An experimental application of machine learning-driven neuro-fuzzy system to predict the wear behaviour of 3D printed bioplastics

Abstract

This study aimed to analyse the wear behaviour of 3D-printed polylactic acid (PLA) samples by machine learning-driven neuro-fuzzy system using digital light processing (DLP). The wear rate and coefficient of friction (COF) in relation to DLP parameters. A Taguchi-based L27 orthogonal design was used to perform a pin-on-disc wear test. The PLA samples with a lower light intensity, shorter exposure time and a 90° orientation yielded a lower COF at a lower load and a higher velocity. The PSI-integrated COPRAS method was employed for multi-objective optimisation. The results of the COPRAS method suggested that the optimal parameters for the improved wear performance of the 3D printed PLA samples were a light intensity of 120%, a 45° orientation, an exposure time of 14 s, an applied load of 5 N and a sliding velocity of 1 m/s. The results of the present study indicated that the machine learning-driven neuro-fuzzy system with DLP could efficiently predict the wear behaviour of 3D-printed bioplastics.