An intelligent control method for residual water treatment in dredging based on dynamic coupling response model

Abstract

The application discloses a kind of based on dynamic coupling response model's dredging residual water treatment intelligent control method, comprising: S1, the water quality of residual water treatment system day, the water quality of inlet and outlet, inlet and outlet flow and operating parameter are collected;S2, form operating variable and inlet disturbance variable, respectively stored in database U and D;S3, calculate the multi-objective benefit variable of day and store to database Y;S4, whether dynamic coupling response model exists and prediction error is judged, error is more than 10% or no model, then execute S5, otherwise execute S6;S5, establish dynamic coupling response model and carry out parameter rating based on database data;S6, input multi-objective benefit variable and inlet disturbance variable to dynamic coupling response model, solve the multi-objective Pareto solution set of next day, filter out the multi-objective coordination scheme of operating parameter, with replacement, return S1.The method of the application realizes multi-objective dynamic collaborative optimization, and improves the accuracy and foresight of control.

Description

Technical Field

[0001] This invention relates to the field of dredging wastewater treatment technology, specifically to an intelligent control method for dredging wastewater treatment based on a dynamic coupling response model. Background Technology

[0002] With increasingly stringent environmental protection requirements, the wastewater treatment industry is facing the severe challenge of shifting from "meeting discharge standards" to "energy conservation, consumption reduction, quality improvement, and efficiency enhancement." The dredging wastewater treatment process exhibits typical characteristics of conflicting and nonlinear objectives (water quality, cost, and output), placing higher demands on traditional control strategies.

[0003] In existing technologies, the automation control of dredging wastewater treatment processes has the following main shortcomings:

[0004] 1. Fragmented control strategies and lack of system coordination. Current systems mostly adopt independent control of single parameters, with each control loop isolated from the others, failing to consider the inherent coupling of the process. For example, increasing aeration to improve nitrification will increase energy consumption and exacerbate membrane fouling; reducing aeration to save energy may affect the effluent quality. This kind of local optimization easily leads to repeated oscillations between multiple objectives, making it impossible to achieve global optimum.

[0005] 2. Operation relies heavily on human experience, resulting in delayed response and poor economic efficiency. Faced with fluctuations in influent, operators mainly adjust based on experience, leading to subjective and delayed decision-making. To ensure water quality safety, conservative strategies such as excessive aeration and overdosing on chemicals are often adopted, resulting in wasted energy and chemical consumption; at low loads, operators are hesitant to reduce the load, leading to persistently high costs.

[0006] 3. Multiple objectives are difficult to quantify and weigh, resulting in a lack of basis for optimization. Objectives such as water quality, cost, and yield are diverse and mutually restrictive. Existing systems lack a unified quantitative evaluation framework, making scientific comparison and dynamic weighing impossible on the same scale. Most control systems simplify multiple objectives into a single objective or a fixed-weight problem, failing to flexibly adjust optimization priorities based on real-time operating conditions (such as influent load). Summary of the Invention

[0007] To address the problems existing in the prior art, this invention provides an intelligent control method for dredged wastewater treatment based on a dynamic coupling response model, which can be used to achieve multi-objective optimization.

[0008] Therefore, the present invention adopts the following technical solution:

[0009] A smart control method for treating dredged wastewater based on a dynamic coupling response model includes the following steps:

[0010] S1, collecting the first Operating data of the wastewater treatment system, including influent water quality, effluent water quality, and influent flow rate. Outflow rate and operating parameters;

[0011] S2, the operating parameters, influent water quality, and influent flow rate are processed to form operating variables. and inflow disturbance variables And store them in databases U and D respectively;

[0012] S3 calculates the multi-objective benefits of the wastewater treatment system on day k based on the operational data collected in S1, including water quality objectives. Cost targets and production targets These constitute multi-objective benefit variables. And store it in database Y;

[0013] S4, Model Adaptive Correction and Feedback Update:

[0014] If a dynamic coupling response model does not exist, then proceed with steps S5 and S6 in sequence.

