IDENTIFICATION OF MECHANICAL ENGINEERING OBJECTS AT THE SAME TIME

Abstract

It is shown that the diagnostics of engineering facilities involves the justification for the expansion of the applied physical effects, oriented primarily to non-destructive testing of the parameters of engineering products. Based on studies that show the promise of experimental and theoretical evidence, the expediency of finding information on the frequency spectra of the resonant acoustic signals of the studied objects caused by broadband resonators of equal amplitude in the acoustic frequency range is given. Examples of the use of broadband emitters for nano-amplitude impacts on the studied objects with the aim of acoustic spectroscopy and the creation of their identification models are provided. For practical use of experimental results to identify dimensional characteristics of diagnosed objects, the authors applied neural network models. Such models serve the practical purposes of diagnosing the state of objects based on the frequency spectra of natural resonance oscillations. It is proved that if the object is diagnosed relative to the reference signal in the form of a broadband constant amplitude, then this approach allows normalizing the output diagnostic signals relative to the reference. Identification models based on a neural network basis are presented, which showed the real possibility of using them to create a system for diagnosing objects using several quantitative criteria. Moreover, the number of such signs that can be controlled simultaneously is practically unlimited. The authors of the work conducted additional studies that showed the possibility of simultaneous monitoring of not only the geometric characteristics of objects, but also their physical and mechanical characteristics, including indicators of stress state, hardness, etc.