A global optimization calculation method for the optimal control parameters of clean air system

Abstract

The present invention relates to the field of clean room technology, and a method for optimizing and calculating the optimal control parameters of a clean air system, the steps of which are as follows: 1) setting the constraint conditions of the control variables of the clean air conditioning system, 2) setting the optimization target parameters, 3) calculating the Pareto optimal solution set, 4) screening the final trade-off solution, taking the fan operating power and the air valve angle as the optimization target, and screening out a group of solutions with small system impedance and low operating energy consumption from the Pareto optimal solution set by the entropy weight method as the final trade-off solution; 5) the final trade-off solution is sent to the control system as a control parameter. The control parameters calculated by the present invention can fully ensure that the room parameters of all clean areas meet the set target requirements, realize high-precision control of the room parameters of the clean areas, and take into account the control parameter combination of the control accuracy of the room parameters of the clean areas and the optimal energy consumption of the system operation, which not only meets the requirements of the clean air system for the control of the air supply volume and the pressure gradient, but also reduces the system operation energy consumption and saves energy.

Description

Technical Field

[0001] The invention relates to the technical field of clean rooms, and in particular to a calculation method for optimizing control parameters of a clean air system. Background Art

[0002] Clean air conditioning systems are important facilities for ensuring the environmental parameters of clean workshops in many industries such as biopharmaceuticals, and are widely used to maintain the cleanliness and stability of the production environment. These systems are mainly composed of fans, air ducts and air valves, and their operating energy consumption usually accounts for more than 45% of the total system energy consumption. The core function of the clean air system is to control the air volume balance and pressure gradient of the clean area by adjusting the fan frequency and air valve angle to ensure proper isolation and pollution control between different areas.

[0003] In the biopharmaceutical industry, clean rooms need to switch between working mode, disinfection mode, duty mode and other operating modes according to different production needs. These mode conversions require precise control of the room's air volume balance and maintenance of the pressure gradient to avoid cross contamination. Therefore, the effective switching of each operating mode is a new adjustment challenge for the clean air conditioning system. The key lies in how to reasonably set the fan frequency and air valve angle. At present, the mode adjustment of the clean air conditioning system usually lacks an efficient strategy, resulting in the system running in the working mode for a long time, resulting in a lot of energy waste.

[0004] In addition, the existing clean air conditioning system still relies on manual adjustment by operators in terms of operating parameter adjustment, which not only leads to low adjustment accuracy, but also causes instability in system control. Although there is an invention patent with application number "CN202211243702.3" that proposes a full-process simulation system for simulation calculation of control parameters of clean air conditioning system, which realizes remote calculation and automatic adjustment of control parameters, it lacks an effective optimization calculation method for the optimal control parameter combination.

[0005] Therefore, the present invention aims to provide a global optimization calculation method for the control parameters of a clean air-conditioning system. By finding the optimal parameter combination through an intelligent algorithm, not only high-precision control of the air volume and pressure gradient in the clean area can be achieved, but also the energy consumption of the system operation can be greatly reduced, thereby improving the overall system efficiency and economy. Summary of the invention

[0006] The invention discloses a global optimization calculation method for the optimal control parameters of a clean air system, the purpose of which is to achieve air volume balance and pressure difference gradient stability in a clean area and reduce system operation energy consumption.

[0007] In order to achieve the above-mentioned invention object, the technical solution of the present invention is as follows: a global optimization calculation method for the optimal control parameters of a clean air system comprises the following steps:

[0008] S1. Set the constraints of the clean air system control variables, including the fan frequency adjustment range and the adjustment range (upper limit and lower limit) of the air valve angle of each clean room.

[0009] f i,min ≤f i ≤f i,max

[0010] 0≤θ j ≤90

[0011] f i,min , the lower limit of frequency regulation of fan i, f i,max , the upper limit of the frequency adjustment of fan i; θ j , the angle of the air valve j, where f i,min =25Hz; f i,max =50Hz.

