Post-Buckling Optimisation of Composite Stiffened Panels Using Neural Networks

This paper deals with the definition of a post-buckling optimisation procedure for the design of composite stiffened panels subjected to compression loads. The optimised structures are then characterised by a local skin buckling between the stiffeners and by a high ratio between the collapse load and the buckling load. To overcome too expensive analyses from a computational point of view, an optimisation procedure is developed. It is based on a global approximation strategy, where the structure response is given by a system of neural networks trained by means of finite element analyses, and on genetic algorithms, that results particularly profitable due to the presence of integer variables. The optimisation procedure reduces considerably the computational costs, offers a complete separation between the system modelling and the optimisation problem and shows that a local skin buckling between the stiffeners allows a weight reduction equal to 18%.