Estimation of machinability performance in wire-EDM on titanium alloy using neural networks
- Authors
- Paturi, Uma Maheshwera Reddy; Cheruku, Suryapavan; Salike, Sriteja; Pasunuri, Venkat Phani Kumar; Reddy, N. S.
- Issue Date
- 4-Jul-2022
- Publisher
- Marcel Dekker Inc.
- Keywords
- WEDM; Ti-6Al-4V; roughness; speed; width; MRR; experimental; ANN
- Citation
- Materials and Manufacturing Processes, v.37, no.9, pp 1073 - 1084
- Pages
- 12
- Indexed
- SCIE
SCOPUS
- Journal Title
- Materials and Manufacturing Processes
- Volume
- 37
- Number
- 9
- Start Page
- 1073
- End Page
- 1084
- URI
- https://scholarworks.gnu.ac.kr/handle/sw.gnu/1060
- DOI
- 10.1080/10426914.2022.2030875
- ISSN
- 1042-6914
1532-2475
- Abstract
- The impact of process factors on wire-cut electrical discharge machining (WEDM) performance is complex and nonlinear. In the present work, initially, the WEDM tests were conducted on titanium alloy (Ti-6Al-4V) with eight input factors and four machinability performance parameters. Later, an artificial neural network (ANN) model was established to estimate the WEDM performance. The ANN model with 8-5-5-4 architecture produced a least mean squared error (MSE) and average prediction error (AE) for both training and test data sets. The precision of the ANN model was assessed by relating model predictions with the experimental values. The combined effect of WEDM variables on the machinability performance was illustrated with the help of visual graphs. The R-value (correlation coefficient) of 0.9995 among WEDM test values and ANN estimated values shows the robustness of the developed ANN model in establishing the link between WEDM process factors and machinability parameters. The proposed model helps in minimizing the time for fixing the process parameter values, thereby increasing production and process efficiency.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - 공과대학 > 나노신소재공학부금속재료공학전공 > Journal Articles

Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.