Flocculation reactor structure parameter optimization system and method combined with physical test

Abstract

The invention relates to the technical field of water treatment equipment and process optimization, and discloses a flocculation reactor structure parameter optimization system combined with a physical test, which comprises a physical test module and a data acquisition module. According to the method, a double-objective function of flocculation efficiency maximization and energy consumption minimization is established, and a hybrid optimization strategy of the improved particle swarm optimization and the response surface method is combined, so that collaborative optimization of structure parameters and operation parameters is realized. Compared with a traditional single-target optimization method, the technical scheme can improve the flocculation treatment effect (the efficiency is improved by 15%-25%) and reduce the operation cost (the energy consumption is reduced by 20%-30%) at the same time. Specifically, the global search capability of the particle swarm algorithm is combined with the local fitting precision of the response surface method, so that the defect that a traditional algorithm is easy to fall into local optimum is avoided, and the stability and reliability of an optimization result are ensured. In addition, a dynamic weight adjustment mechanism enables the algorithm to quickly explore a parameter space in the initial stage of optimization and finely adjust a local area in the later stage, and the optimization efficiency is remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of water treatment equipment and process optimization, in particular to a flocculation reactor structure parameter optimization system and method combined with physical tests. BACKGROUND

[0002] As the core equipment of water treatment process, the flocculation reactor's structure parameters (such as impeller layout, flow velocity distribution) and operation parameters (such as reagent dosage, reaction time) directly determine the treatment effect and operation cost. The traditional optimization method mainly relies on experience or single-factor test, which has the following defects: Low efficiency: a large number of full-factor tests are required, the test period is as long as several months, and all parameter combinations cannot be covered; Insufficient accuracy: single-target optimization (such as only considering flocculation efficiency) may lead to high energy consumption or reagent waste; Poor adaptability: unable to respond to water quality fluctuations (such as sudden changes in raw water turbidity) in real time, manual intervention is required to adjust parameters; Model limitations: existing numerical simulation methods (such as CFD) do not accurately describe the coupling mechanism of turbulence and particle collision, and the prediction results may deviate by more than 20% from the actual situation.

[0003] In recent years, some research has attempted to introduce machine learning algorithms (such as support vector machines, neural networks) to assist parameter optimization, but there are the following problems: The algorithm relies on a large amount of historical data, and the data collection cost is high in actual engineering; The model has poor interpretability and is difficult to guide physical improvements to structure parameters (such as impeller angle); Real-time control function is not integrated, and the optimization results cannot be directly applied to equipment operation.

[0004] Therefore, the present application provides a flocculation reactor structure parameter optimization system and method combined with physical tests, which realizes the whole-process closed-loop optimization from physical test data collection, algorithm model construction, parameter optimization calculation to real-time control of equipment, and is suitable for flocculation process optimization in the fields of water treatment, wastewater treatment and industrial wastewater treatment. SUMMARY

[0005] The present application aims to provide a flocculation reactor structure parameter optimization system and method combined with physical tests to solve the problems raised in the background.

[0006] In order to achieve the above object, the present application provides the following technical scheme: a flocculation reactor structure parameter optimization system combined with physical test, comprising a physical test module, a data acquisition module, an optimization algorithm module and a control module; the physical test module comprises a flocculation reactor body with adjustable structure parameters, the reactor body is provided with a stirring device, a dosing device and a flow rate adjusting device; the data acquisition module comprises a turbidity sensor array arranged at different heights of the reactor, a flow rate sensor and a pH value sensor; the optimization algorithm module is internally provided with a parameter optimization model based on a hybrid optimization strategy, the model takes the maximum flocculation efficiency and the minimum energy consumption as double objective functions, and adopts an optimization strategy combining an improved particle swarm algorithm and a response surface method, wherein the individual velocity update formula of the particle swarm algorithm is: In the formula, is an inertia weight, is a particle is a velocity in the mth iteration, is a velocity in the mth iteration, is a velocity in the mth iteration, , is an acceleration constant, , is a random number in the interval [0, 1], is an individual optimal position, i.e., a particle is a historical optimal position found by the particle itself in the mth iteration, is a historical optimal position found by the particle itself in the mth iteration, is a global optimal position, i.e., a best position shared by all particles in the mth iteration, is a global optimal position, i.e., a best position shared by all particles in the mth iteration, is a current position, i.e., a particle is a current position, i.e., a particle is a current coordinate position in the mth iteration, is a current coordinate position in the mth iteration, is a current coordinate position in the mth iteration, is a current coordinate position in the mth iteration; the control module automatically adjusts the stirring speed, the reagent dosage and the flow rate of the reactor body according to the optimal parameter combination output by the optimization algorithm module.

