Full-band beat vibration identification method of rotating equipment

CN120721213BActive Publication Date: 2026-08-07SHANGHAI MOONS AUTOMATION CONTROL CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202510895246.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-06-30
Publication Date
2026-08-07
Estimated Expiration
2045-06-30

AI Technical Summary

Technical Problem

但是该现有专利依赖转子转频(转速),且仅能分析与转子转频接近的频率范围,识别的过程存在主观因素,不支持实时在线识别等问题

Benefits of technology

[0022]1)本发明首先对振动信号进行VMD分解,获得若干模态函数,并计算各模态的包络谱,通过包络谱特征定量判别出可能为拍振成分的模态,避免依赖人工主观经验,提高了在复杂信号条件下的识别可靠性;

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120721213B_ABST
    Figure CN120721213B_ABST
Patent Text Reader

Abstract

The present application relates to a kind of full-band beat vibration identification methods of rotating equipment, the method comprises: using vibration velocity sensor to collect the vibration velocity signal of rotor;Using variable mode decomposition VMD to decompose vibration velocity signal, obtain multiple intrinsic mode function;The FFT transform of all intrinsic mode functions is obtained, and corresponding frequency spectrum is obtained;While the analytic signal of all intrinsic mode functions is obtained after Hilbert transform, the corresponding amplitude envelope is obtained, and the FFT transform of each amplitude envelope is carried out, and corresponding envelope spectrum is obtained;Envelope spectrum is analyzed, and whether intrinsic mode function can be beat vibration signal is judged, and whether it is beat vibration signal is further determined after analyzing frequency spectrum, and excitation component is obtained.Compared with prior art, the present application can effectively avoid the uncertainty caused by subjective judgment, improve the robustness when tracing analysis, and realize full-band beat vibration detection under the condition of no rotation frequency prior.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis technology for rotating equipment, and in particular to a method for identifying full-frequency band beat vibrations in rotating equipment. Background Technology

[0002] Beating vibration is a common abnormal vibration phenomenon in rotating equipment operation, characterized by slow, periodic fluctuations in the amplitude of the vibration signal in the time domain. This phenomenon is mainly caused by the interference of two excitation sources with similar frequencies. When the difference in excitation frequencies is within a specific range, the vibration energy will produce a significant modulation effect in the time domain. Taking an industrial centrifugal compressor as an example, when the impeller and main shaft become loose due to abnormal clearance, the impact excitation frequency of the loosened parts may approach the rotor's fundamental frequency, thus triggering continuous beating vibration. The periodic impact loads caused by this type of fault accelerate fatigue damage to mechanical components, deteriorate rotor dynamic balance, and may even lead to sealing component failure and media leakage, seriously threatening the safe operation of the equipment. Therefore, accurately identifying the existence of beating vibration and tracing its potential excitation sources and formation mechanisms can provide crucial information for fault diagnosis, equipment maintenance, and health management. Especially under multi-source excitation and complex operating conditions, effective identification and tracing methods help to discover potential risks in advance, improve equipment reliability, extend service life, and ensure the continuous operation of industrial systems.

[0003] Currently, technicians typically use time-domain and frequency-domain analysis to detect beat vibration phenomena in rotating equipment, but these methods have certain limitations. Although VMD decomposition can extract potential beat vibration components from aliased signals, subsequent discrimination is mainly achieved through waveform observation, which is prone to misjudgment when the signal is complex or the beat vibration characteristics are not significant, resulting in poor reliability and repeatability. Secondly, existing methods mostly use Hilbert demodulation spectrum to trace the source of excitation components, but this method is sensitive to boundary effects and usually requires data extension to reduce boundary influences. This may not only introduce artifacts or distortions but also limit the application capability of real-time online identification. In addition, detection methods based on rotor frequency have significant shortcomings in terms of applicability and comprehensiveness. On the one hand, such methods are highly dependent on prior information about the frequency; if the frequency is unknown or has deviations, identification is difficult to implement effectively. On the other hand, even if the frequency is known, the analysis is limited to its adjacent frequency bands, which may miss beat vibration characteristics existing in other frequency bands, making it difficult to achieve identification of signals in the entire frequency domain.

