신경망을 활용한 방파제 파괴확률

Title
신경망을 활용한 방파제 파괴확률
Alternative Title
Failure Probability of Breakwater based on Neural Network
Author(s)
김동현; 박우선
KIOST Author(s)
Park, Woo Sun(박우선)
Alternative Author(s)
김동현; 박우선
Publication Year
2004-02-29
Abstract
Some neural network stability models for rubble mound breakwaters are proposed and analyzed. The proposed models give the more reliable results than the well known van der Meer’s formula. Among them, the neural network model having the slope angle and the wave steepness as independent inputs shows the best performance. But the neural network model having independent input parameter for significant wave height, significant period is found not to be useful for the design of rubble mound breakwater because the design parameters exceed the training data ranges. In addition, a reliability analysis technique using the trained neural network model is proposed. Based on two analysis examples, it was found that van der Meer’s formula gives the larger or smaller failure probabilities than the neural network models. Therefore, one should be very careful in making a decision whether the designed armor units are safe or not if he/she uses only empirical formula. To avoid this situation, it is heavily recommended that more advanced models such as the neural network model proposed here should be simultaneously considered.
URI
https://sciwatch.kiost.ac.kr/handle/2020.kiost/31983
Bibliographic Citation
APAC2003, pp.1, 2004
Publisher
APAC Technical Committee
Type
Conference
Language
English
Publisher
APAC Technical Committee
Files in This Item:
There are no files associated with this item.

qrcode

Items in ScienceWatch@KIOST are protected by copyright, with all rights reserved, unless otherwise indicated.

Browse