Automatic control method for air introduction amount of dry quenching furnace

Abstract

The invention discloses an automatic control method for the air introduction amount of a dry quenching furnace, and relates to the technical field of dry quenching, and the method comprises the following steps: collecting CO and H concentration distribution data and dust concentration gradient data in an annular flue of the dry quenching furnace, and boiler inlet temperature and circulating gas flow in real time; establishing a multi-parameter coupled dynamic prediction model, and predicting coke burn-out rates and boiler thermal efficiency under different air introduction amounts; an optimal air introduction amount threshold value is calculated through a model prediction control algorithm; on the basis of the optimal air introduction amount threshold value, the opening degree of an electric valve of an air introduction pipe is dynamically adjusted, meanwhile, displacement compensation of a servo mechanism at the position of an introduction opening in the annular flue is triggered, and an introduction point is made to be located in the optimization interval of the CO concentration, the Hconcentration and the dust concentration all the time; and a self-adaptive correction module is established, when it is detected that coke maturity is abnormal and coke discharge temperature fluctuation is abnormal, a compensation control mode based on deep reinforcement learning is automatically switched, and dynamic prediction model parameters are updated.

Description

Technical Field

[0001] The present invention relates to the technical field of coke dry quenching, and specifically relates to an automatic control method for the air intake amount of a coke dry quenching furnace. Background Art

[0002] Coke dry quenching is a process of quenching red coke using an inert circulating gas, mainly N2, in a closed system. The circulating gas exchanges heat with the red coke in the coke dry quenching furnace, absorbs the sensible heat of the red coke and then increases in temperature, enters the boiler to generate steam, and the cooled circulating gas is pressurized by a fan and returned to the coke dry quenching furnace for reuse; during this process, the release of residual volatile components in the coke and the combustion reaction of the inhaled air with the coke will generate combustible components such as CO and H2 in the circulating gas; as the cycle progresses, the content of these combustible components will continuously increase, and if it reaches a certain concentration, it may cause safety problems such as explosion, so the combustible components in the circulating gas must be controlled within a certain range.

[0003] As an important part of the modern coking industry, the purpose of coke dry quenching is to recover the sensible heat of red coke through the circulating gas to achieve effective heat utilization and reduce environmental pollution. However, in traditional coke dry quenching processes, the control of the air intake amount often relies on manual experience, resulting in unstable coke burn-off rates and low energy utilization efficiency. Currently, some coke dry quenching systems use fixed parameters to control the air intake amount or automatic control based on simple rules, but these methods do not fully consider the complexity of the production conditions and cannot achieve optimal control of the coke burn-off rate. Summary of the Invention

[0004] To solve the above technical problems, an automatic control method for the air intake amount of a coke dry quenching furnace is provided. This technical solution solves the problems in the above traditional coke dry quenching process, where the control of the air intake amount often relies on manual experience, resulting in unstable coke burn-off rates and low energy utilization efficiency. Some coke dry quenching systems use fixed parameters to control the air intake amount or automatic control based on simple rules, but these methods do not fully consider the complexity of the production conditions and cannot achieve optimal control of the coke burn-off rate.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An automatic control method for the air intake amount of a coke dry quenching furnace, comprising: Real-time collecting the CO and H2 concentration distribution data, dust concentration gradient data in the annular flue of the coke dry quenching furnace, as well as the boiler inlet temperature and the circulating gas flow rate; Establishing a multi-parameter coupled dynamic prediction model, which is based on the oxidation reaction kinetic equations of CO and H2 in the circulating gas, and combines the correction coefficient of the dust concentration on the gas diffusion coefficient to predict the coke burn-off rate and boiler thermal efficiency under different air intake amounts; Through the model predictive control algorithm, taking the minimization of coke burnout rate and the maximization of boiler thermal efficiency as the dual objective functions, calculate the optimal air intake threshold; Based on the optimal air intake threshold, dynamically adjust the opening of the electric valve of the air intake pipe, and at the same time trigger the displacement compensation of the servo mechanism at the air inlet position in the annular flue, so that the inlet point is always in the optimized range of CO concentration, H2 concentration, and dust concentration; establish an adaptive correction module. When abnormal coke maturity or abnormal fluctuation of coke discharging temperature is detected, automatically switch to the compensation control mode based on deep reinforcement learning, and update the parameters of the dynamic prediction model.

