Cleanroom energy-saving pressure control method and system based on dynamic pipe network and model predictive control

Abstract

The present invention relates to the field of clean environment control technology, and in particular to a clean room energy-saving pressure control method and system using a dynamic pipe network and model predictive control, comprising: calculating the current wind pressure; obtaining the current pipe network impedance based on the pipe network impedance curve; calculating the current valve impedance based on the valve opening; calculating the current total pressure drop in the pipe network; constructing a fan dynamic model based on the fan similarity law to calculate the air volume at the next moment; constructing a clean room pressure difference dynamic model to calculate the clean room pressure difference at the next moment; constructing a state space model and the objective function of a model predictive controller; when the clean room pressure difference is unstable, predicting the change trend of the clean room pressure difference in the next time period, and then using the model predictive controller to reversely solve the optimal fan frequency and valve opening of the discretized state space model to achieve the lowest pipe network impedance and obtain the optimal fan frequency. The present invention significantly improves the control accuracy and robustness of the clean room pressure difference while reducing energy consumption.

Description

Technical Field

[0001] The present invention relates to the technical field of clean environment control, and in particular to a clean room energy-saving and pressure control method and system based on a dynamic pipe network and model predictive control. Background Art

[0002] As an important infrastructure supporting high-tech industries, the clean industry is entering a golden period of rapid development. How to improve the energy efficiency of air-conditioning systems while maintaining high cleanliness has become a difficult problem that the clean industry urgently needs to solve.

[0003] Cleanroom differential pressure refers to the pressure difference between a cleanroom and adjacent areas or the external environment. It is a crucial parameter for cleanroom environmental control, playing a key role in maintaining cleanroom cleanliness and preventing the ingress of external contaminants. Existing cleanroom differential pressure control strategies include: multi-fan variable frequency technology, coordinated demand-controlled ventilation strategies, variable residual air volume tracking, zoned air conditioning with independent control, and heat recovery systems combined with variable air volume control.

[0004] Multi-fan joint frequency conversion technology builds a pressure differential control system model based on system air volume balance and adds variable frequency control for fresh air and exhaust air. However, coordinated control of multiple fans is difficult, as fan states influence each other. Any abnormality can easily lead to pressure imbalance and oscillation, making it difficult to stabilize the pressure differential.

[0005] Coordinated demand-controlled ventilation strategies coordinate outdoor and supply ventilation systems to achieve energy savings. However, this strategy can lead to over-ventilation to maintain a pressure differential, increasing energy consumption. If the control strategy is not optimal, high energy consumption will continue to operate, reducing the overall energy efficiency of the system.

[0006] The variable residual air volume tracking method uses a data relationship model to maintain cleanroom air leakage and pressure gradients under variable air volume conditions. However, this method lacks direct monitoring of pressure fluctuations and cannot maintain a stable pressure differential in the face of external disturbances such as door openings and equipment failures.

[0007] Zoned air conditioning and independent control methods use zones based on function and pollution level, with independent air conditioning and pressure differential control systems. However, this significantly increases the complexity of system design, requiring consideration of multiple factors, and increases the difficulty and cost of equipment maintenance.

[0008] Heat recovery systems, combined with variable air volume control, use heat recovery devices to recover heat and cooling from exhaust air to precondition fresh air. However, this approach increases system complexity and initial costs, relies on the coordination of multiple variables, has slow dynamic response, is difficult to maintain, and has limited adaptability in certain scenarios.

[0009] Although the above method can achieve basic pressure differential stability, it still has core problems such as insufficient control accuracy, limited energy consumption optimization, high system complexity, weak dynamic response capability and poor adaptability. It is difficult to meet the stringent requirements of high-tech fields such as biomedicine and semiconductors for pressure differential stability and energy-saving efficiency. Summary of the Invention

[0010] Therefore, the technical problem to be solved by the present invention is to overcome the problems of insufficient control accuracy, limited energy consumption optimization, high system complexity, weak dynamic response capability and poor adaptability in the prior art.

[0011] To solve the above technical problems, the present invention provides a clean room energy-saving and pressure control method based on a dynamic pipe network and model predictive control, comprising:

[0012] The current fan speed is obtained based on the current fan frequency, and the current wind pressure is calculated based on the current fan speed and current air volume;

[0013] Through numerical simulation, the pipe network impedance under different air volumes is obtained, and the pipe network impedance curve is constructed. Based on the pipe network impedance curve, the current pipe network impedance is obtained from the current air volume. The current valve impedance is calculated based on the valve opening. The current total pressure drop of the pipe network is calculated based on the current valve impedance, the current pipe network impedance, and the air handling equipment impedance.

[0014] A fan dynamic model is constructed based on the fan similarity law to calculate the next moment's air volume based on the current air volume, the current total pressure drop in the pipe network, and the current wind pressure. The air volume of the exhaust fan is the exhaust volume, and the air volume of the supply fan is the supply volume.

[0015] Calculate the valve pressure drop based on the valve opening and the current valve impedance, and then calculate the valve output flow;

[0016] Construct a dynamic model of clean room pressure difference, which is used to calculate the clean room pressure difference at the next moment based on the current clean room pressure difference, the exhaust volume and supply volume at the next moment;

[0017] A discretized state-space model was constructed, with fan frequency and valve opening as control inputs, exhaust air volume, supply air volume, and valve output flow as system state variables, and cleanroom differential pressure as output. The objective function of the model predictive controller was constructed with the goal of maintaining strong cleanroom differential pressure stability, minimizing the differential pressure tracking error and the valve opening penalty term.

[0018] The clean room pressure difference is monitored in real time. When the clean room pressure difference is unstable, the clean room pressure difference dynamic model is used to predict the change trend of the clean room pressure difference in the next time period. The model predictive controller is then used to reversely solve the optimal valve opening of the discretized state space model to achieve the lowest pipe network impedance and obtain the optimal fan frequency; the fan frequency and valve opening in the next time period are adjusted based on the optimal valve opening and fan frequency.

[0019] Preferably, the current wind pressure is calculated based on the current fan speed and the current wind volume, using the formula:

[0020]

[0021] Where FTP1 is the current wind pressure, FTP0 is the rated wind pressure, V1 and V0 are the current fan speed and the rated fan speed, respectively, Q1 and Q0 are the current air volume and the rated air volume, respectively, and a0, a1, a2, and a3 are fitting coefficients.

[0022] Preferably, the current valve impedance is calculated according to the valve opening, using the formula:

[0023]

[0024] Among them, θ represents the valve angle corresponding to the valve opening, S v (θ) represents the valve impedance, A ' represents the valve flow area, ρ represents the air density, a, b and c represent the fitting coefficients;

[0025] The total pressure drop of the current pipe network is calculated based on the current valve impedance, the current pipe network impedance, and the air handling equipment impedance. The formula is:

[0026]

[0027] Where ΔP lose Indicates the total pressure drop of the pipe network, S Duct represents the pipe network impedance, S v (θ) represents the valve impedance, S air represents the impedance of the clean room air handling equipment, Q1 represents the air volume;

[0028] When the valve opening is controlled at above 85%, the total pressure drop in the pipeline network is minimized.

