A three-dimensional modeling optimization method and system for a hull fairing

Abstract

The present invention provides a three-dimensional modeling optimization method and system for a hull fairing, and relates to the technical field of three-dimensional modeling of a hull fairing. The method comprises: defining a virtual calculation domain including the hull fairing and its surrounding environment according to a preliminary hull fairing model; setting boundary conditions, dividing the calculation domain into a plurality of grid units, and generating boundary layer grids on the hull surface; performing numerical simulation, setting solution parameters, including time step, convergence criterion and discrete format; and calculating the resistance performance of the hull fairing in water, including friction resistance and pressure difference resistance, through numerical simulation. The present invention improves the accuracy of the design.

Description

Technical Field

[0001] The invention relates to the technical field of three-dimensional modeling of a hull fairing, and in particular to a three-dimensional modeling optimization method and system for a hull fairing. Background Art

[0002] As an important part of ship design, the main function of the hull fairing is to optimize water flow, reduce navigation resistance, and prevent the generation of bubbles, thereby improving navigation efficiency and stability. However, some traditional hull fairing design methods usually design the shape of the fairing based on the basic parameters and functional requirements of the hull, combined with their own experience and intuition. Although this method is simple and intuitive, it has some defects.

[0003] First, due to the lack of precise theoretical analysis and numerical simulation, traditional designs sometimes fail to achieve optimal resistance reduction and bubble prevention effects. Second, the deflector designed by traditional methods may produce adverse flow phenomena in actual navigation, such as flow separation and vortexes, which will increase navigation resistance and reduce navigation efficiency. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a three-dimensional modeling optimization method and system for a hull fairing, thereby improving the design accuracy.

[0005] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0006] In a first aspect, a three-dimensional modeling optimization method for a hull fairing is provided, the method comprising:

[0007] Determine the functional objectives of the fairing according to the purpose, sailing speed and load of the ship;

[0008] Obtain the specific parameters of the hull, including hull length, width, draft, hull line shape and hull surface curvature;

[0009] Create a preliminary hull shroud model based on the shroud's functional objectives and the hull's specific parameters;

[0010] Based on the preliminary hull deflector model, a virtual computational domain including the hull deflector and its surrounding environment is defined;

[0011] Set boundary conditions, divide the computational domain into multiple grid cells, and generate boundary layer grids on the hull surface;

[0012] Perform numerical simulation and set solution parameters, including time step, convergence criterion and discrete format; calculate the resistance performance of the hull fairing in water, including friction resistance and pressure difference resistance, through numerical simulation; analyze the pressure distribution on the surface of the hull fairing and identify high-pressure and low-pressure areas; evaluate the anti-bubble performance of the fairing by analyzing the pressure distribution and streamline distribution around the hull;

[0013] Through the resistance performance and anti-bubble performance, the streamline distribution near the hull is analyzed to identify the flow separation and vortex generation phenomena;

[0014] According to the flow separation and vortex generation phenomena, as well as the high-pressure and low-pressure areas, the shape of the hull fairing is adjusted. The parameters to be adjusted include the curvature, length, width and thickness of the fairing.

[0015] The particle swarm algorithm is used to optimize the shape of the fairing to minimize the drag coefficient and maximize the anti-bubble performance. The shape of the fairing is gradually optimized through multiple iterations. After each iteration, numerical simulation and performance evaluation are performed again until the predetermined optimization goal is achieved to output the final three-dimensional model of the hull fairing.

[0016] Furthermore, the computational domain is divided into multiple grid cells, including:

[0017] Receive the computational domain geometry information provided by the user, including CAD models and point cloud data;

[0018] Preprocess the geometric information and define the global parameters for meshing;

[0019] The entire computational domain is divided into multiple sub-regions according to the geometric features and expected flow characteristics. Each sub-region will be processed independently to meet different mesh generation requirements.

[0020] In each sub-region, an adaptive subdivision strategy is applied to dynamically determine the mesh density of each region based on the geometric complexity, flow characteristics changes, and user-defined global parameters;

[0021] For each subdivided sub-region, a smooth and evenly distributed grid point is generated in the sub-region by solving a specific elliptic equation;

[0022] The generated mesh is quality checked to evaluate the volume ratio, distortion and aspect ratio of each mesh unit. For mesh units that do not meet the quality standards, they are optimized by re-dividing the mesh lines or adjusting the node positions. The optimization process is repeated until the quality of all mesh units meets the preset standards to output the final mesh data, which includes the coordinates and connection relationships of the mesh points.

[0023] Furthermore, the entire computational domain is divided into multiple sub-regions according to the geometric features and expected flow characteristics, including:

[0024] According to the distribution of geometric data points, a KD-Tree data structure is constructed;

[0025] Traverse the KD-Tree data structure and divide the space into a series of voxels according to the set voxel size. Each voxel contains a part of geometric data points. In the voxel segmentation process, the voxels containing geometric data points and adjacent to the external space of the computational domain are marked as boundary voxels.

[0026] Determine the boundary of the voxel, and query the data points related to the boundary through KD-Tree to obtain the divided voxel information, which includes the position, size and geometric data points of each voxel; for each voxel, extract its internal geometric features, including point density, surface curvature and normal vector distribution;

[0027] The K-means clustering algorithm is used to cluster the voxels and obtain preliminary clustering results. Each voxel is regarded as a node of the graph. If two voxels are adjacent, an edge is added between the two voxels.

[0028] Define an energy function, which includes a data term and a smoothing term. The data term measures the cost of assigning each voxel to a subregion, and the smoothing term measures the cost of inconsistent labels between adjacent voxels.

[0029] Use the graph cut algorithm to minimize the energy function to redistribute voxels to different sub-regions to obtain the graph cut optimized result;

[0030] According to the results of graph cut optimization, voxels belonging to the same category are merged into a sub-region to obtain the optimized sub-region division;

[0031] According to the optimized sub-region division, the boundary information of each sub-region is updated to output the divided sub-region information, including the boundary of each sub-region, the included voxel list and the corresponding geometric features and flow characteristics.

[0032] Furthermore, the voxel boundaries are determined, including:

[0033] Randomly select four non-coplanar points from the voxel as the vertices of the initial convex hull, and construct an initial tetrahedron as the convex hull;

[0034] Calculate the adjustment factor based on the data point distribution; for each remaining data point within a voxel, perform the following steps:

[0035] Each face of the convex hull is expanded outward along its normal direction by the length of the adjustment factor to form an expanded convex hull, and whether the current point is located inside the expanded convex hull is determined. If the current point is located outside the expanded convex hull, the convex hull is updated, including finding visible faces, deleting visible faces and adding new faces. After adding new faces, the convex hull optimization operation is performed, and the optimization is continued until all remaining points are processed to obtain the final convex polyhedron;

[0036] Boundary information, including vertices, edges, and faces, is extracted from the final convex polyhedron, and the boundary information is used to describe the adjusted voxel boundaries.

[0037] Furthermore, the adjustment factor The calculation formula is:

[0038] ;

[0039] in, is the standard deviation of the data point density in the local area around the current point; is the local density gradient, indicating how fast the density changes; is the maximum density value in the local area; is the average density value in the local area; is the standard deviation of the local density; is a Gaussian function; is the average radius of curvature in the local area around the current point; is the local curvature gradient, which indicates how fast the curvature changes; is the maximum curvature value in the local area; is the curvature direction angle between the current point and the adjacent point; is the number of neighboring points; is the index value of the neighboring point; , and is the weight coefficient; is the density value of data points in the local area around the current point;

[0040] ;

[0041] in, and are the maximum and minimum pressures in the local area around the current point, respectively; and are the highest and lowest temperatures in the local area around the current point, respectively; is the velocity vector of the current point; is the maximum velocity value in the local area.

[0042] Furthermore, the boundary conditions include inlet boundary, outlet boundary, side boundary and bottom boundary. The inlet boundary is set to the incoming flow direction to simulate the situation of water flow entering the calculation domain. It is set as the velocity inlet to give the velocity and direction of the water flow. The outlet boundary represents the outflow direction of the water flow and is set as the pressure outlet to allow the water to flow out freely. The side boundary is set as the symmetric boundary to simulate the infinitely far water area. The bottom boundary is set as the wall boundary to simulate the influence of the bottom of the water.

[0043] Furthermore, global parameters include overall size, target resolution, maximum and minimum cell sizes.

