Signal Modal Decomposition Method and System for Early Fault Diagnosis of Rotating Machinery

Abstract

The present invention belongs to the technical field of signal processing. A signal modal decomposition method and system for early fault diagnosis of rotating machinery are provided. The vibration signal is decomposed by the MCKD algorithm to obtain the characteristic modal set of the signal, and then fusion is performed according to the cross-correlation spectrum of the modes and the correlation coefficient of the modal envelope power spectral density. Finally, the characteristic modes representing faults are extracted, and the characteristic modes are extracted using the fault period, and the characteristic modes are fused in terms of time-domain and frequency-domain characteristics respectively. It has low sensitivity to complex noise, strong ability to extract weak fault feature signals, high precision, and realizes more efficient and accurate extraction of weak fault feature signals.

Description

Technical Field

[0001] The present invention relates to the technical field of signal processing, and particularly to a signal modal decomposition method and system for early fault diagnosis of rotating machinery. Background Art

[0002] The statements in this part merely provide background art related to the present invention and do not necessarily constitute prior art.

[0003] Rotating machinery takes rotational motion as its core working mode and is widely used in industrial fields such as machining and manufacturing, energy, and metallurgy. Its key components (gearboxes, bearings, etc.) are prone to wear, fracture, and other faults due to long-term high-speed rotation and high-load working conditions, resulting in economic losses and affecting safe production. Early fault diagnosis technology can detect faults in a timely manner by analyzing the vibration signals of rotating machinery and avoid production accidents. Since the early fault vibration signals of rotating machinery are relatively weak and there is interference from noise, it is difficult to extract early fault signals. Therefore, separating weak early fault signals from vibration signals containing a large amount of noise is the key to early fault diagnosis technology.

[0004] In early fault diagnosis technology, currently widely used methods include empirical mode decomposition (EMD), variational mode decomposition (VMD), and feature mode decomposition (FMD), etc. EMD is prone to problems such as mode mixing, endpoint effects, and poor anti-noise performance. The root cause lies in relying on extreme point interpolation to generate the envelope line, lacking a frequency band separation mechanism, and having no explicit denoising mechanism. VMD has problems such as parameter sensitivity, high computational complexity, and the risk of frequency band overlap. Since its optimization model depends on preset parameters and uses iterative solutions, it is difficult to select parameters, has low computational efficiency, and is difficult to completely separate signal components with close frequency bands. FMD faces problems such as dependence on characteristic indicators, serious noise interference, and an imperfect theoretical framework. The decomposition result highly depends on the selection of characteristic indicators, is prone to misjudgment in a strong noise environment, and lacks a unified mathematical convergence proof.

[0005] For example, patent number CN202410252421.7 discloses a fault diagnosis method for wind turbine gearboxes based on complementary ensemble empirical mode decomposition - ResNet - BiLSTM. By collecting gearbox vibration signals, multi - band empirical mode components are extracted using complementary ensemble empirical mode decomposition (CEEMDAN), strong - correlation features are screened using the Pearson correlation coefficient to reduce the dimension, and a ResNet - BiLSTM hybrid model is constructed for training and classification, ultimately achieving high - precision fault diagnosis. However, this method still has the following limitations: 1) The CEEMDAN decomposition depends on the quality of vibration signals. If there are high - frequency noises or non - stationary interferences in sensor data, it may lead to mode aliasing or feature distortion; 2) The ResNet - BiLSTM model has a large number of parameters and complex calculations, making it difficult to be deployed on resource - constrained wind power edge devices; 3) Training depends on a balanced sample set, and the scarcity of fault samples in the actual scenario may lead to a decline in the model's generalization ability; 4) The influence of multiple working conditions of the gearbox (such as variable speed and variable load) on vibration signals is not considered, which may lead to a decline in cross - working - condition diagnosis performance.

[0006] Patent number CN202210756586.9 discloses a rolling bearing early fault diagnosis method based on adaptive variational mode decomposition. By collecting bearing vibration signals, the sparrow search algorithm (SSA) is used to adaptively search for the optimal parameters of variational mode decomposition (VMD) with the goal of minimizing the comprehensive fault index to achieve adaptive decomposition of the signal. Subsequently, the power spectral kurtosis of each mode component is calculated, and the components below the average value are removed to reconstruct the signal. Finally, the fault type is diagnosed through envelope demodulation analysis. However, this method has the following limitations: 1) The global search ability of the sparrow search algorithm is limited by the parameter range and the number of iterations, which may lead to local optimal solutions and affect the accuracy of VMD decomposition; 2) The screening of the power spectral kurtosis threshold depends on the statistical characteristics of the mode components, and effective components may be mis - deleted under strong non - stationary noises or complex working conditions; 3) VMD is sensitive to the non - stationarity of signals and multi - component cross - interferences, and is prone to mode aliasing or under - decomposition problems; 4) Envelope demodulation analysis depends on prior knowledge of accurate fault characteristic frequencies. In practical applications, if the bearing speed fluctuates or the load changes, it may lead to the failure of demodulation.

