23, the core parameters of the gradient dosing algorithm include the carbon-nitrogen ratio coefficient K and the dosing amount adjustment coefficient. The initial value range of the carbon-nitrogen ratio coefficient K is 3-4, and the dosing amount adjustment coefficient includes an increment coefficient of 1.0 and a decrement coefficient of 0.

7.

7. The intelligent dosing method based on a multi-point precise dosing box and a gradient dosing algorithm as described in claim 1, characterized in that, In step S3, the specific method for solving the model is as follows: Set the dataset as (Ci, Qi, Di), i = 1, 2, 3…n, where Di is the expected value of the dosage, Ci is the input vector composed of influent ammonia nitrogen concentration and influent nitrate nitrogen concentration, and Qi is the input vector composed of influent flow rate. The dosage adjustment value is calculated using the core formula of the gradient dosing algorithm. The core formula is as follows: △D=K×Q×(C 当前 -C 控制 ); In the formula, ΔD is the dosage adjustment value, K is the carbon-nitrogen ratio coefficient, Q is the influent flow rate in m³ / h, and C... 当前 The influent nitrate nitrogen concentration is the real-time measured value, in mg / L, C 控制 The preset target value for nitrate nitrogen control is expressed in mg / L. A standardized program was developed in the MATLAB 2024a environment. After training the model using the training set data, the test set data and the real-time collected input data were input into the prediction model to calculate the predicted drug dosage value D'.

8. The intelligent dosing method based on a multi-point precise dosing box and a gradient dosing algorithm as described in claim 7, characterized in that, The target value for nitrate control C 控制 The setting rule is: when the influent ammonia nitrogen concentration is ≤25mg / L, C 控制 Set at 3.0 mg / L; when 25 mg / L < influent ammonia nitrogen concentration ≤ 28 mg / L, C 控制 Set at 2.5 mg / L; when the influent ammonia nitrogen concentration is >28 mg / L and the influent flow rate is >2600 m³ / h, C 控制 When C is set to 1.8 mg / L and the influent flow rate is ≤2600 m³ / h, 控制 Set to 2.3 mg / L.

9. The intelligent dosing method based on a multi-point precise dosing box and a gradient dosing algorithm as described in claim 1, characterized in that, In step S4, the specific method for dynamic adjustment calculation is as follows: Three custom variables are established: the predicted dosage value D', the total nitrogen concentration in the effluent of the process unit, and the total phosphorus concentration in the effluent. Dynamic control calculations are performed based on the predicted value D', effluent feedback data, and the correction rules of the carbon-nitrogen ratio coefficient K. The deviation signal is used to generate control quantities according to the gradient dosing logic. When implementing control measures, the electric water-stop clamps of the multi-point precision dosing distribution box should be adjusted first to optimize the flow distribution ratio of each dosing point. If the distribution ratio adjustment reaches its limit and still cannot meet the effluent water quality requirements, the output flow rate of the dosing pump should be adjusted, and the flow distribution ratio of each dosing point should be recalculated and set simultaneously.

10. The intelligent dosing method based on a multi-point precise dosing box and a gradient dosing algorithm as described in claim 9, characterized in that, The correction rule for the carbon-to-nitrogen ratio coefficient K is as follows: when the total nitrogen concentration in the effluent is >7.5 mg / L, the K value increases by 0.2; when the total nitrogen concentration in the effluent is <5 mg / L, the K value decreases by 0.1; the correction range of the K value is always kept between 3 and 4; when the predicted dosage D' is less than 300 L / h, it is forcibly set to 0; when the nitrate nitrogen concentration in the influent is less than 0.6 mg / L, the dosing of the reagent is stopped.

Images