[0006] Preferably, the real-time acquisition of the CO and H2 concentration distribution data, dust concentration gradient data in the annular flue of the coke dry quenching furnace, as well as the boiler inlet temperature and the circulating gas flow rate specifically includes: Adopt a distributed laser gas analyzer, set detection points at the circumferential positions of the flue, and transmit data in real time through high-temperature resistant fiber optic sensors; Divide the flue from the top to the bottom into several gradient layers, and set electrostatic dust sensors on each layer; combine the dust deposition rate correction model to dynamically adjust the calculated value of the gas diffusion coefficient, and the particle size distribution of dust needs to be collected synchronously; Adopt a K-type thermocouple array, evenly arrange several temperature measurement points along the circumferential direction of the boiler inlet pipe, set a temperature fluctuation threshold, and trigger the adaptive correction module when exceeded; combine the distributed fiber optic sensor to obtain the spatial distribution of the temperature field and correct the single-point measurement error; Use a thermal mass flowmeter, install it on the outlet pipe of the circulating fan, and monitor the total flow rate in real time; synchronously collect the flow distribution ratio of the branch pipes for verifying the data of the main flowmeter; the flow data needs to be coupled with the gas composition and temperature data to calculate the actual gas density and correct the flow value.

[0007] Preferably, for the establishment of the dynamic prediction model with multi-parameter coupling, the model is based on the oxidation reaction kinetic equations of CO and H2 in the circulating gas, combined with the correction coefficient of the gas diffusion coefficient by the dust concentration, to predict the coke burnout rate and boiler thermal efficiency under different air intake amounts, specifically including: Introduce the dust concentration gradient data, correct the gas diffusion coefficient, substitute the corrected diffusion coefficient into the mass transfer equation, and calculate the actual oxidation rates of CO and H2 on the coke surface; Based on the volume fractions of CO2 and CO in the circulating gas and the air intake amount, calculate the carbon loss amount per unit time, and dynamically correct it in combination with the dust deposition rate model; Predict the boiler thermal efficiency through the energy balance equation and thermal efficiency optimization calculation; Couple the circulating gas flow rate, temperature field, and pressure field data, establish a three-dimensional heat transfer - mass transfer coupling equation, and use cross-validation to compare the historical production data with the predicted values.

[0008] Preferably, for establishing a dynamic prediction model with multi-parameter coupling, the model specifically includes, based on the oxidation reaction kinetic equations of CO and H2 in the circulating gas: Establish the oxidation reaction rate equations for CO and H2: In the formula, rCO and rH2 are the consumption amounts of CO and H2 per unit volume per unit time; kCO and kH2 are the pre-exponential factors related to temperature, representing the reaction rate at unit concentration; [CO], [O2], [H2] are the volume concentrations of the reactants in the gas mixture; m, n, p, q are the degrees of dependence of the reaction rate on the concentrations of each reactant; Ea is the minimum energy threshold required for the reaction to occur, determined by the potential barrier of the reaction path; R is the gas constant; T is the absolute temperature of the reaction system; e -Ea / (RT) is the exponential part of the Arrhenius equation, representing the exponential effect of temperature on the reaction rate.

[0009] Preferably, for coupling the circulating gas flow rate, temperature field, and pressure field data, establishing a three-dimensional heat transfer - mass transfer coupling equation, and using cross-validation to compare the historical production data with the predicted values, it specifically includes: Establishing a three-dimensional heat transfer - mass transfer coupling equation generally requires considering the interaction of various factors such as the circulating gas flow rate, temperature field, and pressure field, describing the heat transfer and distribution in the coke dry quenching furnace, and combining the equation of the mass transfer process to achieve the comprehensive prediction of the coke burn-off rate and the boiler thermal efficiency. Its general form is expressed as: In the formula, ρ represents the gas density, with the unit of kilograms per cubic meter; c p represents the specific heat capacity, with the unit of joules per kilogram Kelvin; T represents the temperature, with the unit of Kelvin; t represents the time, with the unit of seconds; k represents the thermal conductivity, with the unit of watts per meter Kelvin; represents the heat source term, with the unit of watts per cubic meter.

[0010] Preferably, for calculating the optimal air inlet threshold through the model predictive control algorithm with the minimum coke burn-off rate and the maximum boiler thermal efficiency as the dual objective functions, it specifically includes: In the model predictive control algorithm, define an optimization objective function that comprehensively considers multiple objectives and constraint conditions of the system. The optimization objective function is expressed as: J = w1·Φ + w2·(1 - η) In the formula, J represents the value of the comprehensive optimization objective function, which is used to evaluate the system performance under different air intake amounts; Φ represents the coke burnout rate, with the unit of percentage, reflecting the mass loss ratio of coke during the dry quenching process; η represents the boiler thermal efficiency, with the unit of percentage, reflecting the efficiency of the boiler in converting the chemical energy of fuel into heat energy and effectively utilizing it; w1 and w2 are the weight coefficients of the coke burnout rate and the boiler thermal efficiency in the comprehensive objective function respectively, used to balance the relative importance of the two objectives, and are dimensionless.