[0012] S2. Set the optimization target parameters, including the total system air volume value (total supply air volume, total fresh air volume, total exhaust air volume, etc.), branch air volume value (air supply volume of each clean area), and clean room pressure difference.

[0013] S3, calculate the Pareto optimal solution set, set the number of global optimization iterations K (K is an integer, and K ≥ 1), and call the simulation system to perform the calculation. The global optimization uses the particle swarm optimization algorithm, and the algorithm flow is as follows: Figure 1 shown.

[0014] S3.1. Initialize the particle swarm. Assume that the size of the particle swarm (the number of particles) is N, the number of control variables to be optimized is D, and N particles are randomly distributed into the D-dimensional space to obtain N groups of random solutions.

[0015] The position (solution) of particle i (i∈N, i=1, 2, 3…, n) in the kth iteration is recorded as k∈[1,K]; l d represents the lower limit of the dth dimension (control variable), u d Represents the upper limit of the d-th dimension.

[0016] The velocity of particle i at the kth calculation is recorded as v d,min represents the minimum value of the velocity of particle i in the dth dimension, v d,max Represents the maximum value of the movement speed of particle i in the dth dimension.

[0017] S3.2, calculate the fitness value of the particle. The purpose of calculating the fitness value is to evaluate the current position (solution) of the particle. The evaluation of the particle fitness is based on two indicators: the clean area pressure difference and the clean area air supply volume.

[0018] Taking particle i as an example, first calculate the deviation between the calculated value and the set value of the pressure difference and air supply volume in each area, and obtain two data sets:

[0019]

[0020]

[0021] Q m,a , the calculated value of the air supply volume in the mth clean area;

[0022] Q m,d , the optimal target value of the air supply volume in the mth clean area;

[0023] P m,a , the calculated value of the pressure difference of the mth clean area;

[0024] P m,d , the optimal target value of the pressure difference of the mth clean area;

[0025] q m , the deviation value of the air supply volume in the mth clean area;

[0026] p m , the deviation value of the pressure difference in the mth clean area.

[0027] Then, the standard deviation of the clean area air supply volume deviation value set and the standard deviation of the clean area pressure difference deviation value set are calculated respectively:

[0028]

[0029]

[0030] Then the fitness value of particle i at the current position (solution) is:

[0031] F(i)=0.5*S q +0.5*S p

[0032] Among them, the standard deviation Sq of the clean area air supply volume deviation value set is given a weight of 0.5, and the standard deviation Sp of the clean area pressure difference deviation value set is given a weight of 0.5. It is believed that the control of branch air supply volume is as important as the room pressure difference control.

[0033] S3.3, calculate the individual historical optimal position. Compare the fitness values ​​calculated for each particle in each iteration, and the position with the smallest fitness value (F(i)) is the historical optimal position P of particle i. best .

[0034] S3.4. Update the Pareto optimal solution set. The Pareto optimal solution set is n groups of Pareto optimal solutions obtained by screening the positions (solutions) of particles through the Pareto method. The Pareto optimal solution can ensure that the air supply volume and room pressure difference of at least one clean area are optimal (close to or reach the set optimization target parameters), and the air supply volume and room pressure difference of other clean areas do not deviate significantly from the set optimization target parameters.

[0035] During the first calculation, the fitness values ​​of N particles are compared, and n groups of solutions (n ​​particle positions, n≤N) with small fitness values ​​are selected to form the Pareto optimal solution set.

[0036] In each subsequent calculation, the new solutions of the N particles are compared with the n groups of solutions in the Pareto optimal solution set. The ones with smaller fitness values ​​enter the Pareto optimal solution set, and the rest are eliminated, always keeping the number of solutions in the Pareto optimal solution set at n groups.

[0037] S3.5. Calculate the historical optimal position of the population. Each time a solution is randomly selected from the updated Pareto optimal solution set as the historical optimal position G of the population. best .

