International Journal of Small Craft Technology
APPLICATION OF ARTIFICIAL NEURAL NETWORKS IN PRELIMINARY SAILING YACHT DESIGN G Ernst, R Birmingham and E Mesbahi, University of Newcastle upon Tyne, UK SUMMARY
This paper describes the application of an Artificial Neural Network (ANN) in the first design steps of a sailing yacht. The network is successfully used to generate values for the main dimensions of a sailing yacht, satisfying given performance targets. By starting the design process with these accurate design parameters the iterative design process can be shortened dramatically. The network is also mapping the often non-linear functional relations between the design parameters and therefore gives the designer a better understanding of the system yacht.
The training data for the ANN is produced by systematic variation of the main dimensions of a standardised yacht.
NOMENCLATURE AS
Sail Area
AR Aspect ratio of sails BOA
Beam over all
DA Δ
Dellenbaugh angle Displacement
GM Metacentric height HA
Heeling Arm
LOA Length over all LWL Water line length
Draft
[m²] [-]
[m] [°]
[kg], [t] [m] [m]
[m], [ft] [m]
MH Mast height above sheer [m] T
[m] 1. INTRODUCTION
Designing a sailing yacht is an iterative process that is characterised by adjusting different design parameters to satisfy previously defined requirements. Most of the times these requirements, such as
performance,
functionality, safety and cost, are in conflict with each other and the designer has to find the best possible compromise. This process is very time consuming.
If the functional relations between the design
parameters are too complex to overlook, coefficients derived from existing designs or empirical values based on experience are used to make first assumptions for the new design. However, these methods are not able to map the functional relations accurately.
Nowadays a variety of design software is available that helps to shorten the design process dramatically. But still, a lot of different steps are necessary and the quality of the very first assumptions made by the designer have major influence on the procedure.
For this reason an attempt is made to apply an Artificial Neural Network to the preliminary design process of sailing yachts to derive accurate design parameters in an early design stage, satisfying given requirements.
©2007: Royal Institution of Naval Architects
2. ARTIFICIAL NEURAL NETWORKS
Artificial Neural Networks (ANNs) are computer systems, inspired by the human brain structure. The network consists of several processing units, each of which is the mathematical model of an organic nerve cell, the neuron.
system these processing units are highly interconnected to form
the network. However,
complexity of the human nervous system, has not been reached by any existing ANN.
The structure of a single processing unit of an ANN, as shown in Figure 1, is a simplified model of a biological neuron.
In analogy to the human nervous until now the
Figure 1: A Processing Unit Reference: [6]
The so called input vector X transports information into the processing unit. Each incoming information is weighted by factors of the weighting vector W before they all get summed up by the input function Σ. F is the activation function which processes the information before passing it to the output function Φ that sends out the computed output value y. [6]
A set of these processing units, that are inter-connected, form the network structure.
B-33
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