Eventos Anais de eventos
COBEF 2021
11th Brazilian Congress on Manufacturing Engineering
IMAGE APPLICATION AND CONVOLUCIONAL NEURAL NETWORK FOR DETERMINING THE END OF TOOL LIFE OF STEP DRILLS
Submission Author:
Henry Scharf , PR , Brazil
Co-Authors:
Henry Scharf, Heraldo Cambraia, João Morais da Silva Neto, Dalberto Dias da Costa
Presenter: Henry Scharf
doi://10.26678/ABCM.COBEF2021.COB21-0328
Abstract
Drilling with stepped drills for machining complex holes in parts produced on a large scale, which commonly have more than one cross section, recesses and chamfers. The deterioration of these tools is generally evidenced through the wear of flank or crater, chipping, breaks or plastic deformation. However, due to the complex sharpening geometries required in the construction of this type of drill and the variety of types of deterioration, making a decision on the right time to change the tool is not a simple task. Visual inspection is the method normally used, however it depends on the inspector's training, his visual acuity and is still e a counterproductive in automated production systems, as it is necessary to remove the tool from the machine for inspection. There are several possible approaches, which are being studied by the scientific community; among them, the techniques that can be loaded on the machine are highlighted. Whether through indirect methods that use data acquisition that can be correlate with the level of deterioration of the tool or direct methods, such as vision systems. In this context, the present work aims to develop an automatic system that uses images of the tool to determine the end of life of stepped drills when used in the production of large batches of parts, basing the decision-making model on the use of artificial neural network.
Keywords
Stepped drills, Tool deterioration, Image inspection, Artificial neural networks
DOWNLOAD PDF VIEW PRESENTATION
