Application of artificial neural networks to predict the hardness of Ni–TiN nanocoatings fabricated by pulse electrodeposition

Article history: Received 28 September 2015 Revised 2 December 2015 Accepted in revised form 11 December 2015 Available online 12 December 2015 A three-layer backward propagation (BP)modelwas used to predict the hardness of Ni–TiNnanocoatings fabricated by pulse electrodeposition. The effect of plating parameters, namely, TiN particle concentration, current density, pulse frequency, and duty ratio on the hardness of Ni–TiN nanocoatings was investigated. The morphology, structure, andhardness of Ni–TiNnanocoatingswere verified using scanning electronmicroscopy,white-light interfering profilometry, high-resolution transmission emissionmicroscopy, andRockwell hardness testing. The results indicated that the surface roughness of the Ni–TiN nanocoating is approximately 0.12 μm. The average grain sizes of Ni and TiN on the Ni–TiN nanocoating are 62 and 30 nm, respectively. The optimum conditions for fabricating Ni–TiN nanocoatings based on the greatest hardness of Ni–TiN deposits are as follows: TiN particle concentration of 8 g/L, current density of 5 A/dm, pulse frequency of 80 Hz, and duty ratio of 0.7. We conclude that the BP model, with a maximum error of approximately 1.03%, can effectively predict the hardness of Ni–TiN nanocoatings. © 2015 Elsevier B.V. All rights reserved.