MODELLING OF THERMAL DRILLING OF AA7075 ALUMINUM ALLOYS USING REGRESSION ANALYSIS AND ARTIFICIAL NEURAL NETWORKS TECHNIQUES | ||
| Engineering Research Journal (Shoubra) | ||
| Volume 49, Issue 1, July 2021, Pages 60-66 PDF (1.23 M) | ||
| Document Type: Research articles | ||
| DOI: 10.21608/erjsh.2021.227490 | ||
| Authors | ||
| Ahmad R.M.M. Alenzi; S. S. Mohammed | ||
| Mechanical Engineering Department, Shoubra Faculty of Engineering, Benha University, Cairo, Egypt. | ||
| Abstract | ||
| In the present research, the effects of the tool spindle speed and the conical angle on the hole diameter, bushing height and bushing thickness of the thermally drilled AA7075 aluminum alloy sheets were investigated. The AA75075 Al sheets have 3.4 mm thickness. Three different tools, made from H13 tool steel, with 25o, 30o and 35o conical angles were manufactured. The holes were drilled at different spindles speeds, typically, 3100 rpm, 3400 rpm and 3700 rpm. Both regression analysis (RA) and artificial neural (AN) modeling techniques were used for prediction the effect of the spindle speed and the conical angle on the hole diameter, bushing height and bushing thickness. The results revealed that the spindle rotational speed and conical angle affects the hole diameter, bushing height and bushing thickness. The mean absolute errors for the developed regression models were about 0.0439506, 0.204691 and 0.0595062 for models used to predict the hole diameter, bushing height and bushing thickness, respectively. The hole diameter, bushing height and thickness were successfully predicted using artificial neural network (ANN) modelling. The MLP neural network architecture 2‐7‐3 with Tanh transfer function exhibited the best performance with 83.62% accuracy. There is a good agreement between the measured results and simulated outputs obtained with the ANN modelling. | ||
| Keywords | ||
| Friction drilling; Aluminum Alloys; Thermal drilling; Artificial Neural Network | ||
|
Statistics Article View: 91 PDF Download: 124 |
||
