Carbon fiber production waste gas treatment device and treatment method

Abstract

This invention discloses a carbon fiber production waste gas treatment device and method. The device consists of a high-temperature incinerator, a waste gas collection device, a pretreatment unit, a core treatment unit, and a post-treatment unit connected in sequence. The waste gas collection device is a trumpet-shaped flange pipe; the pretreatment unit removes particulate matter and liquefies some pollutants through multi-layer filtration and condensation; the core treatment unit includes a wet electrostatic precipitator and a catalytic oxidation reaction tower; the post-treatment unit includes an activated carbon adsorption column and a molecular sieve drying tower. The main control module incorporates a parameter adjustment model based on machine learning, taking waste gas flow rate, key component concentration, and reaction temperature as inputs, and outputting the optimal temperature setpoint of the catalytic oxidation reaction tower and the centrifugal fan speed correction coefficient, achieving real-time nonlinear optimization of key parameters. This invention solves the problems of high-viscosity tar blockage, difficulty in removing highly toxic pollutants, and unstable treatment efficiency due to operating condition fluctuations through graded deep treatment and intelligent collaborative control, achieving waste gas purification.

Description

Technical Field

[0001] This invention relates to the field of industrial waste gas treatment technology, specifically to a waste gas treatment device and method for carbon fiber production. Background Technology

[0002] The carbon fiber production process mainly includes pre-oxidation, low-temperature carbonization, and high-temperature carbonization. During the high-temperature carbonization process, non-carbon elements in the fiber are released as pyrolysis gas, generating complex, high-temperature process waste gas containing large amounts of tar and highly toxic substances. The main characteristics of this waste gas include: high temperature (typically 400–900℃), large fluctuations in pollutant concentration, the presence of viscous tar that easily leads to pipeline blockage, and the presence of highly toxic gases such as HCN, making it extremely difficult to treat.

[0003] Existing technologies mostly employ single or simple combinations of processes such as "water washing and activated carbon adsorption" or "direct combustion". In the former, activated carbon quickly becomes saturated and ineffective when faced with high-concentration, high-temperature carbon fiber exhaust gas, and water washing easily generates cyanide-containing wastewater, causing secondary pollution; the latter consumes a lot of energy and may generate more harmful byproducts when incompletely combusted.

[0004] Although some waste gas treatment devices have been applied to carbon fiber production, traditional single waste gas treatment technologies are difficult to remove all pollutants efficiently at the same time, are prone to secondary pollution, have high operating costs, low automation, and cannot optimize operating parameters in real time according to waste gas conditions, resulting in unstable treatment efficiency, high energy consumption, and a large amount of maintenance work. Summary of the Invention

[0005] In view of this, the present invention provides a carbon fiber production waste gas treatment device and treatment method to solve the technical problems of low treatment efficiency, easy generation of secondary pollution, low degree of automation, and inability to optimize operating parameters in real time according to waste gas conditions in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a carbon fiber production waste gas treatment device, comprising: a waste gas collection device, including a collection pipe for connecting to a high-temperature waste gas source, a gas collection hood disposed at the beginning of the collection pipe, and a centrifugal fan for providing power for waste gas transportation. The pretreatment unit includes a filter and a condensation module, wherein the filter includes a multi-layer composite filter structure and the condensation module includes a shell-and-tube condenser. The core processing unit includes a wet electrostatic precipitator and a catalytic oxidation reaction tower. The catalytic oxidation reaction tower includes a tower body, a multi-layer catalytic reaction layer disposed inside the tower body, a temperature control device, and a third temperature sensor. The post-processing unit includes an activated carbon adsorption column and a molecular sieve drying tower; The main control module and online monitoring equipment, including the frequency converter, various sensors and temperature regulation device of the centrifugal fan, are all electrically connected to the main control module; The main control module has a built-in parameter adjustment model based on machine learning. The input of the parameter adjustment model includes at least the exhaust gas flow rate, the concentration of key components and the reaction temperature, and the output is the optimal temperature set point of the catalytic oxidation reaction tower and / or the speed correction coefficient of the centrifugal fan.

[0007] As a further embodiment of the present invention, the gas collection hood is a trumpet-shaped flange pipe, and the flange pipe is provided with guide vanes. The waste gas collection device also includes a first temperature sensor for detecting the temperature inside the collection pipe.

[0008] As a further embodiment of the present invention, the multi-layer composite filtration structure includes an outer coarse filter, a middle fine filter, and an inner fine filter arranged sequentially from the outside to the inside; the tubular condenser is equipped with a liquid level detector and a second temperature sensor.

[0009] As a further embodiment of the present invention, the multilayer catalytic reaction layer includes multiple honeycomb ceramic catalyst supports and noble metal catalysts supported on the catalyst supports; the temperature regulation device includes an electric heater and a coil-type cooling water heat exchanger.

[0010] As a further embodiment of the present invention, the post-processing unit further includes an online monitoring device for detecting emission gas parameters. The online monitoring device includes a gas concentration sensor group, a wireless transmission module, and a microcontroller. The microcontroller communicates with the main control module through the wireless transmission module.

[0011] As a further embodiment of the present invention, the centrifugal fan is electrically connected to the main control module via a frequency converter. The main control module adjusts the speed of the centrifugal fan via the frequency converter based on the estimated value of the exhaust gas emission or the pipeline pressure signal.

[0012] As a further embodiment of the present invention, the filter is equipped with a differential pressure sensor for detecting the pressure difference across it, and the drain port of the shell-and-tube condenser is connected to an automatic drain valve.

[0013] As a further embodiment of the present invention, the main control module includes a timing module, which is used to record the usage time of the catalyst, activated carbon and molecular sieve and to provide predictive maintenance reminders.

[0014] As a further embodiment of the present invention, a dehumidification section is provided between the wet electrostatic precipitator and the catalytic oxidation reaction tower. The main control module controls the start and stop of the dehumidification section according to the humidity signal, so as to control the relative humidity of the exhaust gas entering the catalytic oxidation reaction tower to be below 30%.

[0015] Specifically, the flange pipe is designed in a trumpet shape, and guide vanes are installed inside the flange pipe to ensure that the exhaust gas can enter the pipeline smoothly and evenly, reducing turbulence and local dust accumulation. The centrifugal fan is electrically connected to the main control module via a frequency converter to provide power for the exhaust gas conveying and can adjust the air volume according to the instructions of the main control module. The first temperature sensor is installed in the collection pipe and connected to the main control module to monitor the exhaust gas temperature in the pipeline in real time, serving as a basis for judging the system's insulation status and abnormal operation.

[0016] Specifically, this filter is used to filter large particulate impurities in exhaust gas. It features a multi-layer composite filtration structure, consisting of an outer coarse filter, a middle fine filter, and an inner fine filter arranged sequentially from the outside in, achieving step-by-step filtration. A differential pressure sensor is also installed on the filter to detect the pressure difference before and after filtration and sends the signal to the main control module to determine whether the filter element needs cleaning or replacement. The condensation module cools the exhaust gas. The tubular condenser achieves heat exchange cooling, causing high-boiling-point organic compounds in the exhaust gas, such as tar, to condense and liquefy. The drain outlet of the tubular condenser is connected to an automatic drain valve, which automatically controls drainage based on control signals. The tubular condenser is equipped with a level detector and a second temperature sensor. The level detector detects the liquid level, and the second temperature sensor detects the liquid temperature. The level detector, the second temperature sensor, and the refrigeration unit are all connected to the main control module. Based on control signals from the main control module, accumulated condensate can be automatically discharged at set times or levels.

[0017] Specifically, wet electrostatic precipitators are used to efficiently capture micron-sized droplets and fine particulate matter carried in condensed exhaust gases. Catalytic oxidation reaction towers are used to catalytically oxidize harmful gases (such as VOCs, HCN, and CO) in exhaust gases into harmless CO2 and H2O. Their specific structure includes a tower body, multiple catalytic reaction layers disposed inside the tower body, a gas distributor, a third temperature sensor, and a temperature control device.

[0018] Specifically, the waste gas treated by the catalytic oxidation reaction tower passes sequentially through an activated carbon adsorption column and a molecular sieve drying tower to further adsorb and remove any remaining trace organic matter and moisture, ensuring that the waste gas is deeply purified and dried. Finally, the treated waste gas that meets the standards is discharged through a chimney.

[0019] Specifically, an online monitoring device is installed at the chimney outlet. This device monitors the concentration, temperature, and humidity of pollutants in the emitted gas in real time and transmits the monitoring information wirelessly to the main control module. The main control module, acting as the brain of the system, receives and processes information such as pressure, temperature, flow rate, and liquid level throughout the entire waste gas treatment process, and executes process control and alarms for abnormal conditions.

