Non-parametric modal decomposition method and system for rotating machinery vibration signals

Abstract

This invention belongs to the field of signal processing technology. It proposes a parameter-free mode decomposition method and system for rotating machinery vibration signals. For the vibration signals of rotating machinery, an auxiliary noise generation process based on local fractional Brownian coloring and endpoint bridging is used. The obtained auxiliary noise is added to the vibration signal to form a noise-assisted signal. This noise-assisted signal undergoes preliminary decomposition processing to generate multiple intrinsic mode functions (EMFs). The eigenvectors of each EMF are solved, and the correlation coefficients are calculated. EMF pairs with correlation coefficients greater than a set threshold are merged to obtain the final mode decomposition result of the vibration signal. This invention effectively suppresses mode aliasing and improves the signal-to-noise ratio.

Description

Technical Field

[0001] This invention relates to the field of signal processing technology, specifically to a method and system for parametric mode decomposition of vibration signals of rotating machinery. Background Technology

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

[0003] Rotating machinery, as a core power component of industrial equipment (bearings, gears, etc.), directly affects the stability and safety of the production system. Vibration signal analysis for fault feature extraction and early diagnosis is a key technology for ensuring the reliable operation of rotating machinery. Because the vibration signals generated by rotating machinery under complex conditions such as variable loads and multi-source excitation exhibit significant nonlinear and non-stationary characteristics, traditional frequency domain analysis or statistical methods based on Fourier transform (such as root mean square value and kurtosis) are insufficient to effectively capture fault characteristics due to the assumption of linear stationarity. In recent years, adaptive mode decomposition techniques, represented by Empirical Mode Decomposition (EMD), have become the mainstream method for nonlinear and non-stationary signal processing due to their advantages of not requiring preset basis functions and adaptively matching local signal features. Building upon this, researchers have improved the mode aliasing problem of EMD by introducing auxiliary noise (such as the CEEMD algorithm), further enhancing the decomposition accuracy. However, existing mode decomposition methods still rely on manually setting key parameters (such as auxiliary noise amplitude and ensemble count), and lack targeted optimization processing of intrinsic mode functions (IMFs) after decomposition, limiting their versatility and reliability in complex vibration signal decomposition scenarios.

[0004] The existing modal decomposition techniques have the following main defects in the analysis of vibration signals of rotating machinery: (1) Strong parameter dependence and insufficient versatility; the current mainstream noise-assisted modal decomposition methods (such as CEEMD) usually use Gaussian noise as an auxiliary signal. However, the amplitude setting of Gaussian noise and the selection of the number of integrations have a decisive influence on the decomposition effect. That is, if the noise amplitude is too small, it cannot effectively improve the extreme value distribution, and if the amplitude is too large, it will introduce additional interference. Insufficient integrations will lead to insufficient decomposition, and too many integrations will significantly increase the computational cost. Due to the lack of an adaptive parameter generation mechanism, parameters need to be set through a large number of experiments or experience in engineering applications, which significantly increases the technical application threshold and limits the versatility of the method; (2) One-time integration and lack of time adaptability; the existing noise-assisted modal decomposition (such as CEEMD / CEEMDAN) usually uses a fixed amplitude and a fixed number of groups for the entire signal in a "global one-time integration". The "Cheng" strategy means that the auxiliary noise does not adjust with the non-stationary change of the original signal on the time axis. For local non-stationary periods such as speed increase / decrease and load change, this type of method cannot adaptively adjust the noise intensity and color according to the changes in local variance, extreme value density or envelope curvature, resulting in a mismatch of "excessive noise injection in smooth segments and insufficient noise injection in drastic segments"; (3) The problems of mode mixing and redundant information are prominent. Traditional mode decomposition methods (including EMD and its improved algorithms) do not perform systematic post-processing optimization of intrinsic mode functions (IMF) after decomposition. Due to the influence of nonlinear non-stationary signal characteristics, mode mixing often exists in the decomposition results (i.e., a single IMF contains cross-scale frequency components or different IMFs contain similar frequency components), which leads to the coupling of fault features and noise components and too much redundant information. This not only reduces the signal-to-noise ratio of the signal, but also increases the complexity of subsequent fault feature extraction and affects the reliability of the diagnostic results. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a parameterless mode decomposition method and system for rotating machinery vibration signals. It introduces a time-adaptive auxiliary noise preparation and injection mechanism, matching the characteristics of the auxiliary noise with the local non-stationary features of the original signal. This improves the extreme value distribution characteristics of the original signal without requiring manual parameter setting, significantly enhancing decomposition accuracy. Furthermore, by optimizing, screening, and reconstructing the intrinsic mode functions obtained from the initial decomposition, mode aliasing is effectively suppressed, redundant information is reduced, and the signal-to-noise ratio is improved.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a method for parametric mode decomposition of vibration signals of rotating machinery.

