An intelligent constant temperature and pressure control system for clean laboratories

Abstract

This invention discloses an intelligent constant temperature and pressure control system for a cleanroom, comprising: a data acquisition module for deploying sensors to collect data and transmit it to a control unit; a data processing module for calibrating, filtering, and synchronizing data, and inputting it into an MLP network to extract features; a coupling prediction module for establishing a prediction model, calculating deviations, and generating a control error vector; a process control module for executing control strategies based on the error vector and determining equipment operating parameters; an execution control module for transmitting adjustment commands and driving synchronous adjustment of the equipment; and a closed-loop feedback module for collecting adjusted data and feeding it back to the control unit to achieve closed-loop regulation. This invention, through an intelligent control system, adjusts the temperature, pressure, and airflow of the cleanroom in real time, improving environmental control accuracy, energy efficiency, and stability, and ensuring the efficient and reliable operation of the experimental environment.

Description

Technical Field

[0001] This invention relates to the field of intelligent control and automation technology, and in particular to an intelligent constant temperature and pressure control system for a clean laboratory. Background Technology

[0002] As cleanrooms place increasingly stringent requirements on temperature, humidity, and air quality, existing constant temperature and pressure control systems have shown limitations in ensuring environmental stability. Traditional cleanroom control systems typically rely on single control methods, such as manual adjustment systems based on conventional airflow, temperature, and pressure sensors, or automated systems based on simple PID control. These systems mostly use preset, fixed parameters for adjustment, lacking the ability to respond to real-time changes in the laboratory environment. This results in system lag when there are large load fluctuations or drastic changes in the external environment, failing to meet the high requirements of modern cleanrooms for environmental precision and energy efficiency.

[0003] In existing technologies, common methods for temperature and humidity control involve adjusting the supply and return air ratios of the air conditioning system to maintain a constant laboratory environment. Pressure control typically relies on simple differential pressure control devices. However, these methods have several limitations in practical applications. First, changes in the external environment (such as seasonal variations and climate conditions) make it difficult for the system to automatically adjust to its optimal state based on the actual load. Second, while traditional PID control algorithms can effectively handle some simple control tasks, they cannot accurately address the complex dynamic interactions between temperature, pressure, and airflow, easily leading to over-adjustment or response lag.

[0004] Furthermore, existing control systems lack intelligent feedback mechanisms. In most traditional systems, the adjustment of the return air ratio and supply air volume is usually controlled based on simple setpoints, and the adjustment process lacks adaptive capabilities and cannot dynamically optimize according to real-time environmental changes. This results in low energy efficiency, and under extreme environmental conditions, the airflow, temperature, and pressure fluctuations within the cleanroom can be significant, affecting the accuracy of experimental results.

[0005] Therefore, how to provide an intelligent constant temperature and pressure control system for clean laboratories is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] One objective of this invention is to propose an intelligent constant temperature and pressure control system for clean laboratories. This invention makes full use of multiple sensor data acquisition, PID control algorithms, and intelligent prediction and adaptive adjustment technologies. It describes in detail the control strategy based on real-time monitoring of temperature, pressure, and wind speed, and makes dynamic adjustments through a closed-loop feedback mechanism. The system can intelligently optimize multiple parameters such as supply air volume, return air ratio, fresh air volume, and exhaust air volume to ensure that the laboratory environment operates stably under constant temperature and pressure conditions.

[0007] An intelligent temperature and pressure control system for a cleanroom according to an embodiment of the present invention includes:

[0008] The data acquisition module is used to deploy temperature sensors, pressure sensors and wind speed sensors in the clean laboratory to collect temperature, pressure and airflow data of the clean laboratory in real time, and transmit the collected environmental data to the control unit.

[0009] The data processing module is used to preprocess the temperature, pressure and airflow data transmitted to the control unit, including data calibration, filtering and timing synchronization, and input the preprocessed data into the MLP network for environmental state feature extraction;

[0010] The coupled prediction module is used to establish a coupled prediction model of temperature, pressure and airflow based on environmental state characteristics, calculate the prediction deviation between the current environmental state and the target isothermal and isobaric values, and generate a control error vector.

[0011] The process control module is used to execute process control strategies based on the control error vector and determine the operating parameters of the energy-saving air conditioning equipment, including adjustment commands for supply air volume, return air ratio, fresh air volume and exhaust air volume.

[0012] The execution control module is used to transmit adjustment commands to energy-saving air conditioning equipment and actuators, drive the equipment to synchronously adjust the supply air, return air, fresh air and exhaust air, and realize the linkage control of temperature and pressure;

[0013] The closed-loop feedback module is used to continuously collect adjusted environmental data through various sensors, feed it back to the control unit, and continuously update the operating parameters of each module to achieve closed-loop intelligent adjustment of the environmental state.

[0014] Optionally, the data acquisition module specifically includes:

[0015] Temperature sensors, pressure sensors, and wind speed sensors are installed in the cleanroom, among which:

[0016] The temperature sensor uses a four-wire PT100 platinum resistance thermometer. The sensor is fixed to the wall and ceiling at different heights in the clean laboratory with a stainless steel protective sleeve, and is connected to a multi-channel signal conditioning board through shielded twisted pair cables.