[0007] Patent No. CN202410382441.6 discloses a rolling bearing fault diagnosis method based on parameter adaptive feature mode decomposition. By collecting vibration signals, calculating the theoretical fault characteristic frequencies, using the Sparrow Search Algorithm (SSA) to optimize the parameters of the Feature Mode Decomposition (FMD), screening the optimal parameter combination with the Envelope Frequency Domain Signal-to-Noise Ratio (EFDSNR) as the index, then decomposing the signal by FMD, and finally extracting the fault characteristic frequencies through Hilbert envelope demodulation to achieve diagnosis. However, this method has the following limitations: 1) The search range of the fault characteristic frequencies is limited to ±3% of the theoretical value. In actual working conditions, if the true frequency deviation exceeds this range due to bearing wear, installation errors, or speed fluctuations, the parameter optimization may fail; 2) The calculation of the theoretical fault frequencies depends on accurate bearing geometric parameters. In practical applications, if the parameters are unknown or the measurement errors are large, it will directly affect the accuracy of fault feature extraction; 3) Hilbert envelope demodulation depends on the significance and singularity of the fault frequencies. In the case of coexisting multiple faults or strong modulation interference, spectral aliasing may occur, leading to misjudgment; 4) The EFDSNR index may not be able to effectively distinguish the noise and fault components under extreme non-stationary noise or complex harmonic interference, resulting in the deletion of effective modes. Summary of the Invention

[0008] To solve the deficiencies of the prior art, the present invention provides a signal mode decomposition method and system for early fault diagnosis of rotating machinery. By decomposing the vibration signal through the MCKD algorithm, obtaining the characteristic mode set of the signal, and then fusing according to the cross-correlation spectrum of the modes and the correlation coefficient of the mode envelope power spectral density, the characteristic mode representing the fault is finally extracted, realizing more efficient and accurate extraction of weak fault feature signals.

[0009] To achieve the above object, the present invention adopts the following technical solutions:

[0010] In the first aspect, the present invention provides a signal mode decomposition method for early fault diagnosis of rotating machinery.

[0011] A signal mode decomposition method for early fault diagnosis of rotating machinery includes the following processes:

[0012] Convert the original vibration signal into a column vector and copy it multiple times to form multiple sub-signals. Divide the entire frequency band into multiple sub-frequency bands, with each sub-channel corresponding to a sub-signal, and calculate the frequency band boundaries of each sub-frequency band. For each sub-frequency band, generate the corresponding filter coefficients;

[0013] Perform iterative decomposition operations on each sub-signal in the form of a column vector to obtain the corresponding modes and filter coefficients of each sub-signal;

[0014] Calculate the correlation coefficient of each mode with the original vibration signal, sort them in descending order according to the magnitude of the correlation coefficient, start from the first mode, calculate its cross-correlation spectrum with the remaining modes, and perform the first fusion of the modes and the corresponding filter coefficients according to the peak value of the normalized cross-correlation spectrum;

[0015] Calculate the power spectral density of the normalized envelope after the first fusion for each mode, sort them in descending order according to the magnitude of the power spectral density, start from the power spectral density of the normalized envelope of the first mode, calculate the correlation coefficient between its power spectral density of the normalized envelope and the power spectral density of the normalized envelope of the remaining modes after the first fusion, and perform the second fusion of the modes and the corresponding filter coefficients according to the correlation coefficient;

[0016] If the number of iterations does not reach the maximum number of iterations, continue to perform the operation of the first fusion on the mode set after the second fusion; if the number of iterations reaches the maximum number of iterations, then use the obtained modes as the final characteristic mode set;

[0017] Obtain the periodic impact fault signal according to the final characteristic mode set.

[0018] As a further limitation of the first aspect of the present invention, find the peak value of the cross-correlation spectrum, judge whether to fuse two modes according to the characteristics of the peak value. If the fusion condition is met, then fuse the two modes, and at the same time fuse the corresponding filter coefficients, and calculate the cross-correlation spectrum of each mode in turn to complete the first fusion.

[0019] As a further limitation of the first aspect of the present invention, the fusion conditions include: the number of peak values is greater than or equal to the first set threshold, the maximum peak value is greater than the second set threshold, and the peak interval has consistency.

[0020] As a further limitation of the first aspect of the present invention, when the correlation coefficient is greater than the third set threshold, fuse the two modes, and at the same time fuse the corresponding filter coefficients to complete the second fusion.

[0021] As a further limitation of the first aspect of the present invention, in the first fusion and the second fusion, the process of fusing two modes and simultaneously fusing the corresponding filter coefficients is the same, including:

[0022] Directly add the two modes as a new mode, and at the same time add the filter coefficients corresponding to the two modes to obtain the filter coefficients corresponding to the new mode.

