Intelligent constant temperature control method based on low temperature constant temperature tank

Abstract

The application discloses an intelligent constant-temperature control method based on a low-temperature constant-temperature tank, and comprises the following steps: S1, providing a power supply through a power supply circuit; S2, placing four PT1000 platinum resistors in four corners of a liquid to be controlled, measuring different temperature measuring points through an analog-to-digital converter, and performing mathematical conversion on four high-precision voltage data collected by the analog-to-digital converter through a microcontroller to obtain real-time temperatures of four points of the liquid to be controlled; S3, performing temperature control through a temperature control device and a PID algorithm based on a genetic algorithm, and collecting a large number of data sets changing with time in constant-temperature control; S4, constructing a temperature control prediction model based on a convolutional neural network and a gated recurrent unit through the data sets based on deep learning, and adding the model to a PID control link; and S5, performing data display and uploading through the microcontroller. According to the application, the intelligent control problem of water constant-temperature is solved, the temperature control speed is improved, the temperature control precision is improved, and the requirement of high-precision rapid constant-temperature control is met.

Description

Technical Field

[0001] This invention relates to the technical field of constant temperature control and electronic automation control, and in particular to an intelligent constant temperature control method based on a low-temperature constant temperature bath. Background Technology

[0002] With the advancement of science, the precision requirements for isothermal environments in metrology, biochemistry, and materials science are becoming increasingly stringent. Existing isothermal control methods primarily utilize traditional PID control, which suffers from problems such as large overshoot and excessively long temperature control times. Furthermore, in industrial production processes where precise mathematical models are difficult to establish, the control precision often fails to meet the accuracy and stability requirements of high-precision time-varying systems, significantly hindering the conduct of scientific experiments. Summary of the Invention

[0003] To address the shortcomings of existing technologies, the present invention aims to provide an intelligent temperature control method based on a low-temperature thermostatic bath, solving the problem of intelligent temperature control of water, improving temperature control speed and accuracy, and meeting the requirements of high-precision and rapid temperature control. To achieve the above-mentioned objectives and other advantages of the present invention, an intelligent temperature control method based on a low-temperature thermostatic bath is provided, comprising:

[0004] S1. Power is supplied through the power supply circuit;

[0005] S2. Place four PT1000 platinum resistance thermometers at the four corners of the liquid to be controlled, measure different temperature points using an analog-to-digital converter, and perform mathematical conversion on the four high-precision voltage data collected by the analog-to-digital converter using a microcontroller to obtain the real-time temperature of the liquid at the four points to be controlled.

[0006] S3. Temperature control is performed using a temperature control device and a PID algorithm based on a genetic algorithm, and a large dataset of data changing over time during constant temperature control is collected.

[0007] S4. Based on deep learning, a temperature control prediction model based on convolutional neural networks and gated loop units is constructed using a dataset and added to the PID control loop.

[0008] S5. Displays and uploads data via a microcontroller.

[0009] Preferably, in step S2, each PT1000 platinum resistor is connected to a balanced bridge with three wires. The balanced bridge signal is connected to a differential operational amplifier, the differential operational amplifier signal is connected to an analog-to-digital converter, and the analog-to-digital converter signal is connected to a microcontroller.

[0010] Preferably, step S2 specifically includes the following steps:

[0011] S21. Place four PT1000 platinum resistance thermometers into the four corners of the liquid to be controlled, and connect each PT1000 platinum resistance thermometer to the bridge circuit using a three-wire connection.

[0012] S22. Acquire the differential voltage of the balanced bridge, amplify it differentially through an operational amplifier, and finally output it to a high-precision analog-to-digital converter.

[0013] S23. The microcontroller then performs mathematical conversion on the four high-precision voltage data collected by the analog-to-digital converter to obtain the high-precision real-time temperature of the four points.

[0014] Preferably, in step S3, the value obtained in the microcontroller is the output code of the analog-to-digital converter. This output code has an equation relationship with the temperature. Let x represent the output code, and we can obtain formula (1):

[0015]

[0016] This formula is used to obtain the temperature data of a temperature measurement point, where a, b, c, and d are all constants, and T is the temperature of the temperature measurement point;

[0017] A mathematical model was constructed using four temperature measurement points, and based on this mathematical model:

[0018]

[0019] The final temperature, T, is calculated. all Z represents the final temperature. n Assuming the weight of the nth temperature measurement point, adjust and calibrate the four Z values ​​based on the actual temperature field. n The value of a. n b n c n d n These represent the constant parameters in the temperature formula for the nth temperature measurement point; C is a constant parameter.

