Name
Abstract | ||
|---|---|---|
The paper presents the application of nonlinear dimensionality reduction methods to shape and physical data in the context of hull-form design. These methods provide a reduced-dimensionality representation of the shape modification vector and associated physical parameters, allowing for an efficient and effective augmented design-space exploration. The data set is formed by shape coordinates and hydrodynamic performance (based on potential flow simulations) obtained by Monte Carlo sampling of a 27-dimensional design space. Nonlinear extensions of the principal component analysis (PCA) are applied, namely kernel PCA, local PCA and a deep autoencoder. The application presented is a naval destroyer sailing in calm water. The reduced-dimensionality representation of shape and physical parameters is set to provide a normalized mean square error smaller than 5%. Nonlinear methods outperform the standard PCA, indicating significant nonlinear interactions in the data structure. The present work is an extension of the authors’ research [1] where only shape data were considered. |
Year | DOI | Venue |
|---|---|---|
2018 | 10.1007/978-3-030-13709-0_13 | LOD |
Field | DocType | Citations |
Nonlinear system,Autoencoder,Pattern recognition,Computer science,Kernel principal component analysis,Artificial intelligence,Shape optimization,Kernel method,Nonlinear dimensionality reduction,Design space exploration,Principal component analysis | Conference | 0 |
PageRank | References | Authors |
0.34 | 4 | 4 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Danny D'Agostino | 1 | 0 | 0.34 |
| Andrea Serani | 2 | 9 | 3.61 |
| Emilio F. Campana | 3 | 26 | 3.02 |
| Matteo Diez | 4 | 9 | 4.29 |