Online monitoring system for hydrogen production reaction and its neural network fuzzy PID self-heating control method

Abstract

The present invention relates to the technical field of online monitoring systems, and discloses an online monitoring system for a hydrogen production reaction and a neural network fuzzy PID self-heating control method thereof. The system comprises an online monitoring system, which is composed of a device layer, a network layer and an application layer. The device layer can transmit data collected by the entire hydrogen production reaction system and the device operating status to intelligent devices such as a laptop computer, a touch screen, and a monitoring center in the application layer through the network layer, and the application layer can also transmit control signals to the device layer through the network layer. The entire online monitoring system can collect device status parameters of the hydrogen production reaction system in real time, record the operating status of each device, display these parameters and status in real time, and control the devices. By performing online monitoring on the entire hydrogen production reaction system, the hydrogen production reaction system can be controlled to be operated quickly, accurately and safely.

Description

Technical Field

[0001] The present invention relates to the technical field of online monitoring systems, in particular to an online monitoring system for a hydrogen production reaction and a neural network fuzzy PID self-heating control method thereof. Background Art

[0002] Hydrogen energy boasts clean, pollution-free operation, high energy density, and rapid energy conversion. Major countries around the world are continuously rolling out hydrogen energy policies, particularly in the transportation sector, where hydrogen energy has its largest application. Methanol steam reforming is a popular and mature hydrogen production technology, offering low reforming temperatures, high hydrogen production rates, and low costs. Controlling autoheating in the hydrogen production reaction system is a crucial component of this technology. In recent years, hydrogen production reaction systems have seen rapid development. Chinese invention patents (Application No. 202222137962.4) and (Application No. 202110398655.9) disclose hydrogen production reaction systems and related devices. However, these systems lack sufficient components to withstand complex operating environments. Furthermore, recent patents offer limited information on methods for controlling the temperature of the hydrogen production reaction. Currently, PID is the mainstream temperature control method. Although PID controllers have many advantages, they also have inherent flaws. The relationship between P, I, and D is a linear combination, which leads to problems such as "overshoot" and "oscillation" in the system. Existing mathematical tools are not enough to support us in finding a "general solution." In practical applications, the controlled processes are often complex, with high nonlinearity, time-varying uncertainty, and pure hysteresis. Process parameters and even model structures vary over time and the operating environment. In addition, the control system has a low level of intelligence and few human-computer interaction modules, which ultimately lead to the system failing to meet control requirements.

[0003] Therefore, in order to improve the temperature control accuracy of the hydrogen production autothermal reforming reaction, cope with various complex working environments, enhance the intelligence of the control system, and better carry out human-computer interaction, it is necessary to invent a reliable, highly applicable, high-control-accuracy and highly intelligent line monitoring system and autothermal control method. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this paper discloses an online monitoring system for a hydrogen production reaction and a neural network fuzzy PID self-heating control method therefor. The system consists of a device layer, a network layer, and an application layer. The entire system is developed using LabVIEW. The device layer comprises the hydrogen production reaction system, a central controller, a data acquisition module, and an execution module. The central controller, comprised of a data acquisition card and a field control terminal, calculates, stores, and displays data collected by the data acquisition module, uploads the processed data to a cloud server via the network, and outputs corresponding control commands to the execution module. The online monitoring system supports operator online modification of control instructions and related algorithms. The network layer transmits information over the network, with routers relaying signals. The application layer utilizes a client / server model. The system employs a neural network fuzzy PID self-heating control method to control the self-heating of the hydrogen production reaction system. The neural network fuzzy PID control algorithm, deployed on a server, controls the temperature of the hydrogen production reaction system based on the difference between the preset temperature and the current hydrogen production reactor temperature and the rate of change of this difference, stabilizing the reaction temperature to the preset temperature. The device layer can transmit data collected by the entire hydrogen production reaction system, equipment operating status, and alarm information to the application layer's client's laptop computer, touch screen, monitoring center, and other intelligent devices through the network layer. The application layer can also transmit control instructions to the device layer through the network layer, allowing operators to always understand the status of the hydrogen production reaction system even when they are not on site, and to perform real-time control operations for various emergencies. The entire intelligent control system can collect parameters such as temperature, pressure, and flow of the hydrogen production reaction system in real time, record the operating status of each device, and display these parameters and status in real time. By monitoring the entire hydrogen production reaction system in real time, the hydrogen production reaction system can be controlled quickly, accurately, and safely. The present invention improves the temperature control accuracy of the hydrogen production reaction and monitors the hydrogen production reaction process in real time, making the control of the hydrogen production reaction process safer, more accurate, and intelligent. The hydrogen production reaction online monitoring system and its neural network fuzzy PID self-heating control method provided by the present invention are characterized in that the online monitoring system is composed of a device layer, a network layer, and an application layer. The online monitoring system and its neural network fuzzy PID self-heating control method are developed in LabVIEW. The equipment layer consists of a hydrogen production reaction system, a central controller, a data acquisition module and an execution module; the central controller consists of a data acquisition card and a field control terminal. The data acquisition card and the field control terminal communicate via the USB protocol. The acquisition and control devices include a temperature acquisition and control device, a pressure acquisition device, a flow acquisition and control device, a cooling device, a gas storage device, a gas-liquid separation device, a liquid evaporation device, a gas-liquid mixing device and a shut-off valve.They can collect the parameters of the hydrogen production reaction system in real time and transmit them to the on-site control terminal to display the status of the hydrogen production reaction system. Based on the neural network fuzzy PID control algorithm, they output the control parameters of each device to achieve rapid response, precise control and safe operation of the hydrogen production reaction system.