[0007] Preferably, the reactor body of the physical test module adopts a cylindrical structure, the height to diameter ratio of which is 3:1 to 5:1, the stirring device adopts a double-layer paddle structure, the upper paddle is a three-blade propeller, and the lower paddle is a four-blade frame, the paddle diameter to reactor diameter ratio is 0.35:1 to 0.45:1; the dosing device comprises a metering pump and a static mixer, the static mixer is arranged at the reactor inlet, and the length to pipe diameter ratio thereof is 10:1 to 15:1; the flow rate adjusting device adopts a centrifugal pump controlled by a frequency converter, and the flow rate adjustment range is 2 times to 5 times of the effective volume of the reactor per hour.

[0008] Preferably, the data acquisition module further comprises a sludge concentration sensor arranged at the bottom of the reactor, which measures in the range of 0 g / L to 10 g / L with an accuracy of ±0.1 g / L; the optimization algorithm module is internally provided with a sludge concentration prediction model, which is constructed using a long short-term memory network, the input layer comprises the current sludge concentration, stirring speed and flow rate, the hidden layer is set to 2 layers, and the number of neurons in each layer is 50 to 100; the loss function of the sludge concentration prediction model adopts mean square error, and the expression is: In the formula, is the number of observation values, is the actual observation value, and is the predicted value. First, the actual sludge concentration data is collected; the predicted value of the sludge concentration is calculated using the prediction model; the MSE between the predicted value and the actual value is calculated; the calculated MSE is compared with the preset threshold value; if the MSE exceeds the threshold value, the sludge discharge operation is triggered.

[0009] Another technical problem to be solved by the present application is to provide a method for optimizing the structure parameters of a flocculation reactor combined with physical tests, comprising the following steps: Step one, establishing an initial database comprising reactor structure parameters, operation parameters and water quality parameters; Step two, designing an orthogonal test scheme, selecting stirring speed, reagent dosage, reaction time and flow rate as test factors, and setting 5 levels for each factor; Step three, executing the orthogonal test in the physical test module, and synchronously obtaining the turbidity value, flow rate value and pH value of each measuring point through the data acquisition module; Step four, constructing a prediction model based on support vector regression, and the kernel function adopts a radial basis function, and the expression is: In the formula, represents the kernel function value between samples and , i.e. the similarity between them, is an exponential function, and the base is the natural constant , is the width parameter of the kernel function, represents the square of the Euclidean distance between samples and ; Step five, inputting the test data into the optimization algorithm module, and performing parameter optimization using a multi-objective genetic algorithm, and the fitness function is set as: In the formula, is the flocculation efficiency, such as sludge settling velocity, turbidity removal rate and COD removal rate, is the theoretical maximum value of the index,​ for the running cost, including the cost of reagent, energy consumption, equipment maintenance cost and economic expenditure related to flocculation process, is the allowable upper limit of the index, is the weight coefficient; Step six, output the optimal parameter combination and verify its effectiveness.