[0017] The pressure sensor uses a piezoresistive differential pressure chip and is installed at both ends of the cleanroom, buffer zone, and airlock. The measuring point is connected to the sensor interface through a polytetrafluoroethylene hose. The interface is equipped with a micro filter and damper to prevent airflow interference.

[0018] The wind speed sensor is a thermal wire mesh anemometer, which is installed at the air supply outlet, air return outlet and the horizontal position in the middle of the clean laboratory. The probe is kept in line with the airflow direction by a fixed bracket. The sensor output is connected to a programmable gain amplifier (PGA) and a low-pass filter for signal amplification and filtering.

[0019] The analog signals from all sensors are converted into digital signals by an analog-to-digital converter (ADC) and transmitted to the data acquisition port of the control unit via an industrial Ethernet interface. At the same time, a data buffer and timestamp marker are set in the control unit.

[0020] Optionally, the data processing module specifically includes:

[0021] The latest sampled temperature, pressure, and wind speed signals are read from the data buffer and preprocessed, including:

[0022] The temperature signal is converted into Celsius temperature by the resistance value of the PT100 sensor through a lookup table method, and a second-order polynomial compensation model is established for each sensor to eliminate linear and quadratic errors.

[0023] The process of establishing the second-order polynomial compensation model includes:

[0024] Standard temperature points were set at 0℃, 25℃, 50℃, 75℃ and 100℃ respectively;

[0025] The actual temperature at each calibration point was measured using a high-precision thermometer, and the sensor output resistance value was recorded.

[0026] The resistance value of each sensor is used as the independent variable and the corresponding standard temperature is used as the dependent variable. The coefficients of the second-order polynomial are fitted using the least squares method and saved to the control unit's cache area.

[0027] In the actual data acquisition process, the sensor resistance value is read and substituted into a second-order polynomial to calculate the temperature value, thereby realizing online compensation for the nonlinear error of the sensor.

[0028] The pressure signal is output through a differential amplifier and detected by zero drift. Zero-point calibration is performed by averaging 10 consecutive sampling points using the exponential moving average (EMA), and the range is calibrated linearly in segments.

[0029] The wind speed signal is compensated for gain and calibrated for temperature drift. The 1000 data points per second are filtered by an FIR low-pass filter of the 50th order with a cutoff frequency of 0.2Hz to filter out high-frequency interference. At the same time, a Kalman filter is used to fuse the measurement values ​​from wind speed sensors at different heights.

[0030] After calibration and filtering, timestamps are used to perform linear interpolation and time alignment on the data from the three types of sensors to ensure that temperature, pressure and wind speed are synchronized at the same sampling time.

[0031] Synchronized data are mapped to the range of 0 to 1 using the Softmax function, and a multidimensional feature matrix is ​​constructed according to sensor type and spatial location order;

[0032] The multidimensional feature matrix is ​​input into the MLP network to obtain the environmental state feature vector, which is then output through the control unit buffer, including:

[0033] The MLP network consists of an input layer, two hidden layers, and an output layer. The number of nodes in the input layer corresponds to the dimension of the multidimensional feature matrix. The number of nodes in the hidden layer is 1.5 times that of the input layer. The activation function is ReLU. The number of nodes in the output layer corresponds to the dimension of the coupled features. Each forward propagation calculates a weighted sum and the activation values ​​to generate an environmental state feature vector.

[0034] Optionally, the coupling prediction module specifically includes:

[0035] Using the environmental state feature vector as input, an initial matrix is ​​constructed. Each row of the initial matrix represents the environmental state feature of a time step, including temperature, pressure, and wind speed.

[0036] LSTM networks are used for modeling, as they are capable of handling long-term dependencies in time-series data.

[0037] To avoid overfitting, the input layer of the LSTM network contains 128 neurons, the first LSTM layer has 256 neurons, the second LSTM layer has 128 neurons, the activation function is tanh, and the output layer has 3 neurons, representing the predicted values ​​of temperature, pressure and airflow respectively.

[0038] The LSTM network is trained using batch gradient descent with a batch size of 64, a learning rate of 0.001, and 200 training epochs. The Adam optimization algorithm is used for parameter updates.

[0039] The training data was generated using a sliding window method with a window size of 30 and a step size of 1 to capture the dynamic changes in the environmental state.

[0040] After training, the LSTM network can predict the changing trends of temperature, pressure and airflow over a future period of time based on the environmental state characteristics of the current time step.

[0041] Once the network has completed training, in real-time applications, the control unit periodically inputs the latest environmental state features collected into the LSTM network to obtain predicted temperature, pressure, and airflow values.

[0042] The difference between the predicted value and the target isothermal and isobaric values ​​is calculated, and the mean square error is used as the loss function to generate a control error vector, which is then normalized.

[0043] Optionally, the process control module specifically includes:

[0044] Based on the control error vector, the process control strategy is executed, and combined with the preset PID control algorithm, the adjustment commands for supply air volume, return air ratio, fresh air volume and exhaust air volume are calculated.

[0045] The parameters of a PID control algorithm include proportional coefficient, integral coefficient, and derivative coefficient, which are used to adjust various operating parameters according to environmental deviations.

[0046] The scaling factor calculates the difference between the current deviation and the target value;

[0047] The integral coefficient is used to calculate the cumulative value of past deviations to prevent the accumulation of long-term errors.

[0048] The differential coefficients assess the rate of change of the current error to prevent over-adjustment;

[0049] After PID control, the resulting control signal will serve as the initial adjustment command for each item.

