Modeling and prediction of surface roughness for running-in wear using Gauss-Newton algorithm and ANN

Authors: Hanief, M. and Wani, M.F.

Journal: Applied Surface Science

Volume: 357

Pages: 1573-1577

ISSN: 0169-4332

DOI: 10.1016/j.apsusc.2015.10.052

Abstract:

In this paper, surface roughness model for running-in and steady state of the wear process is proposed. In this work monotonously decreasing trend of surface roughness during running-in was assumed. The model was developed by considering the surface roughness as an explicit function of time during running-in, keeping other system parameters (velocity, load, hardness, etc.) constant. The proposed model being non-linear, optimal values of the model parameters were evaluated by Gauss-Newton (GN) algorithm. The experimental results adopted from the literature, for steel and Cu-Zn alloy specimens, were used for validation of the model. Artificial neural network (ANN) based model was also developed and was compared with the proposed model on the basis of statistical methods (coefficient of determination (R 2 ), mean square error (MSE) and mean absolute percentage error (MAPE)).

Source: Scopus