Coupled vibration information source separation method based on blind source separation

Abstract

The invention relates to the technical field of vibration signal processing, in particular to a coupling vibration information source separation method based on blind source separation. The method comprises the following steps: processing an obtained multi-channel signal based on local value decomposition and adaptive singular value decomposition, and estimating the number of vibration sources in the signal to obtain the number of information sources; determining a blind source separation algorithm for multi-channel signal processing based on the number of channels and the number of information sources for signal acquisition, and if the blind source separation is underdetermined, decomposing and reconstructing the multi-channel signals, and converting the underdetermined blind source separation into positive determined blind source separation; separating the multi-channel signal or the reconstructed multi-channel signal by adopting rapid independent component analysis to obtain a vibration source estimation signal; and performing cross correlation calculation and gain compensation on the vibration source estimation signal, then performing signal processing on the vibration source estimation signal after gain compensation, setting a characteristic threshold to classify fault types, quantifying fault degrees, and calculating a fault score. According to the invention, the difficulty of underdetermined blind source separation is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of vibration signal processing, and in particular to a coupled vibration source separation method based on blind source separation. Background Art

[0002] Rotating machinery is the core production equipment in the manufacturing industry, and its fault diagnosis and status monitoring are crucial. In industry, feature extraction of vibration signals is usually performed to evaluate the health status of equipment. However, the signals collected by the sensor contain vibration information of various components and other equipment, and are mixed with environmental noise to form coupled signals. It is difficult to directly analyze and process such vibration signals to extract effective features, which affects the judgment of equipment status. Therefore, the research on blind source separation of coupled vibration signals to obtain the source has received widespread attention.

[0003] Blind source separation is a highly potential analysis tool in signal processing. It is a signal processing method that recovers the source signal only from the observed signal when the prior information of the source signal and the propagation channel is unknown or less. It is not affected by the overlap of the time and frequency of the signal, and does not lose the characteristic information in the source signal. It has good application prospects in the state monitoring and fault diagnosis of rotating machinery and equipment. Overdetermined and positive blind source separation refers to the number of signal acquisition channels being greater than or equal to the number of source signals. The existing blind source separation methods of this type have a good effect on the separation of this type of source signals. When the number of channels is less than the number of source signals, it belongs to the underdetermined blind source separation problem. Compared with positive blind source separation, underdetermined blind source separation has two problems: the number of sources and the mixing matrix are difficult to estimate. The separation of source signals is difficult and the separation effect is not ideal. Summary of the invention

[0004] The technical problem to be solved by the present invention is: in order to solve the technical problems of the prior art for underdetermined blind source separation, such as the difficulty in separating source signals and unsatisfactory separation effects, the present invention provides a coupled vibration source separation method based on blind source separation, which reduces the separation difficulty and has a good separation effect.

[0005] The technical solution adopted by the present invention to solve the technical problem is: a coupled vibration source separation method based on blind source separation, the method comprising:

[0006] Step S1, obtaining a multi-channel signal;

[0007] Step S2, processing the multi-channel signal based on local value decomposition and adaptive singular value decomposition, and estimating the number of vibration sources in the multi-channel signal to obtain the number of information sources;

[0008] Step S3, determining a blind source separation algorithm for multi-channel signal processing based on the number of channels and the number of signal sources collected, wherein the blind source separation algorithm is one of positive blind source separation, overdetermined blind source separation, or underdetermined blind source separation;

[0009] Wherein, if the blind source separation algorithm is underdetermined blind source separation, the multi-channel signal is decomposed and the multi-channel signal is reconstructed to convert the underdetermined blind source separation into positive blind source separation;

[0010] Step S4, using fast independent component analysis to separate the multi-channel signal or the reconstructed multi-channel signal to obtain a vibration source estimation signal;

[0011] Step S5, performing cross-correlation calculation and gain compensation on the vibration source estimation signal;

[0012] Step S6: Process the vibration source estimation signal after gain compensation, set a characteristic threshold to classify the fault type, quantify the fault degree, and calculate the fault score.

[0013] Further, specifically, the step S2 includes the following steps:

[0014] S21, processing the multi-channel signal by local value decomposition to obtain a signal matrix;

[0015] S22, perform singular value decomposition on the signal matrix and calculate the proportion of each eigenvalue in all eigenvectors d i ;

[0016] S23, obtain all ratio values, perform difference operation on two adjacent ratio values, adaptively set the maximum difference value as the threshold, estimate the number of multi-channel signals, and obtain the number of signal sources.

