Multi-objective flow field collaborative optimization design method and system for scr denitration system

Abstract

This application relates to a multi-objective flow field collaborative optimization design method and system for SCR denitrification systems, belonging to the technical field of SCR denitrification. The method includes: constructing a CFD model of the SCR denitrification reactor based on current physical data; simulating flue gas flow in the CFD model and obtaining initial flow field simulation results; extracting flue gas velocity distribution cloud maps, flue gas ammonia-nitrogen molar ratio distribution cloud maps, flue gas temperature distribution cloud maps, and flue gas incident angle distribution cloud maps at 500 mm upstream of the first catalyst inlet from the initial flow field simulation results; calculating the relative deviations of the corresponding extracted features based on each extracted distribution cloud map, and determining whether all relative deviations meet the technical objectives; if not, modifying the structure of the virtual SCR denitrification reactor in the CFD model, driving the modified CFD model to re-simulate flue gas flow until the relative deviations of all extracted features meet the technical objectives. This application has the beneficial effect of improving denitrification efficiency.

Description

Technical Field

[0001] This application relates to the technical field of SCR denitrification, and in particular to a multi-objective flow field collaborative optimization design method and system for SCR denitrification systems. Background Technology

[0002] In many industrial production processes, boiler combustion in industries such as thermal power generation and steel smelting generates large amounts of nitrogen oxides. Selective catalytic reduction (SCR) denitrification technology, as a highly efficient method for nitrogen oxide removal, has been widely used in the industrial field. This technology reduces nitrogen oxide emissions by using a reducing agent (such as ammonia) to reduce nitrogen oxides to nitrogen and water under the action of a catalyst.

[0003] In traditional SCR denitrification system design, a combination of empirical design and simple physical models is typically used. Engineers determine the approximate size and structure of the reactor based on past project experience. For flow field design, simple baffles are usually installed inside the reactor to guide the flue gas to flow uniformly. When determining the reductant injection location and method, fixed injection points and angles are often selected based on theoretical calculations and limited experimental data to ensure mixing of the reductant with the flue gas.

[0004] However, empirical design and simple physical models cannot accurately reflect the actual complex flow field, resulting in uneven flow of flue gas in the reactor, which causes some catalysts to fail to play their full role and reduces the denitrification efficiency. Summary of the Invention

[0005] To improve denitrification efficiency, this application provides a multi-objective flow field collaborative optimization design method and system for SCR denitrification systems.

[0006] Firstly, this application provides a multi-objective flow field collaborative optimization design method for SCR denitrification systems, employing the following technical solution: A multi-objective flow field collaborative optimization design method for SCR denitrification systems includes: Based on the current physical data of the SCR denitrification reactor, a CFD model of the SCR denitrification reactor is constructed. The flow of flue gas was simulated in the CFD model, and the initial flow field simulation results were obtained. From the initial flow field simulation results, the flue gas velocity distribution cloud map, flue gas ammonia-nitrogen molar ratio distribution cloud map, flue gas temperature distribution cloud map, and flue gas incident angle distribution cloud map were extracted at 500 mm upstream of the first catalyst inlet. Based on the extracted distribution cloud maps, calculate the relative deviation of the corresponding extracted features, and determine whether all relative deviations meet the technical objectives. If not, the structure of the virtual SCR denitrification reactor in the CFD model is modified, and the modified CFD model is driven to re-simulate the flow of flue gas until the relative deviations of all extracted features meet the technical objectives.

[0007] By adopting the above technical solution and constructing a CFD model based on the physical data of the SCR denitrification reactor, the actual reactor environment can be highly reproduced. The model simulates flue gas flow and obtains initial flow field simulation results. From these results, distribution cloud maps of flue gas velocity, ammonia-nitrogen molar ratio, temperature, and incident angle at 500mm upstream of the first catalyst inlet are extracted. These cloud maps visually present various characteristics of the flow field, providing a comprehensive and detailed basis for evaluating the flow field conditions. Based on these distribution cloud maps, the relative deviations of corresponding characteristics are calculated and compared with the technical targets to accurately determine whether the current flow field meets the requirements. If not, the structure of the virtual SCR denitrification reactor in the CFD model is modified to effectively improve the flow field conditions. Through continuous modification and re-simulation, until the relative deviations of all characteristics meet the technical targets, the performance of the SCR denitrification system is improved. A uniform flow field allows for sufficient contact between the flue gas and the catalyst, enabling a more complete reaction between ammonia and nitrogen oxides, thereby improving the removal efficiency of nitrogen oxides and making emissions more compliant with environmental standards. The optimized flow field ensures uniform distribution of ammonia in the flue gas, reducing the escape of unreacted ammonia and lowering the risk of corrosion and blockage in downstream equipment caused by ammonia escape, thus guaranteeing long-term stable operation of the system. Uniform velocity, temperature, and incident angle distribution reduce wear and aging caused by localized high-speed scouring, ash accumulation and blockage, or uneven temperature, extending catalyst lifespan and reducing catalyst replacement costs.

[0008] Optionally, the multi-objective flow field collaborative optimization design method further includes: After all the relative deviations of the extracted features meet the technical objectives, the pressure distribution cloud map of the SCR system is extracted from the initial flow field simulation results; Calculate the pressure loss of the SCR system based on the pressure distribution cloud map of the SCR system, and determine whether the pressure loss meets the design requirements. If not, the structure of the virtual SCR denitrification reactor in the CFD model is modified, and the modified CFD model is driven to re-simulate the flow of flue gas until the relative deviations of all characteristics meet the technical objectives and the pressure loss meets the design requirements.

