A robust automatic identification method and system for the boundaries of the inner walls of a two-dimensional chamber of arbitrary topology

Abstract

The application discloses a kind of robust automatic identification method and system of arbitrary topology two-dimensional chamber inner wall boundary, original edge element data is automatically sutured with discrete noise edge element filtering, and the neat edge element set of preliminary connection of noise removal, broken edge is obtained;S2 asymmetric inner domain estimation: using multi-directional ray determination algorithm without axisymmetric assumption, the inner region estimation is carried out to the region surrounded by neat edge element set, and the inner domain sampling point set corresponding to the chamber is generated;The technical system of the application is complete, high degree of automation, strong engineering adaptability, can be seamlessly integrated in various CAE pre-processing software, CAD model processing tool, widely used in pressure vessel, heat exchanger, hydraulic component, new energy battery shell and other fields of cavity structure simulation pre-processing, with significant technical progress and engineering application value.

Description

Technical Field

[0001] This invention relates to the fields of computer-aided engineering (CAE) finite element preprocessing and intelligent identification of industrial geometric model boundaries, and particularly to a robust automatic identification method and system for the inner wall boundary of an arbitrary topological two-dimensional cavity. Background Technology

[0002] In the numerical simulation analysis process of pressure vessels, shell and tube heat exchangers, hydraulic valve cavities, and irregularly shaped cavity components, the accurate extraction of the inner wall boundary of the two-dimensional cross section is the core preprocessing step to realize the automatic generation of structured mesh and the automatic loading of inner wall boundary conditions. Its recognition accuracy and efficiency directly determine the calculation accuracy of finite element simulation and the overall engineering efficiency.

[0003] Existing two-dimensional cavity boundary recognition technologies are mostly developed based on the axisymmetric assumption. For rotationally symmetric cavities, they extract the inner wall boundaries through polar coordinate transformation, radial scan lines, and distance field transformation, achieving relatively fast processing speeds in regular circular and elliptical cavity scenarios. However, as industrial equipment structures become more complex, irregular, and integrated, and cavity topologies become increasingly diverse, existing technologies are gradually revealing significant technical shortcomings in practical engineering applications. First, the topology adaptation range is limited, and it cannot adapt to arbitrary asymmetric chambers. The interior domain determination logic of existing algorithms is highly dependent on the axisymmetric characteristics of the cavity. When faced with asymmetric irregular contours and arbitrary topologies of two-dimensional chambers, it is very easy to encounter problems such as interior domain determination deviation and boundary tracking interruption. It cannot complete the automatic boundary recognition of structures such as irregular cross-section containers and multi-channel integrated chambers, and usually relies on manual outlining, which has extremely low processing efficiency.

[0004] Secondly, it suffers from weak anti-interference capabilities and insufficient robustness. In actual engineering, 2D cross-sections exported from CAD software, cross-sectional contours generated from 3D model cutting, and geometric data reconstructed from industrial scanning commonly exhibit data defects such as boundary fracture gaps, discrete noise edge elements, and tiny stray lines. Existing technologies lack targeted pre-processing mechanisms; fractured edges directly lead to the inability to close boundary loops, and discrete noise edge elements interfere with the construction of boundary topological relationships. The final identification results are prone to problems such as burrs, misconnections, and loop breaks, making them unsuitable for direct use in subsequent simulation processes.

[0005] Third, it does not support the layered identification of multi-layered nested chambers. For structures with multiple inner walls, such as jacketed containers, multi-pass heat exchangers, and multi-layered nested cavities, existing methods cannot distinguish the boundaries of different levels of chambers. In areas of cross-obstruction, problems such as confusion between inner and outer layer boundaries and incorrect attribution can easily occur, making it impossible to independently extract the boundaries of each inner wall layer. In engineering, technicians usually need to manually disassemble the boundaries layer by layer, which is not only labor-intensive but also prone to human error.

[0006] Fourth, while the logic for gap completion is simple, its accuracy is poor under complex topologies. Most existing boundary gap completion algorithms rely solely on the Euclidean distance between endpoints, without considering the consistency of the boundary's normal, curvature continuity, and the spatial constraints inside the cavity. In complex and irregular topological scenarios, this can easily lead to problems such as the completion path deviating from the outside of the cavity and discontinuities in the curvature of the completed boundary. Consequently, the geometric accuracy of the extracted inner wall boundary is insufficient, directly affecting the reliability of subsequent finite element calculation results.

