以類神經網路建立建築物風力係數與風力頻譜之估算模式

淡江大學機構典藏 > 工學院 > 土木工程學系暨研究所 > 學位論文 > Item 987654321/87865

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/87865

题名:	以類神經網路建立建築物風力係數與風力頻譜之估算模式
其它题名:	The establishment of wind coefficient and spectrum estimation models for buildings using artificial neural networks
作者:	吳建緯;Wu, Jian-Wei
贡献者:	淡江大學土木工程學系碩士班王人牧;Wang, Jenmu
关键词:	類神經網路;輻狀基底函數;風工程;風力係數;風力頻譜;ANN;RBFNN;Wind Engineering;Wind Coefficient and Wind Force Spectrum
日期:	2013
上传时间:	2013-04-13 11:49:14 (UTC+8)
摘要:	結構物的耐風設計通常需要經由風洞實驗，取得各種風力係數、風力頻譜的實驗數據，其過程相當耗時且費用昂貴。近年，國外風力規範逐步朝資料庫輔助(Database-Assisted)的設計模式發展，使用回歸公式來整理分析實驗數據，常無法得到準確的風力係數，因此，如何更有效的利用風洞實驗氣動力資料庫是一個重要的課題。在風力係數部分，近來重新進行風洞實驗，增加了許多模型(如深寬比 2.5、1.5、0.67、0.4和高寬比3.5、4.5、5.5、6.5)的量測數據，另為配合較為準確的風載重計算模式，嘗試將橫風向、扭轉向不同樓層高度的風力係數加入此次風力係數預測範圍中。在風力頻譜部份，之前淡江大學風工程研究中心的相關研究中，曾應用類神經網路來預測風力頻譜有相當的成效，因此在這次研究上，套用前人之模式方法，重新撰寫RBFNN類神經網路程式，藉由新風洞試驗數據訓練新的類神經網路，探討訓練和驗證案例之分配方式，然後微調網路架構，得到更準確的預測結果。最後將預測結果之類神經網路架構，應用於101年內政部建研所設計風載重資料庫之應用研究，配合發展視窗化高層建築物設計風載重計算軟體之模擬運算，與設計案例之計算結果做進一步的分析探討，初步應用效果良好。 Wind resistant design of buildings oftenneeds to acquire wind coefficents and spectra from wind tunnel tests. Recently, the development of building design wind load standards of other countries has been gradually progresstoward database-assisted design methods. Using regression formulas to process and analyze experimental data of wind coefficients usually are not very accurate. Therefore, one of the most important issue is how to use experimental wind load aerodynamic database more effectively. Wind tunnel test data of new building models, such asD/B 2.5, 1.5, 0.67, 0.4, and H/√A 3.5, 4.5, 5.5, 6.5, were added. In addition, the distribution of acrosswind and torsional windcoefficients were adopted to the new estimation model to increase accuracy. The wind engineering research center of Tamkang University has successfully applied artificial neural networks (ANNs) to simulatewind force spectra. The other part of this research was to follow the previous approach using the new experimental data to train radial basis function neural networks (RBFNNs) to predict spectra. The RBFNN program was updated, the ANN architecture was fine tuned, and the allocating of training and verification data was investigated in order to achieve better accuracy of the model. Finally, the ANN architecture was applied to the 2012 project, Applications of Aerodynamic Database on Building Design Wind Loads, from ABRI, Ministry of the Interior. All the ANNs were coded into the Window based wind load calculation software for wind coefficient and spectrum predictions and calculations. In conclusion, the result of the preliminary application is very well judging from the design cases investigated.
显示于类别:	[土木工程學系暨研究所] 學位論文

文件中的档案:

档案	大小	格式	浏览次数
index.html	0Kb	HTML	199	检视/开启

在機構典藏中所有的数据项都受到原著作权保护.

TAIR相关文章

数据加载中.....