[0009] By adopting the above technical solution, after optimizing the flow field characteristics, the pressure distribution cloud map of the SCR system is extracted and the pressure loss is calculated, enabling a more comprehensive evaluation of the performance of the entire SCR denitrification system. Pressure loss is a key indicator for measuring the system's operating efficiency and economy, directly related to the system's energy consumption and operating costs. When it is determined that the pressure loss does not meet the design requirements, the reactor structure is continuously adjusted, and the flue gas flow is re-simulated until the pressure loss meets the design requirements. At the same time, it is ensured that the relative deviation of the flow field characteristics still meets the technical objectives. This process achieves multi-objective synergistic optimization. A suitable pressure loss can reduce the resistance of flue gas flow within the system and lower operating costs.

[0010] Optionally, the step of modifying the structure of the virtual SCR denitrification reactor in the CFD model includes: Extract the core parameters of the current substandard flow field features and generate a structured feature vector; Select structural modification schemes that match the structured feature vector from the pre-built SCR flow field modification case library; If a match is found, the relevant parameters of the matched structural modification scheme are loaded into the CFD model; If no match is found, the built-in SCR flow field control expert rule engine is invoked to trigger the corresponding structural adjustment logic based on the type of non-compliant characteristics of the current flow field. The structural adjustment logic includes adding structures and adjusting the parameters of the original structures. Load the relevant parameters of the triggered structural adjustment logic into the CFD model.

[0011] By adopting the above technical solution, structural modification schemes matching the structured feature vectors are selected from a pre-built SCR flow field modification case library, making full use of past experience and data. The case library contains a large number of successful flow field modification cases; through matching and filtering, suitable modification schemes for the current problem can be quickly found. If a match is found, the relevant parameters of the matched structural modification scheme are directly loaded into the CFD model, greatly shortening the problem-solving time and improving efficiency. When no suitable scheme is found in the case library, the built-in SCR flow field control expert rule engine is invoked, triggering the corresponding structural adjustment logic based on the type of non-compliant flow field characteristics. The expert rule engine incorporates professional knowledge and experience in the field, providing scientific and reasonable solutions for different types of flow field problems.

[0012] Optionally, the multi-objective flow field collaborative optimization design method further includes: When multiple structural modification schemes are matched or multiple structural adjustment logics are triggered, a weighted evaluation function is established based on the flow field uniformity index, pressure design requirements, and economic objectives. Multiple structural modification schemes or multiple structural adjustment logics triggered by the input are fed into a pre-trained neural network model, which outputs the corresponding flow field uniformity index mapping value, pressure mapping value, and economic mapping value. The flow field uniformity index mapping value, pressure mapping value, and economic mapping value are input into the weighted evaluation function, and the corresponding function value is output. The structural modification scheme with the smallest function value is taken as the optimal structural modification scheme, or the structural adjustment logic with the smallest function value is taken as the optimal structural adjustment logic.

[0013] By adopting the above technical solution, the weighted evaluation function comprehensively considers multiple key factors in the operation of the SCR denitrification system, ensuring that the selected scheme achieves a balance in several important aspects. The mapping values ​​output by the neural network are input into the weighted evaluation function, which outputs the corresponding function value. The structural modification scheme with the smallest function value is selected as the optimal structural modification scheme, or the structural adjustment logic with the smallest function value is selected as the optimal structural adjustment logic. This allows for the selection of the scheme with the best overall performance from numerous options. It not only ensures more uniform flue gas flow within the SCR denitrification system, improves catalyst utilization efficiency and denitrification effect, and reduces ammonia slip, but also ensures that the system pressure meets design requirements, reducing energy consumption and equipment wear. Simultaneously, considering economic objectives effectively controls modification and operating costs, improving the overall economic efficiency and competitiveness of the system.

[0014] Optionally, inputting multiple structural modification schemes or triggered multiple structural adjustment logics into the pre-trained neural network model includes the following steps: Analyze whether there are complementary dimensions between multiple structural modification schemes or multiple structural adjustment logics triggered by them; If not, then input multiple structural modification schemes or triggered multiple structural adjustment logics into the pre-trained neural network model; If so, identify the optimal combination of parameters for each dimension of flow field uniformity, pressure loss, and economy in the structural modification scheme or structural adjustment logic; By integrating the optimal combination of individual parameters and combining them with the boundary conditions of the CFD model, a hybrid structural modification scheme is generated, and the relevant parameters of the hybrid structural modification scheme are loaded into the CFD model.

[0015] By employing the aforementioned technical solutions and analyzing the complementary dimensions between schemes or logics, we can deeply explore the intrinsic connections between different schemes or logics. In the optimization of SCR denitrification systems, different structural modification schemes or structural adjustment logics may have unique advantages in different aspects. By analyzing complementary dimensions, we can determine whether these schemes or logics can cooperate to exert a more powerful effect than a single scheme. When complementary dimensions exist, identifying the optimal parameter combinations for each structural modification scheme or structural adjustment logic in the dimensions of flow field uniformity, pressure loss, and economy can fully leverage the advantages of each scheme or logic. Integrating these advantages and combining them with the boundary conditions of the CFD model to generate a hybrid structural modification scheme ensures that the scheme not only integrates the advantages of each scheme but also adapts to the actual system environment.

