Utilization Of Deep Learning For Wear Analysis Of Mechanical Components Using Self-Oscillating Vibroacoustic Systems

A common method of conducting non-destructive testing of the technical condition of mechanical components is to use the measurement of acoustic waves generated during the operation of machines and devices. The analysis of the frequency spectrum of the signal allows you to estimate the degree of wear of elements such as shafts, bearings or tools used in machine tools. In traditional applications, it is required to make a measurement during the work of the tested element, which is often associated with a time-consuming process of placing measuring devices with the risk of damage to the sensors used, or the health and life of the person carrying out the measurement. For this reason, modern methods of vibroacoustic wear analysis are based on acoustic wave generation systems that allow to obtain accurate measurements without having to take measurements in machinery and equipment operating conditions.The solution to this problem is the use of self-excited systems, which are characterized by lower requirements of measuring equipment and very high sensitivity. The amplitude and frequency of vibrations excited in the system can be constant for a long period of time, which significantly facilitates the collection of measurements. Based on the analysis of the amplitude and frequency of modulated signals generated by the system, it is possible to determine changes in the structure of materials caused by stress or fatigue wear.