Intelligent control system and control method for urea pyrolysis

Abstract

The invention discloses a urea pyrolysis intelligent control system and a control method, belongs to the technical field of flue gas denitration, and solves the problem of how to improve the control precision and the working condition adaptability of a urea pyrolysis control system. Multi-source heterogeneous data are collected in real time according to a deployed sensor network, the multi-source heterogeneous data are synchronized through a PLC and then input into an edge computing node, a small sample reinforcement learning framework is generated by introducing WGAN-GP to generate virtual working condition data, and the working condition data which are verified to be effective are reserved in a data pool; according to the method, the highest denitration efficiency, the lowest ammonia escape, the lowest energy consumption and the lowest crystallization risk are taken as optimization objectives in an optimization layer, the urea injection amount and the combustor power are optimized on line based on a PPO reinforcement learning algorithm, the indexes are balanced, the system can continuously approach the optimal operation point, an optimal control instruction is calculated in a rolling mode based on MPC constraints in a control layer, and the control efficiency of the system is improved. An intelligent algorithm and an ammonia preparation process are deeply fused, and efficient, low-energy-consumption and low-ammonia-escape urea pyrolysis ammonia preparation is realized.

Description

Technical Field

[0001] This invention belongs to the field of flue gas denitrification technology and relates to an intelligent control system and control method for urea pyrolysis. Background Technology

[0002] Selective catalytic reduction (SCR) technology is the mainstream technology for flue gas denitrification in industrial facilities such as coal-fired power units, gas turbines, and cement kilns. Urea pyrolysis for ammonia production is gradually replacing liquid ammonia as the main source of reducing agent in SCR denitrification systems due to its high safety and easy raw material storage. Urea has the chemical formula CO(NH2)2. After pyrolysis, urea decomposes into ammonia (NH3) and isocyanate (HNCO), with the chemical formula CO(NH2)2→NH3+HNCO. Isocyanate (HNCO) further hydrolyzes to produce ammonia and carbon dioxide, with the chemical formula HNCO+H2O→NH3+CO2. However, existing urea pyrolysis systems generally suffer from the following defects in actual operation: (1) High energy consumption: Pyrolysis reactions are generally carried out at high temperatures of 300~650℃. Currently, the system mostly relies on high-power electric heaters, oil or gas burners to maintain this temperature, resulting in huge energy consumption and high operating costs. (2) Ammonia escape: Due to the mismatch between the urea injection volume and the NOx concentration in the flue gas, when there is an excess, the unreacted ammonia gas leaks and escapes downstream, causing waste of raw materials, blockage and corrosion of equipment. (3) Crystallization blockage: Due to poor atomization effect of urea solution and uneven temperature field in the furnace, intermediate products (such as isocyanate HNCO) are easily crystallized and deposited at the pyrolysis furnace and spray gun. (4) Poor dynamic adaptability: Due to the frequent changes in dynamic operating conditions such as boiler load fluctuations (such as changes in the calorific value of coal and start-up and shutdown conditions), traditional PID control is difficult to cope with such dynamic operating conditions, resulting in low system control accuracy and weak anti-interference ability.

[0003] Existing technologies, such as the invention patent with publication number CN107812449A, disclose a urea thermal denitrification control system. This system uses the total flow rate of the urea solution as the control variable and employs predictive control to provide the total flow rate of the urea solution, thereby achieving optimized ammonia injection control. However, this control system uses traditional predictive control, and the control strategy is still based on traditional mathematical models. Furthermore, since it only uses the "total flow rate of the urea solution" as a single control variable, it is still difficult to adapt to complex dynamic operating conditions. Summary of the Invention

[0004] The technical problem to be solved by this invention is how to improve the control accuracy and adaptability of the urea pyrolysis control system.

[0005] The present invention solves the above-mentioned technical problems through the following technical solutions: A smart control system for urea pyrolysis, comprising: The sensing layer is used to deploy a multimodal sensor network in the pyrolysis furnace to collect multi-source heterogeneous data in real time. The fusion layer is used to input the collected data into the edge computing node after synchronization by the PLC, construct the CFD-DEM digital twin synchronization data and output the prediction residual, and expand the effective working condition data based on the adversarial generative network. The decision-making layer is used to continuously calculate the optimal control command based on the online optimization of urea injection quantity and burner power, and switch the control mode of the decision-making layer according to abnormal conditions. The execution layer is used to receive control commands, adjust the spray gun and burner based on the control commands and dynamic zoning strategies, and actively intervene to remove crystals; The energy layer is used to recover waste heat from flue gas and preheat urea solution through the ORC unit, thus achieving an energy-saving closed loop.

[0006] Furthermore, the fusion layer includes: The data synchronization module is used to synchronize multi-source heterogeneous data through the field PLC at a first preset cycle, and uploads a multi-dimensional vector representing the running data using the OPC UA communication protocol, which is then input into the edge computing node. The data fusion module is used to add acoustic energy spectrum and PIV velocity field components to a multidimensional vector at a second preset period through edge computing nodes; the added vector is then packaged and stored in a time-series database. The data cleaning module is used to clean data by imputing missing values, EMA denoising, and Z-score normalization, and to construct derived features; The digital twin construction module is used to build a digital twin based on the CFD-DEM coupled model as a virtual reactor to simulate the reaction process in the pyrolysis furnace, and output the predicted residual by synchronizing physical field data in real time. The working condition generation module is used to generate virtual working condition data using WGAN-GP and to verify the working condition data using a CFD-DEM coupled model and Wasserstein-1 distance.

[0007] Furthermore, the output prediction residual σ in the digital twin construction module is specifically as follows: First, a CFD-DEM coupled model is constructed. Based on the input physical field data, the CFD-DEM coupled model solves the Navier-Stokes equation and the particle momentum equation, and outputs the predicted physical field data. Secondly, a gradient boosting tree is used to fit the nonlinear main trend of the predicted physical field data, and the physical residual Δ is output as the training label. The physical residual Δ is the difference between the predicted physical field data and the true value. Then, Gaussian process regression is used to train the model with training labels, outputting the predicted residual σ and the 95% confidence interval. The model bias is corrected based on the predicted residual σ, and the uncertainty of the prediction is quantified. When the predicted residual is greater than the preset residual threshold, the CFD-DEM fine mesh recalculation is triggered, and the updated model version is stored in the model library.