[0016] Optionally, the multi-objective flow field collaborative optimization design method further includes: The lifetime distribution cloud map of the catalyst layer is extracted from the final optimized flow field simulation results; The local velocity extremes, fly ash concentration gradients, and temperature fluctuation standard deviations at key locations in the catalyst layer are calculated based on the lifetime distribution cloud map. The catalyst's physical parameters and chemical composition are input into the constructed deactivation mechanics model to calculate the combined deactivation rate due to alkali metal poisoning, arsenic poisoning, and thermal sintering. Based on local velocity extremes, fly ash concentration gradients, temperature fluctuation standard deviations, and overall deactivation rates, the coordinates of the minimum lifetime warning region are predicted.

[0017] By employing the above technical solution, a lifetime distribution cloud map of the catalyst layer is extracted from the final optimized flow field simulation results. This cloud map clearly shows the lifetime differences at different locations on the catalyst. Based on the lifetime distribution cloud map, the local velocity extrema, fly ash concentration gradient, and temperature fluctuation standard deviation at key locations in the catalyst layer are calculated. These parameters are crucial factors affecting catalyst lifetime. Excessive local velocity extrema may lead to excessive erosion and wear of the catalyst; the fly ash concentration gradient reflects the distribution of fly ash in the catalyst layer, and excessively high fly ash concentration may cause catalyst blockage; the temperature fluctuation standard deviation reflects temperature stability, and excessive temperature fluctuation may accelerate catalyst aging. By calculating the comprehensive deactivation rate, the rate at which the catalyst deactivates due to various reasons can be quantified, thus more accurately assessing the catalyst lifetime. Then, based on the local velocity extrema, fly ash concentration gradient, temperature fluctuation standard deviation, and comprehensive deactivation rate, the coordinates of the minimum lifetime warning region are predicted, allowing for early identification of the areas where the catalyst is most prone to failure.

[0018] Optionally, the steps following the prediction of the minimum lifespan region warning coordinates include: If the predicted life value corresponding to the minimum life region warning coordinates is less than the life design value, then the minimum life region warning coordinates and the associated local flow field characteristics are back-mapped into the CFD model as new constraint boundary conditions. The influence weights of each structural parameter of the SCR denitrification reactor on the local deactivation rate of the catalyst were calculated, and the key geometric factors affecting the lifetime distribution were identified. The internal structural layout of the virtual SCR denitrification reactor is automatically adjusted based on the key geometric factors driven by the CFD model.

[0019] By adopting the above technical solution, when the predicted lifetime value corresponding to the warning coordinates of the minimum lifetime region is less than the designed lifetime value, the warning coordinates of the minimum lifetime region and the associated local flow field characteristics are back-mapped into the CFD model as new constraint boundary conditions. This allows the CFD model to more accurately simulate the flow field conditions in the region with insufficient catalyst lifetime during actual operation. Different structural parameters have different degrees of influence on the catalyst deactivation rate. By quantifying the influence weights, it is possible to identify which structural parameters are the key factors affecting the catalyst lifetime distribution. Then, based on the key factors, the CFD model automatically adjusts the internal structural layout of the virtual SCR denitrification reactor. Under the premise of meeting the original flow field uniformity, pressure loss, and economic objectives, it prioritizes suppressing the unsteady flow behavior that leads to early catalyst failure.

[0020] Secondly, this application provides a multi-objective flow field collaborative optimization design system for SCR denitrification systems, employing the following technical solution: A multi-objective flow field collaborative optimization design system for SCR denitrification systems includes: The model building and simulation module is used to build a CFD model of the SCR denitrification reactor based on the current physical data of the SCR denitrification reactor, simulate the flow of flue gas in the CFD model, and obtain the initial flow field simulation results. The feature extraction and processing module is used to extract the flue gas velocity distribution cloud map, flue gas ammonia-nitrogen molar ratio distribution cloud map, flue gas temperature distribution cloud map and flue gas incident angle distribution cloud map at 500mm upstream of the first catalyst inlet from the initial flow field simulation results, and calculate the relative deviation of the corresponding extracted features based on each extracted distribution cloud map. The judgment module is used to determine whether all relative deviations meet the technical objectives; if not, the model construction and simulation block modifies the structure of the virtual SCR denitrification reactor in the CFD model and drives the modified CFD model to re-simulate the flow of flue gas until the judgment module determines that the relative deviations of all extracted features meet the technical objectives.

[0021] Thirdly, this application provides a computer device that adopts the following technical solution: A computer device includes a memory, a processor, and a computer program stored in the memory, the processor executing the computer program to implement the multi-objective flow field collaborative optimization design method for an SCR denitrification system as described in the first aspect.

[0022] Fourthly, this application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium storing a computer program that can be loaded by a processor and executed as described in the first aspect for a multi-objective flow field collaborative optimization design method for an SCR denitrification system.

[0023] In summary, this application includes at least one of the following beneficial technical effects: A CFD model is constructed based on the physical data of the SCR denitrification reactor, which can highly reproduce the real reactor environment. The model simulates flue gas flow and obtains initial flow field simulation results. From these results, flue gas velocity, ammonia-nitrogen molar ratio, temperature, and incident angle distribution contour maps at 500mm upstream of the first catalyst inlet are extracted. These contour maps visually present multifaceted characteristics of the flow field, providing a comprehensive and detailed basis for evaluating the flow field conditions. Based on these distribution contour maps, the relative deviations of corresponding characteristics are calculated and compared with the technical targets to accurately determine whether the current flow field meets the requirements. If not, the structure of the virtual SCR denitrification reactor in the CFD model is modified to effectively improve the flow field conditions. Through continuous modification and resimulation, the relative deviations of all characteristics meet the technical targets, thereby improving the performance of the SCR denitrification system. A uniform flow field allows for sufficient contact between the flue gas and the catalyst, enabling a more complete reaction between ammonia and nitrogen oxides, thus improving the removal efficiency of nitrogen oxides and making emissions more compliant with environmental standards. The optimized flow field ensures uniform distribution of ammonia in the flue gas, reducing the escape of unreacted ammonia and lowering the risk of corrosion and blockage in downstream equipment caused by ammonia escape, thus guaranteeing long-term stable operation of the system. Uniform velocity, temperature, and incident angle distribution reduce wear and aging caused by localized high-speed scouring, ash accumulation and blockage, or uneven temperature, extending catalyst lifespan and reducing catalyst replacement costs. Attached Figure Description

