集水區可能最大洪流量之研究

A Study of the Probable Maximum Flood discharge Forecasting for the Catchments

楊元鎮(Yuan-Chen Yang)

指導教授 ： 王安培

中原大學/工學院/土木工程研究所/碩士(2003年)

https://doi.org/10.6840/cycu200300003

摘要

中文摘要

最大洪流量的發生與最大降雨有著密不可分的關係。本文嘗試以可能最大降雨量，配合雨型設計和類神經網路來推估集水區可能的最大洪流量。
首先，以類神經網路(Artifical Neural Network，簡稱ANN)來推測石門水庫集水區在颱風來臨時的洪水流量。先架構一個四層神經網路，其中包含了一個輸入層、二個隱藏層以及一個輸出層的倒傳遞神經網路(簡稱BPN)。將各雨量站之逐時回報的雨量資料做為輸入值，集水區流量做為輸出值。利用倒傳遞運算法則可以修正調整ANN的權值建立降雨—逕流模式，進而預測洪峰到達時間與洪峰量。
本文統計近二十年颱風雨量資料，依設計雨型之方法設計出各雨量站的雨型。最後以石門水庫為例，依可能最大降雨量，做為降雨—逕流模式之輸入值，推估石門水庫在各個延時可能發生的洪流量。模擬之結果發現，以延時48小時之最大降雨量1994 mm，其最大洪流量約為12000 cms。

關鍵字

類神經網路 ； 雨型設計 ； 可能最大降雨

並列摘要

Abstract

There is a very strong relationship between the occurrence of maximum flood discharge and the maximum rainfall. In this study, was calculated based the maximum probable rainfall, the raintype design and Artifical Neural Network (ANN)were applied to calculate the maximum of flood discharge in the basin.
First of all, ANN was used to estimate the discharge in Shihmen basin during the typhoon period. A four layer ANN, which included an input layer, two hidden layers and one output layer for the Back Propagation Network (BPN), was built to predict the discharge in Shihmen basin. The hourly rain datum of rain_stations were used as the input data, and the hourly discharge of Shihmen reservoir as the output data. The weights of rain-runoff system were adjusted by using BPN in order to estimate the time of flood peak and quantity of discharge.
Next, the rainfall data and raintype was used to design the probable maximum precipitation of Shihmen basin, calculated by Water Conservancy Agency. Finally the rain-runoff system of Shihmen basin, the probable maximum input discharge of Shihmen reservoir in each duration was estimated.
It is hope that the result of this study can be taken as the operation rules of Shihmen reservoir.

並列關鍵字

raintype design ； ANN ； probable maximum precipitation

參考文獻

34. 林國峰、張守陽、李汴軍，1993，「台灣地區雨型之研究（二）」，國立台灣大學水工試驗所研究報告第163號。

1. Asaad, Y.S.,Kieran M.C. and Liang, G.C.,1997.”Methods for combining the outputs of different rainfall-runoff models”,Journal of Hydrology,p.203-229.

2. Asaad, Y.S.,1997.”Application of a neural network technique to rainfall-runoff modeling”,Journal of Hydrology,p.272-294.

3. Cameron, M.Z., Donald H.B. and lobodan P.S.,1999.”Short term stream flow forecasting using artifical neural networks”,Journal of Hydrology,p.32-48.

5. Chang,P.T.and Lee,E.S.,1994.”Fuzzy linear regression with spreads unrestricted in sign”,Computers Math,Application 28 No.4, pp.61-70.