[0007] In summary, existing two-dimensional cavity interior wall boundary recognition technologies suffer from technical bottlenecks such as narrow topological adaptability, poor anti-interference robustness, lack of multi-layer nesting processing capability, and insufficient gap completion accuracy. These limitations make it difficult to meet the automated finite element preprocessing requirements of complex cavity structures in modern industry, becoming a key technical obstacle restricting the improvement of CAE simulation efficiency. Summary of the Invention

[0008] To overcome the shortcomings of the prior art, one of the objectives of this invention is to provide a robust automatic identification method and system for the inner wall boundary of an arbitrary topological two-dimensional cavity.

[0009] One of the objectives of this invention is achieved through the following technical solution: A robust automatic identification method for the inner wall boundary of an arbitrary topological two-dimensional cavity includes the step of obtaining the original boundary metadata of the contour of the two-dimensional cavity to be identified, and further includes the following steps: S1 Boundary Preprocessing: Automatic stitching of broken edges and filtering of discrete noise edge elements are performed on the original edge element data to obtain a set of regular edge elements with noise removed and broken edges preliminarily connected. S2 Asymmetric Inner Domain Estimation: A multi-directional ray determination algorithm without axisymmetric assumption is used to estimate the inner region of the area enclosed by the set of regular edge elements, and generate the inner domain sampling point set of the corresponding chamber. S3 Multi-layer Nested Layered Recognition: Based on the sampling point set of the inner domain, layered probes are deployed. The chamber level is divided according to the number of times the probe penetrates the edge element. The edge element is assigned to the corresponding level to form a candidate boundary set for each level, eliminating boundary confusion caused by cross-occlusion of multiple chambers. S4 Adjacency Graph Optimization and Gap Completion: For each level of candidate boundary set, construct an edge element adjacency graph, optimize the adjacency weight function and introduce the inner distance constraint rule to automatically complete the gaps in the boundary, and obtain the closed boundary candidate set for each level. S5 Multidimensional Consistency Check: Perform multidimensional consistency checks on the candidate set of closed boundaries at each level, filter and output the independent valid inner wall boundaries of each level.

[0010] Furthermore, the automatic stitching of the broken edges specifically involves: traversing the two endpoints of all edge elements, calculating the Euclidean distance between any two endpoints, and when the distance between the two endpoints is less than a preset stitching threshold and they do not belong to the same edge element, connecting the two broken endpoints with a straight line segment or a third-order Bézier curve to generate a stitched edge element; the discrete noise edge element filtering specifically involves: calculating the length and local curvature change rate of a single edge element, and identifying edge elements with a length less than a preset length threshold and a local curvature change rate greater than a preset curvature threshold as discrete noise edge elements and removing them.

[0011] Furthermore, the execution process of the multi-directional ray determination algorithm is as follows: a uniformly distributed set of test points is generated within the bounded rectangle of the two-dimensional cavity contour; for each test point, detection rays are emitted in at least three non-parallel arbitrary directions; the number of intersections between each ray and the regular edge element is counted; the intersection count is corrected by combining the local convexity and concavity of the contour at the intersection point; if the number of intersections of all directional rays after correction is odd and has the same parity, then the test point is determined to belong to the cavity interior region; finally, the interior region sampling point set is composed of all interior region test points.

[0012] Furthermore, the layered probe deployment strategy is as follows: taking the outermost contour edge element as a reference, probe rays are deployed along the inner normal direction of each edge element at a preset step size; each probe ray extends into the cavity along the inner normal direction, and the cavity level number is increased by one for each edge element it passes through; based on the level penetration record of all probe rays, each edge element is divided into its corresponding cavity level, forming an independent subset of candidate boundary edge elements for each level.

[0013] Furthermore, for edge elements in the cross-occlusion region, a hierarchy assignment correction step is also included: calculate the shortest distance from each sampling point on the edge element to be judged to the sampling point set of each hierarchy, assign the edge element to the chamber hierarchy corresponding to the shortest distance, and correct the hierarchy misjudgment caused by cross-occlusion.