[0024] Figure 1 This is a first flowchart of an embodiment of the method of this application; Figure 2 This is a second flowchart of an embodiment of the method of this application; Figure 3 This is a third flowchart of an embodiment of the method of this application; Figure 4 This is the fourth flowchart of an embodiment of the method of this application. Detailed Implementation

[0025] To make the purpose, technical solution, and advantages of this application clearer, the following description is provided in conjunction with the appendix. Figures 1-4 The present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.

[0026] The first embodiment of this application discloses a multi-objective flow field collaborative optimization design method for SCR denitrification systems. (Refer to...) Figure 1 The multi-objective flow field collaborative optimization design method includes S110-S160: S110, Construct a CFD model of the SCR denitrification reactor based on the current physical data of the SCR denitrification reactor; S120, simulate the flow of flue gas in the CFD model and obtain the initial flow field simulation results. From the initial flow field simulation results, extract the flue gas velocity distribution cloud map, flue gas ammonia nitrogen molar ratio distribution cloud map, flue gas temperature distribution cloud map and flue gas incident angle distribution cloud map at 500 mm upstream of the first catalyst inlet. S130. Based on the extracted distribution cloud maps, calculate the relative deviation of the corresponding extracted features and determine whether all relative deviations meet the technical objectives. If not, modify the structure of the virtual SCR denitrification reactor in the CFD model and drive the modified CFD model to re-simulate the flow of flue gas until the relative deviations of all extracted features meet the technical objectives. S140, If the relative deviations of all extracted features meet the technical objectives, extract the pressure distribution cloud map of the SCR system from the initial flow field simulation results; S150, calculate the pressure loss of the SCR system based on the pressure distribution cloud map of the SCR system, and determine whether the pressure loss meets the design requirements. S160, if not, then modify the structure of the virtual SCR denitrification reactor in the CFD model, drive the modified CFD model to re-simulate the flow of flue gas until the relative deviations of all characteristics meet the technical objectives and the pressure loss meets the design requirements.

[0027] Specifically, the SCR denitrification reactor is physically located between the boiler economizer and air preheater. A detailed model of the entire system geometry is created using SolidWorks or AutoCAD, encompassing core components such as the economizer outlet flue, main flue guide vane layout, reactor body, catalyst layer support structure, ammonia injection grid (AIG), internal rectification device, and air preheater inlet section. During modeling, key characteristic parameters such as guide vane curvature, AIG nozzle density, and inclination angle are retained, while details with negligible impact on mainstream flow, such as bolt holes and local reinforcing ribs, are removed to ensure the model reflects realistic flow field characteristics without excessively increasing computational burden.

[0028] The main flow channel uses a hexahedral-dominated structured mesh to ensure computational efficiency and accuracy. For complex flow regions such as the guide vane junction area, ammonia injection area, and catalyst front end, unstructured tetrahedral / prism layer meshes are used for local refinement. Specifically, a boundary layer mesh is set within 500 mm upstream of the first catalyst layer, with the wall distance parameter strictly controlled within the range of 30–60 to meet the analytical requirements of the standard k-ε turbulence model for near-wall flow. After mesh generation, boundary marking is completed using ICEM CFD or ANSYS Meshing, clearly defining various boundary conditions such as inlet (economizer outlet), outlet (air preheater inlet), wall, symmetry plane, and ammonia injection port.

[0029] Flow field simulation was performed using the Fluent or STAR-CCM+ solver platform. The standard k-ε turbulence model was used to capture the strong turbulence characteristics of flue gas in a variable cross-section channel, while the Species Transport Model was activated to simulate the dynamic mixing process of NH3 and NOx in the flue gas. Boundary conditions were set as follows: the economizer outlet was set as a mass flow inlet, with the measured total flue gas mass flow rate, temperature (typically around 350℃), O2 / CO2 / H2O volume fractions, and 5%–10% turbulence intensity parameters input; the ammonia injection port was used to inject an NH3 / N2 mixture at the mass flow inlet, with its molar concentration determined by back-calculation based on the theoretical ammonia-to-nitrogen ratio (NSR=1.0); the outlet was set as a pressure outlet with an ambient back pressure (e.g., -500 Pa). The solution process employed the SIMPLE algorithm coupled with the pressure-velocity field. Momentum, energy, and component equations were discretized using a second-order upwind scheme, and iterative calculations were performed until the residuals converged to below 1e-5 and the monitoring point parameters stabilized.

[0030] After completing the initial flow field simulation, four key distribution contour maps are extracted from the standard cross-section 500 mm upstream of the first catalyst layer: flue gas velocity distribution (velocity magnitude contour map), ammonia-nitrogen molar ratio distribution (calculated based on species concentration), temperature distribution (temperature contour), and flue gas incident angle distribution (calculated by the angle between the velocity vector and the vertical normal). Based on this, the uniformity indices for each parameter are calculated: the relative standard deviation of the velocity field must be ≤15%, the relative standard deviation of the ammonia-nitrogen molar ratio must be ≤5%, the absolute deviation of the temperature field must be ≤±10℃, and the average deviation of the incident angle must be ≤10°. If any index exceeds the limit, the virtual reactor structure (such as the angle of the guide vane, the layout of the AIG nozzle, or the position of the rectifier) ​​is parametrically modified, and the CFD simulation is restarted until all flow field indices meet the standards.