[0038] S3.6. Update the speed and position of the particle. The speed of a particle is a position vector. The updated speed is actually the direction and distance that the particle moves when it is calculated next time. Assume that the speed of particle i in dimension d during the kth calculation is Then the speed in dimension d during the k+1th calculation is:

[0039]

[0040] ω, inertia weight, reflects the particle's motion habits and the particle's trust in its previous motion state;

[0041] c1, individual learning factor, i.e., the acceleration weight of the particle converging to its own optimal solution;

[0042] c2, group learning factor, i.e., the acceleration weight of particles converging to the group optimal solution;

[0043] r1, r2, random numbers between (0, 1).

[0044] Assume that the position of particle i calculated for the kth time in dimension d is Then the position calculated for the k+1th time is:

[0045]

[0046] S3.7, determine whether the termination condition is met. Determine whether the number of iterative calculations k is greater than the set number of global optimization iterative calculations K. If it is met, terminate the iterative calculation and proceed to the next step.

[0047] Through the above calculations, we have obtained a solution set that meets the branch air supply volume and room pressure difference as the optimization goals, and takes into account the control accuracy of the wind system operation parameters. The next step is to further consider the energy-saving characteristics of the wind system operation, and further screen the solution set with lower fan frequency and smaller system impedance (smaller air valve angle-larger opening) as the optimization goals.

[0048] S4. Filter the final trade-off solution. After K iterative calculations, the n groups of solutions in the Pareto optimal solution set are n groups of trade-off solutions filtered from the N×K groups of solutions, which can make the air supply volume and room pressure difference of all clean areas reach or approach the set optimization target parameters, ensuring the parameter control accuracy of the clean areas.

[0049] With small system impedance and low operating energy consumption as the optimization goal, the n groups of solutions of the Pareto optimal solution set are comprehensively evaluated by the entropy weight method, and one group of solutions with small clean air system impedance and low operating energy consumption is selected as the final trade-off solution, which is sent to the automatic control system as the control parameter to control the on-site fans and air valves.

[0050] S4.1. Construct the evaluation model matrix. Calculate the sum of the operating power and the average angle of the terminal air valve of n groups of solutions respectively:

[0051]

[0052] W, the sum of the operating powers of the fans in the system;

[0053] N, the number of fans in the system;

[0054] f i , the calculated operating frequency of fan i.

[0055]

[0056] P, the average angle of the terminal damper;

[0057] M, the number of terminal air supply valves, return air valves, and exhaust valves;

[0058] P j , the angle of the air valve j is calculated.

[0059] The two evaluation indexes of the calculated n groups of solutions are shown in the following table.

[0060] Pareto optimal solution Total fan power Average angle of terminal air valve Solution 1 <![CDATA[W1]]> <![CDATA[P1]]> Solution 2 <![CDATA[W2]]> <![CDATA[P2]]> Solution 3 <![CDATA[W3]]> <![CDATA[P3]]> … … … The nth group of solutions <![CDATA[W n ]]> <![CDATA[P n ]]>

[0061] S4.2, data standardization processing. The fan operating energy consumption and the average angle of the terminal air valve are both as small as possible, so they are both negative indicators. The standardization processing method is as follows:

[0062]

[0063] X ij , the jth index of the i-th Pareto optimal solution;

[0064] x ij , the jth indicator of the i-th Pareto optimal solution after data standardization;

[0065] i = 1, 2, ..., n;

[0066] j = 1, 2;

[0067] S4.3. Calculate the proportion of the i-th solution to the j-th indicator:

[0068]

[0069] S4.4. Calculate the entropy value of the j-th indicator:

[0070]

[0071] in, Meet e j ≥0.

[0072] S4.5. Calculate the difference of the jth indicator:

[0073] d j =1-e j

[0074] S4.6. Calculate the weight of the jth indicator:

[0075]

[0076] S4.7. Calculate the scores of the n Pareto optimal solutions respectively and select the group with the highest score as the final trade-off solution.

[0077]

[0078] S5. The final trade-off solution is sent to the control system as the only control parameter set.