[0008] Furthermore, the working condition generation module includes: The generator unit is used to build the GAN generator, fake working condition data, and increase the amount of training data. The verification unit is used to perform preliminary verification of the generated operating condition data. Specifically, the generated operating condition data is input into the CFD-DEM coupled model. If the model output error is less than the error threshold, the generated operating condition data is considered to have passed the preliminary verification. If the model output error is greater than the error threshold, the generated operating condition data is considered to have failed the preliminary verification, and the process is returned to the generator unit. The discrimination unit is used to verify the consistency between the generated working condition data and the actual data distribution based on the Wasserstein-1 distance. If the verification condition of Wasserstein-1 distance is met, the unit will determine the consistency. When the distance threshold is reached, the generated working condition data is considered valid data; the Wasserstein-1 distance is represented using the following logic:

[0009] in, express and Wasserstein-1 distance, Represents the true data distribution. Indicates the distribution of synthetic data. Indicates all and For the joint distribution set of the edges, Indicates the joint distribution. Represents random data pairs. This represents the sum of all data pairs sampled from the joint distribution γ. ,calculate The average value of (x,y) γ represents random data pairs Follows a joint distribution γ, express L1 vector distance between them; The discrimination unit uses the following logic to represent the total loss function of the GAN:

[0010] in, This represents the total loss function of GAN. Indicating resistance to loss, Indicates the fingerprint regularization weight. Represents the fingerprint feature vector. This represents the output of the GAN generator. This represents a random noise vector.

[0011] Furthermore, the decision-making layer includes: The optimization layer aims to achieve the highest denitrification efficiency, lowest ammonia slip, lowest energy consumption, and lowest crystallization risk. Based on a multi-objective Bayesian optimizer and using EHVI as the acquisition function, it searches the Pareto front within the equivalent urea injection rate range of 3.0-8.0 pH. The urea injection rate and burner power are optimized online based on the PPO reinforcement learning algorithm. The control layer calculates the optimal control command based on rolling calculation under MPC constraints, updates the A, B, and C matrices of the MPC internal prediction model using the recursive least squares method, and outputs a multi-dimensional control vector. Safety layer, when the prediction uncertainty is greater than the uncertainty threshold or the crystallization risk index. When the CRI threshold is reached, an anomaly is considered, and the system immediately switches to a conservative PID control mode with a fixed pH of 7.0. Once the anomaly is resolved, it automatically resumes optimized layer control. ( ) Furthermore, the PPO reinforcement learning algorithm in the optimization layer optimizes the urea injection rate and burner power online by designing a reward function, using the following logical representation:

[0012] in, This represents the instantaneous reward value at time t. This represents the weighting coefficient for denitrification efficiency. Indicates denitrification efficiency. This represents the ammonia slip weighting coefficient. This indicates the concentration of ammonia slip at the outlet, in ppm. This represents the energy consumption weighting coefficient. This indicates the total power of the system, in kW. This represents the crystallization risk weighting coefficient. The crystallization risk index is represented by the following logic to quantify the tendency to crystallize a certain time in advance. :

[0013] in, This indicates the HNCO concentration at the furnace outlet at time t, in ppm. This indicates the upper limit of the permissible concentration of HNCO; This represents the lowest furnace wall temperature at time t, in K. Indicates the HNCO dew point temperature. This represents the Heaviside step function, which takes the value 1 if the condition is met, and 0 otherwise.

[0014] Furthermore, the recursive least squares method in the control layer employs an exponentially forgotten weighted objective function. This can be represented using the following logic:

[0015] in, Indicates the forgetting factor, express Time-regression vector, The vector of parameters to be estimated is formed by stacking the elements of matrices A, B, and C, which are updated by recursive least squares; The following logic represents the rolling calculation of the optimal control command:

[0016] in, Indicates the future starting from time k. Step-optimal control sequence Indicates the number of prediction time-domain steps. This represents the model's predicted output vector for the i-th step. This represents the reference trajectory vector for the i-th future step. This represents the output error weight matrix. This represents the control increment vector. This represents the control increment weight matrix.

[0017] Furthermore, the recovery of waste heat from flue gas and preheating of urea solution through the ORC unit in the energy layer specifically involves: The ORC unit recovers the high-temperature flue gas from the outlet of the pyrolysis furnace. The ORC unit absorbs heat and evaporates it into steam to drive the generator to generate electricity. The recovered electricity is used to drive the preheating of the urea solution. After the heat transfer is completed, the flue gas flows into the subsequent dust removal process.

[0018] Furthermore, it also includes a visualization layer for constructing a real-time monitoring panel, which displays trend graphs of multiple key variables, including NH3 concentration, NOx concentration, energy consumption, ORC power, and crystallization risk index.

[0019] This invention also provides a method for intelligent control of urea pyrolysis, comprising the following steps: S1, deploy a multimodal sensor network in the pyrolysis furnace to collect multi-source heterogeneous data in real time; S2, after the collected data is synchronized by the PLC, it is input into the edge computing node to construct the CFD-DEM digital twin synchronization data and output the prediction residual, and expand the effective working condition data based on the adversarial generative network; S3, based on online optimization of urea injection quantity and burner power, continuously calculates the optimal control command of the decision layer, and switches the control mode of the decision layer according to abnormal conditions; S4 receives control commands, adjusts the spray gun and burner based on the control commands and dynamic zoning strategy, and actively intervenes to remove crystals; S5 recovers waste heat from flue gas and preheats urea solution through the ORC unit, achieving a closed-loop energy-saving system.

[0020] The advantages of this invention are: (1) This invention collects multi-source heterogeneous data in real time based on the deployed sensor network, and inputs it into the edge computing node after synchronization by PLC. By introducing the WGAN-GP to generate a small sample reinforcement learning framework, it is used to solve the problem of scarce working condition data in actual industrial production. First, virtual working condition data is generated according to the GAN generator, expanding the original small amount of effective working condition data to more than 10 times the amount of training set data. Then, the effectiveness of the generated working condition data is initially verified based on the CFD-DEM coupling model. Then, the consistency between the distribution of the generated working condition data and the real data is verified according to the Wasserstein-1 distance. The quality of the data distribution output by the generator is improved through multiple games, and the verified effective working condition data is retained in the data pool. It can continuously iterate and quickly adapt to coal-fired units, gas-fired units, and cement kiln tail units, effectively improving the working condition adaptability of the control system.