[0029] Preferably, the formula of the wind turbine dynamic model is:

[0030]

[0031] Among them, Q1 is the current air volume, is the wind volume at the next moment, τ qs is the air volume response time constant, k qs is the fan gain coefficient, fs is the fan frequency, g(fs) is the function relationship between fan speed and fan frequency, FTP1 is the current wind pressure, ΔP all is the total pressure difference;

[0032] Total pressure difference ΔP all =ΔP1+ΔP lose , where ΔP1 is the clean room pressure, ΔP loseis the total pressure drop in the pipe network.

[0033] Preferably, the valve opening in the next time period is adjusted using the valve dynamic model, which is expressed as:

[0034]

[0035] Among them, θ i is the current valve opening, is the valve opening at the next moment, u θi is the valve opening control signal, τ θi is the valve response time constant, K θi is the valve gain coefficient.

[0036] Preferably, calculating the valve pressure drop according to the valve opening and the current valve impedance, and then calculating the valve output flow, includes:

[0037] The valve pressure drop is calculated based on the valve opening and the current valve impedance. The formula is:

[0038]

[0039] Where ΔP valve is the valve pressure drop, S v (θ) represents the valve impedance, Q 11 Indicates the air volume passing through the valve;

[0040] The calculation formula for the valve output flow is:

[0041]

[0042] Among them, Q i is the current valve output flow, is the valve output flow at the next moment, τ i is the valve flow response time constant, K i is the valve flow gain coefficient, θ i is the current valve opening.

[0043] Preferably, the formula of the clean room pressure difference dynamic model is:

[0044]

[0045] Among them, P e is the current pressure difference in the clean room, is the pressure difference of the clean room at the next moment, C e is the clean room gas volume constant, q in,e and q out,e are the supply and exhaust air volumes of clean room area e, k le is the clean room leakage constant.

[0046] Preferably, a discretized state space model is constructed, and the formula is expressed as:

[0047] x(k+1)=Ax(k)+Bu(k)

[0048] y(k)=Cx(k)+Du(k)

[0049] Where x(k) is the system state vector at time k, x(k+1) is the system state vector at time k+1, u(k) is the control input vector at time k, y(k) is the output vector at time k, A is the system matrix, B is the input matrix, C is the output matrix, and D is the direct transfer matrix.

[0050] The objective function of the model predictive controller is constructed and expressed as follows:

[0051]

[0052] st:u1(k)≥85%+ξ(k),ξ(k)≥0

[0053] st:ΔP e,min -∈(k)≤ΔP e (k)≤ΔP e,max +∈(k),∈(k)≥0

[0054] Among them, J1 is the first objective function, J2 is the second objective function, and N p To predict the time domain length, ΔP e (k) is the clean room pressure difference at time k, P e,ref is the preset value of the clean room pressure difference, u1(k) is the valve opening at time k, ∈(k) is the slack variable at time k, ξ(k) is the soft constraint of the valve opening at time k, ΔP e,min and ΔP e,max are the minimum and maximum values of the clean room pressure difference, ω1, ω2, ω ∈ , w2 and w ξ are all weight coefficients.

[0055] Preferably, when it is detected that the door of the clean room is open, the fan frequency is increased to the preset frequency, and the dynamic model of the clean room pressure difference is used to predict the changing trend of the clean room pressure difference in the next time period. Then, the state space model is used to reversely solve the optimal valve opening and adjust the valve opening in the next time period.

[0056] The present invention also provides a clean room energy-saving and pressure-control system with a dynamic pipe network and model predictive control, comprising:

[0057] Sensor module, including:

[0058] Pressure differential sensors are installed between the clean room and adjacent areas, and at the front and back ends of the pipe network and fan to detect the clean room pressure differential and pipe network impedance in real time.

[0059] Air volume sensors are installed at the rear end of the air supply valve, the front end of the exhaust valve, and the front end of the fan to obtain the supply and exhaust air volumes;

[0060] Valve opening sensor, installed on the air supply valve and exhaust valve, is used to obtain the valve opening;

[0061] The control module is connected to the sensor module for communication and is used to obtain the current fan speed according to the current fan frequency, and then calculate the current wind pressure according to the current fan speed and the current air volume; obtain the pipe network impedance under different air volumes through numerical simulation, and construct the pipe network impedance curve; obtain the current pipe network impedance from the current air volume according to the pipe network impedance curve; calculate the current valve impedance according to the valve opening; calculate the current pipe network total pressure drop according to the current valve impedance, the current pipe network impedance and the air handling equipment impedance; construct a fan dynamic model based on the fan similarity law, and calculate the air volume at the next moment according to the current air volume, the current pipe network total pressure drop and the current wind pressure; among them, the air volume of the exhaust fan is the exhaust volume, and the air volume of the supply fan is the supply volume; calculate the valve pressure drop according to the valve opening and the current valve impedance, and then calculate the valve output flow Quantity; construct a dynamic model of clean room pressure difference, which is used to calculate the clean room pressure difference at the next moment based on the current clean room pressure difference, the exhaust volume and the supply volume at the next moment; construct a discretized state space model, in which the fan frequency and valve opening are control inputs, the exhaust volume, the supply volume and the valve output flow are system state variables, and the clean room pressure difference is output; with the strong stability of the clean room pressure difference as the goal, construct the objective function of the model predictive controller, which is used to minimize the pressure difference tracking error and the valve opening penalty term; monitor the clean room pressure difference in real time. When the clean room pressure difference is unstable, use the clean room pressure difference dynamic model to predict the change trend of the clean room pressure difference in the next time period, and then use the model predictive controller to reversely solve the optimal valve opening of the discretized state space model to achieve the lowest pipe network impedance and obtain the optimal fan frequency;

[0062] The execution module is used to adjust the fan frequency and valve opening in the next time period with the optimal valve opening and fan frequency.

[0063] The above technical solution of the present invention has the following beneficial effects compared with the prior art:

[0064] The present invention describes a cleanroom energy-saving and pressure-control method using a dynamic pipe network and model predictive control. Taking into account pipe network impedance and valve impedance, the method dynamically calculates pipe network impedance by constructing a pipe network impedance curve and calculates valve impedance based on valve opening, thereby calculating the total pipe network pressure drop. A fan dynamic model is then constructed based on the fan similarity law. The air volume for the next time period is optimized based on the air volume and the total pipe network pressure drop. The cleanroom pressure differential for the next time period is dynamically predicted based on the optimized air volume. Finally, a state-space model is constructed that includes the fan, pipe network, valve, and cleanroom pressure differential. A model predictive controller is used to reversely solve the discretized state-space model for the optimal valve opening to minimize pipe network impedance and obtain the optimal fan frequency. The state-space model achieves multivariable coordinated control by coupling multiple variables, overcoming the trade-off between pressure differential stability and energy consumption in existing technologies. The present invention combines pipe network impedance and valve opening for joint optimization, enabling early dynamic response based on the predicted cleanroom pressure differential. This significantly improves the control accuracy and robustness of the cleanroom pressure differential while reducing energy consumption, making it suitable for demanding applications such as semiconductors. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments of the present invention in conjunction with the accompanying drawings, wherein:

[0066] Figure 1 This is a flow chart of a clean room energy-saving and pressure-control method using a dynamic pipe network and model predictive control according to the present invention;

