Optimal Design of Truss-Structures with Reliability Constraints Using Improved Genetic Algorithms

• Main Authors: Chao-Yu Chiu, 邱兆宇
• Other Authors: Sen-Lin Lu
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/30934575942420836877

碩士 === 國立臺灣科技大學 === 機械工程系 === 97 === The aim of this thesis is to develop a method to optimize trusses in minimum weight and number of cross-section types used. The optimization of trusses is performed using the force method and genetic method. The cross-section areas and nodal coordinates are taken into account as design variables. In the modified genetic algorithm, both the penalty function of parameter-less adaptive penalty scheme and the narrowing design space technology by contraction of the length of strings are employed simultaneously. The constraints consist of member stresses, slenderness ratios, nodal displacements and buckling stresses. In addition considering the inherent uncertainties on structures and their loads, the truss parameters, loads and strengths are also considered as random variables in the study. Accordingly the optimization is proceeded using reliability based constraints. In the thesis several numerical examples are illustrated and their results are compared with the reference papers. In the case of deterministic system, they are much closed and the present results are even better. In the case of random system, it shows that the optimum weight increases with both the coefficient of variation of random variable and the required reliability.