Multi response prediction of end-milling CFRP with backpropagation neural network

Abstract

The use of CFRP composite materials has experienced a significant increase in the last few years. Such an increase is influenced by the superior mechanical properties of composite materials, such as high strength-to-weight ratio, wear resistance, rust resistance, high stiffness, and good dimensional stability. The milling process is one of the important machining processes in making components from composite materials. This process is used to form surface contours and obtain accurate product dimensions at the final stage. Unlike in metals, the machining process on CFRP composite materials is very complicated. Some common difficulties include high tool wear rates, delamination, and rough surface results. To obtain composite materials from machining process, it is necessary to select the correct parameters, such as tool geometry and tool material, spindle speed, depth of cut, and feed rate. The present study investigated the effect of depth of cut, spindle speed, and feed rate on surface roughness and delamination on the end-milling process of CFRP. It used a full-factorial 2x3x3 experimental design. The machining parameters were two levels of depth of cut, three levels of spindle speed, and three levels of feed rate. The modeling process used Back-Propagation Neural-Network method. This study found that the optimum neural network architecture is 3 x 3 x 2, which successfully predicts the response of surface roughness and delamination as indicated by MSE of 2.49%.