[0005] refer to Figure 2 , the detailed connections of the equipment layer are as follows: the air compressor is connected to the first stop valve, then to the mass flow controller, then to the mixer, and finally to the catalytic combustion chamber of the reforming hydrogen production reactor; the methanol solution storage bottle is connected to the second stop valve, then to the first peristaltic pump, then to the first evaporator; finally to the mixer; the methanol water solution storage bottle is connected to the third stop valve, then to the second peristaltic pump, then to the second evaporator, and finally to the methanol water vapor reforming chamber of the reforming hydrogen production reactor; the first pressure transmitter is inserted between the mixer and the pipeline of the reforming hydrogen production reactor to measure the pressure between the pipelines; the second pressure transmitter is inserted between the second evaporator and the pipeline of the reforming hydrogen production reactor to measure the pressure between the pipelines; the other end of the methanol water vapor reforming chamber of the reforming hydrogen production reactor is connected to the first heat exchanger, then to the first condenser; and then to the first gas-liquid separation device The vortex flowmeter is connected to the first heat exchanger, then to the drying tube, and then to the vortex flowmeter, and finally two channels are separated from the vortex flowmeter, one of which is connected to the gas storage cylinder, and the other is first connected to the third stop valve and then to the sample detection bag; the other end of the catalytic combustion chamber of the reforming hydrogen production reactor is first connected to the second heat exchanger, then to the second condenser, and finally to the second gas-liquid separation device; the third pressure transmitter is inserted between the first heat exchanger and the pipeline of the reforming hydrogen production reactor to measure the pressure between the pipelines; the fourth pressure transmitter is inserted between the second heat exchanger and the pipeline of the reforming hydrogen production reactor to measure the pressure between the pipelines; a J-type thermocouple is inserted in the reforming hydrogen production reactor, the J-type thermocouple is connected to the temperature transmitter, the temperature transmitter is connected to the data acquisition card, the data acquisition card is connected to the field control terminal; a heating rod is inserted in the reforming hydrogen production reactor, the heating rod is connected to the solid-state relay, and the solid-state relay is connected to the data acquisition card.

[0006] The measurement parameters of the hydrogen production reaction system include the instantaneous flow rates of the air compressor, the methanol solution storage bottle, and the methanol-water solution storage bottle channel, the instantaneous flow rate passing through the vortex flowmeter, the instantaneous pressures of the four channels entering and flowing out of the reforming hydrogen production reactor, the instantaneous temperatures of the first evaporator, the second evaporator, the mixer, the reforming hydrogen production reactor, the first heat exchanger, the second heat exchanger, the first condenser, and the second condenser, the current gas storage capacity of the gas cylinder, and the temperature measured by the J-type thermocouple.

[0007] The control parameters output by the control program include electrical signals for controlling the solid-state relay switch, electrical signals for the mass flow controller opening size, electrical signals for all stop valve switches, and alarm signals for corresponding equipment.

[0008] The state variables displayed by the field control terminal include the instantaneous flow rates of the air compressor, the methanol solution storage tank, and the methanol-water solution storage tank channels; the instantaneous flow rate measured by the vortex flowmeter; the instantaneous pressures of the four channels entering and exiting the reforming hydrogen production reactor; the instantaneous temperatures of the evaporator, evaporator, mixer, reforming hydrogen production reactor, heat exchanger, heat exchanger, condenser, and condenser; the current gas storage capacity of the gas tank; the temperature measured by the J-type thermocouple; the Ki, Ti, and Td values output by the current neural network fuzzy PID algorithm; and the overall PID output value.

[0009] The start and stop working status of all the stop valves, peristaltic pumps, and evaporators are controlled by the digital signals output by the field control terminal. All pressure transmitters can collect the current pressure status in the pipeline at all times, output 4-20mA analog current signals, and connect to the field control terminal through a data acquisition card. The collected pressure is displayed on the field control terminal.

[0010] The mass flow controller can measure the flow rate of the current pipeline and control the valve opening size to control the flow rate through the current pipeline. It is connected to the field control terminal through a data acquisition card and outputs a 0-5V analog voltage. The measured flow rate is displayed on the field control terminal, and the valve opening size control signal is controlled by the program outputting a 0-5V analog signal.

[0011] The vortex flowmeter measures the flow rate of gas generated in the methanol steam reforming chamber, outputting a 4-20 mA analog current. This is connected to a field control terminal via a data acquisition card, and the measured flow rate is displayed on the field control terminal's screen. A J-type thermocouple measures the temperature of the methanol steam reforming chamber. A temperature transmitter amplifies and converts the J-type thermocouple's tiny signal into a 0-5 V analog voltage. The temperature measured by the J-type thermocouple is displayed on the field control terminal. The heating rod's on / off state is program-controlled, specifically outputting a PWM pulse signal to control the solid-state relay's heating time, thereby achieving temperature control.

[0012] The network layer acts as a hub to connect the device layer and the application layer. The networking protocol adopts the IP protocol and uses networks such as 3G / 4G / 5G networks, IPv6, Wi-Fi and WiMAX, Bluetooth, ZigBee, etc. to complete information interaction, which can realize access and transmission functions and transmit the data collected by the device layer to the application layer quickly and reliably.