[0050] The control unit inputs preliminary adjustment commands to each execution unit, including supply air volume adjustment, return air ratio adjustment, fresh air volume adjustment, and exhaust air volume adjustment.

[0051] The air supply actuator adjusts the fan speed and controls the air supply volume according to the air supply volume adjustment command;

[0052] The return air actuator controls the return air ratio by adjusting the opening of the return air vent;

[0053] The fresh air actuator controls the amount of fresh air by adjusting the opening of the fresh air valve;

[0054] The exhaust actuator ensures that the exhaust volume meets the set value by adjusting the opening of the exhaust valve;

[0055] Each regulating unit transmits feedback information to the control unit to ensure that the air volume and pressure are always kept within the target range;

[0056] To compensate for the nonlinear response of the fan and valves, the control unit combines the fan's performance curve and historical operating data to optimize the adjustment parameters online and correct the response of each actuator in real time.

[0057] All control commands are processed and output in real time by a digital signal processor (DSP) to ensure precise regulation of supply air, return air, fresh air and exhaust air.

[0058] The effects of each adjustment are fed back to the control unit in real time through various sensors. When the deviation between the actual value and the target value exceeds the preset tolerance range, the control unit will dynamically adjust various control parameters to ensure system stability.

[0059] Optionally, the air supply execution unit specifically includes:

[0060] The air supply actuator receives the air supply volume adjustment command from the control unit. The air supply volume adjustment command is a fan speed adjustment signal calculated based on the PID control algorithm.

[0061] The control unit calculates the control error vectors of temperature, pressure and wind speed, and generates corresponding air volume adjustment commands. The air volume adjustment commands are transmitted to the control unit through the industrial Ethernet interface.

[0062] The control unit adjusts the speed of the fan motor according to the air volume adjustment command, and the relationship between the fan motor and the air volume is precisely calibrated through the fan's performance curve;

[0063] The wind speed signal is converted into a digital signal by an analog-to-digital converter (ADC) and transmitted to the control unit via a multi-channel signal conditioning board. If the deviation between the actual wind speed and the set target air volume exceeds the preset tolerance range, the control unit will correct it by adjusting the fan speed. The fan speed is calculated in real time by a digital signal processor (DSP) and combined with the fan load curve to optimize the control signal in real time to ensure accurate control of the air volume.

[0064] Optionally, the return air execution unit specifically includes:

[0065] The control unit receives the return air proportional adjustment command and calculates the target value of the return air vent opening according to the PID control algorithm;

[0066] The control unit sends the target value to the return air actuator, which precisely adjusts the opening of the return air vent via an electric actuator;

[0067] The electric actuator changes the opening of the return air vent according to the incoming control signal, precisely adjusting the return air ratio;

[0068] The wind speed signal is processed by an analog-to-digital converter (ADC) and then transmitted to the control unit.

[0069] The control unit performs real-time comparison based on the feedback signals of real-time monitored return air flow and return air vent opening. If the deviation between the return air vent opening and the set target return air ratio exceeds the tolerance range, the control unit will dynamically adjust the return air vent opening to ensure that the return air ratio is maintained within the set range.

[0070] The beneficial effects of this invention are:

[0071] First, this invention achieves real-time adjustment of the cleanroom environment through an intelligent control system. It can dynamically adjust the supply air volume, return air ratio, fresh air volume, and exhaust air volume based on real-time data of temperature, pressure, and wind speed, ensuring that environmental parameters are always maintained within the target range. This avoids the lag and errors caused by manual adjustment and fixed parameters in traditional systems.

[0072] Secondly, this invention employs a PID control algorithm and a deep learning prediction model, combined with the fan performance curve, to precisely adjust the equipment, effectively improving the response speed and accuracy of environmental regulation. The closed-loop feedback mechanism enables the system to quickly compensate for errors under any circumstances, avoiding the inaccurate control caused by load fluctuations or changes in the external environment in traditional systems.

[0073] Finally, this invention not only improves the stability of the laboratory environment but also significantly enhances energy efficiency. By optimizing various control parameters in real time, the system can reduce energy consumption while ensuring environmental quality, achieving efficient and energy-saving operation. Attached Figure Description

[0074] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0075] Figure 1 This is a block diagram of the intelligent constant temperature and pressure control system for a clean laboratory proposed in this invention.

[0076] Figure 2 This is a flowchart illustrating the fan control and regulation process of an intelligent constant temperature and pressure control system for a clean laboratory proposed in this invention.

[0077] Figure 3 This is a flowchart illustrating the closed-loop feedback and optimization adjustment process of an intelligent constant temperature and pressure control system for a clean laboratory proposed in this invention. Detailed Implementation

[0078] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0079] refer to Figures 1-3An intelligent temperature and pressure control system for a clean laboratory, comprising:

[0080] The data acquisition module is used to deploy temperature sensors, pressure sensors and wind speed sensors in the clean laboratory to collect temperature, pressure and airflow data of the clean laboratory in real time, and transmit the collected environmental data to the control unit.

[0081] The data processing module is used to preprocess the temperature, pressure and airflow data transmitted to the control unit, including data calibration, filtering and timing synchronization, and input the preprocessed data into the MLP network for environmental state feature extraction;

[0082] The coupled prediction module is used to establish a coupled prediction model of temperature, pressure and airflow based on environmental state characteristics, calculate the prediction deviation between the current environmental state and the target isothermal and isobaric values, and generate a control error vector.

