PID multichannel temperature control system based on neural network algorithm
Through the PID multi-channel temperature control system using neural network algorithm in nuclear magnetic resonance gyroscope, the contradiction between stability and dynamic performance and multi-channel temperature control interference in traditional PID temperature control methods is solved, and more efficient temperature control is achieved and improved The performance and startup speed of the gyroscope are achieved.
Patent Information
- Application Number
- CN202411066294.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-05
- Publication Date
- 2024-11-15
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional PID feedback temperature control methods are difficult to take into account the stability and dynamic performance of the NMR gyroscope, and there is significant mutual interference between multiple temperature controls, making it difficult to achieve ideal temperature distribution control.
Using a PID multi-channel temperature control system based on neural network algorithm, the PID control parameters are adjusted through the BP neural network, the multi-channel temperature control output is coordinated, and the system temperature rise curve is directly input, taking into account the delay characteristics of the temperature control system and the mutual connection between each channel. Influence.
It improves the temperature controllability and dynamic performance of the NMR gyroscope, overcomes the interference between the channels of the traditional PID algorithm, realizes more effective temperature distribution control, and improves the start speed and performance of the gyroscope.
Smart Images

Figure CN118963459A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of nuclear magnetic resonance gyroscope temperature control, and in particular to a PID multi-channel temperature control system based on a neural network algorithm. Background Art
[0002] As a new type of quantum inertial sensor device, the nuclear magnetic resonance gyroscope has the advantages of high precision, small size, and low power consumption. It can support the development of future intelligent and miniaturized transportation equipment such as drones, unmanned vehicles, and robots. Studies have found that the temperature change inside the gyroscope is the main factor restricting the current development of nuclear magnetic resonance gyroscopes. The traditional PID feedback temperature control method is difficult to take into account the stability and dynamic performance of the control system, and there is usually a contradiction between "rapidity" and "overshoot". In addition, the temperature distribution also has a significant impact on the gyroscope signal. To achieve the control of the temperature distribution, a multi-channel temperature control method is required. However, there is usually significant mutual interference between the temperature controls based on traditional PID feedback, and it is difficult to achieve the ideal effect.
[0003] Neural network is a mathematical model algorithm that imitates the behavioral characteristics of animal neural networks to perform distributed parallel information processing. It is composed of a large number of interconnected nodes. Neural networks have strong information integration capabilities and can coordinate and process multiple input information well, thus hopefully solving the mutual interference problem in traditional multi-channel temperature control methods and achieving more stable and effective temperature distribution control.
[0004] Therefore, it is necessary to provide a new PID multi-channel temperature control system based on neural network algorithm to solve the above technical problems. Summary of the invention
[0005] The technical problem solved by the present invention is to provide a PID multi-channel temperature control system based on a neural network algorithm, which can improve the controllability, stability and dynamic performance of the temperature of a nuclear magnetic resonance gyroscope.
[0006] In order to solve the above technical problems, the PID multi-channel temperature control system based on the neural network algorithm provided by the present invention includes a PT1000 temperature measuring resistor, a temperature transmitter, a data acquisition card, a computer, a voltage-controlled current source, and a heating structure, wherein the data acquisition card is connected to the computer, the voltage-controlled current source and the temperature transmitter, the heating structure is connected to the voltage-controlled current source, and the PT1000 temperature measuring resistor is connected to the heating structure; The PT1000 thermistor is used to measure the temperature of the heating structure and provide a resistance value corresponding to the temperature; The temperature transmitter is used to detect the resistance value of the temperature measuring resistor and convert the resistance change caused by temperature into a voltage signal; The data acquisition card converts the voltage signal corresponding to the temperature into a digital signal that can be processed by a computer program and inputs the digital signal into a computer, and converts the digital control signal generated by the computer program into a voltage signal that can control the output current of the voltage-controlled current source; The computer is used to implement the PID temperature control program and the BP neural network program; The voltage-controlled current source converts the voltage signal given by the acquisition card into a high-power current signal for driving the heating structure to heat; The heating structure converts the alternating current signal output by the voltage-controlled current source into thermal energy through the formula U=I2R, thereby changing the temperature of the heating structure.
[0007] Preferably, the number of the PT1000 thermistor, the temperature transmitter, the voltage-controlled current source and the heating structure is at least two.
[0008] Preferably, the temperature control method of the PID multi-channel temperature control system based on the neural network algorithm comprises the following steps: S1: Collect PID temperature control data; S2: Construct and train a BP neural network to obtain multi-channel PID control parameters through the heating curve; S3: Use the trained BP neural network to generate the PID parameters of the required temperature control system according to requirements.
[0009] Preferably, in S1, different PID control parameters and heating curve samples of multiple channels are collected.