[0011] Preferably, the calculation of the optimal air intake amount threshold by using the model predictive control algorithm with the minimization of the coke burnout rate and the maximization of the boiler thermal efficiency as the dual objective functions specifically includes: To find the optimal air intake amount threshold that minimizes the comprehensive optimization objective function J, it is achieved by solving the optimization problem, which is expressed as: In the formula, is the optimal air intake amount threshold, u represents the air intake amount; Φ(u) represents the coke burnout rate when the air intake amount is u; η(u) represents the boiler thermal efficiency when the air intake amount is u; w1 and w2 are the weight coefficients of the coke burnout rate and the boiler thermal efficiency in the comprehensive objective function respectively.

[0012] Preferably, the dynamic adjustment of the opening of the electric valve of the air intake pipe based on the optimal air intake amount threshold, and at the same time triggering the displacement compensation of the servo mechanism at the inner inlet position of the annular flue, so that the inlet point is always in the optimization interval of CO concentration, H2 concentration, and dust concentration specifically includes: Based on the optimal air intake amount threshold, convert this threshold into the opening set value of the electric valve, and through the control system of the electric valve, dynamically adjust the opening of the valve to accurately control the air intake amount; During the adjustment process, the actual opening of the valve is monitored in real time and compared with the set value, and feedback control is performed according to the deviation; displacement compensation is performed on the servo mechanism at the inner inlet position of the annular flue, and according to the detected concentration data and the preset optimization interval, the adjustment amount of the inlet position is calculated, and through the control driving device of the servo mechanism, the accurate displacement of the inlet position is achieved; During the displacement compensation process, consider the response speed and accuracy of the system to ensure that the inlet position can be adjusted in place in a timely and accurate manner.

[0013] Preferably, the dynamic adjustment of the opening of the electric valve of the air intake pipe based on the optimal air intake amount threshold, and at the same time triggering the displacement compensation of the servo mechanism at the inner inlet position of the annular flue, so that the inlet point is always in the optimization interval of CO concentration, H2 concentration, and dust concentration specifically includes: Establish a concentration monitoring system, including a gas sensor and a dust concentration monitor. According to the concentration change trend and the model prediction results, adjust the actions in advance to avoid the concentration exceeding the optimized range. Linkage control is achieved between the adjustment of the electric valve opening and the displacement compensation of the servo mechanism to ensure their coordination and jointly achieve the optimized control of the air inlet volume and the inlet position.

[0014] Preferably, an adaptive correction module is established. When abnormal coke maturity and abnormal fluctuation of the coke discharging temperature are detected, it automatically switches to a compensation control mode based on deep reinforcement learning. The specific steps for updating the dynamic prediction model parameters include: Real-time monitor the parameters of coke maturity and coke discharging temperature through sensors, and use data processing and analysis algorithms to determine whether abnormal situations occur. When abnormal situations are detected, the system automatically switches from the normal control mode to the compensation control mode based on deep reinforcement learning. In the compensation control mode, the deep reinforcement learning algorithm generates corresponding compensation control strategies according to the current state information and environmental feedback of the system to adjust the control variables of the air inlet volume and the inlet position. Update the parameters of the dynamic prediction model according to the new control strategy and actual operation data to improve the model's prediction ability for the future state of the system.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention proposes to construct a dynamic prediction model considering the interaction of multiple factors, which can accurately predict the coke burnout rate and boiler thermal efficiency under different air inlet volumes, providing a reliable theoretical basis for the formulation of optimized control strategies, overcoming the limitations of the traditional model's single parameter consideration, and improving the accuracy and practicality of prediction; the dual-objective optimization strategy reduces coke loss while fully exploring the potential of the boiler's thermal efficiency, realizes the effective control of production costs and the efficient utilization of energy, and improves the economic benefits and market competitiveness of enterprises; the coordinated control mechanism ensures that the inlet point is always within the optimized range of CO concentration, H2 concentration, and dust concentration, further optimizes the chemical reaction environment during the coke dry quenching process, improves the stability of coke quality, and reduces unnecessary energy waste and equipment loss; the adaptive correction module can automatically switch to the compensation control mode based on deep reinforcement learning and update the parameters of the dynamic prediction model. This intelligent fusion mechanism has the ability of self-diagnosis, self-adjustment, and self-optimization, can quickly adapt to working condition changes and external disturbances, ensures the stable operation of the entire coke dry quenching system, and reduces the manual intervention cost and operation risk. Description of the Drawings

[0016] Figure 1 It is a flowchart of an automatic control method for the air inlet volume of a coke dry quenching furnace. Detailed implementation manners

[0017] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments in the following description are only examples, and other obvious variants can be conceived by those skilled in the art.

