Prediction of dynamic mechanical properties of fibre reinforced composites - an ANN approach

摘要

Dynamic mechanical properties (storage modulus and loss factor) of continuous glass fibre reinforced epoxy composites have been investigated as a function of fibre volume fraction, fibre orientation and different measuring temperatures in this work. Dynamic mechanical thermo-analysis (DMTA) is employed in three-point bending mode. On the basis of experimental results an artificial neural networks approach is employed using MATLAB for the prediction of dynamic mechanical properties. An automated 'Bayesian' regularisation of a back propagation algorithm employed here has the capability of automatically identifying the optimal size of the artificial neural network in its hidden layer. It has been found that artificial neural networks can be trained to predict dynamic mechanical properties of continuous fibre reinforced epoxy composites with a fair degree of accuracy.