頁籤選單縮合
題 名 | 時序類神經集水區洪水預測模式=Modeling of Watershed Flood Forecasting with Time Series Artificial Neural Network Algorithm |
---|---|
作 者 | 陳昶憲; 楊朝仲; | 書刊名 | 臺灣水利 |
卷 期 | 46:1=181 1998.03[民87.03] |
頁 次 | 頁84-98 |
分類號 | 443.42 |
關鍵詞 | 類神經網路; 倒傳遞網路; 洪流預報; 時間序列; Artificial neural network; Back-propagation network; Flood forecasting; Time series; |
語 文 | 中文(Chinese) |
中文摘要 | 為了使類神經網路能有效地適用在集水區洪水事件的流量預測上,本文藉由線性 轉換函數法, 配合參數顯著性檢定,設計類神經網路之輸入處理單元數; 並由時序 ARIMA 模式建構各上游流量測站之洪水預測模式,以提供類神經網路進行多時刻洗量預測時所需之 輸入資料。經由烏溪流域實際應用,其下一∼三小時的驗證預測結果表現良好。故可推論此 種設計方法,具有描述集水區洪水傳輸及預報之能力。 |
英文摘要 | In order to forecast the flood discharge of downstream gauging station by the artificial neural netowrk (ANN) algorithm efficiently, the linear transfer function method and parameter significance T-test are proposed to determine the number of the network input elements. In addition, time series ARIMA model for every upstream gauging station are constructed to offer the forecasting discharge which are input data for watershed ANN flood forecasting model. For illustration in Wu-Shi basin, the model verified results of one-step through three-step ahead flood forecasting are good. One may conclude that the algorithm of time series ANN flood forecasting can simulate the phenomena of flood transportation and forecast the flood discharge of watershed efficiently. |
本系統中英文摘要資訊取自各篇刊載內容。