| 摘要 |
<p>A neural network that takes as input an echo signal (e.g. via A/D converter 16) corresponding to an ultrasonic wave emitted underwater and determines water bottom sediment there-from. At least two parameters (e.g. connection weights) are stored (e.g. in memory 174) such that they are associated with positional information to which they correspond (Fig. 5). Positional information is received (e.g. from GPS or a user input) and the corresponding parameters are acquired from the stored data and used by the neural network to characterise water bottom sediment 172. The parameters may be learnt by the neural network in advance for every ocean space or general-purpose parameters may be used. The bottom sediment determining device may be provided by an ultrasound finder and may distinguish sea-bed rocks, sand, mud, stones, etc.</p> |