[0015] If a dynamic coupled response model exists, then calculate the multi-objective benefit variables obtained from S3. Compared with the previous multi-objective benefit forecast Error: When the water quality target Cost targets and production targets If the prediction error of any target is greater than 10%, proceed to step S5; if the prediction error of all targets is less than or equal to 10%, proceed to step S6.

[0016] S5. Establish a dynamic coupling response model and calibrate the parameters of the dynamic coupling response model based on all data in the current database U, D, and Y.

[0017] S6, with the aforementioned , Using the input data of the current dynamic coupling response model, and within the feasible range of system operating parameters, a multi-objective intelligent algorithm is used to solve for the first... The corresponding multi-objective Pareto solution set for the day; the first [objective] is selected from the multi-objective Pareto solution set according to the TOPSIS method. Multi-objective coordination scheme for daily system operation parameters and the corresponding multi-objective benefit prediction values According to the multi-objective coordination scheme Adjust the system's operating parameters, using replace Return to step S1 and repeat the process.

[0018] In step S1 above:

[0019] The The initial value is ≥90;

[0020] The water quality being assessed includes the influent chemical oxygen demand (COD). Total phosphorus in influent Ammonia nitrogen in influent and total nitrogen in the influent ;

[0021] The effluent water quality includes the effluent chemical oxygen demand (COD). Total phosphorus in effluent Ammonia nitrogen in effluent and total nitrogen in effluent ;

[0022] The operating parameters include the system aeration rate. Carbon source usage and dosage of dephosphorizing agent .

[0023] The specific operation of step S2 above is as follows:

[0024] The operating parameters and influent water quality obtained from S1 are processed to form operating variables. =[ , , and inflow disturbance variables =[ , , ], and the operation variable and the aforementioned inflow disturbance variables Stored into databases U and D respectively, resulting in U=[ ] and D=[ ], where, if If it is the initial value, then and According to the system number The operating parameters, influent water quality, and influent flow rate of the day were compiled, and the resulting data was used to... The day's operational variables and water inflow disturbance variables.

[0025] In step S3 above:

[0026] The water quality target The calculation formula is:

[0027] ,

[0028] In the formula, , , , These are the weighting coefficients; , , , The target value for the water quality indicator is mg / L;

[0029] The cost target The calculation formula is:

[0030] ,

[0031] In the formula, For the aerator's power efficiency, m 3 O2 / kW·h; The electricity price is in yuan / kW·h; The unit price of the carbon source is yuan / L; Price of phosphorus removal agent, in yuan / L; The unit is m 3 / h; The unit is L / h; The unit is L / h;

[0032] The production target The calculation formula is:

[0033] ,

[0034] In the formula, To design the water output, m 3 / h.

[0035] The weighting coefficient , , , The sum of is 1.

[0036] The multi-objective benefit variables =[ , , ].

[0037] The specific operation of step S5 above is as follows:

[0038] S51, Establish a dynamic coupled response model, the formula is as follows:

[0039] ,

[0040] in, , , It is a 3x3 parameter matrix. This is the state transition matrix, with the diagonal values ​​ranging from 0.6000 to 0.900 and the off-diagonal values ​​ranging from -0.500 to 0.500. To control the input matrix, the value range is -0.0100 to 0.0100; The perturbation input matrix has a value range of -0.0100 to 0.0100. Nonlinear coupling terms for compensation in RBF neural networks;

[0041] S52, determine the parameters of the dynamic coupled response model through calibration:

[0042] Based on the data in databases U, D, and Y, the parameters of the above dynamic coupling response model are calibrated to determine the parameter matrix. , , and nonlinear coupling terms .

[0043] In step S52 above:

[0044] The parameter matrix , , Fitting using the least squares method;

[0045] The nonlinear coupling term The results are obtained by calling a neural network toolkit. When calling the toolkit, the input layer, hidden layer, and output layer need to be specified. The objective of the nonlinear fitting is... The value is the smallest.

[0046] In step S6 above:

[0047] The multi-objective intelligent algorithm can select from the non-dominated sorting genetic algorithm NSGA-II with an elitist strategy, the non-dominated neighborhood immune algorithm NNIA, and the multi-objective particle swarm optimization algorithm MOPSO.

