|
碩士論文公告區 |
||
| 年度： | 93 | |
| 姓名： | 劉秀鳳(Hsiu-Feng Liu) | |
| 論文題目(中)： | 應用類神經網路評估台14線公路邊坡崩壞潛能之研究 | |
| 論文題目(英)： | Evaluation of Slope Failures along the 14th Provincial Highway Using Artificial Neural Networks | |
| 摘要(中)： | ||
| 台灣位於環太平洋地震帶，地震發生機率頻繁。而地震後容易造成邊坡土質鬆動，以至於颱風或豪雨過後，總帶來許多邊坡崩塌災害。省道台14線為聯繫台灣東、西部之主要幹道，亦是一條重要的觀光道路，其沿線之邊坡經常發生崩塌災害。先前現地調查發現在31k+800~75k+200路段有13處易發生崩塌之邊坡，但以本路線之崩塌潛勢為主題之相關研究並不多，故選擇台14線31k+800~75k+200路段為本研究案例，調查沿線13處易發生崩塌邊坡，研究各影響因子對其崩塌潛勢之影響，利用類神經網路所具有之非線性及平行處理能力，來處理各項參數對邊坡坍塌之影響，預測邊坡是否可能發生崩塌。 研究結果顯示，經由網路之訓練與測試，證明類神經網路對13處易崩塌邊坡462筆全部案例崩壞與否之正確判釋率達90%。而由網路敏感度分析結果發現，崩塌當日之降雨量對邊坡崩塌影響最大，由此可看出在無人為破壞之情況下，降雨乃造成邊坡崩塌之主因，這與一般現實狀況相符合。與高嘉隆（2003）之研究結果比較發現，不論是利用不安定指數法或類神經網路所分析之結果都以「坡度因子」之影響為最大，表示兩種分析方法之結果有其一致性。總之，以類神經網路模式預測邊坡的崩壞潛能可得良好之準確性，故可用來探討各因子對邊坡發生崩塌之影響程度。 |
||
| 摘要(英)： | ||
| Taiwan is located in the Circum-Pacific Seismic Zone. Earthquakes took place frequently, and natural slopes were often weaken after the earthquakes. Consequently, a lot of slopes collapsed after torrential rains during typhoon seasons. The 14th provincial highway is one of the main arterial roads connecting the middle and eastern parts of Taiwan, also is an important tourism highway in central Taiwan. Several slopes along this highway also collapse frequently. However, not much research toward the failure potential of slopes along this highway has been performed. Therefore, this study investigates the failure potential of 13 slopes along the 31k to 75k section of the 14th provincial highway. Four hundred and sixty-two sets of data were collected and analyzed using Artificial Neural Networks (ANN) method. Among these data, 370 sets were used as training data, and the other 92 sets were used as predicting data. Results of this study indicate that the ANN method can predict the slope failure potential with an accuracy up to 90%. Sensibility analysis indicates that the gradient of slope is the most important factor next to rainfall which induced failure of slopes. The outcomes are coincident with study performed by Chia-Lung Kao in 2003 using Instability Index method. Therefore, it is concluded that the ANN method is a good way to predict the failure potential of slopes. |
||
| 檔案： | 沒有相關檔案 | |
| 相關連結： | http://ethesys.lib.cyut.edu.tw/ETD-db/ETD-search-c/view_etd?URN=etd-0221105-142153 | |