Estimation of abrasive wear of nanostructured WC-10Co-4Cr TIG weld cladding using neural network and fuzzy logic approach

Abstract

Surface engineering is a great way to make wear-resistant mechanical components and increase their service life in industrial and commercial applications. The behaviour of surface engineering process parameters and their effect on the output is complex and nonlinear. This research aims to use artificial neural networks (ANN) and fuzzy logic techniques to establish a reliable modeling method and estimate the abrasive wear of nanostructured WC-10Co-4Cr TIG weld claddings. In the TIG welding process, the input variables for ANN modeling are weld current, weld speed, argon flow, and standoff distance, corresponding to the output variable for weld claddings' wear resistance. The ideal ANN model architecture has a topology of 4–7-7–1, while fuzzy logic is based on the Mamdani model. The ANN model predictions were more precise with an R-value of 0.999874 than the fuzzy logic model predictions (R-value:0.959). The results showed that the ANN model accurately predicted the association between TIG welding process parameters and abrasive wear.