A numerical simulation analysis method and system for a flow field of a flash angle type flow regulating valve

Abstract

The present application relates to the technical field of flow field numerical simulation analysis, and particularly relates to a flow field numerical simulation analysis method and system for a flash angle type flow regulating valve. The method obtains initial data of a historical simulation flow field and flow velocity and direction of a fluid in a historical simulation process; divides the historical simulation into scene categories based on similarity of the initial data of the flow field, and obtains a reference historical simulation according to the flow velocity and direction in the scene categories; obtains an optimal grid division strategy based on occurrence of a grid division strategy corresponding to the reference historical simulation in different scene categories; obtains reference grid division parameters based on boundary conditions and grid division parameters of the reference historical simulation corresponding to the optimal grid division strategy, and then determines grid division of a geometric model of the flash angle type flow regulating valve in current simulation. The present application effectively improves accuracy and robustness of flash flow field numerical simulation by obtaining the optimal grid division strategy and the reference grid division parameters.

Description

Technical Field

[0001] This invention relates to the field of flow field numerical simulation and analysis technology, specifically to a flow field numerical simulation and analysis method and system for a flash angle flow control valve. Background Technology

[0002] Flashing is a common phase change phenomenon in fluid systems. When the liquid pressure rapidly drops below its saturated vapor pressure, part of the liquid boils instantly and transforms into a gas phase, forming a two-phase flow. In industrial processes, flow control valves are key components for controlling fluid flow. When high-temperature, high-pressure liquids flow through a flow control valve, the throttling effect between the valve core and seat causes a sharp increase in liquid velocity and a corresponding decrease in pressure. This process easily triggers flashing within the flow control valve. Flashing not only generates noise and vibration, but the high-speed bubbles produced also create a huge impact force when they burst at the valve outlet, causing erosion and damage to key components such as the valve core and seat. This seriously affects the safe and stable operation of industrial processes and the service life of angle flow control valves.

[0003] To ensure the safety of industrial processes and angle flow control valves, it is necessary to deeply analyze the complex flow field characteristics inside the valves. However, since the flow conditions inside angle flow control valves are difficult to observe and obtain directly, existing technologies use numerical simulation methods to analyze the flow field inside the control valves. The conventional procedure for numerically simulating the flow field inside angle flow control valves is to first complete a model based on the valve's structure, then set the model's boundary and initial conditions based on fluid inflow and outflow data, simultaneously selecting appropriate physical models and mesh generation parameters, and finally simulating and analyzing the flow field characteristics inside the valve.

[0004] However, flash evaporation triggers phase transitions and generates tiny bubbles, disrupting the original flow field and causing a series of related changes in the fluid flow within the valve. Furthermore, flash evaporation involves microscopic phase transitions, making the numerical simulation results highly dependent on the quality of the mesh generation within the valve. This necessitates adjusting the mesh generation parameters according to the fluid boundary conditions, but existing mesh generation methods struggle to precisely control the mesh generation at every detail. Even methods that rely on physical fields for adaptive refinement are highly dependent on the quality of the initial mesh generation. If the initial mesh fails to provide a good foundation for adaptive refinement, the numerical simulation results will deviate significantly from the actual situation, making it difficult to quickly and accurately obtain the desired simulation results and failing to meet the precise flow field analysis requirements of flash angle flow control valves in industrial scenarios. Summary of the Invention

[0005] To address the technical problem of inaccurate numerical simulation results caused by the difficulty in precisely controlling the mesh generation at various details using existing mesh generation methods, this invention aims to provide a numerical simulation analysis method and system for the flow field of a flash angle flow control valve. The specific technical solution adopted is as follows:

[0006] In a first aspect, one embodiment of the present invention provides a method for numerical simulation analysis of the flow field of a flash angle flow control valve, the method comprising the following steps:

[0007] Obtain the initial watershed data, boundary conditions, mesh generation strategy and mesh generation parameters for each historical simulation, as well as the flow velocity and direction of the fluid in each unit region at each moment during each historical simulation.

[0008] Based on the similarity of the initial watershed data between historical simulations, the historical simulations are divided into scenario categories. According to the flow velocity and direction of the fluid in each unit area at each moment during each historical simulation in each scenario category, a reference historical simulation in each scenario category is obtained.

[0009] Based on the occurrence of the mesh partitioning strategy corresponding to the reference historical simulation in different scene categories, the optimal mesh partitioning strategy is obtained; based on the boundary conditions and mesh partitioning parameters of the reference historical simulation corresponding to the optimal mesh partitioning strategy, a trained convolutional neural network is obtained; the boundary conditions of the current simulation are input into the trained convolutional neural network to obtain the reference mesh partitioning parameters of the current simulation.

[0010] The meshing of the geometric model of the flash angle flow control valve in the current simulation is determined by the optimal meshing strategy and reference meshing parameters.

[0011] Furthermore, the method for obtaining the scene category is as follows:

[0012] The initial watershed data for each historical simulation is constructed into a vector, which serves as the initial data vector for each historical simulation.

[0013] The cosine similarity of the initial data vectors of any two historical simulations is negatively correlated and normalized, and used as the scene distance metric between any two historical simulations.

