A Two-stage Artifitial Neural Network Model to Predict the Shrinkage of a Polystyrene Matrix Reinforced with Silica Sand and Cement

Prediction of the shrinkage for the manufacturing purposes of composite materials is not an easy task. The use of existing mathematical and statistical tools may help in solving part of the problem.On the other hand, artificial network tools are of a great importance too. In this investigation, a two-stage Artifitial neural network was used to predict the amount of shrinkage. Using an experimentally measured values of the shrinkage under different material and processing parameters to judge about the relevance of the developed model, it was found that the two-stage Artifitial neural network approach is more capable of predicting the shrinkage than the analytical models because the latter lacks consideration of the materials and processing variables.