[0004] A search of Chinese Patent Publication No. CN111397877A reveals a method for detecting and diagnosing beat vibration faults in rotating machinery. Specifically, it discloses a method that first performs bilateral extension preprocessing on the shaft vibration displacement signal at fault-sensitive measurement points, then uses the VMD method to decompose the fault signal into multiple intrinsic mode functions (IMFs), improving upon the mode aliasing problem in existing feature signal extraction methods. Next, it uses waveform diagrams to qualitatively detect whether the IMF with the center frequency closest to the rotor's operating frequency is a beat vibration signal. Finally, it performs Hilbert demodulation analysis on the decomposed beat vibration signal, comprehensively employing Hilbert demodulation spectrum and instantaneous frequency spectrum analysis tools to extract fault feature information, obtaining accurate beat vibration fault characteristic frequencies. This solves the problem of insufficient frequency resolution leading to ineffective identification of beat vibration fault characteristic frequencies. However, this existing patent relies on rotor rotational frequency (speed) and can only analyze frequencies close to the rotor rotational frequency. The identification process is subjective and does not support real-time online identification. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art by providing a full-band beat vibration identification method for rotating equipment, which can effectively avoid the uncertainty caused by subjective judgment, improve the robustness of source analysis, and achieve full-band beat vibration detection without frequency prior.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] According to one aspect of the present invention, a method for identifying full-band beat vibration of a rotating device is provided, the method comprising:

[0008] Vibration velocity sensors are used to collect the vibration velocity signals of the rotor;

[0009] The vibration velocity signal was decomposed using variational mode decomposition (VMD) to obtain multiple intrinsic mode functions.

[0010] Perform an FFT (Fast Fourier Transform) on all intrinsic mode functions to obtain the corresponding spectrum; simultaneously, perform a Hilbert transform on all intrinsic mode functions to obtain the analytic signal, calculate the corresponding amplitude envelope, and perform an FFT on each amplitude envelope to obtain the corresponding envelope spectrum;

[0011] The envelope spectrum is analyzed to determine whether the intrinsic mode function may be a beat signal. Further analysis of the spectrum is then conducted to confirm whether it is a beat signal and to obtain the excitation component.

[0012] As a preferred technical solution, the vibration velocity sensor is installed at the rotor support.

[0013] As a preferred technical solution, the vibration velocity sensor is a vibration velocity sensor that is sensitive to low-frequency modulation.

[0014] As a preferred technical solution, after decomposing the vibration velocity signal using variational mode decomposition (VMD), multiple intrinsic mode functions (imf) are obtained. k (t) and its center frequency f k , where k = 1, 2, ..., k, ..., n, and n represents the number of decompositions.

[0015] As a preferred technical solution, for all intrinsic mode functions (IMF) k (t) Perform Hilbert transform to obtain the corresponding analytic signal, and calculate the corresponding amplitude envelope e. k (t), and then perform FFT transformation on each amplitude envelope to obtain the corresponding envelope spectrum.

[0016] As a preferred technical solution, for an intrinsic mode function (IMF) k The envelope spectrum of (t) The maximum peak value is calculated from its first m frequency components. and the corresponding frequency and the mean of the remaining amplitude Quantitative analysis and The relationship between them with f k The relationship between them, to judge Is it a significant low-frequency dominant component?

[0017] As a preferred technical solution, the number of the first m frequency components is preset, where m is related to the typical beat frequency range and frequency resolution of the device; if and but It is an envelope spectrum The significant low-frequency dominant component in the middle may be the beat frequency, where α is the significance factor of the envelope peak and β is the threshold of the envelope peak frequency; if this condition is not met, then the intrinsic mode function imf k (t) No beat vibration was detected.

[0018] As a preferred technical solution, the intrinsic mode function imf is first analyzed. k (t) is subjected to FFT transformation to obtain the corresponding spectrum. Then calculate the spectrum. The frequency of the first 3 peaks and And check if there is a difference between two peak frequencies that approximates the envelope spectrum. Maximum peak frequency Finally, based on the inspection results, it was determined whether the intrinsic mode function was a beat vibration signal, and two excitation components were obtained.

[0019] As a preferred technical solution, all peak frequency pairs are iterated, and the frequency difference Δf between each pair of peak frequencies is calculated. If there exists a pair of peak frequencies such that... Then determine the intrinsic mode function imf k (t) is the beat vibration signal, and the two excitation components forming the beat vibration are this pair of peak frequencies, where γ is the beat frequency tolerance; if this is not satisfied, then the intrinsic mode function imf k (t) No beat vibration was detected.

