Fault quantitative separation method and device under multi-source excitation signal and medium thereof

Abstract

The invention discloses a fault quantitative separation method under a multi-source excitation signal, and the method comprises the steps: obtaining a multi-channel vibration signal, and carrying out the preprocessing of the vibration signal; performing total local mean decomposition on the preprocessed vibration signals to obtain decomposed signals, calculating covariance matrix eigenvalues of the decomposed signals, and completing estimation of the number of vibration sources in the vibration coupling signals based on all the eigenvalues; the decomposed signals and the preprocessed vibration signals form two-dimensional observation matrixes with the number equal to that of the information sources; processing the two-dimensional observation matrix to complete direction vector estimation, separating time-frequency distribution corresponding to each estimated source signal by using an L1 norm minimization method, and recovering the amplitude of the estimated source signal to ensure that the estimated source signal is close to a real source signal; processing the source signal to realize fault diagnosis classification of the rotating equipment; the rotating equipment state is scored by referring to national standard files, engineering experience and the equipment state, the fault type and the score are presented in the form of a broken line graph and a radar map, and the fault diagnosis accuracy is high.

Description

Technical Field

[0001] The present invention relates to the technical field of rotating equipment fault diagnosis, and in particular to a fault quantization separation method under multi-source excitation signals, a device and a medium thereof. Background Art

[0002] Industrial rotating machinery and equipment usually generate specific periodic vibration signals during operation. The operating status and health of the equipment can be analyzed based on these vibration signals. In actual industrial manufacturing sites, there is often more than one source of fault, which means that the sensor often does not collect a single vibration source, but the coupling of multiple signals. The above phenomenon makes troubleshooting very difficult. Therefore, in order to solve this problem and improve the accuracy of fault detection and diagnosis of modern equipment, the most critical step is to separate the fault signal from the mixed signal. This problem is called blind source separation of signals.

[0003] There are three common methods to solve the problem of blind source separation, namely, Independent Component Analysis (ICA), Principal Component Analysis (PCA) and Sparse Component Analysis (SCA). The premise of the ICA algorithm is that the source signal is statistically independent, and each independent component must conform to a non-Gaussian distribution. However, it is difficult for modern mechanical equipment to meet this assumption. The PCA algorithm separates signals based on the variance of the data, so it can only capture the linear characteristics of the signal. When multiple systems are coupled under actual working conditions, the signal has a complex nonlinear relationship, and PCA will not be able to accurately and effectively achieve signal separation. SCA recovers the original signal from the mixed signal through sparse coding. For comprehensive comparison, SCA is more suitable as a method to solve BSS (Blind Source Separation). When the mixing matrix and source signal are unknown, the source signal sparsity assumption can be used to estimate the mixing matrix and restore the source signal. The SCA algorithm mainly estimates the mixing matrix and restores the source signal. However, due to the complexity of the industrial environment, the on-site noise will destroy the signal sparsity, resulting in a decrease in the signal separation effect, which in turn affects the accuracy of rotating equipment fault diagnosis. Summary of the invention

[0004] The technical problem to be solved by the present invention is: in order to solve the problem that the fault diagnosis accuracy of the prior art is low and it is not applicable to the fault diagnosis of rotating equipment, the present invention provides a fault quantization separation method under multi-source excitation signals, which has high accuracy in the fault diagnosis of rotating equipment and is applicable to the fault diagnosis of rotating equipment.

[0005] The technical solution adopted by the present invention to solve the technical problem is: a fault quantization separation method under multi-source excitation signals, the method comprising the following steps:

[0006] S1, obtaining a multi-channel vibration signal f(n), and preprocessing the vibration signal f(n);

[0007] S2, performing overall local mean decomposition on the preprocessed vibration signal to obtain a decomposed signal, calculating the eigenvalues ​​of the covariance matrix of the decomposed signal, sorting all the eigenvalues, adaptively setting thresholds, and completing the estimation of the number of information sources in the vibration coupling signal;

[0008] S3, respectively composing the decomposed signal and the preprocessed vibration signal into a two-dimensional measurement matrix equal in number to the number of the signal sources;

[0009] S4, performing short-time Fourier time-frequency transform on the two-dimensional observation matrix to obtain a two-dimensional scattered point distribution, setting a threshold according to the device type, filtering the two-dimensional scattered points with amplitudes less than the threshold, mapping the remaining points to the unit circle, using the fuzzy C-means algorithm to estimate the cluster center of the straight line normalization, and after completing the direction vector estimation, using the L1 norm minimization method to separate the time-frequency distribution corresponding to each estimated source signal;

