Learning internal representations by error propagation
Parallel distributed processing: explorations in the microstructure of cognition, vol. 1
Neural Computation
Neural networks for pattern recognition
Neural networks for pattern recognition
Neural network design
Introduction to the Theory of Neural Computation
Introduction to the Theory of Neural Computation
Training feedforward networks with the Marquardt algorithm
IEEE Transactions on Neural Networks
| Hi-index | 0.00 |
The tensile capacity of single adhesive anchors depends on many design parameters. Some of these parameters, such as chemical resin type, resin system and anchor bolt type are difficult to quantify in design models. Due to the complexity of developing rational models for estimating the tensile capacity of such type of anchors, most specifications recommend that the performance of these anchors be determined by product-specific and condition-specific testing. In this study, an attempt to predict the tensile capacity of single adhesive anchors using artificial neural networks (ANNs) is presented. A multilayered feed-forward neural network trained with the back-propagation algorithm is constructed using 7 design variables as network inputs and the uniform bond strength of adhesive anchors as the only output. The ANN was trained and verified using the comprehensive worldwide adhesive anchor database of actual tests compiled by the ACI Committee 355. Different modes of failure observed in experiments but bolt breakage are covered by the trained ANN. The predictions obtained from the trained ANN showed that the tensile capacity of adhesive anchors is linearly proportional to the embedment depth as suggested by the uniform bond stress model. The effect of the concrete compressive strength on the tensile capacity of adhesive anchors is product dependent. The results indicate that ANNs are a useful technique for predicting the tensile capacity of adhesive anchors. rs.