CN-122020335-A - Rolling bearing defect detection method and system based on encoder feedback

CN122020335ACN 122020335 ACN122020335 ACN 122020335ACN-122020335-A

Abstract

The invention relates to the technical field of bearing detection, in particular to a method and a system for detecting the defects of a rolling bearing based on encoder feedback, wherein the method comprises the steps of collecting a vibration signal of the rolling bearing to be detected and obtaining a rotating speed signal by a coaxial encoder; selecting defect impact dominant frequency band with spectral kurtosis for vibration signals, bandpass filtering, then carrying out envelope demodulation on the filtered signals to obtain envelope signals, generating speed measuring pulses by the rotating speed signals, determining equal angle sampling marks and establishing time-angle mapping, resampling the angular domain of the envelope signals according to the marks to form an angular domain envelope sequence, carrying out order spectrum analysis on the angular domain envelope sequence to obtain envelope order spectrums, determining characteristic orders of an outer ring, an inner ring and rolling bodies according to bearing structure parameters, extracting characteristic orders and harmonic peaks, calculating three types of order peak value ratio characteristics by full spectral amplitude and normalization, inputting the three types of order peak value ratio characteristics into a multi-core support vector machine model, and outputting defect types. The invention can solve the problems of difficult identification caused by characteristic drift of the bearing under non-stable conditions and strong dependence of the diagnosis process on expert experience.

Inventors

XIONG GUANGJIE
YANG DONGSHENG
LAI JIUCAI
HU HUA
SHEN HUAQIANG

Assignees

深圳锐特机电技术有限公司

Dates

Publication Date: 20260512
Application Date: 20260131

Claims (10)