[0010] S5, based on the time-frequency distribution corresponding to each estimated source signal, performing amplitude restoration on the estimated source signal to ensure that the estimated source signal is close to the real source signal;

[0011] S6, performing Fourier transform on the source signal, converting the time domain signal into a frequency domain signal, locating the characteristic frequency according to the equipment information, calculating the amplitude at the corresponding characteristic frequency, and realizing fault diagnosis and classification of the rotating equipment;

[0012] S7, refer to national standard documents, engineering experience and equipment status, set feature thresholds, score equipment status based on frequency amplitude and threshold, and present fault types and scores in the form of line graphs and radar charts.

[0013] Further, specifically, the preprocessing of the vibration signal in step S1 includes the following steps:

[0014] Assuming that the vibration signal f(n) is a one-dimensional discrete function on the domain F = (0, 1, ..., N-1), the structural element g(n) is a one-dimensional discrete function on the domain G = (0, 1, ..., M-1), and N ≥ M, then the four basic operators of the vibration signal and the structural element can be defined as the following formulas:

[0015]

[0016]

[0017] in, represents the dilation operator, represents the corrosion operator, represents the opening operator, and represents the closing operator;

[0018] The vibration signal f(n) is filtered using an alternating hybrid filter to obtain a coupled vibration signal The calculation formula is as follows:

[0019] [(f)ah(g)](n)=[(f)oc(g)+(f)co(g)](n) / 2,

[0020]

[0021] Among them, oc represents the alternating filtering operation of first opening and then closing, co represents the alternating filtering operation of first closing and then opening, and ah represents the alternating mixed filtering operation.

[0022] Further, specifically, the step S2 specifically includes the following steps:

[0023] S20, in the coupled vibration signal Add M times white noise n with amplitude k and mean 0 to m (t)(m=1,2,…,M), and get the analysis signal x m (t):

[0024]

[0025] S21, obtaining the analysis signal x m All local extreme points n of (t) i , calculate the average value m of the adjacent local mean extreme points i and the envelope function value a i , the calculation formula is:

[0026]

[0027] The mean value m of all two adjacent local extreme values i and all adjacent envelope function values ​​a i Connect them with straight lines and use the sliding average method to get the local mean function m 11 (t) and envelope estimation function a 11 (t), the two subscripts represent the function and the number of iterations respectively;

[0028] S22, from the analysis signal x m (t) to separate the local mean function m 11 (t), and get the residual h 11 (t):

[0029] h 11 (t) = x m (t)-m 11 (t);

[0030] S23, for the residual h 11 (t) is demodulated to obtain the demodulation function s 11 (t):

[0031]

[0032] S24, judging whether the demodulated signal is a pure FM signal, the judging condition is: |a 12 (t)-1|≤δ;

[0033] If 11 (t) belongs to pure FM signal, δ is the setting threshold;

[0034] On the contrary, if s 11 (t) is not a pure FM signal, repeat steps S20 to S21 to convert s 11 (t) Continue to decompose until s 1n (t) belongs to pure FM signal:

[0035]

[0036] S25, multiplying all the envelope estimation functions obtained to calculate the instantaneous amplitude function a1(t):

[0037]

[0038] S26, based on pure FM signal s 1n (t) and the instantaneous amplitude function a1(t) to calculate the product function PF1(t):

[0039] PF1(t)=a1(t)s 1n (t);

[0040] S27, from the analysis signal x m (t) is subtracted from PF1(t) to obtain the first signal u1(t), and the steps S21 to S26 are repeated k times until u1(t) is k (t) is a monotonic function, and finally k PF components are obtained:

[0041]

[0042] Where K is the total PF component, PF k,m (t) is the kth PF component of the mth decomposition;

[0043] S28, for signal x m (t) The PF components of the same order obtained by decomposition are averaged to reduce the random influence introduced by white noise:

[0044]

[0045] Among them, PF j,m (t) is the jth PF component after adding white noise for the mth time, PF j (t) is the average value of the j-th order component;

[0046] S29, final signal x m (t) is decomposed into multiple averaged PF components, u k (t) is the final residual signal:

[0047]

[0048] S30, all the PF components are combined into a signal matrix, the covariance matrix is ​​calculated, and eigenvalue decomposition is performed, all the obtained eigenvalues ​​are sorted from large to small, and two adjacent eigenvalues ​​are subjected to difference operation:

[0049] The maximum difference is adaptively set as the threshold, and the number of multi-channel signals is estimated. The i corresponding to the maximum difference is the number of signal sources z.

