| Summary: | 碩士 === 大同工學院 === 機械工程研究所 === 87 === The traditional method of shape optimization of structure has to guess the initial shape of the structure. The final optimum shape may be effected by guessing initial conditions which is not fully considerated. The developed approaches, including genetic algorithms, fuzzy set theory, neural network and sensitivity analysis, have been proposed to improve the difficulties by using traditional methods. Instead of using their experiences to guess the final shape, designers only need to know the geometric limits, loads and constraints for structure. The objective of this study is to optimize the two-dimensional structure to yield uniform stresses with least materials by combining finite element method, fuzzy inference theory and sensitivity analysis.
Several fuzzy rules are proposed to infer the probability of removal or add of element in the model of finite analysis by using the information of stress analysis and histogram of stress distribution. According to the result of this study, the proposed algorithm can solve the problem of the topological optimization of two-dimensional structure and fulfill the expecting objective.
|
|---|