[0017]

[0018] Further, specifically, in step S3, determining the blind source separation algorithm for multi-channel signal processing based on the number of channels and the number of signal sources for signal acquisition includes the following steps:

[0019] The number of channels for signal acquisition is ≥ the number of signal sources, and the blind source separation algorithm is positive blind source separation or overdetermined blind source separation;

[0020] The number of signal acquisition channels is less than the number of signal sources, and the blind source separation algorithm is underdetermined blind source separation.

[0021] Further, specifically, in step S3, if the blind source separation algorithm is underdetermined blind source separation, decomposing the multi-channel signal, and reconstructing the multi-channel signal, converting the underdetermined blind source separation into the positive blind source separation specifically includes the following steps:

[0022] S321, performing ensemble empirical mode decomposition on the multi-channel signal to obtain stable IMFs components;

[0023] S322, based on the correlation coefficient calculation method, calculate the correlation between the IMFs component and the multi-channel signal to obtain the source correlation coefficient λ s ,

[0024] S323, based on the number of information sources and the information source correlation coefficient λ s , retain the sensitive IMFs components, combine the multi-channel signals and the sensitive IMFs components to reconstruct a new signal, and transform the underdetermined blind source separation into positive blind source separation.

[0025] Further, specifically, the step S4 specifically includes the following steps:

[0026] Step S41, performing whitening preprocessing on the multi-channel signal x(t) or the reconstructed new signal to obtain the orthogonal quantity X, and selecting the number of components according to the number of information sources;

[0027] Step S42, randomly select an initial weight vector W p ;

[0028] Step S43: Initial weight vector W p Perform iterative calculations, where E is the identity matrix and g is the derivative of any non-quadratic function G:

[0029]

[0030] Step S44, normalization calculation:

[0031]

[0032] Step S45, p Perform fast optimization iterative calculation until convergence, otherwise repeat steps S43 to S44, and the number of iterations is not greater than the number of components.

[0033] Further, specifically, the step S5 specifically includes the following steps:

[0034] S51, calculate the frequency band w adjacent to the source estimation signal after blind source separation m and w m+1 The mutual correlation coefficient is used to determine whether the corresponding frequency bands are the same source estimation signal and solve the sorting problem. The calculation formula is:

[0035]

[0036] Among them, cov(w m ,w m+1 ) is the frequency band w mand w m+1 The covariance of m ) is the frequency band w m The variance of m+1 ) is the frequency band w m+1 The variance of |r w The larger the |, the stronger the correlation;

[0037] S52, using the LMS adaptive non-Gaussian maximization algorithm to perform gain compensation on the vibration source estimation signal, so that the vibration source estimation signal is closer to the vibration source signal, the reconstructed signal vector y(t) at time t = (y(t), y(t-1), ..., y(t-L+1)) T , using FIR filter, h(t) represents the coefficient of the adaptive filter at time t, and the expected output value is:

[0038] Y(t)=h(t)y(t).

[0039] Further, specifically, the step S21 specifically includes the following steps:

[0040] S210, obtaining all local extreme value points n of the multi-channel signal x(t) i , calculate the local extreme value mean m i and the envelope estimate a i , the calculation formula is:

[0041]

[0042] The mean m of all two adjacent local extreme values i and all adjacent envelope estimates 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;

[0043] S211, separate the local mean function m from the multi-channel signal x(t) 11 (t), the center line of the collected signal is changed to the x-axis, and the first function h is obtained 11 (t):

[0044] h 11 (t) = x(t) - m 11 (t);

[0045] S212, through the first function h 11 (t) divided by the envelope estimation function a 11 (t) Implement the first function h 11 (t) modulation, and the demodulated second function s11 (t):

[0046]

[0047] S213, repeating steps S210 and S211 to calculate the second function s 11 (t) envelope estimation function a 12 (t), if a 11 (t) = 1, the second function s 11 (t) is an ideal pure frequency modulation function, the iteration is terminated, otherwise, the next step is executed;

[0048] S214, the second function s 11 (t) as the initial signal, repeat step S213 until a 1(n+1) (t) = 1, the iteration is terminated, and the pure FM signal s is obtained. 1n (t);

[0049] S215, multiply all the envelope estimation functions to obtain an envelope signal a1(t):

[0050]

[0051] S216, multiply the envelope signal a1(t) by the pure FM signal s 1n (t), and obtain the first PF component of LMD decomposition:

[0052] PF1(t)=a1(t)s 1n (t);

[0053] S217, calculate the multi-channel signal x(t) and the first PF component PF1(t) to obtain the first signal u1(t), and repeat steps S21 to S29 based on the first signal u1(t), the number of repetitions is k times, until u1(t) is k (t) is a monotonic function, and the calculation formula is expressed as:

[0054]

[0055] S218, the multi-channel signal x(t) is completely decomposed into k PF components and one residual component u k The sum of (t) is calculated as:

[0056]

[0057] S219, composing the decomposed PF components into a signal matrix.

