A method for extracting multi-channel fault features of rolling bearings

Abstract

The invention discloses a rolling bearing multi-channel fault feature extraction method. First, the FVMMD method is used to decompose the multi-channel signal to obtain multiple intrinsic mode components (IMFs); secondly, the effective weight kurtosis index (EWK) is established to screen the sensitive components for signal reconstruction; finally, the SSA-MOMEDA algorithm is used to reduce the noise of the reconstructed signal, and the envelope demodulation is used to complete the extraction of the rolling bearing fault feature. The invention can effectively extract the characteristic frequency of the rolling bearing fault, which is of great significance to ensure the normal operation of the equipment.

Description

Technical Field

[0001] The invention belongs to the technical field of mechanical fault diagnosis, and in particular relates to a rolling bearing multi-channel fault feature extraction method based on FVMMD and SSA-MOMEDA. Background Art

[0002] Rolling bearings are one of the most important and vulnerable key components in rotating machinery and equipment. They often operate in complex working conditions with high speed and heavy loads. Their health will directly affect the working state of the entire mechanical equipment, causing costly downtime at the least and casualties at the worst. Therefore, it is crucial to ensure the normal operation of rotating equipment and effectively extract the fault characteristics of rolling bearings to evaluate the operating state of bearings.

[0003] At present, most of the feature extraction methods are studied for single-channel signals. However, multi-channel signals contain more fault information and can more comprehensively characterize the operating status of rolling bearings. Therefore, the MVMD method is used to decompose the multi-channel signals of rolling bearings, but the decomposition mode number K and the penalty factor а need to be preset before using MVMD. If the mode number K is too small, the decomposition result of the multi-channel signal will be incomplete and the fault information will be omitted. On the contrary, if the mode number K is too large, it will lead to over-decomposition; and the penalty factor а directly affects the bandwidth of the modal component. Therefore, the FVMMD method is proposed to adaptively select the parameters of MVMD.

[0004] In addition, an effective weighted kurtosis index is proposed to screen and reconstruct sensitive IMF components, and the fault features of the reconstructed signal are enhanced through the SSA-MOMEDA algorithm to complete the rolling bearing fault feature extraction. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a rolling bearing multi-channel fault feature extraction method based on FVMMD and SSA-MOMEDA. Based on the multi-channel rolling bearing signal, FVMMD is used to decompose the multi-channel signal, and the effective components are selected for reconstruction. The reconstructed signal is processed by MOMEDA, and envelope demodulation is used to extract the fault characteristic frequency.

[0006] A rolling bearing multi-channel fault feature extraction method based on FVMMD and SSA-MOMEDA, comprising the following steps:

[0007] S1: Perform FVMMD decomposition on the input multi-channel signal to obtain a series of multi-channel modal components;

[0008] S2: Calculate the EWK of each channel mode, and select the mode with an EWK value greater than 0 as the valid mode for signal reconstruction;

[0009] S3: Use SSA-MOMEDA to reduce noise and enhance periodic impulse components of the reconstructed signal;

[0010] S4: Obtain the characteristic frequency of rolling bearing fault through envelope demodulation;

[0011] Preferably, the method for performing FVMMD decomposition on the signal in S1 is as follows:

[0012] S1.1: Collect multi-channel rolling bearing signals and calculate the EK value of each channel. The EK value is calculated using the following formula:

[0013] EK=KI·KES

[0014]

[0015] Where SE(P) is the envelope signal of the mathematical x(n);

[0016] S1.2: Select the channel with the largest EK value as the main channel, perform spectrum trend segmentation on the main channel signal, and calculate the decomposition mode number K, penalty factor а, and mode center frequency ω;

[0017] S1.3: Bring the modal number K, penalty factor а, and modal center frequency ω into MVMD to complete the decomposition of multi-channel signals;

[0018] Preferably, the spectrum trend segmentation is performed by obtaining the spectrum of the signal through FFT calculation, and the spectrum of the signal is subjected to FFT again to obtain the kernel function of the signal spectrum. The first 30 points of the kernel function are taken and the spectrum trend is obtained by iFFT calculation;