[0018] Referring to Figure 1 As shown, an automatic control method for the air intake amount of a coke dry quenching furnace includes: Real-time collecting the CO and H2 concentration distribution data, dust concentration gradient data, boiler inlet temperature and circulating gas flow rate in the annular flue of the coke dry quenching furnace; Establishing a dynamic prediction model with multi-parameter coupling, where the model is based on the oxidation reaction kinetic equations of CO and H2 in the circulating gas, and combines the correction coefficient of the dust concentration on the gas diffusion coefficient to predict the coke burn-off rate and boiler thermal efficiency under different air intake amounts; Through the model predictive control algorithm, taking the minimization of the coke burn-off rate and the maximization of the boiler thermal efficiency as the dual objective functions, calculating the optimal air intake amount threshold; Based on the optimal air intake amount threshold, dynamically adjusting the opening of the electric valve of the air intake pipe, and at the same time triggering the displacement compensation of the servo mechanism at the air inlet position in the annular flue, so that the inlet point is always in the optimized range of CO concentration, H2 concentration and dust concentration; establishing an adaptive correction module, when abnormal coke maturity or abnormal fluctuation of the coke discharging temperature is detected, automatically switching to the compensation control mode based on deep reinforcement learning to update the parameters of the dynamic prediction model.

[0019] It should be noted that the real-time collection of CO and H2 concentration distribution data in the annular flue of the coke dry quenching furnace: Using a distributed laser gas analyzer: This analyzer can detect specific components in the gas, such as CO and H2, in real time through laser spectroscopy technology, and has the advantages of high precision, fast response and non-contact measurement; in order to comprehensively understand the gas concentration distribution at different positions in the flue, multiple detection points are evenly set in the circumferential direction of the flue to ensure the representativeness and accuracy of the data; the fiber optic sensor has the advantages of high temperature resistance, anti-electromagnetic interference and low transmission loss, and can transmit the detected data to the control system in real time and stably.

[0020] Collection of dust concentration gradient data: In order to obtain the variation trend of dust concentration in the flue height direction, the flue is divided into multiple gradient layers from top to bottom for monitoring; the electrostatic dust sensor measures the dust concentration by utilizing the charge characteristics of dust in the electric field, and has high sensitivity and accuracy; considering the dust deposition situation in the flue, a dust deposition rate correction model is established to dynamically adjust the calculated value of the gas diffusion coefficient so as to more accurately reflect the actual working conditions; the particle size of the dust will affect its deposition rate and the degree of influence on gas diffusion, so it is necessary to synchronously collect the dust particle size distribution data to provide more comprehensive information for the correction model.

[0021] Boiler inlet temperature acquisition: Adopt a K-type thermocouple array: The K-type thermocouple is a commonly used temperature measurement element with characteristics such as a wide temperature measurement range, high accuracy, and fast response. Composing it into an array can measure the temperatures of multiple points simultaneously; in order to accurately grasp the temperature distribution at the boiler inlet, multiple temperature measurement points are evenly arranged in the circumferential direction of the pipeline to ensure the comprehensiveness and accuracy of temperature measurement; a threshold is set according to the temperature fluctuation range under normal operating conditions. When the actual measured temperature exceeds this threshold, it is considered that the system may be abnormal, and the adaptive correction module is automatically triggered for adjustment; the distributed optical fiber sensor can continuously measure the temperature along the length of the optical fiber to obtain the spatial distribution of the temperature field, and use these data to correct the error of the single-point measurement of the thermocouple to improve the reliability of temperature measurement.

[0022] Circulating gas flow rate acquisition: The thermal mass flowmeter measures the gas flow rate based on the heat transfer principle and has advantages such as directly measuring the mass flow rate and being insensitive to gas composition changes; installing the flowmeter at the outlet pipeline of the circulating fan can accurately measure the total amount of circulating gas entering the dry coke quenching furnace; the total flow rate data of the circulating gas is obtained in real time through the flowmeter to provide a basis for the flow control and balance of the system; flow measurement devices are also set on the branch pipes of the circulating gas to collect the flow distribution ratio of each branch pipe. By comparing and verifying these data with the data of the main flowmeter, the accuracy of flow measurement is ensured.

[0023] The flow rate data needs to be coupled with the gas composition and temperature data to calculate the corrected flow rate value at the actual gas density: Since the density of the gas is affected by the composition and temperature, the flow rate data is coupled with the corresponding gas composition and temperature data for calculation to obtain the corrected flow rate value at the actual gas density so as to more accurately reflect the true flow rate situation.

