Design And Development Of Computational Software For Experimental Design And Approximate Model

With the rapid development of the aerospace, automotive, machinery and other areas, application systems(such as mechanical,electrical,structural or dynamical systems) are required to meet more severe requirements. And traditional design methods are difficult to help designer to achieve a suitable design. Design optimization method is becoming a trend. For the simple optimization model with explicit mathematical model, commonly used optimization algorithms can easily obtain the optimal solution, but in practical engineering, design optimization proble-ms are difficult to be expressed in explicit mathematical formulations. For large complex sy-stems(for example, in structural and multidisciplinary optimization), designer have to face foll-owing obstacles:the number of design variables is usually very large, most of index functions (objectives and constraints) have no explicit mathematical formulations, and furthermore it is difficult to conduct integrated multidisciplinary analysis. Therefore corres-ponding design optimization work may lead to large amount computations and long period iterations. A poss-ible way to facilitate design optimization is to incorporate methods of approximate model and experiment design.In this paper, experimental design methods and approximate models were introduced in detail firstly, and then based on the integrated optimization software system, an open software framework design for experimental design and approximate models was proposed using Object Oriented methods, design patterns and UML. Four experiments design methods and two appro-ximate models were implemented and these modules were tested with several benchmarks. It is worth noting that based on the open APIs of the framework, new approximate algorithms, new experiment design methods and new GUIs can be easy to be implemented.Using this software module and the GA algorithm in the integrated optimization software, design optimization of a complex engineering structure was performed successfully. It is shown that the software mod-ules are efficient tools for design optimization of complex application systems with general purposes.