[0021] (2) Based on the extended effective operating data, the present invention takes the highest denitrification efficiency, the lowest ammonia slip, the lowest energy consumption, and the lowest crystallization risk as the optimization objectives at the optimization layer. Based on the PPO reinforcement learning algorithm, the urea injection quantity and burner power are optimized online by designing a reward function to balance the above indicators, so that the system can continuously approach the optimal operating point. At the control layer, the optimal control command is calculated based on MPC constraints and the trend in the next 30 seconds is predicted, which effectively improves the control accuracy of the system.

[0022] When model mismatch, crystallization risk, or severe blockage occurs, the safety layer switches the decision layer's control mode to the traditional PID control mode. After the anomaly is resolved, it switches back to the optimization layer control, which greatly reduces the risk of abnormal situations, reduces the degree of manual intervention, ensures the availability of equipment and long-term safe and stable operation, and reduces economic losses caused by unplanned downtime.

[0023] (3) In addition, the present invention achieves energy-saving closed-loop by recovering waste heat from flue gas and preheating urea solution through the energy layer. It can significantly reduce the plant's electricity consumption and reduce carbon dioxide emissions by thousands of tons per year, and achieve efficient, low-energy consumption and low-ammonia escape urea pyrolysis to produce ammonia. It is particularly suitable for industrial boilers (thermal power generation, gas power generation, biomass power generation, waste incineration power generation and other industries that use fuel combustion to convert heat energy), industrial kilns (including but not limited to glass, cement, ceramics, building materials and other industries that use fuel combustion or electric heating to achieve high-temperature processes such as calcination, smelting and sintering of materials), sintering machines, blast furnaces, pelletizing, coking, chemical manufacturing, ship power systems, metallurgy, VOC treatment industries and other environmental island flue gas treatment scenarios that contain one or more of the following processes: desulfurization, denitrification and dust removal. The flue gas treatment processes include, but are not limited to: desulfurization processes (such as dry desulfurization, wet desulfurization, and semi-dry desulfurization, with desulfurizing agents including, but not limited to, calcium-based desulfurizing agents, sodium-based desulfurizing agents, and magnesium-based desulfurizing agents); denitrification processes (such as selective non-catalytic reduction (SNCR), selective catalytic reduction (SCR), organic polymer denitrification (PNCR), ammonia denitrification, and ammonia-free denitrification); and dust removal processes, such as electrostatic precipitators, bag filters, and electrostatic-bag composite dust collectors.

[0024] This invention provides an intelligent control system for urea pyrolysis that can be integrated into 30MW~1000MW coal / gas / cement kilns with zero modifications. It deeply integrates intelligent algorithms with ammonia production processes, achieving for the first time intelligent, energy-saving, and environmentally friendly urea pyrolysis control systems. This reduces the burden of manual operation and provides technical support for the green and low-carbon transformation of thermal power plants, chemical and building materials industries. It can be integrated into existing vacuum belt filter equipment, supports PLC or DCS platform deployment, and has high modularity and scalability. It is particularly suitable for environmental island systems in industries such as thermal power, steel, glass, cement, ceramics, and building materials that contain one or more of the following processes: desulfurization, denitrification, and dust removal, with any combination of processes. It has significant technical advantages and application prospects. Attached Figure Description

[0025] Figure 1 This is an overall structural diagram of a urea pyrolysis intelligent control system according to Embodiment 1 of the present invention; Figure 2 This is a flowchart illustrating the working condition generation module of Embodiment 1 of the present invention; Figure 3 This is a flowchart illustrating the digital twin construction module of Embodiment 1 of the present invention; Figure 4 This is a schematic diagram of the decision-making layer structure according to Embodiment 1 of the present invention; Figure 5 This is a pressure-specific enthalpy diagram of the ORC waste heat recovery system according to Embodiment 1 of the present invention; Figure 6This is a schematic diagram illustrating the actual benefits of multiple operating conditions in Embodiment 1 of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] The technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments: Example 1 like Figure 1 As shown, specifically, this invention discloses an intelligent control system for urea pyrolysis to produce ammonia, comprising: The sensing layer is used to deploy a multimodal sensor network in the pyrolysis furnace to collect multi-source heterogeneous data in real time, including temperature, flue gas velocity, NH3 concentration, HNCO concentration, NOx concentration, and droplet diameter data.

[0028] In this embodiment, the sensing layer is located in the pyrolysis furnace body and collects multi-source heterogeneous data such as temperature field, flow field, chemical field, and atomization quality of the pyrolysis furnace by deploying a multimodal sensor network. Specifically, the temperature field data is obtained by deploying a thermocouple array to collect temperature data of the pyrolysis furnace; the flow field data is obtained by deploying a laser particle velocimeter (PIV) to monitor the flue gas velocity distribution in real time and thus obtain flue gas velocity data; the chemical field data is obtained by deploying a TDLAS laser spectrometer to measure the concentration data of NH3, HNCO, and NOx online; and the atomization quality data is obtained by acquiring droplet images with a high-speed camera and analyzing the Sauter Mean Diameter (SMD) of the droplets according to existing image recognition algorithms to obtain droplet diameter data; wherein, the Sauter Mean Diameter (SMD) is a core parameter characterizing the average particle size of the spray droplets.

[0029] This embodiment provides a hardware deployment method for the sensing layer, specifically: a 32-channel thermocouple array is arranged at equal intervals along the height of the pyrolysis furnace from 0 to 6 meters; a TDLAS laser spectrometer is deployed across the furnace axis for 1.2 meters, with an accuracy of ±1 ppm; an acoustic sensor array is evenly arranged at 8 points around the furnace wall to collect the high-frequency acoustic emission characteristics of crystalline particles impacting the wall surface. All sensors can be fixed to PTFE gaskets via M5 to M12 threads, and the sensor data signals are connected to the PLC via shielded cables. The above hardware deployment method is only a preferred embodiment and can be adapted to meet actual needs.

[0030] Furthermore, the temperature field data can also be obtained by deploying any one of infrared thermal imagers, fiber optic gratings (FBG), wireless SAW sensors, or RTDs (platinum resistance thermometers), requiring only the use of any sensor model that meets the requirements of spatial resolution ≤5cm and response time ≤50ms.