[0044] In a second aspect, a three-dimensional modeling optimization system for a hull fairing is provided, comprising:

[0045] A determination module, used for determining the functional target of the fairing according to the purpose, sailing speed and load of the ship;

[0046] An acquisition module is used to acquire specific parameters of the hull, including the hull length, width, draft, hull line shape and hull surface curvature;

[0047] A creation module is used to create a preliminary hull shroud model according to the functional objectives of the shroud and the specific parameters of the hull;

[0048] A definition module is used to define a virtual computational domain including the hull deflector and its surrounding environment based on the preliminary hull deflector model; set boundary conditions, divide the computational domain into multiple grid cells, and generate boundary layer grids on the hull surface;

[0049] The simulation module is used to perform numerical simulation and set solution parameters, including time step, convergence criterion and discrete format; calculate the resistance performance of the hull fairing in water through numerical simulation, including friction resistance and pressure difference resistance; analyze the pressure distribution on the surface of the hull fairing and identify high-pressure and low-pressure areas; evaluate the anti-bubble performance of the fairing by analyzing the pressure distribution and streamline distribution around the hull; analyze the streamline distribution near the hull through the resistance performance and anti-bubble performance, and identify the flow separation and vortex generation phenomena;

[0050] An adjustment module is used to adjust the shape of the hull fairing according to flow separation and vortex generation phenomena, as well as high-pressure and low-pressure areas. The adjustment parameters include the curvature, length, width and thickness of the fairing;

[0051] The optimization module is used to optimize the shape of the fairing using a particle swarm algorithm to minimize the drag coefficient and maximize the anti-bubble performance. The shape of the fairing is gradually optimized through multiple iterations. After each iteration, numerical simulation and performance evaluation are performed again until the predetermined optimization goal is achieved to output the final three-dimensional model of the hull fairing.

[0052] According to a third aspect, a computing device includes:

[0053] one or more processors;

[0054] The storage device is used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method described.

[0055] In a fourth aspect, a computer-readable storage medium stores a program, and when the program is executed by a processor, the method described is implemented.

[0056] The above solution of the present invention includes at least the following beneficial effects:

[0057] By determining the functional objectives of the fairing in combination with the specific purpose, sailing speed and load capacity of the hull, and by obtaining detailed parameters of the hull to create a preliminary fairing model, the design is more in line with actual needs and the design accuracy is improved.

[0058] By analyzing the streamline distribution near the hull and the pressure distribution on the surface of the hull fairing, flow separation, vortex generation, and high-pressure and low-pressure areas can be accurately identified. Based on the analysis results of flow separation, vortex and pressure distribution, the shape parameters of the fairing, including curvature, length, width and thickness, can be flexibly adjusted to achieve better performance.

[0059] The particle swarm algorithm is used to optimize the shape of the fairing, and the optimal solution is gradually approached through multiple iterations. This intelligent optimization method not only improves the design efficiency, but also ensures that the final output three-dimensional model achieves the best drag coefficient and anti-bubble performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a flow chart of a three-dimensional modeling optimization method of a hull fairing provided in an embodiment of the present invention.

[0061] Figure 2 It is a flow chart of dividing a calculation domain into a plurality of grid units in a three-dimensional modeling optimization method of a hull fairing provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0063] like Figure 1 and Figure 2 As shown, an embodiment of the present invention provides a three-dimensional modeling optimization method for a hull fairing, the method comprising the following steps:

[0064] Step 100, determining the functional target of the fairing according to the purpose, sailing speed and load of the ship;

[0065] Step 101, obtaining specific parameters of the hull, including hull length, width, draft, hull line shape and hull surface curvature;

[0066] Step 102, creating a preliminary hull shroud model according to the functional objectives of the shroud and specific parameters of the hull;

[0067] Step 103, based on the preliminary hull deflector model, define a virtual computational domain including the hull deflector and its surrounding environment;

[0068] Step 104, setting boundary conditions, dividing the computational domain into multiple grid units, generating boundary layer grids on the hull surface, the boundary conditions include inlet boundary, outlet boundary, side boundary and bottom boundary, the inlet boundary is set as the incoming flow direction, simulating the situation of water flow entering the computational domain, set as the velocity inlet, giving the velocity and direction of the water flow; the outlet boundary represents the outflow direction of the water flow, set as the pressure outlet, allowing the water flow to flow out freely; the side boundary is set as the symmetric boundary, simulating the infinitely far water area; the bottom boundary is set as the wall boundary, simulating the influence of the bottom;

[0069] Step 105, using the SST k-ω turbulence model for numerical simulation, selecting the ANSYS Fluent solver, and setting the solution parameters, including the time step, the convergence criterion, and the discrete format; calculating the resistance performance of the hull fairing in water through numerical simulation, including the friction resistance and the pressure difference resistance; analyzing the pressure distribution on the surface of the hull fairing, identifying the high pressure area and the low pressure area; and evaluating the anti-bubble performance of the fairing by analyzing the pressure distribution and streamline distribution around the hull;

[0070] Step 106, analyzing the streamline distribution near the hull through the resistance performance and anti-bubble performance, and identifying the flow separation and vortex generation phenomena;

[0071] Step 107, adjusting the shape of the hull fairing according to the flow separation and vortex generation phenomena, as well as the high pressure area and the low pressure area, wherein the adjustment parameters include the curvature, length, width and thickness of the fairing;

[0072] Step 108, using a particle swarm algorithm to optimize the shape of the fairing to minimize the drag coefficient and maximize the anti-bubble performance, and gradually optimizing the shape of the fairing through multiple iterations. After each iteration, numerical simulation and performance evaluation are performed again until the predetermined optimization goal is achieved, so as to output the final three-dimensional model of the hull fairing.

[0073] In an embodiment of the present invention, by combining numerical simulation and optimization algorithm, the performance of the fairing can be predicted and evaluated at the early stage of design, thereby reducing the time of repeated experiments and modifications in traditional design methods and significantly improving the design efficiency. Through the numerical simulation of the SST k-ω turbulence model, the resistance performance of the hull fairing in the water, including friction resistance and pressure difference resistance, can be accurately calculated. By optimizing the shape, the resistance of the hull in the water can be effectively reduced, thereby improving the navigation speed and fuel efficiency. The method can analyze the pressure distribution and streamline distribution around the hull, evaluate the anti-bubble performance of the fairing, and through shape optimization, reduce the generation and attachment of bubbles, protect the hull from bubble corrosion and damage, and improve the service life and safety of the hull. By identifying the flow separation and vortex generation phenomena, as well as the high pressure area and the low pressure area, the shape of the hull fairing can be finely adjusted. This adjustment not only takes into account the overall performance requirements, but also takes into account the local flow characteristics, making the design more accurate and effective. The particle swarm algorithm is used to optimize the shape of the fairing, which can find the optimal solution among many possible design schemes. This optimization method not only considers a single performance indicator, but can also handle multi-objective optimization problems, such as minimizing the drag coefficient and maximizing the anti-bubble performance at the same time, thereby comprehensively improving the design quality.

[0074] In another preferred embodiment of the present invention, the above step 100 analyzes the main purpose of the hull, such as cargo transportation, passenger transportation or scientific research exploration; considers the sailing speed requirement of the hull, high-speed sailing requires a more efficient diversion design to reduce resistance; evaluates the load capacity of the hull, heavy load will affect the stability and navigation efficiency of the hull, and a fairing is required to provide corresponding support. Based on the above factors, determine the functional goals that the fairing needs to achieve, such as reducing navigation resistance, improving navigation stability, and preventing water flow from impacting the hull.

[0075] In another preferred embodiment of the present invention, in the above step 101, the design drawings of the hull are obtained, and these drawings include a side view, a top view and a cross-sectional view of the hull, etc.; by measuring the dimensions on the design drawings, basic parameters such as the length (the maximum distance from the bow to the stern), the width (the dimension of the widest part of the hull) and the draft (the immersion depth of the hull in the water) of the hull can be directly obtained; by analyzing the design drawings, the overall outline and streamlined design of the hull can be understood.

[0076] Analyze the shape of the bottom of the ship. The bottom may be flat, V-shaped, or other complex shapes, each of which has its own specific sailing characteristics. At the same time, observe the design of the bow and stern. The bow is designed to be sharp or blunt to reduce water resistance; while the stern contains structures such as propellers or rudders to control heading and speed. Analyze the specific shape of the bottom of the ship. For example, a V-shaped bottom helps stability when sailing at high speeds, while a flat-bottomed boat is more suitable for sailing in shallow waters. Analyze the shape of the bow. A sharp bow can better cut through the water surface, reduce resistance, and is suitable for high-speed sailing; while a blunt bow may provide better stability and cargo space; Scanning the actual hull or hull model using 3D scanning technology can quickly and accurately capture the geometry of the hull surface. By scanning the data, an accurate 3D model of the hull surface can be generated, and the curvature information of the surface can be extracted from this model.

