| 摘要: | 在射出成型產業中，模流分析(Mold flow analysis)的應用已經相當普遍，在模流分析前，需先將CAD模型轉為實體網格模型，以提供求解器進行計算。薄殼件模型是射出成型中常見的模型，且模型上通常有許多凸起特徵。傳統上，使用者會對薄殼件建立四面體網格，因為其實體網格可以自動建構，且可應用於複雜幾何的模型。然而，近年來隨著對模流分析精度要求的提升，需要使用品質與精度更高的三角柱網格或六面體網格，相關網格之建構方式，主要以手動方式建構為主。本研究的目的是開發應用於薄殼件模型上的特徵辨識與特徵分解演算法，以輔助三角柱網格與六面體網格的建構。在本研究所提出特徵辨識的演算法中，包含了孔洞的辨識與凸起特徵的辨識。其中，凸起特徵又可分為肋特徵、管特徵、柱特徵等多種類型。本研究會先在CAD模型上計算一些前處理資料或是辨識Fillet與Loop等資料。接著，再利用這些資料進一步辨識出孔洞與凸起特徵。而關於特徵分解的演算法，本研究會使用特徵辨識後所記錄的各種資訊，計算出每個特徵在模型上的特徵區塊。特徵區塊由多個封閉的輪廓組成，可應用於輔助實體網格的建立。本研究最主要的貢獻為辨識了多種不同類型的特徵，包含孔洞、肋特徵、管特徵、柱特徵、對稱凸起特徵等。這些特徵記錄了自身的組成面、邊界條件、相鄰的特徵等資訊。另一方面，本研究也將各種類型的凸起特徵分解為形狀規則的特徵區塊。特徵區塊記錄了多個封閉的輪廓，而每個輪廓也記錄了與原始模型相對應的資料。;In injection molding, the application of mold flow analysis has become more and more popular. In mold flow analysis, a computer aided design (CAD) model must be converted into sloid meshes so that the solver can perform the required analysis. Thin-shell plastic parts are very common in injection molding. A thin shell part generally involves a thin shell and protrusion features on its inside. Tetrahedral meshes are conventionally used for thin-shell parts owing to the ease for automatic generation and applicability for complex models. Recently, with the increasing on the accuracy requirement, it is getting popular to employ prismatic and hexahedral meshes in mold flow analysis. However, generating prismatic and hexahedral meshes requires the generation of assistant planes and the splitting of the model, which are typically done manually. The purpose of this study is to develop feature recognition and feature decomposition algorithms for thin-shell plastic parts for automating the construction of prismatic and hexahedral meshes. The proposed recognition algorithm includes hole and protrusion recognition. The protrusion features can be divided into the following types: rib, tube, column and symmetric extrusion. In the proposed algorithm, firstly, some pre-processing data and basic features (such as fillet and loop) must be computed. Then, holes and protrusion features are recognized by using the pre-processing data, fillets and loop data. In feature decomposition, feature regions are computed by using the data from various types of protrusion features. A feature region is essentially composed of multiple closed contours. An individual decomposition algorithm is developed for each type of protrusion features to yield the closed contours corresponding to each feature region. The primary contributions of this study are: firstly, various types of features are recognized, including hole, rib, tube, column and symmetrical feature. The composition faces, boundary conditions and related features for each of them are recorded. Secondly, several decomposition algorithms are developed to decompose protrusion features into feature regions, each of which is regular in shape and can be meshed with structured-type mesh. Multiple closed contours of each feature region and the topological data of these contours on the CAD model are also recorded. |
