Eventos Anais de eventos
COBEM 2021
26th International Congress of Mechanical Engineering
PREDICTIVE ASSESSMENT OF THE TURNING TOOL WEAR USING ARTIFICIAL INTELLIGENCE
Submission Author:
Ed Claudio Bordinassi , SP
Co-Authors:
Adalto Farias, Vanessa Seriacopi, SERGIO LUIS RABELO DE ALMEIDA, Ed Claudio Bordinassi
Presenter: Vanessa Seriacopi
doi://10.26678/ABCM.COBEM2021.COB2021-1359
Abstract
Growing developments have been designed in terms of models and algorithms, which involve Artificial Intelligence, and they aim to contribute in the process and machine variables monitoring, by applying Industry 4.0 principles. In this aspect, one of the main challenges consists of monitoring and identifying the point of tool change, which often depends on 100% of the humans. The goal of this work was to determine the wear occurrence of carbide inserts in a machining center, applying turning operations of AISI P20 steel. To reach this purpose, a MTConnect open-source communication protocol was used. Also, a design of experiments based on a composite central design was applied, considering a total of 80 tests, with variation of follow cutting parameters defined based on usual values found in industrial environments: cutting speed; feed rate; cutting depth; and selection of cutting fluid activation. As process outputs, data sets were collected related to the motor powers and loads concerning the spindle, X-axis and Z-axis as a function of the motor use percentages. Moreover, new inserts and artificially worn inserts (corresponding to 0.3 mm of flank tool wear) were utilized in experimental procedure. To analyze the data sets, a model using OLAM (Optimal Linear Associative Memory) neural network was established without previous treatments of these data. From this initial contribution, the results indicated a range between 65 and 90% of correct answers to predict the wear tool according to different conditions. Finally, 3/4 of the data set was destinated for training the neural network, whereas 1/4 was directed to its validation. This work can be seen as an important and useful resource to help the identification of change time of the tool during machining processes.
Keywords
industry 4.0, tool wear, OLAM Neural Network, turning process