[0083] The process control module is used to execute process control strategies based on the control error vector and determine the operating parameters of the energy-saving air conditioning equipment, including adjustment commands for supply air volume, return air ratio, fresh air volume and exhaust air volume.

[0084] The execution control module is used to transmit adjustment commands to energy-saving air conditioning equipment and actuators, drive the equipment to synchronously adjust the supply air, return air, fresh air and exhaust air, and realize the linkage control of temperature and pressure;

[0085] The closed-loop feedback module is used to continuously collect adjusted environmental data through various sensors, feed it back to the control unit, and continuously update the operating parameters of each module to achieve closed-loop intelligent adjustment of the environmental state.

[0086] In this embodiment, the data acquisition module specifically includes:

[0087] Temperature sensors, pressure sensors, and wind speed sensors are installed in the cleanroom, among which:

[0088] The temperature sensor uses a four-wire PT100 platinum resistance thermometer. The sensor is fixed to the wall and ceiling at different heights in the clean laboratory with a stainless steel protective sleeve, and is connected to a multi-channel signal conditioning board through shielded twisted pair cables.

[0089] The pressure sensor uses a piezoresistive differential pressure chip and is installed at both ends of the cleanroom, buffer zone, and airlock. The measuring point is connected to the sensor interface through a polytetrafluoroethylene hose. The interface is equipped with a micro filter and damper to prevent airflow interference.

[0090] The wind speed sensor is a thermal wire mesh anemometer, which is installed at the air supply outlet, air return outlet and the horizontal position in the middle of the clean laboratory. The probe is kept in line with the airflow direction by a fixed bracket. The sensor output is connected to a programmable gain amplifier (PGA) and a low-pass filter for signal amplification and filtering.

[0091] The analog signals from all sensors are converted into digital signals by an analog-to-digital converter (ADC) and transmitted to the data acquisition port of the control unit via an industrial Ethernet interface. At the same time, a data buffer and timestamp marker are set in the control unit.

[0092] In this embodiment, the data processing module specifically includes:

[0093] The latest sampled temperature, pressure, and wind speed signals are read from the data buffer and preprocessed, including:

[0094] The temperature signal is converted into Celsius temperature by the resistance value of the PT100 sensor through a lookup table method, and a second-order polynomial compensation model is established for each sensor to eliminate linear and quadratic errors.

[0095] The process of establishing the second-order polynomial compensation model includes:

[0096] Standard temperature points were set at 0℃, 25℃, 50℃, 75℃ and 100℃ respectively;

[0097] The actual temperature at each calibration point was measured using a high-precision thermometer, and the sensor output resistance value was recorded.

[0098] The resistance value of each sensor is used as the independent variable and the corresponding standard temperature is used as the dependent variable. The coefficients of the second-order polynomial are fitted using the least squares method and saved to the control unit's cache area.

[0099] In the actual data acquisition process, the sensor resistance value is read and substituted into a second-order polynomial to calculate the temperature value, thereby realizing online compensation for the nonlinear error of the sensor.

[0100] The pressure signal is output through a differential amplifier and detected by zero drift. Zero-point calibration is performed by averaging 10 consecutive sampling points using the exponential moving average (EMA), and the range is calibrated linearly in segments.

[0101] The wind speed signal is compensated for gain and calibrated for temperature drift. The 1000 data points per second are filtered by an FIR low-pass filter of the 50th order with a cutoff frequency of 0.2Hz to filter out high-frequency interference. At the same time, a Kalman filter is used to fuse the measurement values ​​from wind speed sensors at different heights.

[0102] After calibration and filtering, timestamps are used to perform linear interpolation and time alignment on the data from the three types of sensors to ensure that temperature, pressure and wind speed are synchronized at the same sampling time.

[0103] Synchronized data are mapped to the range of 0 to 1 using the Softmax function, and a multidimensional feature matrix is ​​constructed according to sensor type and spatial location order;

[0104] The multidimensional feature matrix is ​​input into the MLP network to obtain the environmental state feature vector, which is then output through the control unit buffer, including:

[0105] The MLP network consists of an input layer, two hidden layers, and an output layer. The number of nodes in the input layer corresponds to the dimension of the multidimensional feature matrix. The number of nodes in the hidden layer is 1.5 times that of the input layer. The activation function is ReLU. The number of nodes in the output layer corresponds to the dimension of the coupled features. Each forward propagation calculates a weighted sum and the activation values ​​to generate an environmental state feature vector.

[0106] In this embodiment, the coupling prediction module specifically includes:

[0107] Using the environmental state feature vector as input, an initial matrix is ​​constructed. Each row of the initial matrix represents the environmental state feature of a time step, including temperature, pressure, and wind speed.

[0108] LSTM networks are used for modeling, as they are capable of handling long-term dependencies in time-series data.

[0109] To avoid overfitting, the input layer of the LSTM network contains 128 neurons, the first LSTM layer has 256 neurons, the second LSTM layer has 128 neurons, the activation function is tanh, and the output layer has 3 neurons, representing the predicted values ​​of temperature, pressure and airflow respectively.

[0110] The LSTM network is trained using batch gradient descent with a batch size of 64, a learning rate of 0.001, and 200 training epochs. The Adam optimization algorithm is used for parameter updates.