[0020] Specifically, the main control module is an embedded control system based on an industrial PC, including a control host, an alarm module, a timing module, and a display module. The alarm module provides audible and visual alarms when operating parameters are abnormal or emission gas data exceeds standards; the timing module records the usage time of the catalyst, activated carbon, and molecular sieve and provides predictive maintenance reminders; the display module visualizes the status information and parameter data of the waste gas treatment process. Online monitoring equipment is installed at the chimney outlet and includes a gas concentration sensor group, a fourth temperature sensor, a humidity sensor, a wireless transmission module, and a microcontroller. It monitors the pollutant concentration, temperature, and humidity information in the emission gas in real time and sends the monitoring information to the main control module.

[0021] Preferably, the parameter adjustment model is a deep feedforward neural network, and the main control module is configured to execute model inference with an inference delay of no more than 50ms, and use the model output to adjust the temperature of the catalytic oxidation reaction tower and the speed of the centrifugal fan in real time.

[0022] Preferably, a gas-liquid separator or an electric heat tracing dehumidification section is provided between the wet electrostatic precipitator and the catalytic oxidation reaction tower, and the main control module controls the start and stop of the electric heat tracing dehumidification section according to the humidity signal at the inlet of the catalytic oxidation reaction tower.

[0023] Secondly, the present invention also provides a method for treating waste gas from carbon fiber production, which uses the aforementioned waste gas treatment device for carbon fiber production to treat the waste gas, and includes the following steps: S1. Waste gas collection and transportation: Start the centrifugal fan, and the high-temperature waste gas enters the collection pipeline through the gas collection hood; S2. Pretreatment: After being filtered by a filter, the exhaust gas enters the condensation module for cooling and oil removal; S3, Core Purification: The condensed exhaust gas passes through a wet electrostatic precipitator to capture fine particulate matter, and then enters the catalytic oxidation reaction tower; S4. Catalytic oxidation: During the operation of the catalytic oxidation reaction tower, the main control module uses a built-in machine learning-based parameter adjustment model to dynamically adjust the temperature and / or the speed of the centrifugal fan inside the catalytic oxidation reaction tower. S5. Post-treatment: The waste gas after catalytic oxidation is discharged after passing through an activated carbon adsorption column and a molecular sieve drying tower in sequence.

[0024] As can be seen from the above technical solution, the advantages of the present invention are: 1. By designing the trumpet-shaped flange pipe and internal guide vanes, the flow path of the exhaust gas is optimized, turbulence and resistance are reduced, and the exhaust gas collection efficiency is improved. 2. By combining a multi-layer composite filtration structure, a condensation module, a wet electrostatic precipitator, a catalytic oxidation reaction tower, an activated carbon adsorption column, and a molecular sieve drying tower, the system achieves graded and in-depth treatment of various harmful substances in the waste gas. This results in the concentrations of pollutants such as hydrogen cyanide, carbon monoxide, and volatile organic compounds in the treated waste gas being far lower than those treated by existing technologies, thus stably meeting stringent environmental emission standards. 3. The catalytic oxidation reaction tower uses a honeycomb ceramic catalyst carrier, which increases the contact area between the catalyst and the waste gas, improves the reaction efficiency, and greatly reduces energy consumption compared with the traditional direct combustion method; 4. By monitoring multiple sensors such as differential pressure, liquid level, and temperature, and combining them with a timing module, early warning and predictive maintenance for conditions such as filter blockage, catalyst deactivation, and consumable saturation are achieved, which greatly improves the system's operational reliability and maintenance convenience. 5. The main control module has a built-in parameter adjustment model based on machine learning, which can dynamically optimize the reaction temperature and fan speed according to real-time data such as exhaust gas flow rate and key component concentration. This enables nonlinear and multi-variable collaborative optimization of key parameters, allowing the system to adapt to fluctuations in exhaust gas conditions and reduce energy consumption while ensuring treatment efficiency. 6. The online monitoring equipment and the main control module form a closed-loop feedback control to ensure that the emitted gases always meet the standards. Attached Figure Description

[0025] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0026] Figure 1 This is a schematic diagram of the overall structure of the carbon fiber production waste gas treatment device of the present invention, showing the connection of the waste gas collection device, the pretreatment unit, the core treatment device, the post-treatment unit, and the high-temperature incinerator.

[0027] Figure 2 This is a schematic diagram of the carbon fiber production waste gas treatment device of the present invention, showing the connection relationship of the high-temperature incinerator, collecting air outlet, filter, tubular condenser, wet electrostatic precipitator, catalytic oxidation reaction tower, activated carbon adsorption column, molecular sieve drying tower, chimney and centrifugal fan.

[0028] Figure 3 This is a process flow diagram of the carbon fiber production waste gas treatment method of the present invention.

[0029] Figure 4 This is a three-dimensional structural diagram of the flange pipe of the carbon fiber production waste gas device of the present invention.

[0030] Figure 5 for Figure 4 A magnified view of a portion of point A.

[0031] Figure 6 This is another three-dimensional structural diagram of the flange pipe of the carbon fiber production waste gas device of the present invention.

[0032] Explanation of reference numerals in the attached figures: 10 - High-temperature incinerator; 11 - Flange pipe; 12 - Filter; 13 - Tubular condenser; 14 - Wet electrostatic precipitator; 15 - Catalytic reaction layer; 16 - Catalytic oxidation reaction tower; 17 - Activated carbon adsorption column; 18 - Molecular sieve drying tower; 19 - Chimney; 20 - Centrifugal fan. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0034] Example 1 refer to Figures 1 to 6 This embodiment provides a carbon fiber production waste gas treatment device, which includes a high-temperature incinerator 10, a waste gas collection device, a pretreatment unit, a core treatment device, and a post-treatment unit. It can effectively collect and efficiently purify high-temperature, high-concentration waste gas containing tar and highly toxic substances. Through multi-stage pretreatment and intelligent monitoring, it effectively prevents system blockage and equipment corrosion, improving operational stability. Furthermore, by introducing intelligent control and machine learning algorithms, it achieves automatic optimization of key process parameters, reducing energy and material consumption and improving overall treatment efficiency. Finally, through fault early warning and maintenance reminder mechanisms, it ensures long-term reliable operation and compliance with emission standards.

[0035] like Figure 2 and Figure 4As shown, the high-temperature incinerator 10 is a regenerative thermal oxidizer used for preliminary high-temperature oxidation of the waste gas generated during carbon fiber production, converting most of the combustible components into CO2 and H2O and stabilizing the waste gas temperature. Its heating chamber operating temperature is maintained within the range of 700–900℃. The main function of the high-temperature incinerator 10 is to perform preliminary high-temperature combustion of the pyrolysis gas generated during carbon fiber production, decomposing most of the organic matter and highly toxic components. A waste gas collection device is used for waste gas collection and transmission. One end of the waste gas collection device is connected to the heating chamber of the high-temperature incinerator 10 via a flange pipe 11, and the other end is connected to the pretreatment unit. The waste gas collection device includes a collection pipe, a flange pipe 11, and a centrifugal fan 20. In this embodiment, the collection pipe is made of 316L stainless steel and externally covered with insulation material. This material has good high-temperature resistance and corrosion resistance, enabling it to operate stably for a long time in the harsh environment of the high-temperature incinerator 10. Flange 11 connects to the beginning of the collection pipe. Flange 11 is designed in a trumpet shape, with its larger opening facing the exhaust port of the high-temperature incinerator 10 and its smaller opening connecting to the collection pipe. Flange 11 increases the collection area and, utilizing aerodynamic principles, allows exhaust gas to enter the collection pipe more smoothly, thereby improving collection efficiency. Inside flange 11, multiple guide vanes are also installed. These vanes are evenly distributed circumferentially on the inner wall of flange 11, guiding the exhaust gas to flow axially, reducing turbulence and resistance at flange 11, and further improving the collection effect.

[0036] The flange pipe 11 is designed in a trumpet shape, and multiple guide vanes are installed inside the small end of the flange pipe 11, which can reduce dust accumulation in the flange pipe 11 and reduce turbulence.

[0037] like Figure 4 , Figure 5 and Figure 6 As shown, multiple guide vanes are installed at the small end of flange pipe 11. The inner diameter of the small end is d, the width of the guide vane is W, the length of the guide vane is L, and the height of the guide vane is H. When d∶W∶L∶H=2000∶15∶244∶184, the resistance of the exhaust gas at flange pipe 11 is relatively small.

[0038] A first temperature sensor thermocouple (range 0~600℃) is installed inside the collection pipe. The end of the collection pipe is connected to the inlet of the centrifugal fan 20. The centrifugal fan 20 is electrically connected to the main control module through a frequency converter. The first temperature sensor is connected to the main control module. The centrifugal fan 20 is used to provide power for the transport of waste gas. The first temperature sensor is used to detect the temperature inside the collection pipe. The first temperature sensor is a high-temperature resistant thermocouple.