[0058] Further, specifically, the step S22 specifically includes: performing singular value decomposition on the signal matrix to obtain the main singular value vector S M, eliminate the zero elements of the eigenvector, form a new eigenvector, and calculate each eigenvalue λ i The proportion of all eigenvectors:

[0059]

[0060] Further, specifically, the step S321 specifically includes the following steps:

[0061] S3211, with Gaussian white noise n i (t) is added to the multi-channel signal x(t) to generate the second signal X i (t):

[0062] X i (t) = n i (t)+x(t);

[0063] S3212, the second signal X i (t) Perform empirical mode decomposition to obtain the first IMFs component, and the residual component after decomposition is r n (t), the calculation formula is:

[0064]

[0065] S3213, repeat the above steps S3211 and S3213, adding white noise of the same intensity and different sequences in each calculation process:

[0066]

[0067] S3214, multiple IMFs components are averaged to obtain a stable IMFs component, and the calculation formula is:

[0068]

[0069] Further, specifically, the step S322 specifically includes the following steps:

[0070] S3221, respectively calculate the correlation coefficient u between each multi-channel signal x(t) and its corresponding IMFs component s :

[0071]

[0072] S3222, calculate the RMS value of each IMFs component and normalize it, where N is the length of the IMFs component:

[0073]

[0074] S3223, calculate the correlation coefficient λ between the IMFs component and the multi-channel signals , the calculation formula is:

[0075]

[0076] The beneficial effect of the present invention is that the coupled vibration source separation method based on blind source separation of the present invention, under the premise of realizing the estimation of the number of vibration sources, takes this as a guide to decompose the multi-channel signal, and reconstruct the decomposed signal and the multi-channel signal, thereby solving the underdetermination problem and realizing the separation of coupled vibration sources, reducing the separation difficulty, achieving good separation effect, and improving the accuracy of multi-channel signal fault detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0078] Figure 1 It is a schematic diagram of the method flow of the optimal embodiment 1 of the present invention.

[0079] Figure 2 It is a schematic diagram of the computer device structure of the second embodiment of the present invention.

[0080] In the figure, 10 is a computer device; 1002 is a processor; 1004 is a memory; and 1006 is a transmission device. DETAILED DESCRIPTION

[0081] 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.

[0082] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In addition, features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0083] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0084] Example 1

[0085] The embodiment of the present application provides a coupled vibration source separation method based on blind source separation, such as Figure 1 As shown, the method includes:

[0086] Step S1, acquiring multi-channel signals, that is, collecting vibration signals of mechanical equipment through multiple sensors.

[0087] Step S2, processing the multi-channel signal based on local value decomposition and adaptive singular value decomposition, and estimating the number of vibration sources in the multi-channel signal to obtain the number of information sources.

[0088] Step S3, based on the number of channels and the number of signal sources for signal acquisition, determining a blind source separation algorithm for multi-channel signal processing, wherein the blind source separation algorithm is one of positive blind source separation, overdetermined blind source separation or underdetermined blind source separation.

[0089] If the blind source separation algorithm is underdetermined blind source separation, the multi-channel signal is decomposed and reconstructed to convert the underdetermined blind source separation into positive blind source separation.

[0090] Step S4, using fast independent component analysis to separate the multi-channel signal or the reconstructed multi-channel signal to obtain a vibration source estimation signal, thereby achieving decoupling and separation of the coupled vibration source signal;

[0091] Step S5, performing cross-correlation calculation and gain compensation on the vibration source estimation signal to reduce amplitude and phase distortion of the vibration source estimation signal after blind source separation.

[0092] Step S6: Process the vibration source estimation signal after gain compensation, set a characteristic threshold to classify the fault type, quantify the fault degree, and calculate the fault score.

