Gear vibration signal decomposition method based on improved group sparse mode decomposition

Abstract

The invention relates to the technical field of signal and information processing, and particularly discloses a gear vibration signal decomposition method based on improved group sparse mode decomposition, which comprises the following steps: S1, measuring a gearbox to obtain gear vibration data; s2, using discrete Fourier transform to obtain a vibration signal frequency spectrum; s3, performing group sparse optimization on the frequency spectrum, and constructing a zero-phase filter group; s4, decomposing signal frequency spectrum filtering by using a zero-phase filter bank, and performing inverse Fourier transform to obtain a modal component; s5, calculating the periodicity of all groups of sparse mode components; s6, fusing the group sparse mode components with relatively high period similarity into an intrinsic mode component; and S7, selecting a fault component, extracting a fault characteristic frequency, and carrying out fault diagnosis. According to the method, the problem that the integrity of the gear fault feature information is damaged due to an over-decomposition phenomenon generated based on group sparse mode decomposition in the prior art is solved, the gear fault feature information can be protected, and the vibration feature of the gear signal is effectively extracted.

Description

Technical Field

[0001] The invention relates to the technical field of signal and information processing, and specifically provides a gear vibration signal decomposition method based on improved group sparse modal decomposition. Background Art

[0002] The gearbox is a core component in the mechanical transmission system. Its operating status is directly related to the performance and safety of the equipment. The failure of the gearbox will not only reduce the efficiency of the equipment, but also may cause serious safety accidents and economic losses. Therefore, it is of great significance to diagnose the gearbox failure in a timely and accurate manner. Signal decomposition technology is one of the core tools for gearbox fault diagnosis. Signal decomposition technology decomposes the complex original vibration signal into several physically meaningful components (such as modes or characteristic components of specific frequency bands) to more intuitively identify the fault characteristics.

[0003] As a classic signal decomposition method, EMD can decompose complex nonlinear and non-stationary signals into a set of physically meaningful intrinsic mode functions (IMFs), thereby effectively extracting local features from the signal. It does not require preset basis functions and relies entirely on the data itself. Therefore, it is particularly suitable for processing dynamic signals that cannot be accurately described by traditional methods and is suitable for fault diagnosis of mechanical equipment. However, EMD has problems such as modal aliasing, boundary effects, and sensitivity to noise, which may lead to distortion of the decomposition results, especially when processing strong noise signals or complex frequency components.

[0004] Ensemble empirical mode decomposition (EEMD) is an improved method of EMD, which can effectively alleviate the modal aliasing problem, improve the stability and accuracy of decomposition, and thus extract key modal information from the characteristic signal of mechanical equipment fault. However, EEMD has a large amount of calculation and is sensitive to noise amplitude and number of iterations, which may affect real-time performance and decomposition accuracy. In addition, boundary effects and strong noise interference still have certain limitations on the decomposition effect.

[0005] Group sparse modal decomposition (GSMD) is a signal decomposition method based on sparse representation. It aims to extract sparse key characteristic components in the signal by introducing sparsity constraints. It is particularly suitable for processing vibration signals containing fault characteristics.

[0006] Since the gearbox vibration signal contains complex frequency components, over-decomposition is prone to occur when decomposing the gearbox vibration signal using GSMD, resulting in damage to the gear fault feature information. If envelope demodulation analysis is performed directly, it will seriously affect the extraction of gearbox fault features. Therefore, it is necessary to design a gear vibration signal decomposition method based on improved group sparse modal decomposition that can avoid over-decomposition and effectively extract the gear signal vibration characteristics. Summary of the invention

[0007] Aiming at the problem that GSMD may produce over-decomposition phenomenon and damage gear fault characteristic information in the process of processing gear box vibration signal, the present invention provides a gear vibration signal decomposition method based on improved group sparse modal decomposition.

