頁籤選單縮合
| 題 名 | 自組性演算法及其應用於降雨--逕流模式建立之研究 |
|---|---|
| 作 者 | 張斐章; 黃源義; 徐國麟; | 書刊名 | 國立臺灣大學農學院研究報告 |
| 卷 期 | 32:4 1992.12[民81.12] |
| 頁 次 | 頁293-316 |
| 分類號 | 443.42 |
| 關鍵詞 | 自組性演算法; 降雨; 逕流; 模式; 逐步迴歸; 模擬; 洪水預報; Self-organization algorithm; Stepwise regression; Simulation; Flood forecasting; |
| 語 文 | 中文(Chinese) |
| 中文摘要 | lvakhnenkc提出的自組性演算法,GMDH(Group Method of DataHandling), 對於模擬或推估一具有極多變數、函數架構不清楚、且僅有少量輸入--輸出驗證 資料的複雜系統,係一具有效力之工具。惟GMDH法極易衍生出一多元高次的非 線性模式,為改善此一問題,本研究以逐步迴歸的方式修正GMDH演算程序中的 每一個輸出單元,有效地解決參數過多的問題,並因系統較易收斂使輸出模式轉 化為低階非線性方程式。本研究以該法應用於石門水庫上游集水瓦之洪水預報, 其結果顯示對一特定洪水事件之模擬,可獲得相當接近之流量符合;同時,以該 事件之模式及參數值分別推估其它三場颱風事件之洪水預報,亦顯示極佳之效 果,最後,本研究並應用遞迴GMDH程序,以逐時更新調整模式之參數,更見其 於預報上之卓越性。 |
| 英文摘要 | The self-organization algorithm of lvakhrienko, the GMDH method ,is used to model a rainfall-runoff system. This algorithm appears to be very powerful in a system which is complex with many variables andparameters, ill-defined mathematical structure, and limitedinput-output data. Nevertheless, the GMDH procedure easily producesa high order multivariate nonlinear model. For reducing the number of parameters and simplifing the output model, the stepwiseregression is used to each node of GMDH procedure. The result shows that the modified GMDH version is quicklyconverged and the output model is simplified to a low order nonlinearequation. The method is applied to simulate/predict the flood of anupstream watershed in Shihmen reservoir. The results show that therainfall-runoff relation can be fitted almost perfectly for the identifiedevent, and the real time flood forecast for three verified events arewell matched. With a sequential adjustment of the model's parameters,an improved version of the GMDH, the method is shown to be moreprominent |
本系統中英文摘要資訊取自各篇刊載內容。