[0050] Further, specifically, the step S3 specifically includes the following steps:

[0051] Calculate the correlation coefficient λ between the PF component and the vibration signal f(n) s , the calculation formula is:

[0052]

[0053] According to the number of information sources z and the correlation coefficient λ s Construct z vibration signals composed of PF components and coupling The two-dimensional observation matrix X i .

[0054] Further, specifically, the step S4 specifically includes the following steps:

[0055] S41, set the fuzzy weighted index m, the number of cluster categories is C (2≤C≤n), n represents the number of data sample types, set the iteration stop threshold to ε, and initialize the cluster center value P (0) And the number of iterations l, in the initial state l = 0;

[0056] S42, calculate the partition matrix U l : Calculate the two-dimensional scattered points xk and the cluster center P of the i-th class i The distance norm d between ik :

[0057]

[0058] Where A is a symmetric positive definite matrix of order s×s, and I is the identity matrix; when A=I, d ik is the Euclidean distance. When A≠I, d ik is the Mahalanobis distance;

[0059] For any i, k, if

[0060]

[0061] For any i, k, if And when j≠r,

[0062] S43, update the cluster center value:

[0063]

[0064] S44, if ||P (l+1) -P (l) ||<ε, stop iteration and go to the next step. Otherwise, if ||P (l+1) -P (l) ||≥ε, return to step S42;

[0065] S45, calculate the direction vector of the clustering line from the cluster center value to obtain the mixing matrix T of the estimated source signal

[0066] S46, using the L1 norm minimization method to separate the time-frequency distribution S corresponding to each estimated source signal i (f,t).

[0067] Further, specifically, the step S5 specifically includes the following steps:

[0068] The time-frequency distribution S corresponding to each estimated source signal i (f, t) performs inverse time-frequency transform to obtain the corresponding time domain signal. The calculation formula is:

[0069]

[0070] The calculated time domain signal is multiplied by the corresponding normalization factor to restore the actual amplitude of the signal, ensuring that the estimated source signal is close to the real source signal. The calculation formula is:

[0071]

[0072] Among them, x i ,y i Represents the normalized cluster line center corresponding to the i-th source.

[0073] Further, specifically, the step S7 specifically includes the following steps:

[0074] S71, according to national standard documents, engineering experience and equipment status, set the characteristic frequency threshold corresponding to different faults, combine the frequency amplitude of the source signal with the characteristic threshold, and convert it into a percentage score:

[0075] G=F i -|CA-CA i |γ i

[0076] Among them, F i is the scoring threshold, CA is the characteristic frequency amplitude, CA i is the amplitude threshold, γ i is the score conversion factor;

[0077] S72, divides the operating status into four types according to the scores - healthy, available, warning and faulty. At the same time, the scores within the monitoring time are plotted as a line graph to realize the status trend of the equipment during the operating time, clearly and intuitively realize the quantitative monitoring of the equipment status, and draw a scoring radar chart based on the type of fault diagnosis to reflect the type and degree of fault.

[0078] Further, specifically, the types of fault diagnosis include rotor misalignment, component rubbing, rolling bearing abnormality, poor lubrication of rolling bearing, rotor imbalance and rotor looseness.

[0079] A computer device comprising:

[0080] processor;

[0081] A memory for storing executable instructions;

[0082] The processor is used to read the executable instructions from the memory and execute the executable instructions to implement the fault quantification separation method under the multi-source excitation signal as described above.

[0083] A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the processor implements the fault quantification separation method under multi-source excitation signals as described above.