[0020] As a preferred technical solution, the beat frequency tolerance is used to control the discrimination boundary of beat vibration identification.

[0021] Compared with the prior art, the present invention has the following advantages:

[0022] 1) This invention first performs VMD decomposition on the vibration signal to obtain several mode functions, and calculates the envelope spectrum of each mode. By quantitatively identifying the modes that may be beat vibration components through the envelope spectrum features, it avoids relying on human subjective experience and improves the reliability of identification under complex signal conditions.

[0023] 2) This invention performs FFT analysis on the relevant modal functions and extracts the excitation frequency components by combining the envelope spectrum and spectral information, thus overcoming the problems of dependence on frequency conversion information and incomplete frequency band coverage of traditional methods;

[0024] 3) This invention eliminates the need for data extension processing, reducing the risk of boundary artifacts and distortion, and improving the practicality and stability of the algorithm in real-time online recognition scenarios. Attached Figure Description

[0025] Figure 1 This is a flowchart illustrating an embodiment of the present invention.

[0026] Figure 2 (a) to (c) are time-domain waveforms of each signal component of the simulated signal v(t) in the embodiments; Figure 2 (d) is the time-domain waveform of the noise component of the simulated signal v(t) in the embodiment;

[0027] Figure 3 (a) is the time-domain waveform of the simulated signal v(t) in the embodiment; Figure 3 (b) is the spectrum of the simulated signal v(t) in the embodiment;

[0028] Figure 4 These are all intrinsic mode functions (imf) obtained after VMD decomposition in the embodiment. k A schematic diagram of (t);

[0029] Figure 5 This is a schematic diagram of the spectrum and envelope spectrum of the intrinsic mode function imf3(t) in the embodiment;

[0030] Figure 6 This is a schematic diagram of the beat vibration identification and source tracing results of the simulated signal v(t) in the embodiment;

[0031] Figure 7 This is a schematic diagram of the beat vibration identification and source tracing result of the simulated signal v(t) in the embodiment. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0033] This invention provides a full-band beat vibration identification method for rotating equipment. First, a vibration velocity sensor sensitive to beat vibration is used to acquire the vibration signal of the rotor. Then, VMD is used to decompose the vibration signal, obtaining multiple intrinsic mode functions (IMFs). An FFT transform is performed on all IMFs to obtain their corresponding spectra. Furthermore, a Hilbert transform is performed on all IMFs to obtain analytic signals, and the corresponding amplitude envelopes are calculated. An FFT transform is then performed on each amplitude envelope to obtain its corresponding envelope spectrum. The envelope spectrum is analyzed to determine if the IMFs might be beat vibration signals. Further spectrum analysis confirms whether it is a beat vibration signal and obtains the excitation component. This invention provides a method for beat vibration identification and excitation component tracing across the entire frequency band of a vibration signal without frequency switching. By combining the analysis of IMF spectra and envelope spectra, it effectively solves the problems of mode aliasing and insufficient frequency resolution, improving the accuracy and reliability of signal processing.

[0034] Compared with existing technologies (such as CN111397877A), the present invention has the following specific advantages:

[0035] 1) Existing technologies generally rely on rotor frequency (speed), while this invention does not rely on rotor frequency;

[0036] 2) Scope of analysis: Existing technologies can generally only analyze the frequency range close to the rotor rotation frequency, while this invention performs full-band analysis;

[0037] 3) In the identification process, are there any subjective factors? Existing patents use VMD → to judge the "beating" of the IMF (subjective qualitative judgment of "whether there are time-domain waveform characteristics of beating") → FFT spectrum (quantitative) → Hilbert demodulation spectrum (quantitative) → to use Hilbert instantaneous frequency spectrum to determine the excitation component (subjective qualitative judgment of "whether the maximum instantaneous frequency appears at the beat peak or beat trough") → to obtain a set of beating frequency combinations near the rotor rotation frequency; while this invention uses VMD → FFT spectrum (quantitative) → Hilbert envelope spectrum (quantitative) → to obtain a set of beating frequency combinations across the entire frequency band (no subjective judgment is required throughout the process);

[0038] 4) In the signal analysis process, existing patents use Hilbert demodulation spectrum + Hilbert instantaneous frequency spectrum, which is sensitive to boundary effects and requires extension processing, and does not support real-time online recognition; while this invention uses Hilbert envelope spectrum, which is not sensitive to boundary effects, has higher robustness, does not require extension processing, and performs better in scenarios with fewer sampling points and smaller frequency differences; it supports real-time online recognition.