[0008] A method for parametric mode decomposition of vibration signals of rotating machinery includes the following steps:

[0009] For vibration signals of rotating machinery, auxiliary noise is generated using a noise generation process based on local fractional Brownian coloring and endpoint bridging.

[0010] The obtained auxiliary noise is added to the vibration signal to form a noise auxiliary signal. The noise auxiliary signal is then subjected to preliminary decomposition processing to generate multiple intrinsic mode functions.

[0011] The eigenvectors of the obtained multiple intrinsic mode functions are solved separately, and then the correlation coefficient of the eigenvectors is calculated. The intrinsic mode function pairs with correlation coefficients greater than a set threshold are merged to obtain the final mode decomposition result of the vibration signal.

[0012] Secondly, the present invention provides a parameterless mode decomposition system for vibration signals of rotating machinery.

[0013] A parametric mode decomposition system for vibration signals of rotating machinery includes:

[0014] The noise signal generation unit is configured to generate auxiliary noise for the vibration signal of rotating machinery using a noise generation process based on local fractional Brownian coloring and endpoint bridging.

[0015] The signal preliminary decomposition unit is configured to: add the obtained auxiliary noise to the vibration signal to form a noise auxiliary signal, perform preliminary decomposition processing on the noise auxiliary signal, and generate multiple intrinsic mode functions;

[0016] The signal mode merging unit is configured to: solve for the eigenvectors of the multiple intrinsic mode functions obtained, then calculate the correlation coefficient of the eigenvectors, and merge the intrinsic mode function pairs with a correlation coefficient greater than a set threshold to obtain the final mode decomposition result of the vibration signal.

[0017] Thirdly, the present invention provides a computer device, comprising: a processor and a computer-readable storage medium;

[0018] A processor, adapted to execute computer programs;

[0019] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method for parametric mode decomposition of rotating machinery vibration signals as described in the first aspect of this invention.

[0020] Fourthly, the present invention provides a computer-readable storage medium storing a computer program adapted to be loaded by a processor and executed as described in the first aspect of the present invention, a method for parametric mode decomposition of rotating machinery vibration signals.

[0021] Fifthly, the present invention provides a computer program product comprising a computer program that, when executed by a processor, implements the non-parametric mode decomposition method for rotating machinery vibration signals as described in the first aspect of the present invention.

[0022] Compared with the prior art, the beneficial effects of the present invention are:

[0023] This invention innovatively proposes a parameter-free mode decomposition method for rotating machinery vibration signals. Through three key steps—adaptive windowed noise injection generation, signal mode decomposition, and post-processing optimization—it systematically improves the signal decomposition accuracy, ensuring effective separation of fault features and noise, and providing a highly reliable signal foundation for subsequent fault diagnosis. A time-locally adaptive auxiliary noise preparation and injection mechanism is introduced to match the characteristics of the auxiliary noise with the local non-stationary features of the original signal, further improving decomposition accuracy. The preliminary mode decomposition algorithm can more effectively process nonlinear and non-stationary signals, achieving separation of noise components from healthy state feature components. The post-processing algorithm based on the similarity of signal filter feature vectors effectively improves the mode aliasing characteristics of empirical mode decomposition algorithms, reduces redundant information, improves the signal-to-noise ratio, and makes the processed signal easier for subsequent analysis and fault diagnosis.

[0024] This invention innovatively proposes a parameter-free mode decomposition method for rotating machinery vibration signals. From auxiliary noise generation to signal decomposition and post-processing optimization, no manual parameter setting is required at any stage. The method can automatically adjust according to the characteristics of the input signal, significantly improving its versatility and practicality and lowering the technical application threshold. The process of generating the auxiliary noise sequence using the Brownian motion principle does not require manual input of specific parameters. The auxiliary noise can change the extreme value distribution characteristics of the original signal, providing adaptive support for subsequent decomposition and other operations. The decomposition process is based on the envelope characteristics of the signal itself, and its internal mechanism determines that it can adaptively decompose according to the characteristics of the input signal. Similarly, the post-processing algorithm does not require additional parameter input when processing the intrinsic mode functions.

[0025] This invention innovatively proposes a parametric mode decomposition method for vibration signals of rotating machinery. The parametric mode decomposition method can reduce the impact of noise and redundant information on vibration signal analysis, improve signal quality and signal-to-noise ratio, and make it easier for engineers to analyze and utilize. Based on the decomposed vibration signal, more scientific maintenance strategies can be formulated, such as predictive maintenance and preventive maintenance. This helps to detect potential faults in advance, avoid the occurrence or expansion of faults, reduce operation and maintenance costs, and improve the reliability and service life of rotating machinery.