[0014] Furthermore, the adjacency weight function uses the Euclidean distance between the endpoints of the edge elements as the base term, and superimposes a normal consistency term and a curvature continuity term; wherein the normal consistency term is calculated based on the angle between the normals of two adjacent edge elements at the endpoints, and the smaller the angle, the higher the weight of the normal consistency term; the curvature continuity term is calculated based on the curvature difference between two adjacent edge elements at the endpoints, and the smaller the difference, the higher the weight of the curvature continuity term; the inner distance constraint rule is: the shortest distance from all candidate path points for gap completion to the current level inner domain sampling point set is not greater than a preset inner distance threshold, ensuring that the completion path is located on one side of the cavity.

[0015] Furthermore, the specific process of gap completion is as follows: traverse all endpoints of the current level candidate boundary element subset, and select the endpoints that are adjacent to only one edge element as gap endpoints; in the edge element adjacency graph, with two paired gap endpoints as the start and end points, use a weight-first path search algorithm to search for the optimal completion path under the inner distance constraint, generate the corresponding completion edge element and add it to the current level candidate boundary set to obtain the closed boundary candidate set.

[0016] Furthermore, the multi-dimensional consistency test includes geometric closure test, hierarchical nesting consistency test, and internal normal consistency test: Geometric closure test: Verify whether the boundary forms a single closed loop without self-intersections or repeated edge elements; Nested hierarchy consistency check: Verify that all points of the inner boundary are located inside the outer boundary, and that there is no intersection between different level boundaries; Inner normal consistency check: Verify that the inner normal of all edge elements on the same boundary points to the interior of the corresponding level of cavity; A closed boundary that passes all three tests is considered a valid inner wall boundary.

[0017] Furthermore, the original boundary data of the two-dimensional cavity to be identified comes from CAD two-dimensional profile files, industrial scanning profile extraction data, or finite element model section data; the output effective inner wall boundary data of each layer is directly imported into the finite element preprocessing process for automatic mesh generation, boundary condition definition, and load application.

[0018] A robust automatic identification system for the inner wall boundary of an arbitrary topological two-dimensional cavity includes: The data input module is used to acquire the original edge data of the contour of the two-dimensional cavity to be identified; The boundary preprocessing module is used to perform automatic stitching of broken edges and filtering of discrete noise edge elements on the original edge data, and output a set of regular edge elements; The interior estimation module is used to estimate the internal region of the cavity and generate an interior sampling point set based on a multi-directional ray determination algorithm without axisymmetric assumptions, using a set of regularized edge elements. The layer recognition module is used to perform layer probe deployment and layer division. Based on the probe penetration results, the edge elements are assigned to the corresponding chamber layer, and a candidate boundary set for each layer is generated. The boundary completion module is used to construct an edge adjacency graph for each level of candidate boundary set, optimize the adjacency weight function and introduce the inner distance constraint to complete the automatic completion of boundary gaps and output the candidate set of closed boundaries for each level. The consistency verification module is used to perform multi-dimensional consistency checks on the candidate set of closed boundaries at each level, and to filter out the effective inner wall boundaries of each level. The output module is used to output independent inner wall boundary data for each layer.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: First, this invention abandons the axisymmetric premise of traditional algorithms and adopts an interior domain estimation algorithm based on multi-directional ray joint determination. Through multi-directional parity verification and local convexity-concavity correction, it can accurately determine the interior domain and identify the boundary of any irregular and asymmetric topology two-dimensional cavity. The applicable range is extended from regular rotary cavities to two-dimensional cavities with all topological shapes, and can cover various engineering scenarios such as irregular cross-section containers, non-standard cavities, and multi-channel integrated structures.

[0020] Secondly, the invention adds a boundary preprocessing module, which quickly matches the fracture endpoints through spatial indexing and achieves smooth stitching. At the same time, it combines length and curvature dual features to filter discrete noise edge elements, which can effectively eliminate data defects caused by CAD import errors and scanning noise, and improve the quality of input data from the source. On this basis, with the optimized gap completion algorithm, it can effectively deal with boundary damage scenarios of different degrees, significantly reduce the requirements for input data quality, and still output stable and reliable recognition results under low-quality input with noise and fractures.