[0010] Preferably, the initial database in step one contains optimal parameter combinations under different water quality conditions, and the water quality parameters include raw water turbidity, temperature and organic matter content; the orthogonal test scheme in step two adopts L25(5^6) orthogonal table, and the test factors also include paddle layer spacing and paddle tilt angle, the ratio of paddle layer spacing to reactor height is 0.15:1 to 0.25:1, and the paddle tilt angle is 30° to 60°; the data collection frequency in step three is 1 time per second, and the continuous collection time is not less than 3 hydraulic retention times; the support vector regression model in step four adopts the cross-validation method to determine the optimal hyperparameters, and the cross-validation fold is 5 folds.

[0011] Preferably, in the multi-objective genetic algorithm in step five, the selection operation adopts the tournament selection method, the crossover probability is set to 0.8 to 0.9, the mutation probability is set to 0.05 to 0.1; the algorithm termination condition is that the optimal fitness value change amount is less than 1e-4 for 20 consecutive generations; and the optimal parameter combination needs to meet the constraint condition that the stirring speed is not less than 50 rpm and not higher than 300 rpm, the reagent dosage is not less than 10 mg / L and not higher than 50 mg / L, and the reaction time is not less than 15 min and not higher than 60 min.

[0012] Preferably, the verification process in step six includes physical test verification and numerical simulation verification; the physical test verification repeats the optimal parameter combination test under the same water quality conditions, and the measurement indexes include flocculation efficiency, settling velocity and effluent turbidity; the numerical simulation verification adopts the computational fluid dynamics method to construct a numerical model including a turbulent flow model and a particle collision model, the turbulent flow model adopts the Realizable k-ε model, and the particle collision model adopts the soft sphere model; and the verification standard is that the relative error between the physical test result and the numerical simulation result is not more than 5%, and the flocculation efficiency is improved by not less than 15%.

[0013] Preferably, it also includes a real-time control mode and an offline optimization mode; in the real-time control mode, the control module automatically calls the pre-stored optimal parameter combination according to the current water quality parameters; in the offline optimization mode, the optimization algorithm module periodically executes the optimization method steps to update the parameter database.

[0014] The present application provides a flocculation reactor structure parameter optimization system and method combined with physical tests, which has the following beneficial effects: 1、The present application realizes the collaborative optimization of structure parameters and operation parameters by establishing a double-objective function of maximizing flocculation efficiency and minimizing energy consumption, combining the hybrid optimization strategy of improved particle swarm algorithm and response surface method. Compared with the traditional single-objective optimization method, this technical solution can improve the flocculation treatment effect (efficiency improvement of 15%-25%) and reduce the operating cost (energy consumption reduction of 20%-30%) at the same time. Specifically, the global search ability of particle swarm algorithm and the local fitting accuracy of response surface method are combined to avoid the defect that the traditional algorithm is easy to fall into local optimum, ensuring the stability and reliability of the optimization result. In addition, the dynamic weight adjustment mechanism enables the algorithm to quickly explore the parameter space in the early optimization stage and fine-tune the local area in the later stage, significantly improving the optimization efficiency.

[0015] 2、The real-time control and offline optimization dual-mode system proposed by the present application combines the PID control strategy and the Ziegler-Nichols parameter tuning method to realize the rapid response to water quality fluctuations. When the raw water turbidity changes by more than 20 NTU or the temperature changes by more than 5℃, the system automatically switches to the offline optimization mode, re-executes the orthogonal test and multi-objective genetic algorithm, and updates the parameter database. This design enables the device to maintain stable treatment effect (effluent turbidity stable below 5 NTU) when processing high variability water quality, while avoiding the running interruption caused by frequent optimization. The measured data shows that the mode switching response time is less than 2 minutes and the parameter adjustment accuracy is more than 98% in 24 hours of continuous operation.