[0084] The beneficial effect of the present invention is that the fault quantization separation method under multi-source excitation signals of the present invention can approximately project the noise to zero through preprocessing and retain the main characteristics of the vibration signal, thereby meeting the sparsity requirement of the mixed matrix estimation; then the signal decomposition is completed by the method of overall local mean decomposition, and then the number of source signals is estimated; then an equal two-dimensional observation matrix is ​​constructed, and the mixed matrix is ​​estimated by a clustering algorithm to realize source signal estimation, and the source signal is estimated to perform amplitude recovery; finally, the source signal is processed to realize the fault diagnosis of the rotating equipment and score the equipment status, so as to achieve the purpose of fault quantification. The present invention has high accuracy in rotating equipment fault diagnosis and is suitable for rotating equipment fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0086] Figure 1 It is a schematic diagram of the method flow of embodiment 1 of the present invention.

[0087] Figure 2 This is a radar scoring diagram of the first embodiment of the present invention.

[0088] Figure 3 It is a structural diagram of a computer device according to Embodiment 2 of the present invention.

[0089] In the figure, 10 is a computer device; 1002 is a processor; 1004 is a memory; and 1006 is a transmission device. DETAILED DESCRIPTION

[0090] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0091] Example 1

[0092] The present application embodiment provides a method for quantifying and separating faults under multi-source excitation signals, such as Figure 1 As shown, the method comprises the following steps:

[0093] S1, obtaining a multi-channel vibration signal f(n), and preprocessing the vibration signal f(n);

[0094] S2, perform overall local mean decomposition on the preprocessed vibration signal to obtain the decomposed signal (interpretation), calculate the eigenvalues ​​of the covariance matrix of the decomposed signal, sort all the eigenvalues, adaptively set the threshold, and complete the estimation of the number of information sources in the vibration coupling signal;

[0095] S3, the decomposed signal and the preprocessed vibration signal are respectively formed into a two-dimensional measurement matrix equal to the number of signal sources;

[0096] S4, perform short-time Fourier time-frequency transform on the two-dimensional observation matrix to obtain a two-dimensional scattered point distribution, set a threshold according to the device type, filter the two-dimensional scattered points with amplitudes less than the threshold, map the remaining points to the unit circle, use the fuzzy C-means algorithm to estimate the cluster center of the straight line normalization, and after completing the direction vector estimation, use the L1 norm minimization method to separate the time-frequency distribution corresponding to each estimated source signal;

[0097] S5, based on the time-frequency distribution corresponding to each estimated source signal, performing amplitude restoration on the estimated source signal to ensure that the estimated source signal is close to the real source signal;

[0098] S6, performing Fourier transform on the source signal, converting the time domain signal into a frequency domain signal, locating the characteristic frequency according to the equipment information, calculating the amplitude at the corresponding characteristic frequency, and realizing the fault diagnosis and classification of the rotating equipment;

[0099] S7, refer to national standard documents, engineering experience and equipment status, set feature thresholds, score equipment status based on frequency amplitude and threshold, and present fault types and scores in the form of line graphs and radar charts.

[0100] In this embodiment, in step S1, multiple sensors are used to collect vibration signals in different directions on a certain cross section of the rotating device. The sensors may be displacement sensors, velocity sensors or acceleration sensors. Preprocessing the vibration signals includes the following steps:

[0101] Assume that the vibration signal f(n) is a one-dimensional discrete function on the domain F = (0, 1, ..., N-1), the structural element g(n) is a one-dimensional discrete function on the domain G = (0, 1, ..., M-1), and N ≥ M, then the four basic operators of the vibration signal and the structural element can be defined as follows:

[0102]

[0103]

[0104]

[0105] in, represents the dilation operator, represents the corrosion operator, represents the opening operator, and represents the closing operator;

[0106] The vibration signal f(n) is filtered using an alternating hybrid filter to obtain a coupled vibration signal The calculation formula is as follows:

[0107] [(f)ah(g)](n)=[(f)oc(g)+fco(g)](n) / 2,

[0108]

[0109] Among them, oc represents the alternating filtering operation of first opening and then closing, co represents the alternating filtering operation of first closing and then opening, and ah represents the alternating mixed filtering operation.

[0110] It should be noted that the vibration signal f(n) is subjected to noise reduction by preprocessing to obtain the coupled vibration signal The use of alternating hybrid filters is a morphological filtering method that can approximately map the noise to zero and retain the main features of the vibration signal, thereby achieving signal sparsity representation to a certain extent and improving the accuracy of fault diagnosis of rotating equipment.