[0031] The fusion layer is used to input the collected data into the edge computing nodes after synchronization by the PLC, construct synchronized data for the CFD-DEM digital twin, and output the prediction residuals. It also expands the effective operating condition data based on a generative adversarial network. Specifically, the fusion layer includes the following modules: The data synchronization module is used to synchronize multi-source heterogeneous data through the field PLC at the first preset cycle. It uses the OPC UA (OPC Unified Architecture) communication protocol to upload a 35-dimensional vector representing the running data, including 32 channels of temperature, NH3 concentration, HNCO concentration, and NOx inlet concentration, and inputs it to the edge computing node. The data fusion module is used to add 8-dimensional acoustic FFT energy spectrum and 2-dimensional PIV velocity field components to a 35-dimensional vector at a second preset period via edge computing nodes. The added vector is packaged and stored in a MySQL time-series database using the MQTT (Message Queuing Telemetry Transport) communication protocol. The MySQL time-series database retains two years of data for historical backtracking, data mining, or model training. The acoustic features are acquired through acoustic sensors deployed on S1.

[0032] This embodiment provides a hardware deployment method for the fusion layer, specifically: The Siemens S7-1516-3 PLC is responsible for high-speed I / O synchronization, updating data with a first preset cycle of 4ms to ensure the time consistency of sensor data, and is used for real-time synchronization of temperature field, flow field, chemical field and atomization quality data. The NVIDIA Jetson AGX Orin boasts 275 TOPS of AI computing power and appends multi-dimensional vectors in a second preset cycle of 50ms for running CFD-DEM models and lightweight AI inference. The Jetson AGX Orin is an AI supercomputer released by NVIDIA that can be used for various forms of edge embedded computing with local AI inference latency of less than 10ms.

[0033] In this embodiment, the above hardware deployment method is only a preferred implementation method and can be adapted according to actual needs.

[0034] The data cleaning module cleans the data by imputing missing values, EMA denoising, and Z-score normalization, and constructs derived features, including the NOx concentration change rate (denoted as ΔNOx / Δt), unit urea consumption, and temperature non-uniformity index. .

[0035] In this embodiment, the missing value imputation can employ linear interpolation or nearest neighbor imputation; the EMA denoising specifically uses the exponential moving average method, which reduces the impact of noise by assigning higher weights to closer data points. In this embodiment, the smoothing factor α is set to 0.2 to make the EMA response slower, resulting in a stronger smoothing effect, effectively denoising and avoiding short-term fluctuations from interfering with long-term trends; the Z-score normalization is used to eliminate the dimensional differences between different features in the imputed and denoised data, improving the stability of the model. Z-score normalization is updated based on a rolling window, and the standard deviation is... and mean Whenever a new set of data arrives, the mean and standard deviation of that set of data are calculated within the rolling window.

[0036] In this embodiment, the constructed derived features include the NOx concentration change rate (denoted as ΔNOx / Δt), unit urea consumption, and temperature non-uniformity index. Derived features are used to transform static data collected by raw sensors into dynamic indicators that can reflect process dynamics, early faults, or optimization potential in real time. This allows the model to capture higher-order nonlinear relationships at a lower dimension, improving prediction accuracy, reducing training sample requirements, and enhancing generalization ability under operating conditions. In this embodiment, calculations are performed in real time at Jetson edge nodes, directly outputting the unit urea consumption and temperature non-uniformity index every 50ms frame.

[0037] Furthermore, this embodiment also defines a temperature uniformity index. The following logic is used to evaluate the temperature uniformity of the pyrolysis furnace cross-section:

[0038] in, Indicates the temperature uniformity index. A value closer to 1 indicates a more uniform temperature across the furnace cross-section. The design goal of this embodiment is to meet this requirement. ; This represents the temperature non-uniformity index, which in this embodiment is used to represent the standard deviation of the furnace cross-sectional temperature. This indicates the average temperature of the furnace cross section.

[0039] In this embodiment, the standard deviation of the 32 thermocouples deployed on the same cross-section of the furnace should not exceed 5% of the average temperature of the same cross-section, which is equivalent to a relative uniformity of ≥95% when the temperature difference is ≤15K, denoted as .when When the temperature is below 0.95, the system is considered to have uneven temperature. In this case, you can choose to reduce urea injection or increase burner power to promote temperature uniformity. Recovery.

[0040] The digital twin construction module is used to build a digital twin based on the CFD-DEM coupled model as a virtual reactor to simulate the reaction process in the pyrolysis furnace, and output the predicted residual by synchronizing physical field data in real time.

[0041] This embodiment uses computational fluid dynamics (CFD) to simulate the flow, heat transfer, and mass transfer processes of fluids (such as flue gas and urea droplets), and combines it with the discrete element method (DEM) to simulate the motion, collision, and deposition behavior of particles (such as crystalline particles). The CFD provides flow field information to the DEM, and the DEM then feeds back the influence of particle motion on the flow field to achieve coupled effects. This simulates the entire process of urea atomization, evaporation, and pyrolysis in a pyrolysis furnace. Then, by synchronizing physical field data in real time, the predicted residual σ is output. Specifically, the physical field data is the data output by the data cleaning module.

[0042] Specifically, the output prediction residual σ is as follows: First, a CFD-DEM coupled model is constructed. Based on the input physical field data, the CFD-DEM coupled model solves the Navier-Stokes equation and the particle momentum equation, and outputs the predicted physical field data with a period of 50ms, including NH3 concentration, NOx outlet concentration, cross-sectional average temperature and crystallization risk index CRI.

[0043] Secondly, the gradient boosting tree (GBDT) is used to fit the nonlinear main trend of the predicted physical field data, and the physical residual Δ is output as the training label, where the physical residual Δ is the difference between the predicted physical field data and the true value.

[0044] Then, Gaussian process regression (GPR) is used to train the model with training labels, outputting the predicted residual σ and a 95% confidence interval. Model bias is corrected based on the predicted residual σ, and the uncertainty of the prediction is quantified. When the predicted residual σ > 0.05, a fine-grid recalculation of the CFD-DEM is triggered, and the updated model version is stored in the model library to ensure the continuous evolution of the digital twin. Figure 3 As shown.