[0016] 3、The present application realizes the precise control of sludge discharge operation by integrating the sludge concentration prediction model of LSTM neural network and combining the real-time feedback of the bottom sludge concentration sensor. The processing ability of LSTM model on time series data enables it to predict the sludge accumulation trend 30-60 minutes in advance with a prediction error of less than 5%. When the predicted concentration exceeds the set threshold (such as 8 g / L), the system automatically starts the sludge discharge program, and the sludge discharge amount is precisely controlled by adjusting the centrifugal pump flow through the frequency converter. This design avoids the waste of treatment capacity caused by traditional timed sludge discharge, and the measured sludge discharge efficiency is improved by 40%, the sludge concentration multiple is increased to 3-5 times, and the sludge disposal cost is significantly reduced. BRIEF DESCRIPTION OF DRAWINGS

[0017] Fig. 1 is the connection block diagram of the system modules of the present application; Fig. 2 is the process flow view of the method steps of the present application. DETAILED DESCRIPTION

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. 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.

[0019] Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the invention, and should not be construed as limiting the invention.

[0020] Example 1 A preferred embodiment of the flocculation reactor structural parameter optimization system and method combined with physical experiments provided by the present invention is as follows: Figs. 1-2 The system illustrates a flocculation reactor structural parameter optimization system combining physical experiments. It includes a physical experiment module, a data acquisition module, an optimization algorithm module, and a control module. The physical experiment module comprises a flocculation reactor body with adjustable structural parameters, equipped with a stirring device, a dosing device, and a flow rate regulating device. The data acquisition module includes an array of turbidity sensors, a flow rate sensor, and a pH sensor positioned at different heights within the reactor. The optimization algorithm module incorporates a parameter optimization model based on a hybrid optimization strategy. This model uses maximizing flocculation efficiency and minimizing energy consumption as dual objective functions, employing an optimization strategy combining an improved particle swarm optimization algorithm and a response surface methodology. The individual velocity update formula for the particle swarm optimization algorithm is: In the formula, For inertial weights, For particles In the During the nth iteration, the 1st The velocity dimension represents the particle's current "speed and direction of movement". , The acceleration constant, , A random number in the interval [0,1]. The optimal position for an individual, i.e., the particle In the During the nth iteration, at the... The best historical position found by itself. The globally optimal position, i.e., the position of the entire particle swarm at the [missing position]. During the nth iteration, at the... The optimal position shared by all particles in the dimension. The current position, i.e., the particle In the In the next iteration, in the first current coordinate position in the dimension; the control module automatically adjusts the stirring speed of the reactor body, the dosage of the reagent and the flow rate according to the optimal parameter combination output by the optimization algorithm module. The role of the control module is to automatically apply the optimal parameters (such as the optimal speed, reagent dosage, etc.) found by the algorithm to the actual reactor, realizing intelligent control; wherein, is the influence degree of the speed at the previous moment.

[0021] A larger (such as close to 1): enhances global search ability (particles tend to maintain the original movement direction and explore a larger space).

[0022] A smaller (such as close to 0): enhances local search ability (particles rely more on current optimal information to fine-tune the position).

[0023] Common strategy: linearly decreasing inertia weight (such as gradually decreasing from 0.9 to 0.4) to balance early exploration and late convergence.

[0024] In this embodiment, through this speed update formula, each particle adjusts its own movement direction and speed in the search space according to its own experience (individual optimal) and group experience (global optimal), combined with certain randomness and inertia, thereby gradually approaching the global optimal solution.

[0025] The reactor body of the physical test module adopts a cylindrical structure, and the height to diameter ratio is 3:1 to 5:1. The stirring device adopts a double-blade structure, the upper blade is a three-blade propeller, and the lower blade is a four-blade frame. The ratio of the blade diameter to the reactor diameter is 0.35:1 to 0.45:1. The reagent feeding device includes a metering pump and a static mixer. The static mixer is arranged at the reactor inlet, and the length to pipe diameter ratio is 10:1 to 15:1. The flow rate adjusting device adopts a centrifugal pump controlled by a frequency converter, and the flow rate regulation range is 2 times / hour to 5 times / hour of the effective volume of the reactor.