[0067] Figure 2 This is a main cycle flow chart of the clean room energy-saving and pressure control system with dynamic pipe network and model predictive control in an embodiment of the present invention;

[0068] Figure 3 This is a flow chart of the operation of the clean room energy-saving and pressure control system with dynamic pipe network and model predictive control in an embodiment of the present invention;

[0069] Figure 4 This is a control module prediction and optimization flow chart of a clean room energy-saving and pressure control system with a dynamic pipe network and model predictive control in an embodiment of the present invention;

[0070] Figure 5 This is an example structural diagram of a clean room energy-saving and pressure-control system using a dynamic pipe network and model predictive control in an embodiment of the present invention;

[0071] Figure 6 This is a schematic diagram of the control module sending control input;

[0072] Explanation of the reference numerals in the specification: 1. Wind speed sensor for the return air section of the return air clean room; 2. Pressure sensor for the return air section of the return air clean room; 3. Return air valve opening sensor for the return air clean room; 4. Wind speed sensor for the supply air section; 5. Pressure sensor for the supply air section; 6. Wind speed sensor for the supply air section of the return air clean room; 7. Pressure sensor for the supply air section of the return air clean room; 8. Opening sensor for the supply air valve of the return air clean room; 9. Wind speed sensor for the supply air section of the exhaust clean room; 10. Pressure sensor for the supply air section of the exhaust clean room; 11. Opening sensor for the supply air valve of the exhaust clean room; 12. Pressure sensor for the return air clean room; 13. Pressure sensor for the exhaust clean room; 14. Exhaust clean room exhaust Section wind speed sensor; 15. Exhaust section pressure sensor for exhaust clean room; 16. Exhaust valve opening sensor for exhaust clean room; 17. Exhaust section wind speed sensor; 18. Exhaust section pressure sensor; 19. Return air clean room door magnetic sensor; 20. Exhaust clean room door magnetic sensor; 21. Clean room system; 22. CSV data; 23. Control module; 24. Fresh air valve; 25. Air handling equipment; 26. Supply air fan; 27. Return air clean room return air valve; 28. Return air clean room supply air valve; 29. Return air clean room; 30. Exhaust clean room supply air valve; 31. Exhaust clean room; 32. Exhaust clean room exhaust valve; 33. Exhaust fan. DETAILED DESCRIPTION

[0073] The present invention will be further described below with reference to the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it. However, the embodiments are not intended to limit the present invention.

[0074] Example 1:

[0075] Reference Figure 1 As shown, the present invention provides a clean room energy-saving and pressure control method based on a dynamic pipe network and model predictive control, comprising:

[0076] S1: Obtain the current fan speed based on the current fan frequency, and then calculate the current wind pressure based on the current fan speed and current air volume.

[0077] To accurately monitor the fan's operating status, this embodiment installs air volume and pressure sensors at key locations in the duct system. Simultaneously, a functional relationship between fan frequency and fan speed is constructed through experiments:

[0078] V=g(fs)

[0079] Where V is the fan speed, fs is the fan frequency, and g(fs) is the functional relationship between the fan speed and the fan frequency.

[0080] According to the fan similarity law, the air volume of the fan at different speeds is proportional to the speed, and the wind pressure is proportional to the square of the speed. The current wind pressure is calculated based on the current fan speed and current air volume. The formula is:

[0081]

[0082] Where FTP1 is the current wind pressure, FTP0 is the rated wind pressure, V1 and V0 are the current fan speed and the rated fan speed, respectively, Q1 and Q0 are the current air volume and the rated air volume, respectively, and a0, a1, a2, and a3 are fitting coefficients.

[0083] In practical applications, the fan's initial stable speed (V1) can be calculated using the relationship between fan frequency and speed (V = g(fs)). When the fan frequency changes, the speed (V1) corresponding to the new frequency (fs) is calculated. The corresponding wind pressure (FTP1) at different speeds can then be accurately calculated using the above formula.

[0084] S2: The pipe network impedance under different air volumes is obtained through numerical simulation, and a pipe network impedance curve is constructed. Based on the pipe network impedance curve, the current pipe network impedance is obtained from the current air volume. The current valve impedance is calculated based on the valve opening. The current total pressure drop of the pipe network is calculated based on the current valve impedance, the current pipe network impedance, and the air handling equipment impedance.

[0085] In view of the complex pipe network structure of the clean room, computational fluid dynamics (CFD) software is used to simulate and analyze the airflow in the pipe network during the design phase. By inputting detailed parameters such as the pipe diameter, length, roughness, number of elbows and angles of the pipe network, combined with actual gas physical properties data, and the performance parameters of the air handling equipment unique to the clean room (such as high-efficiency air filters, fan coil units, combined air-conditioning units, etc.), the pressure distribution and flow rate in the pipe network under different working conditions are simulated. During actual operation, the working conditions of the clean room will continue to change, which poses a challenge to the stability of the airflow in the pipe network. In order to achieve precise control of the airflow in the pipe network under different working conditions, it is necessary to understand the impedance characteristics of the pipe network, so it is necessary to design a pipe network impedance curve. On this basis, this embodiment defines the pipe network impedance formula as:

[0086]

[0087] Where ΔP Duct Indicates the pressure drop caused by the pipe network, S Duct represents the pipe network impedance, and Q1 represents the air volume.

[0088] By simulating and outputting wind pressure data corresponding to different flow rates, a series of pipe network impedance values are calculated. A refined valve model is also established. Through experimental testing, the flow rates and pressure drops at different valve openings are obtained, and the relationship between valve impedance and valve opening is constructed, which can be approximately expressed as:

[0089]

[0090] Among them, θ represents the valve angle corresponding to the valve opening, S v (θ) represents valve impedance, A' represents valve flow area, ρ represents air density, and a, b, and c represent fitting coefficients.

[0091] The total pressure drop of the current pipe network is calculated based on the current valve impedance, the current pipe network impedance, and the air handling equipment impedance. The formula is:

[0092]

[0093] Where ΔP lose Indicates the total pressure drop of the pipe network, S Duct represents the pipe network impedance, S v (θ) represents the valve impedance, S air It represents the impedance of the clean room air handling equipment, and Q1 represents the air volume.

[0094] The above formula shows that to minimize pipe network impedance, controlling the valve opening above 85% minimizes the overall pipe network resistance coefficient and the total pipe network pressure drop. At this point, the actual operating air volume and pressure of the fan are reduced, thereby reducing fan frequency and saving energy.

[0095] S3: A fan dynamic model is constructed based on the fan similarity law to calculate the air volume at the next moment based on the current air volume, the current total pressure drop of the pipe network, and the current wind pressure. The air volume of the exhaust fan is the exhaust volume, and the air volume of the supply fan is the supply volume.

[0096] Based on the above-mentioned wind turbine similarity law (static relationship), the wind turbine dynamic model is established as the first-order inertia, and the formula is expressed as follows:

[0097]

[0098] Among them, Q1 is the current air volume, is the wind volume at the next moment, τ qs is the air volume response time constant, k qs is the fan gain coefficient, fs is the fan frequency, g(fs) is the function relationship between fan speed and fan frequency, FTP1 is the current wind pressure, ΔP all is the total pressure difference.