[0014] Based on scene metric distance, historical simulations are clustered using the DBSCAN density clustering algorithm to obtain scene categories.

[0015] Furthermore, the method for obtaining the reference historical simulation is as follows:

[0016] For any scenario category, based on the flow velocity and direction of the fluid in each unit area at each moment during each historical simulation within that scenario category, the flow accuracy of each unit area at each moment during each historical simulation within that scenario category is obtained;

[0017] Based on the accuracy of the flow, the quality importance of each unit area within the scene category is obtained;

[0018] Based on the accuracy of the flow and the importance of the quality, obtain the quality score for each historical simulation within this scenario category;

[0019] When the quality score is greater than the preset quality score threshold, the corresponding historical simulation will be used as the reference historical simulation within that scene category.

[0020] Furthermore, the method for obtaining the accuracy of the flow is as follows:

[0021] The flow velocity and direction of the fluid in each unit area at each moment during each historical simulation in this scenario category are constructed as a flow velocity vector;

[0022] For any unit region and any time, the average vector of the flow velocity vector of the unit region at that time in each historical simulation within the scene category is used as the reference vector of the unit region at that time in the historical simulation within the scene category.

[0023] For any historical simulation within this scenario category, the result of normalizing the cosine similarity between the flow velocity vector of the unit region at that moment in the historical simulation and the reference vector is taken as the accuracy of the flow in the unit region at that moment in the historical simulation.

[0024] Furthermore, the method for obtaining the importance of the quality is as follows:

[0025] For any unit region and any historical simulation within that scenario category, the average flow accuracy of that unit region at all times during the historical simulation is taken as the important analytical value of that unit region in the historical simulation.

[0026] The result of normalizing the mean of the important analytical values ​​of the unit region within the scene category across all historical simulations is used as the quality importance of the unit region within the scene category.

[0027] Furthermore, the method for obtaining the quality score is as follows:

[0028] For any unit region and any historical simulation within that scene category, the product of the important analysis value of that unit region in that historical simulation and the quality importance of that unit region within that scene category is taken as the degree of participation of that unit region in the historical simulation.

[0029] The result of normalizing the mean of the degree of participation of all unit regions in the historical simulation is used as the quality score of the historical simulation within the scenario category.

[0030] Furthermore, the method for obtaining the optimal mesh partitioning strategy is as follows:

[0031] The mesh partitioning strategy corresponding to the reference historical simulation within each scene category is used as the reference mesh partitioning strategy.

[0032] For any reference mesh partitioning strategy, the mesh partitioning strategy with a reference history simulation is the scene category of the reference mesh partitioning strategy, and is used as the reference scene category of the reference mesh partitioning strategy;

[0033] The ratio of the number of reference scene categories to the total number of all scene categories is used as the stability of the reference grid partitioning strategy.

[0034] The reference meshing strategy corresponding to the highest stability is taken as the optimal meshing strategy.

[0035] Furthermore, the method for obtaining the trained convolutional neural network is as follows:

[0036] All historical simulations using the optimal mesh partitioning strategy will be used as historical simulations for analysis.

[0037] The boundary conditions of each historical simulation are used as input to the convolutional neural network (CNN), and the output of the CNN is the grid partitioning parameters of each historical simulation. The CNN is trained to obtain a trained CNN. In the loss function of the CNN training, the quality score of each historical simulation is used as the attention weight.

[0038] Furthermore, the method for obtaining the unit region is as follows:

[0039] A three-dimensional coordinate system is constructed with the center of the smallest circumcircle of the geometric model of the flash angle flow control valve as the origin; where the X-axis in the three-dimensional coordinate system is the axial direction along the bottom inlet of the valve, that is, the initial flow direction of the fluid entering the valve, the Y-axis is the lateral direction of the valve, that is, the flow direction of the fluid at the side outlet, and the Z-axis is the axial direction parallel to the valve core, that is, the core flow direction of the fluid through the valve core.

[0040] The preset distance is used as the unit length of the three-dimensional coordinate system. All coordinate points in the three-dimensional coordinate system with integer dimensions are used as reference points. A cube with a side length of unit length is constructed with each reference point as the center, and each cube is used as the reference area.

[0041] The reference area that is entirely located within the flow domain of the flash angle flow control valve is taken as the unit area.

[0042] Secondly, another embodiment of the present invention provides a flow field numerical simulation analysis system for a flash angle flow regulating valve. The system includes: a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the above methods.