[0077] In another preferred embodiment of the present invention, in the above step 102, three-dimensional modeling software (such as SolidWorks, AutoCAD, etc.) is used to start building a preliminary model of the fairing according to the functional goals determined in step 100 and the hull parameters obtained in step 101. During the modeling process, the connection between the fairing and the hull is considered to ensure that the fairing can fit closely to the surface of the hull. According to design experience and principles of fluid mechanics, the shape, curvature and size of the fairing are preliminarily set to achieve the predetermined functional goals.

[0078] In another preferred embodiment of the present invention, in the above step 103, in the three-dimensional modeling software, a virtual environment including the hull and the fairing is created based on the preliminary fairing model and the hull model, the boundary of the calculation domain is defined to ensure that the calculation domain is large enough to simulate the real water flow environment while avoiding the influence of boundary effects on the simulation results, and appropriate water inlet, outlet, side and bottom boundary conditions are set in the calculation domain to simulate the real water flow conditions, and the calculation domain is meshed, especially generating a fine boundary layer mesh on the surface of the hull and the fairing to improve the accuracy of the numerical simulation.

[0079] In another preferred embodiment of the present invention, in the above step 104, boundary conditions are set:

[0080] Inlet Boundary:

[0081] a. Determine the direction of the incoming current, which is usually based on the direction the vessel is expected to travel.

[0082] b. Set as velocity inlet, which means a fixed water velocity value and direction will be given on this boundary. This velocity and direction should reflect the water flow conditions that the hull may encounter in actual waters.

[0083] c. Make sure the inlet boundary is far enough to allow the incoming flow to fully develop before reaching the hull to avoid the inlet effect affecting the simulation results.

[0084] Export Boundary:

[0085] a. The outlet boundary represents where the water flows out of the computational domain.

[0086] b. Set as a pressure outlet, which means that at this boundary, the pressure is known (usually atmospheric pressure) and water is allowed to flow out freely without backflow.

[0087] c. The outlet boundary should also be far enough away to ensure that the flow field is fully developed and not affected by the outlet boundary conditions before flowing out of the computational domain.

[0088] Side borders:

[0089] a. The side boundaries are usually set as symmetry boundaries, which means that in the calculation, the flow field is assumed to be symmetric on both sides of the symmetry plane.

[0090] b. This setting is used to simulate waters of infinite distance, that is, when the hull is sailing in wide waters, the water flow away from the hull is almost unaffected by the existence of the hull.

[0091] Bottom Boundary:

[0092] a. The bottom boundary is set as the wall boundary to simulate the influence of the bottom on the water flow.

[0093] b. At this boundary, the water velocity is usually set to zero, indicating that the water cannot penetrate the bottom.

[0094] c. Depending on the actual water depth and bottom conditions (such as roughness, silt deposition, etc.), additional friction or slip conditions need to be applied on the bottom boundary.

[0095] In a preferred embodiment of the present invention, the computational domain is divided into a plurality of grid units, including:

[0096] Receive the computational domain geometry information provided by the user, including CAD models and point cloud data;

[0097] Preprocess the geometric information and define the global parameters of mesh division, which include overall size, target resolution, maximum and minimum unit size. Specifically, preprocess the received geometric information, including cleaning invalid data, filling holes, smoothing surfaces, etc., to ensure the integrity and accuracy of the geometric information, and define the global parameters of mesh division, such as overall size (determines the overall range of mesh coverage), target resolution (the fineness of the mesh), maximum and minimum unit size (controls the size range of mesh units).

[0098] According to the geometric features and expected flow characteristics, the entire computational domain is divided into multiple sub-regions, and each sub-region will be processed independently to adapt to different mesh generation requirements, including: According to the geometric features of the computational domain and the expected flow characteristics, the entire computational domain is divided into multiple sub-regions. For example, in the flow field simulation around the hull, the area near the hull and away from the hull may be divided into different sub-regions, and each sub-region will be processed independently according to its specific geometry and flow characteristics to better adapt to different mesh generation requirements.

[0099] In each sub-region, an adaptive subdivision strategy is applied to dynamically determine the mesh density of each region based on the geometric complexity, changes in flow characteristics, and user-defined global parameters. Specifically, in each sub-region, an adaptive subdivision strategy is applied based on the geometric complexity (such as changes in surface curvature), changes in flow characteristics (such as velocity gradients, pressure changes), and global parameters previously defined by the user. This strategy can dynamically determine the mesh density of each sub-region, that is, generate finer meshes in areas with complex geometry or large changes in flow characteristics, and use coarser meshes in relatively simple areas.

[0100] For each subdivided sub-region, a smooth and evenly distributed grid point is generated in the sub-region by solving a specific elliptic equation, specifically including: for each subdivided sub-region, a specific elliptic equation (partial differential equation, used for smoothing in grid generation) is solved to generate grid points. This process ensures that the grid points generated in the sub-region are smooth and evenly distributed, which helps to improve the accuracy and stability of numerical simulation. Among them, the elliptic equation is a type of partial differential equation used to describe the steady-state behavior of physical phenomena. In grid generation, the elliptic equation includes the Poisson equation or the Laplace equation, which can ensure that the generated grid points are smooth and evenly distributed. For example, the Laplace equation can be expressed as:

[0101] ;

[0102] in, is the potential function to be solved, is the Laplacian operator.

[0103] The generated mesh is quality checked, and the volume ratio, distortion and aspect ratio of each mesh unit are evaluated. For mesh units that do not meet the quality standards, they are optimized by re-dividing the mesh lines or adjusting the node positions. The optimization process is repeated until the quality of all mesh units meets the preset standards to output the final mesh data. The final mesh data includes the coordinates and connection relationships of the mesh points. Specifically, the generated mesh is quality checked, and the evaluation indicators include the volume ratio of the mesh unit (the ratio of the unit volume to the ideal volume), distortion (the degree of distortion of the unit shape) and aspect ratio (the ratio of the longest side to the shortest side of the unit), etc. For mesh units that do not meet the quality standards, they are optimized by re-dividing the mesh lines or adjusting the node positions. This optimization process will be repeated many times until the quality of all mesh units meets the preset standards. When the quality of all mesh units meets the requirements, the final mesh data is output. This data includes the coordinates of each mesh point and the connection relationship between them (that is, which points constitute a mesh unit). This data will be used in subsequent numerical simulations to solve the changes in flow fields, temperature fields or other physical fields.

[0104] In an embodiment of the present invention, by receiving the computational domain geometry information (such as CAD model and point cloud data) provided by the user, the method can adapt to geometric bodies of different shapes and complexities, thereby improving the flexibility and versatility of mesh generation. The geometric information is preprocessed and global parameters are defined to make the meshing process more controllable. These global parameters (such as overall size, target resolution, maximum and minimum cell size) allow the user to adjust the overall characteristics and detail level of the mesh according to the simulation requirements. The entire computational domain is divided into multiple sub-regions, and each sub-region is processed independently according to the geometric characteristics and expected flow characteristics. This division method can better capture the flow details of different regions and improve the accuracy of the simulation. An adaptive subdivision strategy is applied to dynamically determine the mesh density of each region. This strategy can generate denser meshes in regions with high geometric complexity or drastic changes in flow characteristics, and use coarser meshes in relatively simple regions, thereby achieving high-precision simulation while optimizing the use of computing resources. By solving specific elliptical equations, smooth and evenly distributed mesh points are generated in the sub-regions. This method can generate high-quality meshes and reduce errors and instabilities in numerical simulations. The generated mesh is quality checked and substandard mesh units are optimized. This iterative optimization process ensures that the quality of all mesh units meets the preset standards, thereby improving the reliability and accuracy of the simulation results. The final mesh data, including the coordinates and connection relationships of the mesh points, is output to facilitate the subsequent reading and processing of numerical simulation software. This standardized data format helps achieve compatibility and data exchange between different software.

[0105] In a preferred embodiment of the present invention, the entire computational domain is divided into a plurality of sub-regions according to geometric features and expected flow characteristics, including:

[0106] According to the distribution of geometric data points, a KD-Tree data structure is constructed, which specifically includes: collecting all geometric data points, and using these data points to construct a KD-Tree (k-dimensional tree), which is a data structure used to organize points in k-dimensional space. KD-Tree is constructed by recursively dividing the data space into two subspaces, and each node corresponds to a hyper-rectangular area in the k-dimensional space partition.