[0111] In this embodiment, step S2 specifically includes the following steps:

[0112] S20, in the coupled vibration signal Add M times white noise n with amplitude k and mean 0 to m (t)(m=1,2,…,M), and get the analysis signal x m (t):

[0113]

[0114] S21, obtain analysis signal x m All local extreme points n of (t) i , calculate the average value m of the adjacent local mean extreme points i and the envelope function value a i , the calculation formula is:

[0115]

[0116]

[0117] The mean m of all two adjacent local extreme values i and all adjacent envelope function values ​​a i Connect them with straight lines and use the sliding average method to get the local mean function m 11 (t) and envelope estimation function a 11 (t), the two subscripts represent the function and the number of iterations respectively;

[0118] S22, from the analysis signal x m (t) to separate the local mean function m 11 (t), and get the residual h 11 (t):

[0119] h 11(t) = x m (t)-m 11 (t);

[0120] S23, for the residual h 11 (t) is demodulated to obtain the demodulation function s 11 (t):

[0121]

[0122] S24, judging whether the demodulated signal is a pure FM signal, the judging condition is: |a 12 (t)-1|≤δ;

[0123] If 11 (t) belongs to pure FM signal, δ is the setting threshold;

[0124] On the contrary, if s 11 (t) is not a pure FM signal, repeat steps S20 to S21 to convert s 11 (t) Continue to decompose until s 1n (t) belongs to pure FM signal:

[0125]

[0126] S25, multiply all the envelope estimation functions to calculate the instantaneous amplitude function a1(t):

[0127]

[0128] S26, based on pure FM signal s 1n (t) and the instantaneous amplitude function a1(t) to calculate the product function PF1(t):

[0129] PF1(t)=a1(t)s 1n (t);

[0130] S27, from the analysis signal x m (t) is subtracted from PF1(t) to obtain the first signal u1(t), and steps S21 to S26 are repeated k times until u1(t) is k (t) is a monotonic function, and finally k PF components are obtained:

[0131]

[0132] Where K is the total PF component, PF k,m (t) is the kth PF component of the mth decomposition;

[0133] S28, for signal x m(t) The PF components of the same order obtained by decomposition are averaged to reduce the random influence introduced by white noise:

[0134]

[0135] Among them, PF j,m (t) is the jth PF component after adding white noise for the mth time, PF j (t) is the average value of the j-th order component;

[0136] S29, final signal x m (t) is decomposed into multiple averaged PF components, u k (t) is the final residual signal:

[0137]

[0138] S30, all PF components are combined into a signal matrix, the covariance matrix is ​​calculated, and eigenvalue decomposition is performed, all the obtained eigenvalues ​​are sorted from large to small, and the difference operation is performed on two adjacent eigenvalues:

[0139] The maximum difference is adaptively set as the threshold, and the number of multi-channel signals is estimated. The i corresponding to the maximum difference is the number of signal sources z.

[0140] In this embodiment, step S3 specifically includes the following steps:

[0141] Calculate the correlation coefficient λ between the PF component and the vibration signal f(n) s , the calculation formula is:

[0142]

[0143] According to the number of information sources z and the correlation coefficient λ s Construct z vibration signals composed of PF components and coupling The two-dimensional observation matrix X i .

[0144] In this embodiment, step S4 specifically includes the following steps:

[0145] S41, set the fuzzy weighted index m, the number of cluster categories is C (2≤C≤n), n represents the number of data sample types, set the iteration stop threshold to ε, and initialize the cluster center value P (0) And the number of iterations l, in the initial state l = 0;

[0146] S42, calculate the partition matrix U l : Calculate the two-dimensional scattered points x k and the cluster center P of the i-th class iThe distance norm d between ik :

[0147]

[0148] Where A is a symmetric positive definite matrix of order s×s, and I is the identity matrix; when A=I, d ik is the Euclidean distance. When A≠I, d ik is the Mahalanobis distance;

[0149] For any i, k, if

[0150]

[0151] For any i, k, if And when j≠r,

[0152] S43, update the cluster center value:

[0153]

[0154] S44, if ||P (l+1) -P (l) ||<ε, stop iteration and go to the next step. Otherwise, if ||P (l+1) -P (l) ||≥ε, then return to step S42;

[0155] S45, calculating the direction vector of the clustering line from the cluster center value to obtain a mixing matrix T of the estimated source signal;

[0156] S46, using the L1 norm minimization method to separate the time-frequency distribution S corresponding to each estimated source signal i (f,t).

