Control method of double-chamber pipeline mixer based on support vector regression-differential optimization algorithm

Abstract

The application discloses a control method of a double-chamber pipeline mixer based on a support vector regression-differential optimization algorithm. The method is aimed at an existing double-chamber pipeline mixer, and combines a support vector regression and a differential optimization algorithm to obtain a corresponding relationship between flow and an inclination angle alpha of an inlet guide vane of a main pipeline, an inclination angle beta of an outlet guide vane, an inclination angle theta of a branch pipeline guide vane and a rotating speed n of an impeller. Based on the corresponding relationship, when the flow deviates, the structure of an adjusting device is adjusted in time, so that the mixer always keeps at a highest efficiency point.

Description

Technical Field

[0001] This invention relates to the field of pipe mixing devices, and more specifically to a control method for a dual-chamber pipe mixer based on a support vector regression-differential optimization algorithm. Background Technology

[0002] Against the backdrop of rapid economic development, wastewater from production and daily life has increased dramatically in recent years, highlighting water environment problems. A pipeline mixer achieves uniform mixing of fluids flowing through a pipeline through the action of a component or mixing element. The mixer structure largely determines the flow field and the final mixing result; therefore, adjusting and controlling the mixer structure can improve the overall mixing efficiency. For the same mixer, once the device structure is determined, its maximum mixing efficiency is also fixed. That is, when a certain parameter changes, the corresponding mixer flow rate and maximum efficiency will also change.

[0003] Current research has shown that changes in mixer structural parameters affect mixing efficiency. For mixing tanks, the mixing effect is best when the impeller placement angle is -15°. Different hyperbolic agitator blade structures also affect the final result, with the mixing effect from best to worst being: high-blade hyperbolic surface, perforated hyperbolic surface, hollow hyperbolic surface, and solid hyperbolic surface. The impeller speed has a significant impact on the tangential velocity and turbulent kinetic energy near the impeller, but a smaller impact on eddy viscosity. Some scholars have used orthogonal experimental schemes to analyze the impeller type, number of blades, installation height, and width-to-diameter ratio, and found that the mixer with the optimal size has higher mixing efficiency.

[0004] Currently used mixers have large head losses and are prone to dead zones. Moreover, most mixers are designed based on simple simulation calculations based on experience. When the flow rate changes, the mixing efficiency of the mixer will also change. Because existing mixers are difficult to adjust with changes in operating conditions, energy loss occurs, which affects subsequent processes. Summary of the Invention

[0005] To address the problem that existing mixers cannot adjust to changes in operating conditions, resulting in low mixing efficiency, this invention proposes a control method for a dual-chamber pipe mixer based on support vector regression-differential optimization algorithm, ensuring that the mixer can operate at maximum efficiency.

[0006] The objective of this invention is achieved through the following technical solution:

[0007] A control method for a dual-chamber pipe mixer based on support vector regression-differential optimization algorithm, wherein the dual-chamber pipe mixer includes a main inlet pipe, a branch pipe, a dosing pipe, an upper mixing chamber, a lower mixing chamber, a main outlet pipe, an impeller, a main pipe inlet guide plate, a main pipe outlet guide plate, and a branch pipe guide plate;

[0008] The main inlet pipe, lower mixing chamber, and main outlet pipe are connected in sequence; one end of the branch pipe is connected to the main inlet pipe, and the other end is connected to the upper mixing chamber; the dosing pipe is installed on the branch pipe; the upper mixing chamber and lower mixing chamber are connected, and the impeller is installed at the connection between the upper and lower mixing chambers; the main pipe inlet guide plate is installed inside the main inlet pipe, the main pipe outlet guide plate is installed inside the main outlet pipe, and the branch pipe guide plate is installed inside the branch pipe and located downstream of the dosing pipe connection; the main inlet pipe and the main outlet pipe have the same diameter;

[0009] The control method includes the following steps:

[0010] Step 1: Select the main pipe diameter ratio λ, the inclination angle of the main pipe inlet guide plate α, the inclination angle of the outlet pipe guide plate β, the inclination angle of the branch pipe guide plate θ, the impeller installation height h, and the impeller speed n as the mixer model parameter variables;

[0011] Step 2: Using an existing two-chamber pipe mixer as a reference, the values ​​of λ, α, β, θ, and n are given to be in the range of 0.91 to 1.11λ. R , 0.5~1.17α R 0.6–1.4β R 0.67~1.33θ R 0.5~1.5n R ; where λ R α R β R θ R n R All parameters are from existing dual-chamber pipe mixers;

[0012] Step 3: The given mixer flow rate Q is in the range of 0.6 to 1.4Q. R Q R The rated flow rate of the mixer is used; simultaneously, the relationship between Q, n, and h is used as a constraint, and λ, α, β, θ, n, and Q are used as variables, and sampling is performed using the super-Latin sampling method;

[0013]

[0014]

[0015]

[0016] Where D1 is the main pipe diameter, D2 is the branch pipe diameter, v is the inlet flow velocity of the main inlet pipe, and k is the proportionality coefficient, which represents the chemical reaction time per unit length between different raw water and the medicine added to the branch pipe.