[0026] The data acquisition module also includes a sludge concentration sensor arranged at the bottom of the reactor, which has a measurement range of 0 g / L to 10 g / L and an accuracy of ±0.1 g / L. The optimization algorithm module has a built-in sludge concentration prediction model. The model is built using a long short-term memory network. The input layer includes the current sludge concentration, stirring speed and flow rate. The hidden layer is set to 2 layers, and the number of neurons in each layer is 50 to 100. The loss function of the sludge concentration prediction model uses mean squared error, and the expression is: wherein, is the number of observations, is the actual observation value, and is the a prediction value; first collecting actual sludge concentration data; calculating a prediction value of the sludge concentration using a prediction model; calculating the MSE between the prediction value and the actual value; comparing the calculated MSE with a preset threshold value; and triggering sludge discharge operation if the MSE exceeds the threshold value.

[0027] Embodiment 2 Please refer to Figs. 1-2 On the basis of Embodiment 1, it is further obtained that another technical problem to be solved by the present application is to provide a flocculation reactor structure parameter optimization method combined with physical test, comprising the following steps: Step one, establishing an initial database containing reactor structure parameters, operation parameters and water quality parameters; Step two, designing an orthogonal test scheme, selecting stirring speed, reagent dosage, reaction time and flow rate as test factors, and setting 5 levels for each factor; Step three, executing the orthogonal test in the physical test module, and synchronously obtaining the turbidity value, flow rate value and pH value of each measuring point through the data acquisition module; Step four, constructing a prediction model based on support vector regression, the kernel function of which adopts radial basis function, and the expression is as follows: In the formula, represents the kernel function value between samples and , i.e. the similarity between them, is an exponential function, and the base is the natural constant , is the width parameter of the kernel function, represents the square of the Euclidean distance between samples and ; Step five, inputting the test data into the optimization algorithm module, and adopting a multi-objective genetic algorithm to optimize the parameters, and the fitness function is set as: In the formula, is the flocculation efficiency, such as sludge settling velocity, turbidity removal rate, COD removal rate, is the theoretical maximum value of the index, is the running cost, including reagent cost, energy consumption, equipment maintenance cost and economic expenditure related to the flocculation process, is the allowed upper limit of the index, , is the weight coefficient; Step six, outputting the optimal parameter combination and verifying its effectiveness.

[0028] In the embodiment, the initial database in step one contains optimal parameter combinations under different water quality conditions, and the water quality parameters include raw water turbidity, temperature and organic matter content; the orthogonal test scheme in step two adopts an L25(5^6) orthogonal table, and the test factors also include paddle layer spacing and paddle tilt angle, the paddle layer spacing to reactor height ratio is 0.15:1 to 0.25:1, and the paddle tilt angle is 30° to 60°; the data collection frequency in step three is 1 time per second, and the continuous collection time is not less than 3 hydraulic retention times; and the support vector regression model in step four adopts a cross-validation method to determine optimal hyperparameters, and the cross-validation fold is 5 folds.

[0029] In the embodiment, in the multi-objective genetic algorithm in step five, a tournament selection method is adopted for selection operation, the crossover probability is set to 0.8 to 0.9, and the mutation probability is set to 0.05 to 0.1; the algorithm termination condition is that the optimal fitness value change amount is less than 1e-4 for 20 consecutive generations; and the optimal parameter combination needs to meet the constraint conditions that the stirring speed is not less than 50 rpm and not more than 300 rpm, the medicament dosage is not less than 10 mg / L and not more than 50 mg / L, and the reaction time is not less than 15 min and not more than 60 min.

[0030] In the embodiment, the verification process in step six includes physical test verification and numerical simulation verification; the physical test verification repeats the optimal parameter combination test under the same water quality conditions, and the measurement indexes include flocculation efficiency, settling velocity and effluent turbidity; the numerical simulation verification adopts a computational fluid dynamics method to construct a numerical model including a turbulent flow model and a particle collision model, the turbulent flow model adopts a Realizable k-ε model, and the particle collision model adopts a soft sphere model; and the verification standard is that the relative error between the physical test result and the numerical simulation result is not more than 5%, and the flocculation efficiency is improved by not less than 15%.

