|
碩士論文公告區 |
||
| 年度： | 91 | |
| 姓名： | 楊竣傑(Chun-Chieh Yang) | |
| 論文題目(中)： | 應用類神經網路預測混凝土受高溫影響之強度折減 | |
| 論文題目(英)： | Prediction of Residual Strength of Heated Concrete Based on Artificial Neural Networks | |
| 摘要(中)： | ||
| 混凝土為一種具有良好耐火性的材料，但在受高溫延時作用後仍會造成材料性質的改變，其中以強度折減為工程上重要評估因素之一。故本研究主要針對混凝土受高溫延時作用後強度以及脈波波速折減程度之行為，並整理出溫度-延時-強度折減三者之間的關係，並對高溫延時實驗結果導入類神經網路觀念，建立以溫度、延時，水灰比或殘餘脈波波速預測混凝土殘餘強度的網路架構，並與現有之溫時分析法加以比較。 研究結果發現火害溫度於320~800℃且延時30分鐘以上混凝土強度和脈波波速折減較為明顯，同時火害溫度550℃以下使用殘餘強度函數來預測火害後強度折減較為準確；而在600℃以上則以類神經網路較為準確。 |
||
| 摘要(英)： | ||
| Concrete is usually considered endurable against high temperatures. The strength of concrete, however, is decreased due to exposure to elevated temperatures over extended period of time. The effect of transient temperature on residual strength of heated concrete is nonlinear and is need of advanced research. Current study employs neural network analysis to construct a non-parametric relationship between transient temperature and residual strength. Two artificial neural networks are proposed and trained based on experimental data and other published data. The inputs of the networks include the maximum exposure temperature, the exposure time, and the water-cement ratio (or residual pulse velocity). The residual strengths predicted by the neural networks agree reasonably well with known experimental results in the literature. Comparison is also made between the predictions of the neural networks and those of the time-temperature analysis. It is concluded that the neural network performs better than the time-temperature analysis at temperatures between 6000 and 800 degree Celsius. | ||
| 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-0808103-025456 | |