[0043] The present invention has the following beneficial effects:

[0044] This invention first categorizes historical simulations into scenario categories based on the similarity of initial watershed data between simulations, accurately identifying the scenario corresponding to each historical simulation and preparing for the subsequent selection of high-quality historical simulations. Then, based on the flow velocity and direction of the fluid in each unit area at each moment during each historical simulation within each scenario category, reference historical simulations are obtained for each scenario category, accurately identifying high-quality historical simulations within each category and preparing for the subsequent accurate simulation of the flow field of the angle flow control valve. Finally, based on the occurrence of the mesh partitioning strategy corresponding to the reference historical simulations in different scenario categories, the optimal mesh partitioning strategy is accurately obtained, which is beneficial for providing a more robust and reliable simulation for the safe design and performance optimization of valves in real and variable working environments. To more accurately simulate the flash flow field numerically, a well-trained convolutional neural network is obtained based on the boundary conditions and meshing parameters of the reference historical simulation corresponding to the optimal meshing strategy. This effectively improves the accuracy, efficiency, and robustness of the flash flow field numerical simulation. Furthermore, the current simulation boundary conditions are input into the trained convolutional neural network to accurately obtain the reference meshing parameters for the current simulation, laying a reliable foundation for subsequent high-precision numerical simulations. Finally, the meshing of the geometric model of the flash angle flow control valve in the current simulation is accurately determined through the optimal meshing strategy and reference meshing parameters, enabling efficient and accurate acquisition of ideal simulation analysis results and effectively meeting the precise requirements for flow field analysis of flash angle flow control valves in complex industrial scenarios. Attached Figure Description

[0045] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 A schematic flowchart of a flow field numerical simulation analysis method for a flash angle flow control valve provided in an embodiment of the present invention;

[0047] Figure 2 A flowchart illustrating a method for obtaining a reference history simulation, as provided in one embodiment of the present invention;

[0048] Figure 3 The diagram shows the structure of a flow field numerical simulation analysis system for a flash angle flow control valve according to an embodiment of the present invention.

[0049] Figure 4 This is a schematic diagram of a computer device provided according to an embodiment of the present invention. Detailed Implementation

[0050] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a flow field numerical simulation analysis method and system for a flash angle flow regulating valve proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0051] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0052] The following description, in conjunction with the accompanying drawings, details the specific scheme of the flow field numerical simulation analysis method and system for a flash angle flow regulating valve provided by the present invention.

[0053] Example 1:

[0054] This invention proposes a numerical simulation analysis method for the flow field of a flash angle flow control valve. Please refer to [link / reference]. Figure 1 The diagram illustrates a flow field numerical simulation analysis method for a flash angle flow control valve according to an embodiment of the present invention. The method includes the following steps:

[0055] Step S1: Obtain the initial watershed data, boundary conditions, mesh generation strategy and mesh generation parameters for each historical simulation, as well as the flow velocity and direction of the fluid in each unit region at each moment during each historical simulation.

[0056] Specifically, flash evaporation is known to generate extreme problems such as high temperature, high pressure, erosion, wear, corrosion, and thermal shock. To ensure the safety of industrial processes and angle flow control valves, it is necessary to deeply analyze the complex flow field characteristics inside the valve. Therefore, numerical simulation of the flow field of flash angle flow control valves is required. The conventional technical approach for numerical simulation of the internal flow field of angle flow control valves is to first establish a geometric model based on the valve's physical structure, then set the simulation boundary conditions and initial states based on measured or design data of the fluid at the valve inlet and outlet, and finally select a suitable fluid dynamics physical model (such as a turbulence model or a multiphase flow model) and determine the mesh generation parameters. This determines a mesh generation strategy to generate the computational mesh, which is the foundation for the entire simulation. However, when flash evaporation occurs in a throttling region (such as between the valve core and seat), a large number of tiny bubbles are generated, accompanied by a violent phase transition. This severely disturbs the original flow field structure, causing drastic fluctuations in velocity and pressure, and complex interphase interactions. This results in a highly unstable and heterogeneous flow field. Considering that the phase transition and bubble nucleation processes involved in flash evaporation occur at the microscale, the results of numerical simulations are extremely sensitive to the mesh generation method. The size, shape, density, and distribution of the mesh in key regions directly determine whether the initiation location of flash evaporation, the distribution of bubbles, and the evolution of the two-phase flow can be accurately captured. In numerical simulations, the computational mesh is the carrier of the physical field; therefore, the quality of the mesh generation—that is, whether it conforms to the requirements of the physical model and accurately represents the true gradient of the flow field—directly determines the accuracy of the simulation results.

[0057] Given that when the valve geometry and flow boundary conditions are fixed, a certain central tendency or statistical consensus exhibited by all simulation results after multiple simulations can be considered the best estimate approximating the actual flow field. Furthermore, a more precise mesh generation strategy naturally leads to more accurate simulation results. Therefore, simulation results that are closer to the central tendency and of higher quality in multiple simulations likely employ a superior mesh generation strategy. Combining these two points, this embodiment can identify high-quality simulations that converge to a consistent trend by analyzing and comparing a large number of historical numerical simulations. Then, it can reverse-engineer, summarize, and construct a preferred mesh generation strategy from these high-quality simulations, enabling more accurate flow field simulations under complex flash evaporation conditions.

[0058] Therefore, this embodiment acquires the initial watershed data, boundary conditions, mesh generation strategy, and mesh generation parameters for each historical simulation, as well as the flow velocity and direction of the fluid in each unit region at each moment during each historical simulation. The initial watershed data consists of preset values ​​for the fluid state within the entire watershed at the start of the corresponding historical simulation, including initial pressure, initial velocity, initial temperature, initial phase fraction, and initial turbulence parameters.

