Micro-vibration data processing method and device based on wiener filter

Abstract

The application discloses a kind of microvibration data processing method and device based on wiener filter, the method comprises: obtaining the microvibration original signal to be processed;The microvibration original signal is input into the wiener filter of pre-training, to obtain the denoising signal output by the wiener filter;In the case where the microvibration original signal exists zero point drift risk, the microvibration original signal is reconstructed using wavelet decomposition algorithm, to obtain the effective signal after removing zero point drift;Wherein, the wiener filter is trained using the original signal of sample and the estimated signal obtained after denoising the original signal of sample.In the analysis process of microvibration test data, the method and device can effectively denoise and remove zero drift to the signal, so as to improve the signal cleanliness, ensure the accuracy of simulation analysis.

Description

Technical Field

[0001] This invention relates to the field of micro-vibration data processing technology, and specifically to a micro-vibration data processing method and apparatus based on Wiener filters. Background Technology

[0002] Remote sensing satellites carry sensitive payloads that require high operational accuracy. Due to the satellite's complex structure and numerous flexible components, it often experiences wide-bandwidth, small-amplitude vibrations during on-orbit operation. Therefore, engineers need to effectively analyze and study the vibration sources, vibration transmission, and the impact of vibration on the payloads. In existing technologies, to analyze satellite vibration data, micro-vibration tests are typically conducted on the ground using simulated raw data. Therefore, during the analysis of micro-vibration test data, how to effectively denoise and remove zero drift from the signal to improve signal cleanliness and ensure the accuracy of the simulation analysis has become a pressing problem for those skilled in the art. Summary of the Invention

[0003] Therefore, embodiments of the present invention provide a micro-vibration data processing method and apparatus based on Wiener filters, which can effectively denoise and remove zero drift from signals during the analysis of micro-vibration test data, thereby improving signal cleanliness and ensuring the accuracy of simulation analysis.

[0004] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0005] This invention provides a method for processing micro-vibration data based on a Wiener filter, the method comprising:

[0006] Acquire the raw micro-vibration signal to be processed;

[0007] The original micro-vibration signal is input into a pre-trained Wiener filter to obtain the denoised signal output by the Wiener filter; in the case that the original micro-vibration signal has the risk of zero-point drift, the original micro-vibration signal is reconstructed using a wavelet decomposition algorithm to obtain an effective signal after removing the zero-point drift.

[0008] The Wiener filter is trained using the original signal of the sample and the estimated signal obtained after denoising the original signal of the sample.

[0009] In some embodiments, the Wiener filter is obtained by training using the original signal of the sample and the estimated signal obtained after denoising the original signal of the sample, specifically including:

[0010] Based on the original signal of the sample, the estimate of the clean signal obtained after denoising is calculated according to the preset objective function;

[0011] At each frequency point, the mean square error between the estimate of the clean signal and the clean signal is calculated;

[0012] The Wiener filter is obtained by training with the goal of minimizing the mean square error.

[0013] In some embodiments, the estimated signal obtained after denoising the original signal of the sample specifically includes:

[0014] The original signal of the sample is preprocessed;

[0015] The preprocessed signal is subjected to a short-time Fourier transform to obtain the decomposed amplitude spectrum and phase spectrum.

[0016] Extract at least one frame of each noise signal from the original signal of the sample as the target signal, and perform a Fourier transform on the target signal;

[0017] The original signal of the sample was subjected to spectral subtraction using oversubtraction to obtain the estimated signal after denoising.

[0018] In some embodiments, the energy spectrum is estimated using the oversubtraction method as follows:

[0019]

[0020] Where Y(ω) is the original signal of the sample, P Y (ω) represents the energy spectrum of Y(ω), E(ω) represents the frequency domain form of the noise signal of the sample, and P E (ω) is the energy spectrum of E(ω), and α and γ are adjustable coefficients in the oversubtraction method. When γ = 1, it is equivalent to energy spectrum subtraction, and when γ = 0.5, it is equivalent to amplitude spectrum subtraction. The larger the value of α, the greater the degree of noise attenuation and the greater the impact on the original signal.

