Optimisation of chemical composition of high speed steel with high vanadium content for abrasive wear using an artificial neural network

Xu Liujie,Xing Jiandong,Wei Shizhong,Zhang Yongzhen,Long Rui

DOI: https://doi.org/10.1016/j.matdes.2005.10.015

2007-01-01

Abstract:The wear weight loss were measured by pin-disk abrasive wear machine after high speed steels with V=5–10% and C=1.66–3.3% were quenched at 1050°C, and tempered at 550°C. By the use of back propagation (BP) network, the non-linear relationship between the wear weight losses (W) and carbon contents, vanadium contents (C, V) has been established on the base of dealing with the experimental data. The results show that the well-trained BP neural network can predict the wear weight loss precisely according to carbon contents and vanadium contents. The prediction results show the optimal V and C contents for abrasive wear are 9–10% and 3–3.4%, respectively. And the prediction values have sufficiently mined the basic domain knowledge of relationship between abrasive wear property and chemical composition of alloys. Therefore, a new way of optimising chemical composition for wear of materials has been provided by the authors.

materials science, multidisciplinary