engleski

An Artificial Neural Network Approach to Wind Loads Estimation

Various aspects of exposed ship structure have major impact on the accuracy of wind load estimation methods. Although several appropriate approaches for dealing with these issues have been proposed so far, there is still room for improvement. In that context, this paper presents an extension of previously proposed approach, which was based on Elliptic Fourier Descriptors (EFD) that are used for ship frontal and lateral closed contour representation. In previous research, the Generalized Regression Neural Network (GRNN) was trained with elliptic Fourier descriptors of a set of closed contours and non-dimensional wind load coefficients obtained from experimental wind tunnel tests. In this paper, training and testing sample is expanded with wind load coefficients derived from 3D steady RANS Computational Fluid Dynamic (CFD) analysis. In this way, the cheaper and faster calculation can bridge the gap between ship shapes for which calculations or experiments have already been made. The obtained neural network (NN) responses are well aligned with results of available experiments and obtained CFD results. Simulations used for this purpose were based on the analysis of the relationship of various container configurations on the deck of a 9000+ TEU container ship and associated wind forces and moments.

Computational Fluid Dynamics (CFD) ; Neural networks ; Wind loads on ships

Online event.