[0099] Total pressure difference ΔP all =ΔP1+ΔPlose , where ΔP1 is the clean room pressure, ΔP lose is the total pressure drop in the pipe network.

[0100] The air volume response time constant and fan gain coefficient are measured experimentally. The input is the fan control signal (such as the frequency converter), the output is the fan air volume, and the total pressure drop of the pipe network is ΔP lose The square root of .

[0101] S4: Calculate the valve pressure drop based on the valve opening and the current valve impedance, and then calculate the valve output flow.

[0102] The valve dynamic model is used to adjust the valve opening in the next time period. The formula is:

[0103]

[0104] Among them, θ i is the current valve opening, is the valve opening at the next moment, u θi is the valve opening control signal, τ θi is the valve response time constant, K θi is the valve gain coefficient.

[0105] The valve pressure drop is calculated based on the valve opening and the current valve impedance:

[0106]

[0107] Where ΔP valve is the valve pressure drop, S v (θ) represents the valve impedance, Q 11 Indicates the air volume passing through the valve.

[0108] The calculation formula for the valve output flow is:

[0109]

[0110] Among them, Q i is the current valve output flow, is the valve output flow at the next moment, τ i is the valve flow response time constant, K i is the valve flow gain coefficient, θ i is the current valve opening.

[0111] S5: Construct a dynamic model of clean room pressure difference, which is used to calculate the clean room pressure difference at the next moment based on the current clean room pressure difference, the exhaust volume and the supply volume at the next moment.

[0112] The formula of the clean room pressure difference dynamic model is:

[0113]

[0114] Among them, P e is the current pressure difference in the clean room, is the pressure difference of the clean room at the next moment, C e is the clean room gas volume constant, q in,e and q out,e are the supply and exhaust air volumes of clean room area e, k le is the clean room leakage constant.

[0115] S6: Construct a discretized state-space model, in which the fan frequency and valve opening are control inputs, the exhaust air volume, supply air volume, and valve output flow rate are system state variables, and the cleanroom pressure difference is output; with the goal of strong stability of the cleanroom pressure difference, construct the objective function of the model predictive controller to minimize the pressure difference tracking error and the valve opening penalty term.

[0116] To achieve multivariable dynamic coordinated control, this embodiment introduces a model predictive controller to construct a state space model of the system including fans, pipe networks, valves, clean room pressure difference, etc. The formula is expressed as:

[0117] x(k+1)=Ax(k)+Bu(k)

[0118] y(k)=Cx(k)+Du(k)

[0119] Among them, x(k) is the system state vector at time k, x(k+1) is the system state vector at time k+1, u(k) is the control input vector at time k, y(k) is the output vector at time k, A is the system matrix, B is the input matrix, C is the output matrix, and D is the direct transfer matrix.

[0120] The objective function of the model predictive controller is set for the valve opening of 85% and the strong pressure difference stability target, minimizing the pressure difference tracking error and the valve opening penalty term. The formula is expressed as:

[0121]

[0122] st:u1(k)≥85%+ξ(k),ξ(k)≥0

[0123] st:ΔP e,min -∈(k)≤ΔP e (k)≤ΔP e,max +∈(k),∈(k)≥0

[0124] Among them, J1 is the first objective function, J2 is the second objective function, and N p To predict the time domain length, ΔP e (k) is the clean room pressure difference at time k, P e,refis the preset value of the clean room pressure difference, u1(k) is the valve opening at time k, ∈(k) is the slack variable at time k, which allows short-term violation of the pressure difference constraint; ξ(k) is the soft constraint of the valve opening at time k, which allows the valve opening to be lower than 85% for a short time; ΔP e,min and ΔP e,max are the minimum and maximum values of the clean room pressure difference, ω1, ω2, ω ∈ , w2 and w ξ are all weight coefficients.

[0125] Through theoretical derivation and experimental verification, this paper proposes an operating principle that minimizes pipe network resistance when the valve opening is ≥85%, breaking through the limitations of traditional fixed opening or single-variable regulation. The objective function prioritizes differential pressure stability while limiting energy consumption through a valve opening penalty term, ensuring differential pressure fluctuations are ≤0.5 Pa and recover within 2 seconds.

[0126] S7: Monitor the clean room pressure difference in real time. When the clean room pressure difference is unstable, use the clean room pressure difference dynamic model to predict the change trend of the clean room pressure difference in the next time period. Then use the model predictive controller to reversely solve the discretized state space model for the optimal valve opening to achieve the lowest pipe network impedance and obtain the optimal fan frequency. The fan frequency and valve opening for the next time period are adjusted based on the optimal valve opening and fan frequency.

[0127] Preferably, the cleanroom door opening can be used as the event source, and a high-precision door magnetic sensor can be installed in the cleanroom door frame. When the door is opened, the door magnetic sensor quickly detects the state change and sends a trigger signal to the control system, increasing the fan frequency to a preset frequency. The dynamic model of the cleanroom pressure differential is used to predict the change trend of the cleanroom pressure differential in the next time period. The state-space model is then used to reversely solve the optimal valve opening and adjust the valve opening for the next time period. This method can quickly replenish air volume and reduce pressure differential disturbances. When the cleanroom system door is closed, the door magnetic sensor sends a signal again, and the control system adjusts the fan frequency and valve opening according to steps S1-S7.

[0128] In summary, the present invention describes a clean room energy-saving pressure control method using a dynamic pipe network and model predictive control. Taking into account the pipe network impedance and valve impedance, the pipe network impedance is dynamically calculated by constructing a pipe network impedance curve, and the valve impedance is calculated using the valve opening, so that the total pipe network pressure drop can be calculated. A fan dynamic model is then constructed based on the fan similarity law. The air volume for the next time period is optimized based on the air volume and the total pipe network pressure drop, and the clean room pressure difference for the next time period is dynamically predicted based on the optimized air volume. Finally, a state space model is constructed that includes the fan, pipe network, valve, and clean room pressure difference. The model predictive controller is used to reversely solve the optimal valve opening of the discretized state space model to achieve the lowest pipe network impedance and obtain the optimal fan frequency. The state space model achieves multivariable coordinated control by coupling multiple variables, breaking through the trade-off between pressure difference stability and energy consumption in the existing technology. The present invention combines pipe network impedance and valve opening for joint optimization, and can respond dynamically in advance based on the predicted clean room pressure difference, significantly improving the control accuracy and robustness of the clean room pressure difference while reducing energy consumption. It is suitable for high-demand scenarios such as semiconductors.

[0129] The invention is applicable to the following fields: pharmaceutical and biopharmaceutical industries, such as GMP-certified clean rooms and biosafety laboratories; electronics manufacturing and semiconductor industries, such as semiconductor dust-free rooms and precision optical component production; medical equipment and food industries, such as sterile medical device production, food and dairy clean rooms; intelligent construction and renovation projects, such as hospital operating rooms and ICUs, and energy-saving renovations of existing clean rooms; nuclear facilities and radioactive laboratories; animal disease isolation centers; and cultural relics restoration laboratories.