[0017] Step 4: For each set of variables obtained in Step 3, model the sample and perform simulation calculations to obtain the outlet mixing variation coefficient CV and impeller power consumption P corresponding to each set of variable samples;

[0018] Step 5: Normalize the series of CVs and Ps obtained in Step 4;

[0019] Step 6: Use the samples obtained by the hyper-Latin sampling method in Step 3 as the input to the support vector regression model, and use the normalized CV and P obtained in Step 5 as the output to train the support vector regression model.

[0020] Step 7: Define the objective function F, and use the differential optimization algorithm to obtain the correspondence E between the flow rate Q and λ, α, β, θ, h, and n at the minimum value of the objective function; where, τ is the weight;

[0021] Step 8: Calculate λ and h using the following formula:

[0022] λ = 0.6λ R +0.2λ 0.8R +0.2λ 1.2R

[0023] h = 0.6h R +0.2h 0.8R +0.2h 1.2R

[0024] Where, λ 0.8R h 0.8R The flow rate is 0.8Q respectively. R Input the λ and h values ​​obtained from the correspondence E; λ 1.2R and h 1.2R The flow rate is 1.2Q. R Input the λ and h values ​​obtained from the correspondence E;

[0025] Step 9: Substitute the λ and h calculated in Step 8 into the relationship E obtained in Step 7 to obtain the correspondence E' between the flow rate Q and α, β, θ, and n under the minimum value of the objective function; input the flow rate Q into the corresponding relationship E' to obtain the values ​​of α, β, θ, and n under that flow rate, thereby controlling the impeller, the main pipe inlet guide plate, the main pipe outlet guide plate, and the branch pipe guide plate to maximize the mixing efficiency of the mixer.

[0026] Further, based on the values ​​of α, β, θ, and n obtained in step nine for each flow rate, modeling and simulation calculations are performed to calculate CV and P. At the same time, the values ​​of α, β, θ, and n obtained in step nine for each flow rate, as well as the corresponding λ and h, are input into the trained support vector regression model to obtain the model output CV and P. If the error between the CV and P predicted by the support vector regression model and the simulation calculation is greater than 3%, then return to step seven to adjust the parameters of the initial population of the differential evolution algorithm.

[0027] Furthermore, the impeller has three blades.

[0028] Furthermore, the formulas for calculating the outlet mixing variation coefficient CV and the impeller power consumption P are as follows:

[0029]

[0030] Where: S is the standard deviation of export data. is the average value of the export data; M is the torque on the impeller.

[0031] The beneficial effects of this invention are as follows:

[0032] This invention combines support vector regression and differential optimization algorithms to obtain the correspondence between flow rate and the tilt angle α of the main pipe inlet guide vane, the tilt angle β of the outlet pipe guide vane, the tilt angle θ of the branch pipe guide vane, and the impeller speed n. Based on this, when the flow rate deviates, the structure of the regulating device is adjusted in time, so that the mixer always remains at the highest efficiency point. Attached Figure Description

[0033] Figure 1 This is a simplified diagram of the mixer model.

[0034] Figure 2 The diagram shows the optimization process of the model.

[0035] Figure 3 A comparison of the outlet mixing situation before and after optimization at 0.6Q.

[0036] Figure 4 To optimize the operation flowchart.

[0037] In the diagram, 1. Main inlet pipe; 2. Main pipe inlet baffle; 3. Branch pipe; 4. Dosing pipe; 5. Branch pipe baffle; 6. Upper mixing chamber; 7. Upper mixing chamber baffle; 8. Variable frequency stirring device; 9. Impeller; 10. Lower mixing chamber; 11. Lower mixing chamber baffle; 12. Main outlet pipe; 13. Main pipe outlet baffle. Detailed Implementation

[0038] The present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments. The purpose and effects of the present invention will become clearer. It should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0039] The control method of the present invention is applicable to dual-chamber pipe mixers, such as... Figure 1 As shown, it includes a main inlet pipe 1, a main inlet baffle 2, a branch pipe 3, a dosing pipe 4, a branch baffle 5, an upper mixing chamber 6, an upper mixing chamber baffle 7, a variable frequency stirring device 8, an impeller 9, a lower mixing chamber 10, a lower mixing chamber baffle 11, a main outlet pipe 12, and a main outlet baffle 13.