[0026] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0027] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0028] Figure 1 A flowchart illustrating a method for parametric mode decomposition of vibration signals of rotating machinery, provided as an exemplary embodiment of the present invention;

[0029] Figure 2 A simulation signal diagram of a partial fault in the sun gear of a planetary gear train is provided as an exemplary embodiment of the present invention. Figure 2 In the diagram, (a) represents the time-domain signal of the simulated gear system. Figure 2 (b) in the figure represents the Gaussian white noise component with a signal-to-noise ratio of 10 dB. Figure 2 (c) in the figure represents the random impulse noise component. Figure 2 (d) in the diagram represents a signal composed of a mixture of three components;

[0030] Figure 3 Time-domain plots of 20 sets of auxiliary noise provided for an exemplary embodiment of the present invention;

[0031] Figure 4 The decomposed modal time-domain diagram is provided as an exemplary embodiment of the present invention;

[0032] Figure 5 Provided as an exemplary embodiment of the present invention The parameter graph shows that the horizontal axis represents the parameter numbers of the 30 feature vectors, and the vertical axis represents the feature vector values.

[0033] Figure 6 A schematic diagram of the signal processing result provided in an exemplary embodiment of the present invention;

[0034] Figure 7 A comparison diagram of the optimal health state components of different signal decomposition methods provided as an exemplary embodiment of the present invention. Figure 7 (a) in the figure is a schematic diagram of the optimal health state component decomposed by the method of the present invention. Figure 7 (b) in the diagram is a schematic diagram of the optimal health status components decomposed by the CEEMDAN method. Figure 7 (c) in the diagram is a schematic diagram of the optimal health state components decomposed by the EEMD method. Figure 7 (d) in the figure is a schematic diagram of the optimal health state component decomposed by the VMD method;

[0035] Figure 8 A schematic diagram of a parametric mode decomposition system for vibration signals of rotating machinery provided as an exemplary embodiment of the present invention;

[0036] Figure 9 A schematic diagram of a computer device provided for an exemplary embodiment of the present invention. Detailed Implementation

[0037] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0038] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0039] As described in the background section, signal decomposition methods based on Empirical Mode Decomposition (EMD) and its improved algorithms (such as CEEMDAN) are widely used, but they still face two major technical bottlenecks: First, mode aliasing is prevalent, resulting in a large amount of redundant information in the decomposition results, leading to insufficient accuracy in fault feature extraction. Second, the one-time integration and lack of time adaptability mean that the auxiliary noise does not adjust to the non-stationarity of the original signal on the time axis, reducing the accuracy of signal decomposition. Third, the decomposition process relies on manually set parameters (such as auxiliary noise amplitude, integration times, etc.), and the subjectivity and cumbersomeness of parameter selection significantly reduce the method's versatility and raise the threshold for engineering applications. These problems make it difficult for existing methods to meet the requirements of high-precision and adaptive decomposition of rotating machinery vibration signals, thus restricting the reliability of subsequent fault diagnosis. In view of this, this implementation proposes a parameter-free mode decomposition method for rotating machinery vibration signals. By integrating the auxiliary noise generation mechanism, the decomposition process, and post-processing of the results, it overcomes the phenomenon of mode aliasing in signal decomposition and achieves accurate decomposition of rotating machinery vibration signals without the need for any parameter input.

[0040] Specifically, such as Figure 1 As shown, the process includes the following:

[0041] S1: Auxiliary noise preparation process, for the collected vibration signals ( (where n is the data length) 20 sets of auxiliary noise sequences were generated using a noise generation process based on local fractional Brownian coloring and endpoint bridging. ,in, ; ;

[0042] S2: Noise-assisted signal decomposition process. The CEEMD algorithm is used to process the acquired vibration signal and auxiliary noise sequence to generate... Individual eigenmode functions ,in, , ;

[0043] S3: Post-processing of the decomposition results, using a post-processing algorithm based on the similarity of signal filter feature vectors. Individual eigenmode functions Processing, generating Individual eigenmode functions ,in, , .

[0044] In S1 of this implementation, the sampling frequency is set to 12.8 kHz and the duration is 4 seconds during the simulation. Figure 2 The time-domain signal of the sun gear in a simulated planetary gear system is shown. Figure 2 (a) shows the time-domain signal of the simulated gear system, where the meshing frequency and periodic pulse components are clearly visible; Figure 2 (b) shows the Gaussian white noise component with a signal-to-noise ratio of 10 dB; Figure 2 (c) in the figure describes the random impulse noise component; Figure 2 (d) in the diagram shows the signal composed of these three components, which will be used in subsequent analyses to evaluate the effectiveness of different signal decomposition methods.