[0130] Example 2:

[0131] Based on the clean room energy-saving and pressure control method using a dynamic pipe network and model predictive control described in Example 1, this embodiment provides a clean room energy-saving and pressure control system using a dynamic pipe network and model predictive control, including:

[0132] Sensor module, including differential pressure sensor, air volume sensor and valve opening sensor;

[0133] The control module is used to obtain the current fan speed according to the current fan frequency, and then calculate the current wind pressure according to the current fan speed and the current air volume; obtain the pipe network impedance under different air volumes through numerical simulation, and construct the pipe network impedance curve; obtain the current pipe network impedance from the current air volume according to the pipe network impedance curve; calculate the current valve impedance according to the valve opening; calculate the current pipe network total pressure drop according to the current valve impedance, the current pipe network impedance and the air handling equipment impedance; construct a fan dynamic model based on the fan similarity law, which is used to calculate the air volume at the next moment according to the current air volume, the current pipe network total pressure drop and the current wind pressure; among them, the air volume of the exhaust fan is the exhaust volume, and the air volume of the supply fan is the supply volume; construct A dynamic model of cleanroom pressure difference is constructed to calculate the cleanroom pressure difference at the next moment based on the current cleanroom pressure difference, the exhaust volume and the supply volume at the next moment; a state-space model is constructed, in which the fan frequency and valve opening are control inputs, the exhaust volume and the supply volume are system state variables, and the cleanroom pressure difference is output; the objective function of the state-space model is constructed with the goal of strong stability of the cleanroom pressure difference to minimize the pressure difference tracking error and the valve opening penalty term; the cleanroom pressure difference is monitored in real time. When the cleanroom pressure difference is unstable, the cleanroom pressure difference dynamic model is used to predict the change trend of the cleanroom pressure difference in the next time period, and then the state-space model is used to reversely solve the optimal fan frequency and valve opening;

[0134] An execution module is used to adjust the fan frequency and valve opening in the next time period;

[0135] The communication module is responsible for data transmission and communication between the sensor module, control module, and execution module. It uses a reliable communication protocol to ensure accurate data transmission and stable system operation.

[0136] The sensor module includes:

[0137] The differential pressure sensor is installed between the cleanroom and adjacent areas, and at the front and rear ends of the pipe network and fan. It monitors the pressure difference and pipe network impedance in the cleanroom in real time and transmits the pressure difference data to the control module. The sensor has high precision and fast response, and can promptly capture small changes in the pressure difference. The sensor has an accuracy of ±0.1Pa and a data transmission frequency of 100ms / time.

[0138] Air volume sensors are installed at the rear end of the air supply valve, the front end of the exhaust valve, and the front end of the fan to measure the supply and exhaust air volumes. Real-time monitoring of air volume provides the control module with accurate airflow information for precise control and adjustment. The sensor measures wind speeds in the range of 0-20 m / s and calculates air volume with an accuracy of ±2% based on the pipe diameter.

[0139] The valve opening sensor, a magnetostrictive displacement sensor, is installed on the supply and exhaust valves to provide real-time feedback on the valve opening status. The control module can calculate the valve impedance based on the valve opening sensor data and perform a comprehensive analysis in combination with the pipe network impedance model. The sensor detects valve opening from 0-100% with a response time of <50ms.

[0140] Preferably, the sensor module further comprises a door magnetic sensor, which is installed on the clean room door frame and is used to detect the open and close status of the door, and has the highest trigger priority.

[0141] The control module includes:

[0142] The pipe network impedance calculation submodule is used to input pipe network parameters (pipe diameter, roughness, number of elbows, etc.) and output pipe network impedance curves under different flow rates;

[0143] Valve model library, used to store the opening-impedance relationship (such as Kv value table) of different valve types (butterfly valve, ball valve);

[0144] Real-time impedance calculation module, used to dynamically update the total impedance of the pipe network based on the current valve opening and air volume;

[0145] The model predictive controller, as the core control unit of the system, constructs a state space model based on the data collected by the sensor module; it uses the fan frequency and valve opening as control inputs, the supply air volume, exhaust air volume, etc. as system state variables, and the clean room pressure difference as the system output; the model predictive controller sets the objective function, which includes the pressure difference tracking error and the valve opening penalty term; by minimizing the objective function, while ensuring the stability of the clean room pressure difference, it optimizes the valve opening, reduces the pipe network resistance, and achieves energy-saving operation; in each control cycle, the model predictive controller predicts the future state of the system, solves the optimization problem based on the prediction results, and dynamically adjusts the fan frequency and valve opening.

[0146] The execution modules include:

[0147] The fan uses a variable frequency fan, which adjusts its speed according to the frequency signal output by the control module, thereby changing the air supply volume. The fan's performance parameters are obtained by experimental fitting to obtain the relationship between frequency and speed, and dynamically calculated in combination with the fan similarity law. The variable frequency fan drive outputs a 0-50Hz frequency signal, and the frequency adjustment accuracy is ±0.1Hz.

[0148] Valves, including air supply valves and exhaust valves, adjust their opening according to the instructions of the control module. A quantitative relationship is established between valve opening and impedance. By adjusting the valve opening in real time, the pipe network impedance is optimized. The electric valve control outputs a 4-20mA signal and the opening resolution is adjusted to 0.1%.

[0149] The dynamic pipe network and model predictive control clean room energy-saving pressure control system modules work together to achieve precise control of the clean room pressure difference and energy-saving operation.

[0150] The specific working process of the clean room energy-saving and pressure control system with dynamic pipe network and model predictive control provided by the present invention is as follows:

[0151] Step 1: Initialization.

[0152] After the system is started, initialization operations are performed, including loading pipe network parameters, sensor calibration, initial parameter settings, etc. The initial control parameters are set to 30Hz fan frequency and 90% valve opening.

[0153] After initialization is completed, enter the main loop. The process of the main loop refers to Figure 2 As shown, it includes model predictive controller prediction and optimization, multi-mode decision-making and instruction execution, disturbance response detection and logging. At the same time, disturbance detection is added to the main loop to emphasize its control priority. Figure 3 This is a flow chart of the operation of the clean room energy-saving and pressure control system with dynamic pipe network and model predictive control in an embodiment of the present invention.

[0154] Step 2: Data collection.

[0155] The sensor module collects data such as the pressure difference, air supply volume, exhaust volume and valve opening of the clean room in real time, and effectively verifies the collected data.

[0156] When data is invalid, such as data with large fluctuations compared to the previous moment, the abnormal data is marked and ignored.

[0157] When the data is valid, data filtering will be performed to reduce the impact of noise on the data, and this data will be used to calculate the real-time pipeline impedance, and then the working condition will be identified. The working condition identification mainly relies on the dynamic data of the pipeline network. The identified working condition mode is then transmitted to the control module so that it can perform different optimization calculations according to the working conditions.

[0158] Step 3: Working condition identification.

[0159] The system adopts an adaptive control strategy to automatically identify different working modes, such as normal mode, intensive mode, high load mode, etc., based on factors such as personnel activities in the clean room and equipment operating status, and based on personnel density and equipment load.

[0160] Preferably, infrared sensors are used to identify the density of people, and current monitoring is used to identify the load of equipment.