[0024] Based on the optimal air inlet threshold, dynamically adjusting the opening degree of the electric valve of the air inlet pipe includes: Converting the threshold to the opening degree setting value: According to the relationship model established in advance between the air inlet amount and the opening degree of the electric valve, the calculated optimal air inlet threshold is converted into the corresponding opening degree setting value of the electric valve; Dynamically adjust the valve opening: Input the opening set value into the control system of the electric valve. The control system drives the actuator of the electric valve according to the set value to dynamically adjust the valve opening, so as to precisely control the air inlet volume and ensure that the actual air inlet volume matches the optimal air inlet volume threshold.

[0025] Feedback control during the adjustment process: Real-time monitor the actual opening: During the adjustment process, use the position sensor installed on the electric valve to real-time monitor the actual opening of the valve, and feedback the monitoring data back to the control system; Deviation feedback control: The control system compares the actual opening with the set value. If there is a deviation, according to the magnitude and direction of the deviation, through the control algorithm of the PID (Proportional-Integral-Derivative) controller, adjust the driving signal of the electric valve to reduce the deviation and make the actual opening of the valve quickly and accurately reach the set value.

[0026] The displacement compensation of the servo mechanism at the air inlet position in the annular flue includes: Calculate the adjustment amount of the air inlet position: According to the detected CO concentration, H2 concentration and dust concentration data, combined with the preset optimization interval, determine whether the current air inlet point position needs to be adjusted; if adjustment is needed, through the established relationship model between the concentration distribution and the adjustment amount of the air inlet position, calculate the specific adjustment amount of the air inlet position.

[0027] Control and drive the servo mechanism: Input the calculated adjustment amount into the control and drive device of the servo mechanism. The control and drive device drives the servo motor according to the adjustment amount to achieve precise displacement of the air inlet position, so that the air inlet point is always in the optimization interval of CO concentration, H2 concentration and dust concentration; System response speed and accuracy guarantee during the displacement compensation process: System response speed: During the displacement compensation process, considering the dynamic and real-time requirements of the coke dry quenching process, the control system of the servo mechanism needs to have a fast response speed, which can be achieved by selecting high-performance servo motors, optimizing control algorithms, reducing the inertia of mechanical transmission links, etc., to ensure that the air inlet position can be adjusted in place in time.

[0028] System accuracy: In order to ensure the adjustment accuracy of the air inlet position, it is necessary to use high-precision position sensors and servo drivers, and at the same time ensure the transmission accuracy of the air inlet position adjustment in the mechanical structure design; in addition, it is also possible to use the closed-loop control method to real-time monitor the actual position of the air inlet, compare it with the target position, and make fine adjustments according to the deviation to improve the overall accuracy of the system.

[0029] Generation of compensation control strategy: The algorithm constructs a policy network that takes the state information of the system as input and outputs corresponding control actions, such as adjusting the air intake volume and changing the position of the intake port. During the training process, the algorithm continuously tries different control actions and evaluates the effects of these actions based on the reward signals feedback by the environment, so as to optimize the parameters of the policy network, making the generated compensation control strategy maximize the performance and production quality of the system.

[0030] Adjustment of control variables: According to the generated compensation control strategy, the system will correspondingly adjust control variables such as the air intake volume and the position of the intake port. For example, if abnormal coke maturity is detected, the algorithm may generate a control strategy to increase or decrease the air intake volume to adjust the oxidation degree of the coke and restore its maturity to the normal range. At the same time, the position of the intake port may also be adjusted to optimize the gas distribution and improve the overall operation effect of the system.

