HIGH-PERFORMANCE CONCRETE USING ARTIFICIAL NEURAL NETWORKS AND MULTIPLE LINEAR REGRESSION

High Performance Concrete is one of the most important masonry buildings created due to recent developments in concrete industry. The structure of this concrete is complicated and its simulation due to wide variation in chemical compositions and physical characteristics of concrete materials is dicult. Slump Test (settlement) is one of the most important eld experiments to determine the concrete downfall ow. Several studies have indicated that the high performance concrete slump is dependent not only on the water content and the size of coarse aggregate, but also on the other components of the concrete. In this study the high performance concrete slump was modeled, using feed-forward back propagation articial neural networks. An articial neural network (ANN) is a computational model based on the structure and functions of biological neural networks. ANNs are considered nonlinear statistical data modeling tools where the complex relationships between inputs and outputs are modeled or patterns are found. Modeling using articial neural networks requires large amounts of data. The required data for this research has been referenced to the UCI Machine Learning Repository. In this reference a number of dierent data sets on dierent aspects exist. They can be used for related research purposes. The slump test data were used in this research. The workability behavior of HPC is a function of the content of all concrete ingredients, including cement, y ash, blast furnace slag, water, super plasticizer, coarse aggregate, and ne aggregate. The Feed-Forward Back propagation, Cascade-Forward neural network, and multiple linear regression methods were used to model slump and 28-day compressive strength of HPC. The neural network developed in this paper has seven neurons in the input layer, one hidden layer with seven neurons, and one neuron in the output layer. The assessment of results based on root mean square and the correlation coecient shows that the Cascade-Forward neural network model performs well for simulation of high performance concrete behavior. Moreover, the proposed methodology provides a guideline to model complex material behaviors, using only a limited amount of experimental data.