English  |  正體中文  |  简体中文  |  Items with full text/Total items : 28611/40652
Visitors : 784301      Online Users : 52
RC Version 4.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
Scope Adv. Search

Please use this identifier to cite or link to this item: http://ntour.ntou.edu.tw:8080/ir/handle/987654321/52867

Title: 類神經網路應用於基隆與龍洞波浪資料之相互推估與預測
Mutual estimations of waves at Keelung and Longdong stations based on artificial neural networks
Authors: Yuan, Shou-Chen
Contributors: NTOU:Department of Marine Environmental Informatics
Keywords: 倒傳遞類神經網路;迴歸;基隆測站;龍洞測站;相關係數;均方根誤差
均方根誤差;均方根誤差;均方根誤差;均方根誤差;均方根誤差;Back-propagation neural network;Keelung station;Correlation coefficient;Root mean squared error;Longdon station;Linear regression
Date: 2018
Issue Date: 2020-01-20T06:20:29Z
Abstract: 目前台灣海域有長期運作的浮標測站有18個,但因海氣象因素或儀器維修等原因,要維持此觀測網不間斷地運作實屬不易。鑑此,本研究將一測站的海氣象資料,以倒傳遞類神經網路預測推估出相鄰位置測站之示性波高及週期。本文以龍洞與基隆兩測站之波浪實測資料,建置數個類神經網路預測模式。模式為將一測站非颱風時期之波浪與風資料,以資料前置時間3小時、6小時、9小時及12小時做為模式中的輸入因子,推估另一測站即時、後1小時、後2小時、後3小時、後4小時、後5小時、後6小時、後9小時及後12小時之示性波高及週期。 結果顯示,二測站相互推算模式以示性波高、平均週期、風速等資料不同前置時間並加入風速分量等因子混合輸入時,類神經網路較能更有效的學習,驗證結果最佳。在推算龍洞測站部分,示性波高相關係數最高可達0.96,均方根誤差為20公分,平均週期相關係數最高可達0.85,均方根誤差為0.35秒;在預測基隆測站部分,示性波高相關係數最高可達0.96,均方根誤差為23公分,平均週期相關係數最高可達0.89,均方根誤差為0.30秒。另再與線性迴歸模式推算結果比較,類神經網路在相關係數與均方根誤差皆明顯優於線性迴歸模式。
Presently, there are eighteen buoy stations in the Taiwan Sea. These stations provide crucial and immediate meteorological and oceanic information. However, operations of these buoy stations may be interrupted due to extreme weather or routine maintenance, resulting in disruption of data. This study developed artificial neural networks (ANN) predictive model for wave height and period using wind and wave information measured at the nearby station. Here, we used the data measured from the two stations on the north coast: Keelung and Longdong. The input parameters for our model include aforementioned parameters measured at -3 hours, -6 hours, -9 hours, and -12 hours and the output parameters are wave height and period at 0 hour, +3 hours, +6 hours, +9 hours and +12 hours. Results showed that models using input parameters with a mixture of different lead times, instead of a fixed lead time, have best results. The correlation coefficients and root mean square errors of the predicted significant wave height for the two stations are about 0.96 and 0.20-0.23 m, respectively. The correlation coefficients and root mean square errors of the predicted wave period are 0.85-0.89 and 0.30-0.35 sec. The ANN models are better than using simple linear regression models. The latter models’ correlation coefficients and root mean square errors are 0.92 and 0.33-0.45 m, respectively, for the wave height and 0.85 and 0.53-0.60 sec for the wave period.
URI: http://ethesys.lib.ntou.edu.tw/cgi-bin/gs32/gsweb.cgi?o=dstdcdr&s=G004044E014.id
Appears in Collections:[海洋環境資訊系] 博碩士論文

Files in This Item:

File SizeFormat

All items in NTOUR are protected by copyright, with all rights reserved.


著作權政策宣告: 本網站之內容為國立臺灣海洋大學所收錄之機構典藏,無償提供學術研究與公眾教育等公益性使用,請合理使用本網站之內容,以尊重著作權人之權益。
網站維護: 海大圖資處 圖書系統組
DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - Feedback