[0019] Preferably, the optimization of the spectrum trend adopts a hard threshold denoising method, the spectrum trend part disturbed by noise is a straight line, and then the spectrum trend minimum point and the endpoint of the straight line trend are used as new segmentation boundaries, and the remaining trend maximum points are used as effective frequency bands, that is, the number of effective frequency bands is equal to the modal number K; through the optimized spectrum trend segmentation diagram, the spectrum trend maximum point is found, and the horizontal axis frequency corresponding to the maximum point is used as the initial center frequency of the mode, that is, ω = [ω1ω2…ωK];

[0020] Preferably, each ω corresponds to an а, that is, а=[а1а2…а K ], а is the penalty factor, а is calculated by the following formula:

[0021]

[0022] In the formula, f s is the data sampling frequency;

[0023] Preferably, the EWK in S2 is calculated by the following formula:

[0024]

[0025] In the formula, KI(k) and C(k) represent the kurtosis and correlation coefficient of data x(i);

[0026] Preferably, the SSA-MOMEDA in S3 optimizes the MOMEDA parameters T and L by using the SSA optimization algorithm, and the specific method is as follows:

[0027] S3.1: Set the number of sparrows in the population and the number of iterations in SSA to 200, and the search range of parameter T is [0.75T i～ 1.25T o ], the search range of parameter L[2f s / f i～ 1000], T i 、T o They are respectively the failure cycles of inner and outer ring rolling bearings;

[0028] S3.2: Decompose the signal using MOMEDA;

[0029] S3.3: Calculate the EK value of the filtered signal, and if the optimization criteria are met, update the position and size of the maximum fitness;

[0030] S3.4: Determine whether the stopping condition is met. If so, output the optimal parameters of MOMEDA. If not, repeat the above steps.

[0031] The beneficial effects of the present invention are:

[0032] Compared with single-channel fault signals, multi-channel fault signals contain more fault information. A rolling bearing multi-channel fault feature extraction method based on FMVMD and SSA-MOMEDA can simultaneously extract the fault feature frequency of multi-channel signals and judge the operating status of rolling bearings. At the same time, the MVMD method parameters K, а; MOMEDA method parameters T, L are optimized during use. This provides a new idea for comprehensively observing the multi-channel signals of rolling bearings to judge the operating status of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0034] Figure 1 is a flow chart of the steps of the present invention;

[0035] Figure 2It is the bearing data center test bench of Case Western Reserve University in the United States in the embodiment of the present invention;

[0036] Figure 3 is a time domain and frequency domain signal diagram in an embodiment of the present invention;

[0037] Figure 4 is a spectrum trend segmentation diagram in an embodiment of the present invention;

[0038] Figure 5 It is an IMF diagram decomposed by FVMMD in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] In order to make a clear and complete description of the technical scheme and effects of the present invention, the following embodiments are described in detail;

[0040] The embodiment of the present invention provides a rolling bearing multi-channel fault feature extraction method based on FVMMD and SSA-MOMEDA. The basic flow chart is as follows: Figure 1 The embodiment uses the data collected by the CWRU rolling bearing fault simulation experimental platform as the experimental object, as shown in FIG. Figure 2 The specific parameters are as follows: bearing model is SKF6205, speed is 1750r / min, sampling frequency is 12kHz.

[0041] Example 1

[0042] The embodiment uses bearing outer ring fault data collected synchronously by acceleration sensors at the drive end and the fan end, and the rolling bearing outer ring theoretical fault frequency BPFO=107.305Hz;

[0043] S1: Use FVMD to decompose the multi-channel signal and obtain several IMF components;

[0044] S1.1: Collect the outer ring fault vibration data measured at the driving end and fan end of the rolling bearing. Its time domain and frequency domain are as follows Figure 3 As shown;

[0045] S1.2: Calculate the EK index of the drive end and the fan end as shown in Table 1:

[0046] Table 1

[0047] Data collection location Outer ring fault signal Drive end 39.4 Fan end 9.6

[0048] S1.3: Use spectrum trend segmentation to process the driver data, such as Figure 4 As shown, it can be concluded that the mode number K = 3, the initial center frequency W F =[37528133375], penalty factor α F =[1616489372];

[0049] S1.4: Substitute the parameters and perform FVMMD decomposition to obtain the IMF components, such as Figure 5 As shown;

[0050] S2: Calculate the EWK of each channel mode. The results are shown in Table 2. Select the mode with an EWK value greater than 0 as the valid mode for signal reconstruction;

