Research On Feature-based Automated Model Simplification In Multi-state Theory

In order to analyze and establish appropriate computational model, the team of the author proposed a kind of multi-state theory which aims to save computing time as much as possible in the presence of meeting the accuracy. This dissertation starts with finite element analysis theory which is an important branch of scientific computation, and then studies the primary reason causing time-consuming computation in the finite element analysis field. The computational performance of simulations depends on the number and complexity of the geometric features present in the CAD model. If we run a finite element analysis on a model with many small features, then the computational time will be very large, and also meshing on complex models often produces inaccurate meshes which may lead to incorrect analysis results. Hence, simply using more powerful computers will not solve the problem associated with highly complex models. In order to get accurate results, one must utilize simplified models which retain the important details and eliminate the irrelevant ones.This dissertation focuses on feature-based automated solid model simplification methods and implements a prototype system. The major achievements and innovations of this work include:First of all, analyze the main reasons causing too dense mesh and recognize two key features (high-curvature features and small features) according to the criteria for meshing. Also, design two kinds of feature recognition algorithm for these two features and implement the conversion from solid model to feature model. Especially, express the feature by volumetric is helpful to carry out model simplification and estimate the error produced in the simplification process. The small feature recognition algorithm introduced in this thesis can recognize common manufacturing features such as hole and groove in despite of the feature types. It could recognize more features as long as that you can control graph decomposition reasonably. In addition, the introduction of high-curvature features enriches the category of features, which would be helpful for us to understand the model simplification process better.Secondly, realize the automatic simplification for these two features which utilizes the technology of feature-removing and feature-replacing respectively. Especially, design an algorithm of feature isovolumetric replacement for high curvature sweeping feature in order to reduce the difference of the solution before and after model simplification process. The realization of automatic model simplification could save analyzer much time and is helpful to engineering.Finally, design and implement a multi-state model simplification system, analyze the important elements of the design process, explain the overall design and modules division, and illustrate the processes associated with application examples. It provides us with a practical foundation for the application of feature recognition and feature automatic simplification by realizing the model simplification system.