[0013] The application layer includes an always-on host computer that serves as a cloud server. This server receives numerous requests from other clients, analyzes them, performs necessary processing, and sends the results to the client or device. Requests include the operating status of each device, alarm signals, collected data, current output control parameters, modified control parameters, and control algorithm modifications. The entire online monitoring system and the neural network fuzzy PID self-heating control method were developed using LabVIEW.

[0014] The neural network fuzzy PID self-heating control method mainly includes the following steps:

[0015] Step 1. In the labview platform on the field control terminal (28), set the preset temperature T1 that the hydrogen production reactor in the hydrogen production reaction system needs to reach and set the safe working pressure P1 of the channel entering the catalytic combustion chamber, the safe working pressure P2 of the channel entering the methanol steam reforming chamber, the safe working pressure P3 of the channel flowing out of the catalytic combustion chamber, the safe working pressure P4 of the channel flowing out of the methanol steam reforming chamber, and the safe working temperature T2.

[0016] Step 2: The first pressure transmitter (5) measures the current pressure P1' in the front channel of the catalytic combustion chamber, the second pressure transmitter (6) measures the current pressure P2' in the front channel of the methanol steam reforming chamber, the fourth pressure transmitter (30) measures the current pressure P3' in the rear channel of the catalytic combustion chamber, the third pressure transmitter (16) measures the current pressure P4' in the rear channel of the methanol steam reforming chamber, and each thermocouple measures the temperature Ti.

[0017] Step 3. When the hydrogen production reaction program is started on LabVIEW or the hydrogen production reaction is in progress, when the pressures P1 ≥ P1', P2 ≥ P2', P3 ≥ P3', P4 ≥ P4', and T2 ≥ Ti, the program starts or continues the hydrogen production process; otherwise, the hydrogen production process stops, and LabVIEW outputs a warning message indicating which channel's pressure exceeds the set safety value.

[0018] Step 4: Initially, the data acquisition card uses a J-type thermocouple to measure the current temperature T3 of the reforming hydrogen production chamber. The field control terminal subtracts this temperature from the preset temperature T1 to determine the difference between the two temperatures. The corresponding control parameters are then calculated using a neural network fuzzy PID algorithm. When the measured temperature T3 is significantly lower than the preset temperature T1, the flow rate of methanol solution and the amount of air injected into the catalytic combustion chamber are increased, triggering an exothermic reaction. Simultaneously, the field control terminal sends a signal to the solid-state relay to increase the heating time of the J-type heater. The methanol solution flow rate, air volume, and heater heating time vary depending on the proximity between the measured temperature T3 and the preset temperature T1. Once temperatures T1 and T3 are close, the methanol solution flow rate, air volume, and heater heating time remain relatively stable. When the measured temperature T3 is greater than the preset temperature T1, the flow rate of methanol solution and the amount of air injected into the catalytic combustion chamber will decrease, shortening the heating time of the heater rod until the measured temperature T1 equals the preset temperature T2. Thereafter, the flow rate of methanol solution, the amount of air, and the heating time of the heater rod will remain at this state. The neural network fuzzy PID algorithm is as follows: A fuzzy neural network optimization algorithm module is added to the PID controller to obtain a neural network fuzzy PID module. This neural network fuzzy PID controller has two-dimensional inputs: the instantaneous error e(t) between the set temperature T1 and the current temperature T2, and the rate of change of the error ec(t); and three-dimensional outputs: Δkp, Δki, and Δkd, respectively. Δkp, Δki, and Δkd are the proportional, integral, and differential coefficients of the PID. Use e and ec as the input of the fuzzy neural network controller and perform fuzzification processing. After fuzzification, variables can be obtained. The variables act as fuzzy inference engines for fuzzy reasoning and memory fuzzy rules, and are combined with fuzzy rules to obtain the results of fuzzy reasoning. Finally, the defuzzification process is performed to obtain the control parameters Δkp, Δki, and Δkd of the output of the fuzzy neural network at the current moment. Then the control parameters are input into the relevant control of PID to obtain the final output parameter u (t) of the neural network fuzzy PID controller.

[0019] Step 5: The neural network is a BP model, consisting of five layers. The first layer is the input layer, with two neurons. The error e(t) and the error rate of change ec(t) are fed into the first layer of the neural network, which then directly feeds these variables into the second layer as outputs. The second layer is the membership function layer, with seven neurons representing seven fuzzy linguistic variables: "Negative Large" (NB), "Negative Medium" (NM), "Negative Small" (NS), "Zero" (ZO), "Positive Small" (PS), "Positive Medium" (PM), and "Positive Large" (PB). This layer primarily fuzzifies its input. The third layer is the fuzzy regularization layer, with 49 neurons. Each neuron in this layer represents a fuzzy rule between the input and output variables, and its function is to calculate the fitness of each rule. The fourth layer is the normalization layer, with a total of 49 neurons. Its function is to normalize the values of the 49 neuron nodes output by the third layer. The fifth layer is the output layer, which defuzzifies the fuzzified input after regularization and normalization. Fuzzy neural networks have the ability to self-learn and can perform online updates of neural network weight parameters through the gradient descent method. Through multiple learning cycles, the output value can be closer to the preset value.