[0059] In one feasible implementation of this embodiment, the method for obtaining the unit area is as follows: using the structural drawings of the flash angle flow regulator, geometric modeling software is used to obtain the geometric model of the flash angle flow regulator valve. The geometric modeling technique is well-known and will not be elaborated further. A three-dimensional coordinate system is constructed with the center of the smallest circumcircle of the geometric model of the flash angle flow regulator valve as the origin. In this three-dimensional coordinate system, the X-axis represents the initial flow direction of the fluid entering the valve along the axis of the valve's bottom inlet; the Y-axis represents the lateral direction of the valve, i.e., the flow direction of the fluid at the side outlet; and the Z-axis represents the axis parallel to the valve core, i.e., the core flow direction of the fluid through the valve core. A preset distance is used as the unit length of the three-dimensional coordinate system. In this embodiment, the preset distance is set to 2mm. The implementer can set the size of the preset distance according to the actual situation, which is not limited here. All coordinate points in the three-dimensional coordinate system where all three dimensions are integers are used as reference points. A cube with a side length of unit length is constructed with each reference point as the center, and these are all used as reference areas. Then, the reference areas completely located within the internal flow domain of the flash angle flow regulator valve are all used as unit areas. The method for obtaining the minimum circumcircle is a well-known technique and will not be elaborated further.

[0060] It should be noted that, in order to better reduce the impact of flash evaporation, the flow channel inside the flash angle flow control valve in this embodiment is set to a downward arc flow channel to prevent the fluid from impacting the valve core and tie rod in a straight line, which would damage the valve core assembly. The valve core assembly is set to be an integral piece to reduce the valve core components from loosening, falling off, and being damaged due to vibration generated during high-pressure flash evaporation. The valve seat and assembly are set to be fitted with an inverted conical interlocking type and have a pre-tightening effect to prevent the fit between the metal valve body and the non-metal valve seat from loosening due to thermal expansion and contraction at high temperature and high pressure. At the same time, it is easier to install and disassemble.

[0061] Step S2: Based on the similarity of the initial data of the watershed between historical simulations, the historical simulations are divided into scenario categories. According to the flow velocity and direction of the fluid in each unit area at each moment during each historical simulation in each scenario category, the reference historical simulation in each scenario category is obtained.

[0062] Specifically, the final structure and evolution of the flow field inside the angle flow control valve in numerical simulations heavily depend on the initial flow field data at the start of the simulation. If two different historical simulations share a set of completely identical or highly similar initial flow field data, their simulation results (such as velocity distribution, pressure field, and gas phase distribution) should also tend to be consistent. Historical simulations with similar initial flow field data are essentially simulation scenarios that are similar. Therefore, this embodiment first divides historical simulations into scenario categories based on the similarity of the initial flow field data between historical simulations. Historical simulations within the same scenario category are considered to be in the same scenario by default. The more concentrated the historical simulations are within the same scenario category, the closer they are to the best estimate of the actual flow field. Therefore, this embodiment obtains a reference historical simulation, i.e., a high-quality historical simulation, within each scenario category based on the flow velocity and direction of the fluid in each unit area at each moment during each historical simulation process, to prepare for a more accurate simulation of the flow field of the flash angle flow control valve.

[0063] Preferably, in one feasible implementation of this embodiment, the method for obtaining the scene category is as follows: The initial watershed data from each historical simulation is constructed into a vector, serving as the initial data vector for each historical simulation; it should be noted that the dimensions of the initial data vectors are kept consistent, and the data types corresponding to elements at the same position in the initial data vectors must be the same. The cosine similarity of the initial data vectors from any two historical simulations is negatively correlated and normalized, and the result is used as the scene metric distance between any two historical simulations; this embodiment uses... The cosine similarity is negatively correlated and normalized, where x represents the cosine similarity of the initial data vectors, and norm is the normalization function. Based on scene distance measurement, historical simulations are clustered using the DBSCAN density clustering algorithm to obtain scene categories. In this embodiment, the minimum number of neighbors for the DBSCAN density clustering algorithm is set to 10, and the neighborhood radius is set to 0.2. Implementers can set the minimum number of neighbors and neighborhood radius for the DBSCAN density clustering algorithm according to actual conditions, and no limitation is imposed here. Both cosine similarity and the DBSCAN density clustering algorithm are well-known technologies and will not be described in detail here.

[0064] Preferably, in one possible implementation of this embodiment, the method for obtaining historical simulations can be found in [reference needed]. Figure 2 The document presents a flowchart of a method for obtaining a reference historical simulation, as provided in this embodiment. The method includes the following steps:

[0065] Step S201: Obtain the accuracy of the flow.

[0066] For any scenario category, the more consistent the flow velocity and direction of fluid in a certain unit area at a certain moment in a historical simulation within that scenario category are with the convergent trend of the flow velocity and direction of fluid in that unit area at that moment in all historical simulations within that scenario category, the more accurate the fluid flow in that unit area at that moment in the historical simulations is. Therefore, this embodiment obtains the flow accuracy of each unit area at each moment in each historical simulation within that scenario category based on the flow velocity and direction of fluid in each unit area at each moment in each historical simulation within that scenario category. The higher the flow accuracy, the more the flow in the corresponding unit area at the corresponding moment in the historical simulation conforms to the actual situation. It should be noted that the simulation process duration is the same for all historical simulations within that scenario category, meaning the number of moments corresponding to the simulation process is the same for all historical simulations within that scenario category.