[0093] In an embodiment, step S2 comprises the following steps:

[0094] S21, processing the multi-channel signal by local value decomposition to obtain a signal matrix; step S21 specifically includes the following steps:

[0095] S210, obtaining all local extreme value points n of the multi-channel signal x(t)i , calculate the local extreme value mean m i and the envelope estimate a i , the calculation formula is:

[0096]

[0097] The mean m of all two adjacent local extreme values i and all adjacent envelope estimates 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;

[0098] S211, separate the local mean function m from the multi-channel signal x(t) 11 (t), the center line of the collected signal is changed to the x-axis, and the first function h is obtained 11 (t):

[0099] h 11 (t) = x(t) - m 11 (t);

[0100] S212, through the first function h 11 (t) divided by the envelope estimation function a 11 (t) Implement the first function h 11 (t) modulation, and the demodulated second function s 11 (t):

[0101]

[0102] S213, repeating steps S210 and S211 to calculate the second function s 11 (t) envelope estimation function a 12 (t), if a 11 (t) = 1, the second function s 11 (t) is an ideal pure frequency modulation function, the iteration is terminated, otherwise, the next step is executed;

[0103] S214, the second function s 11 (t) as the initial signal, repeat step S213 until a 1(n+1) (t) = 1, the iteration is terminated, and the pure FM signal s is obtained. 1n (t);

[0104]

[0105] S215, multiply all the envelope estimation functions to obtain an envelope signal a1(t):

[0106]

[0107] S216, multiply the envelope signal a1(t) by the pure FM signal s 1n (t), and obtain the first PF component of LMD decomposition:

[0108] PF1(t)=a1(t)s 1n (t);

[0109] S217, calculate the multi-channel signal x(t) and the first PF component PF1(t) to obtain the first signal u1(t), and repeat steps S21 to S29 based on the first signal u1(t), the number of repetitions is k times, until u1(t) is k (t) is a monotonic function, and the calculation formula is expressed as:

[0110]

[0111] S218, the multi-channel signal x(t) is completely decomposed into k PF components and one residual component u k The sum of (t) is calculated as:

[0112]

[0113] S219, composing the decomposed PF components into a signal matrix.

[0114] S22, perform singular value decomposition on the signal matrix and calculate the proportion of each eigenvalue in all eigenvectors d i Step S22 specifically includes: performing singular value decomposition on the signal matrix to obtain the main singular value vector S M , eliminate the zero elements of the eigenvector, form a new eigenvector, and calculate each eigenvalue λ i The proportion of all eigenvectors:

[0115]

[0116] S23, obtain all ratio values, perform difference operation on two adjacent ratio values, adaptively set the maximum difference value as the threshold, estimate the number of multi-channel signals, and obtain the number of signal sources.

[0117]

[0118] In an embodiment, determining the blind source separation algorithm for multi-channel signal processing based on the number of channels and the number of signal sources collected in step S3 includes the following steps:

[0119] The number of signal acquisition channels is ≥ the number of signal sources, and the blind source separation algorithm is positive blind source separation or overdetermined blind source separation;

[0120] The number of signal acquisition channels is less than the number of signal sources, and the blind source separation algorithm is underdetermined blind source separation.

[0121] In an embodiment, if the blind source separation algorithm in step S3 is underdetermined blind source separation, decomposing the multi-channel signal and reconstructing the multi-channel signal, converting the underdetermined blind source separation into the well-determined blind source separation specifically includes the following steps:

[0122] S321, performing collective empirical mode decomposition on the multi-channel signal to obtain a stable IMFs component, thereby increasing the number of signal channels; step S321 specifically includes the following steps:

[0123] S3211, with Gaussian white noise n i (t) is added to the multi-channel signal x(t) to generate the second signal X i (t):

[0124] X i (t) = n i (t)+x(t);

[0125] S3212, the second signal X i (t) Perform empirical mode decomposition to obtain the first IMFs component, and the residual component after decomposition is r n (t), the calculation formula is:

[0126]

[0127] S3213, repeat the above steps S3211 and S3213, adding white noise of the same intensity and different sequences in each calculation process:

[0128]

[0129] S3214, multiple IMFs components are averaged to obtain a stable IMFs component, and the calculation formula is:

[0130]

[0131] Where M is the overall average number of times.