[0021] Third, this invention designs a layered probe deployment strategy based on the inner normal. The chamber level is automatically divided by the number of times the probe penetrates the edge element. Combined with the distance correction logic of the cross-occlusion area, the boundaries of each layer in the multi-layer nested structure can be accurately distinguished. This effectively solves the boundary confusion problem caused by the cross-occlusion of multiple chambers, and realizes the independent and fully automatic extraction of the inner wall boundaries of each layer. There is no need for manual layer-by-layer splitting. The efficiency of multi-layer chamber boundary processing can be improved by more than 20 times.

[0022] Fourth, this invention optimizes the weight function of the edge adjacency graph by superimposing a normal consistency term and a curvature continuity term on the distance term, making the completed path more consistent with the geometric continuity characteristics of the real boundary. At the same time, it introduces an inner distance constraint rule to ensure that the completion process is always confined to the inner space of the cavity, fundamentally avoiding the problem of outward erroneous completion, and significantly improving the geometric accuracy and topological rationality of gap completion under complex irregular topology.

[0023] Fifth, this invention sets up a triple verification mechanism of geometric closure, hierarchical nesting consistency, and internal normal consistency to comprehensively verify and automatically correct the initially identified boundaries. This effectively eliminates invalid boundaries that are self-intersecting or misconnected, corrects hierarchical intersections and reversed normals, and ensures that the topologically valid, geometrically accurate, and normal-uniform inner wall boundaries of each layer in the final output are valid. This can be directly imported into the finite element preprocessing workflow for mesh generation and boundary condition loading, effectively guaranteeing the accuracy and reliability of subsequent simulation calculations.

[0024] Overall, the present invention has a complete technical system, a high degree of automation, and strong engineering adaptability. It can be seamlessly integrated into various CAE preprocessing software and CAD model processing tools, and is widely used in the simulation preprocessing of cavity structures in many fields such as pressure vessels, heat exchangers, hydraulic components, and new energy battery shells. It has significant technological advancements and engineering application value.

[0025] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0026] Figure 1 This is a flowchart of the automatic identification method in this embodiment; Figure 2 This is a flowchart of the automatic identification system in this embodiment. Detailed Implementation

[0027] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0028] It should be noted that when a component is described as "fixed to" another component, it can be directly on the other component or may have a component in between. When a component is considered "connected to" another component, it can be directly connected to the other component or may have a component in between. When a component is considered "set on" another component, it can be directly set on the other component or may have a component in between. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0029] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0030] This embodiment uses a two-dimensional CAD cross-section of an industrial plate multi-chamber heat exchanger as the identification object. The original contour edge data is obtained by parsing a DXF format file, containing three nested asymmetric chamber boundaries: outer chamber, intermediate heat exchange chamber, and inner flow channel. The original data contains 12 boundary fracture gaps of 0.5~2mm and 27 discrete noise edge elements, exhibiting an overall irregular topological structure that traditional axisymmetric boundary identification methods cannot effectively handle. The specific implementation process of the method described in this embodiment is as follows: Step S1 Boundary Preprocessing: Automatic Fragment Stitching and Discrete Noise Edge Element Filtering First, the imported raw edge data is normalized to a two-dimensional Cartesian coordinate system in millimeters. Then, a KD-tree spatial index is used to establish a spatial retrieval structure for the endpoints of all edge elements. This involves traversing the two endpoints of all edge elements and retrieving the nearest non-edge element endpoint for each endpoint. Let the plane coordinates of any two endpoints be respectively The Euclidean distance between the two endpoints is calculated using the following formula:

[0031] When the Euclidean distance between the two endpoints is less than the preset suture threshold When the fracture boundary is determined to be a pair of endpoints, the fracture points are directly sutured with straight segments when the distance between them is less than 1 mm, and the fracture points with a distance between them of 1-2 mm are smoothly sutured using a third-order Bézier curve. The parametric equation of the third-order Bézier curve is: In the formula , There are two fracture endpoints. , To ensure the continuity of curvature at the seam, control points are generated based on the tangent directions at both ends, and the generated seam elements are added to the element set.

[0032] Discrete noise edge element filtering: Calculate the geometric length L of a single edge element, and simultaneously take five sampling points (the two endpoints of the edge element and adjacent edge elements) to fit a local circular arc, calculating the local curvature change rate of the edge element. For three consecutive points A, B, and C, the curvature of the fitted circular arc is calculated using the following formula: The criteria for determining discrete noise edge elements are: In the formula: L is the geometric length of a single edge element; Let be the rate of change of local curvature of this edge element; The preset length filtering threshold; The preset curvature change rate threshold.