[0079] Compared with the prior art, the present invention has the following beneficial effects:

[0080] 1) The present invention performs global optimization calculation on the control parameters of the clean air system based on the particle swarm algorithm, and obtains the Pareto optimal solution set through the Pareto method, so that the calculated control parameters can fully ensure that the room parameters of all clean areas meet the set target requirements, and realize high-precision control of the room parameters of the clean area.

[0081] 2) Based on the control parameters to achieve high-precision control of clean area parameters, the present invention further screens the Pareto optimal solution set with the goal of low system impedance and low operating energy consumption, and obtains a control parameter combination that takes into account both the control accuracy of clean area room parameters and the optimal system operating energy consumption. This not only meets the clean air system's requirements for air supply volume and pressure difference gradient control, but also reduces system operating energy consumption and saves energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 The figure is a flow chart of a particle swarm algorithm according to a specific embodiment of the present invention. DETAILED DESCRIPTION

[0083] The technical solution of the present invention is further explained below in conjunction with specific embodiments.

[0084] A global optimization calculation method for optimal control parameters of a clean air system comprises the following steps:

[0085] S1. Set the constraints of the control variables of the clean air conditioning system. Set the upper and lower limits of the operating frequency of all fans in the system, such as the supply air unit, fresh air unit (if any), and exhaust air unit (if any). Set the upper and lower limits of the angles of all adjustable valves in the system, such as the fresh air valve (if any), supply air valve, return air valve, and exhaust air valve.

[0086] S2. Set the optimal target parameters. Set the optimal target values ​​of the main air duct parameters in the system, such as the total supply air volume, total fresh air volume, and total exhaust air volume, and set the optimal target values ​​of the supply air volume and room pressure difference of each clean room.

[0087] S3. Calculate the Pareto optimal solution set. Set the number of global optimization iterations K, such as 100, and then call the simulation system for calculation.

[0088] S3.1. Initialize the particle swarm. The size of the particle swarm (the number of particles) is set to N, for example, 60; the number of control variables to be optimized is D, which is the sum of the number of fans and the number of adjustable valves in S1. The 60 particles are randomly distributed in the D-dimensional space, and the first calculation is performed to obtain 60 solutions.

[0089] The position (solution) of particle i (i∈N, i=1, 2, 3…, 60) at the kth calculation is recorded as k∈[1,100]; l d represents the lower limit of the dth dimension (control variable), u d Represents the upper limit of the d-th dimension.

[0090] The velocity of particle i in the kth calculation is recorded as vd,min represents the minimum value of the velocity of particle i in the dth dimension, v d,max Represents the maximum value of the movement speed of particle i in the dth dimension; for example, the minimum value is set to 0 and the maximum value is set to 5.

[0091] S3.2, calculate the fitness value of the particle. Taking particle i as an example, divide the calculated air supply volume and room pressure difference of each clean room by the set optimization target value, calculate the deviation of the clean room air supply volume and room pressure difference, and obtain two mathematical sets:

[0092]

[0093]

[0094] Q m,a , the calculated value of the air supply volume in the mth clean area;

[0095] Q m,d , the optimal target value of the air supply volume in the mth clean area;

[0096] P m,a , the calculated value of the pressure difference of the mth clean area;

[0097] P m,d , the optimal target value of the pressure difference in the mth clean area;

[0098] q m , the deviation value of the air supply volume in the mth clean area;

[0099] p m , the deviation value of the pressure difference in the mth clean area.

[0100] Then, the standard deviation of the clean area air supply volume deviation value set and the standard deviation of the clean area pressure difference deviation value set are calculated respectively:

[0101]

[0102]

[0103] Then the fitness value of particle i at the current position (solution) is:

[0104] F(i)=0.5*S q +0.5*S p

[0105] S3.3. Calculate the individual historical optimal position. Compare the fitness values ​​calculated for each particle in each iteration. The position with the smallest fitness value (F(i)) is the historical optimal position P of particle i. best .