[0023] As a further limitation of the first aspect of the present invention, for each sub-band, generate the corresponding filter coefficients, including:

[0024] ;

[0025] Wherein, is the Filter coefficients corresponding to each sub - band; is a Hanning window; is the filter length; is the number of sub - bands; is a constant; represents the frequency - band boundary corresponding to the sub - band, and the

[0026] function is used to generate filter coefficients. As a further limitation of the first aspect of the present invention, the entire frequency band is evenly divided into sub - bands, and the set of frequency - band boundaries is:

[0027] As a further limitation of the first aspect of the present invention, for any sub - signal perform iterative decomposition operations, including:

[0028] If the fault period of the sub - signal is not provided, estimate the fault period through Hilbert transform and autocorrelation function;

[0029] Construct a multi - channel delay matrix;

[0030] For the first iteration round, use the multi - channel delay matrix to calculate the vector , and construct the delay matrix according to the calculated vector . Calculate the weights according to the delay matrix , update the filter coefficients according to the weights and normalize them, and output the corresponding eigen - mode and corresponding filter coefficients of this iteration round;

[0031] Perform multiple rounds of iteration until the eigen - mode and corresponding filter coefficients corresponding to the maximum iteration round are output.

[0032] As a further limitation of the first aspect of the present invention, estimate the fault period through Hilbert transform and autocorrelation function , including: , where is the position of the first zero - crossing point of the autocorrelation spectrum, and is the autocorrelation spectrum of the sub - signal .

[0033] In a second aspect, the present invention provides a signal modal decomposition system for early fault diagnosis of rotating machinery.

[0034] A signal modal decomposition system for early fault diagnosis of rotating machinery, comprising:

[0035] The filter bank generation unit is configured to: convert the original vibration signal into a column vector, copy it multiple times to form multiple sub-signals, evenly divide the entire frequency band into multiple sub-frequency bands, each sub-channel corresponding to a sub-signal, calculate the frequency band boundaries of each sub-frequency band, and generate corresponding filter coefficients for each sub-frequency band;

[0036] The iterative decomposition operation unit is configured to: perform iterative decomposition operations on each sub-signal in the form of a column vector to obtain the corresponding modes and corresponding filter coefficients of each sub-signal;

[0037] The first fusion unit is configured to: calculate the correlation coefficient between each mode and the original vibration signal, sort them in descending order according to the magnitude of the correlation coefficient, starting from the first mode, calculate its cross-correlation spectrum with the remaining modes, and perform the first fusion of the modes and the corresponding filter coefficients according to the peaks of the normalized cross-correlation spectrum;

[0038] The second fusion unit is configured to: calculate the power spectral density of the normalized envelope after the first fusion of each mode, sort them in descending order according to the magnitude of the power spectral density, starting from the power spectral density of the normalized envelope of the first mode, calculate the correlation coefficient between its power spectral density of the normalized envelope and the power spectral density of the normalized envelope of the remaining modes after the first fusion, and perform the second fusion of the modes and the corresponding filter coefficients according to the correlation coefficient;

[0039] The characteristic mode set generation unit is configured to: if the number of iterations has not reached the maximum number of iterations, continue to perform the operation of the first fusion on the mode set after the second fusion; if the number of iterations reaches the maximum number of iterations, then use the obtained modes as the final characteristic mode set;

[0040] The fault signal extraction unit is configured to: obtain the periodic impact fault signal according to the final characteristic mode set.

[0041] Compared with the prior art, the beneficial effects of the present invention are:

[0042] 1. The present invention innovatively proposes a signal modal decomposition strategy for early fault diagnosis of rotating machinery. By decomposing the vibration signal through the MCKD algorithm, a characteristic mode set of the signal is obtained, and then fusion is performed according to the cross-correlation spectrum of the modes and the correlation coefficient of the modal envelope power spectral density. Finally, the characteristic modes representing faults are extracted, realizing more efficient and accurate extraction of weak fault characteristic signals.

[0043] 2. The present invention innovatively proposes a signal modal decomposition strategy for early fault diagnosis of rotating machinery. It uses the fault period to extract characteristic modes, fuses the characteristic modes in terms of time-domain and frequency-domain characteristics respectively, has low sensitivity to complex noise, strong ability to extract weak fault characteristic signals, and high accuracy.

[0044] 3. The present invention innovatively proposes a signal modal decomposition strategy for early fault diagnosis of rotating machinery, which uses a fusion algorithm to reduce the number of modes. Multiple modes are fused in each iteration, and the operation efficiency is higher than that of the traditional FMD method.

[0045] Advantages of additional aspects of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings forming a part of this specification are used to provide a further understanding of the present invention. The schematic embodiments and descriptions thereof of the present invention are used to explain the present invention and do not constitute an improper limitation to the present invention.