[0111] The training data was generated using a sliding window method with a window size of 30 and a step size of 1 to capture the dynamic changes in the environmental state.

[0112] After training, the LSTM network can predict the changing trends of temperature, pressure and airflow over a future period of time based on the environmental state characteristics of the current time step.

[0113] Once the network has completed training, in real-time applications, the control unit periodically inputs the latest environmental state features collected into the LSTM network to obtain predicted temperature, pressure, and airflow values.

[0114] The difference between the predicted value and the target isothermal and isobaric values ​​is calculated, and the mean square error is used as the loss function to generate a control error vector, which is then normalized.

[0115] In this embodiment, the process control module specifically includes:

[0116] Based on the control error vector, the process control strategy is executed, and combined with the preset PID control algorithm, the adjustment commands for supply air volume, return air ratio, fresh air volume and exhaust air volume are calculated.

[0117] The parameters of a PID control algorithm include proportional coefficient, integral coefficient, and derivative coefficient, which are used to adjust various operating parameters according to environmental deviations.

[0118] The scaling factor calculates the difference between the current deviation and the target value;

[0119] The integral coefficient is used to calculate the cumulative value of past deviations to prevent the accumulation of long-term errors.

[0120] The differential coefficients assess the rate of change of the current error to prevent over-adjustment;

[0121] After PID control, the resulting control signal will serve as the initial adjustment command for each item.

[0122] The control unit inputs preliminary adjustment commands to each execution unit, including supply air volume adjustment, return air ratio adjustment, fresh air volume adjustment, and exhaust air volume adjustment.

[0123] The air supply actuator adjusts the fan speed and controls the air supply volume according to the air supply volume adjustment command;

[0124] The return air actuator controls the return air ratio by adjusting the opening of the return air vent;

[0125] The fresh air actuator controls the amount of fresh air by adjusting the opening of the fresh air valve;

[0126] The exhaust actuator ensures that the exhaust volume meets the set value by adjusting the opening of the exhaust valve;

[0127] Each regulating unit transmits feedback information to the control unit to ensure that the air volume and pressure are always kept within the target range;

[0128] To compensate for the nonlinear response of the fan and valves, the control unit combines the fan's performance curve and historical operating data to optimize the adjustment parameters online and correct the response of each actuator in real time.

[0129] All control commands are processed and output in real time by a digital signal processor (DSP) to ensure precise regulation of supply air, return air, fresh air and exhaust air.

[0130] The effects of each adjustment are fed back to the control unit in real time through various sensors. When the deviation between the actual value and the target value exceeds the preset tolerance range, the control unit will dynamically adjust various control parameters to ensure system stability.

[0131] In this embodiment, the air supply execution unit specifically includes:

[0132] The air supply actuator receives the air supply volume adjustment command from the control unit. The air supply volume adjustment command is a fan speed adjustment signal calculated based on the PID control algorithm.

[0133] The control unit calculates the control error vectors of temperature, pressure and wind speed, and generates corresponding air volume adjustment commands. The air volume adjustment commands are transmitted to the control unit through the industrial Ethernet interface.

[0134] The control unit adjusts the speed of the fan motor according to the air volume adjustment command, and the relationship between the fan motor and the air volume is precisely calibrated through the fan's performance curve;

[0135] The wind speed signal is converted into a digital signal by an analog-to-digital converter (ADC) and transmitted to the control unit via a multi-channel signal conditioning board. If the deviation between the actual wind speed and the set target air volume exceeds the preset tolerance range, the control unit will correct it by adjusting the fan speed. The fan speed is calculated in real time by a digital signal processor (DSP) and combined with the fan load curve to optimize the control signal in real time to ensure accurate control of the air volume.

[0136] Specifically, let's assume the relationship between fan speed and air volume is as follows:

[0137] ;

[0138] in, Indicates the air supply volume. Indicates the fan speed. Indicates temperature. Indicates pressure, Indicates wind speed;

[0139] The control error vector is:

[0140] ;

[0141] in, This represents the temperature control error, specifically the difference between the target temperature and the actual temperature. This represents the pressure control error, specifically the difference between the target pressure and the actual pressure. This indicates the wind speed control error, which is the difference between the target wind speed and the actual wind speed.

[0142] The air volume adjustment command calculated by the PID control algorithm Given by the following formula:

[0143] ;

[0144] in, Indicates the proportional gain, which adjusts the strength of the proportional control. This represents the integral gain, which adjusts the strength of the integral control and is used to compensate for long-term deviations. It represents the differential gain, which adjusts the strength of the differential control and is used to predict the rate of error change. Indicates time Timing control error, This represents the cumulative sum of control errors. The integral term representing the error. The differential term representing the control error;

[0145] The fan speed adjustment signal is generated by a PID control algorithm, and the required fan speed is calculated by combining the fan performance curve. The expression is:

[0146] ;

[0147] in, Indicates the base fan speed. This indicates the speed adjustment amount generated by PID control;

[0148] air volume From the fan speed and wind turbine performance curves Sure:

[0149] ;

[0150] In practice, if With target air volume If the deviation exceeds the preset tolerance range, the control unit will continue to adjust the fan speed. Make adjustments until the air volume meets the target setting.

