Search

Byline: D. Benny Karunakar, G.L. Datta Defects in castings often lead to rejection, which would result in loss of productivity. Expert systems developed by some investigators mostly act as postmortem tools, discussing and analysing a defect after it has occurred. A tool that could predict all the possible defects before the castings are actually made is not yet developed, which would be highly useful in the shop floor. Hence, in the present work, an attempt has been made to predict the major casting defects using back propagation neural networks. The neural network was trained with moulding properties, composition of the charge and melting conditions as the inputs and the nature of the defects as outputs. After the training was over, the set of inputs of the casting that is going to be made, were fed to the network and the network could predict whether the casting would be sound or defective. If defective, the nature of the defect was specified by the neural network.