[0045] right Figure 2 Vibration signal in (d) Twenty auxiliary noise sequences were generated using a noise generation process based on local fractional Brownian coloring and endpoint bridging. The specific steps in this section include:

[0046] S101: Online estimation of local statistics.

[0047] according to Constructing a sliding window ( For window length, For the first (initial position of each time window), stride Calculate the local variance within each window:

[0048] (1);

[0049] in, This represents the local variance of the vibration signal within the k-th sliding window; Represents a time index.

[0050] Calculate the extreme density:

[0051] (2);

[0052] in, The number of extreme values, after normalization, is:

[0053] (3).

[0054] in, Represents the normalized extreme value density, Represents the minimum value of the extreme density. This represents the maximum value of the extreme density.

[0055] Calculate envelope curvature From the top / bottom envelope (Hilbert envelope or spline envelope) calculation:

[0056] (4);

[0057] Further calculation of normalized envelope curvature :

[0058] (5).

[0059] in, This represents the minimum value of the envelope curvature. This represents the maximum value of the envelope curvature.

[0060] The Hurst exponent is calculated, and detrended volatility analysis is used to estimate the Hurst exponent in each window. .

[0061] S102: White noise generation and spectrum calculation.

[0062] For each group ( (Total number of groups, preferably 20) Generate white noise ( It follows a normal distribution with variance of 1 and mean of 0, and its length is... ).

[0063] Calculate the first The discrete frequency grid is:

[0064] ( (6).

[0065] Calculate the first Group white noise complex spectrum :

[0066] (7);

[0067] in, This refers to the Discrete Fourier Transform (DFT).

[0068] S103: Local fractional Brownian coloring.

[0069] In the window Within this process, amplitude shaping is performed on the white noise spectrum to obtain a fractional Brownian target spectrum.

[0070] (8);

[0071] in, For the first Groups, Windows Target complex spectrum (preservation) (The phase and amplitude are shaped). This refers to the frequency amplitude. To prevent The tiny "anti-shake frequency" that diverges at the point (e.g.) ); For window Hurst exponent estimation; This is the normalization constant.

[0072] Normalization constant Used to control energy or variance within a window, calculated according to the following formula:

[0073] (9);

[0074] After shaping the amplitude of each window Performing inverse Fourier transform yields Then, overlapping weighted splicing was used:

[0075] (10);

[0076] in, For smooth windows (such as Hanning windows). Representing the The group is based on the noise processed by the Hurst exponent H.

[0077] S104: Inverse Transformation and Endpoint Bridging.

[0078] Will Transform to the time domain and perform bridging Brownian processing to reduce endpoint effects.

[0079] For sampling points :

[0080] (11);

[0081] in, This represents the continuous time corresponding to the nth sampling point; Represents the sampling time interval; This represents the discrete time-domain value of the i-th noise group at the n-th sampling point; This represents the discrete-time value of the i-th noise group at the N-th sampling point. represents the discrete time-domain value of the i-th noise group at the n-th sampling point after bridging Brownian processing; N represents the total number of auxiliary noise sampling points; n represents the n-th sampling point.

[0082] S105: Amplitude self-matching.

[0083] Calculate the amplitude self-matching factor to make the local energy of the noise equivalent to that of the target signal:

[0084] (12);

[0085] in, This is the amplitude self-matching factor; For window Inside Local standard deviation (unbiased estimate); For the window Local standard deviation; To prevent small constants with zero denominators (such as...) ), This represents the continuous time-domain signal of the i-th group of noise after bridging Brownian processing.

[0086] S106: Construct positional weights to enhance noise injection in aliasing-sensitive regions (high extrema / high curvature) and suppress noise injection in smooth regions.

[0087] (13)

[0088] (14);

[0089] in, The position weights before and after mapping; These are the mapped position weights; , which is the weighting coefficient between extreme density and curvature; This represents the lower / upper limit of the weight.

[0090] S107: After matching the amplitude of S105 The positional weights of S106 are applied to the bridging noise to obtain the localized noise injection sequence:

[0091] (15);

[0092] in, For the first Group of locally injected noise sequences.

[0093] Time-domain images such as Figure 3As shown, from left to right and from top to bottom, they are: , ... The noise is represented by time (s) on the x-axis of each image, and the auxiliary noise is represented by amplitude (s / m) on the y-axis. 2 ).