[0008] The present invention proposes a gear vibration signal decomposition method based on improved group sparse modal decomposition, which comprises the following steps:

[0009] S1: measure the gearbox and obtain gear vibration data;

[0010] S2: Use discrete Fourier transform to obtain the vibration signal spectrum, denoted as c y ;

[0011] S3: spectrum c y Perform group sparse optimization to construct a set of zero-phase filter banks;

[0012] S4: Filter and decompose the signal spectrum into a series of non-overlapping and spaced frequency bands using a zero-phase filter bank, and perform inverse Fourier transform to obtain a series of group sparse modal components;

[0013] S5: Calculate the periodicity of all groups of sparse modal components;

[0014] S6: Fusion of group sparse modal components with high periodic similarity into intrinsic modal components;

[0015] S7: Select fault components, extract fault characteristic frequencies, and perform fault diagnosis.

[0016] Furthermore, in S3, a zero-phase filter bank is constructed using group sparse optimization as shown in Formula 1:

[0017]

[0018] Among them, c x is the spectral coefficient of the group sparse component, f is the filter group corresponding to the group sparse component, is the Vandermonde product, is a zero-phase filter bank.

[0019] Furthermore, S6 includes the following sub-steps:

[0020] S61: Calculate the periods of d initial effective single components;

[0021] S62: Select the first initial single component Y 1 ;

[0022] S63: The first initial single component Y 1By periodically comparing with the remaining initial valid single components and reorganizing the single components within the allowable error range, the first eigenmode component can be obtained;

[0023] Repeat the above steps to obtain multiple eigenmode components.

[0024] Furthermore, in S7, the relevant kurtosis is used to determine the fault component, and the relevant kurtosis is shown in Formula 2:

[0025]

[0026] Where M is the number of shifts, y n is the intrinsic mode component, T is the fault period, f s is the sampling frequency.

[0027] Furthermore, envelope demodulation is used in S7 to extract fault characteristic frequencies. Specifically, in S4, the signal spectrum is filtered and decomposed into multiple modes using a zero-phase filter group, and the fault mode is determined according to the relevant kurtosis index. Then, the corresponding modal components in the decomposition result are extracted and envelope transformation is performed.

[0028] Furthermore, the method for acquiring the gear vibration data in S1 is as follows: construct a gearbox fault simulation experimental platform, set the sampling frequency of the simulation experimental platform, set the gears in the gearbox to primary meshing, set the number of teeth of the driving wheel teeth, set the number of teeth of the driven wheel teeth, set the gear fault to a crack fault on the tooth surface of the driving wheel; set the rotation speed of the driving wheel and the theoretical fault characteristic frequency.

[0029] Furthermore, in S1, an acceleration sensor is used to measure the gearbox.

[0030] Compared with the prior art, the present invention can achieve the following beneficial effects: the gear vibration signal decomposition method based on improved group sparse modal decomposition proposed in the present invention first obtains the original data of the vibration signal, then performs discrete Fourier transform to obtain the spectrum of the vibration signal, and constructs a series of zero-phase filter groups in combination with the group sparse optimization strategy, and then uses the zero-phase filter group to filter and decompose the signal into a series of group sparse modal components, and calculates the periodic characteristics of the group sparse modal components. Finally, the group sparse modal components are fused and reconstructed according to the periodic similarity, which can effectively extract the vibration characteristics of the gear signal, protect the integrity of the gearbox fault feature information, effectively avoid over-decomposition, and enhance the fault diagnosis capability of the gearbox. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 A flow chart of a gearbox vibration signal decomposition method provided by an embodiment of the present invention;

[0032] Figure 2The original signal waveform diagram and corresponding envelope spectrum of the gearbox crack fault provided by the embodiment of the present invention;

[0033] Figure 3 A time domain diagram of the intrinsic mode component obtained by decomposing the gearbox crack fault signal using IGSMD provided in an example of the present invention;

[0034] Figure 4 The intrinsic mode component envelope spectrum obtained by decomposing the gearbox crack fault signal by IGSMD provided in the example of the present invention;

[0035] Figure 5 The time domain diagram of the intrinsic mode component obtained by decomposing the gearbox crack fault signal with the original GSMD;

[0036] Figure 6 The envelope spectrum of the intrinsic mode component obtained by decomposing the gearbox crack fault signal by the original GSMD. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the following Figure 1-6 It should be understood that the specific embodiments described herein are only used to explain the present invention and do not constitute a limitation of the present invention.

