| Summary: | 碩士 === 國立中正大學 === 機械系 === 91 === In the structural optimization, first it is necessary to obtain an appropriate layout of the structure as an initial model for the sizing or shape optimization. Thus it can find the optimal design of a superior quality. Even for a designer with less experience in this field of the topology optimization, it is a real challenge to find the optimal configuration of structures.
In traditional optimization design, it is dangerous to decide the initial value of design variables by the engineering intuition. Because whether the optimization problem can be solved and converges in the feasible region or not are determined on the reasonableness of the set of the initial guess. The Genetic Algorithm (GA) provides a search strategy, which is based on the rules of the genetic evolution and starts from populations that are estimated by the random number. The special characters of GA can solve the problem when a designer is setting the initial values of design variables strenuous in topology optimization, but it is not yet guaranteed to obtain a feasible design.
In the process of solving an optimization problem, although GA can quickly find the solution near the extreme value in the whole domain, but if we try to find the global optimal solution from a point near the optimal one, it really takes time to evolve into the optimal result. In general, direct optimization method is of great ability to achieve a local minimum value. Therefore, we developed a hybrid method, which combines with GA and direction optimization method to improve the efficiency and quality when solving optimization problems.
In this study, an idea is accomplished to find the reasonable initial distribution of material by means of the characteristic of the genetic algorithm. And then the initial distribution is regarded as the initial guess in conventional optimization. The optimal topology is found through the technique of the sequential linear programming of the direct optimization method.
|
|---|