[0048] In summary, the method of the present invention effectively solves the industry pain points such as difficulty in balancing multiple objectives, high operating costs, and reliance on manual experience in the treatment of dredged wastewater. It provides a complete, efficient, and easy-to-implement technical solution for achieving refined, intelligent, and low-carbon operation of dredged wastewater treatment.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] 1. The method of this invention achieves multi-objective dynamic collaborative optimization, solving the problem of objective conflict in traditional control. Traditional control strategies often adjust a single objective in isolation. This invention constructs a three-dimensional quantitative evaluation system of water quality, cost, and yield, and optimizes multiple objectives. It can automatically find the optimal balance point under different operating conditions, realizing a fundamental shift from "single objective achievement" to "multi-objective collaborative optimization".

[0051] 2. The method of this invention establishes a dynamic coupling response model of key variables and multiple objectives, which improves the accuracy and foresight of control.

[0052] 3. This invention abandons the single-point, single-loop control mode. Through a dynamic coupling response model, it reveals the cascading effects of key operational variables such as aeration rate, carbon source dosage, and phosphorus removal agent dosage on water quality, cost, and yield, forming a global collaborative control strategy. This not only responds to current changes but also predicts the consequences of adjustments, improving the process's resilience and adaptability. Attached Figure Description

[0053] Figure 1 This is a flowchart of the intelligent control method for treating dredged wastewater in an embodiment of the present invention;

[0054] Figure 2 This demonstrates the calibration effect of the dynamic coupling response model in the embodiments of the present invention.

[0055] Figure 3 This is a spatial distribution map of the Pareto target in an embodiment of the present invention;

[0056] Figure 4 This is a schematic diagram of a multi-objective coordination scheme in an embodiment of the present invention;

[0057] Figure 5 This is a comparison diagram of the state before and after optimization in an embodiment of the present invention. Detailed Implementation

[0058] The technical solution of the invention will be clearly and completely described below with reference to the accompanying drawings and embodiments. Obviously, the following embodiments are only some embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] A smart control method for dredged wastewater treatment based on a dynamic coupling response model, such as Figure 1 As shown, it includes the following steps:

[0060] S1, Data Acquisition:

[0061] Collection of the first day( Initial value ≥ 90) Wastewater treatment system operation data, mainly including influent water quality, effluent water quality, and influent flow rate. Outflow rate and operating parameters, where:

[0062] The water quality being assessed includes the influent chemical oxygen demand (COD). Total phosphorus in influent Ammonia nitrogen in influent and total nitrogen in the influent ;

[0063] The effluent water quality includes the effluent chemical oxygen demand (COD). Total phosphorus in effluent Ammonia nitrogen in effluent and total nitrogen in effluent ;

[0064] The operating parameters include the system aeration rate. Carbon source usage and dosage of dephosphorizing agent .

[0065] S2, Determine the influent disturbance variables and operational variables of the wastewater treatment system:

[0066] The wastewater treatment system obtained from S1 Daily operating parameters, influent water quality, and influent flow rate are compiled into operating variables. =[ , , and inflow disturbance variables =[ , , ], and the operation variable and the aforementioned inflow disturbance variables Stored into databases U and D respectively, resulting in U=[ ] and D=[ ], where, if If it is the initial value, then and According to the system number The operating parameters, influent water quality, and influent flow rate of the day were compiled to obtain the first... The day's operational variables and water inflow disturbance variables.

[0067] S3, Evaluate the multi-objective benefits of the wastewater treatment system:

[0068] The multi-objective benefits of the wastewater treatment system on day k are calculated based on the system operation data collected by S1, including water quality objectives. Cost targets and production targets These constitute multi-objective benefit variables. =[ , , ], and multi-objective benefit variables Stored in database Y, resulting in a new database Y=[ ], The system is based on the system number The data obtained from the daily operation The multi-objective benefit variables for each day, including:

[0069] The water quality target The calculation formula is:

[0070] ,

[0071] In the formula, , , , These are weighting coefficients (the specific values ​​are determined based on actual water quality concerns, and the sum of the weighting coefficients is 1). , , , The target value for the water quality indicator is mg / L.