[0051] Table 2

[0052] IMF Components IMF1 IMF2 IMF3 Drive end -2.42 0.13 2.29 Fan end -1.15 -0.14 1.3

[0053] S3: Use SSA-MOMEDA to reduce noise and enhance periodic impulse components of the reconstructed signal;

[0054] S3.1: Since the outer ring theoretical fault period TBPFO = 111.7 and the inner ring theoretical fault period TBPFI = 73.9, according to the above process, the T search range is set to [55.5-134] and the L search range is set to [148-1000];

[0055] S3.2: Decompose the signal using MOMEDA;

[0056] S3.3: Calculate the EK value of the filtered signal, and if the optimization criteria are met, update the position and size of the maximum fitness;

[0057] S3.4: Determine whether the stop condition is met. If so, output the optimal parameters of MOMEDA. If not, repeat S3.2 to S3.3.

[0058] S3.5: Output the optimal parameter combination [T, L] = [113764], and complete the MOMEDA processing;

[0059] S4: Obtain the rolling bearing fault characteristic frequency through envelope demodulation.

[0060] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0061] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A rolling bearing multi-channel fault feature extraction method based on FVMMD and SSA-MOMEDA, characterized in that: The following steps are involved: S1: Perform FVMMD decomposition on the input multi-channel signal to obtain a series of multi-channel modal components; S2: Calculate the EWK of each channel mode, and select the mode with an EWK value greater than 0 as the valid mode for signal reconstruction; S3: Use SSA-MOMEDA to reduce noise and enhance periodic impulse components of the reconstructed signal; S4: Obtain the characteristic frequency of rolling bearing fault through envelope demodulation; The method for S1 to perform FVMMD decomposition on the signal is as follows: S1.1: Collect multi-channel rolling bearing signals and calculate the EK value of each channel. The EK value is calculated using the following formula: EK=KI·KES Where SE(P) is the envelope signal of the mathematical x(n); S1.2: Select the channel with the largest EK value as the main channel, perform spectrum trend segmentation on the main channel signal, and calculate the decomposition mode number K, penalty factor а, and mode center frequency ω; S1.3: Bring the modal number K, penalty factor а, and modal center frequency ω into MVMD to complete the decomposition of multi-channel signals; The spectrum trend segmentation is performed by using FFT to obtain the spectrum of the signal, and the spectrum of the signal is subjected to FFT again to obtain the kernel function of the spectrum of the signal; the first 30 points of the kernel function are taken and the spectrum trend is obtained by using iFFT calculation; The optimization of the spectrum trend adopts a hard threshold denoising method, the spectrum trend part disturbed by noise is a straight line, and then the spectrum trend minimum point and the end point of the straight line trend are used as new segmentation boundaries, and the remaining trend maximum points are used as effective frequency bands, that is, the number of effective frequency bands is equal to the modal number K; through the optimized spectrum trend segmentation diagram, the spectrum trend maximum point is found, and the horizontal axis frequency corresponding to the maximum point is used as the initial center frequency of the mode, that is, ω = [ω1ω2…ωK]; Each ω corresponds to an а, that is, а=[а1а2…аK], а is the penalty factor, and а is calculated by the following formula Where fs is the data sampling frequency.

2. According to claim 1, a rolling bearing multi-channel fault feature extraction method based on FVMMD and SSA-MOMEDA is characterized in that: The S2 EWK Calculated by the following formula: Where KI(k) and C(k) represent the kurtosis and correlation coefficient of the data x(i).

3. According to claim 1, a rolling bearing multi-channel fault feature extraction method based on FVMMD and SSA-MOMEDA is characterized in that: The SSA-MOMEDA in S3 optimizes the MOMEDA parameters T and L through the SSA optimization algorithm. The specific method is as follows: S3.1: Set the number of sparrows in the population and the number of iterations in SSA to 200, and the search range of parameter T is [0.75T i ~1.25T o ], the search range of parameter L[2f s / f i ~1000], T i 、T o They are respectively the failure cycles of inner and outer ring rolling bearings; S3.2: Decompose the signal using MOMEDA; S3.3: Calculate the EK value of the filtered signal, and if the optimization criteria are met, update the position and size of the maximum fitness; S3.4: Determine whether the stopping condition is met. If so, output the optimal parameters of MOMEDA. If not, repeat the above steps.

Images