[0039] The pretreatment unit includes a filter 12, a condenser module, and an automatic drain valve. The filter 12 is connected to the outlet of the centrifugal fan 20. The filter 12 includes a multi-layer composite filtration structure and a differential pressure sensor for detecting the pressure difference before and after filtration. The filter 12 employs a multi-layer composite filtration structure: the outermost layer is a coarse filter screen (stainless steel wire mesh), which filters out larger particles of impurities in the exhaust gas, such as lint; the middle layer is a fine filter screen (glass fiber filter cotton), which filters out smaller dust particles; and the innermost layer is a high-efficiency fine filter screen (PTFE membrane filter cartridge), which traps tiny solid particles and some tar. A differential pressure sensor (range 0–5 kPa) is installed on the filter 12. The differential pressure sensor detects the pressure difference before and after gas filtration and sends the data to the main control module. The condenser module is used to cool the exhaust gas, and the automatic drain valve automatically controls drainage based on the control signal. Specifically, the condensation module includes a shell-and-tube condenser 13, a refrigeration unit that drives the shell-and-tube condenser 13, a level detector for detecting the liquid level of the condensate, and a second temperature sensor for detecting the temperature of the condensate. An automatic drain valve is connected to the drain port of the shell-and-tube condenser 13.

[0040] The shell-and-tube condenser 13 is a horizontal fixed tube sheet heat exchanger. The shell side carries the exhaust gas, and the tube side carries the cooling medium (ethylene glycol aqueous solution), which is used to cool the exhaust gas and condense and liquefy high-boiling-point organic matter.

[0041] The shell-and-tube condenser 13 has multiple heat exchange tubes inside, through which a low-temperature coolant flows, while exhaust gas flows outside the tubes. Through heat exchange, water vapor and some volatile organic compounds in the exhaust gas are cooled, thus achieving the purpose of cooling, removing moisture, and removing some easily condensable impurities. At the bottom of the shell-and-tube condenser 13, a drain outlet is provided for discharging the condensed liquid. The drain outlet is connected to an automatic drain valve, which automatically controls the drainage based on the liquid level to prevent liquid accumulation inside the shell-and-tube condenser 13. The shell-and-tube condenser 13 is equipped with a liquid level detector and a second temperature sensor. The liquid level detector detects the liquid level, and the second temperature sensor detects the liquid temperature. The second temperature sensor uses a high-temperature resistant thermocouple. The liquid level detector, the second temperature sensor, and the refrigeration unit are all connected to the main control module. During installation, the filter 12 and the shell-and-tube condenser 13 are connected sequentially downstream of the exhaust gas collection device, using stainless steel pipes connected via flanges.

[0042] The core treatment unit is used to capture micron-sized droplets and fine particulate matter carried in the condensed exhaust gas, and to convert harmful gases in the exhaust gas into harmless substances. The core treatment unit includes a wet electrostatic precipitator 14 and a catalytic oxidation reaction tower 16. The outlet of the tubular condenser 13 is connected to the inlet of the wet electrostatic precipitator 14, and the inlet of the catalytic oxidation reaction tower 16 is connected to the outlet of the wet electrostatic precipitator 14. The wet electrostatic precipitator 14 adopts a horizontal structure and handles an air volume of 6000 Nm³.3 / h, used to capture droplets and particulate matter with a particle size greater than 1μm.

[0043] The catalytic oxidation reaction tower 16 includes a tower body and a multi-layer catalytic reaction layer 15 disposed inside the tower body, a gas distributor for uniformly distributing exhaust gas, a third temperature sensor for detecting the temperature inside the tower body, and a temperature regulating device for regulating the temperature inside the tower body.

[0044] Specifically, the catalytic oxidation reaction tower 16 is installed downstream of the pretreatment unit, and the connecting pipeline is made of stainless steel and connected by flanges. The tower body is made of high-strength fiberglass, which has the advantages of being lightweight and corrosion-resistant.

[0045] The tower body is a cylindrical fiberglass tower with an inner diameter of Φ1200~1800mm and a height of 4~6m. Three to five multi-layer catalytic reaction layers 15 are arranged along the height of the tower. Each layer includes a honeycomb ceramic catalyst support (cordierite-based, pore density 400cpsi) and a noble metal catalyst (Pt, Pd, or Pt-Pd composite metal) supported on it. The gas distributor is a porous plate structure installed below the gas inlet to ensure uniform distribution of exhaust gas. A third temperature sensor is a K-type wear-resistant thermocouple, one above and one below each catalytic bed layer, used for real-time monitoring of bed temperature distribution. The temperature control device includes electric heating rods and a coil-type cooling water heat exchanger to maintain the optimal reaction temperature range within the tower at 280~330℃.

[0046] The multi-layer catalytic reaction layer 15 consists of multiple catalyst supports and noble metal catalysts (Pt, Pd, or Pt-Pd composite metals) supported on the catalyst supports. The catalyst supports are honeycomb ceramic catalyst supports (cordierite-based, pore density 400 cpsi), which have a large specific surface area and good gas permeability, increasing the contact area between the catalyst and the waste gas and improving the reaction efficiency. The catalysts are noble metals such as platinum and palladium, which can catalytically oxidize harmful substances in the waste gas, such as hydrogen cyanide, carbon monoxide, and volatile organic compounds, at relatively low temperatures, converting them into harmless substances such as carbon dioxide, water, and nitrogen. Inlet and outlet ports are respectively set at the bottom and top of the tower. The inlet port is connected to the outlet of the wet electrostatic precipitator 14, and the outlet port is connected to the inlet of the activated carbon adsorption column 17. A gas distributor is used to evenly distribute the waste gas within the catalytic oxidation reaction tower 16 to ensure the reaction proceeds fully. The gas distributor is made of stainless steel and is installed at the inlet of the catalytic oxidation reaction tower 16. The third temperature sensor is a high-temperature resistant thermocouple. The third temperature sensor can detect the temperature inside the tower in real time. The temperature regulation device includes an electric heater and a coil-type cooling water heat exchanger. The temperature regulation device is used to adjust the temperature inside the tower according to the temperature regulation signal of the main control module to maintain the reaction temperature within the optimal range. Both the electric heater and the coil-type cooling water heat exchanger are connected to the main control module.

[0047] The post-treatment unit includes an activated carbon adsorption column 17 and a molecular sieve drying tower 18. The waste gas treated by the catalytic oxidation reaction tower 16 passes sequentially through the activated carbon adsorption column 17 and the molecular sieve drying tower 18 before being discharged through a chimney 19. Activated carbon has a rich microporous structure and a large specific surface area, enabling it to adsorb residual harmful substances and odors in the waste gas. The molecular sieve selectively adsorbs moisture in the waste gas, ensuring the treated waste gas is dry. The post-treated waste gas is discharged through chimney 19 in compliance with emission standards. In this embodiment, the chimney 19 is made of corrosion-resistant carbon steel and lined with a ceramic anti-corrosion layer to prevent corrosion from the waste gas. The activated carbon adsorption column 17, the molecular sieve drying tower 18, and the chimney 19 are sequentially connected downstream of the catalytic oxidation reaction tower 16, with stainless steel pipes connected via flanges.

[0048] Activated carbon adsorption column 17 is filled with columnar coal-based activated carbon (iodine value ≥1000mg / g) to adsorb residual VOCs and odor components. Molecular sieve drying tower 18 is filled with type 4A spherical molecular sieves to remove moisture, ensuring that the dew point of the final emission gas is ≤-40℃.

[0049] An online monitoring system is installed at the outlet of chimney 19. This system monitors the concentration, temperature, and humidity of pollutants in the emitted gas in real time and sends the information to the main control module. This allows for timely adjustments to the operating parameters of the treatment unit, ensuring that exhaust emissions consistently meet environmental standards. Specifically, the online monitoring system includes a gas concentration sensor array (NOx). x The system includes a SO2 / CO / VOCs sensor, a fourth temperature sensor, a humidity sensor, a wireless transmission module, and a microcontroller, used to monitor emission gas parameters in real time and upload them to the main control module.

[0050] The gas concentration sensor group, the fourth temperature sensor, the humidity sensor, and the wireless transmission module are all connected to the microcontroller. The microcontroller communicates with the main control module via the wireless transmission module. The gas concentration sensor group is used to detect the concentration information of the emitted gas, the fourth temperature sensor is used to detect the temperature information of the emitted gas (the fourth temperature sensor uses a high-temperature resistant thermocouple), and the humidity sensor is used to detect the humidity information of the emitted gas.

[0051] The frequency converter, first temperature sensor, differential pressure sensor, liquid level detector, second temperature sensor, refrigeration unit, third temperature sensor, temperature regulation device, and online monitoring equipment are all electrically connected to the main control module. The main control module is used to receive and process data from each sensor and device, and to perform process control, parameter optimization, and abnormal alarms.

[0052] Specifically, the main control module is used to receive pressure, temperature and flow information during the waste gas treatment process, and to control the treatment process and alarm for abnormal conditions.

[0053] The main control module includes a control host, an alarm module, a timing module, and a display module. The alarm module, timing module, and display module are electrically connected to the control host. The alarm module is used to provide audible and visual alarms when abnormal operating parameters or excessive emission gas data are detected. The timing module is used to record the usage cycle of the catalyst and the after-treatment unit. The display module is used to visually display the status information and parameter data of the exhaust gas treatment process.