[0094] Specifically, S2 of this implementation includes:

[0095] S201: Define the operator For a given signal, the Empirical Mode Decomposition (EMD) method is used to generate the first... 1st mode.

[0096] S202: For simulation signals Add 20 fractional Gaussian noises to create 20 noise-assisted signals. :

[0097] (16);

[0098] In the formula, For the first Sub-integration , The standard deviation is denoted as .

[0099] S203: Decompose noise-assisted signals using EMD To obtain their first-order modes And calculate:

[0100] (17);

[0101] S204: Calculate the first residual in the first stage (m=1):

[0102] (18);

[0103] S205: Decomposition via EMD And define the second mode as:

[0104] (19);

[0105] S206: For m=2,…,M, calculate the m-th residual:

[0106] (20);

[0107] S207: Decomposition , And define the (m+1)th mode as:

[0108] (twenty one);

[0109] S208: Jump to S206 and calculate the next mode until the residual no longer has at least two extrema. The final residual is:

[0110] (twenty two);

[0111] In the formula, It is the total number of modes. Therefore, given a signal It can be represented as:

[0112] (twenty three);

[0113] Decomposition results are as follows Figure 4 As shown, it includes IMF1, IMF2...IMF 15 IMF 16 Of the 16 modes, it can be seen that the random noise component in the signal is concentrated in modes IMF1-IMF6, the fault characteristic component is concentrated in IMF7 and IMF8, and the impulse noise component is concentrated in IMF9-IMF6. 14 In this study, the separation of fault diagnosis signals and noise was effective.

[0114] Specifically, S103 of this implementation includes:

[0115] S301: This process utilizes the principle of a 30-parameter filter to calculate the eigenvectors corresponding to different IMFs, and calculates the original signal. delay matrix , It can be represented as:

[0116] (twenty four);

[0117] S302: Calculation delay vector , Represented as:

[0118] (25);

[0119] S303: Calculation eigenvectors , Represented as:

[0120] (26);

[0121] To more intuitively illustrate the differences in feature vectors corresponding to different IMFs, The parameter graph is shown in Figure 5 From Figure 5 It can be seen that the feature vectors of the health feature components IMF7 and IMF8 have similar morphological features.

[0122] S304: Define an empty list This is used to store the merged IMF. Create a flag array. (Length is 16, representing 16 modes), used to track whether each IMF has participated in the consolidation.

[0123] S305: Each To merge, follow these steps:

[0124] (1) Starting from m=1, check Has it been processed, i.e., has it been satisfied? If it has already been processed, skip this step. .

[0125] (2) Initialize a new merge group and will Set to True.

[0126] (3) For each unprocessed (j>m), calculate its eigenvector. and Correlation coefficient between The formula for calculating the correlation coefficient is as follows:

[0127] (27);

[0128] (4) If the correlation coefficient If >0.95, then Add to In, and will Setting it to True indicates It has been merged.

[0129] S306: When a group of IMFs (e.g., { , When the correlation coefficient threshold r > 0.95 is met, perform the following operations:

[0130] (28);

[0131] In the formula, where, This is the result of the current merge group. It includes all merged IMFs.

[0132] S307: After processing the current merged group, continue to calculate the correlation coefficient for the next unprocessed IMF, and repeat S305 and S306 until all IMFs have participated in the processing. The merging results are shown in Table 1. The key health status feature components IMF7 and IMF8 were merged, which reduced the mode aliasing of the decomposition results and enhanced the signal features.

[0133] Table 1: Correlation coefficients of merged IMFs and their corresponding feature vectors

[0134]

[0135] S307: All IMFs are involved in the process. ( The final signal processing result is as follows: Figure 6 As shown, CIMF7 is a fault characteristic component in the original signal, almost identical to... Figure 2 The simulated fault signal in (a) is consistent with that in the present invention, indicating that the signal processing method proposed in this invention can accurately decompose the vibration signal and recover the fault characteristic components in the noisy vibration signal. The optimal healthy state component decomposed by the method of this invention is compared in the time domain with the optimal healthy state components decomposed by other classic signal decomposition methods (CEEMDAN, EEMD, VMD), as shown below. Figure 7 As shown, obvious signal distortion in the time-domain plot is marked with dashed rectangles. Figure 7 In the figure, (a) represents the optimal health state component decomposed by the method of the present invention. Figure 7 In the diagram, (b) represents the optimal health state component decomposed by the CEEMDAN method. Figure 7 In the diagram, (c) represents the optimal health state component decomposed by the EEMD method. Figure 7 In the figure, (d) represents the optimal health state component decomposed by the VMD method. It can be clearly seen from the figure that, compared with other signal decomposition methods used previously, the signal features decomposed by the method of the present invention are clearer, noise interference is significantly reduced, and the separation effect of signal and noise is significantly improved.