[0021] In some embodiments, the original micro-vibration signal is reconstructed using a wavelet decomposition algorithm to obtain an effective signal after removing zero-point drift, specifically including:

[0022] Determine the initial fluctuation position of the original micro-vibration signal;

[0023] Discrete wavelet decomposition is performed on the signal after the initial fluctuation position in the original micro-vibration signal according to the preset number of decomposition layers.

[0024] Based on the results of wavelet decomposition, it is determined that the original micro-vibration signal has zero-point drift. If so, the original micro-vibration signal is reconstructed to obtain an effective signal after removing the zero-point drift.

[0025] In some embodiments, determining the presence of zero-point drift in the original micro-vibration signal based on the wavelet decomposition result specifically includes:

[0026] Calculate the mean of the approximate values ​​of the decomposed signal. If the mean is greater than a threshold, it is determined that the original micro-vibration signal has zero-point drift.

[0027] In some embodiments, the original micro-vibration signal is reconstructed to obtain an effective signal after removing zero-point drift, and the process further includes:

[0028] The drift effect is judged based on the generated impact response spectrum.

[0029] The present invention also provides a micro-vibration data processing device based on a Wiener filter, the device comprising:

[0030] The signal acquisition unit is used to acquire the original micro-vibration signal to be processed.

[0031] The signal processing unit is used to input the original micro-vibration signal into a pre-trained Wiener filter to obtain a denoised signal output by the Wiener filter; and, in the case that the original micro-vibration signal has the risk of zero-point drift, to reconstruct the original micro-vibration signal using a wavelet decomposition algorithm to obtain an effective signal after removing the zero-point drift.

[0032] The Wiener filter is trained using the original signal of the sample and the estimated signal obtained after denoising the original signal of the sample.

[0033] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method described above.

[0034] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0035] The present invention provides a method and apparatus for micro-vibration data processing based on a Wiener filter. By acquiring the original micro-vibration signal to be processed and inputting it into a pre-trained Wiener filter, a denoised signal output by the Wiener filter can be obtained. If the original micro-vibration signal has a risk of zero-point drift, a wavelet decomposition algorithm is used to reconstruct the original signal to obtain an effective signal after removing the zero-point drift. The method and apparatus provided by the present invention extract weak signals under strong background noise using a trained Wiener filter, and determine whether the signal has zero-point drift using wavelet decomposition, removing any existing zero-point drift. Compared with traditional manually designed filters, using a Wiener filter can achieve adaptive and accurate filtering, thereby more accurately separating the real signal from background noise. In the analysis of micro-vibration test data, effective denoising and zero-drift removal of the signal are achieved, thereby improving signal cleanliness and ensuring the accuracy of simulation analysis.

[0036] Furthermore, the method and apparatus provided by this invention can capture the time-frequency characteristics of a signal more accurately using wavelet decomposition, and can more automatically distinguish the existence of zero drift, thereby achieving a better drift removal effect, making it easier for experimenters to obtain more reliable experimental data, and thus more accurately analyze the vibration source and vibration transmission characteristics. Attached Figure Description

[0037] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0038] The structures, proportions, sizes, etc. illustrated in this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed herein, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0039] Figure 1 This is one of the flowcharts for the Wiener filter-based micro-vibration data processing method provided by the present invention;

[0040] Figure 2 The second flowchart is a micro-vibration data processing method based on Wiener filter provided by the present invention.

[0041] Figure 3A time-domain comparison image before and after background noise removal;

[0042] Figure 4 Frequency domain comparison before and after background noise removal;

[0043] Figure 5 Comparison of time domain before and after zero-point drift removal;

[0044] Figure 6 Comparison of impact response spectra before and after zero-point drift removal.

[0045] Figure 7 This is a structural block diagram of the Wiener filter-based micro-vibration data processing device provided by the present invention.

[0046] Figure 8 This is a structural block diagram of a computer device provided by the present invention. Detailed Implementation

[0047] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] Please refer to Figure 1 , Figure 1 This is one of the flowcharts for the Wiener filter-based micro-vibration data processing method provided by the present invention.

[0049] In one specific embodiment, the micro-vibration data processing method based on Wiener filter provided by the present invention includes the following steps:

[0050] S110: Acquire the raw micro-vibration signal to be processed;

[0051] S120: Input the original micro-vibration signal into a pre-trained Wiener filter to obtain the denoised signal output by the Wiener filter; if the original micro-vibration signal has the risk of zero-point drift, use wavelet decomposition algorithm to reconstruct the original micro-vibration signal to obtain the effective signal after removing the zero-point drift.