[0010] Preferably, the data of each group of samples include the actual temperature rise curve, set temperature and PID parameters of each channel.
[0011] Preferably, in S2, the PID parameters of the required temperature control system are generated according to requirements using the trained BP neural network, including the following steps: S21: extracting a number of time-temperature points that can describe the characteristics of the heating curve from the heating curve as inputs of the BP neural network; S22: Determine the hidden layer and activation function of the BP network according to the actual training effect; S23: giving the output layer of the network according to the number of channels of the temperature control system; S24: After the neural network structure is designed, the data prepared in S21 is substituted into the network for training, and the network structure is optimized during the training process.
[0012] Preferably, in S3, the PID parameters of the required temperature control system are generated according to requirements using the trained BP neural network, including the following steps: S31: Generate an ideal heating curve for each channel according to actual needs.
[0013] S32: inputting the ideal heating curve into the neural network according to the trained BP neural network input format, and calculating the PID parameters of the corresponding channel; S33: Substitute the corresponding channel PID parameters obtained by the neural network into the actual multi-channel temperature control system for experimental verification.
[0014] Compared with the related art, the PID multi-channel temperature control system based on the neural network algorithm provided by the present invention has the following beneficial effects: 1) The present invention applies the neural network algorithm to the temperature control system of the nuclear magnetic resonance gyroscope, which solves the contradiction between stability and dynamic performance in the traditional PID temperature control method, and is of great significance to improving the starting speed of the gyroscope; 2) The present invention uses a neural network algorithm to coordinate multi-channel temperature control output, overcomes the interference between channels of the traditional PID algorithm, and realizes effective control of temperature distribution, which is of great significance to improving the performance of the gyroscope; 3) The neural network of the present invention directly inputs the system heating curve, fully considering the time delay characteristics of the temperature control system and the mutual influence between the channels, and can better realize multi-channel effective temperature control. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 A flow chart of a PID multi-channel temperature control system based on a neural network algorithm provided by the present invention; Figure 2 The principle block diagram of the PID multi-channel temperature control system based on the neural network algorithm provided by the present invention; Figure 3 A schematic diagram of the network structure of the BP neural network provided by the present invention; Figure 4 This is the simulation result of the dual-channel neural network PID control provided by the present invention. DETAILED DESCRIPTION
[0016] The present invention will be further described below in conjunction with the accompanying drawings and implementation modes.
[0017] Please refer to Figure 1 , Figure 2 , Figure 3 and Figure 4 ,in, Figure 1 A flow chart of a PID multi-channel temperature control system based on a neural network algorithm provided by the present invention; Figure 2 The principle block diagram of the PID multi-channel temperature control system based on the neural network algorithm provided by the present invention; Figure 3 A schematic diagram of the network structure of the BP neural network provided by the present invention; Figure 4The simulation result of the dual-channel neural network PID control provided by the present invention. The PID multi-channel temperature control system based on the neural network algorithm includes: it can be divided into a PID control part and a BP neural network part. The process of using the BP neural network to adjust the parameters of the PID controller is as follows: Figure 1 As shown, the specific steps of the PID multi-channel temperature control method based on the neural network algorithm provided in this embodiment are as follows: Step 1: Collect PID temperature control data. Collect enough multi-channel PID control parameters and heating curve samples. The data of each set of samples includes the actual heating curve of each channel, set temperature, PID parameters, etc.
[0018] Step 2: Construct and train a BP neural network that obtains multi-channel PID control parameters through the heating curve. First, extract several time-temperature points that can describe the characteristics of the heating curve from the heating curve as the input of the BP neural network. Then, determine the hidden layer and activation function of the BP network based on the actual training effect. Finally, give the output layer of the network based on the number of channels of the temperature control system. After the neural network structure is designed, the data prepared in the first step can be substituted into the training network, and the network structure can be optimized during the training process.
[0019] Step 3: Use the trained BP neural network to generate the PID parameters of the required temperature control system according to the requirements. First, generate the ideal heating curve of each channel according to the actual needs. Then, input the ideal heating curve into the neural network according to the trained BP neural network input format, and calculate the PID parameters of the corresponding channel. Finally, substitute the corresponding channel PID parameters obtained by the neural network into the actual multi-channel temperature control system for experimental verification.