[0067] In one possible implementation of this embodiment, the method for obtaining the flow accuracy is as follows: The flow velocity and direction of the fluid within each unit region at each moment in each historical simulation within the scenario category are constructed as a flow velocity vector. For any unit region and any moment, the average vector of the flow velocity vector of that unit region at that moment in each historical simulation within the scenario category is used as the reference vector (i.e., the vector corresponding to the central tendency) of that unit region at that moment in the historical simulations within the scenario category. For any historical simulation within the scenario category, the result of normalizing the cosine similarity between the flow velocity vector of that unit region at that moment in the historical simulation and the reference vector is used as the flow accuracy of that unit region at that moment in the historical simulation. In this embodiment, the cosine similarity between the flow velocity vector and the reference vector is normalized using a norm normalization function.

[0068] This allows us to obtain the accuracy of the flow in each unit area at each moment during each historical simulation within this scenario category.

[0069] Step S202: Obtain the importance of quality.

[0070] Within a given scenario category, if the flow accuracy of a certain unit region is higher at every moment in all historical simulations, it indicates that the quality of that unit region is better. Therefore, this embodiment obtains the quality importance of each unit region within the scenario category based on the flow accuracy. The higher the quality importance, the better the quality of the corresponding unit region within the scenario category.

[0071] In one possible implementation of this embodiment, the method for obtaining the quality importance is as follows: For any unit region and any historical simulation within the scene category, the average flow accuracy of the unit region at all times during the historical simulation is taken as the importance analysis value of the unit region in that historical simulation; then, the result of normalizing the mean of the importance analysis values ​​of the unit region in all historical simulations within the scene category is taken as the quality importance of the unit region within the scene category. In this embodiment, the mean of the above-mentioned importance analysis values ​​is normalized using the norm normalization function.

[0072] At this point, the quality importance of each unit area within this scene category is obtained.

[0073] Step S203: Obtain a quality score.

[0074] To accurately identify high-quality historical simulations within the scenario category and subsequently obtain the optimal network partitioning strategy, this embodiment obtains a quality score for each historical simulation within the scenario category based on flow accuracy and quality importance. A higher quality score indicates that the corresponding historical simulation is more likely to be a high-quality simulation that accurately reflects the real-world scenario.

[0075] In one possible implementation of this embodiment, the quality score is obtained as follows: For any unit region and any historical simulation within the scene category, the product of the importance analysis value of the unit region in the historical simulation and the quality importance of the unit region within the scene category is used as the participation analysis degree of the unit region in the historical simulation; the higher the participation analysis degree, the better the quality represented by the unit region in the historical simulation; then, the result of normalizing the mean of the participation analysis degrees of all unit regions in the historical simulation is used as the quality score of the historical simulation within the scene category. In this embodiment, the mean of the participation analysis degrees is normalized using the norm normalization function.

[0076] At this point, the quality score for each historical simulation within this scenario category is obtained.

[0077] Step S204: Obtain the reference historical simulation.

[0078] It is known that the higher the quality score, the more likely the corresponding historical simulation is to be a high-quality simulation that matches the real-world scenario. Therefore, this embodiment sets a preset quality score threshold of 0.7. Implementers can set the size of the preset quality score threshold according to the actual situation, and it is not limited here. When the quality score is greater than the preset quality score threshold, the corresponding historical simulation will be used as a reference historical simulation within that scenario category.

[0079] At this point, reference historical simulations for each scene category have been obtained.

[0080] Step S3: Based on the occurrence of the mesh partitioning strategy corresponding to the reference historical simulation in different scene categories, obtain the optimal mesh partitioning strategy; based on the boundary conditions and mesh partitioning parameters of the reference historical simulation corresponding to the optimal mesh partitioning strategy, obtain the trained convolutional neural network; input the boundary conditions of the current simulation into the trained convolutional neural network to obtain the reference mesh partitioning parameters of the current simulation.

[0081] Specifically, in actual operation, the inlet pressure, temperature, and other boundary conditions of angle flow control valves often fluctuate within a certain range, causing changes in the intensity and location of flashing. This inherent instability makes it difficult to accurately specify the simulation scenario for a new simulation task in advance, thus making it difficult to directly apply a meshing strategy for a specific scenario. A meshing strategy that performs exceptionally well in only one scenario but has significant errors in other scenarios has limited application value because it lacks generalization ability and cannot cope with the parameter fluctuations common in actual engineering. Conversely, if a meshing strategy can produce high-quality calculation results after multiple verifications in various differentiated scenarios, then the meshing strategy has strong adaptability and stability. Therefore, this embodiment obtains the optimal meshing strategy based on the occurrence of the corresponding meshing strategy in different scenario categories in reference historical simulations. This ensures that the selected optimal meshing strategy is not only accurate but also withstands the test of scenario changes, providing a more solid and reliable simulation foundation for the safe design and performance optimization of valves in real and varied working environments.