[0038] like Figure 1 As shown, a gearbox vibration signal decomposition method based on improved group sparse mode decomposition (IGSMD) includes the following steps:

[0039] S1: Measure the gearbox and obtain gear vibration data.

[0040] The method for acquiring gear vibration data is as follows: construct a gearbox fault simulation experimental platform, set the sampling frequency of the simulation experimental platform to 4096Hz, set the gears in the gearbox to primary meshing, set the number of teeth of the driving wheel to 12, set the number of teeth of the driven wheel to 24, set the gear fault to a crack fault on the tooth surface of the driving wheel, set the speed of the driving wheel to 1500rmp, and the theoretical fault characteristic frequency to 25Hz.

[0041] The gearbox vibration signal acquisition device is LMS SCM09, and the gearbox is measured using an acceleration sensor. The acceleration sensor model is PCB 356A25. The time domain waveform and envelope spectrum of the gear crack vibration signal are as follows: Figure 2 As shown, Figure 2 The left side is the time domain waveform of the gear crack vibration signal. Figure 2 On the right is the gear crack vibration signal envelope spectrum.

[0042] S2: Use discrete Fourier transform to obtain the vibration signal spectrum, denoted as cy .

[0043] S3: spectrum c y Perform group sparse optimization to construct a set of zero-phase filter groups. The specific method of using group sparse optimization to construct a zero-phase filter group is shown in Formula 1:

[0044]

[0045] Among them, c x is the spectral coefficient of the group sparse component, f is the filter group corresponding to the group sparse component, is the Vandermonde product, is a zero-phase filter bank.

[0046] S4: Filter the signal spectrum into a series of non-overlapping and spaced frequency bands using a zero-phase filter bank, and perform an inverse Fourier transform to obtain a series of group sparse modal components.

[0047] S5: Calculate the periodicity of all groups of sparse modal components.

[0048] S6: Fusing the group sparse modal components with higher periodic similarity into eigenmodal components. S6 includes the following sub-steps:

[0049] S61: Calculate the periods of d initial effective single components;

[0050] S62: Select the first initial single component Y 1 ;

[0051] S63: The first initial single component Y 1 By periodically comparing with the remaining initial valid single components and reorganizing the single components within the allowable error range, the first eigenmode component can be obtained.

[0052] Repeat the above steps to obtain multiple eigenmode components.

[0053] S7: Select fault components, extract fault characteristic frequencies, and perform fault diagnosis. In practical applications, fault components can be selected based on relevant kurtosis, and envelope demodulation can be used to extract fault characteristic frequencies for fault diagnosis.

[0054] Specifically, in S4, the signal spectrum is filtered and decomposed into multiple modes using a zero-phase filter bank, and the fault mode is determined according to the relevant kurtosis index, and then the corresponding modal components in the decomposition result are extracted and envelope transformation is performed.

[0055] The relevant kurtosis is used to determine the fault component. The relevant kurtosis is shown in Formula 2:

[0056]

[0057] Where M is the number of shifts, y n is the intrinsic mode component, T is the fault period, f s is the sampling frequency.

[0058] Figure 3 The component time domain waveform diagram obtained by decomposing the gear crack signal using IGSMD. Through observation, it can be found that IGSMD decomposes the gear crack signal into 5 intrinsic mode components. According to the relevant kurtosis index, the fourth component can be judged as the fault component. The decomposed result is envelope transformed. Figure 4 is the corresponding envelope spectrum.