[0020] In response to the shortcomings of the reforming hydrogen production reaction process being complex, the control system having low intelligence, low control precision, and slow response, the present invention has developed an online monitoring system for the hydrogen production reaction and a neural network fuzzy PID self-heating control method thereof. The system adopts a combination of data networking, on-site equipment control, and networked remote control to improve the intelligence of the hydrogen production reaction system. The neural network fuzzy PID optimization algorithm improves the control precision of the system, reduces the system delay time, and realizes self-heating control in the reforming hydrogen production reaction process. The specific beneficial effects are as follows:

[0021] 1. The present invention forms an online monitoring system with device layer, network layer and application layer, uses LabVIEW as the development language, and has a simple control program. The control algorithm can be flexibly modified according to actual needs to meet specific control requirements.

[0022] 2. The data acquisition card uploads the data measured by various sensors to the field control terminal for display, calculation and storage. The field control terminal transmits the data to the application layer through various communication methods. The field control terminal can also receive requests from the application layer and control the hydrogen production system equipment online.

[0023] 3. The present invention solves the heating control problem in the hydrogen reaction process, effectively reduces the delay time of the hydrogen production system through the neural network fuzzy PID algorithm, and improves the temperature control accuracy and system control sensitivity. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1Schematic diagram of the overall architecture of the full-process data acquisition and monitoring system of the present invention;

[0025] Figure 2 This is a connection diagram of the self-heating control system device of the present invention;

[0026] Figure 3 This is a flow chart of the self-heating control system of the present invention;

[0027] Figure 4 This is the principle diagram of the fuzzy PID control algorithm of the present invention;

[0028] Figure 5 This is a schematic diagram of the fuzzy controller composition block diagram of the present invention.

[0029] Figure 6 It is a structural diagram of the fuzzy neural network algorithm of the present invention;

[0030] Figure 7 is a membership function diagram of the present invention;

[0031] Figure 8 Schematic diagram of the language rules of the present invention;

[0032] Figure 9 This is the operating interface of the measurement and control system of the reforming hydrogen production reactor of the present invention.

[0033] In the figure: 1, air compressor, 2, first stop valve, 3, mass flow controller, 4, mixer, 5, first pressure transmitter, 6, second pressure transmitter, 7, first evaporator, 8, first peristaltic pump, 9, second stop valve, 10, methanol solution storage bottle, 11, methanol-water solution storage bottle, 12, third stop valve, 13, second peristaltic pump, 14, second evaporator, 15, reforming hydrogen production reactor, 16, third pressure transmitter, 17, first heat exchanger 18. First condenser 19. First gas-liquid separator 20. Drying tube 21. Vortex flowmeter 22. Gas cylinder 23. Sample test bag 24. Second heat exchanger 25. Second condenser 26. Second gas-liquid separator 27. Data acquisition card 28. Field control terminal 29. J-type thermocouple 30. Fourth pressure transmitter 31. Solid-state relay 32. Heating rod 33. Temperature transmitter 34. Fourth stop valve DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0035] See also Figures 1 to 9In an embodiment of the present invention, an online monitoring system for a hydrogen production reaction and a neural network fuzzy PID self-heating control method thereof include an online monitoring system characterized by: a device layer, a network layer, and an application layer. The start and stop status of all shutoff valves, peristaltic pumps, and evaporators are controlled by digital signals output by a field control terminal. All pressure transmitters can constantly collect the current pressure in the pipeline, output a 4-20mA analog current signal, and connect to the field control terminal via a data acquisition card. The collected pressure is displayed on LabVIEW. A mass flow controller can measure the flow rate in the pipeline and control the flow rate by controlling the valve opening. It is connected to the field control terminal via a data acquisition card and outputs a 0-5V analog voltage. The measured flow rate is displayed on LabVIEW. The valve opening control signal is controlled by a 0-5V analog signal output by a LabVIEW program. A vortex flowmeter is used to measure the flow rate of the gas generated in the methanol steam reforming chamber. It outputs a 4-20mA analog current and is connected to the field control terminal via a data acquisition card. The measured flow rate is displayed on LabVIEW. J-type thermocouples are used to measure the temperature of the reforming hydrogen production reaction chamber, condenser, heat exchanger, mixer, and evaporator. Temperature transmitters amplify and convert the J-type thermocouple signals into 0-5V analog voltages. The temperatures measured by the J-type thermocouples are displayed programmatically in LabVIEW. A LabVIEW program controls the on / off switching of the heater rods, specifically outputting PWM signals to control the heating time of the solid-state relays, thereby achieving temperature control. Data transmission between the client, cloud server, router, and on-site control terminal utilizes the IP protocol. Network interfaces can utilize 3G / 4G / 5G networks, IPv6, Wi-Fi, WiMAX, Bluetooth, ZigBee, and other networks. A neural network fuzzy PID algorithm model is deployed on the server. The server responds to client requests, analyzes them, and provides feedback. Requests include the operating status of each device, alarm signals, collected data, current output control parameters, modifications to control parameters, and changes to the control algorithm.The air compressor (1) is connected to the first stop valve (2), then to the mass flow controller (3), then to the mixer (4), and finally to the catalytic combustion chamber of the reforming hydrogen production reactor (15); the methanol solution storage bottle (10) is connected to the second stop valve (9), then to the first peristaltic pump (8), then to the first evaporator (7), and finally to the mixer (4); the methanol aqueous solution storage bottle (11) is connected to the third stop valve (12), then to the second peristaltic pump (13), then to the second evaporator (14), and finally to the reforming hydrogen production reactor (15). The first pressure transmitter (5) is inserted between the pipes of the mixer (4) and the reforming hydrogen production reactor (15) to measure the pressure between the pipes; the second pressure transmitter (6) is inserted between the pipes of the second evaporator (14) and the reforming hydrogen production reactor (15) to measure the pressure between the pipes; the other end of the methanol steam reforming chamber of the reforming hydrogen production reactor (15) is connected to the first heat exchanger (17), and then to the first condenser (18); then to the first gas-liquid separation device (19), and then to the drying pipe (20), and then to the vortex flowmeter ( 21), and finally two channels are separated by the vortex flowmeter (21), one of which is connected to the gas storage bottle (22), and the other is first connected to the fourth stop valve (34), and then connected to the sample detection bag (23); the other end of the catalytic combustion chamber of the reforming hydrogen production reactor (15) is first connected to the second heat exchanger (24), then connected to the second condenser (25), and finally connected to the second gas-liquid separation device (26); the third pressure transmitter (16) is inserted between the first heat exchanger (17) and the pipeline of the reforming hydrogen production reactor (15) to measure the pressure between the pipelines; the fourth pressure transmitter ( 30) is inserted between the pipelines of the second heat exchanger (24) and the reforming hydrogen production reactor (15) and is used to measure the pressure between the pipelines; a J-type thermocouple (29) is inserted in the reforming hydrogen production reactor (15), the J-type thermocouple (29) is connected to the temperature transmitter (33), the temperature transmitter (33) is connected to the data acquisition card (27), the data acquisition card (27) is connected to the field control terminal (28); a heating rod (32) is inserted in the reforming hydrogen production reactor (15), the heating rod (32) is connected to the solid-state relay (31), and the solid-state relay (31) is connected to the data acquisition card (27).