[0082] Preferably, in one feasible embodiment of this method, the optimal mesh partitioning strategy is obtained as follows: The mesh partitioning strategy corresponding to the reference historical simulation within each scene category is taken as a reference mesh partitioning strategy; for any reference mesh partitioning strategy, the scene category of the mesh partitioning strategy with reference historical simulation is taken as the reference scene category of the reference mesh partitioning strategy; the higher the proportion of the reference scene category among all scene categories, the stronger the adaptability and stability of the reference mesh partitioning strategy; the ratio of the number of reference scene categories to the total number of all scene categories is taken as the stability of the reference mesh partitioning strategy; the stability of each reference mesh partitioning strategy is then obtained; the higher the stability, the more likely the corresponding reference mesh partitioning strategy is to be the optimal mesh partitioning strategy; the reference mesh partitioning strategy corresponding to the highest stability is taken as the optimal mesh partitioning strategy.

[0083] It is known that the intensity and morphology of flash evaporation are not fixed but strongly depend on instantaneous fluid inflow parameters (such as pressure and temperature) and back pressure conditions on the outflow side. Any change in these boundary conditions will directly alter the gradient and range of pressure drop inside the angle flow control valve, thereby determining the starting position of flash evaporation, the degree of vaporization, and the spatial distribution of the two-phase flow. Furthermore, a mesh generation parameter that performs well under certain boundary conditions may no longer be optimal after the boundary conditions change; that is, an ideal mesh generation must be able to adaptively adjust with dynamic changes in boundary conditions. Therefore, this embodiment obtains a trained convolutional neural network based on the boundary conditions and mesh generation parameters of historical simulations corresponding to the optimal mesh generation strategy. Then, the boundary conditions of the current simulation are input into the trained convolutional neural network to adaptively obtain the reference mesh generation parameters for the current simulation, significantly improving the accuracy, efficiency, and robustness of the numerical simulation of the flash flow field.

[0084] Preferably, in one feasible embodiment of this method, the method for obtaining a trained convolutional neural network is as follows: all historical simulations using the optimal grid partitioning strategy are used as historical simulations for analysis; the boundary conditions of each historical simulation are used as the input to the convolutional neural network, and the output of the convolutional neural network is the grid partitioning parameters for each historical simulation; the convolutional neural network is then trained to obtain a trained convolutional neural network; wherein, an attention mechanism is introduced into the loss function of the convolutional neural network training, that is, the quality score of each historical simulation is used as the attention weight, making the model more focused on learning and imitating those grid parameter settings that have been verified as high-quality. The convolutional neural network and its training are well-known techniques and will not be described in detail here.

[0085] At this point, the optimal mesh generation strategy and reference mesh generation parameters for the current simulation have been determined.

[0086] Step S4: Determine the meshing of the geometric model of the flash angle flow control valve in the current simulation using the optimal meshing strategy and reference meshing parameters.

[0087] Specifically, after successfully obtaining the optimized mesh, the numerical simulation work enters the core stage of physics field setting and solution. In this embodiment, based on the different flow characteristics of the fluid within the angle regulating valve, the most suitable physics field model is set for each zone and model to ensure simulation accuracy. For example, the flow in the main flow domain of the valve channel is usually in a highly turbulent state; therefore, this embodiment selects the SST k-ω turbulence model because it can accurately capture the flow separation and shear stress in the near-wall region, making it very suitable for simulating the complex overall flow structure inside the valve. In key areas such as the valve core throttling region, which are the core areas of flash vaporization, an Eulerian-Eulerian multiphase flow model is set and coupled with a phase change model (such as a cavitation or flash model) to accurately describe the microscopic physical processes of the instantaneous vaporization of the liquid phase into the gas phase and the mixing and transport of the two phases. The SST k-ω turbulence model, the Eulerian-Eulerian multiphase flow model, and the phase change model are all well-known and will not be elaborated further.

[0088] Based on the actual measured material inflow and outflow data, the corresponding boundary conditions (such as inlet velocity / pressure, outlet pressure) and initial flow domain data are precisely set in each physical field model to drive the physical constraints of the entire simulation. After setting up, a pressure-based solver is selected, and a coupled algorithm is used for solving the problem to complete the numerical simulation of the flow field of the current flash angle flow control valve, obtaining ideal simulation analysis results and meeting the accuracy requirements of the flow field analysis of the flash angle flow control valve. It should be noted that the coupled algorithm can simultaneously solve the momentum equation and the continuity equation of the pressure basis, and has better convergence and computational efficiency for flash flows with significant compressibility changes. The coupled algorithm is a well-known technology and will not be described in detail here.

[0089] Finally, a 3D post-processing plotter is used to visualize the calculated flow field data (such as velocity cloud maps, pressure distribution, gas phase volume fraction, etc.), enabling intuitive analysis of the location, flow pattern, and key parameters of flashing within the valve, providing a direct basis for valve optimization design and safety assessment.