[0136] Figure 8 A parametric mode decomposition system for vibration signals of rotating machinery is shown, comprising:

[0137] The noise signal generation unit 801 is configured to generate auxiliary noise for the vibration signal of rotating machinery using a noise generation process based on local fractional Brownian coloring and endpoint bridging.

[0138] The signal preliminary decomposition unit 802 is configured to: add the obtained auxiliary noise to the vibration signal to form a noise auxiliary signal, perform preliminary decomposition processing on the noise auxiliary signal, and generate multiple intrinsic mode functions;

[0139] The signal mode merging unit 803 is configured to: solve the eigenvectors of the multiple obtained intrinsic mode functions respectively, then calculate the correlation coefficient of the eigenvectors, merge the intrinsic mode function pairs with the correlation coefficient greater than a set threshold, and obtain the final mode decomposition result of the vibration signal.

[0140] It is understood that the aforementioned units can be individually or entirely merged into one or more other units, or some of the units can be further divided into multiple functionally smaller units. This achieves the same operation without affecting the technical effects of the embodiments of this application. The aforementioned units are based on logical functional division. In practical applications, the function of one unit can be implemented by multiple units, or the function of multiple units can be implemented by one unit. In other embodiments of this application, the system may also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and can be implemented collaboratively by multiple units.

[0141] According to another embodiment of this application, the system of this embodiment can be constructed by running a computer program (including program code) capable of performing the steps involved in the corresponding method of the present invention on a general-purpose computing device, such as a computer, which includes processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM). The computer program can be recorded on, for example, a computer-readable recording medium, loaded into the aforementioned computing device through the computer-readable recording medium, and run therein.

[0142] Figure 9 A computer device is shown, which includes a processor 1001, a communication interface 1002, and a computer-readable storage medium 903. The processor 901, communication interface 902, and computer-readable storage medium 903 can be connected via a bus or other means.

[0143] The communication interface 902 is used to receive and send data. The computer-readable storage medium 903 can be stored in the memory of the electronic device. The computer-readable storage medium 903 is used to store computer programs, which include program instructions. The processor 901 is used to execute the program instructions stored in the computer-readable storage medium 903.

[0144] The processor 901 is the computing and control core of electronic devices. It is suitable for implementing one or more instructions, specifically for loading and executing one or more instructions to achieve corresponding methods or functions.

[0145] Processor 901 is configured to perform the following procedure:

[0146] For vibration signals of rotating machinery, auxiliary noise is generated using a noise generation process based on local fractional Brownian coloring and endpoint bridging.

[0147] The obtained auxiliary noise is added to the vibration signal to form a noise auxiliary signal. The noise auxiliary signal is then subjected to preliminary decomposition processing to generate multiple intrinsic mode functions.

[0148] The correlation coefficients of the obtained intrinsic mode functions are calculated separately, and the intrinsic mode function pairs with correlation coefficients greater than a set threshold are combined to obtain the final mode decomposition result of the vibration signal.

[0149] This invention also provides a computer-readable storage medium, which is a memory device in an electronic device for storing programs and data. It is understood that the computer-readable storage medium here may include both built-in storage media in the electronic device and extended storage media supported by the electronic device. The computer-readable storage medium provides storage space for storing the processing system of the electronic device.

[0150] Furthermore, this storage space also contains one or more instructions suitable for loading and execution by the processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM memory or unstable memory, such as at least one disk storage device; optionally, it can also be at least one computer-readable storage medium located remotely from the aforementioned processor.

[0151] In one embodiment, the computer-readable storage medium stores one or more instructions; the processor loads and executes the one or more instructions stored in the computer-readable storage medium to perform the following process:

[0152] For vibration signals of rotating machinery, auxiliary noise is generated using a noise generation process based on local fractional Brownian coloring and endpoint bridging.

[0153] The obtained auxiliary noise is added to the vibration signal to form a noise auxiliary signal. The noise auxiliary signal is then subjected to preliminary decomposition processing to generate multiple intrinsic mode functions.

[0154] The correlation coefficients of the obtained intrinsic mode functions are calculated separately, and the intrinsic mode function pairs with correlation coefficients greater than a set threshold are combined to obtain the final mode decomposition result of the vibration signal.

[0155] The present invention also provides a computer program product or computer program comprising computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the following process:

[0156] For vibration signals of rotating machinery, auxiliary noise is generated using a noise generation process based on local fractional Brownian coloring and endpoint bridging.