[0045] In this embodiment, the physical residual is calculated based on the prediction results output by the CFD-DEM coupled model and used as a training label for machine learning, replacing the training label of traditional experimental data. This enables high-fidelity model training without experiments. At the same time, the data generated by GBDT fitting the CFD-DEM output results can feed back into the CFD-DEM, and the physical residual between the CFD-DEM output results and the true values ​​can train GBDT, enabling the digital twin to have the ability to self-perceive, self-train, self-evaluate, and self-evolve.

[0046] The load case generation module is used to generate virtual load case data using WGAN-GP and to verify the load case data using a CFD-DEM coupled model and Wasserstein-1 distance; for example... Figure 2 As shown, specifically, it includes the following: The generator unit is used to build the GAN generator, which fakes working condition data to increase the amount of training data.

[0047] The verification unit is used to perform preliminary verification of the generated operating condition data. Specifically, the generated operating condition data is input into the CFD-DEM coupled model. If the model output error is ≤5%, the generated operating condition data is considered to have passed the preliminary verification. If the model output error is >5%, the generated operating condition data is considered to have failed the preliminary verification, and the process returns to the generator unit.

[0048] The discrimination unit is used to verify the consistency between the generated working condition data and the actual data distribution based on the Wasserstein-1 distance. If the verification condition of Wasserstein-1 distance is met, the unit will determine the consistency. If so, the generated operating condition data is considered valid data.

[0049] In this embodiment, WGAN-GP (Wasserstein GAN with Gradient Penalty) is a form of Generative Adversarial Network (GAN). It introduces the Wasserstein-1 distribution distance metric to measure the shortest path between the real data distribution and the synthetic data distribution. This embodiment verifies the consistency between the generated working condition data and the real data distribution based on the Wasserstein-1 distance, using the following logical representation:

[0050] in, express and Wasserstein-1 distance, Represents the true data distribution. Indicates the distribution of synthetic data. Indicates all and For the joint distribution set of the edges, Indicates the joint distribution. Represents random data pairs. This represents the sum of all data pairs sampled from the joint distribution γ. ,calculate The average value of (x,y) γ represents random data pairs Follows a joint distribution γ, express The L1 vector distance between them.

[0051] The verification condition designed in this embodiment is when... At that time, the operating condition data generated by the GAN generator was considered to be valid data.

[0052] Furthermore, to prevent the generator from "lazily" copying the original data, this embodiment introduces a fingerprint regularization term based on WGAN-GP, and uses the following logic to represent the total loss function of GAN:

[0053] in, This represents the total loss function of GAN. Indicating resistance to loss, Indicates the fingerprint regularization weight. Represents the fingerprint feature vector. This represents the output of the GAN generator. This represents a random noise vector.

[0054] This embodiment introduces the WGAN-GP generative small-sample reinforcement learning framework to address the scarcity of operating condition data in actual industrial production. First, virtual operating condition data is generated by the GAN generator, expanding the original less than 1,000 valid operating condition data to more than 10 times the amount of training data. Then, the validity of the generated operating condition data is initially verified based on the CFD-DEM coupling model. Next, the consistency between the distribution of the generated operating condition data and the real data is verified based on the Wasserstein-1 distance. The quality of the data distribution output by the generator is improved through multiple game iterations, and the verified valid operating condition data is retained in the data pool. This allows for continuous iteration and rapid adaptation in coal-fired units, gas-fired units, and cement kiln tail units.

[0055] The decision-making layer is used to continuously calculate the optimal control command based on online optimization of urea injection quantity and burner power, and switch the control mode of the decision-making layer according to abnormal conditions. For example... Figure 4 As shown, specifically, the decision-making layer includes the following: The optimization layer aims to achieve the highest denitrification efficiency, lowest ammonia slip, lowest energy consumption, and lowest crystallization risk. Based on a multi-objective Bayesian optimizer and using EHVI as the acquisition function, it searches for the Pareto front within the equivalent urea injection rate range of 3.0-8.0 pH. Finally, it optimizes the urea injection rate and burner power online using the PPO reinforcement learning algorithm.

[0056] In this embodiment, multi-objective Bayesian optimization is used to construct the initial Pareto front offline, providing an initial reference point for PPO with a minute-level response; the PPO reinforcement learning algorithm is used for online policy optimization, adjusting the urea injection quantity and burner power according to the real-time status with a second-level response to adapt to transient conditions such as load fluctuations and NOx mutations.

[0057] In this embodiment, the multi-objective Bayesian optimization process is represented by the following logic: I. Generate initial design points using Latin hypercube sampling; II. Construct Gaussian process proxy models for each objective; III. Calculate the EHVI value of each candidate point and select the point with the largest EHVI for simulation. IV. Update the Pareto frontier and surrogate model, and iterate until convergence; V. Output the Pareto optimal solution set, which is used to initialize the PPO policy network and reward function weights.

[0058] In this embodiment, the PPO reinforcement learning algorithm optimizes the urea injection rate and burner power online by designing a reward function, using the following logical representation:

[0059] in, This represents the instantaneous reward value at time t. This represents the weighting coefficient for denitrification efficiency. Indicates denitrification efficiency. This represents the ammonia slip weighting coefficient. The ammonia escape concentration at the outlet is measured in real time by the TDLAS laser ammonia meter at the outlet, read into the control cycle by the PLC, and directly used as one of the input items of the PPO reward function, in ppm. This represents the energy consumption weighting coefficient. This indicates the total power of the system, in kW. This represents the crystallization risk weighting coefficient. Indicates the crystallization risk index. The value ranges from 0 to 1, and to quantify the crystallization tendency 30 seconds in advance, the crystallization risk index is represented by the following logic. :

[0060] in, This indicates the concentration of isocyanate (HNCO concentration) at the furnace outlet at time t, in ppm. This indicates the upper limit of the permissible concentration of isocyanate; in this example, it can be taken as 50 ppm. This represents the lowest furnace wall temperature at time t, in K. This represents the isocyanate dew point temperature, taken as 588K; This represents the Heaviside step function, which takes the value 1 if the condition is met, and 0 otherwise.

[0061] The control layer calculates the optimal control command based on the constraints of Model Predictive Control (MPC) and updates the A, B, and C matrices of the internal prediction model of MPC using the recursive least squares (RLS) method, outputting a 40-dimensional control vector, including 32 spray gun flow rates and 8 burner power.

