Combined GA-ANN approach for prediction of HAZ and bearing strength in laser drilling of GFRP composite

Abstract

Recently, an increasing interest has been emerged on nontraditional drilling methods of FRP materials. This paper aims to study two critical quality criteria pertaining to both surface quality and mechanical properties of laser drilled GFRP laminates. The effect of major laser drilling parameters was investigated on the HAZ and bearing strength through utilizing the combined GA-ANN approach. At first, the experimental design was performed based on a full factorial design with different process parameters, later on the experimental results were used to develop a GA-ANN model. The GA was integrated to the ANN model to determine the optimal ANN architecture through optimizing the number of hidden layer neurons, momentum coefficient and learning rate. The effectiveness of the models was ultimately evaluated on the basis of experimental data and statistical analysis. It was found that the GA-ANN method could predict the HAZ and bearing strength more accurately than trial-and-error ANN. Furthermore, the SEM micrographs from the contact area of the bearing tests were discussed and compared with mechanical drilling method. The results drawn from this study helps to better understanding of the laser drilling process of FRP materials and for predicting the mechanical properties of laser drilled FRP laminates applicable in aircraft part manufacturing.