[0107] Traverse the KD-Tree data structure, divide the space into a series of voxels according to the set voxel size, each voxel contains a part of geometric data points, and in the voxel segmentation process, mark the voxels containing geometric data points and adjacent to the external space of the computational domain as boundary voxels, specifically including: setting a voxel size, which is the basic unit for dividing the space, traverse the KD-Tree, divide the space into a series of cubes (voxels) according to the voxel size, and in the traversal process, record the geometric data points contained in each voxel, if a voxel contains geometric data points and is adjacent to the external space of the computational domain, it will be marked as a boundary voxel.

[0108] Determine the boundary of the voxel, query the data points related to the boundary through KD-Tree to obtain the voxel information after division, which includes the position, size and geometric data points of each voxel; for each voxel, extract its internal geometric features, including point density, surface curvature and normal vector distribution, specifically including: determine the boundary of the voxel, extract the internal geometric features of each voxel, such as point density (the number of points per unit volume), surface curvature (by fitting the local surface and calculating its curvature) and normal vector distribution (obtained by covariance analysis of the local point set).

[0109] The K-means clustering algorithm is used to cluster the voxels to obtain preliminary clustering results. Each voxel is regarded as a node of the graph. If two voxels are adjacent, an edge is added between the two voxels. Specifically, the K-means clustering algorithm is used to cluster the voxels, where K is the expected number of sub-regions, the cluster centers are initialized, and each voxel is assigned to the nearest cluster center, the cluster centers are iteratively updated, and the voxels are redistributed until the clustering results are stable or the maximum number of iterations is reached, and a preliminary clustering result, i.e., the preliminary assignment of voxels to sub-regions, is obtained.

[0110] Define an energy function, which includes a data term and a smoothing term. The data term measures the cost of assigning each voxel to a sub-region, and the smoothing term measures the cost of inconsistent labels between adjacent voxels. Specifically, define an energy function, which includes two main parts: a data term and a smoothing term. The data term measures the cost of assigning each voxel to a specific sub-region, usually based on the similarity between the geometric features of the voxel and the features of the sub-region. The smoothing term measures the cost of inconsistent labels between adjacent voxels to encourage adjacent voxels to be assigned to the same sub-region. The calculation formula of the energy function is:

[0111] ;

[0112] in, Representing voxels To which subregion is it assigned? It is a discrete variable with values ​​of {1, 2, ..., K}, where K is the number of subregions, which is the variable we hope to optimize. It represents the subregion label to which each voxel belongs. By selecting the appropriate label, the total energy is minimized. N represents the total number of voxels in the image or volume data. This is a constant used to calculate the sum of the data item and the total energy. K represents the number of subregions in the segmentation problem (for example, subregions can be different parts of the segmentation, different classes, etc.). This is a constant that determines how many different subregions a voxel can be assigned to. is an indicator function, when the voxel Tags Equal to sub-area hour, , otherwise 0, this indicator function ensures that only voxels are counted The corresponding label is The data term cost of , which ensures that the energy term only takes effect when the voxel labels match; It is a voxel Assign to sub-area The cost of is the distance between the voxel features (such as color, texture, etc.) and the sub-region features. This term measures the voxel With sub-area The similarity between them is usually measured using the Euclidean distance metric to measure the difference between voxels and sub-regions. The goal of the data item is to assign the voxel to the sub-region that is most similar to it; is a measure of the neighboring voxels and The label inconsistency cost between depends on whether the voxel labels are the same. and When the labels of are different (that is, they are assigned to different sub-regions), the penalty increases. This term is to promote spatial smoothness and prevent too many incoherent regions from appearing in the segmentation. Represents voxels in images or volume data and are adjacent, that is, there is an edge connecting them. This is an index indicating which voxels are adjacent. Each edge in the graph represents a pair of adjacent voxels, the label consistency of these adjacent voxels will affect the smoothing term; It is the total energy function of the entire image or volume data; and is the index of the voxel, representing the different voxels in the image or volume data. For a volume data set or image, all voxels (pixels or volume elements) have a unique number. In the formula, and is the index between voxels, used to refer to the specific voxel position, for example, may represent the first voxel, represents the second voxel, and so on; E is an edge set that represents the adjacency relationship between voxels in the image or volume data; The label of the sub-region is a discrete integer value.

[0113] ;

[0114] in, It is a voxel and The weight between them indicates their similarity, which is usually defined based on features such as spatial distance, color, and texture; is the indicator function, when the voxel and Tags and If they are not equal, the indicator function value is 1, otherwise it is 0.

[0115] Use the graph cut algorithm to minimize the energy function to redistribute voxels to different sub-regions to obtain the graph cut optimized results, including:

[0116] Each voxel is considered as a node in the graph, which means that if there are N voxels, there will be N nodes in the graph. Add edges between adjacent voxels. Adjacent voxels can be determined by spatial position relationship. For example, if two voxels share a face, edge or point in three-dimensional space, they are considered adjacent. Assign a weight to each edge in the graph, which is usually based on the cost function in the smoothing term. For the data term, associate each voxel (node) with the label of the sub-region to which it belongs, and calculate the corresponding cost. In the graph cut algorithm, two special nodes are set: source and sink. These two points represent two possible labels or assignment results; in the scenario, the source and sink can represent different sub-regions respectively. For example, if there are two sub-regions, the source represents sub-region A, and the sink represents sub-region B. The energy of the data term is associated with each node (voxel), which represents the cost of assigning the voxel to a specific subregion (i.e., connected to a source or a sink), and the energy of the smoothness term is associated with the edges between neighboring voxels, which represents the cost that would be incurred if the neighboring voxels were assigned to different subregions (i.e., one connected to a source and the other to a sink).

[0117] Use Graph Cut or GrabCut algorithm to find a cut of the graph, which divides the nodes in the graph into two disjoint sets while minimizing the total weight (i.e., energy) of the cut edges. In the process of graph cutting, try to balance the energy of data terms and smoothing terms to find a global optimal solution, which means that the algorithm will try to assign voxels to sub-regions that are most similar to their geometric features while trying to maintain the consistency of adjacent voxel labels. According to the results of the graph cut, update the label of each voxel, and the label now indicates the sub-region to which the voxel is assigned. Since the graph cut algorithm minimizes the energy function, this allocation method is globally optimized, taking into account both the local features of the voxels and the spatial relationship between the voxels, and outputting the optimized voxel-to-sub-region allocation results, which include the label of each voxel and the sub-region to which it belongs.

[0118] According to the results of graph cut optimization, voxels belonging to the same class are merged into a sub-region to obtain an optimized sub-region division, specifically including: according to the results of graph cut optimization, voxels belonging to the same class are merged into a sub-region, which means that each sub-region now contains a group of voxels with similar geometric features and expected flow characteristics.

[0119] According to the optimized sub-region division, the boundary information of each sub-region is updated to output the divided sub-region information, including the boundary of each sub-region, the list of included voxels, and the corresponding geometric features and flow characteristics, specifically including: for each optimized sub-region, its boundary information is updated, which includes determining the precise boundary of the sub-region, the list of included voxels, and the geometric features and flow characteristics of each voxel, and outputting this information for subsequent use, such as in mesh generation or numerical simulation.

[0120] In the embodiment of the present invention, by constructing a KD-Tree data structure, it is possible to quickly search and query spatial points, thereby improving the efficiency of data processing. KD-Tree not only accelerates the voxel division process, but also facilitates the subsequent boundary determination and geometric feature extraction. Through voxel division and KD-Tree query, the geometric features in each voxel, such as the density of points, surface curvature and normal vector distribution, can be accurately extracted. The K-means clustering algorithm is used to perform preliminary clustering of voxels, and the voxels can be grouped according to the similarity of geometric features. This method is flexible and adaptive and can handle various complex geometric shapes. By defining an energy function and minimizing the energy function using a graph cut algorithm, voxels can be redistributed to different sub-regions, thereby optimizing the boundaries of the sub-regions. This method ensures a smooth transition between sub-regions and reduces unnecessary boundary complexity. Grid generation based on the optimized sub-region division can more accurately capture the flow characteristics in the computational domain, and each sub-region can perform independent grid generation based on its specific geometry and flow characteristics, thereby improving the accuracy and stability of numerical simulation.