[0151] In this embodiment, the return air execution unit specifically includes:

[0152] The control unit receives the return air proportional adjustment command and calculates the target value of the return air vent opening according to the PID control algorithm;

[0153] The control unit sends the target value to the return air actuator, which precisely adjusts the opening of the return air vent via an electric actuator;

[0154] The electric actuator changes the opening of the return air vent according to the incoming control signal, precisely adjusting the return air ratio;

[0155] The wind speed signal is processed by an analog-to-digital converter (ADC) and then transmitted to the control unit.

[0156] The control unit performs real-time comparison based on the feedback signals of real-time monitored return air flow and return air vent opening. If the deviation between the return air vent opening and the set target return air ratio exceeds the tolerance range, the control unit will dynamically adjust the return air vent opening to ensure that the return air ratio is maintained within the set range.

[0157] Specifically, assuming the relationship between the return air ratio and the return air vent opening can be expressed as:

[0158] ;

[0159] in, The actual return air ratio indicates the proportion of the current return air volume to the total air volume. This is a reference value for the return air vent opening. This indicates real-time wind speed, used to assess the actual return air volume. A function representing the nonlinear relationship between the return air ratio, return air vent opening, and wind speed;

[0160] The adjustment amount of the return air vent opening calculated by the PID control algorithm Given by the following formula:

[0161]

[0162] ;

[0163] in, Indicates the proportional gain, which adjusts the strength of the proportional control. This represents the integral gain, which adjusts the strength of the integral control and is used to eliminate long-term deviations. This represents the derivative gain, which adjusts the strength of the derivative control and is used to control the rate of error change. This indicates the target return air ratio, which is set by the control unit according to the requirements of the cleanroom.

[0164] Return air vent opening Given by the following formula:

[0165] ;

[0166] in, This indicates the basic return air vent opening, i.e., the initial state or default setting of the return air vent. This represents the opening adjustment amount dynamically calculated by the PID control algorithm based on the return air proportional error;

[0167] If the actual return air ratio Ratio of return air to target If the deviation exceeds the tolerance range, the control unit will adjust the opening of the return air vent. Make corrections;

[0168] Real-time feedback monitoring of return air flow and return air vent opening The control unit continuously optimizes the opening of the return air vent to ensure the return air ratio. Always meet the set goals.

[0169] Example 1: To verify the feasibility of this invention in practice, it was applied to an intelligent constant temperature and pressure control system for a cleanroom. The goal of this cleanroom is to maintain constant temperature, pressure, and airflow to ensure the stability of the experimental process and the accuracy of the data. In traditional control methods, large environmental fluctuations often lead to instability in temperature, humidity, and airflow, resulting in inaccurate control of the laboratory environment and severely impacting the reliability of experimental results. To overcome this problem, the intelligent constant temperature and pressure control system of this invention monitors environmental data in real time and dynamically adjusts the supply air volume, return air ratio, fresh air volume, and exhaust air volume, ensuring that the system can accurately adjust under different external conditions to achieve optimal control.

[0170] In this scenario, environmental data within the cleanroom is first collected in real time by deploying temperature, pressure, and wind speed sensors. The ambient temperature data collected by the temperature sensors is processed and transmitted to the control unit; the pressure sensor monitors the pressure difference between the indoor and outdoor environments in real time; and the wind speed sensor provides airflow velocity information at the supply and return air vents and different locations within the laboratory. The control unit processes this data comprehensively, calculating adjustment commands for the supply air volume, return air ratio, fresh air volume, and exhaust air volume based on a PID control algorithm. These commands are then sent to the relevant execution units, which precisely adjust the fan speed, return air vent opening, fresh air valve opening, and exhaust valve opening to ensure that the indoor temperature, pressure, and airflow remain within set ranges.

[0171] For example, in an experimental environment with a set temperature of 22°C, a set pressure of 101.3 kPa, and a target wind speed of 0.2 m / s, the system dynamically adjusts the fan speed and the opening of the return and exhaust vents by monitoring data from various sensors in real time and combining this with a control algorithm. When the system detects that the indoor temperature has risen to 23.5°C, the air supply volume adjustment command generated by the PID algorithm requires the fan to accelerate, increasing the air supply volume, while simultaneously adjusting the return air ratio to reduce the input of fresh air and maintain stable indoor pressure. Based on real-time data feedback, the system further optimizes the control strategy, reducing unnecessary power consumption of the fan and improving energy efficiency.

[0172] The data in the examples demonstrate the superiority of the system of the present invention in practical applications. As shown in Table 1, under traditional control methods, the temperature control accuracy is typically ±1.5°C, while the system of the present invention can improve the temperature control accuracy to within ±0.5°C; the pressure control accuracy is also significantly improved, from the traditional ±3 Pa to within ±1.5 Pa, ensuring the stability and accuracy of the experimental environment. Furthermore, the system can adjust promptly under conditions of large load fluctuations, minimizing control errors caused by changes in the external environment and ensuring that the laboratory is always in optimal working condition.