[0106] In the first calculation, the result is the historical optimal position P of each particle. best , in subsequent iterative calculations, the calculation results of each time are consistent with P best Compare, the ones with large fitness values ​​are eliminated, and the ones with small fitness values ​​become the new P best .

[0107] S3.4. Update the Pareto optimal solution set. The 60 groups of solutions calculated by the 60 particles are screened by the Pareto method, and n groups (for example, 6 groups) of Pareto optimal solutions are selected to form the Pareto optimal solution set. The Pareto optimal solution can ensure that the air supply volume and room pressure difference of at least one clean area are optimal (close to or reach the set optimization target parameters), and the air supply volume and room pressure difference of other clean areas do not deviate significantly from the set optimization target parameters.

[0108] During the first calculation, the fitness values ​​of the 60 particles are compared, and the 6 groups of solutions (6 particle positions) with the smallest fitness values ​​are selected to form the Pareto optimal solution set.

[0109] In each subsequent iterative calculation, the new solutions calculated by the 60 particles are compared with the 6 groups of solutions in the Pareto optimal solution set. The ones with smaller fitness values ​​enter the Pareto optimal solution set, and the rest are eliminated, always maintaining the number of solutions in the Pareto optimal solution set at 6 groups.

[0110] S3.5. Calculate the historical optimal position of the population. In each iterative calculation, a set of solutions is randomly selected from the updated Pareto optimal solution set as the historical optimal position G of the population. best .

[0111] S3.6. Update the particle's velocity and position. The velocity of a particle is a position vector. The updated velocity is actually the direction and distance that the particle moves in the next iteration. Suppose the velocity of particle i in dimension d during the kth calculation is Then the speed in dimension d during the k+1th calculation is:

[0112]

[0113] ω, inertia weight, reflects the particle's motion habits and the particle's trust in its previous motion state, and is set to 0.9;

[0114] c1, individual learning factor, i.e., the acceleration weight of the particle converging to its own optimal solution, is set to 1.6;

[0115] c2, group learning factor, i.e., the acceleration weight of particles converging to the group optimal solution, is set to 1.8;

[0116] r1, r2, random numbers between (0, 1).

[0117] Assume that the position of particle i calculated for the kth time in dimension d is Then the position calculated for the k+1th time is:

[0118]

[0119] S3.7, determine whether the termination condition is met. Determine whether the number of iterative calculations k is greater than the set number of global optimization iterative calculations 100. If it is met, terminate the iterative calculation and proceed to the next step.

[0120] 4. Screen the final trade-off solution. After 100 iterative calculations, the 6 groups of solutions in the Pareto optimal solution set are 6 groups of trade-off solutions screened from 6000 (60×100) groups of solutions, which can make the clean area air supply volume and room pressure difference reach or approach the set optimization target parameters as a whole, ensuring the parameter control accuracy of the clean area.

[0121] With the goal of low system impedance and low operating energy consumption, the six groups of solutions in the Pareto optimal solution set were comprehensively evaluated through the entropy weight method, and a group of solutions with low clean air system impedance and low operating energy consumption were selected as the final trade-off solution and sent to the automatic control system as the control parameters to control the on-site fans and air valves.

[0122] S4.1. Construct the evaluation model matrix. Calculate the operating power and average angle of the terminal air valve of the six groups of solutions respectively:

[0123]

[0124] W, the sum of the operating powers of the fans in the system;

[0125] N, the number of fans in the system;

[0126] f i , the calculated operating frequency of fan i.

[0127]

[0128] P, the average angle of the terminal damper;

[0129] M, the number of terminal air supply valves, return air valves, and exhaust valves;

[0130] P j , the angle of the air valve j is calculated.

[0131] The two evaluation indexes of the calculated n groups of solutions are shown in the following table.

