Modelling of machining force in end milling of GFRP composites using MRA and ANN

Abstract

Glass Fibre Reinforced Plastic (GFRP) composites show a tremendous increase in applications due to their superior properties. Some damages on the surface occur due to their complex cutting mechanics in cutting process. Minimization of the machining force is fairly important in terms of product quality. In this study, a GFRP composite material with 15°, 60° and 105° were milled to experimentally minimize the cutting forces on the machined surfaces, using solid carbide end mills with 25°, 35° and 45° helix angles at different combinations of cutting parameters. Experimental results showed that the machining force increased with increasing fibre orientation and feed rate; on the other hand, it was found that the machining force decreased with increasing cutting speed and helix angle of the end mill cutter. In addition, analysis of variance (ANOVA) results clearly revealed that the helix angle of the end mill cutter was the most influential parameter affecting the machining force in end milling of GFRP composites. A model based on an artificial neural network (ANN) is introduced to predict the machining force of GFRP with three different fibre orientations. This model is a feed forward back propagation neural network with a set of machining parameters as its inputs and the machining force as its output. Levenberg–Marquardt learning algorithm was used in predicting the machining force to reduce the number of expensive and time-consuming experiments. The highest performance was obtained by 4-18-18-1 network structure. ANN was notably successful in predicting the damage factor due to higher R2 and lower RMSE and MEP.