[0062] In this embodiment, to enable the model to track changes in the pyrolysis furnace operating conditions as quickly as possible, RLS employs an exponential forgetting weighted objective function. This can be represented using the following logic:

[0063] in, Represents the forgetting factor, taking , express Time-regression vector, This represents the vector of parameters to be estimated, formed by stacking the elements of matrices A, B, and C, which are updated by the recursive least squares method.

[0064] Furthermore, when the prediction residual σ > 0.05 or the system load undergoes a step change, the covariance matrix of the RLS algorithm will be forcibly reset, so that it can quickly discard the influence of old data, prevent old data from continuing to dominate, quickly track new operating conditions, ensure rapid convergence after sudden changes in operating conditions, and improve the adaptability of operating conditions.

[0065] In this embodiment, MPC determines the optimal control sequence by minimizing future errors and control increments. The rolling calculation of the optimal control command utilizes the following logical representation:

[0066] in, Indicates the future starting from time k. Step-optimal control sequence This indicates the prediction time-domain step size; in this embodiment, the time-domain step size is 1 second. Take 30; This represents the model's predicted output vector for the i-th step. This represents the reference trajectory vector for the i-th future step, which is a pre-set value; This represents the output error weight matrix. This represents the control increment vector. This represents the control increment weight matrix. This represents the squared error between the predicted output and the target value, weighted by matrix Q. This represents the squared cost of the control increment vector weighted by matrix E.

[0067] To prevent control actions from exceeding the allowable range of the process, the rolling optimization of MPC must simultaneously satisfy the MPC constraints, as represented by the following logic:

[0068] in, This indicates the predicted ammonia concentration at the outlet in step k, in ppm. This represents the standard deviation of temperature at step k. This represents the temperature at the i-th measurement point in the k-th step. This represents the average temperature at step k. Indicates the number of measurement points. This indicates the HNCO concentration at step k, in ppm.

[0069] Furthermore, this embodiment sets the following constraints on the 40-dimensional control vector: The flow rates of all 32 spray guns are between 0.1 and 5 L / min; The power of each of the eight burners is between 10 and 50 kW; ammonia escape from export .

[0070] Safety layer, when the predicted residual σ > 0.05 or the crystallization risk index When the pH value is greater than 0.8, an anomaly is considered to exist. The system immediately switches to a conservative PID control mode with a fixed pH of 7.0, and the spray gun flow rate is set to 80% of the flow rate corresponding to the previous response cycle. The optimized layer control will automatically resume after the anomaly is resolved.

[0071] Furthermore, if the predicted residual σ < 0.03 and no anomaly is observed after two response cycles, the anomaly is considered to have been contacted.

[0072] In this embodiment, the decision layer is deployed on edge computing nodes, the optimization cycle of the optimization layer is set to 1-5 minutes, the response cycle of the control layer is set to 1-10 seconds, the security layer is triggered in real time, the optimization layer which takes a long time is able to provide the target setpoint to the control layer, and the control layer quickly responds and executes the data and then feeds it back to the optimization layer, which effectively improves the robustness of the decision layer.

[0073] In this embodiment, the optimization layer aims to achieve the highest denitrification efficiency, lowest ammonia slip, lowest energy consumption, and lowest crystallization risk. Based on the PPO reinforcement learning algorithm, the system optimizes the urea injection rate and burner power online by designing a reward function to balance the above indicators, enabling the system to continuously approach the optimal operating point. At the control layer, the optimal control command is calculated based on MPC constraints, predicting the trend within the next 30 seconds. When model mismatch, crystallization risk, or severe blockage occurs, the safety layer switches the control mode of the decision layer to the traditional PID control mode. After the anomaly is resolved, it switches back to the optimization layer control, greatly reducing the risk of abnormal situations, reducing the degree of manual intervention, ensuring the availability of the equipment and long-term safe and stable operation, and reducing the economic losses caused by unplanned downtime.

[0074] The execution layer receives control commands, adjusts the spray gun and burner based on the control commands and dynamic zoning strategies, and actively intervenes to remove crystals.

[0075] In this embodiment, the execution layer first receives a control vector from the decision layer containing the spray gun flow rate and burner power. Based on the control commands, the execution layer adjusts the spray gun flow rate and burner power. Specifically, the dynamic zoning strategy divides the pyrolysis furnace along its height into multiple independent control zones, each associated with a set of spray guns. When the temperature sensor in a certain independent control zone detects a temperature below 320°C, the burner power of that independent control zone is increased and the spray gun flow rate is reduced to prevent unreacted urea crystallization. When the decision layer predicts that the NOx concentration will increase by 20 mg / Nm³ within the next 10 seconds, the urea flow rate in the corresponding zone is increased in advance. Simultaneously, crystallization is removed in real time using an acoustic soot remover.

[0076] This embodiment provides a hardware deployment method for the execution layer, specifically: configuring 32 independent servo spray guns and 8 proportional burners, wherein the spray gun flow rate range is 0.1~5 L / min, and the burner power is 10~50kW; the spray guns are inserted downwards at 15°, with the gun head 450mm from the center of the furnace; at the same time, the sonic soot remover has a power of 500W and a frequency of 20kHz pulsed sonic wave to remove the initial crystallization.

[0077] In this embodiment, the above hardware deployment method is only a preferred implementation method and can be adapted according to actual needs.

[0078] The energy layer is used to recover waste heat from flue gas and preheat urea solution through the ORC unit, thus achieving an energy-saving closed loop.

[0079] In this embodiment, an Organic Rankine Cycle (ORC) waste heat power generation system is used to recover the high-temperature flue gas from the pyrolysis furnace outlet. The ORC unit uses R245fa as the working fluid, with an evaporation pressure of 2.8 MPa, a condensation pressure of 0.45 MPa, and a power generation capacity of 260 kW. It absorbs heat and evaporates it into steam to drive a generator to generate electricity. The recovered electrical energy is used to preheat the urea solution to 80°C. In this embodiment, the above process parameters are set for the specific operation and can be selected according to actual needs, not limited to the scope of this embodiment. The flue gas after heat transfer flows into the subsequent dust removal process. The following logic represents the thermal efficiency of the ORC unit in converting flue gas waste heat into electrical energy:

[0080] in, Indicates ORC thermal efficiency. This represents the net power generation of ORC, which is taken as 260kW in this embodiment. This represents the ORC input thermal power.