[0059] The original GSMD is used to decompose the gear crack vibration signal. GSMD decomposes the gear crack vibration signal into 50 modes. According to the relevant kurtosis index, the 38th mode can be judged as the fault mode. For the purpose of comparison, the 36th to 40th modal components in the decomposition results are extracted and envelope transformed. Figure 5 This is the time domain waveform of the 36th to 40th modes. Figure 6 is the corresponding envelope spectrum.

[0060] It can be seen from the figure that due to the complex frequency characteristics of the gearbox vibration signal, GSMD has a serious over-decomposition phenomenon, and the gear fault feature information is destroyed. The time domain waveform of the fault mode obtained by IGSMD has obvious modulation characteristics, and the corresponding envelope spectrum has obvious amplitude at the fault feature frequency and its multiple positions, and there is no noise interference around it. In addition, IGSMD fuses the sparse components according to the periodic characteristics and finally obtains 5 components, which avoids the over-decomposition phenomenon of GSMD and meets the decomposition purpose of IGSMD.

[0061] Although the embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and cannot be understood as limiting the present invention. Those skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A gear vibration signal decomposition method based on improved group sparse modal decomposition, characterized in that: The method comprises the following steps: S1: measure the gearbox and obtain gear vibration data; S2: Use discrete Fourier transform to obtain the vibration signal spectrum, denoted as c y ; S3: spectrum c y Perform group sparse optimization to construct a set of zero-phase filter banks; S4: Filter and decompose the signal spectrum into a series of non-overlapping and spaced frequency bands using a zero-phase filter bank, and perform inverse Fourier transform to obtain a series of group sparse modal components; S5: Calculate the periodicity of all groups of sparse modal components; S6: Fusion of group sparse modal components with high periodic similarity into intrinsic modal components; S7: Select fault components, extract fault characteristic frequencies, and perform fault diagnosis.

2. The gear vibration signal decomposition method based on improved group sparse modal decomposition according to claim 1, characterized in that: In S3, group sparse optimization is used to construct a zero-phase filter bank as shown in Formula 1: Among them, c x is the spectral coefficient of the group sparse component, f is the filter group corresponding to the group sparse component, is the Vandermonde product, is a zero-phase filter bank.

3. The gear vibration signal decomposition method based on improved group sparse modal decomposition according to claim 1, characterized in that: S6 includes the following sub-steps: S61: Calculate the periods of d initial effective single components; S62: Select the first initial single component Y1; S63: periodically comparing the first initial single component Y1 with the remaining initial valid single components, and reorganizing the single components within the allowable error range, so as to obtain the first eigenmode component; Repeat the above steps to obtain multiple eigenmode components.

4. The gear vibration signal decomposition method based on improved group sparse modal decomposition according to claim 1, characterized in that: S7 uses the relevant kurtosis to determine the fault component. The relevant kurtosis is shown in Formula 2: Where M is the number of shifts, y n is the intrinsic mode component, T is the fault period, f s is the sampling frequency.

5. According to the gear vibration signal decomposition method based on improved group sparse modal decomposition described in claim 4, envelope demodulation is used to extract the fault characteristic frequency in S7. Specifically, in S4, the signal spectrum is filtered and decomposed into multiple modes using a zero-phase filter group, and the fault mode is determined according to the relevant kurtosis index. Then, the corresponding modal components in the decomposition result are extracted and envelope transformation is performed.

6. The gear vibration signal decomposition method based on improved group sparse modal decomposition according to claim 1, characterized in that: The method for acquiring gear vibration data in S1 is as follows: construct a gearbox fault simulation experimental platform, set the sampling frequency of the simulation experimental platform, set the gears in the gearbox to primary meshing, set the number of teeth of the driving wheel, set the number of teeth of the driven wheel, set the gear fault to a crack fault on the tooth surface of the driving wheel; set the speed of the driving wheel and the theoretical fault characteristic frequency.

7. The gear vibration signal decomposition method based on improved group sparse modal decomposition according to claim 1, characterized in that: In S1, the gearbox is measured using an acceleration sensor.

Images