[0052] The Wiener filter is trained using the original signal of the sample and the estimated signal obtained after denoising the original signal of the sample.

[0053] Theoretically, Wiener filters are widely used in speech enhancement as an adaptive filter for removing colored noise. Wavelet decomposition, a signal decomposition method inherited from Fourier analysis, is more suitable for non-stationary signals obtained from actual experiments. Therefore, this invention applies Wiener filters and wavelet decomposition to the data processing of micro-vibration experiments, achieving weak signal extraction under strong noise and making the impact response spectrum of the data more consistent with reality.

[0054] In some embodiments, the Wiener filter is obtained by training using the original signal of the sample and the estimated signal obtained after denoising the original signal of the sample, specifically including:

[0055] Based on the original signal of the sample, the estimate of the clean signal obtained after denoising is calculated according to the preset objective function;

[0056] At each frequency point, the mean square error between the estimate of the clean signal and the clean signal is calculated;

[0057] The Wiener filter is obtained by training with the goal of minimizing the mean square error.

[0058] Specifically, the training process of the Wiener filter is as follows:

[0059] Both the original signal of the sample and the original signal of the micro-vibration to be processed are noisy signals, denoted as the noisy signal Y(ω). The design objective of the Wiener filter is:

[0060]

[0061] in, For the estimation of the clean signal, H(ω) is the target filter.

[0062] The mean square error between the estimated clean signal and the clean signal at each frequency point is:

[0063]

[0064] Where, e(ω) k Let X(ω) be the mean square error between the estimate of the clean signal and the clean signal, and let X(ω) be the clean signal. k This is the kth frequency point.

[0065] The design goal of H(ω) is to make e(ω) k The expected value of ) is minimized, e(ω) k The expectation is:

[0066]

[0067] Here, function E is the expected function of the corresponding parameter, and the parameter marked with * is the conjugate of the corresponding parameter.

[0068] For convenience, E[|Y(ω)] will be used below. k )| 2 ] is denoted as P yy (ω k ), E[X * (ω k )Y(ω k )] is denoted as P yx (ω k ), for E[|e(ω k )| 2 Differentiation yields:

[0069]

[0070] The derivative reaches its minimum value when it is 0.

[0071]

[0072] Where H(ω) k ) represents the value of the target filter at the k-th frequency point.

[0073] because:

[0074] P yx * (ω k )=E[X(ω k )Y(ω k )]=E[X(ω k ){X(ω k )+N(ω k )} * ]

[0075] =E[X(ω) k )X * (ω k )]+E[X(ω k )N * (ω k )]

[0076] Where N(ω) k ) represents the noise signal at ω k The amplitude of the frequency point, due to the clean signal X(ω) k ) and noise signal N(ω) k ) are unrelated, and N(ω) k If the expected value of ) is 0, then we can obtain:

[0077] P yx * (ω k )=E[X(ω k )X * (ωk )]

[0078] E[X(ω) k )X * (ω k )] is denoted as P xx (ω k ), E[N(ω k )N * (ω k )] is denoted as P nn (ω k Then for P yy (ω k )have:

[0079] P yy (ω k )=E[|Y(ω k )| 2 ]=E[|X(ω k )+N(ω k )| 2 ] = P xx (ω k )+P nn (ω k )

[0080] Therefore, we can obtain:

[0081]

[0082] Define the prior signal-to-noise ratio

[0083]

[0084] That is, when the signal-to-noise ratio is high, the signal is allowed to pass through, and when the signal-to-noise ratio is low, the signal is suppressed from passing through.

[0085] In practical applications, a parametric Wiener filter is used to enhance the noise reduction effect.

[0086]

[0087] Where β represents the filter coefficients.

[0088] In some embodiments, such as Figure 2 As shown, the estimated signal obtained after denoising the original signal of the sample specifically includes the following steps:

[0089] First, the original signal of the sample is preprocessed. In order to achieve better noise reduction, the noisy signal is first detrended and min-max normalized. Then, in order to facilitate frame segmentation, the signal is extended.