[0020] Preferably, the temperature control device in step S3 includes a heating rod, a copper coil wound around the edge of the medium container, a compressor, and multiple semiconductor cooling chips. The compressor provides a constant power cooling function. The heating rod acts as the main actuator, and the semiconductor cooling chips act as auxiliary actuators for more precise control. The power of the heating rod is controlled by a PID algorithm, thereby keeping the temperature constant at the set value. The microprocessor adjusts the heating power of the heating rod and the semiconductor cooling chips by adjusting the duty cycle of PWM.

[0021] Preferably, a large amount of temperature control dataset is obtained, which has five dimensions: set temperature, current temperature of the medium, ambient temperature, medium volume, and medium type. The medium type includes water and oil. Each data point has these five dimensions. By collecting a large amount of dataset, historical temperature data points under different set temperatures, different ambient temperatures, different medium volumes, and different medium types are obtained. The large amount of data is divided into a dataset and a validation set for training. A temperature control prediction model based on convolutional neural network and gated loop unit is constructed and finally added to the PID control loop to improve the speed and accuracy of temperature control.

[0022] Preferably, the balanced bridge includes three fixed resistors and one thermistor platinum resistor, wherein the three fixed resistors are all 1K ohms; the differential operational amplifier has a gain of 20 times, and the differential operational amplifier includes two 20K ohm resistors, two 1K ohm resistors and one operational amplifier.

[0023] Preferably, in formula (2), temperature T all The temperature range is divided into fourteen levels from -40℃ to 100℃, with α in each level. n b n c n d n All parameters are refitted for that temperature range. The constant parameters are stored in the array of the microcontroller. After the microcontroller obtains the result of the digital-to-analog converter, it calculates the temperature value through the mathematical model.

[0024] Compared with existing technologies, the advantages of this invention are as follows: By collecting a large dataset, historical temperature data points under different environmental temperatures, media volumes, and media types are obtained. Then, the large amount of data is divided into a dataset and a validation set for training, constructing a temperature control prediction model based on a convolutional neural network (CNN) and a gated recurrent unit (GRU), which is ultimately added to the PID control loop to improve the speed and accuracy of temperature control. A large amount of relatively uniform temperature data (the current temperature fluctuates slightly around the set temperature) is collected and used as input data (five-dimensional data) to construct a prediction model. This model is added to the PID control loop, and is activated when the small temperature oscillation lasts for a longer period. The model predicts the temperature data for the next moment based on the temperature data from a period before the current time. Adding this to the PID control loop allows the actuator to respond to changes more quickly, resulting in more precise control and less oscillation. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the overall temperature control circuit of the intelligent temperature control method based on a low-temperature thermostatic bath according to the present invention.

[0026] Figure 2This is a schematic diagram of the power generation circuit for the intelligent constant temperature control method based on a low-temperature constant temperature bath according to the present invention.

[0027] Figure 3 This is a schematic diagram of a PT1000 resistor, a balanced bridge operational amplifier, and an analog-to-digital converter combined circuit for the intelligent temperature control method based on a low-temperature constant temperature bath according to the present invention.

[0028] Figure 4 This is a schematic diagram of a temperature control device for an intelligent temperature control method based on a low-temperature thermostatic bath according to the present invention.

[0029] Figure 5 This is a basic framework diagram of the CNN-GRU temperature prediction model based on the intelligent constant temperature control method of the low-temperature constant temperature bath according to the present invention.

[0030] Figure 6 This is a schematic diagram of the PID algorithm after incorporating a predictive model in the intelligent constant temperature control method based on a low-temperature constant temperature bath according to the present invention.

[0031] Figure 7 The present invention provides a flowchart of the intelligent temperature control method based on a low-temperature thermostatic bath according to the present invention.