[0036] Figure 3 The present invention is based on Figure 2The flow chart of the self-heating control system. In the LabVIEW platform on the field control terminal (28), the preset temperature T1 and safe working temperature T2 required for the hydrogen production reactor in the hydrogen production reaction system are set, and the safe working pressure P1 of the channel entering the catalytic combustion chamber, the safe working pressure P2 of the channel entering the methanol water vapor reforming chamber, the safe working pressure P3 of the channel flowing out of the catalytic combustion chamber, and the safe working pressure P4 of the channel flowing out of the methanol water vapor reforming chamber are set. When the hydrogen production reaction program is started on LabVIEW or the hydrogen production reaction is in progress, the first pressure transmitter (5) measures the current pressure P1' in the channel before the catalytic combustion chamber, and the second pressure transmitter (6) measures the current pressure P1' in the channel before the methanol water vapor reforming chamber. P2', the pressure P3' of the current catalytic combustion chamber rear channel is measured by the fourth pressure transmitter (30), the pressure P4' of the current methanol steam reforming chamber rear channel is measured by the third pressure transmitter (16), and each thermocouple measures the temperature Ti; when the pressure P1 ≥ P1', and P2 ≥ P2', and P3 ≥ P3', and P4 ≥ P4' and T2 ≥ Ti, the program starts, and the hydrogen production process begins or continues; otherwise, the hydrogen production process stops, and a warning message is output on the LabVIEW to indicate which channel or device state quantity exceeds the set safety value. The data acquisition card collects the current reforming hydrogen production reaction chamber temperature T3 through the J-type thermocouple, and the field control terminal subtracts this temperature from the preset temperature value T1 to obtain the difference between the two temperatures, and uses the neural network fuzzy PID algorithm to calculate the corresponding control parameters. When the measured temperature T3 is significantly lower than the preset temperature T1, the flow rate of methanol solution and the amount of air injected into the catalytic combustion chamber are increased, triggering an exothermic reaction. Simultaneously, the field control terminal initiates a signal to the solid-state relay to increase the heating time of the J-shaped heater. The methanol solution flow rate, air volume, and heater heating time continuously change as the measured temperature T3 approaches the preset temperature T1. Until temperatures T1 and T3 are close, the methanol solution flow rate, air volume, and heater heating time remain relatively stable. When the measured temperature T3 exceeds the preset temperature T1, the methanol solution flow rate and air volume injected into the catalytic combustion chamber are reduced, shortening the heater heating time. This continues until the measured temperature T3 equals the preset temperature T1, at which point the methanol solution flow rate, air volume, and heater heating time remain constant. Since the temperature of the hydrogen production reaction system changes nonlinearly with time, the neural network fuzzy PID model is selected as a dual-input and three-output fuzzy neural controller module for control. The dual input parameters of the fuzzy control module are e(t) and ec(t), and the three output parameters of the fuzzy control module are the calculated adjustment quantities of the three independent proportional, integral and differential coefficients of the corresponding PID regulator: Δkp, Δki, and Δkd. The specific composition of this module is derived from the existing process control method.Fuzzification converts precise input quantities into fuzzy quantities. The database primarily includes the membership functions of various linguistic variables. The rule base includes a series of control rules expressed in fuzzy linguistic variables, reflecting the experience and knowledge acquired by control experts. Fuzzy reasoning is the core of fuzzy control, simulating the reasoning capabilities of basic human fuzzy concepts. This reasoning process is based on the implication relationships and inference rules of fuzzy logic. Clarification transforms the control quantity (fuzzy quantity) derived from fuzzy reasoning into a clear quantity actually used for control. The present invention incorporates a neural network optimization algorithm module that mimics the signal transmission process between neurons in natural organisms, including neuron activation and signal transmission through synapses. This gives the algorithm powerful learning capabilities and the ability to adapt to the dynamic characteristics of uncertain systems.

