| Finite element simulation analysis software, as some kinds of scientific calculation software, is widely applied in various engineering fields. To use this kind of software, proper models should be built first. Usually, researchers tend to build an identical model as the real one in order to get most accurate results. Some features of the real model may bring negative effects to girding and computing and even make the problem unsolvable. An over-complex model will also cost a lot of computing time. The concept of multi-state model is introduced into finite element analysis field on the premise of satisfying the precision of the consumers' requirements. To get the balance between result precision and computing time, proper model state is to be selected on the basis of research on the relationship between model state and result precision. The improvement of computing efficiency is directly related to the quality of model state. Factors that have effects on computing precision and time are preliminarily summarized, which infers relationships that exist among factors. Multi-Objective problems can be transferred to Single-Objective problems according to classical Multi-Objective resolution. The most appropriate model state will be gained by using selection algorithm after the establishment of selection criterion. The main content of the thesis includes:1) Multi state model theory is introduced to finite element fields. Relationship among model precision, computing time and result precision is researched by building scientific experimental models.2) The predictive factors that have effect on result precision are validated through analysis of computing results from scientific experiments, which helps extract model state selection factors. The selection criterion is created considering classical multiple objective method.3) Model state selection algorithm is used to select the best state after the selection criterion is created, which will help users select the most appropriate model state. |
PDF Full Text Request