[0020] The present invention also provides a PID multi-channel temperature control system based on a neural network algorithm: like Figure 1 As shown, first you need to build a temperature control system to collect PID temperature control data. The structure of the temperature control system is as follows Figure 2 As shown, it includes the following 6 parts: PT1000 temperature measuring resistor, temperature transmitter, data acquisition card, computer, voltage-controlled current source, and heating structure. Their functions are: PT1000 thermistor, used to measure the temperature of the heating structure and give a resistance value that corresponds to the temperature; Temperature transmitter, used to detect the resistance value of the temperature measuring resistor and convert the resistance change caused by temperature into a voltage signal; The data acquisition card converts the voltage signal corresponding to the temperature into a digital signal that can be processed by the computer program and inputs it into the computer, and converts the digital control signal generated by the computer program into a voltage signal that can control the output current of the voltage-controlled current source; Computer, used to implement PID temperature control program and BP neural network program; The voltage-controlled current source converts the voltage signal given by the acquisition card into a high-power current signal to drive the heating structure to heat; The heating structure converts the alternating current signal output by the voltage-controlled current source into thermal energy through the formula U=I2R, thereby changing the temperature of the heating structure.
[0021] Since it is a two-channel temperature control system, two sets of PT1000 thermistors, temperature transmitters, voltage-controlled current sources and heating structures are required.
[0022] according to Figure 2 After the circuit is connected by the logic shown, the temperature of the heating structure is first measured by the PT1000 thermistor, and converted into a resistance signal, which is transmitted to the temperature transmitter. The temperature transmitter converts the resistance signal corresponding to the temperature into a voltage signal. The acquisition card converts the voltage signal corresponding to the temperature into a digital signal that can be processed by the computer program and inputs it into the computer. The computer program processes the digital signal and outputs the digital signal for the temperature control of the two channels into the acquisition card. The acquisition card converts the digital signal into a voltage signal that can control the output current of the voltage-controlled current source and outputs it to the voltage-controlled current source of the two channels. The voltage-controlled current source converts the voltage signal output by the acquisition card into a corresponding current signal to drive the heating structure to heat, and the PT1000 measures the temperature of the heating structure at the same time, forming a closed-loop temperature feedback system. The temperature is continuously regulated by the real-time temperature signal fed back to the control program. At the same time, the control program writes the set temperature, real-time temperature and error data of the two channels collected into an Excel spreadsheet for collection.
[0023] Through the above process, the training data of the neural network is obtained. Next, the BP neural network of multi-channel PID control parameters is constructed and trained.
[0024] like Figure 3As shown in the figure, first, a number of time-temperature points that can describe the characteristics of the heating curve are extracted from the heating curve as the input of the BP neural network. In the two-channel temperature control system, 400 points are extracted from the heating curve of each channel, so a total of 2×400 time-temperature data points are extracted as the input of the BP neural network, that is, the input layer of the neural network needs to set 800 nodes. Then, according to the actual training effect, the best hidden layer and activation function of the BP network are found. Finally, the output layer of the network is given according to the number of channels of the temperature control system. The two-channel temperature control system requires 3 PID parameters for each channel, so the number of output nodes of the BP neural network is 2×3=6, and each heating curve corresponds to a set of PID parameters. After the neural network structure is designed, the data prepared in the first step can be substituted into the training network, and the network structure can be optimized during the training process. By substituting the original data into the trained BP neural network, the error analysis is performed between the PID parameters output by the neural network and the PID during training, and the hidden layer and activation function of the BP neural network are modified according to the error to find the BP neural network with the best training effect. Note that the present invention is not limited to a temperature control system with two channels, nor is it limited to taking only 400 data points for each heating curve.
[0025] Then, based on the properties of the exponential function, two exponential functions are set, where T is the real-time temperature and t is the time. Using the boundary conditions of and, it can be seen that T0 is the initial temperature, Ts is the set target temperature, and the coefficient a determines the speed of heating, thereby generating an ideal heating curve. Substituting this ideal heating curve as the input layer into the trained BP neural network can obtain two sets of ideal PID parameters to achieve multi-channel effective temperature control. Note that the ideal heating curve involved in the present invention is not limited to the exponential curve mentioned in this example, and all heating curves with physical meaning are within the scope of consideration of the present invention.
[0026] Figure 4 A set of two-channel temperature control simulation experiments implemented using the above ideas is given, where Curve 1 is the ideal target temperature rise curve, Curve 2 is the curve corresponding to the PID parameters output by the neural network model, and Curve 3 is the corresponding temperature rise curve of the PID parameters in the sample obtained by the traditional tuning method. It can be seen that the parameters trained by the neural network can basically reproduce the ideal temperature rise curve, and the temperature rise rate is significantly higher than the temperature rise curve corresponding to the traditional PID tuning parameters. This illustrates the effectiveness of the method proposed in the present invention.