[0161] The normal mode is limited to optimizing energy consumption. On the basis of stabilizing the clean room pressure difference, it optimizes the objective function, outputs a low fan frequency and limits the valve opening to around 85% to optimize energy consumption. The intensive mode is mainly aimed at scenes with frequent personnel activities (such as production peaks and experimental operation periods). At this time, the control module will reduce the priority of optimizing energy consumption, give priority to ensuring the stability of the pressure difference, increase the fan frequency, and allow the valve opening to be temporarily lower than 85% to optimize the overall airflow distribution. The load mode is mainly aimed at changes in equipment operation, such as the increase in pipe network impedance caused by filter blockage in the air handling equipment or the detection of pressure difference disturbances caused by events such as door opening. The system will switch to load mode. At this time, the control module will no longer consider optimizing energy consumption. The only control purpose includes controlling the clean room pressure difference. The control strategy includes outputting the rated fan frequency + dynamic adjustment of the valve opening. The valve opening is no longer used as a constraint condition, and the clean room pressure difference is adjusted first.

[0162] Intensive mode responds to the cleanroom door opening condition. When the door magnetic sensor detects the door opening, it identifies the operating mode as load mode and performs control calculations, setting the fan frequency to 50Hz and dynamically optimizing the valve opening. When the door closes, normal mode is restored, and the control module simultaneously optimizes the fan frequency and valve opening, continuing the main cycle.

[0163] Through adaptive control strategies, the system can automatically adjust control parameters and strategies for different operating modes. For example, in crowded environments, the system can appropriately increase fan frequency and valve opening to ensure adequate ventilation and stable pressure differentials. When the equipment is under low load and there are fewer people, the system can reduce fan frequency and valve opening to achieve energy-saving operation.

[0164] Step 4: Model calculation and optimization.

[0165] Reference Figure 4 As shown, the control module constructs a state-space model based on the collected data, calculating the current system state and predicted future states. It also performs constraint checks. The different constraints represented by different operating modes enable the control module to determine the optimal fan frequency and valve opening based on these constraints. For example, the quadratic programming problem solved in normal mode yields control inputs that stabilize the pressure differential while minimizing energy consumption.

[0166] Step 5: Control command output: The control module sends the calculated fan frequency and valve opening commands to the execution module.

[0167] Step 6: Execute adjustment: The fan adjusts the speed according to the frequency instruction, and the valve adjusts the opening according to the opening instruction to achieve precise control of the pressure difference in the clean room.

[0168] Step 7: Cyclic monitoring and adjustment: The system continuously cycles through steps such as data collection, model calculation, control instruction output, and execution adjustment to monitor and adjust the clean room’s pressure difference and energy consumption in real time, ensuring the system is always in optimal operating condition.

[0169] This embodiment solves the problem that traditional methods cannot actively optimize system resistance through dynamic pipe network impedance modeling and real-time control technology, achieving a 20%-30% reduction in pipe network resistance and a 15%-25% reduction in fan energy consumption. The multivariable collaborative control structure driven by the model predictive controller can improve the pressure difference control accuracy to ±0.5Pa and shorten the disturbance recovery time to within 2 seconds. The optimization algorithm based on the fan similarity law synchronously optimizes the fan frequency and valve opening, thereby improving the fan operating efficiency by 40%. The multi-mode intelligent switching mechanism realizes fine energy consumption optimization, saving 15%-20% in typical scenarios. The event-driven compensation mechanism shortens the pressure difference recovery time after the door is opened to within 2 seconds. The modular distributed architecture improves system reliability and scalability. These innovations jointly break through the difficult trade-off between pressure difference stability and energy consumption, forming a self-consistent intelligent control system, and significantly improving the energy efficiency and robustness of clean room pressure difference control.

[0170] Example 3:

[0171] This paper takes two clean rooms with two different pressure difference control strategies as an example to provide an example of a clean room energy-saving pressure control system with dynamic pipe network and model predictive control. Figure 5 shown.

[0172] The system includes two cleanrooms with different pressure differential control strategies: return cleanroom 29 uses a return air strategy to control pressure differential, while exhaust cleanroom 31 uses an exhaust strategy to control pressure differential. Within the cleanroom ventilation system, the main ventilation section of the air supply duct network is equipped with a supply section wind speed sensor 4 and a supply section pressure sensor 5 to monitor the main duct air volume and pressure.

[0173] The return cleanroom supply air valve 28, the return cleanroom supply air velocity sensor 6, the return cleanroom supply air pressure sensor 7, and the return cleanroom supply air valve opening sensor 8 comprise the supply air VAV system for the return cleanroom 29. The return cleanroom 29 employs a return air strategy to control differential pressure. Therefore, the return air duct is equipped with the return cleanroom return air velocity sensor 1, the return cleanroom return air pressure sensor 2, the return cleanroom return air valve opening sensor 3, and the return cleanroom return air valve 27. The return cleanroom pressure sensor 12 can be installed anywhere in the return cleanroom 29, but it is recommended to be at least two air vent diameters away from the ventilation and exhaust vents to minimize contamination of the collected cleanroom pressure data.

[0174] The exhaust cleanroom 31 uses an exhaust strategy to control the pressure differential. The VAV system for the exhaust cleanroom 31 is comprised of the supply air velocity sensor 9, the supply air pressure sensor 10, the supply air valve opening sensor 11, and the supply air valve 30. The pressure sensor 13 in the exhaust cleanroom 31 detects the pressure within the exhaust cleanroom 31. The exhaust VAV system for the exhaust cleanroom 31 is comprised of the exhaust air velocity sensor 14, the exhaust air pressure sensor 15, the exhaust valve opening sensor 16, and the exhaust valve 32. The exhaust system is powered by the exhaust fan 33. The exhaust air velocity sensor 17 and the exhaust air pressure sensor 18 detect the exhaust fan's air volume and pressure.

[0175] Specifically, the supply air VAV system is installed at the air outlet of the return cleanroom 29, while a return air system is installed near the fresh air valve 24, and the return air VAV system is installed on the return air system. The return air and fresh air together constitute the supply air volume of the return cleanroom 29 and the exhaust cleanroom 31. The supply air volume of each cleanroom is controlled by the supply air VAV system. The exhaust cleanroom 31 is equipped with an exhaust duct, which is equipped with the exhaust VAV system, exhaust fan 33, exhaust section wind speed sensor 17, and exhaust section pressure sensor 18.

[0176] Specifically, the control module 23 is equipped with a wireless communication module for collecting the operating data of the return air section wind speed sensor 1 of the return air clean room, the return air section pressure sensor 2 of the return air clean room, the return air valve opening sensor 3 of the return air clean room, the supply air section wind speed sensor 4, the supply air section pressure sensor 5, the return air clean room supply air section wind speed sensor 6, the return air clean room supply air section pressure sensor 7, the return air clean room supply air valve opening sensor 8, the return air clean room pressure sensor 12, the exhaust clean room supply air section wind speed sensor 9, the exhaust clean room supply air section wind pressure sensor 10, the exhaust clean room supply air valve opening sensor 11, the exhaust clean room pressure sensor 13, the exhaust clean room exhaust section wind speed sensor 14, the exhaust clean room exhaust section pressure sensor 15, the exhaust clean room exhaust valve opening sensor 16, the exhaust section wind speed sensor 17, the exhaust section pressure sensor 18, the return air clean room door magnetic sensor 19, and the exhaust clean room door magnetic sensor 20, and generating CSV data 22.