[0033] This step yields a set of regularized edge elements with noise removed and initial connections broken.

[0034] Step S2: Asymmetric Interior Domain Estimation: Multi-directional Ray Determination The bounding rectangle of the contour is constructed based on a set of regular edge elements, and expanded outward by 5% as the detection domain. A uniformly distributed set of grid test points is generated with a step size of 0.5 mm. Let the range of the horizontal coordinates of the detection domain be... The range of the vertical axis is Then the test point set The coordinates are: In the formula For grid step size, , It is a non-negative integer.

[0035] For each test point, infinitely long detection rays are emitted along four non-parallel directions: 0°, 45°, 90°, and 135°. The number of intersections between each ray and a regular edge element is counted. Let a certain ray be oriented with respect to the test point... Starting from the origin, the direction vector is... Its parametric equation is: Let the two endpoints of a certain edge element be... Its line segment parametric equation is: Solving for parameters by solving two simultaneous equations If the ray is determined to have a valid intersection with the edge element, the intersection count is incremented by 1.

[0036] For the special case where the intersection point falls exactly on the endpoint of the edge element or the inflection point of the contour, a small offset method is used for counting correction: the ray is offset by 0.01mm in the vertical direction and then counted again to avoid misjudgment of parity caused by concurrent points or overlapping edges.

[0037] Inner domain determination rule: Let the first domain be determined by the second domain. The number of intersection points of the directional rays is If the following conditions are met: If the number of intersections of the four directional rays is the same in terms of parity and is odd, then the test point is determined to belong to the intracavitary region; if there are two or more directions with inconsistent parity, then it is marked as an ambiguous point and removed.

[0038] All test points that meet the judgment criteria constitute the initial inner domain sampling point set, and the total number of ray intersections of each test point is recorded as a preliminary level identifier.

[0039] Step S3: Multi-layer nested hierarchical identification: Hierarchical probe deployment and hierarchical division The chamber hierarchy is divided using an inner normal layered probe strategy. The specific process is as follows: Probe placement: Using all edge elements of the outermost contour as a reference, probe starting points are placed along the inner normal direction of each edge element at 1mm arc length intervals. For endpoints... The straight edge element has a tangent vector as The unit tangent vector is Then the unit normal vector Calculate using the following formula: The normal direction is verified by combining the position of the initial inner domain sampling point set to ensure that the normal vector points to the inside of the cavity; if the direction is opposite, it is reversed.

[0040] Each probe ray extends into the cavity along the inner normal direction, and its parametric equation is: In the formula As the probe starting point, This represents the distance traveled along the inner normal direction.

[0041] Level counting: Each probe ray intersects each layer's edge element sequentially. The chamber level number is incremented by 1 for each successful passage through a valid edge element. The chamber level number corresponding to each segment of the probe ray is recorded. Suppose a probe ray passes through K edge elements in total, corresponding to... If there are K intersection points, then the interval The corresponding chamber level is k, where This represents the maximum extension distance of the ray.

[0042] Edge element hierarchy assignment: For each internal edge element, count the hierarchy values ​​at the intersection points of all probe rays intersecting with that edge element, and take the hierarchy value with the highest frequency as the hierarchy to which the edge element belongs. Suppose that edge element e is traversed by M probe rays, and the corresponding hierarchy value set is as follows: Then the edge element's hierarchy is: In the formula This represents the frequency of the hierarchical value h in the set.

[0043] Cross-occlusion correction: For edge elements located at multi-layer boundaries or with ambiguous hierarchical affiliations, five sampling points are uniformly selected on the edge element. The shortest Euclidean distance from each sampling point to the sampling point set of each layer's inner domain is calculated, and the edge element is assigned to the chamber layer corresponding to the shortest distance. Let sampling point q be the distance to the inner domain point set of the k-th layer. The shortest distance is: The average of the shortest distances of all sampling points on the edge element is taken as the distance criterion and assigned to the level with the smallest average distance to correct the misjudgment of the level caused by cross occlusion.

