|
In the article, the object of research is the development of theoretical (software) tools and experimental methods for analyzing vibroacoustic signals emitted by the analyzed objects, based on the theory of regularization of incorrect singular models, the theory of analysis of Fourier spectra and the theory of pattern recognition. A method is proposed for separating the information signal from the field of external noise interference by measuring two-stage noise in the modes of clean air and dusty air. The technique was tested on a specially designed closed circulation stand for studying diagnostic signals of vibroacoustic emission. The received acoustic signal from the piezoceramic sensor, undergoing digital processing, is realized in the form of an integral Fourier transform in the frequency domain with obtaining spectrograms. Carrying out training experiments with clean air and various concentrations of dust monofractions will create a data bank and provide a partition of the feature space into classes using the methods of pattern recognition theory with the definition of dividing boundaries between classes. The proposed technique is aimed at isolating the information diagnostic signal against the background of non-stationary noise interference with an intensity of up to 50 dB. This should be done in two stages. In the first current operational stage, preliminary express diagnostics are performed using a sound level meter. In cases of a justified forecast about the deterioration of the state of the analyzed objects, i.e. increase in dust content, differential diagnostics is performed using spectrum-analyzing equipment according to the methods of the theory of pattern recognition on a computer. It is possible to recognize and diagnose the dustiness of the flow after a series of identifying experiments and their processing according to the algorithm. Therefore, the task of research is to prepare a theoretical base (software) and an experimental base for the second stage of diagnostics.
|
