Prediction of Hardness of Copper-based Nanocomposites Fabricated by Ball-milling using Artificial Neural Network
Publish Year: 1402
نوع سند: مقاله ژورنالی
زبان: English
View: 61
This Paper With 10 Page And PDF Format Ready To Download
- Certificate
- من نویسنده این مقاله هستم
استخراج به نرم افزارهای پژوهشی:
شناسه ملی سند علمی:
JR_IJE-36-10_017
تاریخ نمایه سازی: 22 مهر 1402
Abstract:
Copper-based alloys are one of the most popular materials in the power distribution, welding industry, hydraulic equipment, industrial machinery, etc. Among different methods for the fabrication of Cu alloys, mechanical alloying (MA) is the major approach due to the fact that this approach is simple, inexpensive, suitable for mass production, and has a high capacity for homogeneous distribution of the second phase. However, the prediction of the hardness of products is very difficult in MA because of a lot of effective parameters. In this work, we designed a feed-forward back propagation neural network (FFBPNN) to predict the hardness of copper-based nanocomposites. First, some of the most common nanocomposites of copper including Cu-Al, Cu-Al۲O۳, Cu-Cr, and Cu-Ti were synthesized by mechanical alloying of copper at varying weight percentages (۱, ۳, and ۶). Next, the alloyed powders were compacted by a cold press (۱۲ tons) and subjected to heat treatment at ۶۵۰˚C. Then, the strength of the alloys was measured by the Vickers microscopy test. Finally, to anticipate the micro-hardness of Cu nanocomposites, the significant variables in the ball milling process including hardness, size, and volume of the reinforcement material, vial speed, the ball-to-powder-weight-ratio (BPR), and milling time; were determined as the inputs, and hardness of nanocomposite was assumed as an output of the artificial neural network (ANN). For training the ANN, many different ANN architectures have been employed and the optimal structure of the model was obtained by regression of ۰.۹۹۱۴. The network was designed with two hidden layers. The first and second hidden layer includes ۱۲ and ۸ neurons, respectively. The comparison between the predicted results of the network and the experimental values showed that the proposed model with a root mean square error (RMSE) of ۳.۷ % can predict the micro-hardness of the nanocomposites.
Keywords:
Authors
R. M. Babaheydari
Department of Materials & Metallurgy, Faculty of Engineering, Shahid Bahonar University of Kerman, Kerman, Iran
S. O. Mirabootalebi
Department of Materials & Metallurgy, Faculty of Engineering, Shahid Bahonar University of Kerman, Kerman, Iran
G. H. Akbari
Department of Materials & Metallurgy, Faculty of Engineering, Shahid Bahonar University of Kerman, Kerman, Iran
H. Salehifar
Department of Computer Science and Automation, Technical University of Ilmenau, Germany
مراجع و منابع این Paper:
لیست زیر مراجع و منابع استفاده شده در این Paper را نمایش می دهد. این مراجع به صورت کاملا ماشینی و بر اساس هوش مصنوعی استخراج شده اند و لذا ممکن است دارای اشکالاتی باشند که به مرور زمان دقت استخراج این محتوا افزایش می یابد. مراجعی که مقالات مربوط به آنها در سیویلیکا نمایه شده و پیدا شده اند، به خود Paper لینک شده اند :