[0031] In the embodiment, real-time control mode and offline optimization mode are further included; in the real-time control mode, the control module automatically calls the pre-stored optimal parameter combination according to the current water quality parameters; in the offline optimization mode, the optimization algorithm module periodically executes the optimization method steps to update the parameter database; and the switching conditions of the two modes are that when the raw water turbidity change amount exceeds 20 NTU or the temperature change amount exceeds 5°C, the system is automatically switched to the offline optimization mode; and the control module adopts a PID control strategy to adjust the frequency converter frequency, and the PID parameters are determined by a Ziegler-Nichols tuning method, wherein the proportional gain is set to 0.5 to 1.5, the integral time is set to 0.2 min to 0.8 min, and the differential time is set to 0.05 min to 0.2 min.

[0032] It is to be noted that, in the present text, relational terms such as first and second and the like can be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0033] Finally, it should be noted that the above-mentioned only constitutes preferred embodiments of the present application and is not intended to limit the present application. Although the present application has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications, equivalent replacements, improvements and the like of the technical solutions described in the foregoing embodiments can still be made. Any modifications, equivalent replacements, improvements and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A flocculation reactor structure parameter optimization system combined with physical tests, characterized by, The physical test module, the data acquisition module, the optimization algorithm module and the control module are included; the physical test module contains a flocculation reactor body with adjustable structure parameters, the reactor body is provided with a stirring device, a dosing device and a flow rate adjusting device; the data acquisition module contains a turbidity sensor array arranged at different heights of the reactor, a flow rate sensor and a pH value sensor; the optimization algorithm module is internally provided with a parameter optimization model based on a hybrid optimization strategy, the model takes the maximum flocculation efficiency and the minimum energy consumption as double objective functions, and adopts an optimization strategy combining an improved particle swarm algorithm and a response surface method, wherein the individual velocity update formula of the particle swarm algorithm is: where, is the inertia weight, is the particle at the first iteration, the velocity of the particle in the dimension, indicating the current "speed and direction of movement" of the particle, , is the acceleration constant, , is a random number in the interval [0, 1], is the individual best position, i.e. the position of the particle at the first iteration, the historical best position found by itself in the dimension, is the global best position, i.e. the best position shared by all particles in the swarm in the dimension at the first iteration, is the current position, i.e. the position of the particle at the first iteration, the current coordinate position in the dimension; The control module automatically adjusts the stirring speed of the reactor body, the medicament dosage and the flow rate according to the optimal parameter combination output by the optimization algorithm module.

2. The system for optimizing the structure parameters of a flocculation reactor combined with a physical test according to claim 1, characterized in that: The reactor body of the physical test module adopts a cylindrical structure, the height to diameter ratio is 3:1 to 5:1, the stirring device adopts a double-layer paddle structure, the upper paddle is a three-blade propeller, the lower paddle is a four-blade frame, and the paddle diameter to reactor diameter ratio is 0.35:1 to 0.45:1; the dosing device contains a metering pump and a static mixer, the static mixer is arranged at the reactor inlet, and the length to pipe diameter ratio is 10:1 to 15:1; the flow rate adjusting device adopts a centrifugal pump controlled by a frequency converter, and the flow rate regulation range is 2 times to 5 times of the effective volume of the reactor per hour.

3. The system for optimizing the structure parameters of a flocculation reactor combined with a physical test according to claim 1, characterized in that: The data acquisition module further contains a sludge concentration sensor arranged at the bottom of the reactor, the measurement range is 0g / L to 10g / L, and the accuracy is ±0.1g / L; the optimization algorithm module is internally provided with a sludge concentration prediction model, the model is constructed by using a long short-term memory network, the input layer contains the current sludge concentration, the stirring speed and the flow rate, the hidden layer is set to 2 layers, and the number of neurons in each layer is 50 to 100; the loss function of the sludge concentration prediction model adopts a mean square error, and the expression is: In the formula, It is the number of observations, and it is the first... The first actual observation is the... First, collect actual sludge concentration data; (Predicted values;) The predicted value of the sludge concentration is calculated by using the prediction model; The MSE between the predicted value and the actual value is calculated, and the calculated MSE is compared with a preset threshold value; If the MSE exceeds the threshold value, the sludge discharge operation is triggered.