[0040] The main inlet pipe 1, lower mixing chamber 10, and main outlet pipe 12 are connected in sequence. One end of the branch pipe 3 is connected to the main inlet pipe 1, and the other end is connected to the upper mixing chamber 6. The dosing pipe 4 is installed on the branch pipe 3. The upper mixing chamber 6 and lower mixing chamber 10 are connected. The impeller 9, driven by the variable frequency stirring device 8, is installed at the connection between the upper mixing chamber 6 and lower mixing chamber 10. The main pipe inlet guide plate 2 is installed inside the main inlet pipe 1, the main pipe outlet guide plate 13 is installed inside the main outlet pipe 12, and the branch pipe guide plate 5 is installed inside the branch pipe 3, located downstream of the connection of the dosing pipe 4. The main inlet pipe 1 and the main outlet pipe 12 have the same diameter. To further increase the mixing uniformity at the outlet of the main outlet pipe 12, an upper mixing chamber baffle 7 is installed in the upper mixing chamber 6, and a lower mixing chamber baffle 11 is installed in the lower mixing chamber 10.

[0041] like Figure 2 As shown, the control method of the present invention specifically includes the following steps:

[0042] Step 1: Select the main pipe diameter ratio λ, the inclination angle of the main pipe inlet guide plate α, the inclination angle of the outlet pipe guide plate β, the inclination angle of the branch pipe guide plate θ, the impeller installation height h, and the impeller speed n as the mixer model parameter variables;

[0043]

[0044] Where D1 is the diameter of the main pipe and D2 is the diameter of the branch pipe.

[0045] Step 2: Using an existing two-chamber pipe mixer as a reference, the values ​​of λ, α, β, θ, and n are given to be in the range of 0.91 to 1.11λ. R , 0.5~1.17α R 0.6–1.4β R 0.67~1.33θ R 0.5~1.5n R ; where λ R α R ,,β R , θR n R All parameters are from existing dual-chamber pipe mixers;

[0046] Step 3: The given mixer flow rate Q is in the range of 0.6 to 1.4Q. R Q R The rated flow rate of the mixer is used; simultaneously, the relationship between Q, n, and h is used as a constraint, and λ, α, β, θ, n, and Q are used as variables, and sampling is performed using the super-Latin sampling method;

[0047]

[0048]

[0049]

[0050] Where v is the inlet flow velocity of the main inlet pipe; k is a proportionality coefficient, representing the chemical reaction time per unit length between different raw water and the medicine added to the branch pipe;

[0051] Step 4: For each set of variables obtained in Step 3, model, mesh, and perform simulation calculations to obtain the outlet mixing variation coefficient CV and impeller power consumption P corresponding to each set of variable samples;

[0052]

[0053]

[0054] Where S is the standard deviation of export data. M represents the average value of the export data, and M is the torque on the impeller.

[0055] Step 5: Normalize the series of CVs and Ps obtained in Step 4;

[0056]

[0057]

[0058] Where CV and P are the processed values, and CV' and P' are the original values, CV max P max CV is the maximum value in the sample data. min P min It is the minimum value in the sample data;

[0059] Step Six: Use the samples obtained from the hyper-Latin sampling method in Step Three as the input to the support vector regression model, and the normalized CV and P obtained in Step Five as the output to train the support vector regression model:

[0060] 1) Creating the hyperplane equation

[0061] f(x) = ω·x + b = 0

[0062] Where ω is the slope vector of the hyperplane equation, b is the hyperplane intercept vector, x is the independent variable of the function, and f(x) is the dependent variable of the function.

[0063] 2) Establish hyperplane constraints

[0064]

[0065] y i ·(ω T x i +b)≥1,i=1,2,……m

[0066] Where, x i For the data of each sample point, y i Here, m represents the function value corresponding to each sample point, and m is the number of samples.

[0067] 3) Separate decision functions

[0068] f(x) = sign(ω) * ·x+b * )

[0069] 4) Selecting the Gaussian kernel function K(x,z) transforms the nonlinear problem into a linear problem:

[0070]

[0071] Where x is the coordinates of the point to be classified, z is the center point of the sample, and σ is the standard deviation of the sample.