[0054] The main control module is an embedded control system based on an industrial PC, electrically connected to the inverter of the centrifugal fan 20, various sensors, refrigeration unit, temperature control device, automatic drain valve, and online monitoring equipment. The main control module includes a control host, an alarm module, a timing module, and a display module. The alarm module provides audible and visual alarms when operating parameters are abnormal or emission gas data exceeds standards; the timing module records the usage time of the catalyst, activated carbon, and molecular sieve and provides predictive maintenance reminders; the display module provides a visual display of the status information and parameter data of the waste gas treatment process.

[0055] The main control module incorporates a machine learning-based parameter tuning model. This model takes at least the waste gas flow rate, key component concentration, and reaction temperature as input, and outputs the optimal temperature setpoint of the catalytic oxidation reaction tower 16 and / or the speed correction coefficient of the centrifugal fan 20. This model is used to achieve real-time nonlinear optimization of key parameters. During the offline training phase, the model uses historical running datasets for supervised learning. After deployment, the inference latency is ≤50ms, meeting the requirements of industrial closed-loop control.

[0056] This machine learning model is trained based on historical operating data. It takes waste gas flow rate, component concentration, etc. as input and optimizes operating parameters as output.

[0057] In this embodiment, the waste gas treatment process includes: waste gas generated during carbon fiber production is fed into a high-temperature incinerator 10 for high-temperature incineration; after high-temperature incineration, the waste gas is discharged from the exhaust port of the high-temperature incinerator 10 and then drawn into a collection pipe through a flange pipe 11. Inside the collection pipe, the waste gas is transported by a centrifugal fan 20. To prevent the waste gas from cooling down during transport, causing some harmful substances to condense and clog the pipe, the collection pipe is wrapped with an insulation material, such as rock wool, to reduce heat loss. The main control module adjusts the speed of the centrifugal fan 20 via a frequency converter based on the estimated waste gas emission or the pipe pressure signal.

[0058] In the pretreatment stage, the exhaust gas first enters filter 12. In filter 12, larger particles are intercepted by the coarse filter, while smaller dust particles and some tar are filtered out by the fine and high-efficiency filters. The filtered exhaust gas then enters the tubular condenser 13, where it exchanges heat with a low-temperature coolant, lowering its temperature and cooling water vapor and some volatile organic compounds. The coolant temperature is controlled by the refrigeration unit to maintain it within a suitable range to ensure effective condensation. The condensed liquid is discharged through a drain, while the exhaust gas continues into the core treatment unit.

[0059] In the core treatment process, the pretreated waste gas enters the catalytic oxidation reaction tower 16. Within the multi-layered catalytic reaction layer 15, harmful substances in the waste gas undergo oxidation under the action of the catalyst. The reaction conditions are a temperature of 250–350℃ and a pressure of 0.1–0.3 MPa. The active components of the catalyst effectively lower the activation energy of the reaction, allowing it to proceed rapidly at a lower temperature. The reaction time is adjusted according to the flow rate and composition of the waste gas, generally 3–5 minutes. During the reaction, the temperature inside the catalytic oxidation reaction tower 16 is monitored in real time using a third temperature sensor. When the temperature is too high, the coil-type cooling water heat exchanger is activated to lower the temperature by introducing cold air into the catalytic oxidation reaction tower 16 or adjusting the coolant flow rate. When the temperature is too low, the electric heater is activated to raise the reaction temperature, ensuring the reaction always proceeds under optimal conditions.

[0060] To prevent residual moisture carried in the outlet gas of the wet electrostatic precipitator 14 from causing hydrothermal aging or activity inhibition of the precious metal catalyst in the catalytic oxidation reaction tower 16, a gas-liquid separator or heating and dehumidification section is added between the wet electrostatic precipitator 14 and the catalytic oxidation reaction tower 16. Specifically, when the fourth temperature sensor (located in the connecting pipe) detects that the relative humidity of the exhaust gas is higher than 60%, the main control module activates the electric heating device installed in the pipe to raise the temperature of the exhaust gas to 100-120°C, reducing the relative humidity to below 30% before it enters the catalytic oxidation reaction tower 16. At the same time, the main control module monitors the humidity sensor signal at the inlet of the catalytic oxidation reaction tower 16. If the humidity continues to exceed the standard, the alarm module is triggered to prompt maintenance personnel to check whether the demister of the wet electrostatic precipitator 14 is malfunctioning or whether the drainage system is blocked.

[0061] In the post-treatment and monitoring stage, the waste gas exiting the catalytic oxidation reaction tower 16 enters the activated carbon adsorption column 17, where residual harmful substances and odors are adsorbed by the activated carbon. Then, the waste gas enters the molecular sieve drying tower 18, where moisture is adsorbed by the molecular sieves, drying the waste gas. The post-treated waste gas is discharged through chimney 19. At the outlet of chimney 19, online monitoring equipment monitors parameters such as pollutant concentration, temperature, and humidity in the waste gas in real time. When the monitoring data exceeds the set emission standards, the main control module automatically adjusts the operating parameters of the treatment device, such as increasing the catalyst dosage, raising the reaction temperature, and extending the reaction time, to ensure that the waste gas meets emission standards. Simultaneously, the monitoring data is recorded and stored for subsequent query and analysis.

[0062] This invention, through the organic combination of the above-mentioned functional units, achieves several non-obvious synergistic technical effects, specifically manifested in the following aspects: (1) Synergy between the trumpet-shaped flange pipe and the condensation module: significantly reduces tar entry into the catalyst layer The waste gas from carbon fiber production contains a large amount of highly viscous tar. If it directly enters the subsequent treatment unit, it can easily cause problems such as filter clogging and catalyst pore covering and deactivation. This invention incorporates a trumpet-shaped flange pipe in the waste gas collection stage. The larger end faces the high-temperature incinerator grate outlet, while the smaller end connects to the collection pipe. Guide vanes are evenly arranged circumferentially inside the smaller end. This structure utilizes the tapering effect of the trumpet opening to accelerate the waste gas entry into the pipe, and the guide vanes force the airflow axially, effectively suppressing turbulence and eddies at the collection inlet and reducing early condensation and deposition of tar on the pipe wall.

[0063] More importantly, this structure forms a synergistic "pre-rectification-deep condensation" chain with the downstream tubular condenser: the flanged pipe ensures that the exhaust gas enters the collection pipe smoothly in a laminar flow state, avoiding uneven deposition of tar on the pipe wall due to turbulent impact during transportation; when the exhaust gas enters the tubular condenser, its uniform flow velocity distribution ensures maximum heat exchange efficiency between the cooling medium inside the heat exchange tubes and the exhaust gas outside the tubes, resulting in more thorough tar condensation. Tests have shown that using this combined structure reduces the residual tar content in the exhaust gas before entering the wet electrostatic precipitator by approximately 65%–80% compared to the traditional straight-pipe collection method, significantly mitigating the risk of the catalyst being covered by tar in the subsequent catalytic oxidation reaction tower and extending the catalyst's lifespan.

[0064] (2) Humidity control logic between wet electrostatic precipitator and catalytic oxidation: protecting catalyst lifespan In the process of capturing fine particulate matter and droplets, wet electrostatic precipitators inevitably cause the outlet exhaust gas to carry a certain amount of saturated water vapor. If this wet exhaust gas directly enters the catalytic oxidation reaction tower, the water vapor will compete for adsorption on the catalyst surface, inhibiting the occupation of active sites for pollutants such as VOCs and HCN by the precious metal catalyst. At the same time, long-term exposure to high-temperature water vapor may cause hydrothermal aging of the catalyst, leading to a decrease in activity.

[0065] To address the aforementioned issues, this invention adds a gas-liquid separator or an electric heating dehumidification section between the wet electrostatic precipitator and the catalytic oxidation reaction tower, with dynamic control implemented by the main control module based on humidity sensor signals. Specifically, the collaborative logic is as follows: when the relative humidity of the exhaust gas is detected to be higher than 60%, the main control module activates the electric heating device to raise the exhaust gas temperature to 100–120°C, reducing the relative humidity to below 30% before the gas enters the catalytic oxidation reaction tower. This humidity control and catalytic oxidation form a synergistic "humidity control and activity preservation" mechanism: on the one hand, the reduced water vapor partial pressure decreases the competitive adsorption of moisture and pollutants on the catalyst surface, improving catalytic oxidation efficiency; on the other hand, it avoids hydrothermal aging of the catalyst under long-term high-temperature and high-humidity conditions, extending the catalyst's activity decay period by approximately 30%–50%. Simultaneously, this control logic only activates when humidity exceeds the limit, avoiding energy waste caused by overheating and achieving a balance between protection and operational economy.

[0066] (3) Dynamic balance of machine learning models under different operating conditions: synergistic optimization of efficiency and energy consumption During carbon fiber production, the flow rate of exhaust gas and the concentration of pollutants fluctuate significantly with each process stage (pre-oxidation, low-temperature carbonization, high-temperature carbonization). Traditional PID control relies on fixed setpoints or manual adjustment based on experience, making it difficult to balance emission compliance and energy minimization when operating conditions change rapidly.

