MULTIPARAMETER STRUCTURAL OPTIMIZATION USING FEM AND MULTIPOINT EXPLICIT APPROXIMATIONS

A unified approach to various problems of structural optimization, bas ed on approximation concepts, is presented. The approach is concerned with the development of the iterative technique, which uses in each it eration the information gained at several previous design points (mult ipoint approximations) in order to better fit constraints and/or objec tive functions and to reduce the total number of FE analyses needed to solve the optimization problem. In each iteration, the subregion of t he initial region in the space of design variables, defined by move li mits, is chosen. In this subregion, several points (designs) are selec ted, for which response analyses and design sensitivity analyses are c arried out using FEM. The explicit expressions are formulated using th e weighted least-squares method. The explicit expressions obtained the n replace initial problem functions. They are used as functions of a p articular mathematical programming problem. Several particular forms o f the explicit expressions are considered. The basic features of the p resented approximations are shown by means of classical test examples, and the method is compared with other optimization techniques.