[0121] In a preferred embodiment of the present invention, determining the boundary of a voxel includes:

[0122] Four non-coplanar points are randomly selected from the voxel as the vertices of the initial convex hull, and an initial tetrahedron is constructed as the convex hull, specifically including: randomly selecting four points from the data points in the voxel, ensuring that the four points are not coplanar, using the four points as vertices, and constructing a tetrahedron, which will serve as the initial shape of the convex hull.

[0123] Calculate the adjustment factor based on the data point distribution; for each remaining data point within a voxel, perform the following steps:

[0124] Each face of the convex hull is expanded outward along its normal direction by the length of the adjustment factor to form an expanded convex hull. It is determined whether the current point is inside the expanded convex hull. If the current point is outside the expanded convex hull, the convex hull is updated, including finding visible faces, deleting visible faces and adding new faces. After adding new faces, the convex hull optimization operation is performed and the optimization is continued until all remaining points are processed to obtain the final convex polyhedron, including:

[0125] For each face in the convex hull, record its vertices, normal vectors, and face boundaries, and for each face in the convex hull, calculate its new vertex position after expansion along the normal direction by the adjustment factor D. Specifically, for each vertex v on the face, the new vertex position v' can be calculated by v'=v+D×n, where n is the normal vector of the face, and the expanded face is constructed using the new vertex set.

[0126] Starting from the current point, select a direction (usually a direction parallel to a coordinate axis) to emit a ray, calculate the intersection of this ray with each face of the extended convex hull, record the number and position of the intersections, if the number of intersections between the ray and the convex hull face is an odd number, the current point is considered to be outside the extended convex hull; if it is an even number, it is inside. If the current point is outside the extended convex hull, traverse each face of the convex hull and check whether the line connecting the current point and the center of the convex hull intersects with the face. If so, the face is visible. Remove all faces marked as visible from the convex hull. For each deleted face's boundary, connect the current point with these boundaries to form new faces. Add the newly formed faces to the convex hull. Check whether the newly added faces are coplanar with the existing faces in the convex hull. If so, merge them. Check whether any faces are completely inside other faces, or whether any vertices are surrounded by other vertices. If so, delete them. Check whether the newly formed convex hull still maintains convexity. Adjust it if necessary. Repeat the above steps until all remaining points have been processed. When all points have been processed, the resulting convex hull is the final convex polyhedron.

[0127] Boundary information is extracted from the final convex polyhedron, including vertices, edges and faces. The boundary information is used to describe the adjusted voxel boundaries, specifically including: From this convex polyhedron, boundary information can be extracted, including its vertices, edges and faces. These boundary information describe the adjusted voxel boundaries, which can be used for subsequent mesh generation, 3D reconstruction, model analysis or other application scenarios that require precise voxel boundaries.

[0128] In an embodiment of the present invention, by randomly selecting an initial point from a voxel and constructing a convex hull, and then dynamically adjusting the convex hull in combination with the distribution of data points, the actual boundary of the voxel can be captured more accurately. Compared with the traditional fixed shape or simple geometric approximation, this method can better reflect the true form of the voxel. The method can process voxel data of various shapes and distributions. Whether the voxel is regular or irregular, dense or sparse, the method can find a reasonable boundary description through iterative optimization. When facing noisy data or outliers, the method can eliminate interference to a certain extent and maintain the stability of the boundary by expanding the convex hull and performing multiple optimization operations. Although the method involves multiple iterations and optimization operations, each processing is adjusted for the local part of the convex hull, avoiding complex calculations in the global range, thereby improving the calculation efficiency to a certain extent. The boundary information finally extracted includes vertices, edges and faces, which not only describe the outer contour of the voxel, but also provide detailed data about the shape and structure of the voxel, which is helpful for subsequent analysis and processing.

[0129] In a preferred embodiment of the present invention, the adjustment factor The calculation formula is:

[0130] ;

[0131] in, is the standard deviation of the data point density in the local area around the current point; is the local density gradient, indicating how fast the density changes; is the maximum density value in the local area; is the average density value in the local area; is the standard deviation of the local density; is a Gaussian function; is the average radius of curvature in the local area around the current point; is the local curvature gradient, which indicates how fast the curvature changes; is the maximum curvature value in the local area; is the curvature direction angle between the current point and the adjacent point; is the number of neighboring points; is the index value of the neighboring point; , and is the weight coefficient; is the density value of data points in the local area around the current point;

[0132] ;

[0133] in, and are the maximum and minimum pressures in the local area around the current point, respectively; and are the highest and lowest temperatures in the local area around the current point, respectively; is the velocity vector of the current point; is the maximum velocity value in the local area.

[0134] In the embodiment of the present invention, by comprehensively considering multiple local characteristics (such as density, curvature, pressure, etc.), the adjustment factor D can more accurately reflect the actual situation of each point in the voxel, which makes the determination of the voxel boundary more accurate and can better adapt to data distributions of different shapes and complexities. Since the calculation of the adjustment factor D involves multiple parameters, these parameters work together to make the algorithm more robust when facing noisy data or outliers. For example, the use of Gaussian functions helps to smooth density changes and reduce the impact of outliers on the results. The weight coefficient in the formula , and The user is allowed to adjust the relative importance of each factor according to the specific application scenario. The adjustment factor D takes into account physical quantities such as pressure, temperature and velocity, which gives the algorithm an advantage in dealing with problems involving multi-physics coupling.

[0135] In order to ensure that the value range of the adjustment factor is within a reasonable range (for example, [0, 1]), the adjustment factor can be Normalization is performed, and when determining the position of the current point relative to the current convex hull, the adjustment factor is used to adjust the expansion distance of the convex hull. The specific steps are as follows:

[0136] For each face of the convex hull, expand it outward along its normal direction by a distance of D×L, where L is a preset expansion length, and determine whether the current point is inside the expanded convex hull. If the point is outside the expanded convex hull, update the convex hull.

[0137] In the above step 105, the SST k-ω turbulence model is used for numerical simulation, the ANSYS Fluent solver is selected, and the solution parameters are set, including the time step, the convergence criterion, and the discrete format; the resistance performance of the hull fairing in water is calculated by numerical simulation, including the friction resistance and the pressure difference resistance; the pressure distribution on the surface of the hull fairing is analyzed to identify the high pressure area and the low pressure area; and the anti-bubble performance of the fairing is evaluated by analyzing the pressure distribution and streamline distribution around the hull, which may include:

[0138] Select a turbulence model to simulate fluid flow. Here, the SST k-ω turbulence model is selected. This model combines the advantages of the k-ω model and the standard k-ε model, and has higher accuracy and stability when simulating complex flow phenomena (such as separated flow, vortex, etc.).

[0139] Open the ANSYS Fluent solver and import the geometric model of the hull fairing. This model is created in CAD software (such as AutoCAD, SolidWorks, etc.) and saved in a format that ANSYS Fluent can recognize (such as .step, .igs, etc.); in the solution settings, first determine the time step, set the convergence criteria, such as the residual convergence standard, and at the same time, monitor the convergence of specific flow variables (such as velocity, pressure, etc.); select the discrete format, which determines how to convert the continuous fluid dynamics equations into a discrete form that can be solved by the computer. For convection terms, you can choose the second-order upwind or QUICK format to improve the calculation accuracy; define the physical properties of the fluid, such as the density and viscosity of water; set boundary conditions, including inlet velocity, outlet pressure, wall conditions, etc., which describe the behavior of the fluid at the boundary of the simulation area; initialize the flow field, that is, give an initial fluid state as the starting point of the simulation, start iterative calculations until the convergence criteria are met, and start the iterative calculation process of ANSYS Fluent.

[0140] In each iteration step, the solver will update the solution of the fluid dynamics equation according to the current flow field state, and determine whether the solution is stable by monitoring the residual (that is, the change in the solution between two consecutive iterations). When the residual is lower than the preset convergence standard, the flow field is considered to have reached a stable state. At the same time, the key physical quantities (such as velocity and pressure at a specific location) can also be monitored as the iteration changes to ensure that they also tend to be stable.

[0141] When the flow field reaches a stable state, the friction resistance and pressure difference resistance data on the surface of the hull fairing are extracted from the simulation results; the friction resistance can be obtained by integrating the shear force on the fairing surface, which usually involves the calculation of the fluid velocity gradient near the wall; the pressure difference resistance is caused by the pressure difference formed by the fluid before and after the fairing, which can be calculated by integrating the pressure difference before and after the fairing.