[0132] Pareto optimal solution Total fan power Average angle of terminal air valve Solution 1 <![CDATA[W1]]> <![CDATA[P1]]> Solution 2 <![CDATA[W2]]> <![CDATA[P2]]> Solution 3 <![CDATA[W3]]> <![CDATA[P3]]> … … … Solution 6 <![CDATA[W6]]> <![CDATA[P6]]>

[0133] S4.2, data standardization processing. The fan operating energy consumption and the average angle of the terminal air valve are both as small as possible, so they are both negative indicators. The standardization processing method is as follows:

[0134]

[0135] X ij , the jth index of the i-th Pareto optimal solution;

[0136] x ij , the jth indicator of the i-th Pareto optimal solution after data standardization;

[0137] i = 1, 2, ..., 6;

[0138] j = 1, 2;

[0139] S4.3. Calculate the proportion of the i-th solution to the j-th indicator:

[0140]

[0141] S4.4. Calculate the entropy value of the j-th indicator:

[0142]

[0143] in, Meet e j ≥0.

[0144] S4.5. Calculate the difference of the jth indicator:

[0145] d j =1-e j

[0146] S4.6. Calculate the weight of the jth indicator:

[0147]

[0148] S4.7. Calculate the scores of the n Pareto optimal solutions respectively, and select the group with the highest score as the final trade-off solution.

[0149]

[0150] S5. The final trade-off solution is sent to the control system as the only control parameter set.

[0151] To sum up, the control parameters calculated by the technical solution adopted in the present invention can fully ensure that the room parameters of all clean areas meet the set target requirements, realize high-precision control of the room parameters of the clean areas, and take into account the control accuracy of the room parameters in the clean areas and the optimal control parameter combination of the system operation energy consumption, which not only meets the requirements of the clean air system for the control of the air supply volume and the pressure difference gradient, but also reduces the system operation energy consumption and saves energy.

Claims

1. A global optimization calculation method for the optimal control parameters of a clean air system, characterized by: The steps include: S1. Set the constraints of the clean air system control variables; S2, setting optimization target parameters; S3, calculate the Pareto optimal solution set; S4, screening the final trade-off solution; S5. The final trade-off solution is sent to the control system as the only control parameter group. The constraints of the clean air system control variables are set as follows: Fan frequency adjustment range: ; Adjustment range of air valve angles in each clean room: ; , the lower limit of the frequency adjustment of fan i, , the upper limit of the frequency adjustment of fan i; , the angle of air valve j, the optimization target parameters include the total air volume value of the system, the total air volume value includes the total supply air volume, the total fresh air volume, the total exhaust air volume, the branch air volume value, the clean room pressure difference, the specific steps of calculating the Pareto optimal solution set are as follows: Set the number of global optimization iterations K (K is an integer, and K≥1), and use the particle swarm optimization algorithm for global optimization: S3.

1. Initialize the particle swarm. Assume that the size of the particle swarm (the number of particles) is N, the number of control variables to be optimized is D, and randomly distribute the N particles into the D-dimensional space to obtain N groups of random solutions. The position of particle i (i∈N, i=1, 2, 3…, n) calculated at the kth iteration is recorded as , k∈[1,K]; , l d represents the lower limit of the dth dimension, u d Represents the upper limit of the d-th dimension; The velocity of particle i at the kth calculation is recorded as ; , represents the minimum value of the velocity of particle i in the dth dimension, represents the maximum value of the velocity of particle i in the dth dimension; S3.

2. Calculate the particle fitness value. The purpose of calculating the particle fitness value is to evaluate the current position of the particle. The evaluation of particle fitness is based on two indicators: the pressure difference in the clean area and the air supply volume in the clean area. Taking particle i as an example, first calculate the deviation between the calculated value and the set value of the pressure difference and air supply volume in each area, and obtain two data sets: ; ; , the calculated value of the air supply volume of the mth clean area; , the optimal target value of the air supply volume in the mth clean area; , the calculated value of the pressure difference of the mth clean area; , the optimal target value of the pressure difference of the mth clean area; , the deviation value of the air supply volume in the mth clean area; , the deviation value of the pressure difference in the mth clean area; Then, the standard deviation of the clean area air supply volume deviation value set and the standard deviation of the clean area pressure difference deviation value set are calculated respectively: ; ; Then the fitness value of particle i at the current position is: ; Among them, the standard deviation Sq of the clean area air supply volume deviation value set is given a weight of 0.5, and the standard deviation Sp of the clean area pressure difference deviation value set is given a weight of 0.