1. A method for detecting a rolling bearing defect based on encoder feedback, the method comprising: Acquiring a vibration signal of a rolling bearing to be tested, and acquiring a rotating speed signal coaxial with the rolling bearing to be tested based on an encoder; the method comprises the steps of preprocessing the vibration signal, namely determining a target frequency band with dominant defect impact based on spectral kurtosis, and carrying out band-pass filtering on the target frequency band on the vibration signal to obtain a filtered vibration signal; generating a speed measuring pulse sequence of angular domain resampling reference by the rotating speed signal, determining an equal-angle sampling mark according to the speed measuring pulse sequence, and establishing a mapping relation between a time axis and a corner axis; angular domain resampling is carried out on the envelope signal according to the sampling mark, and an angular domain envelope sequence with equal angular intervals is obtained; Calculating the passing order of the outer ring rolling body, the passing order of the inner ring rolling body and the spin order of the rolling body according to the structural parameters of the rolling bearing to be measured, and extracting peaks corresponding to the passing order of the outer ring rolling body, the passing order of the inner ring rolling body, the spin order of the rolling body and harmonic waves thereof from the envelope order spectrum; Calculating the outer-order peak value ratio feature, the inner-order peak value ratio feature and the rolling element order peak value ratio feature, wherein the total peak value sum of the envelope order spectrum is the sum of the amplitude values of all order lines in the envelope order spectrum; the outer ring order peak value ratio is characterized by the ratio of the peak value sum of the passing order and the harmonic wave of the outer ring rolling body to the total peak value sum of the envelope order spectrum, the inner ring order peak value ratio is characterized by the ratio of the peak value sum of the passing order and the harmonic wave of the inner ring rolling body to the total peak value sum of the envelope order spectrum, and the rolling body order peak value ratio is characterized by the ratio of the peak value sum of the spinning order and the harmonic wave of the rolling body to the total peak value sum of the envelope order spectrum; and inputting the outer ring order peak value ratio feature, the inner ring order peak value ratio feature and the rolling body order peak value ratio feature into a trained multi-core support vector machine classification model, and outputting the defect type of the rolling bearing to be tested.
2. The encoder feedback based rolling bearing defect detection method of claim 1, wherein the determining the target frequency band based on spectral kurtosis comprises: decomposing the vibration signal to form a candidate frequency band set covering different center frequencies and different bandwidths; and selecting candidate frequency bands with the spectral kurtosis index meeting a threshold condition or a ranking condition as the target frequency bands according to the extremum criterion of the spectral kurtosis index, and constructing passband parameters of the bandpass filtering based on the central position and the bandwidth range of the target frequency bands so as to inhibit background noise and highlight defective impact components.
3. The method for detecting a rolling bearing defect based on encoder feedback according to claim 1, wherein performing envelope demodulation on the filtered vibration signal comprises converting the filtered vibration signal into an analysis signal and obtaining an amplitude value of the analysis signal to obtain an instantaneous amplitude value sequence, or rectifying and smoothing the filtered vibration signal to obtain the instantaneous amplitude value sequence, and performing low-pass filtering on the instantaneous amplitude value sequence to remove carrier frequency components, thereby obtaining the envelope signal.
4. The encoder feedback based rolling bearing defect detection method of claim 1, wherein generating the tachometer pulse sequence from the tachometer signal comprises: When the rotating speed signal is output by the incremental encoder, performing edge detection or phase decoding on the rotating speed signal to obtain a pulse time scale corresponding to the rotating shaft rotation angle increment; Calculating an instantaneous rotational speed based on the adjacent pulse time scale and forming a rotational speed signal; And taking the pulse time scale as a reference of the sampling mark, and enabling the sampling mark to be adaptively updated along with the change of the rotating speed signal.
5. The encoder feedback based rolling bearing defect detection method of claim 1, wherein the angular domain resampling comprises: Determining a target corner sequence with equal angle intervals according to the sampling marks; obtaining the corresponding time position of each target corner by using the mapping relation between the time axis and the corner axis; And performing interpolation sampling on the envelope signal at each time position to form the angular domain envelope sequence, wherein the interpolation sampling adopts linear interpolation, polynomial interpolation, spline interpolation or a combination thereof, and performs trending, endpoint prolongation or boundary smoothing processing on the envelope signal before interpolation to reduce interpolation errors, so that defect impact is approximately distributed at equal intervals in the angular domain envelope sequence.
6. The encoder feedback-based rolling bearing defect detection method according to claim 1, wherein calculating the outer ring rolling element passing order, the inner ring rolling element passing order, and the rolling element spinning order according to the structural parameter of the rolling bearing to be detected, comprises: Acquiring the number of rolling bodies, the diameter of the pitch circle and the contact angle parameters of the rolling bearing to be measured; And respectively determining the passing order of the outer ring rolling bodies, the passing order of the inner ring rolling bodies and the spin order of the rolling bodies based on the number of the rolling bodies, the geometric ratio of the diameter of the rolling bodies to the pitch diameter and the geometric correction term corresponding to the contact angle parameter, and defining integer multiples of each order as corresponding harmonic orders for peak extraction in the envelope order spectrum.
7. The encoder feedback-based rolling bearing defect detection method of claim 1, wherein extracting peaks corresponding to the outer ring rolling element passing order, the inner ring rolling element passing order, and the rolling element spin order and harmonics thereof in the envelope order spectrum comprises: Setting an order search window for each of the feature orders in the envelope order spectrum; When the multi-peak competition is detected, selecting a representative peak value according to the peak value amplitude, the peak width, the peak sharpness or the difference degree of the peak value relative to the noise bottom; and correcting the amplitude of the representative peak value, so that the obtained peak value can more accurately reflect the energy aggregation of the defect impact in the order domain.
8. The encoder feedback-based rolling bearing defect detection method of claim 1, wherein the multi-core support vector machine classification model is constructed by: Collecting training data with known health state labels, and collecting the training data under different rotating speed change modes to cover non-stable working conditions; extracting the outer-ring order peak value ratio feature, the inner-ring order peak value ratio feature and the rolling body order peak value ratio feature from each training sample to form feature vectors; And adopting a kernel mapping training support vector machine formed by weighted combination of a plurality of kernel functions to obtain a classification decision boundary for distinguishing the health state, the outer ring defect, the inner ring defect, the rolling body defect and the composite defect, and solidifying the classification decision boundary into the multi-core support vector machine classification model.
9. The encoder feedback based rolling bearing defect detection method of claim 8, further comprising: dividing each vibration signal into a plurality of fragments along the acquisition time or the corner sequence, taking each fragment as an independent sample and inheriting the health state label corresponding to the vibration signal; Dividing the independent samples into a training set, a verification set and a test set according to preset rules, optimizing the kernel weight, penalty parameters or interval parameters of the multi-core support vector machine classification model based on the verification set, and outputting a confusion matrix or an accuracy index by using the test set after optimizing is finished so as to realize quantitative evaluation of defect recognition effects.
10. A rolling bearing defect detection system based on encoder feedback, the system comprising: The signal acquisition module is used for acquiring a vibration signal of the rolling bearing to be tested and acquiring a rotating speed signal coaxial with the rolling bearing to be tested based on an encoder; The band selecting and demodulating module is used for preprocessing the vibration signal and comprises the steps of determining a target frequency band with dominant defect impact based on spectral kurtosis, carrying out band-pass filtering on the target frequency band on the vibration signal to obtain a filtered vibration signal, and carrying out envelope demodulation on the filtered vibration signal to obtain an envelope signal; the time-angle mapping module is used for generating a speed measurement pulse sequence of angular domain resampling reference by the rotating speed signal, determining an equal-angle sampling mark according to the speed measurement pulse sequence, and establishing a mapping relation between a time axis and a corner axis; The angular domain resampling module is used for carrying out angular domain resampling on the envelope signal according to the sampling mark to obtain an angular domain envelope sequence with equal angular intervals; The order analysis module is used for carrying out order spectrum analysis on the angular domain envelope sequence to obtain an envelope order spectrum, calculating the passing order of the outer ring rolling body, the passing order of the inner ring rolling body and the spin order of the rolling body according to the structural parameters of the rolling bearing to be tested, and extracting peaks corresponding to the passing order of the outer ring rolling body, the passing order of the inner ring rolling body, the spin order of the rolling body and the harmonic thereof from the envelope order spectrum; The characteristic construction module is used for calculating an outer ring order peak value ratio characteristic, an inner ring order peak value ratio characteristic and a rolling body order peak value ratio characteristic, wherein the total peak value sum of an envelope order spectrum is the sum of the amplitude values of all orders in the envelope order spectrum, the outer ring order peak value ratio characteristic is the ratio of the peak value sum of passing orders and harmonic waves of the outer ring rolling body to the total peak value sum of the envelope order spectrum, the inner ring order peak value ratio characteristic is the ratio of the peak value sum of passing orders and harmonic waves of the inner ring rolling body to the total peak value sum of the envelope order spectrum, and the rolling body order peak value ratio characteristic is the ratio of the peak value sum of spinning orders and harmonic waves of the rolling body to the total peak value sum of the envelope order spectrum; and the defect classification module is used for inputting the outer ring order peak value ratio characteristic, the inner ring order peak value ratio characteristic and the rolling body order peak value ratio characteristic into a trained multi-core support vector machine classification model and outputting the defect type of the rolling bearing to be tested.