[0090] Then, a short-time Fourier transform is performed on the preprocessed signal to obtain the decomposed amplitude spectrum and phase spectrum; a short-time Fourier transform with a Hanning window is used with a step size of 1 / 4 of the window length, and the amplitude spectrum and phase spectrum are saved after decomposition.

[0091] Then, at least one frame of each noise signal is extracted from the original signal of the sample as the target signal, and the target signal is subjected to Fourier transform; for ease of calculation, only one frame of data is extracted from each noise signal for analysis, and Fourier transform is also performed after min-max normalization.

[0092] Finally, oversubtraction is used to perform spectral subtraction on the original signal of the sample to obtain the estimated signal after denoising. The Wiener filter adapted to the noisy signal is then trained using the estimated clean signal and the noisy signal. Specifically, the trained Wiener filter is used to denoise the signal, resulting in a denoised curve. The time series and spectrum are then examined to check the denoising effect. Constructing the Wiener filter requires a clean signal. In engineering, purely noisy signals are often available, but the clean signal is actually the target. Therefore, it is necessary to first estimate the clean signal using other methods. In this invention, oversubtraction is used to estimate the clean signal.

[0093] The process of obtaining a clean signal prediction using oversubtraction is described in detail below.

[0094] A noisy signal consists of a clean signal and a noisy signal:

[0095] y(n) = x(n) + e(n)

[0096] Where y(n) represents a noisy signal sequence, x(n) represents a clean signal sequence, and e(n) is a noise signal.

[0097] Its frequency domain is:

[0098] Y(ω)=X(ω)+E(ω)

[0099] Where Y(ω) is the frequency domain form of y(n), X(ω) is the frequency domain form of x(n), and E(ω) is the frequency domain form of e(n).

[0100] There is an estimate for the amplitude of a clean signal:

[0101]

[0102] in, Let |Y(ω)| be the magnitude estimate of X(ω), |Y(ω)| be the magnitude of Y(ω), and |E(ω)| be the magnitude of e(n).

[0103] By combining the phase spectrum of Y(ω), an estimate of X(ω) can be obtained:

[0104]

[0105] in, This is an estimate of X(ω), where j is the imaginary unit. Let Y(ω) be the phase.

[0106] right The inverse Fourier transform yields an estimate of x(n):

[0107]

[0108] in, For the estimate of x(n), IDFT is the inverse Fourier transform function.

[0109] Because spectral subtraction is ineffective in practical applications, an improved oversubtraction method was used. The oversubtraction method modifies the estimation of the X(ω) energy spectrum. In the oversubtraction method, the estimated X(ω) energy spectrum is:

[0110]

[0111] Among them, P Y (ω) is the energy spectrum of Y(ω), P E (ω) represents the energy spectrum of E(ω), and α and γ are adjustable coefficients in the oversubtraction method. When γ = 1, it is equivalent to subtracting the energy spectrum; when γ = 0.5, it is equivalent to subtracting the amplitude spectrum. The larger the value of α, the greater the noise attenuation and the greater the impact on the original signal. To ensure the continuity of the calculation results and avoid excessive discontinuities, the excessively small energy spectrum is corrected after the calculation.

[0112]

[0113] Where β is the correction coefficient.

[0114] In some embodiments, the original micro-vibration signal is reconstructed using a wavelet decomposition algorithm to obtain an effective signal after removing zero-point drift, specifically including:

[0115] Determine the initial fluctuation position of the original micro-vibration signal. Since experimental data often does not have amplitude at the beginning, but rather begins to show amplitude after a period of time, if wavelet decomposition is performed from the first data point during calculation, false fluctuations will be generated in the region where there was no actual amplitude after drift removal. Therefore, it is necessary to first determine the initial position of the signal, and then perform drift removal on the signal after the initial position.

[0116] Discrete wavelet decomposition is performed on the signal after the initial fluctuation position in the original micro-vibration signal according to a pre-set number of decomposition layers. When determining the number of decomposition layers, there is a maximum number of decomposable layers for the signal. The more decomposition layers there are, the less residual energy the approximation coefficients have, and the more likely the mean is to approach zero, thus making it less likely to be judged as containing zero drift. This method selects 70% to 80% of the maximum decomposition layer number as the decomposition layer number.