[0032] Figure 8 This is a schematic diagram illustrating the data display and uploading of the intelligent constant temperature control method based on a low-temperature constant temperature bath according to the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Reference Figure 1-8 An intelligent temperature control method based on a low-temperature thermostatic bath includes the following steps: 1) generating a power supply, such as... Figure 2 As shown

[0035] A 220V AC input is obtained, then a stable 12V DC is obtained through a 220V to 12V transformer, a KBP307 rectifier module, and an L7812CV voltage regulator module. Next, a 5V DC output is obtained through an L7805CV. Finally, a 3.3V output is obtained through an LM1117S-3.3, and a 2.5V output is obtained through a TL431. The 3.3V supplies power to the microcontroller and operational amplifiers, and the 2.5V provides a reference voltage for the analog-to-digital converter and the bridge circuit. C1, C2, and C3 are 1000µF electrolytic capacitors; C4, C6, C7, C9, C11, and C13 are 100nm ceramic capacitors; C5 is a 220µF electrolytic capacitor; and C8, C10, C12, and C14 are 47µF electrolytic capacitors.

[0036] 2) Obtain the real-time temperature of the liquid to be controlled through multi-point measurement.

[0037] A combination circuit of PT1000 resistors, balanced bridge operational amplifiers, and analog-to-digital converters, such as... Figure 3 As shown, the circuit consists of a microcontroller, a high-precision analog-to-digital converter (AD7791), an operational amplifier (HGV8551M5), and resistors. Resistors R1 and R2 are both 30kΩ, R3 is a 1kΩ resistor, Rpt is a PT1000 platinum resistance thermometer, and r1, r2, and r3 are the lead resistances of the PT1000 platinum resistance thermometer. These resistors form a balanced bridge. Connecting the PT1000 platinum resistance thermometer to the bridge with three wires eliminates the influence of lead resistance on the measurement, making the resistance measurement more accurate. By measuring the differential voltage output of the balanced bridge, the resistance change of the PT1000 platinum resistance thermometer is reflected, thus reflecting the temperature change. The operational amplifier (HGV8551M5) and resistors form a differential amplifier. R8 and R9 are both 1kΩ resistors, and R10 and R11 are both 20kΩ resistors. The differential amplification factor is R11 divided by R9, i.e., 20 times amplification. A reference voltage (Vref) is then connected to the bridge circuit, creating a voltage difference across the PT1000 platinum resistance thermometer and across the bridge's internal resistance. As temperature changes, the resistance of the platinum resistance thermometer also changes, leading to a change in the resistive voltage division. The varying differential voltage across the balanced bridge is then acquired and differentially amplified by an operational amplifier (HGV8551M5 / TR). Finally, the output is sent to a high-precision analog-to-digital converter (ADC), model AD7791. A microcontroller (MCU) STM32F103ZET6 performs mathematical conversion on the high-precision voltage data acquired by the AD7791 ADC to obtain a high-precision real-time temperature. Using this method, temperature detection points are deployed at four points on the controlled liquid; multi-point measurement makes control more precise. The microcontroller calculates the measured temperature based on the temperature formula for the platinum resistance thermometer and the bridge resistance calculation formula.

[0038] 3) Calculate real-time temperature

[0039] In a microcontroller (MCU), what we actually obtain is the ADC conversion value, OutputCode. This value has an equation relationship with temperature. Let x represent OutputCode, and we can obtain the formula:

[0040]

[0041] This formula is used to obtain the temperature data of a temperature measurement point, where a, b, c, and d are all constants.

[0042] A mathematical model is constructed using four temperature measurement points, and the following is based on this mathematical model:

[0043]

[0044] The final temperature is calculated, where T is the temperature. Z n Assuming the weight of the nth temperature measurement point, adjust and calibrate the four Z values ​​based on the actual temperature field. n The value of a. n b n c n d n These represent the constant parameters in the temperature formula for the nth temperature measurement point. C is a constant parameter.

[0045] 4) A semiconductor cooling chip is added to the compressor refrigeration and heating rod heating systems to improve the precision of control. Finally, a PID control algorithm based on a genetic algorithm is used to control the temperature, and temperature control data is collected to construct a dataset. For example... Figure 4 The diagram shows a schematic of the temperature control device. Four PT1000 resistors are placed at the four corners of the working chamber to measure the temperature in real time. The compressor and copper coil are used for cooling, the heating rod is used for heating, and the semiconductor cooling chip is used for more precise local temperature control. The stirring paddle is used to stir the liquid to make the temperature control more uniform. All of the above devices are placed in the stirring chamber. The working chamber and the stirring chamber are separated by a partition, and there are gaps at the bottom to allow the liquid to flow, achieving a rapid and uniform temperature control effect.