[0177] The control module monitors the air volume and pressure of the main duct according to the data of the air supply section wind speed sensor 4 and the air supply section pressure sensor 5, and detects the air volume and pressure of the exhaust duct according to the data of the exhaust section wind speed sensor 17 and the exhaust section pressure sensor 18.

[0178] The control module obtains the return air volume of the return cleanroom from the return air section wind speed sensor 1, the return air pressure of the return cleanroom from the return air section pressure sensor 2, and the return air valve opening of the return cleanroom from the return air valve opening sensor 3. The control module then uses the fan dynamic model to calculate the return air volume of the return cleanroom at the next moment. The control module also obtains the supply air volume of the return cleanroom from the return air section wind speed sensor 6, the supply air pressure of the return cleanroom from the return air section pressure sensor 7, and the supply air valve opening of the return cleanroom from the return air valve opening sensor 8. The control module then uses the fan dynamic model to calculate the supply air volume of the return cleanroom at the next moment. The control module obtains the pressure differential of the return cleanroom from the return air section pressure sensor 12. The control module then uses the cleanroom pressure differential dynamic model to predict the pressure differential change of the return cleanroom in the next time period based on the pressure differential of the return cleanroom, the return air volume of the return cleanroom at the next moment, and the supply air volume. Using the state space model, the valve openings of the return air valve 27 and the return air supply valve 28 of the return air clean room are optimized according to the pressure difference change, return air volume and supply air volume of the return air clean room in the next time period to adjust the return air volume and supply air volume.

[0179] The control module obtains the cleanroom's supply air volume from the cleanroom's supply air section wind speed sensor 9, obtains the cleanroom's supply air pressure from the cleanroom's supply air section wind pressure sensor 10, and obtains the cleanroom's supply air valve opening from the cleanroom's supply air valve opening sensor 11. The control module then uses the fan dynamic model to calculate the cleanroom's supply air volume at the next moment. The cleanroom's exhaust air volume is obtained from the cleanroom's exhaust section wind speed sensor 14, obtains the cleanroom's exhaust air pressure from the cleanroom's exhaust section pressure sensor 15, and obtains the cleanroom's supply air valve opening from the cleanroom's exhaust valve opening sensor 16. The control module then uses the fan dynamic model to calculate the cleanroom's exhaust air volume at the next moment. The cleanroom's pressure differential is obtained from the cleanroom's pressure sensor 13. The cleanroom pressure differential dynamic model is then used to predict the cleanroom's pressure differential change in the next time period based on the cleanroom's pressure differential, the cleanroom's supply air volume, and the exhaust air volume at the next moment. By using the state space model, the valve openings of the exhaust clean room supply valve 30 and the exhaust clean room exhaust valve 32 are optimized according to the pressure difference change, the supply air volume and the exhaust air volume of the exhaust clean room in the next time period to adjust the supply air volume and the exhaust air volume.

[0180] When a door opening and closing event causes the pressure difference in the return clean room or the exhaust clean room to be unstable, the fan frequency of the supply fan 26 and the exhaust fan 33 will be increased to the preset frequency; after the pressure difference stabilizes, the state space model will be used to optimize the fan frequency of the supply fan 26 and the exhaust fan 33, as well as the valve openings of the fresh air valve 24, the return clean room supply air valve 28, the exhaust clean room supply air valve 30, and the exhaust clean room exhaust valve 32.

[0181] In this embodiment, constructing a discretized state space model includes:

[0182] The state vector is selected as:

[0183]

[0184] Q supply_total (k) is the air supply fan flow at time k, Q exhaust_total (k) is the exhaust fan air volume at time k, is the output flow of the return air clean room air supply valve at time k, is the return air valve output flow rate of the return air clean room at time k, ΔP1(k) is the pressure difference of the return air clean room at time k, is the output flow of the exhaust clean room air supply valve at time k, is the exhaust valve output flow rate of the exhaust clean room at time k, and ΔP2(k) is the pressure difference of the exhaust clean room at time k.

[0185] Select the input variables as:

[0186]

[0187] f fan_supply (k) is the frequency of the air supply fan at time k, f fan_exhaust (k) is the exhaust fan frequency at time k, is the opening of the return air clean room air supply valve at time k, is the opening of the return air valve of the return air clean room at time k, is the opening of the air supply valve of the exhaust clean room at time k, is the opening of the exhaust valve of the exhaust clean room at time k.

[0188] Then the system matrix is:

[0189]

[0190] Among them, T s is the discretized time step, τ s is the response time of the air supply fan, τ e is the exhaust fan response time, τ v1 is the flow response constant of the return air clean room air supply valve, τ v2 is the return air valve flow response constant of the return air clean room, C1 is the return air clean room gas volume constant, C2 is the exhaust clean room gas volume constant, τ v3 is the flow response constant of the exhaust clean room air supply valve, τ v4 is the flow response constant of the exhaust valve in the clean room.

[0191] The input matrix is:

[0192]

[0193] Among them, Q sr and f sr are the rated air volume and rated frequency of the air supply fan, Q er and f er are the rated air volume and rated frequency of the exhaust fan, K s is the air supply fan gain coefficient, K e is the exhaust fan gain coefficient, K v1 K is the gain coefficient of the return air clean room air supply valve, v2 K is the gain coefficient of the return air valve in the return air clean room, v3 K is the gain coefficient of the exhaust clean room air supply valve, v4 is the exhaust valve gain coefficient of the clean room exhaust valve, ΔP vs1 is the pressure drop of the return air clean room air supply valve, ΔP ve2 is the pressure drop of the return air valve in the clean room, ΔP vs3 is the pressure drop of the exhaust clean room air supply valve, ΔP ve4 The valve pressure drop of the exhaust valve for exhausting clean rooms.

[0194] The output matrix and direct transfer matrix are:

[0195]

[0196] The schematic diagram of the control module sending control inputs is shown in the following figure: Figure 6 The control module 23 transmits the calculated control input to each actuator through the communication module, including the fresh air valve 24, the return air clean room return air valve 27, the supply air fan 26, the return air clean room supply air valve 28, the exhaust air clean room supply air valve 30, the exhaust air clean room exhaust valve 32 and the exhaust air fan 33.

[0197] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will appreciate that other variations or modifications can be made based on the above description. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A clean room energy-saving and pressure control method based on dynamic pipe network and model predictive control, characterized in that: include: The current fan speed is obtained based on the current fan frequency, and the current wind pressure is calculated based on the current fan speed and current air volume; The pipe network impedance under different air volumes is obtained through numerical simulation, and the pipe network impedance curve is constructed; According to the pipe network impedance curve, the current pipe network impedance is obtained from the current air volume; the current valve impedance is calculated based on the valve opening; the current pipe network total pressure drop is calculated based on the current valve impedance, the current pipe network impedance and the air handling equipment impedance; A fan dynamic model is constructed based on the fan similarity law to calculate the next moment's air volume based on the current air volume, the current total pressure drop in the pipe network, and the current wind pressure. The air volume of the exhaust fan is the exhaust volume, and the air volume of the supply fan is the supply volume. Calculate the valve pressure drop based on the valve opening and the current valve impedance, and then calculate the valve output flow; Construct a dynamic model of clean room pressure difference, which is used to calculate the clean room pressure difference at the next moment based on the current clean room pressure difference, the exhaust volume and supply volume at the next moment; A discretized state-space model was constructed, with fan frequency and valve opening as control inputs, exhaust air volume, supply air volume, and valve output flow as system state variables, and cleanroom differential pressure as output. The objective function of the model predictive controller was constructed with the goal of maintaining strong cleanroom differential pressure stability, minimizing the differential pressure tracking error and the valve opening penalty term. The clean room pressure difference is monitored in real time. When the clean room pressure difference is unstable, the clean room pressure difference dynamic model is used to predict the change trend of the clean room pressure difference in the next time period. The model predictive controller is then used to reversely solve the optimal valve opening of the discretized state space model to achieve the lowest pipe network impedance and obtain the optimal fan frequency; the fan frequency and valve opening in the next time period are adjusted based on the optimal valve opening and fan frequency.