[0117] Based on the wavelet decomposition results, it is determined that the original micro-vibration signal exhibits zero-point drift. If so, the original micro-vibration signal is reconstructed to obtain an effective signal after removing the zero-point drift. The mean of the approximate values ​​of the decomposed signal is calculated. If the mean is greater than a threshold, it is determined that the original micro-vibration signal exhibits zero-point drift.

[0118] The drift removal effect is determined based on the generated impulse response spectrum. Ignoring the approximation coefficients of the drift-containing signal, the drift-containing signal is reconstructed using all wavelet coefficients. The reconstructed signal is the de-drift signal. The drift removal effect is then observed by examining the impulse response spectra of the signal before and after de-drifting.

[0119] Specifically, in the wavelet decomposition process, this method uses the discrete Mayer wavelet as the wavelet basis for wavelet decomposition, which is essentially an FIR filter approximation of the Mayer wavelet. According to the definition of the discrete Mayer wavelet, a high-pass filter and a low-pass filter are specified to filter the original signal to obtain the first-level approximation coefficients and wavelet coefficients respectively.

[0120]

[0121]

[0122] Where, x 1,L (n) represents the first-level approximation coefficients, x 1,H (n) represents the wavelet coefficients of the first-level decomposition, K is the sum of the filter sequence length and the model sequence length, x(n) is the original signal, h(n) represents the high-pass and low-pass filters defined by the definition of discrete Mayer wavelets, and g(n) represents the high-pass filter defined by the definition of discrete Mayer wavelets.

[0123] The first-level approximation coefficients are further filtered to obtain the second-level approximation coefficients:

[0124]

[0125]

[0126] Where, x 2,L (n) represents the second-level approximation coefficients, x 2,H (n) represents the wavelet coefficients of the second-level decomposition.

[0127] Repeat this process until the decomposition reaches a specified number of levels α. Then, you can view the approximation coefficients x of level α. α,L If the mean of (n) is less than the threshold, the signal is considered to have no zero drift; otherwise, the signal is considered to have zero drift. The approximation coefficients of this layer are ignored, and the signal is reconstructed using all wavelet coefficients to obtain the zero-drift-free signal of the original signal.

[0128]

[0129] in, The signal after zero drift removal is called IDWT, which is the inverse discrete wavelet transform.

[0130] The following uses a specific application scenario as an example to briefly describe the implementation process and technical effects of the method provided by this invention.

[0131] like Figure 3 The Original Signal shown is the original signal containing background noise. The horizontal axis represents the test time, and the vertical axis represents the vibration amplitude. The Background Noise is the background noise data corresponding to that location. Both the original signal and the background noise are sampled at 6400Hz. The original signal has 2,598,656 sampling points, and the background noise has 548,608 sampling points. To maximize frequency resolution, the frame length of the Short-Time Fourier Transform (SFT) is 524,288 (219). The original signal is symmetrically extended to 2,883,584 points (5.5 frames). A SFT is then performed on the extended original signal with a Hanning window length of 524,288 and an overlap length of 393,216, yielding the time spectrum of the original signal. The background noise is truncated to 548,608 points, and then subjected to a Fourier Transform to obtain its spectrum. Spectral subtraction is then used on the time spectrum of the original signal to obtain the estimated spectrum of the clean signal. The filter coefficients of the Wiener filter are calculated using the estimated spectrum of the clean signal and the spectrum of the noise signal. The Wiener filter is then used to filter the original signal to obtain the denoised signal. Figure 3 The denoised signal is shown below. Fourier transforms of the original signal, background noise signal, and denoised signal yield the following results. Figure 4 The corresponding spectra are shown below, with the horizontal axis representing frequency and the vertical axis representing amplitude and energy.

[0132] Figure 5 The image shows a signal containing zero drift. In the original signal, the latter part of the signal clearly deviates from the baseline. After performing a 9-level wavelet decomposition on the original signal, a trend curve is obtained. This trend curve component is then removed, and the signal is reconstructed using wavelet coefficients to obtain the final signal. Figure 5 The drift curve in the curve is de-dashed, and the phenomenon of deviation from the baseline is eliminated. Figure 6It compares the impulse response spectra of the original signal and the de-drifted signal. "Positive" represents the positive impulse response spectrum, and "negative" represents the reciprocal of the negative impulse response spectrum. After removing zero drift, the positive and negative impulse response spectra of the signal are closer.

