Implementation of neural network for the thrust force prediction in hot drilling of 6082 aluminium alloy

A multilayered neural network have been implemented for predicting force in hot drilling of the 6082 aluminium alloy. Experimental tests were performed in dry drilling operation, using a conventional milling machine and HSS-Co 8% (DIN338) twist drills, 2.5, 5 and 7 mm in diameter. The spindle speed has been changed in the range 5,000-15,000 rev/min, the feed in the range 0.0076-0.042 mm/rev, the temperature in the range 25-140 ° C. As test temperature increases, a remarkable reduction in thrust forces was observed, low wear and no significant damage of the hole surface was also found. The influence of each parameter was investigated and a neural network was implemented for the force prediction obtaining a good agreement between experimental and numerical results. Copyright © 2010 Inderscience Enterprises Ltd.