[0067] The parameter adjustment model built into this invention, based on machine learning, takes waste gas flow rate, key component concentration, and current reaction temperature as inputs, and outputs in real time the optimal temperature setpoint of the catalytic oxidation reaction tower and the centrifugal fan speed correction coefficient, forming a collaborative optimization mechanism of "feedforward prediction - feedback correction". Specifically, it manifests as follows: High-load conditions: When the exhaust gas flow rate or pollutant concentration increases sharply, the model predicts the emission risk and raises the temperature set point of the catalytic oxidation reaction tower to 320-350℃ (upper limit of the activity window) in advance. At the same time, it appropriately reduces the fan speed correction coefficient (0.90-0.95) to extend the residence time of exhaust gas in the catalytic bed, prioritizes the removal efficiency, and ensures that emissions meet the standards. Low load conditions: When the exhaust gas flow rate decreases or the pollutant concentration is low, the model lowers the temperature setpoint to 280-300℃ (lower limit of the active window), while increasing the fan speed correction coefficient (1.05-1.10) to increase the ventilation volume and reduce the energy consumption of the electric heater and the power consumption of the fan while ensuring that the emissions meet the requirements. Steady-state operating condition: The model continuously learns the optimal operating point during long-term operation, gradually converging to the balance region between "emission compliance margin" and "minimum energy consumption", thus achieving adaptive optimization.

[0068] Through the aforementioned dynamic adjustments, compared to traditional fixed-parameter control modes, this invention can reduce overall energy consumption by 15% to 25% under the same emission standards, while effectively avoiding the risk of exceeding emission limits due to fluctuations in operating conditions. This synergistic effect relies on the machine learning model's ability to fit multivariate nonlinear relationships, and the two-level architecture of "setpoint optimization + closed-loop tracking" formed by it and the underlying PID control. It is not a simple superposition of the functions of each unit, demonstrating the substantial innovation of this invention in intelligent control and process synergy.

[0069] Example 2 refer to Figures 1 to 6 This embodiment provides a method for treating waste gas from carbon fiber production, using the waste gas treatment device for carbon fiber production described in Embodiment 1, specifically including the following steps: S01. High-temperature incineration: Start the high-temperature incinerator 10 and pass the waste gas generated in the carbon fiber production into the high-temperature incinerator 10 for high-temperature incineration. The working temperature is maintained at 750-850℃ and the residence time is ≥1.2s, so that most of the organic matter undergoes pyrolysis and decomposition.

[0070] S02. Waste Gas Collection and Transportation: The centrifugal fan 20 is started. The main control module adjusts the speed of the centrifugal fan 20 through the frequency converter based on the estimated waste gas emission volume. The waste gas generated by the high-temperature incinerator 10 is discharged from the exhaust port and enters the collection pipeline through the trumpet-shaped flange pipe 11 with guide vanes. The main control module receives the pipeline temperature information detected by the first temperature sensor. If the pipeline temperature is lower than the set threshold (e.g., 450℃), the alarm module is activated to remind staff to check whether the insulation material is damaged or detached and to check for leaks in the pipeline. S03. Pretreatment and Monitoring: After the exhaust gas enters the filter 12 through the exhaust gas collection device, the main control module receives the differential pressure signal detected by the differential pressure sensor and determines whether the differential pressure value exceeds the preset threshold. If the differential pressure value exceeds the preset threshold (e.g., 3.2 kPa), an alarm is triggered to prompt cleaning or replacement of the filter element, reminding the staff that filter 12 needs cleaning or replacement. Subsequently, the exhaust gas enters the shell-and-tube condenser 13. The main control module receives the detection signals from the liquid level detector and the second temperature sensor, and compares the liquid level information detected by the liquid level detector and the temperature information detected by the second temperature sensor with the corresponding liquid level threshold and temperature threshold, respectively. If the liquid level information is lower than the liquid level threshold, the coolant level is too low, and the alarm module reminds the staff to replenish the coolant in time. If the temperature information is higher than the temperature threshold, the temperature is too high, and the alarm module reminds the staff to check the operating parameters of the refrigeration unit and make adjustments. The automatic drain valve is also periodically opened according to a preset time cycle (e.g., every 2 hours) to discharge the condensate at the bottom of the condenser and prevent condensate accumulation.

[0071] To prevent excessively high exhaust gas temperatures at the outlet of the high-temperature incinerator 10 from causing thermal damage to downstream equipment, the main control module incorporates high-temperature protection logic in the exhaust gas collection device. A first temperature sensor monitors the exhaust gas temperature in the collection pipeline in real time. When the detected temperature exceeds a preset safety threshold (e.g., 550℃), the main control module first reduces the speed of the centrifugal fan 20 via the frequency converter to decrease the amount of high-temperature exhaust gas entering. If the temperature continues to rise above 600℃, the main control module automatically opens the bypass cold air valve located between the flange pipe 11 and the filter 12, introducing ambient air for mixing and cooling. Simultaneously, it triggers the alarm module to issue a high-temperature warning, reminding operators to check the operating status of the high-temperature incinerator 10. Once the temperature returns to the normal range (≤450℃), the system automatically resumes normal operation.

[0072] S04, Core Purification: After filtration and condensation, micron-sized droplets and fine particulate matter carried in the exhaust gas are captured by a wet electrostatic precipitator 14, operating at a voltage of 40–60 kV, with the spray solution being alkaline circulating water with a pH of 8–9. The remaining exhaust gas enters a catalytic oxidation reaction tower 16 connected to the wet electrostatic precipitator 14; S05, Catalytic Oxidation: Before starting the catalytic oxidation reaction tower 16, the catalyst is activated. The main control module controls the electric heater to heat the tower to the catalyst activation temperature (e.g., 280-300℃), and oxygen-containing gas is introduced for catalyst activation. During operation, the main control module uses a parameter adjustment model based on machine learning. According to the real-time flow and composition data of the exhaust gas, it outputs adjustment signals to dynamically adjust the reaction temperature and pressure in the catalytic oxidation reaction tower 16. Temperature adjustment is achieved through the electric heater and the coil-type cooling water heat exchanger, and pressure adjustment is achieved by adjusting the opening of the inlet valve and the outlet valve, so that the temperature inside the tower is stabilized within the optimal reaction window of 280-330℃. Specifically, this model is a machine learning model. Its inputs include exhaust gas flow rate, temperature, and key component concentrations, while its outputs are the optimal reaction temperature setpoint and the fan speed correction coefficient. The model is trained using historical best operating data and can achieve real-time, nonlinear, and multivariate collaborative optimization of key parameters. Temperature is controlled by adjusting the electric heater and the coil-type cooling water heat exchanger, while pressure is adjusted by regulating the opening of the inlet and outlet valves. The parameter tuning model employs a deep feedforward neural network structure. Its offline training phase utilizes historical operating datasets derived from at least six months of actual operating records, covering typical operating conditions such as start-up, steady state, load surges, and shutdown. Each training sample contains a synchronously acquired input feature vector and a target output label validated by expert calibration or closed-loop optimization. The training objective is to minimize the mean squared error loss function.

[0073] The input data to the model is normalized before being fed into the neural network: the exhaust gas flow rate is normalized to the [0,1] interval, the concentration of key components is normalized using Z-score, and the reaction temperature is mapped to the [-1,1] interval. The output layer uses a linear activation function, and the output value is converted into actual control commands after inverse normalization.

[0074] The parameter tuning model based on machine learning uses two sources for its "key component concentration" input: First, a Fourier transform infrared spectrometer (FTIR) or gas chromatograph configured in the online monitoring equipment detects the concentration of characteristic pollutants such as HCN and NMHC in the exhaust gas in real time, with a sampling cycle of 30 seconds. This data is pre-processed by the microcontroller and then uploaded to the main control module. Second, a bypass sampling port is set at the inlet of the catalytic oxidation reaction tower (position 16) for periodic manual or automatic sampling and analysis to correct the online monitoring data and ensure the accuracy of the model input. Model inference calculations are performed locally on the industrial control computer, with a single inference delay ≤50ms, meeting the requirements of industrial closed-loop control. When abnormal fluctuations occur in the online monitoring data, the main control module prioritizes error compensation using the deviation between the forward predicted value and the actual value to prevent drastic fluctuations in control commands.

[0075] In this embodiment, the parameter tuning model based on machine learning adopts a deep feedforward neural network (DFNN) structure, and its specific design is as follows: (1) Model structure Input layer: Contains 6 input nodes, each corresponding to the exhaust gas flow rate (Nm³). 3 / h), catalytic oxidation reaction tower inlet temperature (°C), HCN concentration in exhaust gas (mg / m³) 3 NMHC concentration in exhaust gas (mg / m³) 3 The input variables are: the average reaction temperature (°C) of each layer in the current catalytic oxidation reaction tower, and the current operating frequency (Hz) of the centrifugal fan. All input variables are normalized before being fed into the network. The exhaust gas flow rate is normalized to the [0,1] interval using MinMax, concentration variables are normalized using Zscore, and temperature variables are mapped to the [1,1] interval.

