Application of Neural Networks for Predicting the Workability of Self-Compacting Concrete
Moosa Mazloom *
Civil Engineering Department, Shahid Rajaee Teacher Training University, Tehran, Iran.
*Author to whom correspondence should be addressed.
Abstract
Self-compacting concrete (SCC) is a complex material and modeling its workability is a complicated task. To evaluate the workability of SCC, five different tests have been conducted, which are slump flow, V-funnel, J-Ring, L-box and U-box. In fact, executing L-box and U-box tests are more difficult than the other ones, especially on sites. Therefore, this research studies the possibility of predicting the results of L-box and U-box tests from the results of the other tests utilizing artificial neural networks (ANN). For this purpose, multi layer perceptron (MLP) networks and radial basis (RB) networks were chosen. The conclusion was that the MLP networks could foresee the L-box and U-box test results in all situations.
Keywords: Concrete, self-compacting, workability, neural networks, perceptron, radial basis