[0031] Once the flow field uniformity meets the requirements, the global pressure distribution cloud map of the SCR system is further extracted, and the system pressure loss value is calculated. If the pressure loss exceeds the design threshold, the virtual reactor structure is parametrically modified, and the CFD simulation is restarted until all flow field indicators and pressure loss values ​​meet the requirements.

[0032] Reference Figure 2 The steps for modifying the structure of the virtual SCR denitrification reactor in the CFD model include S210-S250: S210: Extract the core parameters of the current flow field's substandard features and generate a structured feature vector; S220, select structural modification schemes that match the structured feature vectors from the pre-built SCR flow field modification case library; S230, if a match is found, load the relevant parameters of the matched structural modification scheme into the CFD model; If no match is found, the built-in SCR flow field control expert rule engine is invoked to trigger the corresponding structural adjustment logic based on the type of non-compliant flow field characteristics. The structural adjustment logic includes adding structures and adjusting the parameters of existing structures. S250, loads the relevant parameters of the triggered structural adjustment logic into the CFD model.

[0033] Specifically, a pre-built Python script is used to intelligently diagnose the CFD flow field simulation results. Based on abnormal spatial distribution patterns of flow field parameters, such as high-speed region aggregation coordinates, ammonia concentration deficit zones, temperature gradient distortion, or large-angle flue gas impact zones, the script automatically extracts core feature parameters including the three-dimensional coordinates of abnormal locations, gradient change direction, exceedance threshold, and spatial influence range, and encapsulates them into a structured feature vector. This vector, serving as a digital fingerprint of flow field defects, is pushed in real-time to a pre-built SCR flow field remediation case library for intelligent matching.

[0034] The case library is built from engineering data of historically successful retrofit projects. Each record includes an original defect description map, implemented structural modification measures, post-modification performance improvement data, and applicable boundary condition labels. Structural modification measures include adding guide vanes with specific curvature, adjusting the elevation angle distribution of AIG ammonia injection grilles, or adding rectifier grille layers. The system uses a cosine similarity algorithm or a K-nearest neighbor (KNN) classification model to quickly retrieve feature vectors. If a historical solution with a similarity exceeding the 85% threshold exists, its structural parameters are directly retrieved and mapped to the current CFD model. For example, for the downstream vortex problem at a bend, the installation position and tilt angle parameters of the inclined guide vane are automatically loaded; for the defect of insufficient lateral mixing, the optimized AIG nozzle opening distribution matrix is ​​inherited.

[0035] When no matching solution is found in the case library, the system activates the built-in SCR flow field control expert rule engine. This engine integrates heuristic logic derived from multi-dimensional engineering experience, embedding multiple "IF-THEN" heuristic logics, prioritized accordingly. For example: "If significant velocity deflection occurs in the inlet section accompanied by right-side vortex accumulation → add a curvature-gradient guide vane to the left side of the front horizontal flue"; "If ammonia-nitrogen mixing is uneven and concentrated in the bottom region → increase the diameter ratio of the bottom ammonia injection branch pipe and fine-tune the injection angle upward by 5°"; "If the temperature field is high in the center and low at the edges → install a temperature equalization baffle in the expansion section in front of the reactor to promote thermal diffusion." These rules support combined triggering and parameter co-optimization, ultimately outputting a set of candidate structural adjustment instructions. The CFD geometric model is dynamically updated through a parametric modeling interface to complete a single structural iteration optimization.

[0036] Reference Figure 3 Multi-objective flow field collaborative optimization design methods also include S310-S360: S310, when multiple structural modification schemes are matched or multiple structural adjustment logics are triggered, establish a weighted evaluation function based on flow field uniformity index, pressure design requirements and economic objectives; S320, analyze whether there are complementary dimensions between multiple structural modification schemes or multiple structural adjustment logics triggered by them; S330, if not, input multiple structural modification schemes or triggered multiple structural adjustment logics into the pre-trained neural network model, and output the corresponding flow field uniformity index mapping value, pressure mapping value and economic mapping value. S340 inputs the flow field uniformity index mapping value, pressure mapping value and economic mapping value into the weighted evaluation function, outputs the corresponding function value, and takes the structural modification scheme with the smallest function value as the optimal structural modification scheme or the structural adjustment logic with the smallest function value as the optimal structural adjustment logic. S350, if so, then identify the optimal combination of parameters for each dimension of flow field uniformity, pressure loss, and economy in the structural modification scheme or structural adjustment logic. S360 integrates the optimal combination of individual parameters and combines them with the boundary conditions of the CFD model to generate a hybrid structure modification scheme, and loads the relevant parameters of the hybrid structure modification scheme into the CFD model.

[0037] Specifically, when the system identifies multiple potential structural modification schemes or simultaneously triggers multiple structural adjustment logics, it enters a multi-objective collaborative decision-making process. The core objective of this process is to seek the globally optimal solution across three key dimensions: flow field uniformity, pressure loss control, and economic cost. The starting point for decision-making is to establish a quantified weighted evaluation function. In this function, U represents the comprehensive flow field uniformity index, which incorporates the distribution deviation assessment results of key flow field parameters such as velocity, concentration, and temperature through normalization processing. The system pressure drop after implementing the scheme is represented by the increment relative to the baseline condition; C is the economic cost factor, which quantifies economic indicators such as the increase or decrease in steel consumption, changes in manufacturing process complexity, and ease of maintenance involved in the scheme. The weighting coefficients w1, w2, and w3 are not fixed, but can be dynamically configured according to specific application scenarios. For example, the retrofitting of old units focuses more on economic efficiency w3, while new high-performance projects emphasize flow field performance w1 and pressure drop control w2.

