论文部分内容阅读
关于结构抗震最优设计的一般概念可见文献[1].通常,在采用非线性规划方法以获取最优解时,需要选用合适的优化方法.目前解决工程非线性问题的方式,大致有三类途径:第一类是用线性规划去逐次逼近原来的问题,叫做序列线性化(SLP);第二类是直接处理约束,例如可行方向法、梯度法等;第三类是将约束问题转比为一系列无约束最优化问题(SUMT).各类方法的选用要尽量使设计的结构在优化进程中精度高、收敛快,同时又要顾及目标函数不易求得,甚至导数不存在时,就需要采用所谓“直接搜索法”去寻找最优解,或是用差分去接近导数,但一般情况下它的精度与收敛程度较差,同时不允许结构具有太多的设计变量,因而在工程应用上往往会受到一定的限制.
The general concept of structural seismic optimal design can be found in the literature [1]. Generally, when using nonlinear programming methods to obtain the optimal solution, it is necessary to select the appropriate optimization method. Currently there are three ways to solve the nonlinear problems of engineering. The first category is to use linear programming to successively approximate the original problem, called sequence linearization (SLP); the second category is to directly deal with constraints, such as feasible direction methods, gradient methods, etc.; the third category is to transfer the constraint problem to A series of unconstrained optimization problems (SUMT). The selection of various methods should try to make the structure of the design high accuracy and convergence in the optimization process, while taking into account that the objective function is not easy to obtain, even when the derivative does not exist, it needs to The so-called “direct search method” is used to find the optimal solution, or differential is used to approximate the derivative. However, in general, its accuracy and convergence are poor, and at the same time, the structure is not allowed to have too many design variables. Therefore, it is applied to engineering applications. Often subject to certain restrictions.