Predictive Model for Specific Energy Consumption in the Turning of AISI 316L Steel

Abstract

This article presents an approach for the simulation of machining operations through Artificial Intelligence, which guarantees an automatic learning of the distinctive features in the processes of metal cutting. In the research, an Artificial Neural Network was designed, which establishes the relationships between the parameters of cutting regime and the technological indexes of machining, based on the information generated in real experimentation. For the conception of suitable cutting strategies, the following magnitudes were considered for the input of the model: lubrication regime, cutting speed, feed rate and machining time; which determined the behavior of the cutting forces in the turning of the AISI 316L steel, in order to obtain the cutting powers that define the specific energy consumption. Several designs were considered according to the features of Multi-Layer Perceptron architecture and the selected model was evaluated according to the mean square error and the regression coefficient R2, reflecting high precision in the approximation. The deviation for the error made in the estimation of the cutting force values represents approximately 2% of the average value. These results showed a good level of reliability in the prediction of energy consumption under various machining conditions, in order to adopt relevant saving measures.