[0047] Figure 1 Schematic flowchart of the signal modal decomposition method for early fault diagnosis of rotating machinery provided in Embodiment 1 of the present invention;

[0048] Figure 2 Original signal diagram provided in Embodiment 1 of the present invention;

[0049] Figure 3 Schematic diagram of the MFMD decomposition result (iterated 5 times) provided in Embodiment 1 of the present invention; wherein, (a) is the time-domain diagram of mode M1, (b) is the frequency-domain diagram of mode M1, (c) is the time-domain diagram of mode M2, and (d) is the frequency-domain diagram of mode M2;

[0050] Figure 4 Schematic diagram of the FMD decomposition result (iterated 5 times) provided in Embodiment 1 of the present invention; wherein, (a) is the time-domain diagram of mode M1, (b) is the frequency-domain diagram of mode M1, (c) is the time-domain diagram of mode M2, (d) is the frequency-domain diagram of mode M2, (e) is the time-domain diagram of mode M3, (f) is the frequency-domain diagram of mode M3, (g) is the time-domain diagram of mode M4, (h) is the frequency-domain diagram of mode M4, (i) is the time-domain diagram of mode M5, (j) is the frequency-domain diagram of mode M5, (k) is the time-domain diagram of mode M6, (l) is the frequency-domain diagram of mode M6, (m) is the time-domain diagram of mode M7, and (n) is the frequency-domain diagram of mode M7;

[0051] Figure 5 Schematic diagram of the FMD decomposition result (iterated 13 times) provided in Embodiment 1 of the present invention; wherein, (a) is the time-domain diagram of mode M1, (b) is the frequency-domain diagram of mode M1, (c) is the time-domain diagram of mode M2, and (d) is the frequency-domain diagram of mode M2;

[0052] Figure 6 Schematic diagram of a signal modal decomposition system for early fault diagnosis of rotating machinery provided in Embodiment 2 of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0053] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0054] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0055] In the case of no conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0056] Embodiment 1:

[0057] As described in the background art, although the existing feature mode decomposition method can accurately and adaptively separate the early fault components of vibration signals and is widely used in the field of early fault diagnosis technology, the operation efficiency of this method is relatively low, which is not conducive to real-time monitoring and rapid diagnosis. In view of this, this implementation provides a signal mode decomposition method for early fault diagnosis of rotating machinery. The vibration signal is decomposed by the MCKD algorithm to obtain the characteristic mode set of the signal, and then fused according to the cross-correlation spectrum of the modes and the correlation coefficient of the mode envelope power spectral density. Finally, the characteristic mode representing the fault is extracted. Experiments compared the operation efficiency of MFMD and FMD, and the results show that the operation efficiency of MFMD proposed by the present invention is significantly improved compared with FMD.

[0058] Specifically, as Figure 1 shown, it includes the following processes:

[0059] Step S1: The process of initializing the original vibration signal and the filter. Initialize the original signal, divide the frequency band, and design the FIR filter. The original vibration signal ( is the signal length) is initialized as a column vector, the signal decomposition range is constrained by frequency band division, and the initial filter bank , is the number of sub-frequency bands;

[0060] Step S2: The decomposition process, that is, the iterative mode extraction process. The characteristic modes are extracted band by band through the maximum correlation kurtosis deconvolution (MCKD) algorithm, and the filter is dynamically updated, is the total number of characteristic modes in this iteration;

[0061] Step S3: The post-processing process of mode fusion. The characteristic modes obtained by decomposing in Step S2 are subjected to two similarity fusion operations based on the modal cross-correlation sequence and the power spectral density similarity number, and the redundant modes are fused. When the maximum number of iterations is reached (that is, ), output the final characteristic mode set .

[0062] In step S1 of this implementation method, specifically, it includes:

[0063] S101: Convert the original vibration signal into a column vector:

[0064] (1).

[0065] Among them, is the signal length.

[0066] S102: Divide the entire frequency band evenly into sub-frequency bands, and the frequency band boundary set :

[0067] (2).

[0068] S103: For each sub-frequency band, call function to generate filter coefficients:

[0069] (3);

[0070] In the formula, is the Hanning window, and according to the filter length , call function and generate; is a tiny constant to avoid frequency band overlap; represents the frequency band boundary corresponding to the i-th sub-frequency band, represents the th sub-frequency band corresponding filter coefficient.

[0071] S104: Copy the original vibration signal times to obtain the signal set , and use the signal set as the input signal set of the MCKD decomposition algorithm in S2.

[0072] In step S2 of this implementation method, specifically, it includes:

[0073] S201: Enter the MCKD decomposition algorithm and perform iterative decomposition operations on each ( ).

[0074] S202: If the input signal fault period is not provided, estimate the fault period through Hilbert transform and autocorrelation function, and call function to calculate the envelope of the input signal:​

[0075] (4);

[0076] Wherein, is to take the average value, |hilbert(x)| is the envelope (a sequence), and taking the average value of it is to calculate the DC component of the envelope through operation. is to take of the modulus (the signal is converted into a complex signal after Hilbert transform), and call function to calculate the autocorrelation spectrum of the signal envelope .

[0077] Step S203: Determine fault period :

[0078] (5);

[0079] Wherein, is the position of the first zero crossing of the autocorrelation spectrum.

