In order to make the technical means, creative features, objectives and effects of the present invention easy to understand, the technical solutions in the specific embodiments of the present invention are described clearly and completely below to further illustrate the present invention. Obviously, the described specific implementations The modes are only a part of the embodiments of the present invention, rather than all modes.
 In order to verify the effectiveness and superiority of the aero-engine rotor fault feature extraction method based on VNCMD and Birge-Massart threshold noise reduction in the actual rotor system fault diagnosis, this paper is based on the aero-engine rotor in the Intelligent Diagnosis and Expert System Laboratory of Nanjing University of Aeronautics and Astronautics Fault data collection is performed on the test bench. Taking the engine rotor rubbing fault as an example, the aero-engine rotor rubbing fault experiment is simulated on the above test bench. In the experiment, the signal is collected by the acceleration sensor, and the rotor speed is set to 2700r/min, sampling The frequency is 10240 Hz.
 Time-frequency analysis of the rubbing vibration signal collected in the experiment, we get figure 1 The time-domain waveform diagram and envelope demodulation spectrum diagram of the original fault vibration signal are shown by figure 1 It can be seen from the envelope spectrum of the original fault vibration signal that the speed frequency of the original fault vibration signal is 45 Hz, and there is only an impulse response at the speed frequency 1×, 3× frequency multiplication, and the other fault characteristic frequency multiplication components are submerged in strong noise. It shows that the shock pulse that can reflect the characteristics of the fault is almost covered by noise. Therefore, it is difficult to completely extract the fault characteristic information from the envelope demodulation spectrum of the original fault vibration signal. For this reason, a threshold reduction based on VNCMD and Birge-Massart is proposed. Noisy rotor system fault feature extraction method.
 Analysis of decomposition effect
 Firstly, three methods of EMD, VMD, and VNCMD are used to decompose the fault vibration signal. The time domain diagram of the signal components after decomposition is as follows: figure 2 , image 3 , Figure 4 As shown, the EMD decomposes a total of 12 IMF components. For the EMD method, the fault information is mainly contained in the first few components, so the first 5 components are selected for analysis.
 Calculate the kurtosis values and correlation coefficients of each component decomposed by the three methods and the original signal. The calculation results are shown in Table 1, Table 2, and Table 3. Among them, Table 1 is the calculation of the kurtosis value and correlation coefficient of each component of EMD decomposition As a result, Table 2 shows the calculation results of the kurtosis values and correlation coefficients of each component of the VMD decomposition, and Table 3 shows the calculation results of the kurtosis values and correlation coefficients of each component of the VNCMD decomposition. According to the kurtosis value and correlation coefficient criterion, it can be known that when the kurtosis value and correlation coefficient of the signal component are large, it contains the most fault characteristic information of the original fault signal. Therefore, through the analysis of Table 1, Table 2, Table 3 It can be obtained that the components corresponding to the maximum fault characteristic information of the original fault signal corresponding to the three methods are IMF1, IMF4, IMF4, and their corresponding kurtosis values and correlation coefficients are relatively large.
 Table 1. Calculation results of kurtosis values and correlation coefficients of each component of EMD decomposition
 Table 2. Calculation results of kurtosis value and correlation coefficient of each component of VMD decomposition
 Table 3. Calculation results of kurtosis value and correlation coefficient of each component of VNCMD decomposition
 In order to verify the effectiveness of the method VNCMD of the present invention, firstly, the signal components IMF1, IMF4, and IMF4 corresponding to the three selected methods are directly subjected to envelope analysis to extract the rotor fault frequency, and the results are as follows: Figure 5 , Image 6 , Figure 7 Shown.
 by Figure 5 with Image 6 It can be seen that the signal components decomposed by the EMD and VMD methods can extract the 1×, 2×, and 3× frequency spectrum components of the rotor fault characteristic frequency, indicating that both methods can extract the characteristic information of the rotor fault, but they can only extract Up to 3× octave spectrum components, the effect is not ideal. Figure 7 shows that the VNCMD method can extract the 1×, 2×, 3×, 4× frequency spectrum components of the rotor fundamental frequency, and the noise interference frequency components are relatively small. Therefore, the VNCMD method has a better decomposition effect and has more advantages in fault feature extraction.
 Noise reduction effect analysis
 In order to verify the effectiveness of the combination of VNCMD and Birge-Massart threshold noise reduction methods, the effective components decomposed by the above three methods are combined with the Birge-Massart threshold noise reduction method for noise reduction, and the signal components and fault feature extraction results after noise reduction Respectively Figure 8 , Picture 9 , Picture 10 Shown.
 by Figure 8 , Picture 9 , Picture 10 It can be seen that the EMD-Birge-Massart method extracts the 1×, 2×, 3×, 4× frequency octave spectrum components of the rotor fault characteristic frequency, but because the overall noise interferes with more frequency components, other characteristic frequency components are submerged in the noise among. Compared with the EMD-Birge-Massart method, the VMD-Birge-Massart threshold noise reduction method has improved the suppression of noise interference frequency components, but also only extracts 1×, 2×, 3×, 4 of the rotor fault characteristic frequency × frequency spectrum components, and the method proposed in this paper not only improves the suppression of noise interference frequency components, but also can extract 1×, 2×, 3×, 4×, 5× and 6× times the characteristic frequency of the rotor fault Frequency spectrum components. At the same time, comparing the extraction results of the highest 4× frequency multiplier that can be extracted by the direct VNCMD method, it can be seen that the method in this paper can extract more fault frequencies and extract fault feature information more completely, which verifies that the method proposed in this paper is in terms of fault feature information extraction. Has a better effect.
 in conclusion
 (1) The VNCMD algorithm can be used to effectively decompose the rotor fault vibration signal. The VNCMD decomposition result of the vibration signal based on the kurtosis value and the correlation coefficient criterion can accurately filter the signal component containing the largest fault information and eliminate the influence of irrelevant IMF.
 (2) Birge-Massart noise reduction is performed on the rotor fault vibration signal decomposed by the VNCMD algorithm, which can further extract the characteristic frequency of the rotor fault under the background of strong noise, and realize a more effective diagnosis of the rotor fault.
 (3) The rotor fault feature extraction results based on EMD, VMD, VNCMD combined with the Birge-Massart noise reduction method are compared, and the superiority of the rotor fault feature extraction based on the VNCMD and Birge-Massart noise reduction method is verified.
 The main technical features, basic principles and related advantages of the present invention are described above. For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and does not deviate from the concept or basic features of the present invention. In this case, the present invention can be implemented in other specific forms. Therefore, no matter from which point of view, the above-mentioned specific embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description, so it is intended to fall on All changes within the meaning and scope of equivalent elements of the claims are encompassed in the present invention.
 In addition, it should be understood that although this specification is described in accordance with each embodiment, not each embodiment only contains an independent technical solution. This narration in the specification is only for clarity, and those skilled in the art should regard the specification as a Overall, the technical solutions in the various embodiments can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.