A contribution to cylindrical plunge grinding monitoring

Author(s)

Celso Minoru Hara

• Document type: Doctoral Thesis
• Press: Campinas, SP.
• Institution: Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Mecânica
• Defense date: 1999-12-23

Examining board members

Anselmo Eduardo Diniz; João Fernando Gomes de Oliveira; Paulo Correa Lima; Antonio Batocchio; Reginaldo Teixeira Coelho

• Advisor: Anselmo Eduardo Diniz

Abstract

The main goal of this work is to verify the feasibility of employing artificial neural networks to analyse signals from several sensors (acoustic emission, vibration and electric current) attached to the tailstock of a cylindrical plunge grinding machine in order to increase the accuracy of the automatic decision of the wheel dressing momento A second goal is to verify the behavior of surface quality parameters (roughness and circularity deviation) as workpiece speed, feed and wheel sharpness vary. For these purposes, several plunge grinding tests were carried out, grinding ABNT 52100 steel with different cutting conditions. During these operations, the signals from those sensors were sampled and stored in the memory of a computer. The best cutting conditions and the best acquired signals were used in the Back-Propagation neural network training stage. The goal of this network was to establish a moment to interrupt the grinding process to perform the wheel dressing, based in the workpiece quality, using parameters as surface roughness and circularity deviation. The most important conclusions are: the neural network was not able to identify the phenomena occuring in grinding process, and so this is not a good tool to determine the exact moment of dressing; the vibration signal, however, showed a little growth trend as the cutting time increased, but not closely correlated to the workpiece quality parameters. Moreover, there was not any direct relationship either between roughness and workpiece speedor between roughness and feed. Besides, the circularity deviation didn't show direct relationship either with the workpiece speed or with feed and the surface roughness didn't keep itself constant for cutting conditions with equal equivalent chip thickness, in opposition with what the literature traditionally states (AU)