[0080] Step S204: Construct a multi-channel delay matrix , and calculate the inverse matrix of the delay matrix :

[0081] (6);

[0082] Wherein, represents the row index, (i.e., ) is the filter length; represents the column index, is the input signal length; represents the third dimension index, is the polynomial order, default value is 3, means the two-dimensional matrix of the first layer of the three-dimensional matrix is the signal page with a delay of T points.

[0083] S205: Enter the iterative decomposition process in the MCKD algorithm, and use the delay matrix to calculate the vector:

[0084] (7);

[0085] S206: Update the filter. First, construct the delay matrix :

[0086] (8);

[0087] In the formula, is the length of the input signal, , and then calculate the weights:

[0088] (9);

[0089] In the formula, , , , , , The calculation of is for calculating each element of the matrix, and represents a column vector used for matrix multiplication with means the two-dimensional matrix of the first layer of the three-dimensional matrix XmT, which is the signal page with mT points delayed.

[0090] Finally, update the filter and normalize it:

[0091] (10).

[0092] S207: Apply the vector in formula (7) to update the periodic fault according to S203 , and update the delay matrix and the inverse matrix according to S204.

[0093] S208: Calculate the output signal of this iteration and update the iteration number:

[0094] (11);

[0095] In the formula, is the function for filtering the signal, represents the number of this iteration.

[0096] S209: Return to S205 until the maximum number of iterations is reached to obtain the characteristic mode in the frequency band where the filter (denoted as ) and its corresponding filter (denoted as ), and at this time is the maximum number of iterations.

[0097] In step S3 of this implementation manner, specifically, it includes:

[0098] S301: Perform the first fusion operation. Call the function to calculate the correlation coefficient between each mode and the original vibration signal:

[0099] (12).

[0100] S302: Sort in descending order according to the magnitude of the correlation coefficient between the mode and the original vibration signal. The purpose is to preferentially retain the modes with strong correlation in the subsequent mode fusion and fuse the modes with weak correlation into the modes with strong correlation.

[0101] S303: According to the sorting of the modes in step S302, call function, starting from the first mode, calculate its cross-correlation spectrum with the remaining modes , take ( is the time delay, is 80% of the length of the original signal), and perform normalization processing on :

[0102] (13);

[0103] In the formula, and represent the mean value of the signal, and represent the standard deviation of the signal.

[0104] S304: Call function to find the peak value of (the minimum peak prominence is set to 0.2). According to the characteristics of the peak value of , judge whether to fuse the two modes. The specific conditions for fusing the two modes are: 1) The number of peak values ≥ 3 (i.e., the first set threshold); 2) The maximum peak value > 0.25 (i.e., the second set threshold); 3) The peak intervals are consistent (if the peak value is greater than 15, only the first 15 peak values are tested, and the 3 rule is used for testing). For the two modes that meet the fusion conditions ( and ) and their corresponding filter coefficients ( and ), perform a fusion operation to obtain the results of the first fusion operation and :

[0105] (14);

[0106] Repeat the operations of S303 and S304 until all modes participate in the fusion.

[0107] S305: Perform the second fusion operation. Use the modes after the first fusion as the input signals for the second fusion operation. Calculate the power spectral density (PSD) of the envelope of each mode and perform normalization.

[0108] Calculate the envelope signal of the modal:

[0109] (15);

[0110] In the formula, is the modal after the first fusion. Call the function to calculate the PSD of the envelope signal:

[0111] (16);

[0112] In the formula, represents the sampling frequency of the signal, represents using a 256-point Hamming window; 256 * 3 / 4 is the overlapping point number of 192; 256 is the FFT length of 256 points; ' ' indicates calculating the bilateral spectrum, is the output parameter, representing the frequency vector; is the output parameter, representing the power spectral density (PSD).

[0113] Convert the power spectral density to decibel units and normalize it:

[0114] (17).

[0115] Among them, represents the power spectral density, represents the minimum value of, represents the maximum value of.

[0116] S306: Sort the input modalities, with the same operations as in S301 and S302.

[0117] S307: According to the sorting of the modalities in S306, call the function, starting from the PSD of the normalized envelope of the first modality, calculate the correlation coefficient between its normalized envelope PSD and the normalized envelope PSD of the remaining modalities .

[0118] S308: If (i.e., the third set threshold), then fuse the two modalities and their corresponding filters, with the same fusion operation part as in S304, and calculate the PSD of the envelope of the fused modality. Repeat S307 and S308 until all modalities participate in the fusion.

[0119] S308: After performing the two fusion operations, if the number of iterations has not reached the maximum number of iterations, return to step S2 to continue the decomposition and fusion operation on the decomposed and fused modal set; if the number of iterations has reached the maximum number of iterations, use the mode obtained in S308 as the final characteristic modal set. 。

[0120] Based on specific data, the present implementation provides the following detailed exemplary steps:

[0121] Step 1: Import the original vibration signal (sampling frequency is 25600 Hz, duration is 1 s), as Figure 2 shown, initialize the vibration signal as (n is the signal length) as a column vector, set the number of divided frequency bands to 10, generate the frequency band boundaries , set the filter length to 30, call function to generate a Hanning window, and then call function to design the filter , copy the input signal into 10 copies for extracting 10 sets of characteristic modes.

