CN-122020260-A - Precision bearing full life cycle fault early warning method based on multi-source vibration signal decoupling

CN122020260ACN 122020260 ACN122020260 ACN 122020260ACN-122020260-A

Abstract

The invention belongs to the technical field of bearing fault monitoring, and particularly relates to a precision bearing full life cycle fault early warning method based on multi-source vibration signal decoupling, which comprises the steps of obtaining vibration signals of three channels of a bearing, and calculating impact liveness according to local energy of all micro segments of a space vector mode sequence and the times that a space vector mode in the space vector mode sequence is larger than the average value of the space vector mode sequence in a current time window; the method comprises the steps of calculating an autocorrelation coefficient and a bias coefficient of a space vector mode sequence, obtaining background noise coupling strength, obtaining envelope spectrum entropy of the space vector mode sequence, weighting the envelope spectrum entropy by utilizing impact liveness and the background noise coupling strength to obtain a fault characteristic index, and outputting grading early warning based on an early warning threshold and fitting slopes of the fault characteristic indexes of a plurality of time windows in succession. The invention overcomes the defect of covering weak features by strong background noise, obviously reduces false alarm of early missing report, and realizes high-precision monitoring of the whole life cycle.

Inventors

ZUO BAOJUN
XU YONGJIN
WU KAI
FAN CHUNJU
WANG HONGWANG

Assignees

山东黑石轴承科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (10)

1. The precision bearing full life cycle fault early warning method based on multi-source vibration signal decoupling is characterized by comprising the following steps: Obtaining vibration signals of three channels of a bearing and dividing a plurality of time windows; The method comprises the steps of calculating a space vector module of a vibration signal at each moment in a current time window to obtain a space vector module sequence, dividing the sequence into a plurality of micro-segments, forming local energy of each micro-segment by the sum of the space vector modules in each micro-segment, and calculating impact activity according to the standard deviation of the local energy of the micro-segments and the times that the space vector module in the sequence is larger than the average value of the sequence; calculating an autocorrelation coefficient and a bias coefficient of the sequence, and calculating the coupling strength of background noise according to the autocorrelation coefficient and the bias coefficient; the sequence is subjected to Hilbert transformation to obtain an envelope spectrum and an envelope spectrum entropy is calculated, and the background noise coupling strength and the impact liveness are used for weighting the envelope spectrum entropy to obtain a fault characteristic index; and carrying out grading early warning based on the comparison result of the fault characteristic index and the acquired early warning threshold value, the fitted slope values of the fault characteristic index and the previous continuous multiple time windows.
2. The precision bearing life cycle fault early warning method based on multi-source vibration signal decoupling of claim 1, wherein the impact liveness satisfies the expression: ; In the formula, Impact liveness of the current time window; 、 Standard deviation and mean of local energy of all micro-segments in the current time window; the method comprises the steps that the number of times that a space vector module is larger than the average value of a space vector module sequence in a current time window is counted, and the ratio of the space vector module to the time window length is calculated; Is a coefficient of sensitivity.
3. The precision bearing life cycle fault early warning method based on multi-source vibration signal decoupling according to claim 1, wherein the obtaining mode of the autocorrelation coefficient comprises the following steps: Obtaining the sampling point number of one engagement period of the bearing and recording as The front of the sequence The sampling points are used as reference subsequences, the first is And from the first to the second The sample points act as a lag sub-sequence, And calculating the Pearson correlation coefficient of the reference subsequence and the lag subsequence to obtain an autocorrelation coefficient.
4. The precision bearing full life cycle fault early warning method based on multi-source vibration signal decoupling according to claim 1, wherein the skewing coefficient is obtained through third-order central moment calculation.
5. The precision bearing life cycle fault early warning method based on multi-source vibration signal decoupling of claim 1, wherein the background noise coupling strength satisfies the expression: ; In the formula, The background noise coupling strength of the current time window; The autocorrelation coefficient of the space vector mode sequence in the current time window; the method comprises the steps of obtaining a bias coefficient of a space vector mode sequence in a current time window; Is a weight coefficient; Taking an absolute value; as a natural exponential function.
6. The precision bearing life cycle fault early warning method based on multi-source vibration signal decoupling of claim 1, wherein the calculating envelope spectrum entropy comprises: the sequence is subjected to Hilbert transformation demodulation to obtain an envelope signal, the envelope signal is subjected to fast Fourier transformation to obtain an envelope spectrum, the probability of occurrence of the amplitude of each frequency point on the envelope spectrum is obtained, and the entropy of the envelope spectrum is calculated by combining an information entropy formula.
7. The precision bearing full life cycle fault early warning method based on multi-source vibration signal decoupling of claim 1, wherein the fault signature index satisfies the expression: ; In the formula, The fault characteristic index is the fault characteristic index of the current time window; Impact liveness of the current time window; The background noise coupling strength of the current time window; the envelope spectrum entropy of the current time window; Is a penalty coefficient.
8. The precision bearing life cycle fault early warning method based on multi-source vibration signal decoupling according to claim 1, wherein the early warning threshold obtaining mode comprises the following steps: And intercepting fault characteristic indexes in an initial time window after the precise bearing is put into operation to construct a reference sequence, and taking the sum of products of the average value of the reference sequence and the standard deviation of the preset safety coefficient and the reference sequence as an early warning threshold.
9. The precision bearing life cycle fault early warning method based on multi-source vibration signal decoupling of claim 1, wherein the step early warning comprises: recording the fault characteristic index of the current time window as The early warning threshold is The fitting slope of the fault characteristic indexes of the current time window and the continuous multiple time windows before the current time window is as follows If (1) And is also provided with Outputting red early warning signal if But is provided with Or alternatively But is provided with Outputting yellow early warning signal if And is also provided with The early warning signal is not triggered.
10. The precision bearing full life cycle fault pre-warning method based on multi-source vibration signal decoupling according to claim 9, wherein the fitted slope is obtained by least square linear fitting of the fault characteristic indexes of the current time window and a plurality of time windows in succession before the current time window.

