臺灣博碩士論文加值系統

　　本文應用時間序列單變量ARIMA模式及多元變量TFN模式，分別針對烏溪流域大肚橋及其上游三測詀之流量進行分析及比較，且嘗試結合時序即時分析與模糊控制分析，利用時間序列單變量ARIMA之代表模式，產生即時預測分析之參數資料，建構模糊控制規則進行時序參數值之推估，以此模糊時序模式來簡化即時序率分析之步驟，增進即時預測效率；時序分析結果顯示加入上游測詀之TFN模式，其大肚橋流量預測結果顯然比單一測站ARIMA模式更具準確性。其差異尤其在多階前進預測上更為明顯，而模糊時序分析之結果，顯示其的確較時序即時預測分析有效率，且能增加預測之精確度。

　　In this study, univariate ARIMA and multivariate TFN time series models are applied to analysis and forecast the catchment flood discharge on the Wu - Shi basin. And combined Time Series analysis and Fuzzy control analysis, try to simplify the work of real - time analysis by Fuzzy Time Series Analysis. The gauging stations which were discussed in this artecle, included the downstream gauging station, Da - Du bridge and the upstream gauging stations, His - Nan bridge, Nan - Gan bridge and Gan - Feng bridge. The Time Series Analysis forecasting result of Da - Feng bridge. The Time series Analysis forecasting result of Da - Du bridge was shown that TFN model is much precise than ARIMA model, especially on multi - steps ahead forecasting. And Fuzzy Time Series indeed improve complex steps of real - time analysis, and increase the accuracy of real - time forecasting result.