[0122] Step 2: Enter the iterative process of the MFMD algorithm, apply the MCKD decomposition algorithm to decompose each input signal ( the maximum value of is the number of signals to be decomposed in the current iteration, and the initial value is 10) to obtain the characteristic modes (the same as the number of input signals). The specific description of the MCKD decomposition algorithm is as follows:

[0123] (1) Call to calculate the envelope of the input signal, call to calculate the autocorrelation spectrum of the signal, and take the fault period as the moment when the autocorrelation spectrum reaches the maximum value after passing through the zero point;

[0124] (2) Construct the delay matrix and the inverse matrix ;

[0125] (3) Enter the iterative process within the MCKD algorithm, calculate the vector , construct the delay matrix from the vector ;

[0126] (4) Calculate the weights, update the filter and normalize the filter, call function to calculate the output signal of this iteration, is the number of iterations;

[0127] (5) Use the obtained characteristic modes as the input to update the fault period and recalculate the delay matrix and the inverse matrix ;

[0128] (6) Determine whether the maximum number of iterations of the MCKD algorithm has been reached. If not, return to step 2 (2); if reached, This is the characteristic mode decomposed by the MCKD algorithm , the maximum number of iterations of the MCKD algorithm is 2 times less than the maximum number of iterations of the MFMD algorithm.

[0129] Step 3: Apply the fusion algorithm to perform similar fusion on the characteristic modes, as described below:

[0130] (1) First enter the first fusion process and calculate the correlation coefficient between each input mode and the original vibration signal , sort the input modes in descending order according to the obtained correlation coefficient;

[0131] (2) Starting from the first mode, determine in order whether the subsequent modes are integrated with it, and call The function calculates the cross-correlation spectrum of the two modes, and determines whether to fuse them according to the characteristics of the peak values ​​of the cross-correlation spectrum of the two modes. The conditions for fusing the two modes are: 1) the number of peaks ≥ 3 (i.e. the first set threshold); 2) the maximum peak value > 0.25 (i.e. the second set threshold); 3) the peak interval is consistent (if the peak value is greater than 15, only the first 15 peaks are tested, using 3 The two modes that meet the conditions and their corresponding filters are merged;

[0132] (3) The characteristic mode obtained after the first fusion is used as the input mode, the power spectrum density of each mode envelope is calculated and normalized to obtain , sort the input modalities in the same way as step (1) of step 3;

[0133] (4) Starting from the first mode, determine in order whether the subsequent modes are integrated with it, and call The function calculates the correlation coefficient of the envelope power spectral density after normalization of the two modes ,like (i.e., the second set threshold), the two modes and the corresponding filters are fused, and the fusion operation is completed.

[0134] (5) Determine whether the number of iterations of the MFMD algorithm exceeds the maximum number of iterations. If not, return to step 2 and use the characteristic modes obtained in this iteration as the input signal to perform decomposition and fusion operations again. If it exceeds, use the characteristic modes obtained in this iteration as the final decomposition mode set. , determine the mode that characterizes the early fault according to the characteristics of the fault signal.

[0135] The number of iterations of the MFMD algorithm is taken as 5. The decomposition and fusion results of the vibration signal of the fan gearbox in the present invention are as follows Figure 3 shown. For the FMD signal decomposition algorithm, the number of iterations is taken as 5 times, and the remaining parameters are the same as those of the original input signal and the MFMD algorithm settings. The final decomposition results are respectively as follows Figure 4 , Figure 5 shown.

[0136] It can be seen from Figure 3 that the characteristic modes obtained by MFMD decomposition (iterated 5 times). For mode M1, it can be seen from the figure that M1 has periodic peak value impacts, indicating that there are local damages on the gear teeth, resulting in greater impacts when some teeth are meshing. The time interval between the impacts is consistent with the gear rotation frequency. For mode M2, it can be seen from the figure that M2 shows continuous and dense vibrations, and M2 is a noise signal.

[0137] In summary, it can be seen that the MFMD algorithm can better extract the periodic impact fault signal (the characteristic of the early fault signal of the gearbox is periodic impact, obviously mode M1 is a periodic impact signal, that is, the early fault signal of the gearbox) in the vibration signal of the fan gearbox after 5 iterations.

[0138] As Figure 4 shown, the characteristic modes obtained by FMD decomposition (iterated 5 times). Each mode (that is, Figure 4 mode M1, mode M2, mode M3, mode M4, mode M5, mode M6 and mode M7 in) are not clearly distinguishable on the time domain diagram and the frequency domain diagram, indicating that the characteristic modes representing early faults are not extracted. As Figure 5 shown, the characteristic modes obtained by FMD decomposition (iterated 13 times) are similar to the characteristic modes obtained by the MFMD method (modes M1 and M2 are obtained), and the periodic vibration impact characteristics representing early faults are decomposed, but more iteration rounds are required. In summary, it can be seen that the MFMD algorithm has great advantages in operation efficiency compared with the FMD algorithm.

