CN113670612B - Rolling bearing fault diagnosis method based on weighted combined envelope spectrum - Google Patents

Abstract

The invention discloses a rolling bearing fault diagnosis method based on weighted combined envelope spectrum, which is characterized in that the diagnosis method includes: S1: acquiring the vibration acceleration signal of the rolling bearing; S2: estimating the two-dimensional spectrum of the rolling bearing vibration acceleration signal Coherence; S3: Construct the envelope spectrum slice weight function of the two-dimensional spectrum coherence; S4: Obtain the weighted combined envelope spectrum of the vibration acceleration signal according to the two-dimensional spectrum coherence and the envelope spectrum slice weight function; S5 : analyzing the weighted combined envelope spectrum according to the relevant fault information of the rolling bearing to obtain an analysis result; S6 : diagnosing the fault of the rolling bearing according to the analysis result. The rolling bearing fault diagnosis method based on the weighted combined envelope spectrum provided by the present invention can solve the problem that the existing fault diagnosis method cannot effectively extract bearing fault characteristic information when analyzing bearing vibration signals with multiple resonance frequency bands.

Description

A Fault Diagnosis Method for Rolling Bearings Based on Weighted Combined Envelope Spectrum

technical field

The invention relates to the technical field of rolling bearing fault diagnosis, in particular to a rolling bearing fault diagnosis method based on a weighted combined envelope spectrum.

Background technique

The surface damage of rolling bearing elements usually triggers repetitive transient shocks. Correspondingly, the vibration signals of rolling bearings often appear as transient pulses that appear periodically at a certain characteristic frequency. Therefore, the extraction of periodic pulse features is an important prerequisite for rolling bearing fault diagnosis. The transient pulses caused by rolling bearing faults have second-order cyclostationarity, and an effective analysis method is spectral coherence. Since spectral coherence is a two-dimensional representation of spectral frequency and cyclic frequency, in applications, spectral coherence is often integrated along the spectral frequency axis to construct an enhanced envelope spectrum (Enhanced Envelope Spectrum), and the fault detection of rolling bearings is realized by analyzing the enhanced envelope spectrum and diagnosis. However, the enhanced envelope spectrum is obtained by integrating the spectral coherence over the entire spectral frequency range (that is, zero to Nyquist frequency) and does not consider the distribution difference of fault information in the entire frequency band, so that in the case of strong interference noise, it cannot Effectively eliminate the influence of interference components on the identification of fault characteristic frequencies. The Improved Envelope Spectrum (Improved Envelope Spectrum) obtained by integrating spectral coherence in the resonance frequency band can improve the fault detection ability of the envelope spectrum based on spectral coherence, but this method usually only selects a fault-sensitive resonance frequency band for integration and Other resonance frequency bands containing fault information are not considered, and the characteristic information of bearing faults cannot be effectively extracted when analyzing bearing vibration signals with multiple resonance frequency bands.

Contents of the invention

The purpose of the present invention is to provide a rolling bearing fault diagnosis method based on weighted combined envelope spectrum to solve the problem that the existing fault diagnosis method cannot effectively extract bearing fault characteristic information when analyzing bearing vibration signals with multiple resonance frequency bands.

The technical scheme that the present invention solves the problems of the technologies described above is as follows:

The present invention provides a rolling bearing fault diagnosis method based on a weighted combined envelope spectrum, the diagnosis method comprising:

S1: Obtain the vibration acceleration signal of the rolling bearing;

S2: Estimating the two-dimensional spectral coherence of the vibration acceleration signal of the rolling bearing;

S3: constructing the envelope spectrum slice weight function of the two-dimensional spectrum coherence;

S4: Obtain a weighted combined envelope spectrum of the vibration acceleration signal according to the two-dimensional spectral coherence and the envelope spectrum slice weight function;

S5: According to the relevant fault information of the rolling bearing, analyze the weighted combined envelope spectrum to obtain an analysis result;

S6: Diagnose the fault of the rolling bearing according to the analysis result.

Optionally, in the step S1, a vibration acceleration signal of the rolling bearing is obtained by using a vibration acceleration sensor and a data acquisition device.

Optionally, in the step S2, the two-dimensional spectral coherence of the vibration acceleration signal of the rolling bearing is:

Wherein, γ _x (α, f) is the two-dimensional spectral coherence of the acceleration signal, S _x (α, f) is the spectral correlation of the vibration acceleration signal, α is the cycle frequency, f is the spectral frequency, and S _x (0, f) is the slice of the spectral correlation of the vibration acceleration signal at α=0, and S _x (0,f-α) is the result of shifting the slice of the spectral correlation of the vibration acceleration signal at α=0 along the spectral frequency by α.

Optionally, the spectral correlation of the vibration acceleration signal is expressed as:

是期望算子，*表示复数共轭，t_n＝n/F_s，n＝0,1,2,…,N-1，τ_m＝m/F_s，m＝0,1,2,…N-1，t_n和τ_m分别表示采样时刻和时间延迟。Wherein, S _x (α, f) is the spectral correlation of the vibration acceleration signal, α is the cycle frequency, f is the spectral frequency, N is the sampling length of the vibration acceleration signal, and F _s is the sampling frequency of the vibration acceleration signal, R _x (t _n , τ _m ) is the instantaneous autocorrelation function of the vibration acceleration signal, and

is the expectation operator, * means complex conjugate, t _n =n/F _s , n=0,1,2,…,N-1, τ _m =m/F _s , m=0,1,2,… N-1, t _n and τ _m denote the sampling instant and time delay, respectively.

Optionally, in the step S3, the envelope spectrum slice weight function of the two-dimensional spectrum coherence is:

Wherein, w(f) represents the envelope spectral slice weight function of the two-dimensional spectral coherence, represents the frequency-domain signal-to-noise ratio measure of the envelope spectral slice of the spectral coherence at each spectral frequency, and thres is a threshold.

Optionally, the frequency-domain signal-to-noise ratio measure of the envelope spectrum slice of the spectral coherence at each spectral frequency is expressed as:

Among them, FDSNRM(f) represents the frequency-domain signal-to-noise ratio measure of the envelope spectrum slice of the spectral coherence at each spectral frequency, H and L are the harmonic number and cycle frequency of the fault characteristic frequency in the envelope spectrum slice, respectively , γ _x (α, f) is the two-dimensional spectral coherence of the acceleration signal, A _h represents a narrow cyclic frequency band centered on frequency hf _m and A _h ={α|(h-δ)f _m ≤α≤(h+δ)f _m }, h=1,2,…,H, δ is a small positive number, f _m is the fault characteristic frequency of the rolling bearing, α and f represent the cycle frequency and spectrum frequency respectively.

Optionally, the threshold is expressed as:

thres=μ(FDSNRM(f))+η·σ(FDSNRM(f))

Among them, μ( ) and σ( ) are the mean operator and standard deviation operator respectively, η is a non-negative coefficient used to adjust the threshold, FDSNRM(f) represents the spectral coherence at each spectral frequency A frequency-domain signal-to-noise ratio measure for envelope spectral slices of .

Optionally, in the step S4, the weighted combined envelope spectrum of the vibration acceleration signal is expressed as:

Wherein, WCES(α) is the weighted combined envelope spectrum of the vibration acceleration signal, w(f) represents the envelope spectrum slice weight function of the two-dimensional spectral coherence, and γ _x (α, f) is the two-dimensional spectrum of the acceleration signal Dimensional spectral coherence, F _s is the sampling frequency of the vibration acceleration signal, α and f represent the cyclic frequency and spectral frequency respectively, α _i represents the ith discrete cyclic frequency and α _i =iF _s /N, N and F _{s are} respectively is the sampling length and sampling frequency of the vibration acceleration signal.

Optionally, the rolling bearing-related fault information includes:

Dimensional parameters and rotational speed information of the rolling bearing; and/or

According to the size parameter and rotational speed information of the rolling bearing, the fault characteristic frequency of each element of the rolling bearing is obtained.

The present invention has the following beneficial effects:

The method proposed in the present invention does not need to divide the spectral frequency bands of spectral coherence, and the calculation process is simple; the frequency-domain signal-to-noise ratio measurement is used to evaluate the distribution difference of fault characteristic information in the entire spectral frequency band, and the information threshold is introduced to identify faults in spectral coherence Envelope spectrum slices that are rich in information and dominated by interference components; the constructed weight function can enhance the magnitude of fault-related components and weaken the influence of interference components. This method can effectively reveal the fault feature information of rolling bearings, and is an effective fault diagnosis method for rolling bearings.

Description of drawings

Fig. 1 is a flow chart of a rolling bearing fault diagnosis method based on a weighted combined envelope spectrum provided by an embodiment of the present invention;

Fig. 2 is the vibration acceleration signal of the outer ring fault bearing and its spectral coherence, enhanced envelope spectrum, frequency domain signal-to-noise ratio measurement and weighted combination package of the rolling bearing fault diagnosis method based on weighted combined envelope spectrum provided by the embodiment of the present invention network spectrum;

Fig. 3 is the vibration acceleration signal of the inner ring fault bearing and its spectral coherence, enhanced envelope spectrum, frequency domain signal-to-noise ratio measurement and weighted combination package of the rolling bearing fault diagnosis method based on weighted combined envelope spectrum provided by the embodiment of the present invention network spectrum.

Detailed ways

The principles and features of the present invention are described below in conjunction with the accompanying drawings, and the examples given are only used to explain the present invention, and are not intended to limit the scope of the present invention.

Example 1

The present invention proposes a method for diagnosing rolling bearing faults based on weighted combined envelope spectrum. The flow chart of the method is shown in FIG. 1 . This method designs a frequency-domain signal-to-noise ratio measure to quantify and evaluate the bearing fault feature information contained in each envelope spectrum slice of the spectral coherence, and introduces an information threshold to identify the envelope with rich fault information and interference components in the spectral coherence Spectrum slices, and then a weight function is constructed from the frequency-domain SNR measure and information threshold to enhance the fault feature extraction and interference noise removal capabilities of the spectral coherence-based envelope spectrum.

The technical scheme that the present invention solves the problems of the technologies described above is as follows:

The present invention provides a rolling bearing fault diagnosis method based on a weighted combined envelope spectrum, as shown in FIG. 1 , the diagnosis method includes:

S1: Obtain the vibration acceleration signal of the rolling bearing;

S2: Estimating the two-dimensional spectral coherence of the vibration acceleration signal of the rolling bearing;

S3: constructing the envelope spectrum slice weight function of the two-dimensional spectrum coherence;

S4: Obtain a weighted combined envelope spectrum of the vibration acceleration signal according to the two-dimensional spectral coherence and the envelope spectrum slice weight function;

S5: According to the relevant fault information of the rolling bearing, analyze the weighted combined envelope spectrum to obtain an analysis result;

S6: Diagnose the fault of the rolling bearing according to the analysis result.

The present invention has the following beneficial effects:

The method proposed in the present invention does not need to divide the spectral frequency bands of spectral coherence, and the calculation process is simple; the frequency-domain signal-to-noise ratio measurement is used to evaluate the distribution difference of fault characteristic information in the entire spectral frequency band, and the information threshold is introduced to identify faults in spectral coherence Envelope spectrum slices that are rich in information and dominated by interference components; the constructed weight function can enhance the magnitude of fault-related components and weaken the influence of interference components. This method can effectively reveal the fault feature information of rolling bearings, and is an effective fault diagnosis method for rolling bearings.

Optionally, in the step S1, a vibration acceleration signal of the rolling bearing is obtained by using a vibration acceleration sensor and a data acquisition device.

The data acquisition equipment here refers to the equipment used to acquire vibration acceleration signals of bearings, including data acquisition cards, computers with data acquisition software installed, etc.

Specifically, the mathematical symbol of the vibration acceleration signal is: x(t _n ).

Wherein, t _n =n/F _s , n=0,1,...,N-1 is the sampling time, F _s is the sampling frequency, and N is the signal sampling length.

Optionally, in the step S2, the two-dimensional spectral coherence (Spectral Coherence) of the vibration acceleration signal of the rolling bearing is:

Wherein, γ _x (α, f) is the two-dimensional spectral coherence of the acceleration signal, S _x (α, f) is the spectral correlation of the vibration acceleration signal, α is the cycle frequency, f is the spectral frequency, and S _x (0, f) is the slice of the spectral correlation of the vibration acceleration signal at α=0, and S _x (0,f-α) is the result of shifting the slice of the spectral correlation of the vibration acceleration signal at α=0 along the spectral frequency by α.

Optionally, the spectral correlation (Spectral Correlation) of the vibration acceleration signal is expressed as:

是期望算子，*表示复数共轭，t_n＝n/F_s，n＝0,1,2,…,N-1，τ_m＝m/F_s，m＝0,1,2,…N-1，t_n和τ_m分别表示采样时刻和时间延迟。Wherein, S _x (α, f) is the spectral correlation of the vibration acceleration signal, α is the cycle frequency, f is the spectral frequency, N is the sampling length of the vibration acceleration signal, and F _s is the sampling frequency of the vibration acceleration signal, R _x (t _n , τ _m ) is the instantaneous autocorrelation function of the vibration acceleration signal, and

is the expectation operator, * means complex conjugate, t _n =n/F _s , n=0,1,2,…,N-1, τ _m =m/F _s , m=0,1,2,… N-1, t _n and τ _m denote the sampling instant and time delay, respectively.

Optionally, in the step S3, the envelope spectrum slice weight function of the two-dimensional spectrum coherence is:

Among them, w(f) represents the envelope spectrum slice weight function of the two-dimensional spectral coherence, FDSNRM(f) represents the frequency-domain signal-to-noise ratio measure of the envelope spectrum slice of the spectrum coherence at each spectral frequency, and thres is the threshold .

Optionally, the Frequency Domain Signal-to-Noise Ratio Measure (FDSNRM) of the envelope spectral slice of the spectral coherence at each spectral frequency is expressed as:

Among them, FDSNRM(f) represents the frequency-domain signal-to-noise ratio measure of the envelope spectrum slice of the spectral coherence at each spectral frequency, H and L are the harmonic number and cycle frequency of the fault characteristic frequency in the envelope spectrum slice, respectively , γ _x (α, f) is the two-dimensional spectral coherence of the acceleration signal, A _h represents a narrow cyclic frequency band centered on frequency hf _m and A _h ={α|(h-δ)f _m ≤α≤(h+δ)f _m }, h=1,2,…,H, δ is a small positive number, f _m is the fault characteristic frequency of the rolling bearing, α and f represent the cycle frequency and spectrum frequency respectively, α _i represents the ith discrete cycle frequency and α _i =iF _s /N, N and F _s are the sampling length and sampling frequency of the vibration acceleration signal, respectively.

Optionally, the threshold is expressed as:

thres=μ(FDSNRM(f))+η·σ(FDSNRM(f))

Among them, μ( ) and σ( ) are the mean operator and standard deviation operator respectively, η is a non-negative coefficient used to adjust the threshold, FDSNRM(f) represents the spectral coherence at each spectral frequency A frequency-domain signal-to-noise ratio measure for envelope spectral slices of .

Optionally, in the step S4, the weighted combined envelope spectrum (WeightedCombined Envelope Spectrum, WCES) of the vibration acceleration signal is expressed as:

Wherein, WCES(α) is the weighted combined envelope spectrum of the vibration acceleration signal, w(f) represents the envelope spectrum slice weight function of the two-dimensional spectral coherence, and γ _x (α, f) is the two-dimensional spectrum of the acceleration signal Dimensional spectral coherence, F _s is the sampling frequency of the vibration acceleration signal, and α and f represent the cyclic frequency and the spectral frequency respectively.

Optionally, the rolling bearing-related fault information includes:

Dimensional parameters and rotational speed information of the rolling bearing; and/or

According to the size parameter and rotational speed information of the rolling bearing, the fault characteristic frequency of each element of the rolling bearing is obtained.

Specifically, the fault characteristic frequency of each element of the rolling bearing is estimated according to the size parameters and rotational speed information of the rolling bearing to be detected. According to the fault characteristic frequency of the rolling bearing, it is judged whether the spectral lines at the fault characteristic frequency and its harmonic components in the weighted combined envelope spectrum can be observed. If the fault characteristic frequency of a certain component and the spectral line at its harmonic component are very obvious, the fault and the fault type of the rolling bearing can be identified.

The parameters of the method proposed by the present invention include the type of window function, window length, observed maximum cycle frequency, threshold coefficient η and fault characteristic frequency f _m . In the following examples, the window function is a Hanning (Hanning) window, the window length is 128 sampling points, and the threshold coefficient is 1.5; the maximum cycle frequency observed in embodiment 2 is 250 Hz, and the maximum cycle frequency observed in embodiment 3 is 1200Hz; the fault characteristic frequency of embodiment 2 is 66.42Hz, and the fault characteristic frequency of embodiment 3 is 325.8Hz.

Example 2

Fig. 2 is the vibration acceleration signal and its spectral coherence, enhanced envelope spectrum, frequency-domain signal-to-noise ratio measure and weighted combined envelope spectrum of a rolling bearing with an outer ring fault. The sampling frequency of the bearing vibration acceleration signal is 150kHz, and the analyzed signal length is 4s. In the spectral coherence and enhanced envelope spectra shown in Figure 2(b) and (c), the spectral lines corresponding to the characteristic frequency f _o of the bearing outer ring fault and its harmonics cannot be observed, and the outer ring fault of the rolling bearing cannot be identified. The frequency-domain signal-to-noise ratio measurement shown in Fig. 2(d) shows that the characteristic information of bearing outer ring faults is mainly distributed in the spectral frequency band around 47kHz. The weighted combined envelope spectrum of the bearing vibration acceleration signal shown in Figure 2(e) can clearly detect the fault characteristic frequency f _o of the bearing outer ring and the spectral lines at the first 2 harmonics 2f _o and 3f _o , which can be judged It is a failure of the outer ring of the bearing. Therefore, the method proposed by the invention can effectively detect the fault of the outer ring of the rolling bearing.

Example 3

Fig. 3 is the vibration acceleration signal and its spectral coherence, enhanced envelope spectrum, frequency-domain signal-to-noise ratio measure and weighted combined envelope spectrum of a rolling bearing with an inner ring fault. The sampling frequency of the bearing vibration acceleration signal is 51.2kHz, and the analyzed signal length is 4s. In the spectral coherence shown in Fig. 3(b), the spectral lines corresponding to the characteristic frequency fi _and its harmonics of the bearing inner ring fault cannot be clearly observed, and the inner ring fault of the rolling bearing cannot be identified. In the enhanced envelope spectrum shown in Fig. 3(c), the spectral lines corresponding to the characteristic frequency f _i of the bearing inner ring fault and its first two harmonics 2f _i and 3f _i can be observed. The frequency-domain signal-to-noise ratio measurement shown in Fig. 3(d) shows that the characteristic information of bearing inner ring faults is mainly distributed in three spectral frequency bands near 7kHz, 11kHz and 17.5kHz. From the weighted combined envelope spectrum of the bearing vibration acceleration signal shown in Fig. 3(e), the spectral lines at the characteristic frequency f _i of the bearing inner ring fault and its first 2 harmonics 2f _i and 3f _i can be clearly detected. It is judged that the inner ring of the bearing is faulty. Therefore, the method proposed by the present invention can effectively detect the fault of the inner ring of the rolling bearing.

The above descriptions are only some embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the protection scope of the present invention within.

Claims (6)

1. A rolling bearing fault diagnosis method based on weighted combined envelope spectrum, characterized in that, the diagnosis method comprises: S1: Obtain the vibration acceleration signal of the rolling bearing; S2: Estimating the two-dimensional spectral coherence of the vibration acceleration signal of the rolling bearing; S3: constructing the envelope spectrum slice weight function of the two-dimensional spectrum coherence; S4: Obtain a weighted combined envelope spectrum of the vibration acceleration signal according to the two-dimensional spectral coherence and the envelope spectrum slice weight function; S5: According to the relevant fault information of the rolling bearing, analyze the weighted combined envelope spectrum to obtain an analysis result; S6: Diagnose the fault of the rolling bearing according to the analysis result; In the step S4, the weighted combined envelope spectrum of the vibration acceleration signal is expressed as:

Wherein, WCES(α) is the weighted combined envelope spectrum of the vibration acceleration signal, w(f) represents the envelope spectrum slice weight function of the two-dimensional spectral coherence, and γ _x (α, f) is the two-dimensional spectrum of the acceleration signal Dimensional spectral coherence, F _s is the sampling frequency of the vibration acceleration signal, and α and f represent cycle frequency and spectral frequency respectively; In the step S3, the envelope spectrum slice weight function of the two-dimensional spectrum coherence is:

Among them, w(f) represents the envelope spectrum slice weight function of the two-dimensional spectral coherence, FDSNRM(f) represents the frequency-domain signal-to-noise ratio measure of the envelope spectrum slice of the spectrum coherence at each spectral frequency, and thres is the threshold ; The frequency-domain signal-to-noise ratio measure of the envelope spectral slice of the spectral coherence at each spectral frequency is expressed as:

Among them, FDSNRM(f) represents the frequency-domain signal-to-noise ratio measure of the envelope spectrum slice of the spectral coherence at each spectral frequency, H and L are the harmonic number and cycle frequency of the fault characteristic frequency in the envelope spectrum slice, respectively , γ _x (α, f) is the two-dimensional spectral coherence of the acceleration signal, A _h represents a narrow cyclic frequency band centered on frequency hf _m and A _h ={α|(h-δ)f _m ≤α≤(h+δ)f _m }, h=1,2,…,H, δ is a small positive number, f _m is the fault characteristic frequency of the rolling bearing, α and f represent the cycle frequency and spectrum frequency respectively, α _i represents the i-th discrete cycle frequency and α _i =iF _s *N, where N and F _s are the sampling length and sampling frequency of the vibration acceleration signal, respectively. 2. The rolling bearing fault diagnosis method based on weighted combined envelope spectrum according to claim 1, characterized in that, in the step S1, the vibration acceleration signal of the rolling bearing is obtained by using a vibration acceleration sensor and a data acquisition device. 3. The rolling bearing fault diagnosis method based on weighted combined envelope spectrum according to claim 1, characterized in that, in the step S2, the two-dimensional spectral coherence of the rolling bearing vibration acceleration signal is:

Wherein, γ _x (α, f) is the two-dimensional spectral coherence of the acceleration signal, S _x (α, f) is the spectral correlation of the vibration acceleration signal, α is the cycle frequency, f is the spectral frequency, and S _x (0, f) is the slice of the spectral correlation of the vibration acceleration signal at α=0, and S _x (0,f-α) is the result of shifting the slice of the spectral correlation of the vibration acceleration signal at α=0 along the spectral frequency by α. 4. the rolling bearing fault diagnosis method based on weighted combined envelope spectrum according to claim 3, is characterized in that, the spectral correlation of described vibration acceleration signal is expressed as:

Wherein, S _x (α, f) is the spectral correlation of the vibration acceleration signal, α is the cycle frequency, f is the spectral frequency, N is the sampling length of the vibration acceleration signal, and F _s is the sampling frequency of the vibration acceleration signal, R _x (t _n , τ _m ) is the instantaneous autocorrelation function of the vibration acceleration signal, and

is the expectation operator, * means complex conjugate, t _n =n*F _s , n=0,1,2,…,N-1, τ _m =m/F _s ,m=0,1,2,… t _n and τ _m denote the sampling instant and time delay, respectively. 5. the rolling bearing fault diagnosis method based on weighted combined envelope spectrum according to claim 1, is characterized in that, described threshold value is expressed as: thres=μ(FDSNRM(f))+η·σ(FDSNRM(f)) Among them, μ( ) and σ( ) are the mean operator and standard deviation operator respectively, η is a non-negative coefficient used to adjust the threshold, FDSNRM(f) represents the spectral coherence at each spectral frequency A frequency-domain signal-to-noise ratio measure for envelope spectral slices of . 6. The rolling bearing fault diagnosis method based on the weighted combined envelope spectrum according to any one of claims 1-5, wherein the rolling bearing-related fault information includes: Dimensional parameters and rotational speed information of the rolling bearing; and/or According to the size parameter and rotational speed information of the rolling bearing, the fault characteristic frequency of each element of the rolling bearing is obtained.

Images