[0037] On the other hand, the self-heating control method of the hydrogen production reaction system includes the following steps:

[0038] The time domain function equation of the PID regulator is as follows:

[0039]

[0040] Discretize it:

[0041]

[0042] kp, Ti, and Td are the coefficients of the proportional term, integral term, and differential term, respectively. A fuzzy neural network optimization algorithm module is added to the PID controller to obtain a neural network fuzzy PID module. This neural network fuzzy PID controller has two-dimensional inputs, namely the instantaneous error e(t) between the set temperature T1 and the current temperature T3 and the rate of change of the error ec(t); and three-dimensional outputs, represented by kp, Ti, and Td, respectively. To reduce the time delay of the system, the kp, Ti, and Td of the controller are changed to:

[0043] kp=kp1+Δkp

[0044] ki=ki1+Δki

[0045] kd=kd1+Δkd

[0046] kp1, ki1, kd1 are the initial proportional term, integral term, and differential term parameters of PID.

[0047] Use e and ec as the input of the fuzzy neural network controller and perform fuzzification processing. After fuzzification, the variable xi can be obtained. The variable xi acts as a fuzzy inference engine for fuzzy reasoning and memory of fuzzy rules, and is combined with the fuzzy rule wij to obtain the fuzzy reasoning result Oj. Finally, the defuzzification process is performed to obtain the control parameters Δkp, Δki, and Δkd of the output of the fuzzy neural network at the current moment. Then the control parameters are input into the relevant control of PID to obtain the final output parameter u (t) of the neural network fuzzy PID controller.

[0048] The fuzzy neural network is a BP model with five layers. The first layer is the input layer, which has two neurons. The error e(t) and the rate of change of the error ec(t) are input into the first layer of the neural network. These variables are directly input as outputs into the second layer. The fuzzy domain for the error and the rate of change of the error is [-6, 6].

[0049] , ,

[0050] is the input of the cth neuron node in the first layer, is the output variable of the jth neuron in the first layer; the superscript indicates the number of network layers.

[0051] The second layer is the membership function layer, which has 7 neurons, representing seven fuzzy subsets: "Negative Large" (NB), "Negative Middle" (NM), "Negative Small" (NS), "Zero" (ZO), "Positive Small" (PS), "Positive Middle" (PM), and "Positive Large" (PB). It is mainly used to fuzzify its input. The membership function of the fuzzy subset is as follows: Figure 8 shown.

[0052] ,

[0053] m ij is the mean of the j-th membership function of the i-th input signal, σ ij is the standard deviation of the d-th membership function of the c-th input signal.

[0054] The third layer is the fuzzy regularization layer, which contains 49 neurons. Each neuron in this layer represents a fuzzy rule between the input and output variables. Its function is to calculate the fitness of each rule. Based on the experience of temperature control, fuzzy control rules are established, using the form "If e is 1 and ec is 2, then Δkp is 3 and Δki is 4 and Δkd is 5". Figure 9 shown.

[0055]

[0056] Ni is the i-th node in the current layer.

[0057] The fourth layer is the normalization layer, which has a total of 49 neurons. Its function is to normalize the values of the 49 neuron nodes output by the third layer. The calculation formula is:

[0058]

[0059] The fifth layer is the output layer, which is used to defuzzify the fuzzy input after regularization and normalization, and use the centroid method to defuzzify the output structure and accurately calculate Δkp, Δki, and Δkd. The calculation formula is:

[0060]

[0061] Fuzzy neural networks have the ability to self-learn and can perform online updates of neural network weight parameters through the gradient descent method. Through multiple learning cycles, the output value is closer to the preset value. According to the error back propagation algorithm, the evaluation function of the error signal is:

[0062]

[0063] R represents the desired output temperature, and r represents the current temperature.

[0064] BP fuzzy neural network uses gradient descent method to calculate m ij , σ ij and ω ij The optimization algorithm is:

[0065]

[0066] Where η m ,η σ ,η ω are parameters m ij , σ ij and ω ij The learning rate is α, and α is the momentum factor.

