Estimation of Wear Resistance in Acid Solution of Dental Ceramics by Neural Network

• Published in: Acta stomatologica Croatica Vol. 36; no. 3; p. 343
• Main Authors: Lisjak, D., Čurković, L., Živko-Babić, J., Jakovac, M.
• Format: Web Resource
• Language: English
• Published: Stomatolški fakultet Sveučilišta u Zagrebu, Hrvatsko Stomatološko društvo - Hrvatskog liječničkog zbora 15.09.2002
• ISSN: 0001-7019; 1846-0410

It is known that exposure to acid causes damage to the glass surface. The aim of this study was to examine wear resistance, measuring the mass change of dental ceramics after contact with 10-3 mol dm-3 HCl at temperature of 50°C. Four samples of dental ceramics were analyzed: feldspatic ceramic, hydrothermal ceramic, glass ceramic for staining and glass ceramic for layering. The mass concentrations of eluted Na+, K+ and Ca2+ were determined by ion chromatography (IC) and mass concentrations of Si4+ and Al3+ by UV/VIS spectrometry. Measurements were conducted after 1, 2, 3, 6 and 12 months of emersion. For the subject issue, using experimental data, the feedforward backpropagation neural network for estimation of wear resistance of dental ceramics was modelled. The results of 1, 2 and 12 months of emersion were used for the training 13-20-5 model of neural network. Comparison of experimental data and data obtained by estimation (results of 3 and 6 month intervals) of neural network shows that the applied network model provided a very good prediction of wear behavior of dental ceramics with high correlation coefficient (R) and low sum of squared error (SSE) between measurement and estimated output values.