[0157] In this embodiment, step S5 specifically includes the following steps:

[0158] The time-frequency distribution S corresponding to each estimated source signal i (f, t) performs inverse time-frequency transform to obtain the corresponding time domain signal. The calculation formula is:

[0159]

[0160] The calculated time domain signal is multiplied by the corresponding normalization factor to restore the actual amplitude of the signal, ensuring that the estimated source signal is close to the real source signal. The calculation formula is:

[0161]

[0162] Among them, x i ,yi Represents the normalized cluster line center corresponding to the i-th source.

[0163] In this embodiment, step S7 specifically includes the following steps:

[0164] S71, according to national standard documents, engineering experience and equipment status, set the characteristic frequency threshold corresponding to different faults, combine the frequency amplitude of the source signal with the characteristic threshold, and convert it into a percentage score:

[0165] G=F i -|CA-CA i |γ i

[0166] Among them, F i is the scoring threshold, CA is the characteristic frequency amplitude, CA i is the amplitude threshold, γ i is the score conversion factor;

[0167] S72, divides the operation status into four types according to the score - healthy, available, warning and fault, and draws the score during the monitoring time as a line chart to realize the status trend of the equipment during the operation time, clearly and intuitively realize the quantitative monitoring of the equipment status, and draw a score radar chart based on the type of fault diagnosis, see Figure 2 As shown, it reflects the type and degree of fault.

[0168] In this embodiment, the types of fault diagnosis include rotor misalignment, component rubbing, rolling bearing abnormality, poor lubrication of rolling bearing, rotor imbalance and rotor looseness.

[0169] The fault quantification separation method under multi-source excitation signals of the present invention can approximately project the noise to zero through preprocessing, and retain the main characteristics of the vibration signal, thereby meeting the sparsity requirement of the mixed matrix estimation; then complete the signal decomposition by the method of overall local mean decomposition, and then realize the estimation of the number of source signals; then construct an equal amount of two-dimensional observation matrix, estimate the mixed matrix and realize source signal estimation by clustering algorithm, and estimate the source signal for amplitude recovery; finally, process the source signal to realize the fault diagnosis of the rotating equipment and score the equipment status, so as to achieve the purpose of fault quantification. The present invention has high accuracy in rotating equipment fault diagnosis and is suitable for rotating equipment fault diagnosis.

[0170] Example 2

[0171] An embodiment of the present application provides a computer device, which includes a processor and a memory, wherein the memory stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by the processor to implement a fault quantization separation method under a multi-source excitation signal as provided in the above method embodiment.

[0172] Figure 3 A schematic diagram of the hardware structure of a device for implementing a method for quantitatively separating faults under a multi-source excitation signal provided in an embodiment of the present application is shown. The device may participate in constituting or include the apparatus or system provided in an embodiment of the present application. Figure 3 As shown, the computer device 10 may include one or more processors 1002 (the processor may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 1004 for storing data, and a transmission device 1006 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 3 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 3 More or fewer components as shown, or with Figure 3 Different configurations are shown.

[0173] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer device 10 (or mobile device). As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0174] The memory 1004 can be used to store software programs and modules of application software, such as a program instruction / data storage device corresponding to a method for quantitative separation of faults under a multi-source excitation signal in an embodiment of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 1004, that is, to implement the above-mentioned method. The memory 1004 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 1004 may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the computer device 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0175] The transmission device 1006 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer device 10. In one example, the transmission device 1006 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 1006 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0176] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer device 10 (or mobile device).

[0177] Example 3

[0178] An embodiment of the present application also provides a computer-readable storage medium, which can be set in a server to store at least one instruction or at least one program related to a fault quantization separation method under a multi-source excitation signal in a method embodiment. The at least one instruction or the at least one program is loaded and executed by the processor to implement a fault quantization separation method under a multi-source excitation signal provided in the above method embodiment.

[0179] Optionally, in this embodiment, the storage medium may be located in at least one of the multiple network servers of the computer network. Optionally, in this embodiment, the storage medium may include, but is not limited to, various media that can store program codes, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk, or an optical disk.

[0180] Example 4

[0181] The embodiment of the present invention further provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes a fault quantization separation method under a multi-source excitation signal provided in the above various optional implementations.

[0182] It should be noted that the above-mentioned sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above-mentioned specific embodiments of the present application are described. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0183] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment and storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0184] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.

