Artificial neural networks (ANNs) have been employed by various researchers for variety of purposes e.g. modeling, forecasting, classification, and control of complex engineering systems. In this study, ANNs are employed for the estimation of concrete slump using concrete mix proportions. Linear and multiple regression models are also developed for comparison purposes. Many multi-layer feed-forward ANN architectures with back-propagation training algorithm were used for developing ANN models. A wide variety of error statistics was used to determine the best ANN model. The concrete mix constituent and slump data obtained from IIT Kanpur laboratory were employed for model development. The results obtained show that the ANN models consistently outperformed both linear and non-linear regression models in terms of the standard statistical measures during both training and testing data sets.
展开▼