The Applications of Neural Networks to Non-Destructive Testing Techniques

Abstract

This study aims to investigate the prediction of eccentric loading and strength of bolts made up of Brass using Artificial Neural Networks (ANN). A simple experimental model has been developed that contains a C-Channel, a base plate and a load plate where Base plate is directly fixed to supporting members that is the C-Channel with the help of studs & the load plate is fixed to this with the help of four Bolts along with the simply supported boundary condition. Use of Artificial Neural Networks to model the eccentric loading and other strength parameters have been explored in this paper. Multi-Layered Neural Networks (MLNN), Back-Propagation Neural Networks are used to associate the Load carrying capacity with the keen consideration of its point of failure or breaking point. Result of experimental study on eccentrically loaded threaded bolt members was used to train and validate the proposed ANN model. And on comparing experimental results with the predicted data obtained from the neural networks we get to know that these networks have achieved good agreement with the training data and have yielded satisfactory generalization. A neural network could be effectively implemented for estimating the eccentric load and other strength parameters. ANN and its methodology can be applied for numerous other science and engineering solutions saving time and effort to predict unknown data accurately.