[0081] like Figure 5 The diagram shown is a thermodynamic pressure-specific enthalpy (Ph) diagram of the ORC waste heat recovery system. It illustrates how the energy layer uses the ORC unit to generate 260kW of electricity from 400–450℃ flue gas and drive the urea solution to preheat from 25℃ to 80℃. The ORC unit does not connect the electricity to the grid but uses it directly within the system, forming an inherent "electricity-based ammonia" energy-saving closed loop. This not only achieves waste heat recovery for power generation but also directly applies the recovered electrical energy to the urea preheating process, greatly reducing net energy consumption and system power consumption.

[0082] Furthermore, calculations show that utilizing ORC to recover electrical energy and preheat the urea solution can save power:

[0083] in, Indicates annual electricity savings, in kW; This indicates the mass flow rate of the urea solution. This indicates the specific heat capacity of the urea solution; in this example, the specific heat capacity of a 50wt% solution is 3.3. This indicates the preheating temperature difference, which is 25~80℃ in this embodiment; This represents the electric heating efficiency, taken as 0.95.

[0084] Furthermore, the annual CO2 emission reduction from ORC waste heat power generation is calculated:

[0085] in, This represents the annual reduction in carbon dioxide emissions, in units of... ; This indicates the number of operating hours per year, in hours (h). This represents the power grid emission factor. In summary, the intelligent control system for urea pyrolysis provided in this embodiment achieves an energy-saving closed loop by recovering waste heat from flue gas and preheating the urea solution through an energy layer, significantly reducing plant power consumption and reducing carbon dioxide emissions by thousands of tons annually.

[0086] This invention provides a urea pyrolysis intelligent control system that can be directly deployed in 30MW~1000MW coal-fired units, 350MW gas-fired combined cycle units, or cement kiln tail gas purification systems, as shown in Table 1 below. Figure 6 The diagram shows the actual benefits of applying the present invention to coal-fired power units, gas-fired power units, and cement kiln tail gas purification systems.

[0087] Table 1. Actual benefits of the control system in different units

[0088] Under a wide operating range with flue gas temperature of 300~450℃, NOx inlet concentration of 100–800mg / Nm³, SO2 inlet concentration not exceeding 5000mg / Nm³, and boiler load change rate ≤5% / min, the control system can maintain ammonia slip ≤3ppm (strictly lower than the national standard GB13223 limit of 5ppm), urea utilization rate ≥98%, and system power consumption reduced by 35%, while simultaneously recovering 260kW of clean electricity through ORC waste heat power generation.

[0089] Furthermore, the control system also includes a visualization layer for building a real-time monitoring panel based on Vue3 and ECharts 8.0. The real-time monitoring panel is used to display trend graphs of multiple key variables, including NH3 concentration, NOx concentration, energy consumption, ORC power, and crystallization risk index.

[0090] The real-time monitoring panel also includes a draggable timeline for replaying past data.

[0091] In this embodiment, a real-time monitoring panel is built based on the Vue3 front-end framework and the ECharts 8.0 visualization chart library technology. The real-time monitoring panel collects and displays real-time data from PLC and edge nodes through OPC UA and MQTT protocols, and stores long-term data through a MySQL time-series database for data tracing and playback. In this embodiment, all data in the past 72 hours can be played back by dragging the timeline.

[0092] Furthermore, alarm thresholds can be set for each key variable according to the actual needs of the project. When the relevant variable in the system exceeds the alarm threshold, an alarm is triggered, and the alarm information is pushed to the relevant technical personnel in real time using the API of mainstream office software (such as WeChat and DingTalk).

[0093] This invention also provides a method for intelligent control of urea pyrolysis, comprising the following steps: S1, deploy a multimodal sensor network in the pyrolysis furnace to collect multi-source heterogeneous data in real time; S2, after the collected data is synchronized by the PLC, it is input into the edge computing node to construct the CFD-DEM digital twin synchronization data and output the prediction residual, and expand the effective working condition data based on the adversarial generative network; S3, based on online optimization of urea injection quantity and burner power, continuously calculates the optimal control command of the decision layer, and switches the control mode of the decision layer according to abnormal conditions; S4 receives control commands, adjusts the spray gun and burner based on the control commands and dynamic zoning strategy, and actively intervenes to remove crystals; S5 recovers waste heat from flue gas and preheats urea solution through the ORC unit, achieving a closed-loop energy-saving system.

[0094] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A smart control system for urea pyrolysis, characterized in that, include: The sensing layer is used to deploy a multimodal sensor network in the pyrolysis furnace to collect multi-source heterogeneous data in real time. The fusion layer is used to input the collected data into the edge computing node after synchronization by the PLC, construct the CFD-DEM digital twin synchronization data and output the prediction residual, and expand the effective working condition data based on the adversarial generative network. The decision-making layer is used to continuously calculate the optimal control command based on the online optimization of urea injection quantity and burner power, and switch the control mode of the decision-making layer according to abnormal conditions. The execution layer is used to receive control commands, adjust the spray gun and burner based on the control commands and dynamic zoning strategies, and actively intervene to remove crystals; The energy layer is used to recover waste heat from flue gas and preheat urea solution through the ORC unit, thus achieving an energy-saving closed loop.

2. The intelligent control system for urea pyrolysis according to claim 1, characterized in that, The fusion layer includes: The data synchronization module is used to synchronize multi-source heterogeneous data through the field PLC at a first preset cycle, and uploads a multi-dimensional vector representing the running data using the OPC UA communication protocol, which is then input into the edge computing node. The data fusion module is used to add acoustic energy spectrum and PIV velocity field components to a multidimensional vector at a second preset period through edge computing nodes; the added vector is then packaged and stored in a time-series database. The data cleaning module is used to clean data by imputing missing values, EMA denoising, and Z-score normalization, and to construct derived features; The digital twin construction module is used to build a digital twin based on the CFD-DEM coupled model as a virtual reactor to simulate the reaction process in the pyrolysis furnace, and output the predicted residual by synchronizing physical field data in real time. The working condition generation module is used to generate virtual working condition data using WGAN-GP and to verify the working condition data using a CFD-DEM coupled model and Wasserstein-1 distance.