[0142] The extracted friction resistance and pressure difference resistance are added together to obtain the total resistance. The variation of the total resistance with flow conditions (such as flow velocity, medium density, etc.) is analyzed. This can be achieved by conducting multiple groups of simulation experiments, changing one or more flow condition parameters in each group of experiments, and then comparing the total resistance of each group of experiments. The simulation results are compared with the experimental results to ensure the accuracy of the simulation.

[0143] During the simulation, shear forces are generated when the fluid contacts the surface of the fairing. The solver calculates these shear forces and integrates them over the entire fairing surface to obtain the total frictional resistance. This process involves factors such as fluid viscosity, velocity gradient, and wall roughness.

[0144] Since the fluid forms a pressure difference before and after the fairing, pressure differential resistance will be generated. The solver will calculate the average pressure difference before and after the fairing and multiply it by the corresponding area to obtain the pressure differential resistance. This process is closely related to the fluid velocity, medium density and shape of the fairing.

[0145] In numerical simulation, pressure is obtained by solving the pressure equation in the fluid dynamics equation. This process involves multiple factors such as fluid velocity, density, viscosity, and boundary conditions. The solver will calculate the pressure value at each calculation grid point and form the pressure distribution of the entire flow field. These pressure data can be used for further analysis, such as identifying high-pressure and low-pressure areas, and evaluating the anti-bubble performance of the guide cover.

[0146] According to the output data of the numerical simulation, find the part containing the surface pressure data of the shroud, read the pressure value of each point and its corresponding coordinates, perform a preliminary check on the read pressure data, identify and remove possible outliers, which may be caused by errors in the numerical simulation or grid quality problems. If the data points are not uniformly distributed in space, you can use data interpolation methods (such as linear interpolation, spline interpolation, etc.) to create a more regular data grid to fill in the missing data points. If there are missing values ​​in the data, you can use the average value of the surrounding points or interpolation methods to estimate.

[0147] Perform statistical analysis on the processed pressure data, calculate the average value of the entire pressure data set, calculate the standard deviation of the pressure data set to quantify the degree of dispersion of the data, and set the threshold:

[0148] High pressure area threshold (P high ) is set to the mean pressure plus one standard deviation; the low pressure area threshold (P low ) is set to the value of the mean pressure minus one standard deviation; traverse each point in the processed pressure data set, and for each point, compare the pressure value with the set threshold. If the pressure value of a point is higher than the high pressure area threshold (P high ), then the point is marked as belonging to the high pressure area. If the pressure value of a point is lower than the low pressure area threshold (P low ), then the point is marked as belonging to the low pressure area, and the pressure value is at P low and P high Points between 0 and 1 are considered to be within the normal pressure range.

[0149] For the marked high-pressure and low-pressure areas, calculate their average pressures respectively, and calculate the pressure difference between the high-pressure and low-pressure areas. This can be achieved by taking the average pressure of each area and then calculating the difference. Evaluate the contribution of the pressure difference to the overall resistance, which usually involves combining the pressure difference with parameters such as the flow rate of the fluid and the density of the medium, and using the resistance formula (such as the resistance formula in fluid dynamics) to calculate; analyze the specific impact of the distribution of high-pressure and low-pressure areas on the resistance performance of the fairing, such as high-pressure areas may cause greater resistance, while low-pressure areas may trigger the formation of bubbles or vortices.

[0150] After ANSYS FLUENT completes the numerical simulation, select "Pressure" as the physical quantity to be extracted, and specify the extraction location as around the hull or a specific area of ​​interest. Extract the pressure data at each grid point. These data are usually given in numerical form and correspond to the coordinates of the grid points.

[0151] In the simulation results of ANSYS Fluent, select the streamline generation tool, which can generate streamlines based on the velocity vector data in the flow field, that is, the motion trajectory of fluid particles, set the appropriate streamline starting point (which can be a specific point, line or area), and adjust the streamline generation parameters, such as the number, length and resolution of streamlines, to ensure that the characteristics of the flow field can be captured.

[0152] Once the streamlines are generated, use the export function of ANSYS Fluent to export the data. In the export options, make sure to select a format that contains the coordinates and velocity information of the streamlines, such as CSV or TXT. This data will be used for subsequent analysis and processing.

[0153] Import the exported streamline data into data analysis software, such as Excel, MATLAB, or Python environment (such as using the pandas library), clean the data, remove any outliers or erroneous records, and ensure that the data format is correct for subsequent analysis. Extract key information such as the coordinates, speed, and direction of each point from the streamline data, and organize this information into a structured data set to facilitate subsequent data analysis and calculations.

[0154] Among them, the specific identification process of the flow separation and vortex generation phenomenon in the above step 106 can be carried out through the following steps: for each streamline, calculate the velocity change between adjacent points, which can be done by taking the difference between the velocity vectors of adjacent points; analyze these velocity change values, and find those areas with small and stable changes. In these areas, the changes in velocity magnitude and direction are relatively small, indicating that the fluid flows smoothly at these locations; in addition to stability, it is also necessary to evaluate the continuity of velocity changes. Continuity refers to whether the velocity change occurs gradually and smoothly along the streamline without sudden jumps or interruptions. By checking the distribution of velocity data along the streamline, those areas with continuous and smooth velocity changes can be identified. Combined with the results of stability and continuity analysis, smooth streamline areas can be determined. These areas are where the fluid flows smoothly, usually characterized by small velocity changes, smooth and continuous flow.

[0155] For each point in the streamline data, extract the magnitude (i.e., rate) and direction of its velocity. The direction can be represented by an angle or a vector. Analyze the differences in velocity magnitude and direction between adjacent points, and pay special attention to areas with large differences. When the magnitude or direction of the velocity changes dramatically, this may indicate the rotation of the fluid (vortex) or separation from the solid surface (flow separation). Drastic changes can be identified by setting a threshold. For example, when the change in velocity magnitude exceeds a certain percentage, or the change in direction exceeds a certain angle, it can be considered that a dramatic change has occurred. After identifying the areas of dramatic changes, it is necessary to further analyze the characteristics of these areas to determine whether they are vortices or flow separations. Vortices are usually manifested as fluids rotating around a center with a circular change in velocity direction, while flow separation refers to the separation of the fluid from the solid surface to form an independent flow area. The velocity direction may change dramatically. Once vortex or flow separation phenomena are determined, the location and range of these phenomena are recorded.

[0156] Conduct quantitative analysis on the identified key areas, such as calculating the average, maximum, and minimum values ​​of the velocity gradient, and counting the number and distribution of smooth areas and eddy / flow separation areas. Output the analysis results in a data table or report, including the location of the key areas, velocity change characteristics, etc. If the analysis results show that the guide cover can effectively isolate or reduce the low-pressure area, maintain the stability and continuity of the streamline, and reduce the occurrence of eddy and flow separation phenomena, it can be considered to have good anti-bubble performance.

[0157] Vortex and flow separation areas may lead to bubble accumulation. In these areas, the kinetic energy of the fluid is reduced, making it easier for bubbles to stay and gather in these locations. The previously extracted pressure distribution data around the hull is combined with the streamline data for analysis. By comparing the pressure distribution and streamline characteristics, a more comprehensive understanding of the impact of the flow field on bubble behavior can be obtained.

[0158] In high-pressure areas, bubbles may have difficulty attaching because positive pressure pushes them away from the surface, while in low-pressure areas, especially in areas of eddy currents or flow separation, bubbles may accumulate more easily. High-pressure areas may indicate that the fluid is strongly compressed, while low-pressure areas may be potential areas for bubble formation. Analyze the distribution and size of low-pressure areas, especially those near the guide shield. If low-pressure areas are widespread and close to the guide shield, it may indicate poor anti-bubble performance. On the contrary, if the low-pressure areas are effectively isolated or reduced, it can be regarded as a positive indicator of anti-bubble performance.

[0159] In summary, this step involves extracting streamline data and pressure distribution data from the simulation results, through detailed analysis and identification of key flow features (such as smooth streamline areas, vortices or areas of flow separation), and combining with pressure data to comprehensively evaluate the impact of the hull fairing on bubble behavior, which helps to guide design optimization and improve the anti-bubble performance of the hull.