[0132] S322, based on the correlation coefficient calculation method, calculate the correlation between the IMFs component and the multi-channel signal to obtain the source correlation coefficient λ s ; Through the source correlation coefficient λ s It can help to clarify the sensitive IMFs components. Step S322 specifically includes the following steps:

[0133] S3221, respectively calculate the correlation coefficient u between each multi-channel signal x(t) and its corresponding IMFs component s :

[0134]

[0135] S3222, calculate the RMS value of each IMFs component and normalize it, where N is the length of the IMFs component:

[0136]

[0137] S3223, calculate the correlation coefficient λ between the IMFs component and the multi-channel signal s , the calculation formula is:

[0138]

[0139] S323, based on the number of information sources and the information source correlation coefficient λ s , retain the sensitive IMFs components, combine the multi-channel signals and the sensitive IMFs components to reconstruct a new signal, and transform the underdetermined blind source separation into positive blind source separation.

[0140] It should be noted that the sensitive IMFs component determination includes: calculating the source correlation coefficient λ s , sort the corresponding IMFs components from large to small, and calculate the difference between two adjacent correlation coefficients, find the index v of the maximum difference corresponding to each group, and consider the first v IMFs components to be the sensitive IMFs components.

[0141] In an embodiment, step S4 specifically includes the following steps:

[0142] Step S41, performing whitening preprocessing on the multi-channel signal x(t) or the reconstructed new signal to obtain the orthogonal quantity X, and selecting the number of components according to the number of information sources;

[0143] Step S42, randomly select an initial weight vector W p ;

[0144] Step S43: Initial weight vector W p Perform iterative calculations, where E is the identity matrix and g is the derivative of any non-quadratic function G:

[0145]

[0146] Step S44, normalization calculation:

[0147]

[0148] Step S45, pPerform fast optimization iterative calculation until convergence, otherwise repeat steps S43 to S44, and the number of iterations is not greater than the number of components.

[0149] In an embodiment, step S5 specifically includes the following steps:

[0150] S51, calculate the frequency band w adjacent to the source estimation signal after blind source separation m and w m+1 The mutual correlation coefficient is used to determine whether the corresponding frequency bands are the same source estimation signal and solve the sorting problem. The calculation formula is:

[0151]

[0152] Among them, cov(w m ,w m+1 ) is the frequency band w m and w m+1 The covariance of m ) is the frequency band w m The variance of m+1 ) is the frequency band w m+1 The variance of |r w The larger the |, the stronger the correlation;

[0153] S52, using the LMS adaptive non-Gaussian maximization algorithm to perform gain compensation on the vibration source estimation signal, so that the vibration source estimation signal is closer to the vibration source signal, the reconstructed signal vector y(t) at time t = (y(t), y(t-1), ..., y(t-L+1)) T , using FIR filter, h(t) represents the coefficient of the adaptive filter at time t, and the expected output value is:

[0154] Y(t)=h(t)y(t);

[0155] Use J(h) as the cost function to minimize:

[0156]

[0157] Where the parameter β depends on the sign of the extracted signal kurtosis, and k4(y) is the normalized kurtosis, which is defined for a zero-mean signal as:

[0158]

[0159] Use the gradient descent method to minimize the cost function J(h):

[0160]

[0161] Among them, μ(t) is a learning rate greater than 0, and the nonlinear adaptive activation function It can be expressed as:

[0162]

[0163] Real-time estimation of the second-order moment m2(Y) and the fourth-order moment m4(Y) of Y:

[0164] m q (t)=(1-η)m(t-1)+η|y(t)| q ,η∈(0,1],q=2,4;

[0165] The real-time LMS adaptive algorithm is obtained, and the calculation formula is:

[0166]

[0167] In this embodiment, the signal processing of the vibration source estimation signal after gain compensation in step S6 includes Fourier transform (TFT) or Hilbert-Huang transform (HHT), and the threshold value of the equipment operation status is set with reference to national standard documents and engineering project experience. The corresponding characteristic amplitude is calculated according to TFT and HHT, and the characteristic amplitude is compared with the set threshold value to realize fault diagnosis classification. At the same time, the fault status is scored on a percentage basis, and four states of health, availability, warning and fault are set. The equipment operation status is divided according to the score to realize quantitative evaluation of the equipment.

[0168] To sum up, the coupled vibration source separation method based on blind source separation of the present invention, under the premise of realizing the estimation of the number of vibration sources, takes this as a guide to decompose the multi-channel signal, and reconstruct the decomposed signal and the multi-channel signal, thereby solving the underdetermination problem and realizing the separation of coupled vibration sources, reducing the separation difficulty, achieving good separation effect, and improving the accuracy of multi-channel signal fault detection.

