以實驗計劃法與類神經網路建構混凝土的坍流度模型

Modeling Workability of Concrete Using Design of Experiments and Artificial Neural Networks

葉怡成(I-Cheng Yeh) ； 陳家偉(Jia-Wei Chen)

《技術學刊》 20卷2期 (2005/06) Pp. 153-162

https://doi.org/10.29507/JT.200506.0006

摘要

在混凝土科技中，工作度的重要性是明顯的。在規範與標準中，用以估計工作度的經驗圖表是基於未添加強塑劑與礦物攙料（如飛灰與爐灰）等材料的混凝土。對添加這些材料的混凝土，這些關係之妥當性應該要加以研究。由於這些關係的高度複雜性，傳統的迴歸分析可能不足以建構精確的模型。類神經網路是建構非線性模型的有效工具。因此，本研究以實驗計劃法與類神經網路建構一個坍流度模型。在這個模型中，坍流度是混凝土所有成份用量的函數，包括水泥、飛灰、爐灰、水、強塑劑、粗骨材，與細骨材。本研究導出下列結論：(1)利用雛形模式找出可疑的實驗數據，並予以重新實驗，對建構精確之模型有非常顯著的助益。(2)類神經網路可以建構一個比二階多項式迴歸分析更精確的坍流度模型。

關鍵字

混凝土 ； 坍流度 ； 實驗設計 ； 模式化 ； 類神經網路

並列摘要

The significance of workability in concrete technology is obvious. The current empirical diagrams and tables presented in codes and standards for estimating workability are based on tests of concrete without supplementary cementitious materials (fly ash, blast furnace slag, etc.). The validity of these relations for concrete with supplementary cementitious materials should be investigated. Because of the high complexity of these relations, conventional regression analysis is not sufficient to build an accurate model. The artificial neural network (ANN) is a powerful tool for modeling complex nonlinear models. Therefore, in this study, a slump flow model has been built using design of experiments (DOE) and ANN. In this model, the slump flow is a function of the content of all concrete ingredients, including cement, fly ash, blast furnace slag, water, superplasticizer, coarse aggregate, and fine aggregate. This study led to the following conclusions: (1)Discovering doubtful experimental data produced by using the prototype model and repeating these experiments is very significantly beneficial for building a reliable model. (2) ANN can build a more accurate slump flow model than a 2-order polynomial regression can.

並列關鍵字

concrete ； slump flow ； design of experiment ； modeling ； artificial neural networks