relations are founded in the formulae used for the generation of the training data. However, the sensitivity analysis reveals the weight of each different relation that cannot be derived from the formulae. These weights are the important information to allow a more efficient adjustment of the parameters to satisfy defined requirements.
9. LIMITATIONS IN THE USE OF THE NETWORK
When using the network, one has always to bear in mind that the values presented by an ANN can only be as good as its training data. This means, the network can only produce feasible results when used within the limitations laid down by the training data.
The main restrictions for the use of this network are
Monohull Bermudan sloop rig Length within 10 m to 15 m Displacement within 5 tons to 18 tons
Although ANNs have been proven in previous
applications to be feasible not only for interpolations but also for extrapolations, this has not been tested for the problem at hand.
The training data for the network is created, based on basis yachts of a certain type - cruisers and performance cruisers. Therefore the network can only produce feasible data for yachts of the same type.
It cannot be expected that innovative designs can be developed, because the training data existing yachts.
is based on
In the training data set the beam is expressed as a linear function of the length over all. Only one value for the beam is calculated for each value of the length. This should be kept
in mind while using the ANN, when
changes are made to the length beam ratio. The data
sets produced by the network do not
automatically imply that the design of a yacht with these parameters is possible. This is especially the case for values for the vertical centre of gravity. It is left to the responsibility of the designer to judge, based on the output data of the network whether a design satisfying the given performance targets can be realised.
10. CONCLUSION
An ANN is successfully applied to the preliminary design process of sailing yachts. Using the network, assumptions for important design parameters can be derived quickly. The values of these parameters are proven to be of satisfying accuracy.
The network is also able to map the functional relations between input and output data. However, due to the relatively small number of in- and output parameters the strengths of the ANN is not fully exhausted. In further applications these numbers can be increased and the relations between the parameters can become more complex. This could allow an application to a wider range of yachts.
The insertion of the ANN in an optimisation process for single or combinations of parameters could be another possible application.
11. REFERENCES [1]
Larsson, L. ; Eliasson, R. (2000) Principles of Yacht Design Blacklick, The McGraw-Hill Companies
[2] Marchaj, C. [3] Cepowski, T. A. (1996) Sail Performance
Theory and Practice London, Adlard Coles Nautical
; Szelangiewicz, T. (2001)
Application of artificial neural networks to investigations of ship seakeeping ability Polish Maritime Res, v 8 n 4: 26
[4] Mesbahi, E. ; Atlar, M. (2000) Artificial Neural Networks: Application in
Design and Modelling 1st International EuroConference on Computer Applications and Information Technology in the Maritime Industries, COMPIT2000, Berlin
[5] Mesbahi, E. ; Bertram, V. Emperical Design Formulae Using Artificial Neural Nets, 1st International EuroConference on Computer Applications and Information Technology in the Maritime Industries, COMPIT2000, Berlin
[6] Mesbahi, E. (2000) Modelling Simulation and control of Diesel Engines Using Artificial Neural Networks PhD Thesis, University, Newcastle upon Tyne.
Newcastle [7] Stergiou, C.; Siganos, D. NEURAL
NETWORKS Retrieved 26 October 2005 from:
http://www.doc.ic.ac.uk/~nd/surprise_96/journa l/vol4/cs11/
report.html
[8]
Important Neural Networks Terms Retrieved May 3rd 2006 from:
http://www.rpi.edu/dept/chem-eng/Biotech- Environ/SYSTEMS/neural/
terms.ht
Marine
B-38
©2007: Royal Institution of Naval Architects
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54