5. It is believed that the control of branch air supply volume is as important as the room pressure difference control; S3.3, calculate the individual historical optimal position, compare the fitness value of each particle calculated in each iteration, and the fitness value The minimum position is the historical optimal position of particle i ; S3.

4. Update the Pareto optimal solution set. The Pareto optimal solution set is n groups of Pareto optimal solutions obtained by screening the positions of particles through the Pareto method. In the first calculation, the fitness values ​​of N particles are compared, and n groups of solutions with small fitness values ​​(n particle positions, n≤N) are selected to form the Pareto optimal solution set. In each subsequent calculation, the new solutions of N particles are compared with the n groups of solutions in the Pareto optimal solution set. The solutions with small fitness values ​​enter the Pareto optimal solution set, and the rest are eliminated, so that the number of solutions in the Pareto optimal solution set is always kept at n groups. S3.

5. Calculate the historical optimal position of the population. Each time you calculate, randomly select a set of solutions from the updated Pareto optimal solution set as the historical optimal position of the population. ; S3.

6. Update the speed and position of the particle. The speed of the particle is a position vector. The updated speed is actually the direction and distance the particle moves when the particle is calculated next time. Assume that the speed of particle i in dimension d during the kth calculation is , then the speed in dimension d during the k+1th calculation is: ; w, inertia weight, reflects the particle's motion habits and the particle's trust in its previous motion state; c1, individual learning factor, i.e., the acceleration weight of the particle converging to its own optimal solution; c2, group learning factor, i.e., the acceleration weight of particles converging to the group optimal solution; r1, r2, random numbers between (0, 1); Assume that the position of particle i calculated for the kth time in dimension d is , then the position calculated for the k+1th time is: ; S3.

7. Determine whether the termination condition is met, and determine whether the number of iterative calculations k is greater than the set number of global optimization iterative calculations K. If so, terminate the iterative calculation and proceed to the next step.

2. A global optimization calculation method for optimal control parameters of a clean air system according to claim 1, characterized in that: After K iterative calculations, the n groups of solutions in the Pareto optimal solution set are n groups of trade-off solutions screened from N×K groups of solutions. With small system impedance and low operating energy consumption as the optimization goals, the n groups of solutions in the Pareto optimal solution set are comprehensively evaluated through the entropy weight method, and 1 group of solutions with small clean air system impedance and low operating energy consumption is screened out as the final trade-off solution and sent to the automatic control system as a control parameter to control the on-site fans and air valves.

3. A global optimization calculation method for optimal control parameters of a clean air system according to claim 2, characterized in that: The screening steps for the final trade-off solution are as follows: S4.

1. Construct the evaluation model matrix and calculate the sum of the operating power and the average angle of the terminal air valve of n groups of solutions respectively: ; W is the sum of the operating power of the fans in the system; N is the number of fans in the system; , the calculated operating frequency of fan i; ; P, the average angle of the terminal damper; M, the number of terminal air supply valves, return air valves, and exhaust air valves; P j , the calculated angle of air valve j; S4.2, data standardization processing, the fan operating energy consumption and the average angle of the terminal air valve are both as small as possible, so they are negative indicators. The standardization processing method is as follows: ; , the jth index of the i-th Pareto optimal solution; , the jth index of the i-th Pareto optimal solution after data standardization; i=1, 2…, n; j=1, 2; S4.

3. Calculate the proportion of the i-th solution to the j-th indicator: ; S4.

4. Calculate the entropy value of the j-th indicator: ; in, ,satisfy , S4.

5. Calculate the difference of the jth indicator: ; S4.

6. Calculate the weight of the jth indicator: ; S4.

7. Calculate the scores of n Pareto optimal solutions respectively, and select the highest score as the final trade-off solution; 。

Images