Claims

1. A neural network fuzzy PID self-heating control method, applied to an online monitoring system for hydrogen production reaction, characterized in that: The steps include: Step 1: In the LabVIEW platform, set the preset temperature T1 required for the hydrogen production reactor in the hydrogen production reaction system, and set the safe working pressure P1 of the channel entering the catalytic combustion chamber, the safe working pressure P2 of the channel entering the methanol steam reforming chamber, the safe working pressure P3 of the channel flowing out of the catalytic combustion chamber, the safe working pressure P4 of the channel flowing out of the methanol steam reforming chamber, and the safe working temperature T2; Step 2: The current reforming hydrogen production reactor ambient temperature T3 is detected by the J-type thermocouple (29), and the current pressure P1' in the front channel of the catalytic combustion chamber is measured by the first pressure transmitter (5), the current pressure P2' in the front channel of the methanol steam reforming chamber is measured by the second pressure transmitter (6), the current pressure P3' in the rear channel of the catalytic combustion chamber is measured by the fourth pressure transmitter (30), and the current pressure P4' in the rear channel of the methanol steam reforming chamber is measured by the third pressure transmitter (16) and the temperature Ti measured by each thermocouple; the data collected by the J-type thermocouple (29), the first pressure transmitter (5), the second pressure transmitter (6), the fourth pressure transmitter (30), and the third pressure transmitter (16) are transmitted to the field control terminal through the data acquisition card (27) and displayed on the labview; Step 3. When the hydrogen production reaction program is started on LabVIEW or the hydrogen production reaction is in progress, if the pressures P1 ≥ P1', P2 ≥ P2', P3 ≥ P3', P4 ≥ P4', and T2 ≥ Ti, the program starts or continues the hydrogen production process; otherwise, the hydrogen production process stops, and LabVIEW outputs a warning message indicating which channel's pressure exceeds the set safety value. Step 4: When the hydrogen production system is operating normally, the corresponding control parameters are calculated using a neural network fuzzy PID algorithm based on the difference between the current temperature T3 and the set temperature T1 on LabVIEW. The field control terminal (28) controls the heating rod (32) to heat the reforming hydrogen production reactor (15) through the data acquisition card (27) and the solid-state relay (31). The field control terminal (28) controls the valve opening size of the mass flow controller (3) through the data acquisition card (27) to control the flow rate of air. The characteristics of the neural network fuzzy PID algorithm in step 4 are as follows: a fuzzy neural network optimization algorithm module is added to the PID controller to obtain a neural network fuzzy PID module; the neural network fuzzy PID controller has two-dimensional inputs, namely the instantaneous error e(t) between the set temperature T1 and the current temperature T2 and the rate of change of the error ec(t); the three-dimensional outputs are expressed as Δkp, Δki, and Δkd respectively; e and ec are used as inputs of the fuzzy neural network controller for fuzzification processing, and the variable xi can be obtained after fuzzification. The variable xi acts as a fuzzy inference engine for fuzzy reasoning and memory fuzzy rules, and is combined with the fuzzy rules to obtain the result of fuzzy reasoning. Finally, a defuzzification process is performed to obtain the control parameters Δkp, Δki, and Δkd of the output of the fuzzy neural network at the current moment, and then the control parameters are added to the relevant control parameters of the PID to obtain the final output parameter u(t) of the neural network fuzzy PID controller; Step 5. The fuzzy neural network is a BP model, which includes five layers. The fuzzy neural network is a BP model, which includes five layers; the first layer is the input layer, which has two neurons, and the error e(t) and the error change rate ec(t) are imported into the first layer of the neural network, and the variables are directly input into the second layer as output; the second layer is the membership function layer, which has 7 neurons, representing seven fuzzy language variables, namely "negative large", "negative medium", "negative small", "zero", "positive small", "positive medium", and "positive large", which are mainly used to fuzzify their inputs; the third layer is the fuzzy regularization layer, which includes 49 neurons. Each neuron in this layer represents a fuzzy rule between the input variable and the output variable, and its function is to calculate the fitness of each rule; the fourth layer is the normalization layer, which has a total of 49 neurons, and its function is to normalize the values of the 49 neuron nodes output by the third layer; the fifth layer is the output layer, and its function is to defuzzify the fuzzy input after regularization and normalization; Fuzzy neural networks have self-learning capabilities and can perform online updates of neural network weight parameters through the gradient descent method. Through multiple learning cycles, the output value can be closer to the preset value. It includes an online monitoring system. The hydrogen production reaction online monitoring system consists of an equipment layer, a network layer and an application layer. The entire hydrogen production reaction online monitoring system is developed based on the LabVIEW language. The equipment layer consists of a hydrogen production reaction system, a central controller, a data acquisition module and an execution module; the central controller consists of a data acquisition card and a field control terminal.

2. The neural network fuzzy PID self-heating control method according to claim 1, characterized in that: The detailed connection of the equipment layer is as follows: the air compressor (1) is connected to the first stop valve (2), then connected to the mass flow controller (3), then connected to the mixer (4), and finally connected to the catalytic combustion chamber of the reforming hydrogen production reactor (15); The methanol solution storage bottle (10) is connected to the second stop valve (9), then connected to the first peristaltic pump (8), then connected to the first evaporator (7); and finally connected to the mixer (4); the methanol aqueous solution storage bottle (11) is connected to the third stop valve (12), then connected to the second peristaltic pump (13), then connected to the second evaporator (14), and finally connected to the methanol water vapor reforming chamber of the reforming hydrogen production reactor (15); the first pressure transmitter (5) is inserted between the pipes of the mixer (4) and the reforming hydrogen production reactor (15) to measure the pressure of the pipe. The second pressure transmitter (6) is inserted between the pipelines of the second evaporator (14) and the reforming hydrogen production reactor (15) to measure the pressure between the pipelines; the other end of the methanol steam reforming chamber of the reforming hydrogen production reactor (15) is connected to the first heat exchanger (17), and then to the first condenser (18); then to the first gas-liquid separation device (19), and then to the drying pipe (20), and then to the vortex flowmeter (21), and finally the vortex flowmeter (21) is divided into two channels, one of which is connected to the gas storage bottle (22) The other end of the catalytic combustion chamber of the reforming hydrogen production reactor (15) is first connected to the second heat exchanger (24), then connected to the second condenser (25), and finally connected to the second gas-liquid separation device (26); the third pressure transmitter (16) is inserted between the first heat exchanger (17) and the pipeline of the reforming hydrogen production reactor (15) to measure the pressure between the pipelines; the fourth pressure transmitter (30) is inserted between the second heat exchanger (24) and the reforming hydrogen production reactor (15) The pressure between the pipelines of the reforming hydrogen production reactor (15) is measured; a J-type thermocouple (29) is inserted in the reforming hydrogen production reactor (15), the J-type thermocouple (29) is connected to the temperature transmitter (33), the temperature transmitter (33) is connected to the data acquisition card (27), the data acquisition card (27) is connected to the field control terminal (28); a heating rod (32) is inserted in the reforming hydrogen production reactor (15), the heating rod (32) is connected to the solid-state relay (31), and the solid-state relay (31) is connected to the data acquisition card (27).