[0031] The usage process of the present invention is as follows: Step 1: Adopt a distributed laser gas analyzer, set detection points at the circumferential positions of the flue, and transmit data in real time through high-temperature-resistant fiber optic sensors. Step 2: Divide the flue from top to bottom into several gradient layers, and set electrostatic dust sensors on each layer. Combine the dust deposition rate correction model to dynamically adjust the calculated value of the gas diffusion coefficient, and the particle size distribution of the dust needs to be collected synchronously. Step 3: Adopt a K-type thermocouple array, evenly arrange several temperature measurement points along the circumferential direction of the boiler inlet pipe, set a temperature fluctuation threshold, and trigger the adaptive correction module when it is exceeded. Combine the distributed fiber optic sensor to obtain the spatial distribution of the temperature field and correct the single-point measurement error. Step 4: Use a thermal mass flowmeter, install it on the outlet pipe of the circulation fan, and monitor the total flow in real time. Synchronously collect the flow distribution ratio of the branch pipes for verifying the data of the main flowmeter. The flow data needs to be coupled with the gas composition and temperature data to calculate the actual gas density and correct the flow value. Step 5: Introduce the dust concentration gradient data to correct the gas diffusion coefficient, substitute the corrected diffusion coefficient into the mass transfer equation, and calculate the actual oxidation rates of CO and H2 on the coke surface. Step 6: Calculate the carbon loss amount per unit time based on the volume fractions of CO2 and CO in the circulating gas and the air intake volume, and dynamically correct it in combination with the dust deposition rate model. Step 7: Calculate the prediction of the boiler thermal efficiency through the energy balance equation and thermal efficiency optimization. Step 8: Couple the circulating gas flow, temperature field, and pressure field data, establish a three-dimensional heat transfer - mass transfer coupling equation, and use cross-validation to compare the historical production data with the predicted values. Step 9: Through the model predictive control algorithm, with the minimization of coke burnout rate and the maximization of boiler thermal efficiency as the dual objective functions, calculate the optimal air inlet threshold; Step 10: Based on the optimal air inlet threshold, convert this threshold into the opening set value of the electric valve, and through the control system of the electric valve, dynamically adjust the opening of the valve to accurately control the air inlet volume; Step 11: During the adjustment process, monitor the actual opening of the valve in real time, compare it with the set value, and perform feedback control according to the deviation; Step 12: Perform displacement compensation on the servo mechanism at the air inlet position in the annular flue. According to the detected concentration data and the preset optimization interval, calculate the adjustment amount of the air inlet position, and through the control drive device of the servo mechanism, achieve the accurate displacement of the air inlet position; Step 13: During the displacement compensation process, consider the response speed and accuracy of the system to ensure that the air inlet position can be adjusted in place in a timely and accurate manner; Step 14: Establish a concentration monitoring system, including gas sensors and dust concentration monitors. According to the concentration change trend and the model prediction results, perform adjustment actions in advance to avoid the concentration exceeding the optimization interval; Step 15: Realize the interlocking control between the electric valve opening adjustment and the servo mechanism displacement compensation to ensure their coordination and jointly achieve the optimal control of the air inlet volume and the air inlet position; Step 16: Through sensors, monitor the parameters of coke maturity and coke discharging temperature in real time, and use data processing and analysis algorithms to judge whether abnormal conditions occur; Step 17: When abnormal conditions are detected, the system automatically switches from the normal control mode to the compensation control mode based on deep reinforcement learning; Step 18: In the compensation control mode, the deep reinforcement learning algorithm generates corresponding compensation control strategies according to the current system state information and environmental feedback to adjust the control variables of the air inlet volume and the air inlet position; Step 19: According to the new control strategy and the actual operation data, update the parameters of the dynamic prediction model to improve the model's prediction ability for the future state of the system.

[0032] In summary, the advantages of the present invention are as follows: The constructed multi-parameter coupled dynamic prediction model comprehensively considers the oxidation reaction kinetic equations of CO and H2 in the circulating gas and the correction coefficient of dust concentration on the gas diffusion coefficient, and can accurately predict the coke burn-off rate and boiler thermal efficiency under different air intake amounts, overcoming the limitations of traditional models and improving prediction accuracy; The model predictive control algorithm is adopted, with the minimization of the coke burn-off rate and the maximization of the boiler thermal efficiency as the dual objective functions, to calculate the optimal air intake threshold, realizing the effective control of production costs and the efficient utilization of energy, and enhancing the economic benefits of enterprises; Based on the optimal air intake threshold, the opening degree of the electric valve is dynamically adjusted, and at the same time, the displacement compensation of the servo mechanism is triggered to ensure that the intake point is always in the optimized range, further optimizing the chemical reaction environment, improving the quality stability of coke, reducing energy waste and equipment loss; An adaptive correction module is established, which automatically switches to the compensation control mode based on deep reinforcement learning and updates the model parameters when an anomaly is detected, enabling the system to have the capabilities of self-diagnosis, adjustment and optimization, quickly adapting to changes in working conditions and external disturbances, ensuring stable operation, and reducing the cost of manual intervention and operation risks.

[0033] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification is only the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. An automatic control method for the air intake of a coke dry quenching furnace, characterized in that, Including: Real-time collect the CO and H2 concentration distribution data, dust concentration gradient data, boiler inlet temperature and circulating gas flow rate in the annular flue of the coke dry quenching furnace; Establish a dynamic prediction model with multi-parameter coupling. The model is based on the oxidation reaction kinetic equations of CO and H2 in the circulating gas, combined with the correction coefficient of the dust concentration on the gas diffusion coefficient, to predict the coke burn-off rate and boiler thermal efficiency under different air intake amounts; Through the model predictive control algorithm, with the minimization of the coke burn-off rate and the maximization of the boiler thermal efficiency as the dual objective functions, calculate the optimal air intake threshold; Based on the optimal air intake threshold, dynamically adjust the opening of the electric valve of the air intake pipe, and at the same time trigger the displacement compensation of the servo mechanism at the air inlet position in the annular flue, so that the inlet point is always in the optimized interval of CO concentration, H2 concentration and dust concentration; Establish an adaptive correction module. When abnormal coke maturity or abnormal fluctuation of the coke discharging temperature is detected, automatically switch to the compensation control mode based on deep reinforcement learning to update the parameters of the dynamic prediction model.