[0027] In summary, the method proposed in the present invention first uses the BP neural network to learn the characteristics of the temperature control system through the heating curve, and then uses the ideal heating curve to give the PID control parameters that produce the ideal heating curve through the trained BP neural network. It is suitable for complex temperature control processes that are difficult to establish an accurate mathematical model, and has the advantages of better robustness, smaller steady-state error, shorter heating time, etc. Compared with the related art, the PID multi-channel temperature control system based on the neural network algorithm provided by the present invention has the following beneficial effects: 1) The present invention applies the neural network algorithm to the temperature control system of the nuclear magnetic resonance gyroscope, which solves the contradiction between stability and dynamic performance in the traditional PID temperature control method, and is of great significance to improving the starting speed of the gyroscope; 2) The present invention uses a neural network algorithm to coordinate multi-channel temperature control output, overcomes the interference between channels of the traditional PID algorithm, and realizes effective control of temperature distribution, which is of great significance to improving the performance of the gyroscope; 3) The neural network of the present invention directly inputs the system heating curve, fully considering the time delay characteristics of the temperature control system and the mutual influence between the channels, and can better realize multi-channel effective temperature control.
[0028] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A PID multi-channel temperature control system based on a neural network algorithm, characterized in that: include: PT1000 temperature measuring resistor, temperature transmitter, data acquisition card, computer, voltage-controlled current source, heating structure, the data acquisition card is connected to the computer, the voltage-controlled current source and the temperature transmitter, the heating structure is connected to the voltage-controlled current source, and the PT1000 temperature measuring resistor is connected to the heating structure; The PT1000 thermistor is used to measure the temperature of the heating structure and provide a resistance value corresponding to the temperature; The temperature transmitter is used to detect the resistance value of the temperature measuring resistor and convert the resistance change caused by temperature into a voltage signal; The data acquisition card converts the voltage signal corresponding to the temperature into a digital signal that can be processed by a computer program and inputs the digital signal into a computer, and converts the digital control signal generated by the computer program into a voltage signal that can control the output current of the voltage-controlled current source; The computer is used to implement the PID temperature control program and the BP neural network program; The voltage-controlled current source converts the voltage signal given by the acquisition card into a high-power current signal for driving the heating structure to heat; The heating structure converts the alternating current signal output by the voltage-controlled current source into thermal energy through the formula U=I2R, thereby changing the temperature of the heating structure.
2. The PID multi-channel temperature control system based on the neural network algorithm according to claim 1 is characterized in that: The number of the PT1000 thermistor, the temperature transmitter, the voltage-controlled current source and the heating structure are at least two.
3. The PID multi-channel temperature control system based on the neural network algorithm according to claim 1 is characterized in that: The temperature control method of the PID multi-channel temperature control system based on the neural network algorithm includes the following steps: S1: Collect PID temperature control data; S2: Construct and train a BP neural network to obtain multi-channel PID control parameters through the heating curve; S3: Use the trained BP neural network to generate the PID parameters of the required temperature control system according to requirements.
4. The PID multi-channel temperature control system based on the neural network algorithm according to claim 3 is characterized in that: In S1, different PID control parameters and heating curve samples of multiple channels are collected.
5. The PID multi-channel temperature control system based on the neural network algorithm according to claim 4 is characterized in that: The data of each group of samples include the actual temperature rise curve of each channel, the set temperature and the PID parameters.
6. The PID multi-channel temperature control system based on the neural network algorithm according to claim 3 is characterized in that: In S2, the trained BP neural network is used to generate the PID parameters of the required temperature control system according to the requirements, including the following steps: S21: extracting a number of time-temperature points that can describe the characteristics of the heating curve from the heating curve as inputs of the BP neural network; S22: Determine the hidden layer and activation function of the BP network according to the actual training effect; S23: giving the output layer of the network according to the number of channels of the temperature control system; S24: After the neural network structure is designed, the data prepared in S21 is substituted into the network for training, and the network structure is optimized during the training process.
7. The PID multi-channel temperature control system based on the neural network algorithm according to claim 3 is characterized in that: In S3, the trained BP neural network is used to generate the PID parameters of the required temperature control system according to the requirements, including the following steps: S31: Generate an ideal heating curve for each channel according to actual needs; S32: inputting the ideal heating curve into the neural network according to the trained BP neural network input format, and calculating the PID parameters of the corresponding channel; S33: Substitute the corresponding channel PID parameters obtained by the neural network into the actual multi-channel temperature control system for experimental verification.
Citation Information
Patent Citations
Multi-heating point coordinate temperature control device for realizing uniform temperature field of gyroscope
CN104950950A
Temperature control method of dynamically predicting PID based on RBF neural network
CN108958020A
HVPE temperature control system and method based on fuzzy control
CN112680789A
Temperature self-tuning method and system for GaN substrate growth heating furnace
CN112782970A
Anti-interference temperature control method for heat treatment of aerospace complex casting parts
CN116987878A
Cited By
Photoelectric pod temperature control system based on SOC and use method thereof
CN119960518A
Gain dynamic adjustment method, device and system for controlling temperature of special gas equipment
CN121326039A