2. A clean room energy-saving and pressure-control method based on a dynamic pipe network and model predictive control according to claim 1, characterized in that: The current wind pressure is calculated based on the current fan speed and current air volume. The formula is: ； in, is the current wind pressure, is the rated wind pressure, and are the current fan speed and the rated fan speed respectively, and They are the current air volume and the rated air volume respectively. 、 、 and is the fitting coefficient.

3. The clean room energy-saving and pressure-control method based on a dynamic pipe network and model predictive control according to claim 1, characterized in that: Calculate the current valve impedance based on the valve opening. The formula is: ； in, Indicates the valve angle corresponding to the valve opening, represents the valve impedance, Indicates the valve flow area, represents the air density, 、 and represents the fitting coefficient; The total pressure drop of the current pipe network is calculated based on the current valve impedance, the current pipe network impedance, and the air handling equipment impedance. The formula is: ； in, Indicates the total pressure drop in the pipe network. represents the pipe network impedance, represents the valve impedance, Represents the impedance of the clean room air handling equipment, Indicates air volume; When the valve opening is controlled at above 85%, the total pressure drop in the pipeline network is minimized.

4. The clean room energy-saving and pressure-control method based on a dynamic pipe network and model predictive control according to claim 1, characterized in that: The formula of the fan dynamic model is: ； in, is the current wind volume, is the wind volume at the next moment, is the air volume response time constant, is the fan gain coefficient, is the fan frequency, is the function relationship between fan speed and fan frequency, is the current wind pressure, is the total pressure difference; Total pressure difference ,in For clean room pressure, is the total pressure drop in the pipe network.

5. The clean room energy-saving and pressure-control method based on a dynamic pipe network and model predictive control according to claim 1, characterized in that: The valve dynamic model is used to adjust the valve opening in the next time period. The formula is: ； in, is the current valve opening, is the valve opening at the next moment, is the valve opening control signal, is the valve response time constant, is the valve gain coefficient.

6. The clean room energy-saving and pressure-control method using a dynamic pipe network and model predictive control according to claim 1, characterized in that: Calculate the valve pressure drop based on the valve opening and the current valve impedance, and then calculate the valve output flow, including: The valve pressure drop is calculated based on the valve opening and the current valve impedance. The formula is: ； in, is the valve pressure drop, represents the valve impedance, Indicates the air volume passing through the valve; The calculation formula for the valve output flow is: ； in, is the current valve output flow, The valve output flow at the next moment, is the valve flow response time constant, is the valve flow gain coefficient, is the current valve opening.

7. The clean room energy-saving and pressure-control method based on a dynamic pipe network and model predictive control according to claim 1, characterized in that: The formula of the clean room pressure difference dynamic model is: ； in, is the current pressure difference in the clean room, is the pressure difference of the clean room at the next moment, is the clean room gas volume constant, and are the supply and exhaust air volumes of the clean room area e, is the clean room leakage constant.

8. The clean room energy-saving and pressure-control method based on a dynamic pipe network and model predictive control according to claim 1, characterized in that: Construct a discretized state space model, the formula is expressed as: ； ； in, is the system state vector at time k, is the system state vector at time k+1, is the control input vector at time k, is the output vector at time k, is the system matrix, is the input matrix, is the output matrix, is a direct transfer matrix; The objective function of the model predictive controller is constructed and expressed as follows: ； in, is the first objective function, is the second objective function, To predict the time domain length, is the clean room pressure difference at time k, Preset value for clean room pressure difference, is the valve opening at time k, is the slack variable at time k, is the soft constraint of valve opening at time k, and are the minimum and maximum values of the clean room pressure difference, 、 、 、 and are all weight coefficients.

9. The clean room energy-saving and pressure-control method using a dynamic pipe network and model predictive control according to claim 1, characterized in that: When the clean room door is detected to be open, the fan frequency is increased to the preset frequency, and the dynamic model of the clean room pressure difference is used to predict the changing trend of the clean room pressure difference in the next time period. Then, the model prediction controller is used to reversely solve the optimal valve opening of the discretized state space model and adjust the valve opening for the next time period.

10. A clean room energy-saving and pressure control system based on dynamic pipe network and model predictive control, characterized in that: include: Sensor module, including: Pressure differential sensors are installed between the clean room and adjacent areas, and at the front and back ends of the pipe network and fan to detect the clean room pressure differential and pipe network impedance in real time. Air volume sensors are installed at the rear end of the air supply valve, the front end of the exhaust valve, and the front end of the fan to obtain the supply and exhaust air volumes; Valve opening sensor, installed on the air supply valve and exhaust valve, is used to obtain the valve opening; The control module is connected to the sensor module for communication and is used to obtain the current fan speed according to the current fan frequency, and then calculate the current wind pressure according to the current fan speed and the current air volume; obtain the pipe network impedance under different air volumes through numerical simulation, and construct the pipe network impedance curve; obtain the current pipe network impedance from the current air volume according to the pipe network impedance curve; calculate the current valve impedance according to the valve opening; calculate the current pipe network total pressure drop according to the current valve impedance, the current pipe network impedance and the air handling equipment impedance; construct a fan dynamic model based on the fan similarity law, and calculate the air volume at the next moment according to the current air volume, the current pipe network total pressure drop and the current wind pressure; among them, the air volume of the exhaust fan is the exhaust volume, and the air volume of the supply fan is the supply volume; calculate the valve pressure drop according to the valve opening and the current valve impedance, and then calculate the valve output flow Quantity; construct a dynamic model of clean room pressure difference, which is used to calculate the clean room pressure difference at the next moment based on the current clean room pressure difference, the exhaust volume and the supply volume at the next moment; construct a discretized state space model, in which the fan frequency and valve opening are control inputs, the exhaust volume, the supply volume and the valve output flow are system state variables, and the clean room pressure difference is output; with the strong stability of the clean room pressure difference as the goal, construct the objective function of the model predictive controller, which is used to minimize the pressure difference tracking error and the valve opening penalty term; monitor the clean room pressure difference in real time. When the clean room pressure difference is unstable, use the clean room pressure difference dynamic model to predict the change trend of the clean room pressure difference in the next time period, and then use the model predictive controller to reversely solve the optimal valve opening of the discretized state space model to achieve the lowest pipe network impedance and obtain the optimal fan frequency; The execution module is used to adjust the fan frequency and valve opening in the next time period with the optimal valve opening and fan frequency.

Images