Analysis of the Vibration Characteristic of an Experimental Turning Lathe Using Artificial Neural Networks

Abstract

Due to the complexity of the tool and cutting geometry in machining, many undesirable effects occur during the cutting. Vibrations affecting the poor surface quality of the workpiece and the tool life are one of the most important undesirable effects. These vibrations causing loss of production should be detected and the most appropriate working conditions in which the machine tools will work more stable should be determined. In this work, an artificial neural network estimator has been designed in order to analyze the oscillating effects on computer numerical control (CNC) lathes. Firstly, experimental studies have been carried out, and then vibration analysis has been performed by using two types of artificial neural networks. The experimental part of this manuscript analyses has been carried out for the different material types, spindle speeds and feed rates which are affecting the vibration. After the experimental part, two different artificial neural network structures have been proposed in order to analyze the vibration data measured from the CNC lathe under the relevant working conditions. As a result of these experimental and simulation-based approaches, it was seen that the proposed RBNN structure has a good performance even in real-time parameters in capturing the vibrations occurring in the CNC lathe.