EA034627B1 - Method for vibration diagnostics of rotary equipment to detect defects of rolling bearings - Google Patents

Abstract

The invention relates to the field of vibration diagnostics of rotary equipment using systems and methods to process vibration signals and can be used for early detection of defects in industrial equipment that develop in the course of its operation, enabling timely maintenance and repair. The method for vibration diagnostics of rotary equipment to detect defects of rolling bearings by processing vibration signals is proposed, namely, the vibration signal received from the accelerometer installed on the equipment is subjected to wavelet transform using the basis function ψ(t)where the parameter k sets the decay rate of the exponent, and the parameter ω sets the dominant cyclic wavelet frequency; a matrix of wavelet coefficients is formed and a signal scalogram is built; and the dominant eigenfrequencies of the equipment containing shock processes are determined via searching for scaleogram maxima; sets of wavelet coefficients corresponding to the frequencies found on the scalogram are selected; an envelope is constructed for each selected set of wavelet coefficients using the Hilbert transform, that envelope yields the location of shock pulses in the temporal signal; Fourier transformation of the envelope of the set of wavelet coefficients is calculated to search for a set of frequencies of the bearings in the spectrum and a matrix is formed of the bearing frequencies thus found; comparing the totality of the bearing frequencies retrieved with their template, a conclusion is drawn about the technical state of the bearing.

Description

The invention relates to the field of vibration diagnostics of rotary equipment using systems and methods for processing vibration signals and can be used for early detection of defects in industrial equipment that arise during operation, which will allow for timely maintenance and repair.

A known method of vibration diagnostics of rolling bearings, which consists in analyzing the Fourier spectrum of the envelope of the vibration signal. The vibration signal is subjected to bandpass filtering and its envelope is calculated. The Fourier spectrum of the envelope of the vibration signal is calculated. The main bearing frequencies are calculated. A bearing frequency is searched in the envelope spectrum. The conclusion about the presence of a bearing defect and its severity is based on the similarity of the found spectral components with the pattern of a specific defect [1].

The disadvantage of this method is the strong dependence of the quality of the method on the selected filtering band, which affects the reliability of the diagnostic result.

There are also known methods of vibration diagnostics of rotary equipment, which include the use of spectral kurtosis to estimate the center frequency and passband of a band-pass filter [2, 3].

The disadvantage of these methods is their low sensitivity to incipient defects due to the lack of analysis of the shape of the temporary signal (shock processes).

There is also a method of extracting physiological information based on basis functions and a continuous wavelet transform, which consists in searching for areas of interest in the frequency domain using a set of basis functions and a continuous wavelet transform [4].

The disadvantage of this method is the low sensitivity of the method when detecting incipient defects at an early stage of their development due to the use of basic functions that are not adapted to solve the problems of vibration diagnostics.

The problem solved by the invention is to increase the sensitivity of the method of vibration diagnostics of rotary equipment to detect defects in rolling bearings and increase the reliability of diagnostic results. The problem is solved by the claimed method of vibration diagnostics of rotary equipment for detecting defects in rolling bearings based on the analysis of the spectrum of the envelope of vibration using preliminary wavelet filtering using a basic function optimized to highlight shock processes of rolling bearings.

In accordance with the invention, a method of vibration diagnostics of rotary equipment for detecting defects in rolling bearings by processing a vibration signal is that the vibration signal received from the accelerometer installed on the equipment is subjected to continuous wavelet transform using the basic function (t):

where the parameter k determines the rate of decrease of the exponent, and the parameter ω is the dominant cyclic frequency of the wavelet;

form a matrix of wavelet coefficients and build a signalogram of the signal;

determine the dominant natural frequencies of the equipment containing shock processes, based on the search for maximums of the scaleogram;

sets of wavelet coefficients corresponding to the frequencies found on the scaleogram are selected;

for each selected set of wavelet coefficients, an envelope is constructed using the Hilbert transform, based on which the location of the shock pulses in the time signal is determined;

calculate the Fourier transform from the envelope of the set of wavelet coefficients, search for a set of bearing frequencies in the spectrum and form a matrix of the found bearing frequencies;

by comparing the totality of the found bearing frequencies with the template, a conclusion is drawn about the technical condition of the bearing.

Compared with the known solution [1], the proposed method selects a frequency band for filtering adaptively based on the analysis of a skylogram, which increases the severity of the spectral components in the analyzed frequency band and, therefore, increases the sensitivity and accuracy of vibration diagnostics of rolling bearings.

Compared with the known solutions [2, 3], the proposed method analyzes the shape of a temporary signal, i.e. selects a filtration band containing shock processes of rolling bearings. The proposed method has great sensitivity in detecting incipient defects in rolling bearings.

