Optimize Structural Topology and Computer Experimental Design: with Simulation and Optimizers - Softcover

Inhaltsangabe

Two new topology optimization methods are presentedand many sampling design methodologies for computerexperiments are compared. HIMO combines a sizingoptimizer with a metamodeling technique. A sizingoptimizer finds feasible and optimal solutions insize (e.g. plate thickness). All performanceconstraints are handled. Then a metamodel is to fitall optimal solutions and is then used to find theoptimal topology design. Only the objective (e.g.weight) is approximated, thus large-scale structuralsystems can be optimized. HIMO resulted in 18% and36% weight savings for two real projects. SOTOdirectly uses a sizing optimizer to optimizetopology as well as thickness. The thinner elementsare gradually deleted to achieve improved topology.The design problem is then reformulated with muchfewer design variables for final sizingoptimization. To improve metamodeling in HIMO, 18experimental design methods are compared. Samplesizes affect accuracy more than design types. Enoughsamples are needed to achieve low error with one-stage sampling. More uniform sampling does notgenerally lead to more accurate prediction unlessinvolving extremely non-uniformity.

Die Inhaltsangabe kann sich auf eine andere Ausgabe dieses Titels beziehen.