[0173] Table 1 Comparison data between traditional control methods and the control method of this invention

[0174] Control methods Temperature control accuracy (°C) Pressure control accuracy (Pa) Wind speed control accuracy (m / s) Wind turbine power consumption (kW) Experimental environment stability score (0-100) Traditional control methods ±1.5 ±3 ±0.2 5.2 75 Control method of the present invention ±0.5 ±1.5 ±0.05 3.2 95

[0175] The above data demonstrates that this invention offers significant advantages in improving environmental control precision and optimizing energy efficiency. Temperature and pressure control precision are significantly improved, wind speed control precision is substantially enhanced, and fan power consumption is reduced by approximately 38%. Furthermore, the experimental environment stability score increased from 75 using traditional methods to 95, indicating that this invention can greatly improve the stability of the laboratory environment and ensure the reliability of experimental results.

[0176] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An intelligent constant temperature and pressure control system for a clean laboratory, characterized in that, include: The data acquisition module is used to deploy temperature sensors, pressure sensors and wind speed sensors in the clean laboratory to collect temperature, pressure and airflow data of the clean laboratory in real time, and transmit the collected environmental data to the control unit. The data processing module is used to preprocess the temperature, pressure and airflow data transmitted to the control unit, including data calibration, filtering and timing synchronization, and input the preprocessed data into the MLP network for environmental state feature extraction; The coupled prediction module is used to establish a coupled prediction model of temperature, pressure and airflow based on environmental state characteristics, calculate the prediction deviation between the current environmental state and the target isothermal and isobaric values, and generate a control error vector. The process control module is used to execute process control strategies based on the control error vector and determine the operating parameters of the energy-saving air conditioning equipment, including adjustment commands for supply air volume, return air ratio, fresh air volume and exhaust air volume. The execution control module is used to transmit adjustment commands to energy-saving air conditioning equipment and actuators, drive the equipment to synchronously adjust the supply air, return air, fresh air and exhaust air, and realize the linkage control of temperature and pressure; The closed-loop feedback module is used to continuously collect adjusted environmental data through various sensors, feed it back to the control unit, and continuously update the operating parameters of each module to achieve closed-loop intelligent adjustment of the environmental state.

2. The intelligent constant temperature and pressure control system for a clean laboratory according to claim 1, characterized in that, The data acquisition module specifically includes: Temperature sensors, pressure sensors, and wind speed sensors are installed in the cleanroom, among which: The temperature sensor uses a four-wire PT100 platinum resistance thermometer. The sensor is fixed to the wall and ceiling at different heights in the clean laboratory with a stainless steel protective sleeve, and is connected to a multi-channel signal conditioning board through shielded twisted pair cables. The pressure sensor uses a piezoresistive differential pressure chip and is installed at both ends of the cleanroom, buffer zone, and airlock. The measuring point is connected to the sensor interface through a polytetrafluoroethylene hose. The interface is equipped with a micro filter and damper to prevent airflow interference. The wind speed sensor is a thermal wire mesh anemometer, which is installed at the air supply outlet, air return outlet and the horizontal position in the middle of the clean laboratory. The probe is kept in line with the airflow direction by a fixed bracket. The sensor output is connected to a programmable gain amplifier (PGA) and a low-pass filter for signal amplification and filtering. The analog signals from all sensors are converted into digital signals by an analog-to-digital converter (ADC) and transmitted to the data acquisition port of the control unit via an industrial Ethernet interface. At the same time, a data buffer and timestamp marker are set in the control unit.

3. The intelligent constant temperature and pressure control system for a clean laboratory according to claim 1, characterized in that, The data processing module specifically includes: The latest sampled temperature, pressure, and wind speed signals are read from the data buffer and preprocessed, including: The temperature signal is converted into Celsius temperature by the resistance value of the PT100 sensor through a lookup table method, and a second-order polynomial compensation model is established for each sensor to eliminate linear and quadratic errors. The process of establishing the second-order polynomial compensation model includes: Standard temperature points were set at 0℃, 25℃, 50℃, 75℃ and 100℃ respectively; The actual temperature at each calibration point was measured using a high-precision thermometer, and the sensor output resistance value was recorded. The resistance value of each sensor is used as the independent variable and the corresponding standard temperature is used as the dependent variable. The coefficients of the second-order polynomial are fitted using the least squares method and saved to the control unit's cache area. In the actual data acquisition process, the sensor resistance value is read and substituted into a second-order polynomial to calculate the temperature value, thereby realizing online compensation for the nonlinear error of the sensor. The pressure signal is output through a differential amplifier and detected by zero drift. Zero-point calibration is performed by averaging 10 consecutive sampling points using the exponential moving average (EMA), and the range is calibrated linearly in segments. The wind speed signal is compensated for gain and calibrated for temperature drift. The 1000 data points per second are filtered by an FIR low-pass filter of the 50th order with a cutoff frequency of 0.2Hz to filter out high-frequency interference. At the same time, a Kalman filter is used to fuse the measurement values ​​from wind speed sensors at different heights. After calibration and filtering, timestamps are used to perform linear interpolation and time alignment on the data from the three types of sensors to ensure that temperature, pressure and wind speed are synchronized at the same sampling time. Synchronized data are mapped to the range of 0 to 1 using the Softmax function, and a multidimensional feature matrix is ​​constructed according to sensor type and spatial location order; The multidimensional feature matrix is ​​input into the MLP network to obtain the environmental state feature vector, which is then output through the control unit buffer, including: The MLP network consists of an input layer, two hidden layers, and an output layer. The number of nodes in the input layer corresponds to the dimension of the multidimensional feature matrix. The number of nodes in the hidden layer is 1.5 times that of the input layer. The activation function is ReLU. The number of nodes in the output layer corresponds to the dimension of the coupled features. Each forward propagation calculates a weighted sum and the activation values ​​to generate an environmental state feature vector.

