Minimization of Hole Entry and Exit Surface Delamination on Carbon Fiber Reinforced Polymer (CFRP) Drilling Process Using BPNN-ALNS

Delamination was the primary reason for carbon fiber reinforced polymer (CFRP) parts rejection in the aircraft industry which led to a tremendous amount of profit losses. Therefore, a lot of research aimed to tackle this problem by studying the influence of drilling parameters and their optimization to minimize delamination in CFRP drilling processes. In this study, a backpropagation neural network (BPNN) is deployed to model the CFRP drilling process. Further, adaptive large neighborhood search (ALNS) is utilized to minimize the surface delamination on the CFRP materials' hole entry (SDen) and the hole exit (SDex). ALNS is used to find the best combination of drilling bit geometry (DG), feed rate (FR), and spindle speed (SS). The best obtained BPNN architecture from the trial-and-error method is 2-12-2, which means that there are two neurons in the input layer, twelve neurons in the hidden layer, and two neurons in the output layer tansig activation function type. Optimum CFRP drilling parameters obtained using ALNS optimization are X-type drill bit geometry, 136 mm/min feed rate, and 3000 rpm spindle speed. The minimized SDen and SDex using the optimum drilling parameters are 1.074 and 1.009. © 2024 American Institute of Physics Inc.. All rights reserved.