[0072] The cost target The calculation formula is:

[0073] ,

[0074] In the formula, For the aerator's power efficiency, m 3 O2 / kW·h; The electricity price is in yuan / kW·h; The unit price of the carbon source is yuan / L; Price of phosphorus removal agent, in yuan / L; The unit is m 3 / h; The unit is L / h; The unit is L / h.

[0075] The production target The calculation formula is:

[0076] ,

[0077] In the formula, To design the water output, m 3 / h.

[0078] S4, Model Adaptive Correction and Feedback Update:

[0079] If this invention is being run on a system for the first time and a dynamic coupling response model does not currently exist, then steps S5 and S6 are executed directly in sequence.

[0080] If the method of this invention is not being run on the system for the first time, and a dynamic coupling response model currently exists, then the multi-objective benefit variables obtained from calculating S3 are... Compared with the previous multi-objective benefit forecast Error:

[0081] When water quality target Cost targets and production targets If the prediction error of any target is greater than 10%, proceed to step S5;

[0082] When water quality target Cost targets and production targets If the prediction error of all targets is less than or equal to 10%, proceed to step S6.

[0083] S5, Calibration of dynamic coupling response model parameters:

[0084] S51, Establish a dynamic coupled response model, the formula is as follows:

[0085] ,

[0086] The dynamic coupling response model consists of two parts. For the mechanism prediction part, This is the data-driven compensation part.

[0087] in, , , It is a 3x3 parameter matrix. The state transition matrix represents the dynamic characteristics of the target variable itself over time. The diagonal values ​​range from 0.6000 to 0.900, and the off-diagonal values ​​range from -0.500 to 0.500. The input matrix is ​​used to control the impact of the manipulated variables on the target, and its value ranges from -0.0100 to 0.0100. The disturbance input matrix represents the impact of uncontrollable disturbances such as inflow load on the target, with values ​​ranging from -0.0100 to 0.0100; This is the nonlinear coupling term for compensation in the RBF neural network.

[0088] S52, determine the parameters of the dynamic coupled response model through calibration:

[0089] Based on all the data in the current databases U, D, and Y, the parameters of the aforementioned dynamic coupling response model are calibrated to determine the parameter matrix. , , and nonlinear coupling terms The specific steps are as follows:

[0090] The parameter matrix , , Fitting using the least squares method;

[0091] The nonlinear coupling term The results are obtained by calling a neural network toolkit. When calling the toolkit, the input layer, hidden layer, and output layer need to be specified. The objective of the nonlinear fitting is... The value is the smallest.

[0092] S6, Real-time optimization decision-making based on multi-objective intelligent algorithms:

[0093] by , Using the input data of the current dynamic coupling response model, and within the feasible range of system operating parameters, a multi-objective intelligent algorithm is employed to solve for the... The multi-objective Pareto solution set corresponding to the day is used to select the first [objective] from the multi-objective Pareto solution set according to the TOPSIS method. Multi-objective coordination scheme for daily system operation parameters and the corresponding multi-objective benefit prediction values .

[0094] According to the multi-objective coordination scheme Adjust the system's operating parameters and use... replace Return to step S1 to achieve a loop.

[0095] The multi-objective intelligent algorithm can be selected from: Non-dominated sorting genetic algorithm with elitist strategy (NSGA-II), Non-dominated neighborhood immune algorithm (NNIA), and Multi-objective particle swarm optimization algorithm (MOPSO).

[0096] Example

[0097] In this embodiment, the intelligent wastewater control method of the present invention is applied to a wastewater treatment platform system for dredging in Taihu Lake. The system is designed to produce a certain amount of wastewater. 45m 3 The treatment capacity is [ / h], with the core process consisting of an AO tank + a V-MBR tank + an effluent protection system. The treatment target is to meet the Class III surface water standard (i.e., COD < 20 mg / L, TN < 1 mg / L, NH3-N < 1 mg / L, TP < 0.05 mg / L). The specific process is as follows:

[0098] S1, Data Acquisition:

[0099] Data on the operation of the wastewater treatment system were collected on day 361. The influent water quality, effluent water quality, and influent and effluent flow rates are shown in Table 1.