[0133] In the above specific embodiments, the Wiener filter-based micro-vibration data processing method provided by the present invention obtains the original micro-vibration signal to be processed, inputs the original micro-vibration signal into a pre-trained Wiener filter, and obtains the denoised signal output by the Wiener filter. When the original micro-vibration signal has the risk of zero-point drift, a wavelet decomposition algorithm is used to reconstruct the original micro-vibration signal to obtain an effective signal after removing the zero-point drift. The method provided by the present invention uses a trained Wiener filter to extract weak signals under strong background noise, uses wavelet decomposition to determine whether the signal has zero-point drift, and removes any existing zero-point drift. Compared with traditional manually designed filters, using a Wiener filter can achieve adaptive and accurate filtering, thereby more accurately separating the real signal from background noise. In the analysis of micro-vibration test data, effective denoising and zero-drift removal of the signal are achieved, thereby improving signal cleanliness and ensuring the accuracy of simulation analysis.

[0134] Furthermore, the method provided by this invention uses wavelet decomposition to more accurately capture the time-frequency characteristics of a signal and more automatically identify the presence of zero drift, thereby achieving a better drift removal effect. This makes it easier for experimenters to obtain more reliable experimental data, and thus more accurately analyze the vibration source and vibration transmission characteristics.

[0135] In addition to the methods described above, this invention also provides a micro-vibration data processing device based on a Wiener filter, such as... Figure 7 As shown, the device includes:

[0136] The signal acquisition unit 710 is used to acquire the original micro-vibration signal to be processed.

[0137] The signal processing unit 720 is used to input the original micro-vibration signal into a pre-trained Wiener filter to obtain a denoised signal output by the Wiener filter; and when the original micro-vibration signal has the risk of zero-point drift, it uses a wavelet decomposition algorithm to reconstruct the original micro-vibration signal to obtain an effective signal after removing the zero-point drift.

[0138] The Wiener filter is trained using the original signal of the sample and the estimated signal obtained after denoising the original signal of the sample.

[0139] In some embodiments, the Wiener filter is obtained by training using the original signal of the sample and the estimated signal obtained after denoising the original signal of the sample, specifically including:

[0140] Based on the original signal of the sample, the estimate of the clean signal obtained after denoising is calculated according to the preset objective function;

[0141] At each frequency point, the mean square error between the estimate of the clean signal and the clean signal is calculated;

[0142] The Wiener filter is obtained by training with the goal of minimizing the mean square error.

[0143] In some embodiments, the estimated signal obtained after denoising the original signal of the sample specifically includes:

[0144] The original signal of the sample is preprocessed;

[0145] The preprocessed signal is subjected to a short-time Fourier transform to obtain the decomposed amplitude spectrum and phase spectrum.

[0146] Extract at least one frame of each noise signal from the original signal of the sample as the target signal, and perform a Fourier transform on the target signal;

[0147] The original signal of the sample was subjected to spectral subtraction using oversubtraction to obtain the estimated signal after denoising.

[0148] In some embodiments, the energy spectrum is estimated using the oversubtraction method as follows:

[0149]

[0150] Where Y(ω) is the original signal of the sample, P Y (ω) represents the energy spectrum of Y(ω), E(ω) represents the frequency domain form of the noise signal of the sample, and P E (ω) is the energy spectrum of E(ω), and α and γ are adjustable coefficients in the oversubtraction method. When γ = 1, it is equivalent to energy spectrum subtraction, and when γ = 0.5, it is equivalent to amplitude spectrum subtraction. The larger the value of α, the greater the degree of noise attenuation and the greater the impact on the original signal.

[0151] In some embodiments, the original micro-vibration signal is reconstructed using a wavelet decomposition algorithm to obtain an effective signal after removing zero-point drift, specifically including:

[0152] Determine the initial fluctuation position of the original micro-vibration signal;

[0153] Discrete wavelet decomposition is performed on the signal after the initial fluctuation position in the original micro-vibration signal according to the preset number of decomposition layers.

[0154] Based on the results of wavelet decomposition, it is determined that the original micro-vibration signal has zero-point drift. If so, the original micro-vibration signal is reconstructed to obtain an effective signal after removing the zero-point drift.

