Artificial neural network methodology: Application to predict magnetic properties of nanocrystalline alloys

Abstract

This paper is dedicated to the optimization of magnetic properties of iron based magnetic materials with regard to milling and coating process conditions using artificial neural network methodology. Fe–20 wt.% Ni and Fe–6.5 wt.% Si, alloys were obtained using two high-energy ball milling technologies, namely a planetary ball mill P4 vario ball mill from Fritsch and planetary ball mill from Retch. Further processing of Fe–Si powder allowed the spraying of the feedstock material using high-velocity oxy-fuel (HVOF) process to obtain a relatively dense coating. Input parameters were the disc Ω and vial ω speed rotations for the milling technique, and spray distance and oxygen flow rate in the case of coating process. Two main magnetic parameters are optimized namely the saturation magnetization and the coercivity. Predicted results depict clearly coupled effects of input parameters to vary magnetic parameters. In particular, the increase of saturation magnetization is correlated to the increase of the product Ωω (shock power) and the product of spray parameters. Largest coercivity values are correlated to the increase of the ratio Ω/ ω (shock mode process) and the increase of the product of spray parameters.