[0100] Table 1

[0101]

[0102] System aeration rate =285m 3 / h, carbon source usage =25.5L / h, phosphorus removal agent dosage =14.4L / h.

[0103] S2, Determine the influent disturbance variables and operational variables of the wastewater treatment system:

[0104] Get the operation variable =[ , , ], inflow disturbance variables =[ , , ], and the operation variable and inflow disturbance variables The data is stored in databases U and D respectively. Databases U and D also include operational variables compiled based on the system's operating parameters, influent water quality, and influent flow rate from day 1 to day 360. and inflow disturbance variables .

[0105] S3, Evaluate the multi-objective benefits of the wastewater treatment system:

[0106] The multi-objective benefits of the wastewater treatment system on day 361 are calculated based on S1.

[0107] =0.7;

[0108] 2.38;

[0109] 0.79;

[0110] Obtaining multi-objective benefit variables =[ , , ], and The data is stored in database Y, which also includes multi-objective benefit variables calculated based on the system's operational data for days 1-360. ;

[0111] S4, Model Adaptive Correction and Feedback Update:

[0112] Since this is the first time the system is running, proceed directly to step S5;

[0113] S5, Calibration of dynamic coupling response model parameters:

[0114] Based on the databases U, D, and Y, the following dynamic coupled response model is fitted using a two-stage method combining least squares and a neural network. Perform parameter calibration to determine the parameter matrix. , , as follows:

[0115] = , = , = ;

[0116] Nonlinear coupling terms The input layer has 9 layers, the hidden layer has 15 layers, and the output layer has 3 layers. The neural network toolbox can be used to obtain the fitting results. The calibration effect of the dynamic coupling response model is as follows: Figure 2 As shown, the actual value is quite close to the predicted value. The value is relatively high, indicating a good calibration effect.

[0117] S6, Real-time optimization decision-making based on multi-objective intelligent algorithms:

[0118] by , The input data for the current dynamic coupling response model consists of A, B, C, and the neural network compensation term determined in S5. For model parameters, within the feasible range of system operating parameters ( Aeration rate range: 94~328 (Values ​​range 3-31, phosphorus removal agent dosage range 3-24). Using the multi-objective intelligent algorithm NSGA-II, the multi-objective Pareto solution set for day 362 is solved. The spatial stepwise partitioning of the Pareto objective is as follows: Figure 3 As shown, the TOPSIS method is used to filter out, such as Figure 4 The multi-objective coordination scheme shown and the corresponding multi-objective benefit prediction values ,according to Adjust the system's operating parameters and return to step S1. When step S4 is executed again: , The prediction error range is [5.7%, 3.1%, -2.5%], all less than 10%. Skip step S5, do not recalibrate the parameters of the dynamic coupled response model, and directly run S6.

[0119] The multi-objective benefit variables on day 361 The predicted multi-objective benefits on day 362 The comparison yielded the following results: Figure 5As shown, comparing the states before and after optimization, it can be seen that the water quality has been significantly improved, the water quality target has been optimized by about 23.5%, the cost target has been reduced by about 5.9%, and the surplus water production (production target) has increased by about 2.7%.

Claims

1. A smart control method for treating dredged wastewater based on a dynamic coupling response model, characterized in that, Includes the following steps: S1, collecting the first Operating data of the wastewater treatment system, including influent water quality, effluent water quality, and influent flow rate. Outflow rate and operating parameters; S2, the operating parameters, influent water quality, and influent flow rate are processed to form operating variables. and inflow disturbance variables And store them in databases U and D respectively; S3 calculates the multi-objective benefits of the wastewater treatment system on day k based on the operational data collected in S1, including water quality objectives. Cost targets and production targets , constituting multi-objective benefit variables And store it in database Y; S4, Model Adaptive Correction and Feedback Update: If a dynamic coupling response model does not exist, then proceed with steps S5 and S6 in sequence. If a dynamic coupled response model exists, then calculate the multi-objective benefit variables obtained from S3. Compared with the previous multi-objective benefit forecast Error: When the water quality target Cost targets and production targets If the prediction error of any target is greater than 10%, proceed to step S5; if the prediction error of all targets is less than or equal to 10%, proceed to step S6. S5. Establish a dynamic coupling response model and calibrate the parameters of the dynamic coupling response model based on all data in the current database U, D, and Y. S6, with the aforementioned , Using the input data of the current dynamic coupling response model, and within the feasible range of system operating parameters, a multi-objective intelligent algorithm is used to solve for the first... The corresponding multi-objective Pareto solution set for the day; the first solution is selected from the multi-objective Pareto solution set according to the TOPSIS method. Multi-objective coordination scheme for daily system operation parameters and the corresponding multi-objective benefit prediction values According to the multi-objective coordination scheme Adjust the system's operating parameters, using replace Return to step S1 and repeat the process.