Description

Rolling bearing defect detection method and system based on encoder feedback Technical Field The invention relates to the technical field of bearing detection, in particular to a rolling bearing defect detection method and system based on encoder feedback. Background Rolling bearings are used as critical components in rotating machines, the operating state of which is usually assessed by vibration monitoring. However, under non-stationary working conditions such as variable rotation speed, start-stop, acceleration and deceleration, the statistical characteristics of signals change with time, and characteristic components corresponding to defect excitation drift and spread and crosstalk are generated in the traditional frequency domain representation, so that the accuracy of a diagnosis method based on fixed frequency assumption is reduced. In the prior art, although time-frequency analysis, demodulation methods, order tracking/synchronous processing related to rotating speed and other means are used for improving diagnosability under non-stationary conditions, a few schemes depend on additional measurement or complex parameter setting, and diagnosis staff with abundant experience is often needed for interpretation of results, meanwhile, machine learning can be used for automatic identification, but stable differentiation of the same defect types under variable rotating speed still faces the problems of insufficient feature mobility, poor robustness and the like. Disclosure of Invention In view of the technical problems, the invention provides a rolling bearing defect detection method and a rolling bearing defect detection system based on encoder feedback, which solve the problems that identification is difficult due to characteristic drift under non-stationary conditions and the dependence of a diagnosis process on expert experience is strong. Other features and advantages of the present disclosure will be apparent from the following detailed description, or may be learned in part by the practice of the disclosure. According to an aspect of the present invention, there is provided a rolling bearing defect detection method based on encoder feedback, the method comprising: Acquiring a vibration signal of a rolling bearing to be tested, and acquiring a rotating speed signal coaxial with the rolling bearing to be tested based on an encoder; the method comprises the steps of preprocessing the vibration signal, namely determining a target frequency band with dominant defect impact based on spectral kurtosis, and carrying out band-pass filtering on the target frequency band on the vibration signal to obtain a filtered vibration signal; generating a speed measuring pulse sequence of angular domain resampling reference by the rotating speed signal, determining an equal-angle sampling mark according to the speed measuring pulse sequence, and establishing a mapping relation between a time axis and a corner axis; angular domain resampling is carried out on the envelope signal according to the sampling mark, and an angular domain envelope sequence with equal angular intervals is obtained; Calculating the passing order of the outer ring rolling body, the passing order of the inner ring rolling body and the spin order of the rolling body according to the structural parameters of the rolling bearing to be measured, and extracting peaks corresponding to the passing order of the outer ring rolling body, the passing order of the inner ring rolling body, the spin order of the rolling body and harmonic waves thereof from the envelope order spectrum; Calculating the outer-order peak value ratio feature, the inner-order peak value ratio feature and the rolling element order peak value ratio feature, wherein the total peak value sum of the envelope order spectrum is the sum of the amplitude values of all order lines in the envelope order spectrum; the outer ring order peak value ratio is characterized by the ratio of the peak value sum of the passing order and the harmonic wave of the outer ring rolling body to the total peak value sum of the envelope order spectrum, the inner ring order peak value ratio is characterized by the ratio of the peak value sum of the passing order and the harmonic wave of the inner ring rolling body to the total peak value sum of the envelope order spectrum, and the rolling body order peak value ratio is characterized by the ratio of the peak value sum of the spinning order and the harmonic wave of the rolling body to the total peak value sum of the envelope order spectrum; and inputting the outer ring order peak value ratio feature, the inner ring order peak value ratio feature and the rolling body order peak value ratio feature into a trained multi-core support vector machine classification model, and outputting the defect type of the rolling bearing to be tested. Further, the determining the target frequency band based on the spectral kurtosis comprises: decomposing the vibration signal to form a ca