[0076] Hidden layers: A total of 3 hidden layers are set, each containing 64 neurons. The ReLU (Rectified LinearUnit) activation function is used to introduce non-linear fitting ability and alleviate the gradient vanishing problem. To prevent overfitting, a Dropout layer is added after each layer, with a dropout rate of 0.2.

[0077] Output layer: Contains two output nodes, corresponding to the optimal temperature setpoint (°C) of the catalytic oxidation reaction tower and the speed correction coefficient of the centrifugal fan (dimensionless, ranging from 0.85 to 1.15). The output layer uses a linear activation function, and the output values ​​are converted into actual control commands after inverse normalization.

[0078] (2) Training methods Offline Training: The offline training phase of the model utilizes at least 12 months of historical operating data. The dataset covers various typical operating conditions, including start-up, steady-state operation, load surges, and shutdown, collecting approximately 500,000 valid samples. Each sample contains a synchronously recorded input feature vector and an optimal control label calibrated by experts (the label is determined by combining emission compliance and energy efficiency during operation). The training uses Mean Squared Error (MSE) as the loss function, with Adam as the optimizer. The initial learning rate is set to 0.001, the batch size is 256, and the training epochs are 200. An early stopping mechanism monitors the validation set loss; training stops if the loss does not decrease for 10 consecutive epochs.

[0079] Online Updates: After model deployment, the main control module automatically triggers incremental learning every 30 days. The system extracts high-quality operational data from the past 30 days that meets the criteria of "stable emissions compliance and unit energy consumption lower than the historical median," and performs small-batch fine-tuning in conjunction with the original training set. During incremental learning, the model parameters are frozen in the first three hidden layers, and only the weights of the last hidden layer and the output layer are updated to maintain model stability while adapting to changes in operating conditions. Incremental learning is executed in the background and does not affect the main control logic. Before replacing the old model with the new model after the update, it must pass the validation set error verification.

[0080] (3) Control Logic The model output and the underlying PID control form a two-level control architecture of "setpoint optimization + closed-loop tracking". The main control module reads data from each sensor with a sampling period of 1 second, and after filtering and normalization, it is input into the model for inference, with a single inference delay of ≤50ms.

[0081] The "optimal temperature setpoint" output by the model serves as the target value for the PID control loop of the catalytic oxidation reaction tower temperature. The PID controller outputs adjustment signals for the electric heater or coil-type cooling water heat exchanger based on the deviation between the current temperature and the target value. The "fan speed correction coefficient" output by the model is multiplied by the base speed setpoint estimated based on the exhaust gas flow rate, serving as the final frequency command for the centrifugal fan inverter.

[0082] To avoid frequent actions of the actuators due to sudden changes in model output, the main control module introduces a first-order low-pass filter (filter coefficient 0.3) into the model output results and sends them to each execution unit at a period of 500ms, forming a closed-loop control link of "sampling, inference, filtering, execution, and resampling".

[0083] (4) Robust design Sensor Failure Response: When any critical input sensor (such as an exhaust gas flow meter or gas concentration analyzer) experiences a communication interruption or exceeds data limits, the main control module automatically switches to standby control mode. In standby mode, model inference is paused, and the system uses preset values ​​based on operating conditions for control, while simultaneously triggering an alarm module to prompt maintenance personnel for repair. If a temperature sensor fails, the weighted average of temperature sensors on adjacent layers is used as a substitute input; if a concentration sensor fails, the 95th percentile is used as a conservative estimate based on historical concentration distribution data under the same operating conditions, ensuring the system can still operate safely in degraded conditions.

[0084] Model output anomaly handling: The main control module performs real-time verification of the model output's rationality. If the output temperature setpoint exceeds the catalyst's safe temperature window (250℃~380℃), or the fan speed correction coefficient exceeds the range of 0.8~1.2, the system determines that the model output is abnormal, automatically blocks the current model inference result, reverts to the last valid output value, and uses a PID controller to maintain operation at the current setpoint. At the same time, it triggers a "model output anomaly" alarm and records the input data at the moment of the anomaly for subsequent analysis.

[0085] Control Timeout and Communication Failure: If the model inference time exceeds 100ms (normally ≤50ms), the system will automatically skip the current cycle inference, use the valid output value of the previous cycle, and accumulate the timeout count. After 5 consecutive timeouts, the system will switch to standby control mode and notify maintenance personnel to check the industrial control computer resource usage or model service status.

[0086] Through the above-mentioned model structure design, training strategy, two-level control architecture and multiple robustness protection measures, the machine learning model of this invention can achieve stable, reliable and low-latency intelligent control in complex industrial environments, effectively cope with the fluctuations in the operating conditions of carbon fiber production exhaust gas, and achieve synergistic optimization of treatment efficiency and operating energy consumption.

[0087] Every 1000 hours of operation, the main control module automatically triggers a model performance review: it compares the operating data of the most recent 30 days with the model's predicted values. If the mean square error exceeds a preset threshold, it prompts the maintenance personnel to retrain the model or adjust the parameters.

[0088] During the catalyst activation stage before the start-up of the catalytic oxidation reaction tower 16, the main control module controls the electric heater to raise the tower temperature to the activation temperature (280-300℃) and opens the activation gas source valve. The activation gas source uses clean air filtered and condensed by the pretreatment unit or a mixture of nitrogen and air, ensuring an oxygen content of no less than 10% (volume fraction). Untreated process waste gas is avoided to prevent impurities such as tar and cyanide in the waste gas from adsorbing onto the catalyst active sites during activation, thus affecting the activation effect. The activation duration is 2-4 hours, until the third temperature sensor shows that the temperature distribution of each catalytic reaction layer 15 is uniform and stable, indicating that the active sites on the catalyst surface have been fully activated.

[0089] S06. Catalyst Maintenance Reminder and Regeneration: Based on the records from the timing module, the main control module issues a replacement or regeneration reminder when the catalyst approaches its theoretical service life (e.g., 12,000 hours). During catalyst regeneration, the exhaust gas in the catalytic oxidation reaction tower 16 is vented, and oxygen-containing high-temperature gas is introduced to oxidize the catalyst and restore its activity. The timing module records the cumulative operating time of the catalyst, activated carbon, and molecular sieve respectively. The main control module has built-in theoretical lifespan thresholds for each consumable (catalyst: 12,000 hours; activated carbon: 8,000 hours; molecular sieve: 10,000 hours). When the remaining lifespan drops to 10%, the display module pops up a replacement reminder; when the remaining lifespan is 0, the alarm module upgrades to a forced replacement warning.

[0090] S07. Post-treatment and Consumables Management: The waste gas treated by the catalytic oxidation reaction tower 16 is sequentially adsorbed and dried through the activated carbon adsorption column 17 and the molecular sieve drying tower 18, and finally discharged through the chimney 19. The main control module controls the alarm module to detect and remind the post-treatment unit. The staff regularly checks the adsorption effect of the activated carbon and the molecular sieve drying tower 18. When the activated carbon or molecular sieve is saturated, it needs to be replaced. S08. Emission Monitoring and Feedback Control: The emission parameter NO of the exhaust gas is monitored in real time by online monitoring equipment installed at the outlet of chimney 19. x (Concentrations of SO2, CO, VOCs, temperature, and humidity). When the monitored data exceeds the standard, the main control module triggers an alarm and adjusts the operating parameters of each unit in conjunction with the control (such as increasing the temperature of the catalytic oxidation reaction tower and adjusting the fan speed), forming a closed-loop control to ensure that the emitted gas consistently meets the standards.

[0091] When the online monitoring equipment detects that the concentration of VOCs in the exhaust gas exceeds the set limit for 3 consecutive minutes, the main control module will perform the following linkage operations in sequence: (1) raise the target temperature set point of the catalytic oxidation reaction tower 16 by 5 to 10°C; (2) reduce the speed of the centrifugal fan 20 by 5 to 8% through the frequency converter to prolong the residence time of the exhaust gas in the catalytic bed; (3) trigger the alarm module to prompt the operation and maintenance personnel to check the catalyst activity status.

[0092] The main control module synchronously reads data from each sensor with a sampling period of 1 second. After filtering, calibration, and unit normalization, it constructs a process variable vector containing 12 dimensions, including exhaust gas flow rate, key component concentration, reaction temperature at each stage, and emission gas concentration. This vector serves as the input to the PID controller or machine learning model. Control commands are issued to each execution unit with a period of 500ms, forming a closed-loop control chain of sampling-calculation-execution-re-sampling.

[0093] Application Implementation Taking the waste gas treatment system of a carbon fiber production line with an annual output of 1000 tons as an example, the specific implementation of the present invention will be described in detail. The waste gas flow rate emitted by the high-temperature incinerator 10 is approximately 5000 m³ / s. 3 The temperature fluctuates between 380 and 400°C per hour. The main pollutants include tar and HCN (hydrogen cyanide), with a concentration of approximately 150–300 mg / m³. 3NMHC (non-methane total hydrocarbons, concentration approximately 200–500 mg / m³) 3 ) and charcoal dust.