[0044] This step ultimately divides all edge elements into three independent hierarchical candidate boundary sets: the first layer (inner wall of the outer cavity), the second layer (inner wall of the intermediate heat exchange cavity), and the third layer (inner wall of the inner flow channel).

[0045] Step S4: Adjacency Graph Optimization and Gap Completion For each level of candidate boundary set, an edge element adjacency graph is constructed, where nodes are the endpoints of edge elements and edges correspond to the original edge elements. The adjacency weight function is optimized and an inner distance constraint is introduced to complete the gap filling. 1. Adjacency Weight Function: Defines the adjacency weight between two endpoints as follows: In the formula, Let be the Euclidean distance between the two endpoints. Let be the angle between the normals of the edge elements belonging to the two endpoints at the endpoints. The difference in curvature of the edge elements belonging to the two endpoints at the endpoints; , As the weighting coefficients, in this embodiment, we take respectively... =1.0、 =1.5、 =1.8; the smaller the weight value, the higher the geometric rationality of the connection between the two endpoints.

[0046] 2. Inner Distance Constraint Rule: Set an inner distance threshold. The requirement is that the shortest distance from all sampling points on the candidate path for gap completion to the sampling point set of the current level's inner domain is no greater than [missing value]. This ensures that the completion path is always located inside the cavity, avoiding incorrect completion to the outside of the cavity.

[0047] 3. Gap Completion Execution: Traverse all edge element endpoints in the current level, count the number of connected edge elements for each endpoint, and mark the endpoint with a connection count of 1 as a gap endpoint; pair gap endpoints within the same level, with the pairing condition being that the distance between the two endpoints is less than 5mm and their normals are opposite; using the two paired gap endpoints as the start and end points, use Dijkstra's shortest path algorithm to search for the path with the minimum weight in the adjacency graph, while also satisfying the inner distance constraint; discretize the searched path into completion edge elements, add them to the candidate boundary set of this level, and finally form the closed boundary candidate set of this level.

[0048] In this embodiment, the three-layer chamber automatically completes the repair of 12 fractures and cracks, and after repair, the boundaries of each layer form a continuous closed loop.

[0049] Step S5: Multi-dimensional Consistency Check and Output Perform three consistency checks sequentially on the candidate sets of closed boundaries at each level: Geometric closure test: Starting from one endpoint of any edge element on the boundary, traverse all edge elements sequentially along their adjacency relationships, verifying that the process eventually returns to the starting point and that all edge elements are traversed only once. Let the closed boundary contain N edge elements, and the endpoint sequence in traversal order be... Then the closure criterion is: In the formula To ensure closure tolerance, a line segment intersection detection algorithm is used to verify that the boundary has no self-intersecting line segments and no repeated edge elements. If these conditions are not met, the boundary is determined to be invalid.

[0050] Hierarchical Nesting Consistency Check: Take all vertices of the inner boundary and use the ray casting method to determine whether each vertex is inside the outer boundary. Verify that all inner vertices are within the bounded area of ​​the outer boundary, and that there are no segment intersections between different hierarchical boundaries. For test point q, emit a ray horizontally to the right and count the number of intersections with the outer boundary. If the number is odd, the point is determined to be inside the polygon; if there is an intersection, return to step S3 to re-correct the level.

[0051] Inner normal consistency check: Calculate the unit inner normal vector of each edge element on the boundary. Take any reference point c in the current level's inner domain, and compare the midpoint of the edge element. Construct a vector pointing to the reference point. Determining normal direction consistency using dot product: If the above formula holds true, the normal points to the inner domain, verifying that the inner normals of all edge elements on the same boundary point in the same direction; if there are edge elements with reversed normals, they are automatically flipped and corrected.

[0052] A closed boundary that passes all three tests is considered a valid inner wall boundary, and the final output is the independent inner wall boundary vector data of layers 1 to 3.

[0053] In this embodiment, the boundary recognition accuracy of the cross-section of the multi-chamber heat exchanger reaches 99.2%, and the recognition time for a single frame cross-section is less than 0.8s. The output boundary data can be directly imported into finite element preprocessing software for structured mesh generation and automatic loading of thermal boundary conditions, which is more than 20 times more efficient than manual recognition and processing.