- 1 034627

Compared with the known solution [4], the basic function of the proposed method is optimized for highlighting the shock processes of rolling bearings, therefore, it has greater accuracy when searching for the natural frequencies of the bearing and, as a result, is more sensitive when detecting defects in rolling bearings.

The inventive method increases the sensitivity of the spectral methods of vibration diagnostics for detecting defects in rolling bearings by 8-12%. It is obvious to a specialist that this diagnostic method can be extended to other elements of rotary equipment, such as gears, couplings, etc.

The invention is illustrated using figures.

FIG. 1 is a diagram of diagnosed equipment with a vibration sensor installed;

FIG. 2 - vibration signal of a rolling bearing 6213 with an outer ring defect;

FIG. 3 - basic wavelet function;

FIG. 4 - a scaleogram based on the wavelet function v _m (t);

FIG. 5 - a scaleogram and found natural frequencies Fr1 and Fr2;

FIG. 6 - Fourier spectrum of the envelope of vibration for the frequency Fr1;

FIG. 7 - Fourier spectrum of the envelope of vibration for the frequency Fr2.

The inventive method is illustrated by a diagnostic example of a bearing 6213 mounted on a test bench schematically depicted in FIG. 1, where 1, 4 - deep groove ball bearing 6213 (bearing 001, bearing 002); 2 - shaft (shaft 001); 3 - electric motor АИР16084, 5 - vibration sensor.

At the first stage, from the accelerometer (vibration sensor) installed on the diagnosed item of equipment (Fig. 1 bearing 002), receive a vibration signal (Fig. 2).

At the second stage, using the wavelet function v _m (t) (1) (Fig. 3) and continuous wavelet transform, a vibro-signal scalegram is calculated (Fig. 4), which describes the frequency distribution of the signal energy (scale of the wavelet transform). It is normalized by amplitude. They search for maxima (ridges) of the skylogram and determine the dominant eigenfrequencies of the equipment (scaled wavelet coefficients V _m (t)) (Fig. 5). For example, for FIG. 2 vibrational signal on the scalegram found natural frequencies Fr1 = 1115 Hz and Fr2 = 4172 Hz.

At the third stage, the Hilbert envelope is calculated for the sets of wavelet coefficients corresponding to the natural frequencies Fr1 and Fr2. The Fourier spectrum for the envelope of each obtained sets of wavelet coefficients is calculated.

At the fourth stage, in accordance with the bearing model and the shaft rotation speed at which the bearing is mounted, the values of the main bearing frequencies are calculated. They search for bearing frequencies in the Fourier spectra of the vibration envelope for the natural frequencies Fr1 and Fr2.

At the fifth step, based on the results of comparing the sets of found bearing frequencies for Fr1 and Fr2 with the templates stored in the database, a conclusion is made about the presence / absence of defects in the rolling bearing and their severity (Fig. 6 and 7). The Fourier spectrum of the envelope of vibration for the natural frequency Fr2 shown in FIG. 7, indicates a pronounced defect in the outer ring of the bearing 6213 (bearing 002), similarity to the pattern 78%. The Fourier spectrum of the vibration envelope for the natural frequency Fr1 shown in FIG. 6, indicates the runout of the shaft, shaft 001, similarity to the 61% pattern. The similarity is calculated as the sum of the weights of the found bearing frequencies.

Sources of information:

[1] Barkov A.V., Barkova N.A. Vibration diagnostics of machinery and equipment. Vibration analysis: textbook. allowance / SPb .: SPbGMTU, 2004 .-- 156 p.

[2] CN 106053069 A. 2016.02.04, G 06 M 13/00.

[3] CN 102866010 B, 2015.10.26, G 06 N 13/04.

[4] US 8226568 (B2), 2012.07.24, G 06 A 15/00; G 11 C 17/00.

Claims (2)

A method for vibration diagnostics of rotary equipment for detecting defects in rolling bearings by processing a vibration signal received from an accelerometer, characterized in that the vibration signal is subjected to continuous wavelet transform using the basic function Vm (t): where parameter k determines the decay rate of the exponential, and parameter ω determines the dominant cyclic - 2,034,627 wavelet frequency; form a matrix of wavelet coefficients and build a signal skeletogram; determine the dominant natural frequencies of the equipment containing shock processes, based on the search for maximums of the scaleogram; sets of wavelet coefficients corresponding to the frequencies found on the scaleogram are selected; for each selected set of wavelet coefficients, an envelope is constructed using the Hilbert transform, based on which the location of the shock pulses in the time signal is determined; calculate the Fourier transform from the envelope of the set of wavelet coefficients, search for a set of bearing frequencies in the spectrum and form a matrix of the found bearing frequencies; by comparing the totality of the found bearing frequencies with the template, a conclusion is drawn about the technical condition of the bearing.