Multi-response optimization of cutting force and surface roughness in carbon fiber reinforced polymer end milling using back propagation neural network and genetic algorithm

Abstract

Carbon Fiber Reinforced Polymer (CFRP) composite material has been widely used in various applications, especially in the field of aerospace and aviation, automotive, maritime, and sports equipment manufacturing. CFRP is replacing conventional materials with excellent strength and low specific weight properties. Manufacturing capabilities in a variety of combinations with adjustable strength properties, high fatigue, high toughness and temperature resistance and oxidation resistance capability make this material an excellent choice in engineering applications. As in machining all anisotropic and heterogeneous materials, the failure mechanism also occurs in the machining of CFRP materials. In this research, we conduct End milling machining parameter optimization on CFRP using Back Propagation Neural Network (BPNN) and Genetic Algorithm (GA) to predict the final factory process response. The experiment is carried out with a complete factorial design with 18 combinations (2 x 3 x 3), full factorial use so that we could collect more data, which would improve the accuracy of neural network training results. The measured response is cutting force and surface roughness. From the results of the genetic algorithm, we find that the most optimal network structure is 2 hidden layers with 10 x 2 nodes in each layer. While the prediction error using BPNN is 0.0073%. We conclude that this network structure is capable of predicting the response well so that it can be used to optimize the machining parameters in the CFRP end milling.