[0094] The specific configuration of the carbon fiber production waste gas treatment device is as follows: High-temperature incinerator 10: It adopts a regenerative thermal oxidizer (RTO) to perform preliminary high-temperature oxidation on the waste gas generated by the original production line, converting most of the combustible components into CO2 and H2O, and stabilizing the waste gas temperature.

[0095] The exhaust gas collection device is configured with a collection pipe made of 310S stainless steel, covered with 150mm thick aluminum silicate insulation cotton. The flange pipe 11 is trumpet-shaped with 4 guide vanes. The inner diameter of the small end is d=500mm. It is equipped with 4 evenly distributed guide vanes with a width W=37.5mm, a length L=61mm, and a height H=46mm (d:W:L:H≈2000:15:244:184).

[0096] The centrifugal fan 20 is a high-temperature resistant type with a power of 55kW and frequency conversion control. A first temperature sensor (thermocouple, range 0~600℃) is installed on the collection pipeline.

[0097] Filter 12: Employs a three-stage composite filter element, with an outer layer of stainless steel wire mesh (100 mesh), a middle layer of glass fiber filter cotton, and an inner layer of PTFE membrane fine filter screen. Differential pressure sensors (range 0-5 kPa) are installed at the filter inlet and outlet.

[0098] The heat exchange area of ​​the shell-and-tube condenser 13 is 50m². 2 The cooling medium is an aqueous solution of ethylene glycol. The matching refrigeration unit has a power of 30kW. A liquid level sensor and a second temperature sensor are installed at the bottom of the condenser, and are connected to an automatic drain valve.

[0099] The wet electrostatic precipitator 14 adopts a horizontal structure. It handles an air volume of 6000 Nm³. 3 / h, used to capture droplets and particulate matter with a particle size greater than 1μm.

[0100] Catalytic oxidation reaction tower 16: The tower body is made of 316L stainless steel. Catalytic oxidation reaction tower 16 is filled with Pt-Pd / TiO2-Al2O3 honeycomb ceramic catalyst. A third temperature sensor array with three temperature measurement points is installed inside the tower. The temperature control device includes a 60kW electric heater and a coil-type cooling system. The control accuracy of the temperature control device is ±5℃. The main control module combines an industrial PLC with a host computer, and incorporates a machine learning model trained based on historical data for parameter prediction.

[0101] Post-treatment unit: Activated carbon adsorption column 17 is filled with 1.5m of coal-based activated carbon with an iodine value ≥1000. 3Molecular sieve drying tower 18 is filled with 1.0m³ of 13X type molecular sieve. 3 .

[0102] Main Control Module and Online Monitoring: The main control module combines an industrial PLC (Programmable Logic Controller) with a host industrial computer. The online monitoring equipment is installed at chimney outlet 19 and integrates a non-dispersive infrared (NDIR) analyzer (measuring CO and CO2), a chemiluminescence analyzer (measuring NOx, if present), a Fourier transform infrared (FTIR) spectrometer or a gas chromatography-mass spectrometry (GC-MS) bypass system (for periodic monitoring of NMHC and benzene compounds, etc.), as well as temperature and humidity sensors. All data is transmitted to the main control module via an industrial Ethernet network.

[0103] The process is as follows: System initialization: The main control module is started, and all sensors perform online self-tests. Operators set target emission standards (e.g., NMHC < 50 mg / m³) via the display module. 3 HCN < 1.0 mg / m² 3 ).

[0104] The high-temperature incinerator 10 is started, and the carbonization process waste gas is introduced into the high-temperature incinerator 10 for combustion at a temperature above 850°C. After combustion, the waste gas (temperature approximately 400°C) is smoothly introduced into the collection pipeline through the flared flange pipe 11. The main control module sets the speed of the centrifugal fan 20 to 45Hz according to the preset value. The first temperature sensor displays a pipeline temperature of 380°C, which is within the normal range. If the temperature is lower than 300°C (the set threshold), the alarm module issues an audible and visual alarm stating "Low pipeline temperature, please check insulation."

[0105] The exhaust gas passes through filter 12 with an initial differential pressure of 200 Pa. The main control module reads the differential pressure sensor data in real time. After operation, the differential pressure sensor reading rises to 800 Pa, with a preset threshold of 1000 Pa. If the differential pressure remains above 1000 Pa for 2 consecutive hours, the main control module displays a message on the screen: "Filter element differential pressure is too high; inspection recommended soon." In the condensation module, the second temperature sensor shows the coolant temperature at 5°C, indicating a normal level. The automatic drain valve opens for 30 seconds every 4 hours.

[0106] After being cooled, the exhaust gas passes through a wet electrostatic precipitator 14 to remove fine droplets, and then enters a catalytic oxidation reaction tower 16.

[0107] Based on the concentration and flow rate of the emitted gas at the chimney outlet fed back by the online monitoring equipment, the main control module calculates the current optimal reaction temperature as 320℃ using a machine learning model, and then sends a command to the temperature regulation device to control the electric heater to maintain this temperature.

[0108] The timing module showed that the catalyst had been running continuously for 4,500 hours, and the main control module issued a reminder message that "the catalyst is about to reach its maintenance cycle".

[0109] The NMHC concentration of the gas after post-processing was measured to be 12 mg / m³ by online monitoring equipment. 3 well below 50 mg / m 3 The emission standards are met. All data is displayed in real time on the display module.

[0110] This machine learning model is trained using historical best operating data. Its input variables include real-time exhaust gas flow rate, concentrations of key pollutants (such as HCN and NMHC), and current reaction temperature. Output variables are the optimal temperature setpoint of the catalytic oxidation reaction tower 16 and the speed correction coefficient (0.9–1.1) for the centrifugal fan 20. The model fine-tunes the fan's base speed setpoint to achieve optimal overall system pressure drop. This model enables the system to cope with fluctuations in exhaust gas operating conditions, achieving nonlinear, multivariate real-time collaborative optimization, reducing energy consumption while ensuring treatment efficiency.

[0111] The main control module sends the temperature setpoint output from the model to the temperature regulation device of the catalytic oxidation reaction tower. Through PID control combined with feedforward compensation, the temperature is stabilized at the setpoint (e.g., dynamically adjusted within the range of 300–350℃). Simultaneously, a speed correction coefficient is applied to the fan frequency converter to achieve precise airflow control.

[0112] The exhaust gas after catalytic oxidation passes sequentially through activated carbon adsorption column 17 and molecular sieve drying tower 18 for further adsorption of residual trace organic matter and moisture. The timing module of the main control module records the cumulative operating time of the catalyst, activated carbon, and molecular sieve. When the catalyst's operating time approaches its design life (e.g., 4800 hours), the display module will show "Catalyst life is about to expire; regeneration or replacement is recommended" 240 hours in advance. Similarly, replacement reminders are set for activated carbon and molecular sieve.

[0113] Comparative experiments and effect verification To verify the actual effect of the technical solution of the present invention, the following comparative experiments were carried out in the application scenario of a carbon fiber production line with an annual output of 1,000 tons. All experiments were carried out under the same working conditions, and the cumulative operating period was no less than 6 months.

[0114] (1) Comparison of emission concentration fluctuations when the machine learning model is on and off Under the same exhaust gas treatment device and operating conditions, the continuous changes in VOCs (as NMHC) concentration at the chimney outlet were recorded in two modes: "machine learning model enabled" and "machine learning model disabled (using fixed setpoint PID control)". The sampling period was 1 minute, and the monitoring was carried out continuously for 72 hours. The results are shown in Table 1: Table 1

[0115] Experimental results show that: After adopting a machine learning model, the average NMHC emission concentration decreased from 18.7 mg / m³. 3 Decreased to 12.3 mg / m³ 3 The decrease was approximately 34.2%; The standard deviation of emission concentration was 12.4 mg / m³. 3 Decreased to 5.6 mg / m³ 3 The fluctuation amplitude has been significantly reduced, indicating that the model can effectively cope with operating condition fluctuations and achieve stable control; Under fixed PID control mode, due to the failure to respond promptly to sudden changes in exhaust gas flow and concentration, there were a total of 8 instances of instantaneous emission exceeding the standard (≥50 mg / m³). 3 Under the machine learning model control mode, the emission concentration remained stable at 40 mg / m³. 3 No instances of exceeding the standard were found below.

[0116] This comparison verifies the dynamic optimization capability of the machine learning model under the coordinated control of "feedforward prediction-feedback correction", which significantly improves the system's adaptability to operating condition fluctuations and the reliability of emission compliance.

[0117] (2) Comparative experiment between the device of the present invention and the traditional "water washing + activated carbon" process At different operating stages of the same production line, the device of this invention and the traditional "water washing tower + activated carbon adsorption" process were used for waste gas treatment, and their performance in terms of tar blockage frequency, system energy consumption, and consumable replacement cycle was compared. Data was recorded over a total of 6 months of operation, as shown in Table 2. Table 2

[0118] Note: In the traditional "water washing + activated carbon" process, the cyanide-containing wastewater generated in the water washing section needs to be treated separately. Therefore, the energy consumption and cost of wastewater treatment are not included in this comparison. In the device of this invention, the activated carbon adsorption column is only used as a fine treatment unit, and its load is much lower than that of the traditional process, so the replacement cycle is significantly extended.