[0090] In summary, this embodiment acquires the initial watershed data from historical simulations, as well as the flow velocity and direction of the fluid during the historical simulations. Based on the similarity of the initial watershed data, the historical simulations are divided into scenario categories, and reference historical simulations are obtained according to the flow velocity and direction within each scenario category. Based on the occurrence of the meshing strategy corresponding to the reference historical simulations in different scenario categories, the optimal meshing strategy is obtained. Based on the boundary conditions and meshing parameters of the reference historical simulations corresponding to the optimal meshing strategy, reference meshing parameters are obtained, thereby determining the meshing of the geometric model of the flash angle flow control valve in the current simulation. This invention effectively improves the accuracy, efficiency, and robustness of numerical simulations of flash flow fields by acquiring the optimal meshing strategy and reference meshing parameters.

[0091] Example 2:

[0092] This invention also proposes a flow field numerical simulation analysis system for a flash angle flow control valve. Please refer to [link / reference]. Figure 3 The diagram shows a flow field numerical simulation analysis system for a flash angle flow regulating valve according to an embodiment of the present invention. The system includes: a data acquisition module 10, a reference history simulation acquisition module 20, a mesh generation related content acquisition module 30, and a mesh generation module 40.

[0093] The data acquisition module 10 is used to acquire the initial watershed data, boundary conditions, grid generation strategy and grid generation parameters for each historical simulation, as well as the flow velocity and direction of the fluid in each unit area at each moment during each historical simulation.

[0094] The reference historical simulation acquisition module 20 is used to divide historical simulations into scenario categories based on the similarity of the initial data of the watershed between historical simulations. Based on the flow velocity and direction of the fluid in each unit area at each moment during each historical simulation in each scenario category, the reference historical simulation in each scenario category is acquired.

[0095] The mesh partitioning-related content acquisition module 30 is used to obtain the optimal mesh partitioning strategy based on the occurrence of the mesh partitioning strategy corresponding to the reference historical simulation in different scene categories; to obtain a trained convolutional neural network based on the boundary conditions and mesh partitioning parameters of the reference historical simulation corresponding to the optimal mesh partitioning strategy; and to input the boundary conditions of the current simulation into the trained convolutional neural network to obtain the reference mesh partitioning parameters of the current simulation.

[0096] Mesh generation module 40 is used to determine the mesh generation of the geometric model of the flash angle flow control valve in the current simulation using the optimal mesh generation strategy and reference mesh generation parameters.

[0097] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the flow field numerical simulation analysis system for a flash angle flow regulating valve and the flow field numerical simulation analysis method embodiment for a flash angle flow regulating valve provided in the above embodiments belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0098] Example 3:

[0099] This invention also proposes a flow field numerical simulation analysis device for a flash angle flow control valve. The device includes a memory and a processor. The memory stores executable program code, and the processor calls and executes the executable program code to perform a flow field numerical simulation analysis method for a flash angle flow control valve provided in the embodiments of this application. Specifically, the device may be a chip, component, or module. The chip may include a connected processor and memory; the memory stores instructions, and when the processor calls and executes the instructions, the chip can perform the flow field numerical simulation analysis method for a flash angle flow control valve provided in the above embodiments.

[0100] In addition, this embodiment also protects a computer device; please refer to [link to relevant documentation]. Figure 4 The computer device includes a memory 401, a processor 402, and a computer program 403 stored in the memory 401 and running on the processor 402. When the processor 402 executes the computer program 403, the computer device can execute any of the aforementioned methods for numerical simulation analysis of the flow field of a flash angle flow control valve.

[0101] Example 4:

[0102] The present invention also provides a computer-readable storage medium storing computer program code, which, when executed on a computer, causes the computer to perform the aforementioned method steps to implement the flow field numerical simulation analysis method for a flash angle flow regulating valve provided in the above embodiments.

[0103] Example 5:

[0104] The present invention also provides a computer program product, which, when run on a computer, causes the computer to perform the above-mentioned related steps to realize the flow field numerical simulation analysis method for a flash angle flow regulating valve provided in the above embodiments.

[0105] In this embodiment, the device, computer-readable storage medium, computer program product, or chip are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0106] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0107] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A numerical simulation analysis method for the flow field of a flash angle flow control valve, characterized in that, The method includes the following steps: Obtain the initial watershed data, boundary conditions, mesh generation strategy and mesh generation parameters for each historical simulation, as well as the flow velocity and direction of the fluid in each unit region at each moment during each historical simulation. Based on the similarity of the initial watershed data between historical simulations, the historical simulations are divided into scenario categories. According to the flow velocity and direction of the fluid in each unit area at each moment during each historical simulation in each scenario category, a reference historical simulation in each scenario category is obtained. Based on the occurrence of the mesh partitioning strategy corresponding to the reference historical simulation in different scene categories, the optimal mesh partitioning strategy is obtained; based on the boundary conditions and mesh partitioning parameters of the reference historical simulation corresponding to the optimal mesh partitioning strategy, a trained convolutional neural network is obtained; the boundary conditions of the current simulation are input into the trained convolutional neural network to obtain the reference mesh partitioning parameters of the current simulation. The meshing of the geometric model of the flash angle flow control valve in the current simulation is determined by the optimal meshing strategy and reference meshing parameters; The method for obtaining the optimal grid partitioning strategy is as follows: The mesh partitioning strategy corresponding to the reference historical simulation within each scene category is used as the reference mesh partitioning strategy. For any reference mesh partitioning strategy, the mesh partitioning strategy with a reference history simulation is the scene category of the reference mesh partitioning strategy, and is used as the reference scene category of the reference mesh partitioning strategy; The ratio of the number of reference scene categories to the total number of all scene categories is used as the stability of the reference grid partitioning strategy. The reference meshing strategy corresponding to the highest stability is taken as the optimal meshing strategy.

