Using an Artificial Neural Network for Nondestructive Evaluation of the Heat Treating Processes for D2 Tool Steels

Autor: Saeed Kahrobaee, Mehrdad Kashefi, Sadegh Ghanei
Rok vydání: 2019
Předmět:
Zdroj: Journal of Materials Engineering and Performance. 28:3001-3011
ISSN: 1544-1024
1059-9495
DOI: 10.1007/s11665-019-04057-4
Popis: In nondestructive evaluation (NDE) of heat-treated steels, different variables of the heat treating process have complex effects on outputs of NDE methods, and hence, the effect of one desired variable on NDE outputs should be evaluated to help interpreting the changes. In the present paper, the potential of the magnetic hysteresis method was evaluated for simultaneous detection of the austenitizing and tempering temperatures of AISI D2 samples parts subjected to different heat treatment conditions. To produce the microstructural changes, five groups of the samples were austenitized at 1025-1130 °C, quenched in oil and finally each group was tempered in the range 200-650 °C. SEM and x-ray diffractometry techniques were used to characterize different produced microstructures. For accurate and simultaneous prediction of tempering and austenitizing temperatures, an artificial neural network (ANN) was implemented for magnetic hysteresis outputs including magnetic saturation, coercivity and maximum differential permeability. The study revealed that the magnetic NDE system coupled to ANN has the ability to be adopted as an effective expert NDE tool to predict heat treatment effects on D2 tool steels.
Databáze: OpenAIRE
Externí odkaz: