Rotary machinery fault positioning method based on adaptive frequency band and medium

Abstract

The invention relates to a rotating machine fault positioning method based on a self-adaptive frequency band and a medium, and the method comprises the steps: synchronously collecting acoustic signals under a multi-channel stable working condition through an acoustic array, and carrying out the preprocessing of the acoustic signals under the multi-channel stable working condition; performing coherence-kurtosis combined filtering on the preprocessed acoustic signal to obtain a dominant frequency point in the acoustic signal; adaptive filtering is carried out on dominant frequency points in the acoustic signals, and candidate frequency bands are generated; and performing sound field imaging on each frequency band in the candidate frequency bands, generating n sound images for each frequency band, and dynamically selecting an optimal sound image through an energy saliency scoring mechanism and a weighted scoring function to complete fault positioning of the rotating machinery. According to the invention, self-adaptive imaging frequency band selection can be completed, and the fault positioning precision and effectiveness are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of rotating machinery fault diagnosis, and in particular to a rotating machinery fault locating method and medium based on an adaptive frequency band. Background Art

[0002] Rotating machinery is the core equipment in the fields of energy, chemical industry, aviation, etc. The accuracy of its fault location is directly related to production safety and economic efficiency. Although traditional vibration analysis can detect faults, it is limited by the installation constraints and frequency band limitations of contact measurement, making it difficult to achieve spatial positioning under complex working conditions. Acoustic imaging technology realizes non-contact, wide-band fault visualization through sound field imaging. It is particularly suitable for locating abnormal sound sources and detecting early weak faults. However, its performance is highly dependent on the selection of imaging frequency. It is necessary to match the fault characteristic frequency band (such as high-frequency resonance of bearings or gear meshing modulation sidebands) to overcome industrial site noise interference and frequency band aliasing problems. Variable speed operation causes dynamic changes in characteristic frequencies, requiring the frequency band selection algorithm to have adaptive capabilities. The commonly used imaging frequency band selection methods are mainly:

[0003] (1) Traditional frequency band selection method based on signal processing

[0004] Extracting signal features through fast Fourier transforms, Hilbert transforms, or wavelet transforms requires prior knowledge to determine the analysis frequency band. Without requiring large-scale data training, characteristic frequency bands can be directly selected using frequency domain energy distribution or statistical indicators (such as kurtosis). Preset thresholds (such as wavelet hard thresholds) or frequency band energy ratios (such as spectral kurtosis) can be used to filter out non-sensitive frequency bands, providing a stable suppression effect on periodic interference (such as power frequency noise). However, frequency band selection requires a predefined fault characteristic frequency range (such as a bearing fault characteristic formula), resulting in limited signal separation effectiveness for unknown fault types or complex faults, and the potential for missing potentially sensitive frequency bands.

[0005] (2) Frequency band selection method based on intelligent algorithm

[0006] Intelligent algorithms (such as gray wolf optimization, genetic algorithms, and CNN networks) can automatically search for the optimal frequency band combination by dynamically adapting to signal characteristics. However, black box models such as CNN find it difficult to explain the physical mechanism of frequency band selection, which limits the credibility of technical applications in safety-critical scenarios (such as nuclear power plant equipment monitoring). Summary of the Invention

[0007] The technical problem to be solved by the present invention is to provide a rotating machinery fault location method and medium based on adaptive frequency band, which can complete adaptive imaging frequency band selection and improve the fault location accuracy and effectiveness.

[0008] The present invention solves the technical problem by adopting a technical solution: providing a rotating machinery fault location method based on an adaptive frequency band, comprising the following steps:

[0009] Acoustic arrays are used to synchronously collect acoustic signals under multi-channel stable working conditions, and the acoustic signals under multi-channel stable working conditions are preprocessed;

[0010] Perform coherence-kurtosis joint filtering on the preprocessed acoustic signal to obtain the dominant frequency points in the acoustic signal;

[0011] Adaptively filtering the dominant frequency points in the acoustic signal to generate candidate frequency bands;

[0012] Performing sound field imaging on each of the candidate frequency bands, generating n acoustic images for each frequency band, obtaining the optimal acoustic image through an energy significance scoring mechanism and dynamically selecting the weighted scoring function, and completing the rotating machinery fault location.

[0013] The preprocessing of the acoustic signal under the multi-channel stable working condition specifically includes:

[0014] Decentralize the collected multi-channel acoustic signals;

[0015] Perform partial coherence analysis on the decentralized acoustic signal.

[0016] The coherence-kurtosis combined filtering is performed on the pre-processed acoustic signal to obtain the dominant frequency points in the acoustic signal, specifically including:

[0017] Calculating the coherence function of the preprocessed acoustic signal and retaining the frequency bands whose coherence coefficient exceeds the coherence coefficient threshold;

[0018] The spectral kurtosis value is calculated for the retained frequency band, and the frequency band whose spectral kurtosis value is greater than the spectral kurtosis threshold is extracted to obtain the dominant frequency point in the acoustic signal.

[0019] The coherence function is calculated as follows: Where γ(f) is the coherence coefficient at frequency f, S xy (f) is the cross power spectrum of signal x(t) and signal y(t) at frequency f, S xx (f) is the autopower spectrum of the signal x(t) at frequency f, S yy (f) is the autopower spectrum of signal y(t) at frequency f.

[0020] The spectral kurtosis value is calculated as follows: Where SK(f) is the spectral kurtosis value at frequency f, X i (k) is the discrete Fourier transform of the signal x(t), and K is the total number of data segments.

[0021] The adaptive filtering of the dominant frequency points in the acoustic signal to generate candidate frequency bands specifically includes:

[0022] Applying a Butterworth bandpass filter to the acoustic signal of the extracted dominant frequency point in time sequence to obtain a first intermediate signal;

[0023] Inverting the time sequence of the first intermediate signal and applying the Butterworth bandpass filter to the first intermediate signal to obtain a second intermediate signal;

[0024] The second intermediate signal is time-reversed to obtain a candidate frequency band signal.

[0025] The transfer function of the Butterworth bandpass filter is: Where H(s) is the transfer function of the Butterworth bandpass filter, H0 is the passband gain, Q is the quality factor, ω0 is the center angular frequency, and s is the operator.

[0026] The acquisition of the optimal sonogram by the energy saliency scoring mechanism and the dynamic selection of the weighted scoring function specifically include:

[0027] Calculate the main lobe energy ratio and side lobe suppression level evaluation of each sonogram in each frequency band;

[0028] Superimposing the acoustic images in which the main lobe energy ratio exceeds the sufficient main lobe energy ratio threshold and the side lobe suppression level evaluation is less than the side lobe suppression level evaluation threshold to obtain a dominant acoustic image in each frequency band;

[0029] The dominant sonogram of each frequency band is designed with a weighted scoring function to dynamically select the optimal sonogram of the target sound source.

[0030] The weighted scoring function is: Score = αE r +β(S sup / 15)+γcosθ, where Score is the weighted score, E r is the energy proportion of the main lobe, S sup is the sidelobe suppression level evaluation, cosθ is the cosine similarity between the dominant sonogram and the template sonogram in each frequency band, and α, β, and γ are all weight coefficients.

[0031] The technical solution adopted by the present invention to solve its technical problem is: providing a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned rotating machinery fault locating method based on adaptive frequency band are implemented.

[0032] Beneficial effects

[0033] Due to the adoption of the above-mentioned technical solution, the present invention offers the following advantages and positive effects compared to existing technologies: By adaptively acquiring dominant frequency points for fault imaging through coherence-kurtosis combined filtering, combined with improved adaptive bandpass filtering to generate candidate frequency bands, the present invention can directly achieve sound field imaging, obtain optimal acoustic images, and locate rotating machinery faults. Compared to manually predefined frequency bands, the present invention is more convenient to process and can acquire sensitive frequency bands in the signal without the need for large-scale data training, thus improving the accuracy and reliability of rotating machinery fault location. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 is a flow chart of a rotating machinery fault location method based on adaptive frequency band according to an embodiment of the present invention;

[0035] Figure 2 is a flow chart of coherence-kurtosis joint filtering in an embodiment of the present invention;

[0036] Figure 3 is a flow chart of adaptive filtering in an embodiment of the present invention;

[0037] Figure 4 This is the optimal acoustic image fault location result diagram of the target sound source in an embodiment of the present invention. DETAILED DESCRIPTION

[0038] Below in conjunction with specific embodiment, further set forth the present invention.Should be understood that these embodiments are only used to illustrate the present invention and are not used in limiting the scope of the present invention.In addition, should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall equally within the scope limited by the appended claims of the application.

[0039] The first embodiment of the present invention relates to a rotating machinery fault location method based on adaptive frequency band, such as Figure 1 As shown, the following steps are included:

[0040] Step 1: Use an acoustic array to synchronously collect acoustic signals under multi-channel stable working conditions and pre-process the acoustic signals under multi-channel stable working conditions. This step is as follows:

[0041] In step S11 , a multi-channel array is used to collect acoustic signals. The acoustic array is placed according to the size of the experimental object, the spatial position and the experimental environment, and the measurement distance and measurement method are set.

[0042] Step S12, centralization preprocessing: Decentralize the collected multi-channel acoustic signals to eliminate the influence of the DC component, as shown in the following formula:

[0043]

[0044] Where x(t)=(x1(t),x2(t),…,x n (t)) Τ ,t=1,2,…,N is the signal sample, N is the sample length, n is the number of samples, is the decentralized acoustic signal.

[0045] Step S13, partial coherence analysis preprocessing: Based on decentralization, partial coherence analysis processing is performed on the acoustic signal to enhance the signal-to-noise ratio of the rotating machinery periodic signal, as shown in the following formula:

[0046]

[0047] Where X1 is the acoustic signal x 1 The spectrum of the sound signal x 2 The spectrum, S 12 is x 1 and x 2 The spectrum of the cross-correlation function, S 11 is x 1 Autocorrelation function spectrum, x 3 is the acoustic signal x 1 The enhanced signal based on partial coherence analysis.

[0048] Step 2: Perform coherence-kurtosis combined filtering on the pre-processed acoustic signal to obtain the dominant frequency points in the acoustic signal. This step uses coherence-kurtosis combined filtering to process the multi-channel acoustic signal to obtain the dominant frequency band with strong coherence and high impact in the acoustic signal. The coherence-kurtosis combined filtering process is as follows: Figure 2 The specific steps are as follows:

[0049] Step S21 performs a coherence-kurtosis joint filtering on the preprocessed acoustic signal. Specifically, the coherence function of the preprocessed multi-channel acoustic signal is calculated, and the frequency bands where the coherence coefficient exceeds the coherence coefficient threshold are retained. That is, only the signal in the high-coherence frequency band is retained, and the low-coherence region dominated by noise is eliminated. In this embodiment, the coherence function is the ratio of the square of the cross-power spectral density of the two signals to the product of their respective auto-power spectral densities. Its calculation formula is:

[0050]

[0051] Where γ(f) is the coherence coefficient at frequency f, S xx (f) is the autopower spectrum of the signal x(t) at frequency f,

[0052] S yy (f) is the autopower spectrum of signal y(t) at frequency f, S xy(f) is the cross power spectrum of signal x(t) and signal y(t) at frequency f. In this embodiment, the coherence coefficient threshold can be set to 0.7, and the frequency band components with a coherence coefficient greater than 0.7 are retained.

[0053] Step S22 further screens the frequencies with high coherence, i.e., calculates the spectral kurtosis value for the retained frequency bands and extracts the frequency bands with spectral kurtosis values ​​greater than the spectral kurtosis threshold to obtain the dominant frequency points in the acoustic signal. This step can extract high-impact component frequency bands with high kurtosis values. For frequency f, the theoretical definition of spectral kurtosis SK(f) is:

[0054]

[0055] Where SK(f) is the spectral kurtosis value at frequency f, X i (k) is the discrete Fourier transform of the signal x(t), and K is the total number of data segments. In this embodiment, the spectral kurtosis threshold can be set to 3, and the frequency band components with spectral kurtosis values ​​greater than 3 are retained.

[0056] Step 3, adaptively filter the dominant frequency point in the acoustic signal to generate a candidate frequency band. This step can use an improved adaptive bandpass filter to retain the frequency band near the dominant frequency point in the rotating machinery and suppress low-frequency environmental noise. The improved adaptive bandpass filter implementation process is as follows: Figure 3 As shown in Figure 1, the phase distortion introduced by traditional filtering is eliminated through bidirectional filtering. The specific steps are as follows:

[0057] Forward filtering: Apply the Butterworth bandpass filter to the input signal x[n] in time sequence to obtain the intermediate signal y1[n]. Butterworth bandpass analog filter transfer function:

[0058]

[0059] in, is the center angular frequency, is the quality factor, which determines the passband width. H0 is the passband gain, and s is the operator.

[0060] Converting the analog filter H(s) to a digital filter H(z) yields:

[0061]

[0062] The high-order filter directly expressed in the form of transfer function will lead to numerical instability, so the transfer function is decomposed into the product of multiple second-order sections. The entire filter is cascaded through the second-order sections to make the numerical calculation more stable. Each second-order section structure H k (z) as follows:

[0063]

[0064] Reverse filtering: The intermediate signal y1[n] is time-reversed and passed through the same Butterworth bandpass filter again to generate the intermediate signal y2[n].

[0065] Result inversion: The intermediate signal y2[n] is inverted in time sequence and the final result y[n] is output. The final result y[n] is the candidate frequency band signal.

[0066] Through the above-mentioned forward and reverse filtering, the phase responses can be canceled out, so that the total phase delay is 0.

[0067] Step 4: Perform sound field imaging on each frequency band in the candidate frequency bands, generate n sound images for each frequency band, obtain the optimal sound image through the energy significance scoring mechanism and dynamically select the weighted scoring function to complete the rotating machinery fault location.

[0068] The details of this step are as follows:

[0069] Step S41: The energy significance scoring mechanism includes main lobe energy proportion and side lobe suppression level evaluation, and the main lobe energy proportion and side lobe suppression level evaluation are calculated for each sonogram in each frequency band. Specifically, the main lobe energy proportion is defined as the range of the main lobe region of the sonogram as the peak energy ±3 octaves, and the energy proportion in this region is calculated:

[0070]

[0071] Among them, E m is the main lobe energy, E t is the total energy of the sonogram, E r For the main lobe energy ratio, E is preferred. r Higher sonogram ensures focused energy.

[0072] The sidelobe suppression level is evaluated by calculating the ratio of the maximum sidelobe energy outside the main lobe to the main lobe peak energy:

[0073]

[0074] Among them, E side Maximum sidelobe energy, S sup To evaluate the sidelobe suppression level, screen S sup Smaller sonogram, suppressing artifact interference.

[0075] Step S42: Superimpose the acoustic images that meet both the mainlobe energy ratio threshold and the sidelobe suppression level assessment threshold to generate a dominant acoustic image for each frequency band. In this embodiment, the mainlobe energy ratio threshold can be set to 0.7, and the sidelobe suppression level assessment threshold can be set to -15dB. The acoustic images with a mainlobe energy ratio greater than 0.7 and a sidelobe suppression level assessment less than -15dB are retained. The retained acoustic images are then superimposed for each frequency band to generate the dominant acoustic image for each frequency band.

[0076] Step S43: Design a weighted scoring function for the dominant acoustic image of each frequency band to dynamically select the optimal acoustic image of the target sound source. The designed weighted scoring function is shown in the following formula:

[0077] Score = αE r +β(S sup / 15)+γcosθ

[0078] Where Score is a weighted score, cosθ is the cosine similarity between the dominant acoustic image in each frequency band and the template acoustic image. A greater cosine similarity indicates a more accurate acoustic image fault location. α, β, and γ are weight coefficients, which are set based on real-time stable operating conditions. If there is no template acoustic image, the weight coefficient γ can be set to 0. In this embodiment, the weight coefficients can be set based on real-time stable operating conditions as follows: α = 0.6, β = 0.3, and γ = 0.1.

[0079] Step S44: Obtain the target sound source position coordinates on the optimal sound image, calculate the positioning error, and achieve the fault location of the target sound source of the rotating machinery. The calculation formula of the positioning error is as follows:

[0080]

[0081] Among them, err is the positioning error, and the true position is (x t ,y t ), measuring position (x m ,y m ).

[0082] In this embodiment, the fault location result is as follows: Figure 4 As shown in the figure, the coordinates of the target sound source fault location are [-0.35, 0.44], the actual estimated fault location is [-0.344, 0.441], and the positioning error is 1.5 cm.

[0083] The present invention, through adaptively acquiring dominant frequency points for fault imaging through coherence-kurtosis combined filtering and combining it with improved adaptive bandpass filtering to generate candidate frequency bands, can directly achieve sound field imaging, obtain optimal acoustic images, and locate rotating machinery faults. Compared to manually predefined frequency bands, this method is more convenient to process and can capture sensitive frequency bands in the signal without requiring large-scale data training, thus improving the accuracy and reliability of rotating machinery fault location.

[0084] A second embodiment of the present invention relates to a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the computer program implements the steps of the rotating machinery fault location method based on adaptive frequency bands of the first embodiment.

[0085] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage and optical storage, etc.) that contain computer-usable program code.

[0086] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0087] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction method, which is implemented in the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0088] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are generated to realize the computer-implemented processes in the computer or other programmable devices, and the instructions executed in the computer or other programmable devices provide the steps for realizing the functions specified in the flowchart Figure 1 one flow or multiple flows and / or the functions specified in the block Figure 1 one block or multiple blocks.

[0089] The above description is merely a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A rotating machinery fault location method based on adaptive frequency band, characterized in that: The following steps are involved: Acoustic arrays are used to synchronously collect acoustic signals under multi-channel stable working conditions, and the acoustic signals under multi-channel stable working conditions are preprocessed; Perform coherence-kurtosis joint filtering on the preprocessed acoustic signal to obtain the dominant frequency points in the acoustic signal; Adaptively filtering the dominant frequency points in the acoustic signal to generate candidate frequency bands; Performing sound field imaging on each of the candidate frequency bands, generating n acoustic images for each frequency band, obtaining the optimal acoustic image through an energy significance scoring mechanism and dynamically selecting the weighted scoring function, and completing the rotating machinery fault location.

2. The rotating machinery fault location method based on adaptive frequency band according to claim 1, characterized in that: The preprocessing of the acoustic signal under the multi-channel stable working condition specifically includes: Decentralize the collected multi-channel acoustic signals; Perform partial coherence analysis on the decentralized acoustic signal.

3. The rotating machinery fault location method based on adaptive frequency band according to claim 1, characterized in that: The coherence-kurtosis combined filtering is performed on the pre-processed acoustic signal to obtain the dominant frequency points in the acoustic signal, specifically including: Calculating the coherence function of the preprocessed acoustic signal and retaining the frequency bands whose coherence coefficient exceeds the coherence coefficient threshold; The spectral kurtosis value is calculated for the retained frequency band, and the frequency band whose spectral kurtosis value is greater than the spectral kurtosis threshold is extracted to obtain the dominant frequency point in the acoustic signal.

4. The rotating machinery fault location method based on adaptive frequency band according to claim 3, characterized in that: The coherence function is calculated as follows: Where γ(f) is the coherence coefficient at frequency f, S xy (f) is the cross power spectrum of signal x(t) and signal y(t) at frequency f, S xx (f) is the autopower spectrum of the signal x(t) at frequency f, S yy (f) is the autopower spectrum of signal y(t) at frequency f.

5. The rotating machinery fault location method based on adaptive frequency band according to claim 3, characterized in that: The spectral kurtosis value is calculated as follows: Where SK(f) is the spectral kurtosis value at frequency f, X i (k) is the discrete Fourier transform of the signal x(t), and K is the total number of data segments.

6. The rotating machinery fault location method based on adaptive frequency band according to claim 1, characterized in that: The adaptive filtering of the dominant frequency points in the acoustic signal to generate candidate frequency bands specifically includes: Applying a Butterworth bandpass filter to the acoustic signal of the extracted dominant frequency point in time sequence to obtain a first intermediate signal; Inverting the time sequence of the first intermediate signal and applying the Butterworth bandpass filter to the first intermediate signal to obtain a second intermediate signal; The second intermediate signal is time-reversed to obtain a candidate frequency band signal.

7. The rotating machinery fault location method based on adaptive frequency band according to claim 6, characterized in that: The transfer function of the Butterworth bandpass filter is: Where H(s) is the transfer function of the Butterworth bandpass filter, H0 is the passband gain, Q is the quality factor, ω0 is the center angular frequency, and s is the operator.

8. The rotating machinery fault location method based on adaptive frequency band according to claim 1, characterized in that: The acquisition of the optimal sonogram by the energy saliency scoring mechanism and the dynamic selection of the weighted scoring function specifically include: Calculate the main lobe energy ratio and side lobe suppression level evaluation of each sonogram in each frequency band; Superimposing the acoustic images in which the main lobe energy ratio exceeds the sufficient main lobe energy ratio threshold and the side lobe suppression level evaluation is less than the side lobe suppression level evaluation threshold to obtain a dominant acoustic image in each frequency band; The dominant sonogram of each frequency band is designed with a weighted scoring function to dynamically select the optimal sonogram of the target sound source.

9. The rotating machinery fault location method based on adaptive frequency band according to claim 8, characterized in that: The weighted scoring function is: Score = αE r +β(S sup / 15)+γcosθ, where Score is the weighted score, E r is the energy proportion of the main lobe, S sup is the sidelobe suppression level evaluation, cosθ is the cosine similarity between the dominant sonogram and the template sonogram in each frequency band, and α, β, and γ are all weight coefficients.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the rotating machinery fault location method based on adaptive frequency bands as claimed in any one of claims 1 to 9 are implemented.

Images