Multi-performance Optimization in End Milling of GFRP Composites Using Backpropagation Neural Network and Differential Evolution Algorithm

Abstract

AbstractAn integrated approach has been applied to predict and optimize multi-performance characteristics, namely, cutting force (CF) and surface roughness (SR), in the end-milling process of glass fiber-reinforced polymer (GFRP) composites. The experiments were performed by varying spindle speed, feeding speed, and depth of cut. The quality characteristics of cutting force and surface roughness were the smaller, the better. Full factorial design 3 × 3 × 3 was used as the design of experiments. Backpropagation neural network (BPNN) was used to model the end-milling experiment and also to determine the objective function. This objective function will be modified into a fitness function optimized by using a differential evolution algorithm (DEA) to find the combination of drilling parameters’ levels that yield minimum cutting force and surface roughness simultaneously. As a result, the minimum cutting force can reduce the energy consumption, and the end-milling process can be performed with higher energy efficiency. Based on BPNN-DEA, the depth of cut of 2 mm, the spindle speed of 4797.5 rpm, and the feeding speed of 579.7 mm/min can simultaneously minimize the cutting force and surface roughness in the end milling of GFRP.KeywordsGlass fiber-reinforced polymerEnd millingBackpropagation neural networkDifferential evolution algorithmEnergy consumption