VIBRATION-BASED INTELLIGENT DIAGNOSTICS OF METAL CUTTING MACHINE TOOLS USING FFT AND CNN ALGORITHMS

Abstract

Modern manufacturing systems require highly reliable and accurate metal cutting machine tools to ensure stable machining quality and production efficiency. One of the major problems in industrial practice is the early detection of defects in spindle units, bearings, and gear transmission systems. This paper presents an intelligent vibration-based diagnostic methodology for evaluating the technical condition of machine tools using Fast Fourier Transform (FFT) and Convolutional Neural Network (CNN) algorithms. Experimental investigations were conducted on an HT-250M lathe machine under various operating conditions. Vibration signals were processed using FFT spectral analysis to identify characteristic defect frequencies. A CNN model was developed for automatic classification of machine conditions based on vibration spectra. The obtained results demonstrated that the proposed methodology allows accurate identification of bearing and gearbox defects at early stages, improves machine reliability, and reduces maintenance costs. The developed system can be effectively applied in predictive maintenance and Industry 4.0 manufacturing environments.