| 摘要: | 在進行有限元素分析前，需將CAD模型轉成實體網格，而實體網格建構前，必需先完成表面網格的建構。在表面網格建構過程中，需先在3D輪廓上撒點，再將點轉至uv參數平面上，接著在uv參數平面建構表面網格，最後再將網格映射回3D空間。本研究主要對一般性CAD模型進行表面網格的建構，過程中使用本實驗室開發的輪廓撒點技術進行撒點，並以該撒點結果來建構網格，可能會導致網格建構失敗，例如:3D輪廓邊長與uv參數平面上的2D輪廓邊長不一致時，將導致在3D點轉換至2D點後，點距發生改變，進而影響網格建構。另外，當外輪廓邊與內輪廓間的距離小於輪廓上的撒點間距時，可能會導致網格自交或建構失敗。因此本研究提出2D輪廓點縮放及非均勻輪廓撒點的方法，藉此改善輪廓上的點分佈，使網格能夠成功建構。完成輪廓撒點後，即可進行四邊形網格的建構。為了將網格輸出為STL檔案，本研究將透過將建構完成的四邊形網格轉成三角形網格，以符合STL檔案格式。本研究將以每個節點相鄰六個三角形網格為目標來建構網格，藉此提升網格的結構性。在所有測試案例中，與使用Rhinoceros 5 將四邊形網格轉換為三角形網格的方法相比，本研究所提出的方法產生之三角形網格中，相鄰六個三角形網格的節點在總節點數中所佔比例較高，顯示本建構方法能夠建構出結構性較佳的網格。;Before conducting finite element analysis, the CAD model must be converted into a solid mesh. Prior to constructing the solid mesh, the surface mesh must be generated. During the surface mesh construction process, seeding is first performed on the 3D contour, and the points are then mapped to the uv parameter plane. Subsequently, the surface mesh is constructed in the uv parameter plane and finally mapped back to the 3D space. This study focuses on the construction of surface meshes for general CAD models. During the process, the contour seeding technique developed in this laboratory is used for seeding, and the mesh is constructed based on the seeding results, which may lead to mesh construction failure. For instance, when the edge lengths of the 3D contour and the 2D contour in the uv parameter plane are inconsistent, the point spacing may change after converting 3D points to 2D points, thereby affecting mesh construction. Additionally, when the distance between the inner and outer contours is smaller than the seeding interval on the contours, it may result in mesh self-intersections or failure in mesh construction. Therefore, this study proposes the 2D contour point scaling and non-uniform contour seeding methods to improve the distribution of points along the contours, thereby ensuring successful mesh construction. After completing contour seeding, quadrilateral mesh construction can proceed. To enable mesh export in STL format, this study converts the constructed quadrilateral mesh into a triangular mesh to comply with the STL file requirements. The mesh is generated with the goal of achieving a valence of six at each node, thereby improving the structural regularity of the mesh. In all test cases, compared to the method of converting quadrilateral meshes into triangular meshes using Rhinoceros 5, the triangular meshes generated by the method proposed in this study exhibit a higher proportion of nodes with a valence of six, indicating that the proposed approach can produce meshes with better structural regularity. |