3. The intelligent control system for urea pyrolysis according to claim 2, characterized in that, The output prediction residual σ in the digital twin construction module is specifically: First, a CFD-DEM coupled model is constructed. Based on the input physical field data, the CFD-DEM coupled model solves the Navier-Stokes equation and the particle momentum equation, and outputs the predicted physical field data. Secondly, a gradient boosting tree is used to fit the nonlinear main trend of the predicted physical field data, and the physical residual Δ is output as the training label. The physical residual Δ is the difference between the predicted physical field data and the true value. Then, Gaussian process regression is used to train the model with training labels, outputting the predicted residual σ and the 95% confidence interval. The model bias is corrected based on the predicted residual σ, and the uncertainty of the prediction is quantified. When the predicted residual is greater than the preset residual threshold, the CFD-DEM fine mesh recalculation is triggered, and the updated model version is stored in the model library.

4. The intelligent control system for urea pyrolysis according to claim 2, characterized in that, The working condition generation module includes: The generator unit is used to build the GAN generator, fake working condition data, and increase the amount of training data. The verification unit is used to perform preliminary verification of the generated operating condition data. Specifically, the generated operating condition data is input into the CFD-DEM coupled model. If the model output error is less than the error threshold, the generated operating condition data is considered to have passed the preliminary verification. If the model output error is greater than the error threshold, the generated operating condition data is considered to have failed the preliminary verification, and the process is returned to the generator unit. The discrimination unit is used to verify the consistency between the generated working condition data and the actual data distribution based on the Wasserstein-1 distance. If the verification condition of Wasserstein-1 distance is met, the unit will determine the consistency. When the distance threshold is reached, the generated working condition data is considered valid data; the Wasserstein-1 distance is represented using the following logic: in, express and Wasserstein-1 distance, Represents the true data distribution. Indicates the distribution of synthetic data. Indicates all and For the joint distribution set of the edges, Indicates the joint distribution. Represents random data pairs. This represents the sum of all data pairs sampled from the joint distribution γ. ,calculate The average value of (x,y) γ represents random data pairs Follows a joint distribution γ, express L1 vector distance between them; The discrimination unit uses the following logic to represent the total loss function of the GAN: in, This represents the total loss function of GAN. Indicating resistance to loss, Indicates the fingerprint regularization weight. Represents the fingerprint feature vector. This represents the output of the GAN generator. This represents a random noise vector.

5. The intelligent control system for urea pyrolysis according to claim 1, characterized in that, The decision-making body includes: The optimization layer aims to achieve the highest denitrification efficiency, lowest ammonia slip, lowest energy consumption, and lowest crystallization risk. Based on a multi-objective Bayesian optimizer and using EHVI as the acquisition function, it searches the Pareto front within the equivalent urea injection rate range of 3.0-8.0 pH. The urea injection rate and burner power are optimized online based on the PPO reinforcement learning algorithm. The control layer calculates the optimal control command based on rolling calculation under MPC constraints, updates the A, B, and C matrices of the MPC internal prediction model using the recursive least squares method, and outputs a multi-dimensional control vector. Safety layer, when the prediction uncertainty is greater than the uncertainty threshold or the crystallization risk index. When the value exceeds the CRI threshold, an anomaly is considered to exist. The system immediately switches to a conservative PID control mode with a fixed pH of 7.

0. Once the anomaly is resolved, the system automatically resumes optimized layer control.

6. The intelligent control system for urea pyrolysis according to claim 5, characterized in that, The PPO reinforcement learning algorithm in the optimization layer optimizes the urea injection rate and burner power online by designing a reward function, as shown in the following logical representation: in, This represents the instantaneous reward value at time t. This represents the weighting coefficient for denitrification efficiency. Indicates denitrification efficiency. This represents the ammonia slip weighting coefficient. This indicates the concentration of ammonia slip at the outlet, in ppm. This represents the energy consumption weighting coefficient. This indicates the total power of the system, in kW. This represents the crystallization risk weighting coefficient. The crystallization risk index is represented by the following logic to quantify the tendency to crystallize a certain time in advance. : in, This indicates the HNCO concentration at the furnace outlet at time t, in ppm. This indicates the upper limit of the permissible concentration of HNCO; This represents the lowest furnace wall temperature at time t, in K. Indicates the HNCO dew point temperature. This represents the Heaviside step function, which takes the value 1 if the condition is met, and 0 otherwise.

7. The intelligent control system for urea pyrolysis according to claim 5, characterized in that, The recursive least squares method in the control layer employs an exponential forgetting weighted objective function. This can be represented using the following logic: in, Indicates the forgetting factor, express Time-regression vector, The vector of parameters to be estimated is formed by stacking the elements of matrices A, B, and C, which are updated by recursive least squares; The following logic represents the rolling calculation of the optimal control command: in, Indicates the future starting from time k. Step-optimal control sequence Indicates the number of prediction time-domain steps. This represents the model's predicted output vector for the i-th step. This represents the reference trajectory vector for the i-th future step. This represents the output error weight matrix. This represents the control increment vector. This represents the control increment weight matrix.

8. The intelligent control system for urea pyrolysis according to claim 1, characterized in that, Specifically, the recovery of waste heat from flue gas and preheating of urea solution in the energy layer via ORC units are as follows: The ORC unit recovers the high-temperature flue gas from the outlet of the pyrolysis furnace. The ORC unit absorbs heat and evaporates it into steam to drive the generator to generate electricity. The recovered electricity is used to drive the preheating of the urea solution. After the heat transfer is completed, the flue gas flows into the subsequent dust removal process.

9. The intelligent control system for urea pyrolysis according to claim 1, characterized in that, It also includes a visualization layer for building a real-time monitoring panel, which displays trend graphs of multiple key variables, including NH3 concentration, NOx concentration, energy consumption, ORC power, and crystallization risk index.

10. A method for intelligent control of urea pyrolysis, characterized in that, Includes the following steps: S1, deploy a multimodal sensor network in the pyrolysis furnace to collect multi-source heterogeneous data in real time; S2, after the collected data is synchronized by the PLC, it is input into the edge computing node to construct the CFD-DEM digital twin synchronization data and output the prediction residual, and expand the effective working condition data based on the adversarial generative network; S3, based on online optimization of urea injection quantity and burner power, continuously calculates the optimal control command of the decision layer, and switches the control mode of the decision layer according to abnormal conditions; S4 receives control commands, adjusts the spray gun and burner based on the control commands and dynamic zoning strategy, and actively intervenes to remove crystals; S5 recovers waste heat from flue gas and preheats urea solution through the ORC unit, achieving a closed-loop energy-saving system.

Images