[0157] The obtained auxiliary noise is added to the vibration signal to form a noise auxiliary signal. The noise auxiliary signal is then subjected to preliminary decomposition processing to generate multiple intrinsic mode functions.

[0158] The correlation coefficients of the obtained intrinsic mode functions are calculated separately, and the intrinsic mode function pairs with correlation coefficients greater than a set threshold are combined to obtain the final mode decomposition result of the vibration signal.

[0159] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0160] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic cable, digital cable) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0161] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for parametric mode decomposition of vibration signals of rotating machinery, characterized in that, The process includes the following: For vibration signals of rotating machinery, auxiliary noise is generated using a noise generation process based on local fractional Brownian coloring and endpoint bridging. The obtained auxiliary noise is added to the vibration signal to form a noise auxiliary signal. The noise auxiliary signal is then subjected to preliminary decomposition processing to generate multiple intrinsic mode functions. The eigenvectors of the multiple intrinsic mode functions are solved separately, and then the correlation coefficient of the eigenvectors is calculated. The pairs of intrinsic mode functions with a correlation coefficient greater than a set threshold are merged to obtain the final mode decomposition result of the vibration signal. Auxiliary noise is generated using a noise generation process based on local fractional Brownian coloring and endpoint bridging, including: A sliding window with a fixed stride is constructed for the acquired vibration signal, and the normalized extreme value density, normalized envelope curvature, Hurst exponent, and local standard deviation within each sliding window are calculated. Multiple sets of white noise with a variance of 1 and a mean of 0 are generated. For each set of white noise, its corresponding discrete frequency grid is calculated. Then, based on the discrete frequency grid, the complex spectrum of each set of white noise is solved by discrete Fourier transform. Within each sliding window, the complex spectrum of the white noise is amplitude shaped by combining the frequency amplitude of the window, the preset anti-shake frequency, the Hurst exponent, and the preset normalization constant, so that the spectrum exhibits fractional Brownian characteristics, thus obtaining the fractional Brownian target spectrum of each group of white noise within the corresponding window. An inverse Fourier transform is performed on the fractional Brownian target spectrum after amplitude shaping within each window. The transform results are then integrated into a continuous signal using a smooth window overlapping weighted splicing method. The spliced ​​signal is then converted to the time domain. The signal endpoint effect is reduced by bridging Brownian processing to obtain the noise of each group after bridging. The local standard deviation of the noise of each group after bridging within the corresponding sliding window is calculated. Based on the local standard deviation of the vibration signal and the local standard deviation of the bridged noise, combined with a preset small constant, the amplitude self-matching factor is solved to make the bridged noise consistent with the local energy of the vibration signal within the corresponding window. We set weighting coefficients for extreme density and curvature, and upper and lower limits for position weights. Based on normalized extreme density and normalized envelope curvature, we calculate the position weights before mapping by weighted fusion, and then normalize the position weights before mapping to obtain the position weights after mapping. The amplitude self-matching factor and the mapped position weight are applied together to the bridged noise to generate each set of local noise injection sequences, which serve as auxiliary noise to match the local non-stationary characteristics of the original vibration signal.

2. The method for parametric mode decomposition of rotating machinery vibration signals as described in claim 1, characterized in that, The number of extreme values ​​is counted within each sliding window. The extreme value density is calculated based on the number of extreme values, and the extreme value density is normalized to obtain the normalized extreme value density. Solve for the upper and lower envelopes of the vibration signal within each sliding window, calculate the envelope curvature based on the envelope, and normalize the envelope curvature to obtain the normalized envelope curvature; The Hurst exponent is estimated within each sliding window using a detrended fluctuation analysis method, and the local standard deviation of the vibration signal within each sliding window is calculated simultaneously.

3. The method for parametric mode decomposition of rotating machinery vibration signals as described in claim 1, characterized in that, The window length of the sliding window is determined based on the sampling frequency of the vibration signal and the basic rotation frequency of the rotating machinery. The step size of the sliding window is set to 1 / 4 to 1 / 2 of the window length. The total number of auxiliary noise groups is set to 20 groups, and the length of each auxiliary noise group is consistent with the length of the vibration signal. In the estimation process of the Hurst exponent, the scale range of the detrended fluctuation analysis is set to 1 / 10 to 1 / 5 of the length of the original vibration signal data.