[0139] Example 2:

[0140] As Figure 6 shown, this implementation provides a signal modal decomposition system for early fault diagnosis of rotating machinery, including:

[0141] A filter bank generation unit configured to: convert the original vibration signal into a column vector and copy multiple copies to form multiple sub-signals, divide the entire frequency band into multiple sub-frequency bands, each sub-channel corresponds to a sub-signal, and calculate the frequency band boundaries of each sub-frequency band, and generate corresponding filter coefficients for each sub-frequency band;

[0142] An iterative decomposition operation unit, configured to perform iterative decomposition operations on each sub-signal in the form of a column vector to obtain the corresponding mode and corresponding filter coefficients of each sub-signal;

[0143] A first fusion unit, configured to calculate the correlation coefficient between each mode and the original vibration signal, sort them in descending order according to the magnitude of the correlation coefficient, start from the first mode, calculate its cross-correlation spectrum with the remaining modes, and perform the first fusion of the mode and the corresponding filter coefficients according to the peak value of the normalized cross-correlation spectrum;

[0144] A second fusion unit, configured to calculate the power spectral density of the normalized envelope of each mode after the first fusion, sort them in descending order according to the magnitude of the power spectral density, start from the power spectral density of the normalized envelope of the first mode, calculate its correlation coefficient with the power spectral density of the normalized envelope of the remaining modes after the first fusion, and perform the second fusion of the mode and the corresponding filter coefficients according to the correlation coefficient;

[0145] A characteristic mode set generation unit, configured to: if the number of iterations has not reached the maximum number of iterations, continue to perform the operation of the first fusion on the mode set after the second fusion; if the number of iterations reaches the maximum number of iterations, then use the obtained modes as the final characteristic mode set;

[0146] A fault signal extraction unit, configured to obtain a periodic impact fault signal according to the final characteristic mode set.

[0147] For the working processes of specific units, see the description in Embodiment 1 and will not be elaborated here.

[0148] It can be understood that the above units can be separately or all combined into one or several other units to form, or some of them can be further split into multiple smaller units with more specific functions to form, which can achieve the same operation without affecting the realization of the technical effects of the embodiments of this application. The above units are divided based on logical functions. In practical applications, the function of one unit can also be realized by multiple units, or the functions of multiple units can be realized by one unit. In other embodiments of this application, the system can also include other units. In practical applications, these functions can also be assisted by other units and can be realized through the cooperation of multiple units.

[0149] According to another embodiment of the present application, the system described in this embodiment can be constructed by running a computer program (including program code) that can execute the respective steps involved in the corresponding method described in Embodiment 1 on a general computing device such as a computer including processing elements and storage elements such as a Central Processing Unit (CPU), a Random Access Memory (RAM), and a Read-Only Memory (ROM), and the method of Embodiment 1 of the present application can be implemented. The computer program can be recorded on, for example, a computer-readable recording medium, loaded into the above computing device through the computer-readable recording medium, and run therein.

[0150] The foregoing are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A signal modal decomposition method for early fault diagnosis of rotating machinery, characterized in that: The process includes: The original vibration signal is converted into a column vector and replicated to form multiple sub-signals. The entire frequency band is divided into multiple sub-bands. Each sub-channel corresponds to a sub-signal. The frequency band boundaries of each sub-band are calculated. For each sub-band, the corresponding filter coefficient is generated. Perform iterative decomposition operation on each sub-signal in column vector form to obtain the mode and filter coefficient corresponding to each sub-signal; Calculate the correlation coefficient between each mode and the original vibration signal, sort them in descending order according to the size of the correlation coefficient, start from the first mode, calculate its cross-correlation spectrum with the remaining modes, and perform the first fusion of the modes and the corresponding filter coefficients according to the peak value of the standardized cross-correlation spectrum; Calculate the power spectral density of the normalized envelope of each mode after the first fusion, sort them in descending order according to the size of the power spectral density, start from the normalized envelope power spectral density of the first mode, calculate its correlation coefficient with the normalized envelope power spectral density of the other modes after the first fusion, and perform the second fusion of the mode and the corresponding filter coefficient according to the correlation coefficient; If the number of iterations does not reach the maximum number of iterations, the modal set after the second fusion is iteratively decomposed and calculated, and the first fusion operation is continued; if the number of iterations reaches the maximum number of iterations, the obtained mode is used as the final characteristic mode set; A periodic impact fault signal is obtained according to the final characteristic mode set; Among them, for any sub-signal Perform iterative decomposition operations, including: If no subsignal is provided Fault period, the fault period is estimated by Hilbert transform and autocorrelation function; Construct a multi-channel delay matrix; For each iteration, the multi-channel delay matrix is ​​used to calculate the vector , according to the calculation vector Constructing the delay matrix , according to the delay matrix Calculate the weights, update the filter coefficients according to the weights and normalize them, and output the eigenmodes and the corresponding filter coefficients corresponding to this iteration round; Perform multiple rounds of iterations until the characteristic mode corresponding to the maximum iteration round and the corresponding filter coefficient are output; Estimation of Failure Period by Hilbert Transform and Autocorrelation Function ,include: ； In the formula, is the position of the first zero crossing point of the autocorrelation spectrum, Sub-signal The autocorrelation spectrum of .

