Application of artificial neural network to predict flow stress of as quenched A357 alloy

• Published in: Materials science and technology Vol. 28; no. 2; pp. 151 - 155
• Main Authors: Yang, X W, Zhu, J C, Lai, Z H, Kong, Y R, Zhao, R D, He, D
• Format: Journal Article
• Language: English
• Published: London, England Taylor & Francis 01.02.2012; SAGE Publications; Taylor & Francis Ltd
• Subjects: A357 alloy; Alloys; Aluminum base alloys; Artificial neural network; Artificial neural networks; Back propagation; Compression tests; Flow stress; Mathematical models; Neural networks; Quenching; Quenching (cooling); Strain rate; Temperature effects; Thermal stresses; Yield strength; A357 alloy; Artificial neural network; Flow stress
• ISSN: 0267-0836; 1743-2847
• DOI: 10.1179/1743284710Y.0000000051

In order to establish an accurate thermal stress mathematical model of the quenching operation for A357 alloy, isothermal compression tests were performed in the temperature range of 200-500°C and the strain rate range of 0·001-1 s −1 on as quenched A357 (Al-7Si-0·6Mg) alloy. Based on the experimental results, the deformation behaviour of as quenched A357 alloy was investigated in terms of artificial neural network with a back propagation learning algorithm. Using the deformation temperature, strain rate and strain were used as inputs, and flow stress was used as the output in the network. The average absolute relative error between the predicted results that used the back propagation network and the experimental data obtained from compression tests is 2·89%, which indicates that this artificial neural network model is able to predict the flow stress with high precision. Therefore, it can be used as an accurate thermal stress model to solve the problems of quench distortion of parts.