[0169] Example 2

[0170] 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 coupled vibration source separation method based on blind source separation as provided in the above method embodiment.

[0171] Figure 2 The hardware structure diagram of a device for implementing a coupled vibration source separation method based on blind source separation 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 2As 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 2 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 2 More or fewer components as shown, or with Figure 2 Different configurations are shown.

[0172] 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).

[0173] 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 coupled vibration source separation method based on blind source separation 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, implementing 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 instances, 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.

[0174] 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.

[0175] 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).

[0176] Example 3

[0177] 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 coupled vibration source separation method based on blind source separation 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 coupled vibration source separation method based on blind source separation provided in the above method embodiment.

[0178] 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.

[0179] Example 4

[0180] The embodiment of the present invention also 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 coupled vibration source separation method based on blind source separation provided in the above various optional implementations.

[0181] 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.

[0182] 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.

[0183] 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.

[0184] 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 coupled vibration source separation method based on blind source separation, characterized in that: The method comprises: Step S1, obtaining a multi-channel signal; Step S2, processing the multi-channel signal based on local value decomposition and adaptive singular value decomposition, and estimating the number of vibration sources in the multi-channel signal to obtain the number of information sources; Step S3, determining a blind source separation algorithm for multi-channel signal processing based on the number of channels and the number of signal sources collected, wherein the blind source separation algorithm is one of positive blind source separation, overdetermined blind source separation, or underdetermined blind source separation; Wherein, if the blind source separation algorithm is underdetermined blind source separation, the multi-channel signal is decomposed and the multi-channel signal is reconstructed to convert the underdetermined blind source separation into positive blind source separation; Step S4, using fast independent component analysis to separate the multi-channel signal or the reconstructed multi-channel signal to obtain a vibration source estimation signal; Step S5, performing cross-correlation calculation and gain compensation on the vibration source estimation signal; Step S6: Process the vibration source estimation signal after gain compensation, set a characteristic threshold to classify the fault type, quantify the fault degree, and calculate the fault score.

2. The coupled vibration source separation method based on blind source separation as claimed in claim 1, characterized in that: The step S2 comprises the following steps: S21, processing the multi-channel signal by local value decomposition to obtain a signal matrix; S22, perform singular value decomposition on the signal matrix and calculate the proportion of each eigenvalue in all eigenvectors d i ; S23, obtain all ratio values, perform difference operation on two adjacent ratio values, adaptively set the maximum difference value as the threshold, estimate the number of multi-channel signals, and obtain the number of signal sources.

3. The coupled vibration source separation method based on blind source separation according to claim 1, characterized in that: Determining the blind source separation algorithm for multi-channel signal processing based on the number of channels and the number of information sources of signal acquisition in step S3 includes the following steps: The number of channels for signal acquisition is ≥ the number of signal sources, and the blind source separation algorithm is positive blind source separation or overdetermined blind source separation; The number of signal acquisition channels is less than the number of signal sources, and the blind source separation algorithm is underdetermined blind source separation.

4. The coupled vibration source separation method based on blind source separation as claimed in claim 3, characterized in that: In step S3, if the blind source separation algorithm is underdetermined blind source separation, decomposing the multi-channel signal and reconstructing the multi-channel signal, converting the underdetermined blind source separation into the positive blind source separation specifically includes the following steps: S321, performing ensemble empirical mode decomposition on the multi-channel signal to obtain stable IMFs components; S322, based on the correlation coefficient calculation method, calculate the correlation between the IMFs component and the multi-channel signal to obtain the source correlation coefficient λ s , S323, based on the number of information sources and the information source correlation coefficient λ s , retain the sensitive IMFs components, combine the multi-channel signals and the sensitive IMFs components to reconstruct a new signal, and transform the underdetermined blind source separation into positive blind source separation.

5. The coupled vibration source separation method based on blind source separation according to claim 1, characterized in that: The step S4 specifically comprises the following steps: Step S41, performing whitening preprocessing on the multi-channel signal x(t) or the reconstructed new signal to obtain the orthogonal quantity X, and selecting the number of components according to the number of information sources; Step S42, randomly select an initial weight vector W p ; Step S43: Initial weight vector W p Perform iterative calculations, where E is the identity matrix and g is the derivative of any non-quadratic function G: Step S44, normalization calculation: Step S45, p Perform fast optimization iterative calculation until convergence, otherwise repeat steps S43 to S44, and the number of iterations is not greater than the number of components.