4. A method of optimizing the parameters of a flocculator structure in a flocculation reactor in combination with a physical test according to any one of claims 1 to 3, characterized in that, The following steps are included: Step one, establishing an initial database containing reactor structure parameters, operation parameters and water quality parameters; Step two, designing an orthogonal test scheme, selecting stirring speed, medicament dosage, reaction time and flow rate as test factors, and setting 5 levels for each factor; Step three, executing the orthogonal test in the physical test module, and synchronously acquiring the turbidity value, flow rate value and pH value of each measuring point by the data acquisition module; Step four, constructing a prediction model based on support vector regression, the kernel function adopts a radial basis function, and the expression is: wherein, denotes the similarity between samples and , is an exponential function with base the natural constant , is a width parameter of the kernel function, denotes the squared Euclidean distance between samples and . Step five, input the test data into the optimization algorithm module, use multi-objective genetic algorithm to optimize the parameters, and set the fitness function as: , wherein, is the flocculation efficiency, such as sludge settling velocity, turbidity removal rate, COD removal rate, is the theoretical maximum value of the index, is the operating cost, including reagent cost, energy consumption, equipment maintenance cost and economic expenditure related to flocculation process, is the allowable upper limit of the index, , is the weight coefficient; Step six, outputting the optimal parameter combination and verifying the effectiveness.

5. The method of optimizing the structure parameters of a flocculation reactor in combination with a physical test according to claim 4, characterized in that: The initial database in step one contains optimal parameter combinations under different water quality conditions, including raw water turbidity, temperature, and organic matter content; the orthogonal test scheme in step two uses an L25(5^6) orthogonal table, and the test factors also include paddle layer spacing and paddle tilt angle, with a paddle layer spacing to reactor height ratio of 0.15:1 to 0.25:1 and a paddle tilt angle of 30° to 60°; the data collection frequency in step three is 1 time per second, and the continuous collection time is not less than 3 hydraulic retention times; and the support vector regression model in step four uses a cross-validation method to determine the optimal hyperparameters, with a 5-fold cross-validation.

6. The method of optimizing the structure parameters of a flocculation reactor in combination with a physical test according to claim 4, characterized in that: In the multi-objective genetic algorithm in step five, the selection operation uses a tournament selection method, the crossover probability is set to 0.8 to 0.9, and the mutation probability is set to 0.05 to 0.1; the algorithm termination condition is that the optimal fitness value change amount is less than 1e-4 for 20 consecutive generations; and the optimal parameter combination needs to meet the constraint conditions: the stirring speed is not less than 50 rpm and not higher than 300 rpm, the reagent dosage is not less than 10 mg / L and not higher than 50 mg / L, and the reaction time is not less than 15 min and not higher than 60 min.

7. The method of optimizing the structure parameters of a flocculation reactor in combination with a physical test according to claim 4, characterized in that: The verification process in step six includes physical test verification and numerical simulation verification; the physical test verification repeats the optimal parameter combination test under the same water quality conditions, and the measurement indicators include flocculation efficiency, settling velocity, and effluent turbidity; the numerical simulation verification uses computational fluid dynamics methods to construct a numerical model containing a turbulence model and a particle collision model, the turbulence model uses a Realizable k-ε model, and the particle collision model uses a soft sphere model; and the verification standard is that the relative error between the physical test results and the numerical simulation results is not more than 5%, and the flocculation efficiency is improved by not less than 15%.

8. The method of optimizing the structure parameters of a flocculation reactor in combination with a physical test according to claim 4, characterized in that: It also includes real-time control mode and offline optimization mode; in the real-time control mode, the control module automatically calls the pre-stored optimal parameter combination according to the current water quality parameters; and in the offline optimization mode, the optimization algorithm module periodically executes its optimization method steps to update the parameter database.

Images