[0038] Next, an in-depth analysis is conducted to determine whether there is complementarity in the dimensions of action among these coexisting schemes or logics. For example, it might be found that scheme A is highly effective in significantly improving the uniformity of downstream velocity distribution, but it will lead to a certain increase in pressure loss; while scheme B can effectively reduce system flow resistance, but its effect on improving the mixing uniformity of key substances (such as ammonia) is limited. If the analysis shows that the influence dimensions of these schemes are independent (orthogonal) and their effects are theoretically additive without significant conflict, then they are judged to have fusion potential and belong to the "fusionable" case. Conversely, if the mechanisms of action of the schemes overlap, they are mutually restrictive, or their objectives conflict and are difficult to reconcile, then they are considered "independent" schemes.

[0039] For schemes or logics deemed "independent," the system inputs their corresponding detailed structural parameter change vectors (such as the specific number of guide vanes, installation angle, spatial coordinate distribution, or AIG opening matrix configuration) into a pre-trained deep fully connected neural network model. This neural network model is trained based on thousands of sets of historically accumulated high-fidelity CFD simulation data, enabling it to efficiently establish a nonlinear mapping relationship between structural parameter changes and key performance indicators. The model outputs predicted flow field uniformity mapping values, pressure loss mapping values, and economic cost mapping values, representing the expected performance of the scheme in the three target dimensions. Subsequently, these predicted values ​​are substituted into the aforementioned weighted evaluation function F for calculation. The system compares the calculated function value F and ultimately selects the scheme with the minimum F value as the optimal single structural modification scheme or the optimal structural adjustment logic for adoption and implementation, because the function is designed as cost-based, and a smaller value indicates better overall performance.

[0040] For combinations of schemes deemed "integrable," the decision-making path shifts to identifying the individual advantages of each scheme and innovatively integrating them. The system first deeply analyzes each scheme, identifying its optimal combination of parameters across three dimensions: flow field uniformity (U), pressure loss (ΔP), and economy (C). For example, it might identify scheme A as having unique parameters for achieving optimal velocity uniformity, scheme B as possessing a key structural configuration that minimizes pressure loss, and scheme C as achieving optimal material utilization efficiency (economy). After identifying these optimal factors, the system enters the fusion design phase. It analyzes the core improvement elements of each scheme and their effective sub-modules (such as the guiding structure form in specific regions, unique rectification concepts, etc.), and performs intelligent topology integration while considering the boundary conditions of the CFD physical model (such as geometric compatibility, flow continuity, and process feasibility). Integration methods include: merging efficient guiding structures from different schemes in different functional regions, or applying the most advantageous rectification strategies at different levels of the flow channel (such as inlet section, mixing section, and outlet section). Through this deep integration, a completely new "hybrid structure modification scheme" is ultimately generated. After the parameters of this scheme are verified for rationality, they are loaded back into the CFD model to start a new round of simulation verification.

[0041] Reference Figure 4 Multi-objective flow field collaborative optimization design methods also include S410-S470: S410, extract the lifetime distribution cloud map of the catalyst layer from the final optimized flow field simulation results; S420, calculates the local velocity extremes, fly ash concentration gradients and temperature fluctuation standard deviations at key locations in the catalyst layer based on the lifetime distribution cloud map; S430 inputs the catalyst physical parameters and chemical components into the constructed deactivation mechanics model to calculate the combined deactivation rate of alkali metal poisoning, arsenic poisoning and hot sintering. S440 predicts the early warning coordinates of the minimum lifetime region based on local flow velocity extremes, fly ash concentration gradients, temperature fluctuation standard deviations, and overall deactivation rates. S450, if the predicted life value corresponding to the minimum life region warning coordinates is less than the life design value, then the minimum life region warning coordinates and associated local flow field characteristics are back-mapped into the CFD model as new constraint boundary conditions. S460, calculate the influence weight of each structural parameter of the SCR denitrification reactor on the local deactivation rate of the catalyst, and identify the key geometric factors affecting the lifetime distribution; S470 automatically adjusts the internal structural layout of the virtual SCR denitrification reactor based on a CFD model driven by key geometric factors.

[0042] Specifically, once both the flow field uniformity and pressure loss objectives are met, based on the final flow field simulation results, the lifetime distribution cloud map of the catalyst layer is first extracted. Then, three key local flow field characteristics are quantified using post-processing tools: the average velocity extreme value within the catalyst layer (characterizing the intensity of airflow scouring), the fly ash concentration gradient (derived from particle trajectory statistics from the discrete phase model DPM), and the temperature fluctuation standard deviation (calculated through transient condition sampling or multi-condition comparison). Simultaneously, the actual physical parameters of the catalyst layer (such as pore density / wall thickness / pitch) and chemical composition (such as the V2O5-WO3 / TiO2 ratio and the thickness of the alkali-resistant metal coating) are input into an integrated deactivation kinetic model. This model couples the alkali metal (K / Na) adsorption kinetic equation, the arsenic compound permeation diffusion model, and the thermal sintering specific surface area decay function, outputting the comprehensive deactivation rate per unit time. .