[0039] The following specific embodiments verify the effectiveness of the invention using simulated vibration velocity signals of the rotor. The expression for the simulated signal v(t) is shown in formula (1), and it consists of four parts. Wherein v1(t) represents a single component with a frequency of 15Hz, as shown in... Figure 2 As shown in (a); v2(t) represents a simulated beat signal with excitation components of 48Hz and 50Hz, a frequency difference of 2Hz, forming a beat frequency of 1Hz. Figure 2 (b) shows that the amplitude of the time-domain waveform v2(t) exhibits a slow, periodic change; v3(t) represents another simulated beat signal, with excitation components of 99Hz and 100Hz, a frequency difference of 1Hz, forming a beat frequency of 0.5Hz. Figure 2 As can be seen in (c), the amplitude of the time-domain waveform of v3(t) also exhibits a slow, periodic change; n(t) represents Gaussian white noise with a standard deviation of 2.5, such as Figure 2 As shown in (d);

[0040] v(t)=v1(t)+v2(t)+v3(t)+n(t)(1)

[0041] in:

[0042]

[0043] The simulation signal was discretely sampled for 1 second using a sampling rate of 2048Hz, and the resulting simulation signal is shown in Figure (3). Figure 3 As can be seen in the time-domain waveform diagram of (a), due to the aliasing of noise and various frequency components, it is difficult to directly observe obvious beat vibration phenomena from the time-domain waveform diagram. Furthermore, in Figure 3 In the spectrum diagram of (b), due to the limitation of the actual number of sampling points, the resolution of the spectrum is only 1 Hz. The frequency difference between the excitation components in the simulated beat vibration signals v2(t) and v3(t) is 1 Hz, which is very close. This makes it easy for the excitation component method to mistakenly identify them as a single frequency component, leading to inaccurate identification of the beat vibration phenomenon. Next, the method proposed in this invention will be used to analyze the simulated signal v(t) to verify its effectiveness and accuracy in beat vibration identification and source tracing.

[0044] According to Figure (1), a method for identifying and tracing the vibration of a rotating device specifically includes the following steps:

[0045] 1) Use the simulated signal v(t) as the object of analysis and research;

[0046] 2) From Figure 3 The spectrum diagram in (b) reveals three distinct frequency range components and a noise component in the simulated signal v(t). Based on the spectral characteristics of v(t), the number of intrinsic mode functions n for VMD decomposition is set to 6, and the bandwidth balancing parameter is set to 6000 to obtain modes with higher separation and more concentrated frequencies. Then, VMD is used to decompose the simulated signal v(t), resulting in six intrinsic mode functions from imf1(t) to imf6(t), as follows: Figure 4 As shown; the center frequencies of imf1(t) to imf3(t) are 14.75Hz, 99.38Hz and 48.28Hz respectively, which correspond to the 15Hz component v1(t), the analog beat signal v3(t) and the analog beat signal v2(t). Figure 2 (a) Figure 2 (b) and Figure 2 (c) respectively with Figure 3 (a) Figure 2 (c) and Figure 2 (d) By comparison, it can be found that VMD can effectively extract the effective frequency components from the simulated signal v(t), thus avoiding the occurrence of mode aliasing.

[0047] 3) Identify and trace the source of beat vibration across the entire frequency band of the signal v(t) without relying on the rotor frequency. This involves examining all intrinsic mode functions (IMF). k (t) Perform Hilbert transform to obtain the respective analytic signals, and then calculate the respective amplitude envelope e. k (t), and then perform FFT transform on the amplitude envelope to obtain their respective envelope spectra.

[0048] 4) The frequency differences between the excitation components of the simulated beat vibration signals v2(t) and v3(t) in the simulated signal v(t) are 2Hz and 1Hz, respectively. Therefore, a beat vibration identification range of 5Hz is sufficient for this rotating device. Furthermore, the frequency resolution of the simulated signal is 1Hz, so the number of the first m frequency components is set to 5. The envelope peak significance factor α is set to 2, and the envelope peak frequency threshold β is set to 0.25. Here, we take… Figure 4 (c) is used as an example for detailed analysis of imf3(t), whose center frequency f3 = 48.28 Hz, and the envelope spectrum of imf3(t). like Figure 5 As shown. The maximum peak value is calculated as follows. Frequency corresponding to the maximum peak Mean of remaining amplitude Quantitative analysis and illustrate It is an envelope spectrum The low-frequency dominant component is significant and is twice the beat frequency; proceed to step 5.

