Prediction and Process Analysis of Tensile Properties of Sinter-Hardened Alloy Steel by Artificial Neural Network

Abstract

Sinter-hardening is an emerging powder metallurgy process by which the consolidation of powder and the hardening of dense bulk samples are integrated into one step. In this study, to understand the complex effects of sinter-hardening parameters on the properties of the Fe-Cr-Ni (Cu)-C alloy, an artificial neural network (ANN) with the topology of a nonlinear multi-layered perceptron was designed to predict the ultimate tensile strength and elongation, considering parameters including chemical composition, sintering temperature, and cooling rate. The predictability of the ANN was verified by experiments, indicating that this method is adequate to quantitatively ascribe steel properties to powder metallurgy parameters in the view of improving process robustness.