[0043] Combining the aforementioned local flow field parameters and deactivation rate, a spatially resolved lifetime prediction model is constructed: , where the coefficient Regression calibration based on aging sample data from on-site service units. L(x,y,z) represents the initial lifetime of the catalyst layer under ideal conditions, L(x,y,z) represents the predicted lifetime value corresponding to the three-dimensional coordinates, and v represents the extreme value of the average flow velocity. The value represents the fly ash concentration gradient, and T represents the standard deviation of temperature fluctuation. The model automatically identifies the spatial coordinates of the minimum lifetime region and the corresponding lifetime value. If the predicted value is lower than the design threshold, the high-risk characteristic parameters associated with the coordinates (abnormal flow rate peak, fly ash enrichment gradient, temperature fluctuation amplitude) are back-mapped to the CFD model boundary condition library to generate local constraint equations that are embedded in subsequent optimization iterations.

[0044] Furthermore, sensitivity analysis methods such as the Sobol index or local regression SHAP values ​​are used to calculate the influence weights of various structural parameters of the SCR reactor (such as inlet cone angle, guide vane curvature radius, rectifier grid spacing, and AIG height from the catalyst) on the local deactivation rate of the catalyst. This identifies key geometric factors that dominate the lifetime distribution. For example, the "inlet diffuser length" has been shown to strongly influence the peak velocity in the central region, while the "top guide vane inclination angle" determines the risk of low-velocity deposition at the edge. Based on this, the CFD model is driven into a closed-loop adaptive optimization mode. With the help of parametric modeling interfaces such as ANSYS DesignXplorer or modeFRONTIER, the key geometric factors are automatically adjusted, such as dynamic optimization of the diffuser length by ±15% and fine adjustment of the guide vane inclination angle in 0.5° increments. While meeting the original flow field and pressure loss targets, the catalyst lifetime consistency is maximized, ultimately outputting an optimal design scheme that balances performance, energy consumption, and durability throughout the entire life cycle.

[0045] Based on the above method embodiments, the second embodiment of this application discloses a multi-objective flow field collaborative optimization design system for SCR denitrification systems. The multi-objective flow field collaborative optimization design system for SCR denitrification systems of this application embodiment can implement any of the above-described methods for multi-objective flow field collaborative optimization design of SCR denitrification systems, and the specific working process of each module in the multi-objective flow field collaborative optimization design system for SCR denitrification systems can be referred to the corresponding process in the above method embodiments.

[0046] For ease of understanding, an example is as follows: A multi-objective flow field collaborative optimization design system for SCR denitrification systems includes: The model building and simulation module is used to build a CFD model of the SCR denitrification reactor based on the current physical data of the SCR denitrification reactor, simulate the flow of flue gas in the CFD model, and obtain the initial flow field simulation results. The feature extraction and processing module is used to extract the flue gas velocity distribution cloud map, flue gas ammonia-nitrogen molar ratio distribution cloud map, flue gas temperature distribution cloud map, and flue gas incident angle distribution cloud map from the initial flow field simulation results at 500 mm upstream of the first catalyst inlet, and calculate the relative deviation of the corresponding extracted features based on each extracted distribution cloud map. The judgment module is used to determine whether all relative deviations meet the technical objectives. If not, the model building and simulation block modifies the structure of the virtual SCR denitrification reactor in the CFD model and drives the modified CFD model to re-simulate the flow of flue gas until the judgment module determines that the relative deviations of all extracted features meet the technical objectives.

[0047] The third embodiment of this application provides a computer device, which may include a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement a multi-objective flow field collaborative optimization design method for an SCR denitrification system.

[0048] The memory can communicate with the processor via a communication bus, which can be an address bus, a data bus, a control bus, etc.

[0049] Additionally, the memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device.

[0050] Furthermore, the processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0051] The fourth embodiment of this application provides a computer-readable storage medium storing a computer program that can be loaded and executed by a processor for a multi-objective flow field collaborative optimization design method for an SCR denitrification system.

[0052] The computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device; the program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0053] It should be noted that the computer device and storage medium in this application embodiment are respectively electronic devices and storage media that apply the above-described multi-objective flow field collaborative optimization design method for SCR denitrification systems. Therefore, all embodiments of the above-described multi-objective flow field collaborative optimization design method for SCR denitrification systems are applicable to the computer device and storage medium, and can achieve the same or similar beneficial effects. For the computer device / storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple; relevant details can be found in the descriptions of the method embodiments.

[0054] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce a good effect.

[0055] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.

Claims

1. A multi-objective flow field collaborative optimization design method for SCR denitrification systems, characterized in that, include: Based on the current physical data of the SCR denitrification reactor, a CFD model of the SCR denitrification reactor is constructed. The flow of flue gas was simulated in the CFD model, and the initial flow field simulation results were obtained. From the initial flow field simulation results, the flue gas velocity distribution cloud map, flue gas ammonia-nitrogen molar ratio distribution cloud map, flue gas temperature distribution cloud map, and flue gas incident angle distribution cloud map were extracted at 500 mm upstream of the first catalyst inlet. Based on the extracted distribution cloud maps, calculate the relative deviation of the corresponding extracted features, and determine whether all relative deviations meet the technical objectives. If not, the structure of the virtual SCR denitrification reactor in the CFD model is modified, and the modified CFD model is driven to re-simulate the flow of flue gas until the relative deviations of all extracted features meet the technical objectives.