[0054] Corresponding System Implementation Examples This embodiment also provides a robust automatic identification system for the inner wall boundary of an arbitrary topological two-dimensional cavity that implements the above method, including: Data input module: Supports importing data in various formats such as DXF, DWG, STL cross sections, and scanned contour point clouds, completes coordinate analysis and edge element extraction, and outputs raw edge data; built-in coordinate normalization and unit conversion unit to adapt to input data from different sources.

[0055] Boundary preprocessing module: Built-in KD tree spatial indexing unit, performs automatic stitching of broken edges and filtering of discrete noise edge elements, and outputs a set of regular edge elements; supports custom configuration of stitching threshold and noise threshold.

[0056] Interior estimation module: It adopts a multi-directional ray determination algorithm without axisymmetric assumption, generates a uniform test point set based on the regular edge element set, and completes the determination of interior points and output of interior sampling point set.

[0057] Layer recognition module: performs internal normal layer probe deployment and layer counting, completes edge cell layer allocation and cross occlusion correction, and outputs candidate boundary sets for each layer.

[0058] Boundary completion module: Constructs an adjacency graph of edge elements, and uses the shortest path search to complete the gap completion based on the optimized weight function and the inner distance constraint, outputting the candidate set of closed boundaries at each level.

[0059] Consistency verification module: Integrates three verification logics: geometric closure, hierarchical nesting consistency, and internal normal consistency, to perform validity screening and correction on closed boundaries.

[0060] Output module: Exports the final effective inner wall boundary to common CAD formats such as STEP and IGES, or directly outputs it as a boundary file that can be read by finite element preprocessing software.

[0061] The above embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of protection of the present invention. Any non-substantial changes and substitutions made by those skilled in the art based on the present invention shall fall within the scope of protection claimed by the present invention.

Claims

1. A robust automatic identification method for the inner wall boundary of a two-dimensional cavity with arbitrary topology, comprising the step of acquiring the original boundary data of the contour of the two-dimensional cavity to be identified, characterized in that, It also includes the following steps: S1 Boundary Preprocessing: Automatic stitching of broken edges and filtering of discrete noise edge elements are performed on the original edge element data to obtain a set of regular edge elements with noise removed and broken edges preliminarily connected. S2 Asymmetric Inner Domain Estimation: A multi-directional ray determination algorithm without axisymmetric assumption is used to estimate the inner region of the area enclosed by the set of regular edge elements, and generate the inner domain sampling point set of the corresponding chamber. S3 Multi-layer Nested Layered Recognition: Based on the sampling point set of the inner domain, layered probes are deployed. The chamber level is divided according to the number of times the probe penetrates the edge element. The edge element is assigned to the corresponding level to form a candidate boundary set for each level, eliminating boundary confusion caused by cross-occlusion of multiple chambers. S4 Adjacency Graph Optimization and Gap Completion: For each level of candidate boundary set, construct an edge element adjacency graph, optimize the adjacency weight function and introduce the inner distance constraint rule to automatically complete the gaps in the boundary, and obtain the closed boundary candidate set for each level. S5 Multidimensional Consistency Check: Perform multidimensional consistency checks on the candidate set of closed boundaries at each level, filter and output the independent valid inner wall boundaries of each level.

2. The robust automatic identification method according to claim 1, characterized in that, In step S1, the automatic stitching of the broken edge specifically involves: traversing the two endpoints of all edge elements, calculating the Euclidean distance between any two endpoints, and when the distance between the two endpoints is less than a preset stitching threshold and they do not belong to the same edge element, connecting the two broken endpoints with a straight line segment or a third-order Bézier curve to generate a stitched edge element; the discrete noise edge element filtering specifically involves: calculating the length and local curvature change rate of a single edge element, and identifying edge elements with a length less than a preset length threshold and a local curvature change rate greater than a preset curvature threshold as discrete noise edge elements and removing them.

3. The robust automatic identification method according to claim 1, characterized in that, In step S2, the execution process of the multi-directional ray determination algorithm is as follows: a uniformly distributed set of test points is generated within the bounded rectangle of the two-dimensional cavity contour; for each test point, detection rays are emitted in at least three non-parallel arbitrary directions; and the number of intersections between each ray and the regular edge element is counted. The intersection point count is corrected by combining the local convexity and concavity of the contour at the intersection point. If the number of intersection points of all directional rays is odd and has the same parity after correction, the test point is determined to belong to the cavity intra-domain. Finally, the intra-domain sampling point set is composed of all intra-domain test points.