2. The intelligent control method for treating dredged wastewater according to claim 1, characterized in that, In step S1: The The initial value is ≥90; The water quality being assessed includes the influent chemical oxygen demand (COD). Total phosphorus in influent Ammonia nitrogen in influent and total nitrogen in the influent ; The effluent water quality includes the effluent chemical oxygen demand (COD). Total phosphorus in effluent Ammonia nitrogen in effluent and total nitrogen in effluent ; The operating parameters include the system aeration rate. Carbon source usage and dosage of dephosphorizing agent .

3. The intelligent control method for treating dredged wastewater according to claim 2, characterized in that, The specific operation of step S2 is as follows: The operating parameters and influent water quality obtained from S1 are processed to form operating variables. =[ , , and inflow disturbance variables =[ , , ], and the operation variable and the aforementioned inflow disturbance variables Stored into databases U and D respectively, resulting in U=[ ] and D=[ ], where, if If it is the initial value, then and According to the system number The operating parameters, influent water quality, and influent flow rate of the day were compiled to obtain the first... The day's operational variables and water inflow disturbance variables.

4. The intelligent control method for treating dredged wastewater according to claim 3, characterized in that, In step S3: The water quality target The calculation formula is: ， In the formula, , , , These are the weighting coefficients; , , , The target value for the water quality indicator is mg / L; The cost target The calculation formula is: ， In the formula, For the aerator's power efficiency, m 3 O2 / kW·h; The electricity price is in yuan / kW·h; The unit price of the carbon source is yuan / L; Price of phosphorus removal agent, in yuan / L; The unit is m 3 / h; The unit is L / h; The unit is L / h; The production target The calculation formula is: ， In the formula, To design the water output, m 3 / h.

5. The intelligent control method for treating dredged wastewater according to claim 4, characterized in that: The weighting coefficient , , , The sum of is 1.

6. The intelligent control method for treating dredged wastewater according to claim 4, characterized in that: The multi-objective benefit variables =[ , , ].

7. The intelligent control method for treating dredged wastewater according to claim 6, characterized in that, The specific operation of step S5 is as follows: S51, Establish a dynamic coupled response model, the formula is as follows: ， in, , , It is a 3x3 parameter matrix. This is the state transition matrix, with the diagonal values ​​ranging from 0.6000 to 0.900 and the off-diagonal values ​​ranging from -0.500 to 0.

500. To control the input matrix, the value range is -0.0100 to 0.0100; The perturbation input matrix has a value range of -0.0100 to 0.0100. Nonlinear coupling terms for compensation in RBF neural networks; S52, determine the parameters of the dynamic coupled response model through calibration: Based on the data in databases U, D, and Y, the parameters of the above dynamic coupling response model are calibrated to determine the parameter matrix. , , and nonlinear coupling terms .

8. The intelligent control method for treating dredged wastewater according to claim 7, characterized in that, In step S52: The parameter matrix , , Fitting using the least squares method; The nonlinear coupling term The results are obtained by calling a neural network toolkit. When calling the toolkit, the input layer, hidden layer, and output layer need to be specified. The objective of the nonlinear fitting is... The value is the smallest.

9. The intelligent control method for treating dredged wastewater according to claim 1, characterized in that, In step S6: The multi-objective intelligent algorithm can select from the non-dominated sorting genetic algorithm NSGA-II with an elitist strategy, the non-dominated neighborhood immune algorithm NNIA, and the multi-objective particle swarm optimization algorithm MOPSO.

Images