2. The multi-objective flow field collaborative optimization design method for an SCR denitrification system according to claim 1, characterized in that, The multi-objective flow field collaborative optimization design method also includes: After all the relative deviations of the extracted features meet the technical objectives, the pressure distribution cloud map of the SCR system is extracted from the initial flow field simulation results; Calculate the pressure loss of the SCR system based on the pressure distribution cloud map of the SCR system, and determine whether the pressure loss meets the design requirements. If not, the structure of the virtual SCR denitrification reactor in the CFD model is modified, and the modified CFD model is driven to re-simulate the flow of flue gas until the relative deviations of all characteristics meet the technical objectives and the pressure loss meets the design requirements.

3. The multi-objective flow field collaborative optimization design method for an SCR denitrification system according to claim 2, characterized in that, The steps for modifying the structure of the virtual SCR denitrification reactor in the CFD model include: Extract the core parameters of the current substandard flow field features and generate a structured feature vector; Select structural modification schemes that match the structured feature vector from the pre-built SCR flow field modification case library; If a match is found, the relevant parameters of the matched structural modification scheme are loaded into the CFD model; If no match is found, the built-in SCR flow field control expert rule engine is invoked to trigger the corresponding structural adjustment logic based on the type of non-compliant characteristics of the current flow field. The structural adjustment logic includes adding structures and adjusting the parameters of the original structures. Load the relevant parameters of the triggered structural adjustment logic into the CFD model.

4. The multi-objective flow field collaborative optimization design method for an SCR denitrification system according to claim 3, characterized in that, The multi-objective flow field collaborative optimization design method also includes: When multiple structural modification schemes are matched or multiple structural adjustment logics are triggered, a weighted evaluation function is established based on the flow field uniformity index, pressure design requirements, and economic objectives. Multiple structural modification schemes or multiple structural adjustment logics triggered by the input are fed into a pre-trained neural network model, which outputs the corresponding flow field uniformity index mapping value, pressure mapping value, and economic mapping value. The flow field uniformity index mapping value, pressure mapping value, and economic mapping value are input into the weighted evaluation function, and the corresponding function value is output. The structural modification scheme with the smallest function value is taken as the optimal structural modification scheme, or the structural adjustment logic with the smallest function value is taken as the optimal structural adjustment logic.

5. The multi-objective flow field collaborative optimization design method for an SCR denitrification system according to claim 4, characterized in that, The steps preceding inputting multiple structural modification schemes or triggered multiple structural adjustment logics into a pre-trained neural network model include: Analyze whether there are complementary dimensions between multiple structural modification schemes or multiple structural adjustment logics triggered by them; If not, then input multiple structural modification schemes or triggered multiple structural adjustment logics into the pre-trained neural network model; If so, identify the optimal combination of parameters for each dimension of flow field uniformity, pressure loss, and economy in the structural modification scheme or structural adjustment logic; By integrating the optimal combination of individual parameters and combining them with the boundary conditions of the CFD model, a hybrid structural modification scheme is generated, and the relevant parameters of the hybrid structural modification scheme are loaded into the CFD model.

6. The multi-objective flow field collaborative optimization design method for an SCR denitrification system according to claim 3, characterized in that, The multi-objective flow field collaborative optimization design method also includes: The lifetime distribution cloud map of the catalyst layer is extracted from the final optimized flow field simulation results; The local velocity extremes, fly ash concentration gradients, and temperature fluctuation standard deviations at key locations in the catalyst layer are calculated based on the lifetime distribution cloud map. The catalyst's physical parameters and chemical composition are input into the constructed deactivation mechanics model to calculate the combined deactivation rate due to alkali metal poisoning, arsenic poisoning, and thermal sintering. Based on local velocity extremes, fly ash concentration gradients, temperature fluctuation standard deviations, and overall deactivation rates, the coordinates of the minimum lifetime warning region are predicted.

7. The multi-objective flow field collaborative optimization design method for an SCR denitrification system according to claim 6, characterized in that, The steps following the prediction of the minimum lifespan region warning coordinates include: If the predicted life value corresponding to the minimum life region warning coordinates is less than the life design value, then the minimum life region warning coordinates and the associated local flow field characteristics are back-mapped into the CFD model as new constraint boundary conditions. The influence weights of each structural parameter of the SCR denitrification reactor on the local deactivation rate of the catalyst were calculated, and the key geometric factors affecting the lifetime distribution were identified. The internal structural layout of the virtual SCR denitrification reactor is automatically adjusted based on the key geometric factors driven by the CFD model.

8. A multi-objective flow field collaborative optimization design system for SCR denitrification systems, characterized in that, Performing the multi-objective flow field collaborative optimization design method for an SCR denitrification system as described in any one of claims 1 to 7, comprising: The model building and simulation module is used to build a CFD model of the SCR denitrification reactor based on the current physical data of the SCR denitrification reactor, simulate the flow of flue gas in the CFD model, and obtain the initial flow field simulation results. The feature extraction and processing module is used to extract the flue gas velocity distribution cloud map, flue gas ammonia-nitrogen molar ratio distribution cloud map, flue gas temperature distribution cloud map and flue gas incident angle distribution cloud map at 500mm upstream of the first catalyst inlet from the initial flow field simulation results, and calculate the relative deviation of the corresponding extracted features based on each extracted distribution cloud map. The judgment module is used to determine whether all relative deviations meet the technical objectives; if not, the model construction and simulation block modifies the structure of the virtual SCR denitrification reactor in the CFD model and drives the modified CFD model to re-simulate the flow of flue gas until the judgment module determines that the relative deviations of all extracted features meet the technical objectives.

9. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the multi-objective flow field collaborative optimization design method for an SCR denitrification system as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program capable of being loaded by a processor and executed as described in any one of claims 1 to 7 for the multi-objective flow field collaborative optimization design method for SCR denitrification systems.

Images