2. The method of numerical simulation analysis of flow field of a flash angle type flow regulating valve according to claim 1, characterized in that, The method for obtaining the scene category is as follows: The initial watershed data for each historical simulation is constructed into a vector, which serves as the initial data vector for each historical simulation. The cosine similarity of the initial data vectors of any two historical simulations is negatively correlated and normalized, and used as the scene distance metric between any two historical simulations. Based on scene metric distance, historical simulations are clustered using the DBSCAN density clustering algorithm to obtain scene categories.

3. The method of numerical simulation analysis of flow field of a flash angle type flow regulating valve according to claim 1, characterized in that, The method for obtaining the reference historical simulation is as follows: For any scenario category, based on the flow velocity and direction of the fluid in each unit area at each moment during each historical simulation within that scenario category, the flow accuracy of each unit area at each moment during each historical simulation within that scenario category is obtained; Based on the accuracy of the flow, the quality importance of each unit area within the scene category is obtained; Based on the accuracy of the flow and the importance of the quality, obtain the quality score for each historical simulation within this scenario category; When the quality score is greater than the preset quality score threshold, the corresponding historical simulation will be used as the reference historical simulation within that scene category.

4. The method of numerical simulation analysis of flow field of a flash angle type flow regulating valve according to claim 3, characterized in that, The method for obtaining the accuracy of the flow is as follows: The flow velocity and direction of the fluid in each unit area at each moment during each historical simulation in this scenario category are constructed as a flow velocity vector; For any unit region and any time, the average vector of the flow velocity vector of the unit region at that time in each historical simulation within the scene category is used as the reference vector of the unit region at that time in the historical simulation within the scene category. For any historical simulation within this scenario category, the result of normalizing the cosine similarity between the flow velocity vector of the unit region at that moment in the historical simulation and the reference vector is taken as the accuracy of the flow in the unit region at that moment in the historical simulation.

5. The flow field numerical simulation analysis method for a flash angle flow regulating valve as described in claim 3, characterized in that, The method for obtaining the importance of the quality is as follows: For any unit region and any historical simulation within that scenario category, the average flow accuracy of that unit region at all times during the historical simulation is taken as the important analytical value of that unit region in the historical simulation. The result of normalizing the mean of the important analytical values ​​of the unit region within the scene category across all historical simulations is used as the quality importance of the unit region within the scene category.

6. The method of numerical simulation analysis of flow field of a flash angle type flow regulating valve according to claim 5, characterized in that, The method for obtaining the quality score is as follows: For any unit region and any historical simulation within that scene category, the product of the important analysis value of that unit region in that historical simulation and the quality importance of that unit region within that scene category is taken as the degree of participation of that unit region in the historical simulation. The result of normalizing the mean of the participation of all unit regions in the historical simulation is used as the quality score of the historical simulation within the scenario category.

7. The method of numerical simulation analysis of flow field of a flash angle type flow regulating valve according to claim 3, characterized in that, The method for obtaining the trained convolutional neural network is as follows: All historical simulations using the optimal mesh partitioning strategy will be used as historical simulations for analysis. The boundary conditions of each historical simulation are used as input to the convolutional neural network (CNN), and the output of the CNN is the grid partitioning parameters of each historical simulation. The CNN is trained to obtain a trained CNN. In the loss function of the CNN training, the quality score of each historical simulation is used as the attention weight.

8. The method of numerical simulation of flow field of a flash angle type flow regulating valve according to claim 1, wherein The method for obtaining the unit region is as follows: A three-dimensional coordinate system is constructed with the center of the smallest circumcircle of the geometric model of the flash angle flow control valve as the origin; where the X-axis in the three-dimensional coordinate system is the axial direction along the bottom inlet of the valve, that is, the initial flow direction of the fluid entering the valve, the Y-axis is the lateral direction of the valve, that is, the flow direction of the fluid at the side outlet, and the Z-axis is the axial direction parallel to the valve core, that is, the core flow direction of the fluid through the valve core. The preset distance is used as the unit length of the three-dimensional coordinate system. All coordinate points in the three-dimensional coordinate system with integers in all three dimensions are used as reference points. A cube with a side length of unit length is constructed with each reference point as the center, and all of them are used as reference areas. The reference area that is entirely located within the flow domain of the flash angle flow control valve is taken as the unit area.

9. A flow field numerical simulation analysis system for a flash angle flow control valve, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the flow field numerical simulation analysis method for a flash angle flow regulating valve as described in any one of claims 1-8.

Images