Description

Precision bearing full life cycle fault early warning method based on multi-source vibration signal decoupling Technical Field The invention relates to the technical field of bearing fault monitoring. More particularly, the invention relates to a precision bearing full life cycle fault early warning method based on multi-source vibration signal decoupling. Background With the rapid development of high-end equipment manufacture, the precision bearing is used as a core component of a rotary machine, the health state of the precision bearing directly determines the machining precision and operation safety of the equipment, the precision bearing usually experiences long running-in period, stable period and degradation period in the whole life cycle, the bearing usually only generates extremely weak spalling or pitting corrosion in the early degradation period, the weak fault characteristic signals are extremely low in energy and are often submerged in the strong background noise generated by the equipment operation and the coupling interference of multi-source vibration signals, if the weak characteristics cannot be accurately identified and decoupling early warning can be carried out in the early stage, and the service life of the equipment is shortened once the advanced severe wear stage is entered. At present, fault monitoring of bearings mainly depends on vibration signal analysis, and a fixed threshold is adopted to judge the state of the bearings, and the basic logic of the method is to assume that background noise is Gaussian white noise, and high-frequency or low-frequency interference is filtered through frequency domain filtering, so that obvious fault characteristic frequency can be identified. Because early failure impact energy is extremely weak, complex nonlinear coupling occurs on a transmission path of a multi-source vibration signal, deterministic mechanical interference and sudden failure impact cannot be effectively separated by traditional frequency domain filtering, strong background noise such as gear meshing is often non-Gaussian colored noise with strong periodicity, and complex modulation coupling exists between the strong background noise and bearing failure frequency, so that the prior art is often subjected to missing report or false report in early bearing degradation, an accurate health degree evolution model is difficult to construct, and accurate trend early warning of a full life cycle cannot be realized. Disclosure of Invention In order to solve the technical problems that the early weak fault characteristics of the precision bearing are easy to be covered by multi-source strong background noise coupling, so that the full life cycle early warning is not timely and the false alarm rate is high, the invention provides a precision bearing full life cycle fault early warning method based on multi-source vibration signal decoupling, which comprises the steps of obtaining vibration signals of three channels of the bearing and dividing a plurality of time windows; the method comprises the steps of calculating a space vector module of a vibration signal at each moment to obtain a space vector module sequence in a current time window, dividing the sequence into a plurality of micro-segments, forming local energy of each micro-segment by the sum of the space vector modules in each micro-segment, calculating impact activity according to the standard deviation of the local energy of the micro-segments and the times that the space vector modules in the sequence are larger than the average value of the sequence, calculating an autocorrelation coefficient and a bias coefficient of the sequence, calculating background noise coupling strength according to the autocorrelation coefficient and the bias coefficient, performing Hilbert transformation on the sequence to obtain an envelope spectrum, calculating envelope spectrum entropy, weighting the envelope spectrum entropy by utilizing the background noise coupling strength and the impact activity, obtaining fault characteristic indexes, and carrying out grading early warning based on comparison results of the fault characteristic indexes and the obtained early warning threshold values, and slope fitting values of the fault characteristic indexes of a plurality of time windows. According to the method, the impact activity of local energy and the background noise coupling strength based on the autocorrelation and the bias coefficient are respectively evaluated in a time domain, sudden fault impact and Gaussian colored background noise are effectively distinguished, in a frequency domain, the noise and the impact strength are used as mutually exclusive weighting factors to dynamically correct the envelope spectrum entropy, the pure fault characteristic index is adaptively demodulated from a multi-source coupling signal, and the false alarm and false alarm rate in early degradation are remarkably reduced by combining the threshold value and the dual