Parameters optimization for 2.5D needled Cf/SiC in longitudinal torsional ultrasonic-assisted laser milling on PSO-BP-PSO

Abstract

The 2.5D needle-punched C_f/SiC was processed by longitudinal torsional ultrasound-assisted laser milling (LTUALM), and the removal form was found to be closely related to the fiber cutting angle in this paper. Compared to conventional machining (CM), laser-assisted machining (LAM), and ultrasonic-assisted machining (UAM), the main cutting force (Fx), radial cutting force (Fy) and axial cutting force (Fz) of LTUALM were reduced by 37.79 %, 22.79 %, 10.32 %; by 68.94 %, 65.89 %, 16.22 %; and by 48.12 %, 37.96 %, 16.96 %, respectively. The reductions in surface roughness were 60.78 %, 32.28 %, and 46.16 %, respectively. The response surface method (RSM) analysis indicated that with a laser power of 349.932 W, an ultrasonic amplitude of 2.849 µm, a cutting speed of 41.699 m/rev, and a cutting depth of 0.040 mm, the surface roughness was minimized to 1.137 µm. Moreover, the surface roughness was optimized by machine learning, and the results showed that the two-time particle swarm optimization for back propagation neural network (PSO-BP-PSO) has significant effectiveness, with the model predicting a minimum surface roughness of 1.119 µm.