4. The robust automatic identification method according to claim 1, characterized in that, In step S3, the layered probe deployment strategy is as follows: taking the outermost contour edge element as the reference, probe rays are deployed along the inner normal direction of each edge element at a preset step size; each probe ray extends into the cavity along the inner normal direction, and the cavity level number is increased by one for each edge element it passes through; according to the layer penetration record of all probe rays, each edge element is divided into its corresponding cavity level, forming an independent subset of candidate boundary edge elements for each level.

5. The robust automatic identification method according to claim 4, characterized in that, In step S3, for edge elements in the cross-occlusion area, a level assignment correction step is also included: calculate the shortest distance from each sampling point on the edge element to be judged to the sampling point set of each level's inner domain, assign the edge element to the chamber level corresponding to the shortest distance, and correct the level misjudgment caused by cross-occlusion.

6. The robust automatic identification method according to claim 1, characterized in that, In step S4, the adjacency weight function uses the Euclidean distance between the endpoints of the edge element as the base term, and superimposes the normal consistency term and the curvature continuity term. The normal consistency term is calculated based on the angle between the normals of two adjacent edge elements at their endpoints. The smaller the angle, the higher the weight of the normal consistency term. The curvature continuity term is calculated based on the curvature difference between two adjacent edge elements at their endpoints. The smaller the difference, the higher the weight of the curvature continuity term. The inner distance constraint rule is that the shortest distance from all candidate path points for gap completion to the current level inner domain sampling point set is not greater than the preset inner distance threshold, ensuring that the completion path is located on one side of the cavity.

7. The robust automatic identification method according to claim 6, characterized in that, In step S4, the specific process of gap completion is as follows: traverse all endpoints of the current level candidate boundary element subset, and select the endpoints that are adjacent to only one edge element as gap endpoints; in the edge element adjacency graph, with two paired gap endpoints as the start and end points, use a weight-first path search algorithm to search for the optimal completion path under the inner distance constraint, generate the corresponding completion edge element and add it to the current level candidate boundary set to obtain the closed boundary candidate set.

8. The robust automatic identification method according to claim 1, characterized in that, In step S5, the multi-dimensional consistency check includes geometric closure check, hierarchical nesting consistency check, and internal normal consistency check: Geometric closure test: Verify whether the boundary forms a single closed loop without self-intersections or repeated edge elements; Nested hierarchy consistency check: Verify that all points of the inner boundary are located inside the outer boundary, and that there is no intersection between different level boundaries; Inner normal consistency check: Verify that the inner normal of all edge elements on the same boundary points to the interior of the corresponding level of cavity; A closed boundary that passes all three tests is considered a valid inner wall boundary.

9. The robust automatic identification method according to claim 1, characterized in that, The original boundary data of the two-dimensional cavity to be identified comes from CAD two-dimensional profile files, industrial scanning profile extraction data, or finite element model section data; the output effective inner wall boundary data of each layer is directly imported into the finite element preprocessing process for automatic mesh generation, boundary condition definition, and load application.

10. A robust automatic identification system for the inner wall boundary of an arbitrary topological two-dimensional cavity, characterized in that, include: The data input module is used to acquire the original edge data of the contour of the two-dimensional cavity to be identified; The boundary preprocessing module is used to perform automatic stitching of broken edges and filtering of discrete noise edge elements on the original edge data, and output a set of regular edge elements; The interior estimation module is used to estimate the internal region of the cavity and generate an interior sampling point set based on a multi-directional ray determination algorithm without axisymmetric assumptions, using a set of regularized edge elements. The layer recognition module is used to perform layer probe deployment and layer division. Based on the probe penetration results, the edge elements are assigned to the corresponding chamber layer, and a candidate boundary set for each layer is generated. The boundary completion module is used to construct an edge adjacency graph for each level of candidate boundary set, optimize the adjacency weight function and introduce the inner distance constraint to complete the automatic completion of boundary gaps and output the candidate set of closed boundaries for each level. The consistency verification module is used to perform multi-dimensional consistency checks on the candidate set of closed boundaries at each level, and to filter out the effective inner wall boundaries of each level. The output module is used to output independent inner wall boundary data for each layer.

Images