3. The neural network fuzzy PID self-heating control method according to claim 1, characterized in that: The parameters collected for the hydrogen production reaction system include the instantaneous flow rates of the air compressor (1), the methanol solution storage bottle (10), and the methanol-water solution storage bottle (11), the instantaneous flow rate through the vortex flowmeter (21), the instantaneous pressures of the four channels entering the reforming hydrogen production reactor (15) and flowing out of the reforming hydrogen production reactor (15), the instantaneous temperatures of the first evaporator (7), the second evaporator (14), the mixer (4), the reforming hydrogen production reactor (15), the first heat exchanger (17), the second heat exchanger (24), the first condenser (18), and the second condenser (25), the current gas storage capacity of the gas storage bottle (22), and the temperature measured by the J-type thermocouple (29).

4. The neural network fuzzy PID self-heating control method according to claim 1, characterized in that: The output control instructions include an electrical signal for controlling the switch of the solid-state relay (31), an electrical signal for controlling the opening size of the mass flow controller (3), an electrical signal for controlling the switch of all stop valves, and an alarm signal.

5. The neural network fuzzy PID self-heating control method according to claim 1, characterized in that: The state quantity displayed by the field control terminal (28) includes the various parameters measured in claim 3, the Ki, Ti, Td values output by the current neural network fuzzy PID algorithm and the overall output value of PID.

6. The neural network fuzzy PID self-heating control method according to claim 1, wherein: The development language of the hydrogen production reaction online monitoring system is LabVIEW. The mass flow controller (3) can output and receive analog signals and is controlled by the field control terminal (28). The first peristaltic pump (8) provides flow power for the methanol solution. The first evaporator (7) evaporates the methanol solution into a gaseous state and mixes it with air in the mixer (4). The second stop valve (9) controls the on-off of the methanol solution channel. The second stop valve (9), the first peristaltic pump (8) and the first evaporator (7) are all controlled by the field control terminal (28). The mixer (4) and the first evaporator The real-time temperature of the device (7) is collected and displayed on LabVIEW; the second peristaltic pump (13) provides the flow power for the methanol aqueous solution, the second evaporator (14) evaporates the methanol aqueous solution into gas, the third stop valve (12) controls the on-off of the methanol aqueous solution channel, the third stop valve (12), the second peristaltic pump (13) and the second evaporator (14) are all controlled by the field control terminal (28), the real-time temperature of the second evaporator (14) is collected and displayed on LabVIEW, the first heat exchanger (17) is a first-level cooling device, which can reduce the temperature after the reforming reaction. The temperature of the product, the first condenser (18) is a two-stage cooling device, which converts water vapor, methanol, etc. into liquid, and then separates the gas from the liquid through the first gas-liquid separation device (19) to obtain gas. The gas is dried through the drying tube (20) and then passes through the vortex flowmeter (21). The vortex flowmeter (21) can output an analog signal and display the flow rate on the field control terminal (28). Finally, it passes to the gas storage bottle (22) or the sample detection bag (23). The temperature of each level of cooling device is collected and displayed on the labview; the second heat exchanger (24) is The first-stage cooling device can reduce the temperature of the product after the catalytic combustion reaction. The second condenser (25) is a second-stage cooling device, which further reduces the temperature of the product after the catalytic combustion reaction, and then separates the gas and liquid through the second gas-liquid separation device (26); the temperatures of each stage of the cooling device are collected and displayed on LabVIEW; the solid-state relay (31) controls the on and off of the heating rod (32) according to the received electrical signal, and the mass flow controller (3) controls the valve opening size according to the received electrical signal to adjust the air flow at the inlet of the combustion channel of the reforming hydrogen production reactor.

7. The neural network fuzzy PID self-heating control method according to claim 1, characterized in that: The network layer acts as a hub connecting the device layer and the application layer, and can realize access and transmission functions. It uses networks such as 3G / 4G / 5G networks, IPv6, Wi-Fi and WiMAX, Bluetooth, ZigBee, etc. to complete information interaction and quickly and reliably transmit the data collected by the device layer to the application layer.

8. The neural network fuzzy PID self-heating control method according to claim 1, characterized in that: The application layer has an always-on host as a cloud server, which can receive many requests from other clients, analyze the requests, perform necessary processing, and send the results to the client; the requests include the operating status of each device, alarm signals, collected data, current output control parameters, modified control parameters and modified control algorithms.

Images