6. The coupled vibration source separation method based on blind source separation according to claim 1, characterized in that: The step S5 specifically comprises the following steps: S51, calculate the frequency band w adjacent to the source estimation signal after blind source separation m and w m+1 The mutual correlation coefficient is used to determine whether the corresponding frequency bands are the same source estimation signal. The calculation formula is: Among them, cov(w m ,w m+1 ) is the frequency band w m and w m+1 The covariance of m ) is the frequency band w m The variance of m+1 ) is the frequency band w m+1 The variance of |r w The larger the |, the stronger the correlation; S52, using the LMS adaptive non-Gaussian maximization algorithm to perform gain compensation on the vibration source estimation signal, so that the vibration source estimation signal is closer to the vibration source signal, the reconstructed signal vector y(t) at time t = (y(t), y(t-1), ..., y(t-L+1)) T , using FIR filter, h(t) represents the coefficient of the adaptive filter at time t, and the expected output value is: Y(t)=h(t)y(t).

7. The coupled vibration source separation method based on blind source separation as claimed in claim 2, characterized in that: The step S21 specifically includes the following steps: S210, obtaining all local extreme value points n of the multi-channel signal x(t) i , calculate the local extreme value mean m i and the envelope estimate a i , the calculation formula is: The mean m of all two adjacent local extreme values i and all adjacent envelope estimates 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; S211, separate the local mean function m from the multi-channel signal x(t) 11 (t), the center line of the collected signal is changed to the x-axis, and the first function h is obtained 11 (t): h 11 (t)=x(t)-m 11 (t); S212, through the first function h 11 (t) divided by the envelope estimation function a 11 (t) Implement the first function h 11 (t) modulation, and the demodulated second function s 11 (t): S213, repeating steps S210 and S211 to calculate the second function s 11 (t) envelope estimation function a 12 (t), if a 11 (t) = 1, the second function s 11 (t) is an ideal pure frequency modulation function, the iteration is terminated, otherwise, the next step is executed; S214, the second function s 11 (t) as the initial signal, repeat step S213 until a 1(n+1) (t) = 1, the iteration is terminated, and the pure FM signal s is obtained. 1n (t); S215, multiply all envelope estimation functions to obtain envelope signal a1(t): S216, multiply the envelope signal a1(t) by the pure FM signal s 1n (t), and obtain the first PF component of LMD decomposition: PF1(t)=a1(t)s 1n (t); S217, calculate the multi-channel signal x(t) and the first PF component PF1(t) to obtain the first signal u1(t), and repeat steps S21 to S29 based on the first signal u1(t), the number of repetitions is k times, until u1(t) is k (t) is a monotonic function, and the calculation formula is expressed as: S218, the multi-channel signal x(t) is completely decomposed into k PF components and one residual component u k The sum of (t) is calculated as: S219, composing the decomposed PF components into a signal matrix.

8. The coupled vibration source separation method based on blind source separation as claimed in claim 7, characterized in that: The step S22 specifically includes: performing singular value decomposition on the signal matrix to obtain the main singular value vector S M , eliminate the zero elements of the eigenvector, form a new eigenvector, and calculate each eigenvalue λ i The proportion of all eigenvectors:

9. The coupled vibration source separation method based on blind source separation as claimed in claim 4, characterized in that: The step S321 specifically includes the following steps: S3211, with Gaussian white noise n i (t) is added to the multi-channel signal x(t) to generate the second signal X i (t): X i (t)=n i (t)+x(t); S3212, the second signal X i (t) Perform empirical mode decomposition to obtain the first IMFs component, and the residual component after decomposition is r n (t), the calculation formula is: S3213, repeat the above steps S3211 and S3213, adding white noise of the same intensity and different sequences in each calculation process: S3214, multiple IMFs components are averaged to obtain a stable IMFs component, and the calculation formula is:

10. The coupled vibration source separation method based on blind source separation according to claim 4, characterized in that: The step S322 specifically includes the following steps: S3221, respectively calculate the correlation coefficient u between each multi-channel signal x(t) and its corresponding IMFs component s : S3222, calculate the RMS value of each IMFs component and normalize it, where N is the length of the IMFs component: S3223, calculate the correlation coefficient λ between the IMFs component and the multi-channel signal s , the calculation formula is:

Images