[0072] Step 7: Define the objective function F, and use the differential optimization algorithm to obtain the correspondence E between the flow rate Q and λ, α, β, θ, h, and n at the minimum value of the objective function; where, τ is the weight.

[0073] 1) Set the population size S, iteration count N, mutation factor t, and crossover factor c to initialize the population and obtain the population G;

[0074] 2) Evaluate the initial population G, using the parameter values ​​from the population as inputs into the prediction model M. Depending on the application scenario, the weight ratio of CV to P is set to τ:1-τ, and this ratio is defined as the objective function F.

[0075]

[0076] Substitute each sample in population G into the prediction model M to obtain the corresponding individual fitness, i.e., the F value, thus obtaining the array R. The minimum value R' in R is taken as the optimal solution in the first generation, i.e., the point with the highest efficiency;

[0077] 3) Perform mutation and crossover between the parent and offspring generations, check the boundaries, calculate the fitness of individuals in the new population, continuously update R', and select the optimal solution;

[0078] 4) Compare the results of the parent generation with those of the child generation. If the result of the child generation is better than that of the parent generation, continue the evolution; otherwise, abandon the evolution and return to step 3). Continue this process until the same value is maintained after multiple evolutions. When the evolution stops, output the values ​​of each parameter at this point and record the optimization path as the optimization function E.

[0079] Step 8: Calculate λ and h using the following formula:

[0080] λ = 0.6λ R +0.2λ 0.8R +0.2λ 1.2R

[0081] h = 0.6h R +0.2h 0.8R +0.2h 1.2R

[0082] Where, λ 0.8R h 0.8R The flow rate is 0.8Q respectively. R Input the λ and h values ​​obtained from the correspondence E; λ 1.2R and h 1.2R The flow rate is 1.2Q. R Input the λ and h values ​​obtained from the correspondence E.

[0083] Step 9: Substitute λ and h obtained in Step 8 into the relationship E obtained in Step 7 to obtain the correspondence E' between flow rate Q and α, β, θ, and n under the minimum value of the objective function; at the mixer inlet flow meter, detect the flow rate Q during operation, input the flow rate Q into the corresponding relationship E', and obtain the values ​​of α, β, θ, and n corresponding to the minimum value of F under this flow rate, which is the point of highest mixing efficiency, thereby controlling the impeller, main pipe inlet guide plate, main pipe outlet guide plate and branch pipe guide plate to make the mixer's mixing efficiency the highest.

[0084] The effects of the method of the present invention are illustrated below with a specific embodiment. The range of mixer model parameters obtained in this embodiment is shown in Table 1.

[0085] Table 1 Range of Mixer Model Parameters

[0086] variable α / ° β / ° θ / ° λ v / m / s k / s / m η / % lower limit 15 18 20 0.36 0.54 0.23 0.82 upper limit 35 42 40 0.44 1.11 0.87 0.94

[0087] The fitness of the individual in the final optimal solution is as follows: Figure 3 With fixed parameters λ and h, the correspondence between quantity Q and α, β, θ, and n at the minimum of the objective function is obtained. When the flow rate changes, the device structure is actively adjusted to improve mixing efficiency. (Using 0.6Q) R For example, with λ = 2.29 and h = 622 mm fixed, the final system obtained α = 21.4°, β = 27.8°, θ = 20°, and n = 176 r / min. Compared with the prediction model, the error of CV value and P value is 1.02% compared with numerical calculation, which is consistent with the optimized model. Figure 4 The figure compares the outlet mixing uniformity before and after optimization at this flow rate. It can be seen from the figure that the optimized outlet mixing uniformity is significantly better than the unoptimized one. Therefore, this further illustrates that the control method of the present invention can promptly adjust the structure of the regulating device when the flow rate deviates, thereby ensuring that the mixer always remains at its highest efficiency point.

[0088] It will be understood by those skilled in the art that the above descriptions are merely preferred examples of the invention and are not intended to limit the invention. Although the invention has been described in detail with reference to the foregoing examples, those skilled in the art can still modify the technical solutions described in the foregoing examples or make equivalent substitutions for some of the technical features. All modifications and equivalent substitutions made within the spirit and principles of the invention should be included within the scope of protection of the invention.

