Rotating machine fault diagnosis method, medium and equipment

Abstract

The invention provides a rotating machinery fault diagnosis method, a medium and equipment, solves the problem that CYCBD needs priori estimation of a cyclic frequency and a filter length, and proposes that IESCFFOgram is used for identifying an accurate information frequency band of the CYCBD, and the cyclic frequency is estimated through an enhanced harmonic product spectrum of the information frequency band. According to the method, the advantages of PSE and DE are fused, and the PSE-DE index is proposed to be used for effectively selecting the optimal filter length of CYCBD. Compared with a traditional deconvolution method, the method can effectively weaken the interference of harmonic components, increases the number and amplitude of fault characteristic frequencies, and has wider applicability.

Description

Technical Field

[0001] The invention belongs to the field of fault diagnosis, and in particular relates to a rotating machinery fault diagnosis method, medium and equipment. Background Art

[0002] Rotating machinery is widely used in the industrial field, such as power, chemical, metallurgy and other industries. The normal operation of these rotating machinery (such as generators, compressors, pumps, etc.) is crucial to the stability and safety of the entire industrial production process. Once a failure occurs, it may cause production interruption, equipment damage, and even cause safety accidents. For example, in the power generation industry, the failure of a large steam turbine generator set may cause a large-scale power outage, causing huge losses to society and the economy. According to statistics, rotating machinery failures account for about 60%-70% of industrial equipment failures. Therefore, timely and accurate diagnosis of rotating machinery failures and taking corresponding maintenance measures is a key technical issue in the industrial field.

[0003] Early fault diagnosis mainly relied on manual experience and simple vibration analysis. Traditional time domain analysis methods, such as peak detection and RMS value calculation, can only provide limited fault information. For example, in bearing fault diagnosis, it is difficult to distinguish different types and degrees of faults based on simple time domain amplitude characteristics.

[0004] However, the traditional CYCBD method has some problems in practical application. Due to the complex and diverse operating conditions of rotating machinery, the fixed-parameter CYCBD method may not be able to adapt to the signal characteristics under different operating conditions. For example, different operating conditions such as speed and load will cause the cyclic frequency and noise level of the signal to change. Therefore, an adaptive CYCBD method is needed that can automatically adjust the parameters according to the actual characteristics of the signal to more effectively extract fault features. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a rotating machinery fault diagnosis method, medium and equipment.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides a method for diagnosing a fault in a rotating machine, comprising the following steps:

[0008] S1: Collect the vibration signal x of the rolling bearing under the noise background;

[0009] S2: According to the collected vibration signal x, the temporary filtering signal s0 is obtained by CYCBD calculation;

[0010] S3: For the temporary filtered signal s0, estimate the cyclic frequency α using IESCFFOgram;

[0011] S4: When the cycle frequency α is determined, the filter length L is determined using the sample entropy and the power spectrum entropy;

[0012] S5: According to the cycle frequency α and the filter length L, a new temporary filter signal s0 is obtained by using CYCBD calculation;

[0013] S6: Determine whether the convergence criterion is met or the maximum number of iterations is exceeded: if not, return to step S3; if satisfied, output the temporary filtered signal s0 calculated in step S5 as the final filtered signal s;

[0014] S7: Analyze the envelope spectrum of the filtered signal s to determine whether the rotating machinery has a fault.

[0015] Optionally, in step S2, the temporary filtering signal s0 is obtained by using CYCBD calculation, as follows:

[0016] s=Xf≈s0;

[0017] Where X is the matrix representation of the vibration signal x; f is the inverse filter, which is calculated by the cyclostationary index ICS2:

[0018]

[0019] Where, the superscript H is the conjugate transpose of the matrix; R XWX is the weighted correlation matrix of X; R XX is the correlation matrix of X; the weight matrix W is expressed as:

[0020]

[0021] |s| 2 =[|s[L-1]| 2 ,…,|s[N-1]| 2 ] T ;

[0022]

[0023] Where N is the length of signal s0; L is the length of f; s[L-1] is the L-1th number of the filtered signal s; α represents the cycle frequency.

[0024] Optionally, in step S3, estimating the cycle frequency α by using IESCFFOgram specifically includes the following steps:

[0025] S3.1: Use the spectral correlation SC to estimate the spectral coherence normalized SCoh of the temporary filtered signal s0, and divide the full spectrum band of SCoh into multiple narrow bands;

[0026] S3.2: Take the narrow band with the maximum energy ratio ER and calculate the enhanced harmonic product spectrum H(α) of the narrow band;

[0027] S3.3: Output the estimated cycle frequency α corresponding to the global maximum of the H(α) amplitude.

[0028] Optionally, step S3.1 is specifically as follows:

[0029] Calculate the double discrete Fourier transform of the instantaneous autocorrelation function of s0, denoted as S x (α, f c ), f c is the carrier frequency; γ x (α, f c ) is SCoh, which is S x (α, f c ) is a normalized version of SCoh; the full spectrum band division strategy of the 1 / 3 binary tree filter structure is used to divide SCoh into multiple narrowbands. For the j-th narrowband of the h-th layer, the enhanced envelope spectrum IES formula is as follows:

[0030]

[0031] In the formula, F s is the sampling frequency of s0.

[0032] Optionally, step S3.2 is specifically as follows:

[0033] Calculate all local fault frequencies and IES h，j The energy ratio ER of (α), h* and j* are the h and j values ​​corresponding to the maximum value of ER, and the enhanced harmonic product spectrum H(α) is obtained as follows:

[0034] H(α)=IES h*，j* (α) IES h*，j* (2α)·...·IES h*，j* (Mα).

[0035] Optionally, in step S4, the filter length L is determined by using sample entropy and power spectrum entropy, specifically:

[0036] Calculate the power spectrum entropy DE(s0) and sample entropy PSE(s0) corresponding to s0 and normalize them;

[0037] According to the normalized power spectrum entropy β(DE(s0)) and sample entropy β(PSE(s0), the PSE-DE corresponding to s0 is calculated as the measurement indicator:

[0038] PSE-DE(s0)=β(DE(s0))·β(PSE(s0));

[0039] Output the filter length L corresponding to the maximum value of PSE-DE(s0).

[0040] In a second aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the rotating machinery fault diagnosis method as described in the first aspect.

[0041] In a third aspect, the present invention provides an electronic device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the rotating machinery fault diagnosis method as described in the first aspect is implemented.

[0042] The beneficial effects of the present invention are as follows: the present invention solves the problem that CYCBD requires a priori estimation of the cycle frequency and filter length, proposes using IESCFFOgram to identify the accurate information band of CYCBD, and estimates the cycle frequency through the enhanced harmonic product spectrum of the information band. The present invention combines the advantages of PSE and DE, and proposes the PSE-DE indicator for effectively selecting the optimal filter length of CYCBD. Compared with the traditional deconvolution method, the present invention can effectively weaken the interference of harmonic components, increase the number and amplitude of fault characteristic frequencies, and has a wider applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 The present invention is a flow chart of a rotating machinery fault diagnosis method.

[0044] Figure 2 The experimental vibration signal and its envelope spectrum of the present invention: (a) original vibration signal; (b) its envelope spectrum.

[0045] Figure 3 SCoh and IESC FFOgram of the experimental vibration signal of the present invention: (a) Scoh; (b) EES; (c) IESC FFOgram; (d) its IES.

[0046] Figure 4 SCoh and IESCFFOgram of the temporary filtered signal of the present invention: (a) Scoh; (b) EES; (c) IESCFFOgram; (d) its IES.

[0047] Figure 5 These are five indicators of the temporary filtering signal of the present invention.

[0048] Figure 6 The envelope spectra of the signals after different L filtering of the present invention: (a) original vibration signal (without filtering); (b) L==25; (c) L=167; (d) L=196. DETAILED DESCRIPTION

[0049] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0050] In one embodiment, the present invention provides a rotating machinery fault diagnosis method, such as Figure 1 As shown in Figure 1, the steps of fault feature extraction are:

[0051] Step 1: Collect the vibration signal x of the rolling bearing under the noise background;

[0052] Step 2: Use CYCBD (Maximum second order cyclostationarity blind deconvolution) to calculate and obtain the temporary filtered signal s0;

[0053] Step 3: Estimate the cycle frequency α using IESCFFOgram (Improved envelope spectrum via candidate faultfrequencyoptimization-gram);

[0054] Step 4: PSE-DE determines the filter length L;

[0055] Step 5: Calculate the correlation matrix R XX , weighted matrix W, weighted correlation matrix R XWX ;

[0056] Step 6: Update f and ICS2;

[0057] Step 7: Update α and L until the convergence criterion is met or the maximum number of iterations is exceeded;

[0058] Step 8: Output the final CYCBD filtered signal s, analyze the envelope spectrum of s to see if the rotating machinery has any faults.

[0059] In the above Step 2, the temporary filtering signal s0 is obtained by using CYCBD calculation as follows:

[0060] s=x*f≈s0;

[0061] Among them, f is the inverse filter; * is the convolution operation; the above formula can be expressed in matrix as:

[0062] s = Xf;

[0063]

[0064] Among them, x[L-1] is the L-1th number of signal x, and the same is true for s[L-1] and f[L-1]. The solution process of f can be expressed as an optimization problem under conditional constraints. In CYCBD, in order to obtain f, the cyclostationarity index ICS2 is introduced. The larger the value of this index, the more obvious the periodicity of the signal. ICS2 can be defined as:

[0065]

[0066] Where, the superscript H is the conjugate transpose of the matrix; R XWX is the weighted correlation matrix of X; R XX is the correlation matrix of X; the weight matrix W can be expressed as:

[0067]

[0068] |s| 2 =[|s[L-1]| 2 ,…,|s[N-1]| 2 ] T ;

[0069]

[0070] Among them, α represents the cycle frequency, N is the length of signal s0; L is the length of f.

[0071] In order to make CYCBD more effective, it is necessary to find the optimal α and L.

[0072] In the above Step 3, α is estimated using IESCFFOgram as follows: SCoh (Spectral coherence) estimated by fast SC (spectral correlation) is used to divide the full spectrum band of SCoh into multiple groups of narrow bands; the narrow band with the maximum ER (Energy Ratio) is taken to calculate the enhanced harmonic product spectrum H(α) of the narrow band; and the estimated cyclic frequency α corresponding to the global maximum amplitude of H(α) is output.

[0073] Fast SC is the double discrete Fourier transform of the instantaneous autocorrelation function of the temporary filtered signal s0, expressed as S x (α, f c ); γ x (α, f c ) is SCoh, which is S x (α, f c) is a normalized version of SCoh; the full spectrum band division strategy of the 1 / 3 binary tree filter structure is used to divide SCoh into multiple narrowbands. For the jth (j = 0, 1, 1.6, ..., 5) narrowband of the hth (h = ±1, ±2, ±3) layer, the enhanced envelope spectrum (IES) formula is as follows:

[0074]

[0075] The energy of all local fault frequencies is compared with IES h，j (α) is called ER; h* and j* are the values ​​of h and j* corresponding to the maximum value of ER, then the enhanced harmonic product spectrum H(α) can be defined as follows:

[0076] H(α)=IES h*，j* (α) IES h*,j* (2α)·...·IES h*，j* (Mα);

[0077] In the above Step 4, PSE-DE determines the filter length L, as follows: when α is determined, calculate all PSE-DE(s0) in the range of L values; output L corresponding to the maximum value of PSE-DE(s0), PSE-DE(s0) is:

[0078] PSE-DE(s0)=β(DE(s0))·β(PSE(s0));

[0079] In the formula, DE(s0) is the power spectrum entropy, PSE(s0) is the sample entropy, and the above formula is expressed as the product of the normalized power spectrum entropy and sample entropy.

[0080] Next, the experimental signal obtained from the Case Western Reserve University (CWRU) Bearing Data Center is used to verify the effectiveness and universal applicability of the method proposed in this embodiment. The accelerometer is installed on the motor housing at the motor drive end to collect the vibration signal generated by the experimental bearing. The data acquisition system channel sampling frequency is 12kHz, and the signal length is set to 1.3653s.

[0081] Analyze the vibration signal of the drive end bearing with a 0.021 rolling element defect collected at a shaft speed of 1797 rpm. The ball spin frequency (BSF) is 119.4 Hz. Figure 2 The time domain waveform and corresponding envelope spectrum of the rolling element fault signal to be analyzed are given. Figure 3 As shown in (b), the envelope spectrum of the original vibration signal is composed of complex frequency components. Under the interference of the surrounding spectrum lines, the BSF and its harmonics are almost undetectable. Using SCoh to identify the fault-related cyclostationary features hidden in the vibration signal, the results are shown in Figure 3 (a) and Figure 3As shown in (c), the EES of SCoh shows that the original vibration signal is contaminated by background noise, making the BSF and its harmonics invisible. Figure 3 (b) and Figure 3 (d) shows the IESCFFOgram of the original vibration signal, from which the identified information band is centered at 800 Hz with a bandwidth of 250 Hz.

[0082] Figure 3 (d) depicts the IES of IESCFFOgram. No BSF and its harmonics are found in the IES, indicating that the correct information frequency band has not been identified. This embodiment uses IESCFFOgram to adaptively select α of ACYCBD. The results are shown in Figure 2. Figure 4 shown. Figure 4 (a) and Figure 4 The EES of SCoh plotted in (c) shows that the temporarily filtered signal still has the influence of many interfering spectral lines. Figure 4 (b) shows the IESCFFOgram of the original vibration signal, from which the identified information band is centered at 4125 Hz with a bandwidth of 250 Hz. Figure 4 In (d), prominent amplitudes of 119.4 Hz and 238.8 Hz can be observed, so 119.4 Hz is the α of ACYCBD.

[0083] This embodiment uses PSE-DE to select the optimal L of CYCBD and compares it with some related entropy indicators, such as Figure 5 As shown. Taking L as 25, 167 (optimal value of PSE-DE) and 196 (optimal value of FE and SE) and substituting them into CYCBD, the corresponding envelope spectrum is as follows Figure 6 shown. Figure 6 In (c), the BSF and its first three harmonics are very prominent, with no interfering spectral lines around them. Figure 6 (b) and Figure 6 (d) There are still interference spectral lines, and the amplitude corresponding to the BSF is not obvious.

[0084] In another embodiment, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the rotating machinery fault diagnosis method of the aforementioned embodiment.

[0085] In another embodiment, the present invention proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the rotating machinery fault diagnosis method of the aforementioned embodiment is implemented.

[0086] In the embodiments disclosed in the present application, the computer storage medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. The computer storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the above. More specific examples of computer storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0087] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0088] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A method for diagnosing a rotating machinery fault, characterized in that: The steps include: S1: Collect the vibration signal x of the rolling bearing under the noise background; S2: According to the collected vibration signal x, the temporary filtering signal s0 is obtained by CYCBD calculation; S3: For the temporary filtered signal s0, estimate the cyclic frequency α using IESCFFOgram; S4: When the cycle frequency α is determined, the filter length L is determined using the sample entropy and the power spectrum entropy; S5: According to the cycle frequency α and the filter length L, a new temporary filter signal s0 is obtained by using CYCBD calculation; S6: Determine whether the convergence criterion is met or the maximum number of iterations is exceeded: if not, return to step S3; if satisfied, output the temporary filtered signal s0 calculated in step S5 as the final filtered signal s; S7: Analyze the envelope spectrum of the filtered signal s to determine whether the rotating machinery has a fault.

2. A rotating machinery fault diagnosis method according to claim 1, characterized in that: In step S2, the temporary filtering signal s0 is obtained by using CYCBD calculation, as follows: s=Xf≈s0; Where X is the matrix representation of the vibration signal x; f is the inverse filter, which is calculated by the cyclostationary index ICS2: Where, the superscript H is the conjugate transpose of the matrix; R XWX is the weighted correlation matrix of X; R XX is the correlation matrix of X; the weight matrix W is expressed as: |s| 2 =[|s[L-1]| 2 ,…,|s[N-1]| 2 ] T ; Where N is the length of signal s0; L is the length of f; s[L-1] is the L-1th number of the filtered signal s; α represents the cycle frequency.

3. A rotating machinery fault diagnosis method according to claim 1, characterized in that: In step S3, the estimating the cycle frequency α by using IESCFFOgram specifically includes the following steps: S3.1: Use the spectral correlation SC to estimate the spectral coherence normalized SCoh of the temporary filtered signal s0, and divide the full spectrum band of SCoh into multiple narrow bands; S3.2: Take the narrow band with the maximum energy ratio ER and calculate the enhanced harmonic product spectrum H(α) of the narrow band; S3.3: Output the estimated cycle frequency α corresponding to the global maximum of the H(α) amplitude.

4. A rotating machinery fault diagnosis method as claimed in claim 3, characterized in that: Step S3.1 is as follows: Calculate the double discrete Fourier transform of the instantaneous autocorrelation function of s0, denoted as S x (α, f c ), f c is the carrier frequency; γ x (α, f c ) is SCoh, which is S x (α, f c ) is a normalized version of SCoh; the full spectrum band division strategy of the 1 / 3 binary tree filter structure is used to divide SCoh into multiple narrowbands. For the j-th narrowband of the h-th layer, the enhanced envelope spectrum IES formula is as follows: In the formula, F s is the sampling frequency of s0.

5. A rotating machinery fault diagnosis method according to claim 4, characterized in that: Step S3.2 is as follows: Calculate all local fault frequencies and IES h，j The energy ratio ER of (α), h* and j* are the h and j values ​​corresponding to the maximum value of ER, and the enhanced harmonic product spectrum H(α) is obtained as follows: H(a)=IES h*，j* (a)·IES h*，j* (2a)·…·IES h*，j* (Yes).

6. A rotating machinery fault diagnosis method according to claim 1, characterized in that: In step S4, the filter length L is determined by using sample entropy and power spectrum entropy, specifically: Calculate the power spectrum entropy DE(s0) and sample entropy PSE(s0) corresponding to s0 and normalize them; According to the normalized power spectrum entropy β(DE(s0)) and sample entropy β(PSE(s0), the PSE-DE corresponding to s0 is calculated as the measurement indicator: PSE-DE(s0)=β(DE(s0))·β(PSE(s0)); Output the filter length L corresponding to the maximum value of PSE-DE(s0).

7. A computer-readable storage medium storing a computer program, characterized in that: The computer program enables a computer to execute the rotating machinery fault diagnosis method according to any one of claims 1 to 6.

8. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the rotating machinery fault diagnosis method according to any one of claims 1 to 6 is implemented.

Images