4. The intelligent constant temperature and pressure control system for a clean laboratory according to claim 1, characterized in that, The coupling prediction module specifically includes: Using the environmental state feature vector as input, an initial matrix is ​​constructed. Each row of the initial matrix represents the environmental state feature of a time step, including temperature, pressure, and wind speed. LSTM networks are used for modeling, as they are capable of handling long-term dependencies in time-series data. To avoid overfitting, the input layer of the LSTM network contains 128 neurons, the first LSTM layer has 256 neurons, the second LSTM layer has 128 neurons, the activation function is tanh, and the output layer has 3 neurons, representing the predicted values ​​of temperature, pressure and airflow respectively. The LSTM network is trained using batch gradient descent with a batch size of 64, a learning rate of 0.001, and 200 training epochs. The Adam optimization algorithm is used for parameter updates. The training data was generated using a sliding window method with a window size of 30 and a step size of 1 to capture the dynamic changes in the environmental state. After training, the LSTM network can predict the changing trends of temperature, pressure and airflow over a future period of time based on the environmental state characteristics of the current time step. Once the network has completed training, in real-time applications, the control unit periodically inputs the latest environmental state features collected into the LSTM network to obtain predicted temperature, pressure, and airflow values. The difference between the predicted value and the target isothermal and isobaric values ​​is calculated, and the mean square error is used as the loss function to generate a control error vector, which is then normalized.

5. The intelligent constant temperature and pressure control system for a clean laboratory according to claim 1, characterized in that, The process control module specifically includes: Based on the control error vector, the process control strategy is executed, and combined with the preset PID control algorithm, the adjustment commands for supply air volume, return air ratio, fresh air volume and exhaust air volume are calculated. The parameters of a PID control algorithm include proportional coefficient, integral coefficient, and derivative coefficient, which are used to adjust various operating parameters according to environmental deviations. The scaling factor calculates the difference between the current deviation and the target value; The integral coefficient is used to calculate the cumulative value of past deviations to prevent the accumulation of long-term errors. The differential coefficients assess the rate of change of the current error to prevent over-adjustment; After PID control, the resulting control signal will serve as the initial adjustment command for each item. The control unit inputs preliminary adjustment commands to each execution unit, including supply air volume adjustment, return air ratio adjustment, fresh air volume adjustment, and exhaust air volume adjustment. The air supply actuator adjusts the fan speed and controls the air supply volume according to the air supply volume adjustment command; The return air actuator controls the return air ratio by adjusting the opening of the return air vent; The fresh air actuator controls the amount of fresh air by adjusting the opening of the fresh air valve; The exhaust actuator ensures that the exhaust volume meets the set value by adjusting the opening of the exhaust valve; Each regulating unit transmits feedback information to the control unit to ensure that the air volume and pressure are always kept within the target range; To compensate for the nonlinear response of the fan and valves, the control unit combines the fan's performance curve and historical operating data to optimize the adjustment parameters online and correct the response of each actuator in real time. All control commands are processed and output in real time by a digital signal processor (DSP) to ensure precise regulation of supply air, return air, fresh air and exhaust air. The effects of each adjustment are fed back to the control unit in real time through various sensors. When the deviation between the actual value and the target value exceeds the preset tolerance range, the control unit will dynamically adjust various control parameters to ensure system stability.

6. The intelligent constant temperature and pressure control system for a clean laboratory according to claim 5, characterized in that, The air supply actuator specifically includes: The air supply actuator receives the air supply volume adjustment command from the control unit. The air supply volume adjustment command is a fan speed adjustment signal calculated based on the PID control algorithm. The control unit calculates the control error vectors of temperature, pressure and wind speed, and generates corresponding air volume adjustment commands. The air volume adjustment commands are transmitted to the control unit through the industrial Ethernet interface. The control unit adjusts the speed of the fan motor according to the air volume adjustment command, and the relationship between the fan motor and the air volume is precisely calibrated through the fan's performance curve; The wind speed signal is converted into a digital signal by an analog-to-digital converter (ADC) and transmitted to the control unit via a multi-channel signal conditioning board. If the deviation between the actual wind speed and the set target air volume exceeds the preset tolerance range, the control unit will correct it by adjusting the fan speed. The fan speed is calculated in real time by a digital signal processor (DSP) and combined with the fan load curve to optimize the control signal in real time to ensure accurate control of the air volume.

7. The intelligent constant temperature and pressure control system for a clean laboratory according to claim 5, characterized in that, The return air execution unit specifically includes: The control unit receives the return air proportional adjustment command and calculates the target value of the return air vent opening according to the PID control algorithm; The control unit sends the target value to the return air actuator, which precisely adjusts the opening of the return air vent via an electric actuator; The electric actuator changes the opening of the return air vent according to the incoming control signal, precisely adjusting the return air ratio; The wind speed signal is processed by an analog-to-digital converter (ADC) and then transmitted to the control unit. The control unit performs real-time comparison based on the feedback signals of real-time monitored return air flow and return air vent opening. If the deviation between the return air vent opening and the set target return air ratio exceeds the tolerance range, the control unit will dynamically adjust the return air vent opening to ensure that the return air ratio is maintained within the set range.

Images