[0049] 5) Perform an FFT transform on the intrinsic mode function imf3(t) to obtain the spectrum. like Figure 5 As shown, the frequencies of the first three peaks are calculated. The beat frequency tolerance γ is set to 1Hz; and The frequency difference Δf = 2Hz; This indicates that imf3(t) is the beat signal. It is twice the beat rate. and These are the two excitation components that form the beat vibration.

[0050] 6) For all intrinsic mode functions imf k (t) Proceed to steps 4) and 5), and the final identification and tracing results are as follows: Figure 6 , Figure 7 As shown: imf2(t) was identified as a beat vibration signal with excitation components of 99Hz and 100Hz, consistent with the simulated beat vibration signal v3(t); imf3(t) was also identified as a beat vibration signal with excitation components of 48Hz and 50Hz, consistent with the simulated beat vibration signal v2(t). This analysis result is consistent with the theoretical assumptions of the simulated signal. Therefore, this invention can achieve quantitative, non-subjective identification of beat vibration signals and possesses the ability to trace excitation sources across the entire frequency band without relying on frequency conversion, solving the problems of traditional methods that rely on human experience, have limited identification range, and poor real-time performance.

[0051] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for identifying full-frequency beat vibrations in rotating equipment, characterized in that, The method includes: Vibration velocity sensors are used to collect the vibration velocity signals of the rotor; Variational mode decomposition (VMD) was used to decompose the vibration velocity signal, yielding multiple intrinsic mode functions. and its center frequency ,in , Indicates the number of decompositions; Perform an FFT transform on all intrinsic mode functions to obtain the corresponding spectrum; simultaneously, perform an FFT transform on all intrinsic mode functions. After performing the Hilbert transform, the analytic signal is obtained, and the corresponding amplitude envelope is calculated. Then, an FFT transformation is performed on each amplitude envelope to obtain the corresponding envelope spectrum. ; For an eigenmode function envelope spectrum From its front The maximum peak value was calculated from each frequency component. and the corresponding frequency and the mean of the remaining amplitude Quantitative analysis and The relationship between them and The relationship between them, to judge Is it a significant low-frequency dominant component? Before setting The number of frequency components, It is related to the typical beat range and frequency resolution of the equipment; if and ,but It is an envelope spectrum The dominant low-frequency component is the beat frequency, in which... As a significant factor for the envelope peak, This is the envelope peak frequency threshold; if it is not satisfied, then the intrinsic mode function... No beat vibration was detected; The envelope spectrum is analyzed to determine whether the intrinsic mode function could be a beat signal. Further analysis of the spectrum confirms whether it is a beat signal and obtains the excitation component. The specific process is as follows: First, the intrinsic mode functions... Perform an FFT transform to obtain the corresponding spectrum. Then calculate the spectrum. The frequency of the first 3 peaks , and And check if the difference between two peak frequencies approximates the envelope spectrum. Maximum peak frequency Finally, based on the inspection results, it is determined whether the intrinsic mode function is a beat vibration signal, and two excitation components are obtained. Iterate through all peak frequency pairs and calculate the frequency difference between each pair of peak frequencies. If there exists a pair of peak frequencies such that Then determine the intrinsic mode function. It is a beat vibration signal, and the two excitation components that form the beat vibration are this pair of peak frequencies, where This is the beat frequency tolerance; if it is not satisfied, then the intrinsic mode function... No beat vibration was detected.

2. The method for full-frequency band beat vibration identification of a rotating device according to claim 1, characterized in that, The vibration velocity sensor is installed at the rotor support.

3. The method for full-frequency band beat vibration identification of a rotating device according to claim 1 or 2, characterized in that, The vibration velocity sensor is a vibration velocity sensor that is sensitive to low-frequency modulation.

4. The method for full-band beat vibration identification of rotating equipment according to claim 1, characterized in that, The beat frequency tolerance is used to control the discrimination boundary for beat vibration identification.

Citation Information

Patent Citations

  • Rotary machine beat vibration fault detection and diagnosis method

    CN111397877A

  • Method for distinguishing true and false beat vibration

    CN106197655A

  • Multi-resolution dynamic signal time domain and spectrum feature analysis method and device

    CN117606724A