[0185] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A fault quantization separation method under multi-source excitation signals, characterized in that: The method comprises the following steps: S1, obtaining a multi-channel vibration signal f(n), and preprocessing the vibration signal f(n); S2, performing overall local mean decomposition on the preprocessed vibration signal to obtain a decomposed signal, calculating the eigenvalues ​​of the covariance matrix of the decomposed signal, sorting all the eigenvalues, adaptively setting thresholds, and completing the estimation of the number of information sources in the vibration coupling signal; S3, respectively composing the decomposed signal and the preprocessed vibration signal into a two-dimensional measurement matrix equal in number to the number of the signal sources; S4, performing short-time Fourier time-frequency transform on the two-dimensional observation matrix to obtain a two-dimensional scattered point distribution, setting a threshold according to the device type, filtering the two-dimensional scattered points with amplitudes less than the threshold, mapping the remaining points to the unit circle, using the fuzzy C-means algorithm to estimate the cluster center of the straight line normalization, and after completing the direction vector estimation, using the L1 norm minimization method to separate the time-frequency distribution corresponding to each estimated source signal; S5, based on the time-frequency distribution corresponding to each estimated source signal, performing amplitude restoration on the estimated source signal to ensure that the estimated source signal is close to the real source signal; S6, performing Fourier transform on the source signal, converting the time domain signal into a frequency domain signal, locating the characteristic frequency according to the equipment information, calculating the amplitude at the corresponding characteristic frequency, and realizing fault diagnosis and classification of the rotating equipment; S7, refer to national standard documents, engineering experience and equipment status, set feature thresholds, score equipment status based on frequency amplitude and threshold, and present fault types and scores in the form of line graphs and radar charts.

2. The fault quantization separation method under multi-source excitation signals according to claim 1, characterized in that: The preprocessing of the vibration signal in step S1 comprises the following steps: Assuming that the vibration signal f(n) is a one-dimensional discrete function on the domain F = (0, 1, ..., N-1), the structural element g(n) is a one-dimensional discrete function on the domain G = (0, 1, ..., M-1), and N ≥ M, then the four basic operators of the vibration signal and the structural element can be defined as the following formulas: in, represents the dilation operator, represents the corrosion operator, represents the opening operator, and represents the closing operator; The vibration signal f(n) is filtered using an alternating hybrid filter to obtain a coupled vibration signal The calculation formula is as follows: [(f)ah(g)](n)=[(f)oc(g)+(f)co(g)](n) / 2, Among them, oc represents the alternating filtering operation of first opening and then closing, co represents the alternating filtering operation of first closing and then opening, and ah represents the alternating mixed filtering operation.

3. The fault quantization separation method under multi-source excitation signals according to claim 2, characterized in that: The step S2 specifically includes the following steps: S20, in the coupled vibration signal Add M times white noise n with amplitude k and mean 0 to m (t)(m=1,2,…,M), and get the analysis signal x m (t): S21, obtaining the analysis signal x m All local extreme points n of (t) i , calculate the average value m of the adjacent local mean extreme points i and the envelope function value a i , the calculation formula is: The mean value m of all two adjacent local extreme values i and all adjacent envelope function values ​​a i Connect them with straight lines and use the sliding average method to get the local mean function m 11 (t) and envelope estimation function a 11 (t), the two subscripts represent the function and the number of iterations respectively; S22, from the analysis signal x m (t) to separate the local mean function m 11 (t), and get the residual h 11 (t): h 11 (t)=x m (t)-m 11 (t); S23, for the residual h 11 (t) is demodulated to obtain the demodulation function s 11 (t): S24, judging whether the demodulated signal is a pure FM signal, the judging condition is: |a 12 (t)-1|≤δ; If 11 (t) belongs to pure FM signal, δ is the setting threshold; On the contrary, if s 11 (t) is not a pure FM signal, repeat steps S20 to S21 to convert s 11 (t) Continue to decompose until s 1n (t) belongs to pure FM signal: S25, multiplying all the envelope estimation functions obtained to calculate the instantaneous amplitude function a1(t): S26, based on pure FM signal s 1n (t) and the instantaneous amplitude function a1(t) to calculate the product function PF1(t): PF1(t)=a1(t)s 1n (t); S27, from the analysis signal x m (t) is subtracted from PF1(t) to obtain the first signal u1(t), and the steps S21 to S26 are repeated k times until u1(t) is k (t) is a monotonic function, and finally k PF components are obtained: Where K is the total PF component, PF k,m (t) is the kth PF component of the mth decomposition; S28, for signal x m (t) The PF components of the same order obtained by decomposition are averaged to reduce the random influence introduced by white noise: Among them, PF j,m (t) is the jth PF component after adding white noise for the mth time, PF j (t) is the average value of the j-th order component; S29, final signal x m (t) is decomposed into multiple averaged PF components, u k (t) is the final residual signal: S30, all the PF components are combined into a signal matrix, the covariance matrix is ​​calculated, and eigenvalue decomposition is performed, all the obtained eigenvalues ​​are sorted from large to small, and two adjacent eigenvalues ​​are subjected to difference operation: The maximum difference is adaptively set as the threshold, and the number of vibration sources in the vibration coupling signal is estimated. The i corresponding to the maximum difference is the number of information sources z.

