以可行域极点为初始迭代点求结构优化全局最优解  

A structural global optimization approach by usingpolar points in feasible regions as initial starting points

作　　者：余小勇[1] 

机构地区：[1]西安石油学院机械工程系,西安710065

出　　处：《计算力学学报》2002年第3期376-378,共3页Chinese Journal of Computational Mechanics

摘　　要：探索将结构优化可行域的极点选取作为初始迭代点 ,从而解决了结构优化的全局最优解问题。文中给出三个算例均得到满意解决 ,并显示了解题时的巨大优越性。From mechanical criterion of the structural optimization, it is known that the restraint conditions and the object function have clear-cut physics sense, and thus it wouldn't happen to have the optimum solution in which all restraints don't work. That is to say, the design points of the structural optimization under general case are located on the common surface of boundary restraints in the feasible regions. Even more researches discover that the global optimum of the structural optimization is usually located at a polar point in feasible regions or on the constraint surface near it. If the locations of polar points in feasible regions are found, the object function values of every polar point are simply compared to get the global optimum for the linear programming. However, for the nonlinear programming, every polar point is selected as the initial starting points and the classical optimization methods are used to get the global optimum. This method avoids the blindness of selecting the initial starting points and generally ensures to obtain global optimum. Three examples are given. Comparing with other method shows the superiority of proposed method.

关 键 词：可行域 初始迭代点 结构优化 全局最优解 非线性规划 

分 类 号：O342[理学—固体力学] TU311[理学—力学]