2. The automatic control method for the air intake of a coke dry quenching furnace according to claim 1, characterized in that, The real-time collection of the CO and H2 concentration distribution data, dust concentration gradient data, boiler inlet temperature and circulating gas flow rate in the annular flue of the coke dry quenching furnace specifically includes: Adopt a distributed laser gas analyzer, set detection points in the circumferential direction of the flue, and transmit data in real time through high-temperature resistant fiber optic sensors; Divide the flue from top to bottom into several gradient layers, and set electrostatic dust sensors on each layer; combine the dust deposition rate correction model to dynamically adjust the calculated value of the gas diffusion coefficient, and the particle size distribution of the dust needs to be collected synchronously; Adopt a K-type thermocouple array, evenly arrange several temperature measurement points along the circumferential direction of the boiler inlet pipeline, set a temperature fluctuation threshold, and trigger the adaptive correction module when it is exceeded; combine the distributed fiber optic sensor to obtain the spatial distribution of the temperature field and correct the single-point measurement error; use a thermal mass flowmeter, installed in the outlet pipeline of the circulating fan, to monitor the total flow rate in real time; synchronously collect the flow distribution ratio of the branch pipes for verifying the data of the main flowmeter; the flow data needs to be coupled with the gas composition and temperature data to calculate the actual gas density correction flow value.

3. The automatic control method for the air intake amount of a coke dry quenching furnace according to claim 2, characterized in that, The establishment of the dynamic prediction model with multi-parameter coupling. The model is based on the oxidation reaction kinetic equations of CO and H2 in the circulating gas, combined with the correction coefficient of the dust concentration on the gas diffusion coefficient, to predict the coke burn-off rate and boiler thermal efficiency under different air intake amounts specifically includes: Introduce the dust concentration gradient data, correct the gas diffusion coefficient, substitute the corrected diffusion coefficient into the mass transfer equation, and calculate the actual oxidation rate of CO and H2 on the coke surface; Based on the volume fractions of CO2 and CO in the circulating gas and the air intake amount, calculate the carbon loss amount per unit time, and dynamically correct it in combination with the dust deposition rate model; Predict the boiler thermal efficiency through the energy balance equation and thermal efficiency optimization calculation; Couple the circulating gas flow rate, temperature field and pressure field data, establish a three-dimensional heat transfer and mass transfer coupling equation, and use cross-validation to compare the historical production data with the predicted values.

4. The automatic control method for the air intake amount of a coke dry quenching furnace according to claim 3, characterized in that, The establishment of the dynamic prediction model with multi-parameter coupling. The model is based on the oxidation reaction kinetic equations of CO and H2 in the circulating gas specifically includes: Establish the oxidation reaction rate equations for CO and H2: Where, rCO and rH2 are the consumption rates of CO and H2 per unit volume per unit time; kCO and kH2 are the pre-exponential factors related to temperature, representing the reaction rate at unit concentration; [CO], [O2], [H2] are the volume concentrations of the reactants in the gas mixture; m, n, p, q characterize the dependence of the reaction rate on the concentrations of the respective reactants; E a is the minimum energy threshold required for the reaction to occur, determined by the potential barrier of the reaction path; R is the gas constant; T is the absolute temperature of the reaction system; e -Ea / (RT) is the exponential part of the Arrhenius equation, representing the exponential effect of temperature on the reaction rate.

5. The automatic control method for the air inlet volume of a coke dry quenching furnace according to claim 3, characterized in that Based on the coupled circulating gas flow rate, temperature field, and pressure field data, establish a three-dimensional heat transfer - mass transfer coupling equation. Using cross-validation, compare the historical production data with the predicted values, specifically including: Establishing a three-dimensional heat transfer - mass transfer coupling equation generally requires considering the interaction of various factors such as circulating gas flow rate, temperature field, and pressure field to describe the heat transfer and distribution in the coke dry quenching furnace. Combining with the equation of the mass transfer process, realize the comprehensive prediction of the coke burnout rate and boiler thermal efficiency. Its general form is expressed as: Where ρ represents the gas density, with the unit of kilograms per cubic meter; c p represents the specific heat capacity, with the unit of joules per kilogram Kelvin; T represents the temperature, with the unit of Kelvin; t represents the time, with the unit of seconds; k represents the thermal conductivity, with the unit of watts per meter Kelvin; represents the heat source term, with the unit of watts per cubic meter.

