期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2018
卷号：96
期号：23
出版社：Journal of Theoretical and Applied
摘要：This work aims to optimize the mix proportions of Reactive Powder Concrete (RPC) mixtures by Artificial Neural Networks (ANN) technique. Ninety-nine sets of RPC mixes with their results from six different sources are used to check the reliability of the model. The values of compressive strength (Fc), Splitting Tensile Strength (Fsp) and Flexural strength (Fr) were specified as the input parameters. The values of sand to powder ratio (S\P), water to powder ratio (W\P) and volume of steel fiber (Vf) are computed and specified as the output parameters. (Fc) model with an architecture Multi Layers Perceptron (MLP) 3-40-1 had (0.95) training performance, (0.4%) training error, (0.93) testing performance and (0.4%) testing error. Fsp model with MLP 4-13-1 has (0.99) training performance, (0.014%) training error, (0.99) testing performance and (0.011%) testing errors. The primary predicting model has the architecture MLP 3-14-3. It also has training performance, training error, testing performance and testing error values of (0.96), (0.8%), (0.93), and (1.2%) respectively. All of the ANN models show very good percentages of correlation between target and output values with very low values of error, and high percentage of matching between targets and outputs, and no clear trend to overestimation or underestimation.