[0046] First use Figure 4 The temperature control device shown uses a PID algorithm based on a genetic algorithm for temperature control and collects a large dataset of time-varying data during constant temperature control. The large dataset is divided into training and validation sets in an 8:2 ratio. After simple data preprocessing (filtering, normalization, etc.), the data is input into a system such as... Figure 5 The CNN-GRU temperature prediction model shown above. The input data first passes through a CNN network to extract its high-dimensional features, and then enters a GRU network to predict the time series data. The output of the GRU network passes through a fully connected layer to obtain the predicted value of the temperature data.

[0047] like Figure 6 The diagram shows the PID control algorithm after incorporating the temperature prediction model mentioned above. Compared with traditional PID control, the addition of the prediction model after real-time temperature acquisition makes the control faster and more accurate.

[0048] like Figure 7 For use Figure 4 The temperature control flowchart of the constant temperature control device is as follows: First, temperature is acquired; then, PID control with a predictive model is applied; finally, temperature is acquired again, and this cycle is repeated to achieve high-precision constant temperature control.

[0049] Step 3) completes the temperature acquisition stage of the PID control. In this method, the actuator in the PID control stage uses a combined control of a heating rod and a compressor, further refined by incorporating a thermoelectric cooler for finer control. The compressor provides constant-power cooling, the heating rod acts as the primary actuator, and the thermoelectric cooler precisely controls the temperature at four measurement points. A PID algorithm based on a genetic algorithm controls the power of the heating rod and the thermoelectric cooler, thus maintaining the temperature at the set value. The microprocessor (MCU) adjusts the heating power by regulating the duty cycle via PWM. A large amount of data is collected during the temperature control process, forming a temperature control dataset.

[0050] 5) Based on deep learning, a predictive model is built using a temperature control dataset and incorporated into PID control.

[0051] A temperature control prediction model based on a convolutional neural network (CNN) and a gated recurrent unit (GRU) is constructed and incorporated into the PID control loop. The model is continuously trained and iterated using collected temperature control data to improve control accuracy.

[0052] To collect a complete temperature control process, the steps are as follows:

[0053] a) First, set the target temperature.

[0054] b) Collect the ambient temperature (i.e., the temperature at the start of the temperature control phase).

[0055] c) Start temperature control using a traditional PID temperature control algorithm.

[0056] d) Collect temperature data throughout the entire process and incorporate it into the temperature control dataset.

[0057] Through the above steps, a large temperature control dataset is obtained, which has four dimensions: current medium temperature, ambient temperature, medium volume, and medium type. The medium type includes both water and oil. Each data point has these four dimensions. By collecting a large amount of data, historical temperature data points under different ambient temperatures, different medium volumes, and different medium types are obtained. Then, the large amount of data is divided into a dataset and a validation set for training, constructing a temperature control prediction model based on a convolutional neural network (CNN) and a gated recurrent unit (GRU). This model is ultimately incorporated into the PID control loop to improve the speed and accuracy of temperature control.

[0058] The specific implementation method is as follows: Collect a large amount of relatively uniform temperature data (the current temperature fluctuates slightly around the set temperature), and use this data as input data (five-dimensional data) to build a predictive model. This model is then added to the PID control loop. When the period of slight temperature oscillation exceeds a certain threshold, the model is activated. This model predicts the temperature data for the next moment based on temperature data from a period preceding the current time. Adding it to the PID control loop allows the actuator to respond to changes more quickly, resulting in more precise control and less oscillation.

[0059] 6) Data display and uploading:

[0060] The microcontroller (MCU) controls the OLED display module via the SPI protocol to display the current actual temperature value (PV) and the set temperature value (SV). At the same time, the microcontroller (MCU) controls the WIFI module via the serial port protocol to send the above data to the IoT cloud platform.

[0061] like Figure 8 As shown, the microcontroller (MCU) controls the WIFI module through a serial communication protocol to send the data obtained by the MCU through the front-end acquisition circuit to the IoT cloud platform. The TX port is used to send data, and the RX port is used to receive data. At the same time, the microcontroller (MCU) controls the OLED display through the SPI protocol to display the real-time measured temperature and the set temperature. CS is the chip select signal port, MOSI is the port for the microcontroller (MCU) to send data to the OLED display, and SCK is the clock signal port.

