基于離散變量桁架結(jié)構(gòu)截面優(yōu)化及軟件實現(xiàn)

1、湘潭大學(xué)碩士學(xué)位論文基于離散變量桁架結(jié)構(gòu)截面優(yōu)化及軟件實現(xiàn)姓名：楊濤申請學(xué)位級別：碩士專業(yè)：結(jié)構(gòu)工程指導(dǎo)教師：劉忠20080520湘潭大學(xué)碩士學(xué)位論文 II Abstract According to the rod sections can just be real existence types in process of truss structure design, therefore, taking discrete vari

2、able optimum methods to design truss m structure is more tally with the actual situation than taking with continuous variable optimum methods. this paper take two level optimization method which is based on discrete vari

3、able to solve the minimize weight problem of truss that is subjected to multi-case loads and restricted by size, stress, local stability and displacement constraint. In the first level optimization, the solution of minim

4、ize weight under stress restriction is found by full stress method. For the compress rods subjected to axis force, the equivalent allowable stress has been got by compressed rod stability calculating formulation. Fo

5、r avoiding non-convergence，the stress ratio is multiplied by a lower relaxation factor, and the optimization solution is got by full stress method, Then, the solution of the first level optimization is rounded up, and i

6、t will be taken as the dynamic dimension constraint lower bound of the second level optimization. In the second level optimization, the most light weight solution under displacement restriction is found. Firstly, an appr

7、oximate local optimum solution is found by using the heuristic algorithm relative difference quotient algorithm, with high computational efficiency and high performance demonstrated by the performance test of rand

8、om samples. Secondly, a mathematical mode1 of (-1,0,1) programming is established first, and then it is changed into (0,1) programming model． The local optimum solution will be get from the (0,l) programming by using the

9、 delimitative and combinatorial algorithm. by this algorithm, the local optimum solution can be obtained certainly. Repeated iteration until get the optimal solution. Finally, based on MSC.PATRAN/NASTRAN software, this p