4. The method for parametric mode decomposition of rotating machinery vibration signals as described in claim 1, characterized in that, The noise-assisted signal undergoes preliminary decomposition processing to generate multiple intrinsic mode functions, including the following steps: Select all generated auxiliary noise, superimpose each group of auxiliary noise with the vibration signal, and adjust the amplitude based on the standard deviation of each group of auxiliary noise to construct multiple groups of noise auxiliary signals; The empirical mode decomposition algorithm is used to decompose each group of noise-assisted signals and extract the first-order intrinsic mode function of each group of noise-assisted signals. Calculate the mean of all first-order intrinsic mode functions, take this mean as the first intrinsic mode function, and then subtract this first intrinsic mode function from the original vibration signal to obtain the first residual. The first residual is decomposed using the empirical mode decomposition algorithm, and the first-order mode in the decomposition result is extracted as the second intrinsic mode function. For the m-th residual, where m starts from 2 and increases sequentially, the current m-th residual is obtained by subtracting the corresponding m-th eigenmode function from the previous residual. The m-th residual is decomposed using the empirical mode decomposition algorithm, and the first mode in the decomposition result is extracted as the (m+1)-th eigenmode function. Continue calculating the next-order residual and its corresponding intrinsic mode function until the current residual no longer has at least two extrema, then stop the decomposition. Collect all the intrinsic mode functions obtained from the decomposition and the final residuals to form a set of multiple intrinsic mode functions after the initial decomposition of the original vibration signal.

5. The method for parametric mode decomposition of rotating machinery vibration signals as described in claim 1, characterized in that, Calculate the delay matrix of the vibration signal. Based on the delay matrix, calculate the delay vector of each intrinsic mode function after preliminary decomposition. Calculate the eigenvector of each intrinsic mode function based on the delay vector. Calculate the correlation coefficient based on the eigenvector. Combine intrinsic mode function pairs with correlation coefficients greater than a set threshold.

6. A parameter-free mode decomposition system for vibration signals of rotating machinery, characterized in that, include: The noise signal generation unit is configured to generate auxiliary noise for the vibration signal of rotating machinery using a noise generation process based on local fractional Brownian coloring and endpoint bridging. The signal preliminary decomposition unit is configured to: add the obtained auxiliary noise to the vibration signal to form a noise auxiliary signal, perform preliminary decomposition processing on the noise auxiliary signal, and generate multiple intrinsic mode functions; The signal mode merging unit is configured to: solve the eigenvectors of the multiple obtained intrinsic mode functions respectively, then calculate the correlation coefficient of the eigenvectors, merge the intrinsic mode function pairs with the correlation coefficient greater than a set threshold, and obtain the final mode decomposition result of the vibration signal; Auxiliary noise is generated using a noise generation process based on local fractional Brownian coloring and endpoint bridging, including: A sliding window with a fixed stride is constructed for the acquired vibration signal, and the normalized extreme value density, normalized envelope curvature, Hurst exponent, and local standard deviation within each sliding window are calculated. Multiple sets of white noise with a variance of 1 and a mean of 0 are generated. For each set of white noise, its corresponding discrete frequency grid is calculated. Then, based on the discrete frequency grid, the complex spectrum of each set of white noise is solved by discrete Fourier transform. Within each sliding window, the complex spectrum of the white noise is amplitude shaped by combining the frequency amplitude of the window, the preset anti-shake frequency, the Hurst exponent, and the preset normalization constant, so that the spectrum exhibits fractional Brownian characteristics, thus obtaining the fractional Brownian target spectrum of each group of white noise within the corresponding window. An inverse Fourier transform is performed on the fractional Brownian target spectrum after amplitude shaping within each window. The transform results are then integrated into a continuous signal using a smooth window overlapping weighted splicing method. The spliced ​​signal is then converted to the time domain. The signal endpoint effect is reduced by bridging Brownian processing to obtain the noise of each group after bridging. The local standard deviation of the noise of each group after bridging within the corresponding sliding window is calculated. Based on the local standard deviation of the vibration signal and the local standard deviation of the bridged noise, combined with a preset small constant, the amplitude self-matching factor is solved to make the bridged noise consistent with the local energy of the vibration signal within the corresponding window. We set weighting coefficients for extreme density and curvature, and upper and lower limits for position weights. Based on normalized extreme density and normalized envelope curvature, we calculate the position weights before mapping by weighted fusion, and then normalize the position weights before mapping to obtain the position weights after mapping. The amplitude self-matching factor and the mapped position weight are applied together to the bridged noise to generate each set of local noise injection sequences, which serve as auxiliary noise to match the local non-stationary characteristics of the original vibration signal.

7. A computer device, characterized in that, include: Processor and computer-readable storage media; A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the parametric mode decomposition method for rotating machinery vibration signals as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed as described in any one of claims 1 to 5, the method for parametric mode decomposition of rotating machinery vibration signals.

9. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the parametric mode decomposition method for rotating machinery vibration signals as described in any one of claims 1 to 5.

Images