[0062] The number of devices and processing scale described herein are for the purpose of simplifying the description of the invention, and applications, modifications and variations thereof will be apparent to those skilled in the art.

[0063] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. An intelligent temperature control method based on a low-temperature thermostatic bath, characterized in that, Includes the following steps: S1. Power is supplied through the power supply circuit; S2. Place four PT1000 platinum resistance thermometers at the four corners of the liquid to be controlled. Measure different temperature points using an analog-to-digital converter (ADC). The microcontroller performs mathematical conversion on the four high-precision voltage data acquired by the ADC to obtain the real-time temperature of the four points of the liquid to be controlled. In step S2, each PT1000 platinum resistance thermometer is connected to a balanced bridge circuit using a three-wire connection. The balanced bridge signal is connected to a differential operational amplifier, the differential operational amplifier signal is connected to the ADC, and the ADC signal is connected to the microcontroller. Step S2 specifically includes the following steps: S21. Place four PT1000 platinum resistance thermometers into the four corners of the liquid to be controlled, and connect each PT1000 platinum resistance thermometer to the bridge circuit using a three-wire connection. S22. Acquire the differential voltage of the balanced bridge, amplify it differentially through an operational amplifier, and finally output it to a high-precision analog-to-digital converter. S23. The microcontroller then performs mathematical conversion on the four high-precision voltage data collected by the analog-to-digital converter to obtain the high-precision real-time temperature of the four points. S3. Temperature control is performed using a temperature control device and a PID algorithm based on a genetic algorithm, and a large dataset of data changing over time during constant temperature control is collected; the value obtained in the microcontroller in step S3 is the conversion value of the analog-to-digital converter. This conversion value has an equation relationship with temperature, denoted by x. Formula (1) can be obtained: ； This formula is used to obtain the temperature data for a single temperature measurement point, where a, b, c, and d are all constants. The temperature at the measuring point; A mathematical model was constructed using four temperature measurement points, and based on this mathematical model: ； The final temperature was calculated. The final temperature; The weights for the nth temperature measurement point are adjusted and calibrated based on the actual temperature field conditions. The value; , , , These represent the constant parameters in the temperature formula for the nth temperature measurement point; It is a constant parameter; S4. Based on deep learning, a temperature control prediction model based on convolutional neural networks and gated loop units is constructed using a dataset and added to the PID control loop. S5. Displays and uploads data via a microcontroller.

2. The intelligent temperature control method based on a low-temperature thermostatic bath as described in claim 1, characterized in that, In step S3, the temperature control device includes a heating rod, a copper coil wound around the edge of the medium container, a compressor, and multiple semiconductor cooling chips. The compressor provides a constant power cooling function. The heating rod acts as the main actuator, and the semiconductor cooling chips act as auxiliary actuators for more precise control. The power is controlled by a PID algorithm to keep the temperature constant at the set value. The microprocessor adjusts the heating power of the heating rod and the semiconductor cooling chips by adjusting the duty cycle through PWM.

3. The intelligent temperature control method based on a low-temperature thermostatic bath as described in claim 2, characterized in that, A large dataset of temperature control data was obtained, which has five dimensions: set temperature, current temperature of the medium, ambient temperature, medium volume, and medium type. The medium type includes water and oil. Each data point has these five dimensions. Through the large dataset collection, historical temperature data points under different set temperatures, different ambient temperatures, different medium volumes, and different medium types were obtained. The large amount of data was divided into a dataset and a validation set for training. A temperature control prediction model based on convolutional neural network and gated loop unit was constructed and finally added to the PID control loop to improve the speed and accuracy of temperature control.

4. The intelligent temperature control method based on a low-temperature thermostatic bath as described in claim 1, characterized in that, The balanced bridge includes three fixed resistors and one thermistor platinum resistance, all of which are 1K ohms; the differential operational amplifier has a gain of 20 and includes two 20K ohm resistors, two 1K ohm resistors and one operational amplifier.

5. The intelligent temperature control method based on a low-temperature thermostatic bath as described in claim 1, characterized in that, Temperature in formula (2) The temperature range is divided into fourteen levels, from -40℃ to 100℃, with each level containing... , , , All parameters are refitted for that temperature range. The constant parameters are stored in the array of the microcontroller. After the microcontroller obtains the result of the digital-to-analog converter, it calculates the temperature value through the mathematical model.

Images