6. The automatic control method for the air intake amount of a coke dry quenching furnace according to claim 5, characterized in that, Through the model predictive control algorithm, with the minimization of the coke burnout rate and the maximization of the boiler thermal efficiency as the dual objective functions, calculate the optimal air inlet threshold, specifically including: In the model predictive control algorithm, define an optimization objective function that comprehensively considers multiple objectives and constraint conditions of the system. The optimization objective function is expressed as: J = w1·Φ + w2·(1 - η) In the formula, J represents the value of the comprehensive optimization objective function, which is used to evaluate the system performance under different air inlet amounts; Φ represents the coke burnout rate, in percentage, reflecting the mass loss ratio of coke during the coke dry quenching process; η represents the boiler thermal efficiency, in percentage, reflecting the efficiency of the boiler in converting the chemical energy of the fuel into heat energy and effectively utilizing it; w1 and w2 are the weight coefficients of the coke burnout rate and boiler thermal efficiency in the comprehensive objective function respectively, used to balance the relative importance of the two objectives, dimensionless.

7. The automatic control method for the air inlet volume of a coke dry quenching furnace according to claim 6, characterized in that, Through the model predictive control algorithm, with the minimization of the coke burnout rate and the maximization of the boiler thermal efficiency as the dual objective functions, calculate the optimal air inlet threshold, specifically including: To find the optimal air intake threshold that minimizes the comprehensive optimization objective function J, it is achieved by solving the optimization problem, expressed as: In the formula, is the optimal air intake threshold, u represents the air intake; Φ(u) represents the coke burnout rate when the air intake is u; η(u) represents the boiler thermal efficiency when the air intake is u; w1 and w2 are the weight coefficients of the coke burnout rate and the boiler thermal efficiency in the comprehensive objective function, respectively.

8. The automatic control method for the air intake amount of a coke dry quenching furnace according to claim 7, characterized in that Based on the optimal air inlet threshold, dynamically adjust the opening of the electric valve of the air inlet pipe, and at the same time trigger the displacement compensation of the servo mechanism at the air inlet position in the annular flue, so that the inlet point is always in the optimization interval of CO concentration, H2 concentration, and dust concentration, specifically including: Based on the optimal air inlet threshold, convert this threshold into the opening setting value of the electric valve, and through the control system of the electric valve, dynamically adjust the opening of the valve to accurately control the air inlet amount; During the adjustment process, real-time monitor the actual opening of the valve, compare it with the set value, and perform feedback control according to the deviation; Perform displacement compensation on the servo mechanism at the air inlet position in the annular flue. According to the detected concentration data and the preset optimization interval, calculate the adjustment amount of the air inlet position, and through the control drive device of the servo mechanism, realize the accurate displacement of the air inlet position; During the displacement compensation process, consider the response speed and accuracy of the system to ensure that the air inlet position can be adjusted in place in a timely and accurate manner.

9. The automatic control method for the air intake amount of a coke dry quenching furnace according to claim 8, characterized in that, Based on the optimal air inlet threshold, dynamically adjust the opening of the electric valve of the air inlet pipe, and at the same time trigger the displacement compensation of the servo mechanism at the air inlet position in the annular flue, so that the inlet point is always in the optimization interval of CO concentration, H2 concentration, and dust concentration, specifically including: Establish a concentration monitoring system, including gas sensors and dust concentration monitors, and based on the concentration change trend and model prediction results, perform adjustment actions in advance to avoid the concentration exceeding the optimization interval; Linkage control is achieved between the opening adjustment of the electric valve and the displacement compensation of the servo mechanism to ensure their coordination and jointly achieve the optimal control of the air intake volume and the inlet position.

10. The automatic control method for the air intake of a coke dry quenching furnace according to claim 9, characterized in that, The establishment of the adaptive correction module, when abnormal coke maturity and abnormal fluctuation of the coke discharging temperature are detected, automatically switches to the compensation control mode based on deep reinforcement learning. The specific update of the dynamic prediction model parameters includes: Real-time monitoring of the parameters of coke maturity and coke discharging temperature through sensors, and using data processing and analysis algorithms to judge whether abnormal conditions occur; When abnormal conditions are detected, the system automatically switches from the normal control mode to the compensation control mode based on deep reinforcement learning; In the compensation control mode, the deep reinforcement learning algorithm generates corresponding compensation control strategies according to the current system state information and environmental feedback to adjust the control variables of the air intake volume and the inlet position; According to the new control strategy and actual operation data, the parameters of the dynamic prediction model are updated to improve the model's prediction ability for the future state of the system.

Images