4. The fault quantization separation method under multi-source excitation signals according to claim 3 is characterized in that: The step S3 specifically comprises the following steps: Calculate the correlation coefficient λ between the PF component and the vibration signal f(n) s , the calculation formula is: According to the number of information sources z and the correlation coefficient λ s Construct z vibration signals composed of PF components and coupling The two-dimensional observation matrix X i .

5. The fault quantization separation method under multi-source excitation signals according to claim 1, characterized in that: The step S4 specifically comprises the following steps: S41, set the fuzzy weighted index m, the number of cluster categories is C (2≤C≤n), n represents the number of data sample types, set the iteration stop threshold to ε, and initialize the cluster center value P (0) And the number of iterations l, in the initial state l = 0; S42, calculate the partition matrix U l : Calculate the two-dimensional scattered points x k and the cluster center P of the i-th class i The distance norm d between ik : Where A is a symmetric positive definite matrix of order s×s, and I is the identity matrix; when A=I, d ik is the Euclidean distance. When A≠I, d ik is the Mahalanobis distance; For any i, k, if For any i, k, if And when j≠r, S43, update the cluster center value: S44, if ||P (l+1) -P (l) ||<ε, stop iteration and go to the next step. Otherwise, if ||P (l+1) -P (l) ||≥ε, return to step S42; S45, calculating the direction vector of the clustering line from the cluster center value to obtain a mixing matrix T of the estimated source signal; S46, using the L1 norm minimization method to separate the time-frequency distribution S corresponding to each estimated source signal i (f,t).

6. The fault quantization separation method under multi-source excitation signals according to claim 5, characterized in that: The step S5 specifically comprises the following steps: The time-frequency distribution S corresponding to each estimated source signal i (f, t) performs inverse time-frequency transform to obtain the corresponding time domain signal. The calculation formula is: The calculated time domain signal is multiplied by the corresponding normalization factor to restore the actual amplitude of the signal, ensuring that the estimated source signal is close to the real source signal. The calculation formula is: Among them, x i ,y i Represents the normalized cluster line center corresponding to the i-th source.

7. The fault quantization separation method under multi-source excitation signals according to claim 1, characterized in that: The step S7 specifically includes the following steps: S71, according to national standard documents, engineering experience and equipment status, set the characteristic frequency threshold corresponding to different faults, combine the frequency amplitude of the source signal with the characteristic threshold, and convert it into a percentage score: G=F i -|CA-CA i |γ i Among them, F i is the scoring threshold, CA is the characteristic frequency amplitude, CA i is the amplitude threshold, γ i is the score conversion factor; S72, divides the operating status into four types according to the scores - healthy, available, warning and faulty. At the same time, the scores within the monitoring time are plotted as a line graph to realize the status trend of the equipment during the operating time, clearly and intuitively realize the quantitative monitoring of the equipment status, and draw a scoring radar chart based on the type of fault diagnosis to reflect the type and degree of fault.

8. The fault quantization separation method under multi-source excitation signals according to claim 3, characterized in that: The types of fault diagnosis include rotor misalignment, component rubbing, rolling bearing abnormality, poor rolling bearing lubrication, rotor imbalance and rotor looseness.

9. A computer device, characterized in that: include: processor; A memory for storing executable instructions; The processor is used to read the executable instructions from the memory, and execute the executable instructions to implement the fault quantification separation method under multi-source excitation signals according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the processor implements the fault quantification separation method under multi-source excitation signals according to any one of claims 1 to 8.

Images