Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/13287
標題: 類神經網路在混凝土配比設計之應用
Application of Neural Networks on Concrete Mix Proportion Design
作者: 李立渠
Li-Chiu, LI
關鍵字: 類神經網路
Artificial neural networks
倒傳遞法
最陡坡降法
Back-propagation Network
The Gradient Steepest Descent Method
出版社: 土木工程學系
摘要: 本研究之主要工作在開發一個電腦程式,以VB6程式語言撰寫兩個執行檔程式(配比設計.exe) 及(預測強度.exe)來完成兩項主題工作,程式(配比設計.exe)可以針對某一特定設計強度提出建議之配比設計,程式(預測強度.exe)可以針對某一特定配比設計預測其混凝土強度。程式(配比設計.exe) 之撰寫乃是先收集針對水利署需求所進行之一系列配比設計試驗結果為基礎,再根據ACI配比設計方法寫出之程式,執行該程式等於根據那些試驗結果執行了一個新的ACI配比設計一樣,此程式求得之配比設計與手算之結果幾乎相同。(預測強度.exe)之撰寫設定了三個預測模式,此三個模式分別採用三個不同資料庫。測試結果,三個預測模式之可信度相近,但由於資料庫中混凝土強度之標準偏差不同,造成預測結果精度不同。前述水利署需求之混凝土性質為:混凝土強度140 ~ 420 kgf/cm2、坍度7.5 cm ~25 cm、最大粗骨材粒徑為3/4 in ~ 1.5 in及卜作嵐材料使用量為30% ,在此一範圍內,程式(配比設計.exe)及程式(預測強度.exe)之第一模式均能提供具有優良精度之參考。
The main purpose of this study is to develop a computer program which includes two parts, Part-I to make concrete proportion design and Part-II to predict concrete strengths. These programs were written using VB (visual basic) language. Program Part-I was written on the basis of test results of a series of specific concrete proportion experiments requested by Water Resource Agency and ACI proportion design method. As a result, performing the program Part-I will have the same results as carrying out a proportion design following ACI Specifications. Program Part-II has three options for predicting concrete strengths based on different selection of three databases. The prediction results show that although reliabilities of the three prediction ways are close, but the standard deviations of the prediction results are quite different because the standard deviations of the three databases are quite different. The property demands of concrete above-mentioned for Water Resource Agency are as follows: Concrete strength : 140~ 420 kgf/cm2 Slump : 7.5 ~25 cm Dmax : 1.5”~ 3/4” Puzzolan : ≦30% In this range, both programs, program Part-I and program part-II with the first option, can offer very precise services.
URI: http://hdl.handle.net/11455/13287
Appears in Collections:土木工程學系所

文件中的檔案:

取得全文請前往華藝線上圖書館



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.