2. The signal modal decomposition method for early fault diagnosis of rotating machinery according to claim 1, characterized in that: Find the peak of the cross-correlation spectrum and decide whether to fuse the two modes based on the characteristics of the peak. If the fusion conditions are met, the two modes are fused, and the corresponding filter coefficients are fused at the same time. The cross-correlation spectrum of each mode is calculated in turn to complete the first fusion.

3. The signal modal decomposition method for early fault diagnosis of rotating machinery according to claim 2, characterized in that: The fusion conditions include: the number of peaks is greater than or equal to a first set threshold, the maximum peak is greater than a second set threshold, and the peak intervals are consistent.

4. The signal modal decomposition method for early fault diagnosis of rotating machinery according to claim 1, characterized in that: When the correlation coefficient is greater than the third set threshold, the two modes are fused, and the corresponding filter coefficients are fused at the same time to complete the second fusion.

5. The signal modal decomposition method for early fault diagnosis of rotating machinery according to claim 2 or 4, characterized in that: The two modes are fused, and the corresponding filter coefficients are fused at the same time, including: The two modes are directly added together as a new mode, and the filter coefficients corresponding to the two modes are added together to obtain the filter coefficients corresponding to the new mode.

6. The signal modal decomposition method for early fault diagnosis of rotating machinery according to claim 1, characterized in that: For each sub-band, generate the corresponding filter coefficients, including: ； in, For the The filter coefficients corresponding to the sub-bands; It is the Hanning window; is the filter length; is the number of sub-bands; is a constant; Representative The frequency band boundaries corresponding to the sub-bands are The function is used to generate the filter coefficients.

7. The signal modal decomposition method for early fault diagnosis of rotating machinery according to claim 6, characterized in that: Divide the entire frequency band into subbands, the set of band boundaries for: 。 8. A signal modal decomposition system for early fault diagnosis of rotating machinery, characterized in that: include: The filter bank generating unit is configured to: convert the original vibration signal into a column vector and replicate it into multiple sub-signals, divide the entire frequency band into multiple sub-bands, each sub-channel corresponds to a sub-signal, and calculate the frequency band boundary of each sub-band, and generate a corresponding filter coefficient for each sub-band; The iterative decomposition operation unit is configured to: perform iterative decomposition operation on each sub-signal in the form of a column vector to obtain a mode and a corresponding filter coefficient corresponding to each sub-signal; The first fusion unit is configured to: calculate the correlation coefficient between each mode and the original vibration signal, sort them in descending order according to the size of the correlation coefficients, start from the first mode, calculate its cross-correlation spectrum with the remaining modes, and perform the first fusion of the modes and the corresponding filter coefficients according to the peak value of the standardized cross-correlation spectrum; The second fusion unit is configured to: calculate the power spectral density of the normalized envelope of each mode after the first fusion, sort them in descending order according to the size of the power spectral density, start from the normalized envelope power spectral density of the first mode, calculate its correlation coefficient with the normalized envelope power spectral density of the remaining modes after the first fusion, and perform a second fusion of the mode and the corresponding filter coefficient according to the correlation coefficient; The characteristic mode set generation unit is configured to: if the number of iterations does not reach the maximum number of iterations, perform iterative decomposition calculation on the mode set after the second fusion, and continue the operation of the first fusion; if the number of iterations reaches the maximum number of iterations, use the obtained mode as the final characteristic mode set; The fault signal extraction unit is configured to: obtain a periodic impact fault signal according to the final characteristic mode set; Among them, for any sub-signal Perform iterative decomposition operations, including: If no subsignal is provided Fault period, the fault period is estimated by Hilbert transform and autocorrelation function; Construct a multi-channel delay matrix; For each iteration, the multi-channel delay matrix is ​​used to calculate the vector , according to the calculation vector Constructing the delay matrix , according to the delay matrix Calculate the weights, update the filter coefficients according to the weights and normalize them, and output the eigenmodes and the corresponding filter coefficients corresponding to this iteration round; Perform multiple rounds of iterations until the characteristic mode corresponding to the maximum iteration round and the corresponding filter coefficient are output; Estimation of Failure Period by Hilbert Transform and Autocorrelation Function ,include: ； In the formula, is the position of the first zero crossing point of the autocorrelation spectrum, Sub-signal The autocorrelation spectrum of .

Images