[0119] Comparative analysis of results: 1. Tar Clogging Frequency: In traditional processes, the nozzles and pipes of the water washing tower are frequently clogged due to tar adhesion, resulting in an average of 3.2 malfunctions per month, requiring shutdown for cleaning. This invention employs a step-by-step pretreatment process involving a trumpet-shaped flange pipe for flow guidance, multi-layer composite filtration, condensation and coking removal, and wet electrostatic dust removal. Tar is effectively removed before entering the activated carbon adsorption column, reducing tar clogging malfunctions to 0.3 times per month, a reduction of 90.6%, significantly improving the reliability of continuous system operation.

[0120] 2. System Energy Consumption: In traditional processes, the water washing tower requires a high-power circulating water pump and fan, and the activated carbon adsorption column experiences a significant pressure drop, resulting in an average system operating energy consumption of 385 kW·h / 10,000 m³. 3 Exhaust gas. This invention employs a combination of a wet electrostatic precipitator and a catalytic oxidation reaction tower. The catalytic oxidation reaction is carried out at 280–330°C, significantly reducing energy consumption compared to traditional direct combustion methods (typically >750°C). Simultaneously, a machine learning model dynamically optimizes fan speed and reaction temperature, avoiding excessive air supply and overheating, reducing the system's average operating energy consumption to 278 kW·h / 10,000 m³. 3 Exhaust emissions decreased by 27.8%.

[0121] 3. Consumable Replacement Cycle: In traditional processes, activated carbon directly withstands the impact of high-concentration pollutants, requiring replacement every 3200 hours on average. In this invention, the activated carbon adsorption column is located after the catalytic oxidation reaction tower, serving only as a fine treatment unit to handle low-concentration residual pollutants. This extends the activated carbon replacement cycle to 8500 hours, an extension of approximately 165.6%. Under the dual protection of humidity control and tar pre-removal, the precious metal catalyst in the catalytic oxidation reaction tower achieves an actual operating life of 12500 hours, exceeding the theoretical design value (12000 hours), significantly reducing consumable replacement costs and maintenance workload.

[0122] 4. Annual maintenance downtime: Due to the reduced frequency of tar blockage, extended replacement cycle of consumables, and the application of predictive maintenance reminder mechanism, the annual maintenance downtime of the device of this invention is reduced from 168 hours in the traditional process to 48 hours, a reduction of 71.4%, which significantly improves the effective operating time of the production line.

[0123] (3) Specific verification of the impact of humidity control logic on catalyst lifetime To further verify the protective effect of the humidity control logic between wet electrostatic precipitator and catalytic oxidation on catalyst life, two modes, "humidity control enabled (electric heating dehumidification)" and "humidity control disabled," were run on the same device, and the running time required for the catalyst activity to decay to 80% of its initial activity was recorded. The results are shown in Table 3. Table 3

[0124] Experimental results show that after humidity control is enabled, the catalyst activity retention time is extended by approximately 55.6%, and the overall catalyst replacement / regeneration cycle is extended by 56.3%. This effect is attributed to the fact that electric heating dehumidification controls the relative humidity of the exhaust gas entering the catalytic oxidation reaction tower to below 30%, effectively inhibiting the competitive adsorption of water vapor on the surface of the precious metal catalyst and the hydrothermal aging effect, thus verifying the significant technical effect of the present invention in the synergistic mechanism of "humidity control and activity retention".

[0125] Comparative experimental conclusions The comparative experimental data above show that, through the collaborative design of hierarchical deep processing units and the dynamic optimization of key parameters by machine learning models, the present invention has achieved significantly better technical effects than traditional processes and single control modes in terms of tar blockage control, system energy consumption reduction, consumable service life extension and emission stability improvement. The units form a non-obvious synergistic effect, rather than a simple combination of existing technical features.

[0126] This invention features a high degree of automation, effectively solving problems such as clogging, unstable efficiency, and frequent maintenance encountered in traditional processes, achieving long-term, efficient, and compliant purification results. This invention can be widely applied to the treatment of waste gas from high-temperature carbonization processes in carbon fiber production enterprises, and is also suitable for other industrial waste gas treatment scenarios containing tar, cyanide, volatile organic compounds, and large temperature fluctuations.

[0127] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. In some cases, the actions or steps described in the specification and claims can be performed in a different order than that shown in the embodiments, and the desired result can still be achieved. In addition, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result; in some embodiments, multitasking and parallel processing are also feasible or advantageous.

[0128] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing the relevant hardware to implement them. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0129] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations can be made to the embodiments of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A carbon fiber production off-gas treatment device characterized by, include: The waste gas collection device includes a collection pipe for connecting to a high-temperature waste gas source, a gas collection hood installed at the beginning of the collection pipe, and a centrifugal fan for providing power for waste gas transportation. The pretreatment unit includes a filter and a condensation module, wherein the filter includes a multi-layer composite filter structure and the condensation module includes a shell-and-tube condenser. The core processing unit includes a wet electrostatic precipitator and a catalytic oxidation reaction tower. The catalytic oxidation reaction tower includes a tower body, a multi-layer catalytic reaction layer disposed inside the tower body, a temperature control device, and a third temperature sensor. The post-processing unit includes an activated carbon adsorption column and a molecular sieve drying tower; The main control module and online monitoring equipment, including the frequency converter, various sensors and temperature regulation device of the centrifugal fan, are all electrically connected to the main control module; The main control module has a built-in parameter adjustment model based on machine learning. The input of the parameter adjustment model includes at least the exhaust gas flow rate, the concentration of key components and the reaction temperature, and the output is the optimal temperature set point of the catalytic oxidation reaction tower and / or the speed correction coefficient of the centrifugal fan.

2. The carbon fiber production off gas treatment device according to claim 1, characterized by, The gas collection hood is a trumpet-shaped flange pipe, and guide vanes are installed inside the flange pipe. The waste gas collection device also includes a first temperature sensor for detecting the temperature inside the collection pipe.

3. The carbon fiber production off gas treatment device according to claim 1, characterized by, The multi-layer composite filtration structure includes an outer coarse filter, a middle fine filter, and an inner fine filter arranged sequentially from the outside to the inside; the tubular condenser is equipped with a liquid level detector and a second temperature sensor.

4. The carbon fiber production off gas treatment device according to claim 1, characterized by, The multilayer catalytic reaction layer includes multiple honeycomb ceramic catalyst supports and noble metal catalysts supported on the catalyst supports; the temperature control device includes an electric heater and a coil-type cooling water heat exchanger.

5. The carbon fiber production off gas treatment device according to claim 1, characterized by, The post-processing unit also includes an online monitoring device for detecting emission gas parameters. The online monitoring device includes a gas concentration sensor group, a wireless transmission module, and a microcontroller. The microcontroller communicates with the main control module through the wireless transmission module.

6. The carbon fiber production off gas treatment device according to claim 1, characterized by, The centrifugal fan is electrically connected to the main control module via a frequency converter. The main control module adjusts the speed of the centrifugal fan via the frequency converter based on the estimated value of the exhaust gas emission or the pipeline pressure signal.

7. The carbon fiber production off gas treatment device according to claim 1, characterized by, The filter is equipped with a differential pressure sensor for detecting the pressure difference across it, and the drain port of the tubular condenser is connected to an automatic drain valve.

8. The carbon fiber production off gas treatment device according to claim 1, characterized by, The main control module includes a timing module, which is used to record the usage time of the catalyst, activated carbon, and molecular sieve and to provide predictive maintenance reminders.

9. The carbon fiber production waste gas treatment device according to claim 1, characterized in that, A dehumidification section is provided between the wet electrostatic precipitator and the catalytic oxidation reaction tower. The main control module controls the start and stop of the dehumidification section according to the humidity signal to control the relative humidity of the exhaust gas entering the catalytic oxidation reaction tower to below 30%.

10. A method for treating waste gas from carbon fiber production, characterized in that, The waste gas treatment using the carbon fiber production waste gas treatment device as described in any one of claims 1-9 includes the following steps: S1. Waste gas collection and transportation: Start the centrifugal fan, and the high-temperature waste gas enters the collection pipeline through the gas collection hood; S2. Pretreatment: After being filtered by a filter, the exhaust gas enters the condensation module for cooling and oil removal; S3, Core Purification: The condensed exhaust gas passes through a wet electrostatic precipitator to capture fine particulate matter, and then enters the catalytic oxidation reaction tower; S4. Catalytic oxidation: During the operation of the catalytic oxidation reaction tower, the main control module uses a built-in machine learning-based parameter adjustment model to dynamically adjust the temperature and / or the speed of the centrifugal fan inside the catalytic oxidation reaction tower. S5. Post-treatment: The waste gas after catalytic oxidation is discharged after passing through an activated carbon adsorption column and a molecular sieve drying tower in sequence.

Images