Claims

1. A control method for a two-chamber pipe mixer based on support vector regression-difference optimization algorithm, characterized in that, The dual-chamber pipeline mixer includes a main inlet pipeline, branch pipelines, a dosing pipeline, an upper mixing chamber, a lower mixing chamber, a main outlet pipeline, an impeller, a main pipe inlet guide plate, a main pipe outlet guide plate, and a branch pipe guide plate. The main inlet pipe, lower mixing chamber, and main outlet pipe are connected in sequence; one end of the branch pipe is connected to the main inlet pipe, and the other end is connected to the upper mixing chamber; the dosing pipe is installed on the branch pipe; the upper mixing chamber and lower mixing chamber are connected, and the impeller is installed at the connection between the upper and lower mixing chambers; the main pipe inlet guide plate is installed inside the main inlet pipe, the main pipe outlet guide plate is installed inside the main outlet pipe, and the branch pipe guide plate is installed inside the branch pipe and located downstream of the dosing pipe connection; the main inlet pipe and the main outlet pipe have the same diameter; The control method includes the following steps: Step 1: Select the main pipe diameter ratio λ, the inclination angle of the main pipe inlet guide plate α, the inclination angle of the outlet pipe guide plate β, the inclination angle of the branch pipe guide plate θ, the impeller installation height h, and the impeller speed n as the mixer model parameter variables; Step two: based on the existing double-chamber pipeline mixer, the value range of λ, α, β, θ, n is given as 0.91~1.11λ R , 0.5~1.17α R , 0.6~1.4β R , 0.67~1.33θ R , 0.5~1.5n R ; wherein, λ R , α R , β R , θ R , n R are all parameters of the existing double-chamber pipeline mixer; Step 3: The given mixer flow rate Q is in the range of 0.6~1.4Q. R Q R The rated flow rate of the mixer is used; simultaneously, the relationship between Q, n, and h is used as a constraint, and λ, α, β, θ, n, and Q are used as variables, and sampling is performed using the super-Latin sampling method; ； ； ； Where D1 is the main pipe diameter, D2 is the branch pipe diameter, v is the inlet flow velocity of the main inlet pipe, and k is the proportionality coefficient, which represents the chemical reaction time per unit length between different raw water and the medicine added to the branch pipe. Step 4: For each set of variables obtained in Step 3, model the sample and perform simulation calculations to obtain the outlet mixing variation coefficient CV and impeller power consumption P corresponding to each set of variable samples; Step 5: Normalize the series of CVs and Ps obtained in Step 4; Step 6: Use the samples obtained by the hyper-Latin sampling method in Step 3 as the input to the support vector regression model, and use the normalized CV and P obtained in Step 5 as the output to train the support vector regression model. Step 7: Define the objective function F, and use the differential optimization algorithm to obtain the correspondence E between the flow rate Q and λ, α, β, θ, h, and n at the minimum value of the objective function; where, , As weight; Step 8: Calculate λ and h using the following formula: ； ； in, , The flow rate is 0.8Q respectively. R Input the λ and h values ​​obtained from the correspondence E; and The flow rate is 1.2Q. R Input the λ and h values ​​obtained from the correspondence E; Step 9: Substitute the λ and h calculated in Step 8 into the relationship E obtained in Step 7 to obtain the correspondence E' between the flow rate Q and α, β, θ, and n under the minimum value of the objective function; input the flow rate Q into the corresponding relationship E' to obtain the values ​​of α, β, θ, and n under that flow rate, thereby controlling the impeller, the main pipe inlet guide plate, the main pipe outlet guide plate, and the branch pipe guide plate to maximize the mixing efficiency of the mixer.

2. The control method for a dual-chamber pipe mixer based on support vector regression-difference optimization algorithm according to claim 1, characterized in that, Based on the values ​​of α, β, θ, and n obtained in step nine for each flow rate, modeling and simulation calculations are performed to obtain CV and P. At the same time, the values ​​of α, β, θ, and n obtained in step nine for each flow rate, as well as the corresponding λ and h, are input into the trained support vector regression model to obtain the model output CV and P. If the error between the CV and P predicted by the support vector regression model and the simulation calculation is greater than 3%, return to step seven to adjust the parameters of the initial population of the differential evolution algorithm.

3. The control method for a dual-chamber pipe mixer based on support vector regression-difference optimization algorithm according to claim 1, characterized in that, The impeller has 3 blades.

4. The control method for a dual-chamber pipe mixer based on support vector regression-difference optimization algorithm according to claim 1, characterized in that, The formulas for calculating the outlet mixing variation coefficient CV and the impeller power consumption P are as follows: ； ； Where: S is the standard deviation of export data. is the average value of the export data; M is the torque on the impeller.

Images