Predicting crystallite size of Mg-Ti-SiC nanocomposites using an adaptive neuro-fuzzy inference system model modified by termite life cycle optimizer

Abstract

In this study, Mg-Ti-SiC composite powders with varied micron and nano silicon carbide (SiC) particle sizes were fabricated utilizing the ball milling technology at various milling times. The effect of reinforcement particles sizes and milling time on the morphology and microstructure of the magnesium composite powders was characterized. Then, we developed a machine-learning model based on Adaptive Neuro-fuzzy Inference System (ANFIS) modified with termite life cycle optimizer to predict the crystallite size of the produced composites. The average particles size in all composites including micron SiC (µSiC) and nano SiC (nSiC) always decreased with increasing milling time and SiC content, and the most optimal reduction in particle size was achieved after 16 h of milling for both configurations, which were 5.12 µm and 1.96 µm, respectively. Changing reinforcement particle size from micron to nano caused the peak intensities of Mg and Ti more decreased and phases Ti5Si3 and TiC were observed after milling for 16 h in ND composite powder. With increasing milling time in Mg-25 wt% Ti-5 wt% µSiC, the crystallite size decreased from 31 nm to 13.62 nm after 1 h and 32 h milled, respectively. The most optimum reduction in crystallite size occurred in the composite powders including nSiC, in which crystallite size decreased to 13.35 nm. The developed Machine learning model was able to predict the evolution of the crystallite size of the produce d composites with very good accuracy.

Keywords

• Machine learning
• Termite Life Cycle Optimizer (TLCO)
• Magnesium composite
• Silicon carbide
• Crystallite size
• Ball milling