Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/13009
標題: 類神經網路在長時期潮汐預報之應用
Artificial Neural Network in Long-Term Tidal-Level Forecasting
作者: 謝榮哲
Hsieh, Rong-Jer
關鍵字: Artificial Neural Network;類神經網路;Back-Propagation Network;Tide;倒傳遞類神經網路;潮汐
出版社: 土木工程學系
摘要: 
摘要
本文係藉由倒傳遞類神經網路結合調和方程式,建立可以預報長時期潮汐變化的模式。一般的潮汐預報,往往需要使用長達1年以上的實際觀測資料,以調和分析方法,求得主要分潮,再代入調和方程式或數值模式來預測潮汐水位。然而,本文的預報模式可在有限的資料中,學習過去的潮汐變化特性,進而推算主要分潮,並預測潮汐序列的變化。本文研究結果顯示,使用2個月的實測資料,就能順利地推算出主要分潮,而以半個月的觀測資料作為類神經網路的學習,即可相當準確地預測半日潮、全日潮及混合潮等不同潮汐形態,在未來一年內的潮位變化。

ABSTRACT
Accurate forecasting for tidal-level variation is of great importance for construction installations or human activities in maritime areas. The tidal level could be predicted conventionally by the harmonic analysis based on the least square method. Good resolution in the conventional methods demands a sufficiently long records to ascertain the parameters of the major constituents. Alternatively, while applying the harmonic equation, this paper reports an application of the artificial neural network for forecasting the long-term tide level. The present model can determine the harmonic parameters using a very short-term observed tidal records based on a learning process. Field data of three types of tides, referred as the diurnal, semidiurnal and mixed types, are used to test the performance of the present model. The results show that the major constituents can be determined only using a two-months measured data. The results also present that one-year tidal level forecasting can be satisfactorily achieved using a half-month length of observed data.
URI: http://hdl.handle.net/11455/13009
Appears in Collections:土木工程學系所

Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.