[0160] In the above step 107, the shape of the hull fairing is adjusted according to the flow separation and vortex generation phenomena, as well as the high pressure area and the low pressure area. The adjusted parameters include the curvature, length, width and thickness of the fairing, which may include:

[0161] According to the data of flow separation and vortex generation phenomenon obtained by analysis in step 106, as well as the distribution data of high-pressure and low-pressure areas, these data are the basis for adjusting the shape of the shroud. According to the areas where flow separation and vortex generation occur, as well as the specific locations of high-pressure and low-pressure areas, key areas that need to be focused on and adjusted are identified. For example, areas where flow separation is severe or areas where pressure difference is large. For areas where flow separation and vortex generation occur, it is possible to consider adjusting the curvature of the shroud to make it more in line with the natural flow path of the fluid and reduce the sudden change and separation of the flow. For example, the curvature is increased in the edge area where flow separation is severe to guide the fluid to transition more smoothly. According to the results of the flow simulation, the length and width of the shroud are adjusted. If the shroud is too short, the fluid may leave the shroud body before it is fully stable. Increasing the length can provide more stable flow areas, and the adjustment of the width can affect the channel size of the fluid, thereby affecting the flow rate and pressure distribution. The thickness of the shroud will also affect its performance. The thickness can be increased in areas where the structural strength needs to be enhanced, and can be appropriately thinned in areas where the fluid resistance needs to be reduced.

[0162] Import the original shroud geometry model into the CAD software, and accurately locate the area that needs to be adjusted. Use the parametric design function of the CAD software to adjust the parameters such as the curvature, length, width and thickness of the shroud. Fine adjustment can be achieved by setting variables and constraints. During the adjustment process, the real-time rendering or simulation function of the CAD software can be used to preliminarily verify the adjustment effect, which helps to promptly discover and correct unreasonable designs. Save each step of the adjustment record in the CAD software for subsequent tracking and modification, and export the adjusted shroud model to a format acceptable to the numerical simulation software, such as STL or IGES. Import the adjusted model into the numerical simulation software, set the corresponding solution parameters, boundary conditions and flow field initialization, run a new numerical simulation, and analyze whether the resistance performance and anti-bubble performance of the adjusted shroud in water are improved.

[0163] For example, a ship manufacturer discovered that the fairing of its existing ship model had flow separation and vortex phenomena during actual navigation, which caused bubble accumulation on the hull surface, affecting navigation efficiency and hull life. In order to improve this problem, the manufacturer decided to adjust the shape of the fairing.

[0164] Adjustment steps:

[0165] The original fairing was simulated using numerical simulation software, and data on flow separation, vortex generation, high-pressure and low-pressure areas were collected. The data were analyzed to determine the areas with severe flow separation and vortex phenomena, as well as the specific locations of high-pressure and low-pressure areas. It was found that some edge areas of the fairing had severe flow separation, and these areas were exactly at the junction of high-pressure and low-pressure areas, and these areas were identified as key areas that needed to be focused on and adjusted.

[0166] Shape Adjustment:

[0167] Curvature adjustment: In the edge area where flow separation is serious, the curvature of the guide cover is increased to allow the fluid to transition more smoothly and reduce mutations and separation.

[0168] Length adjustment,According to the simulation results, the length of the deflector cover is appropriately increased to provide more stable flow areas and ensure that the fluid can be fully stabilized before leaving the cover.

[0169] Width adjustment: adjust the width of certain parts of the guide cover to change the channel size of the fluid, thereby affecting the flow rate and pressure distribution and reducing bubble accumulation.

[0170] Thickness adjustment: increase thickness in areas where structural strength needs to be enhanced, and appropriately reduce thickness in areas where fluid resistance needs to be reduced.

[0171] CAD Design and Simulation:

[0172] Import the original fairing model into the CAD software, accurately locate the area that needs to be adjusted, and use the parametric design function of CAD to adjust the curvature, length, width and thickness. During the adjustment process, use the real-time rendering function of CAD to preliminarily verify the adjustment effect, and save the adjustment records of each step to facilitate subsequent tracking and modification.

[0173] The adjusted fairing model is exported to STL format and imported into the numerical simulation software. The solution parameters, boundary conditions and flow field initialization are set. A new numerical simulation is run to analyze the resistance performance and anti-bubble performance of the adjusted fairing in water.

[0174] In the above step 108, the particle swarm algorithm is used to optimize the shape of the fairing to minimize the drag coefficient and maximize the anti-bubble performance. The shape of the fairing is gradually optimized through multiple iterations. After each iteration, numerical simulation and performance evaluation are performed again until the predetermined optimization goal is achieved to output the final three-dimensional model of the hull fairing. The process may include:

[0175] Define the minimization of the drag coefficient and the maximization of the anti-bubble performance as the optimization goals, and set specific quantitative indicators for these goals, such as the specific value of the drag coefficient or the evaluation criteria for the anti-bubble performance; initialize the particle swarm, where each particle represents a possible set of fairing shape parameters; define the particle fitness function according to the optimization goal, which should reflect the combined effect of the drag coefficient and the anti-bubble performance, and search for the optimal solution by iteratively updating the particle speed and position. In each iteration, use the updated particles (i.e., the shape parameter set) to generate a new fairing geometry model, perform numerical simulation on the new model, and evaluate its drag coefficient and anti-bubble performance. Update the particle fitness value based on the evaluation results, and adjust the particle speed and position accordingly. When the predetermined optimization goal or number of iterations is reached, stop the iteration and output the final hull fairing three-dimensional model, which should be the design with the best performance under given constraints. Fitness function The calculation formula is:

[0176] ;

[0177] in, is the resistance exerted by the fluid on the deflector; is the fluid density; is the flow rate of the fluid; is the orthographic projection area of ​​the shroud; The surface of the air deflector Pressure on a certain point; It is a tiny curved element on the surface of the fairing; is the bubble generation area on the surface of the shroud; is the velocity vector at a point on the surface of the shroud; is the curl of the velocity field, indicating the vortex intensity of the flow. The particle swarm algorithm optimizes the design through continuous iteration, and the velocity and position of the particles are updated according to the following formula:

[0178] The speed update formula is:

[0179] ;

[0180] in, is the velocity of particle i in the dth dimension; w is the inertia weight; and are learning factors, which control the extent to which particles move toward the individual optimal position and the global optimal position respectively; and It is a randomly generated number in the range of [0, 1] to increase the randomness of the algorithm; the position update formula is:

[0181]

[0182] in, is the position of particle i in the dth dimension, indicating a parameter value of the deflector design.

[0183] An embodiment of the present invention further provides a three-dimensional modeling optimization system for a hull fairing, comprising:

[0184] A determination module, used for determining the functional target of the fairing according to the purpose, sailing speed and load of the ship;

[0185] An acquisition module is used to acquire specific parameters of the hull, including the hull length, width, draft, hull line shape and hull surface curvature;

[0186] A creation module is used to create a preliminary hull shroud model according to the functional objectives of the shroud and the specific parameters of the hull;

[0187] A definition module is used to define a virtual computational domain including the hull deflector and its surrounding environment based on the preliminary hull deflector model; set boundary conditions, divide the computational domain into multiple grid cells, and generate boundary layer grids on the hull surface;

[0188] The simulation module is used to perform numerical simulation and set solution parameters, including time step, convergence criterion and discrete format; calculate the resistance performance of the hull fairing in water through numerical simulation, including friction resistance and pressure difference resistance; analyze the pressure distribution on the surface of the hull fairing and identify high-pressure and low-pressure areas; evaluate the anti-bubble performance of the fairing by analyzing the pressure distribution and streamline distribution around the hull; analyze the streamline distribution near the hull through the resistance performance and anti-bubble performance, and identify the flow separation and vortex generation phenomena;

[0189] An adjustment module is used to adjust the shape of the hull fairing according to flow separation and vortex generation phenomena, as well as high-pressure and low-pressure areas. The adjustment parameters include the curvature, length, width and thickness of the fairing;

[0190] The optimization module is used to optimize the shape of the fairing using a particle swarm algorithm to minimize the drag coefficient and maximize the anti-bubble performance. The shape of the fairing is gradually optimized through multiple iterations. After each iteration, numerical simulation and performance evaluation are performed again until the predetermined optimization goal is achieved to output the final three-dimensional model of the hull fairing.

