The inverse design and optimization for composite materials with random uncertainty

Abstract

The inverse design is a contemporary novel modern design method and it can provide strong theoretical support for composite materials. The GMDH-NN and Kriging surrogate models are carried out to construct the transfer relations between input variables and output responses by combining finite element analysis and sensitivity analysis. In view of the objective existence of random uncertainties, structural mechanical behaviors of composite are considered as deterministic and uncertainty responses respectively, and then the corresponding optimization mathematical models are constructed. Meanwhile, genetic algorithm (GA) is employed to solve the presented inverse optimization design. Furthermore, several examples of laminated and 2D-woven composite materials are delivered to show that presented procedures are reliable and effective to obtain the dispersive of input parameters, which are of great significance in practical engineering.