[0191] It should be noted that the system is a system corresponding to the above method, and all implementation methods in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0192] The embodiment of the present invention further provides a computing device, comprising: a processor, a memory storing a computer program, wherein when the computer program is executed by the processor, the method described above is executed. All implementations in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0193] The embodiment of the present invention also provides a computer-readable storage medium storing instructions, which, when executed on a computer, enable the computer to execute the method described above. All implementations in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

Claims

1. A three-dimensional modeling optimization method for a hull fairing, characterized in that: The method comprises: Determine the functional objectives of the fairing according to the purpose, sailing speed and load of the ship; Obtain the specific parameters of the hull, including hull length, width, draft, hull line shape and hull surface curvature; Create a preliminary hull shroud model based on the shroud's functional objectives and the hull's specific parameters; Based on the preliminary hull deflector model, a virtual computational domain including the hull deflector and its surrounding environment is defined; Set boundary conditions, divide the computational domain into multiple grid cells, and generate boundary layer grids on the hull surface; Perform numerical simulation and set solution parameters, including time step, convergence criterion and discrete format; calculate the resistance performance of the hull fairing in water, including friction resistance and pressure difference resistance, through numerical simulation; analyze the pressure distribution on the surface of the hull fairing and identify high-pressure and low-pressure areas; evaluate the anti-bubble performance of the fairing by analyzing the pressure distribution and streamline distribution around the hull; Through the resistance performance and anti-bubble performance, the streamline distribution near the hull is analyzed to identify the flow separation and vortex generation phenomena; According to the flow separation and vortex generation phenomena, as well as the high-pressure and low-pressure areas, the shape of the hull fairing is adjusted. The parameters to be adjusted include the curvature, length, width and thickness of the fairing. The particle swarm algorithm is used to optimize the shape of the fairing to minimize the drag coefficient and maximize the anti-bubble performance. The shape of the fairing is gradually optimized through multiple iterations. After each iteration, numerical simulation and performance evaluation are performed again until the predetermined optimization goal is achieved to output the final three-dimensional model of the hull fairing.

2. A three-dimensional modeling optimization method for a hull fairing according to claim 1, characterized in that: Divide the computational domain into multiple grid cells, including: Receive the computational domain geometry information provided by the user, including CAD models and point cloud data; Preprocess the geometric information and define the global parameters for meshing; The entire computational domain is divided into multiple sub-regions according to the geometric features and expected flow characteristics. Each sub-region will be processed independently to meet different mesh generation requirements. In each sub-region, an adaptive subdivision strategy is applied to dynamically determine the mesh density of each region based on the geometric complexity, flow characteristics changes, and user-defined global parameters; For each subdivided sub-region, a smooth and evenly distributed grid point is generated in the sub-region by solving a specific elliptic equation; The generated mesh is quality checked to evaluate the volume ratio, distortion and aspect ratio of each mesh unit. For mesh units that do not meet the quality standards, they are optimized by re-dividing the mesh lines or adjusting the node positions. The optimization process is repeated until the quality of all mesh units meets the preset standards to output the final mesh data, which includes the coordinates and connection relationships of the mesh points.

3. The three-dimensional modeling optimization method of a hull fairing according to claim 2, characterized in that: The entire computational domain is divided into multiple sub-regions based on the geometric features and expected flow characteristics, including: According to the distribution of geometric data points, a KD-Tree data structure is constructed; Traverse the KD-Tree data structure and divide the space into a series of voxels according to the set voxel size. Each voxel contains a part of geometric data points. In the voxel segmentation process, the voxels containing geometric data points and adjacent to the external space of the computational domain are marked as boundary voxels. Determine the boundary of the voxel, and query the data points related to the boundary through KD-Tree to obtain the divided voxel information, which includes the position, size and geometric data points of each voxel; for each voxel, extract its internal geometric features, including point density, surface curvature and normal vector distribution; The K-means clustering algorithm is used to cluster the voxels and obtain preliminary clustering results. Each voxel is regarded as a node of the graph. If two voxels are adjacent, an edge is added between the two voxels. Define an energy function, which includes a data term and a smoothing term. The data term measures the cost of assigning each voxel to a subregion, and the smoothing term measures the cost of inconsistent labels between adjacent voxels. Use the graph cut algorithm to minimize the energy function to redistribute voxels to different sub-regions to obtain the graph cut optimized result; According to the results of graph cut optimization, voxels belonging to the same category are merged into a sub-region to obtain the optimized sub-region division; According to the optimized sub-region division, the boundary information of each sub-region is updated to output the divided sub-region information, including the boundary of each sub-region, the included voxel list and the corresponding geometric features and flow characteristics.

4. The three-dimensional modeling optimization method of a hull fairing according to claim 3, characterized in that: Determine the boundaries of the voxel, including: Randomly select four non-coplanar points from the voxel as the vertices of the initial convex hull, and construct an initial tetrahedron as the convex hull; Calculate the adjustment factor based on the data point distribution; for each remaining data point within a voxel, perform the following steps: Each face of the convex hull is expanded outward along its normal direction by the length of the adjustment factor to form an expanded convex hull, and whether the current point is located inside the expanded convex hull is determined. If the current point is located outside the expanded convex hull, the convex hull is updated, including finding visible faces, deleting visible faces and adding new faces. After adding new faces, the convex hull optimization operation is performed, and the optimization is continued until all remaining points are processed to obtain the final convex polyhedron; Boundary information, including vertices, edges, and faces, is extracted from the final convex polyhedron, and the boundary information is used to describe the adjusted voxel boundaries.

5. The three-dimensional modeling optimization method of a hull fairing according to claim 4, characterized in that: Adjustment Factor The calculation formula is: ； in, is the standard deviation of the data point density in the local area around the current point; is the local density gradient, indicating how fast the density changes; is the maximum density value in the local area; is the average density value in the local area; is the standard deviation of the local density; is a Gaussian function; is the average radius of curvature in the local area around the current point; is the local curvature gradient, which indicates how fast the curvature changes; is the maximum curvature value in the local area; is the curvature direction angle between the current point and the adjacent point; is the number of neighboring points; is the index value of the neighboring point; , and is the weight coefficient; is the density value of data points in the local area around the current point; ； in, and are the maximum and minimum pressures in the local area around the current point, respectively; and are the highest and lowest temperatures in the local area around the current point, respectively; is the velocity vector of the current point; is the maximum velocity value in the local area.

6. A three-dimensional modeling optimization method for a hull fairing according to claim 5, characterized in that: The boundary conditions include inlet boundary, outlet boundary, side boundary and bottom boundary. The inlet boundary is set to the incoming flow direction to simulate the situation of water flow entering the calculation domain. It is set as the velocity inlet to give the velocity and direction of the water flow. The outlet boundary represents the outflow direction of the water flow and is set as the pressure outlet to allow the water to flow out freely. The side boundary is set as the symmetric boundary to simulate the water area at infinite distance. The bottom boundary is set as the wall boundary to simulate the influence of the bottom of the water.

7. A three-dimensional modeling optimization method for a hull fairing according to claim 6, characterized in that: Global parameters include overall size, target resolution, and maximum and minimum cell sizes.

8. A three-dimensional modeling optimization system for a ship hull, characterized in that: Applied to the method according to any one of claims 1 to 7, comprising: A determination module, used for determining the functional target of the fairing according to the purpose, sailing speed and load of the ship; An acquisition module is used to obtain specific parameters of the hull, including the hull length, width, draft, hull line shape and hull surface curvature; A creation module is used to create a preliminary hull shroud model according to the functional objectives of the shroud and the specific parameters of the hull; A definition module is used to define a virtual computational domain including the hull deflector and its surrounding environment based on the preliminary hull deflector model; set boundary conditions, divide the computational domain into multiple grid cells, and generate boundary layer grids on the hull surface; The simulation module is used to perform numerical simulation and set solution parameters, including time step, convergence criterion and discrete format; calculate the resistance performance of the hull fairing in water through numerical simulation, including friction resistance and pressure difference resistance; analyze the pressure distribution on the surface of the hull fairing and identify high-pressure and low-pressure areas; evaluate the anti-bubble performance of the fairing by analyzing the pressure distribution and streamline distribution around the hull; analyze the streamline distribution near the hull through the resistance performance and anti-bubble performance, and identify the flow separation and vortex generation phenomena; An adjustment module is used to adjust the shape of the hull fairing according to flow separation and vortex generation phenomena, as well as high-pressure and low-pressure areas. The adjustment parameters include the curvature, length, width and thickness of the fairing; The optimization module is used to optimize the shape of the fairing using a particle swarm algorithm to minimize the drag coefficient and maximize the anti-bubble performance. The shape of the fairing is gradually optimized through multiple iterations. After each iteration, numerical simulation and performance evaluation are performed again until the predetermined optimization goal is achieved to output the final three-dimensional model of the hull fairing.

9. A computing device, characterized in that include: one or more processors; A storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the method as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program, which, when executed by a processor, implements the method according to any one of claims 1 to 7.

Images