[0155] In some embodiments, determining the presence of zero-point drift in the original micro-vibration signal based on the wavelet decomposition result specifically includes:

[0156] Calculate the mean of the approximate values ​​of the decomposed signal. If the mean is greater than a threshold, it is determined that the original micro-vibration signal has zero-point drift.

[0157] In some embodiments, the original micro-vibration signal is reconstructed to obtain an effective signal after removing zero-point drift, and the process further includes:

[0158] The drift effect is judged based on the generated impact response spectrum.

[0159] In the above specific embodiments, the Wiener filter-based micro-vibration data processing device provided by the present invention acquires the original micro-vibration signal to be processed, inputs the original micro-vibration signal into a pre-trained Wiener filter, and obtains the denoised signal output by the Wiener filter. When the original micro-vibration signal has a risk of zero-point drift, a wavelet decomposition algorithm is used to reconstruct the original micro-vibration signal to obtain an effective signal after removing the zero-point drift. The device provided by the present invention uses a trained Wiener filter to extract weak signals under strong background noise, uses wavelet decomposition to determine whether the signal has zero-point drift, and removes any existing zero-point drift. Compared with traditional methods of manually designing filters, using a Wiener filter can achieve adaptive and accurate filtering, thereby more accurately separating the real signal from background noise. In the analysis of micro-vibration test data, effective denoising and zero-drift removal of the signal are achieved, thereby improving signal cleanliness and ensuring the accuracy of simulation analysis.

[0160] Furthermore, the device provided by this invention can more accurately capture the time-frequency characteristics of a signal using wavelet decomposition, and more automatically distinguish the existence of zero drift, thereby achieving a better de-drift effect, making it easier for experimenters to obtain more reliable experimental data, and thus more accurately analyze the vibration source and vibration transmission characteristics.

[0161] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 8As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and model predictions. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The model predictions of the computer device store static and dynamic information data. The network interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the steps in the above method embodiments.

[0162] Those skilled in the art will understand that Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the computer device to which the present invention is applied. A specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0163] Corresponding to the above embodiments, this invention also provides a computer storage medium containing one or more program instructions. These one or more program instructions are used to execute the method described above.

[0164] The present invention also provides a computer program product, the computer program product including a computer program, the computer program being stored on a non-transitory computer-readable storage medium, and the computer being able to perform the above-described method when the computer program is executed by a processor.

[0165] In this embodiment of the invention, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0166] The various methods, steps, and logic diagrams disclosed in the embodiments of this invention can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The processor reads information from the storage medium and, in conjunction with its hardware, completes the steps of the above methods.

[0167] The storage medium can be memory, such as volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.

[0168] Among them, non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory.

[0169] Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (Synchlink DRAM, SLDRAM), and direct memory bus RAM (DRRAM).

[0170] The storage media described in the embodiments of the present invention are intended to include, but are not limited to, these and any other suitable types of memory.

[0171] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this invention can be implemented using a combination of hardware and software. When applied as software, the corresponding functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one place to another. Storage media can be any available medium accessible to general-purpose or special-purpose computers.

[0172] The above specific embodiments further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for processing micro-vibration data based on a Wiener filter, characterized in that, The method includes: Acquire the raw micro-vibration signal to be processed; The original micro-vibration signal is input into a pre-trained Wiener filter to obtain the denoised signal output by the Wiener filter; in the case that the original micro-vibration signal has the risk of zero-point drift, the original micro-vibration signal is reconstructed using a wavelet decomposition algorithm to obtain an effective signal after removing the zero-point drift. The Wiener filter is obtained by training the original signal of the sample and the estimated signal obtained after denoising the original signal of the sample. The Wiener filter is obtained by training using the original signal of the sample and the estimated signal obtained after denoising the original signal of the sample, specifically including: Based on the original signal of the sample, the estimate of the clean signal obtained after denoising is calculated according to the preset objective function; At each frequency point, the mean square error between the estimate of the clean signal and the clean signal is calculated; The Wiener filter is obtained by training with the goal of minimizing the mean square error. The estimated signal obtained after denoising the original signal of the sample specifically includes: The original signal of the sample is preprocessed; The preprocessed signal is subjected to a short-time Fourier transform to obtain the decomposed amplitude spectrum and phase spectrum. Extract at least one frame of each noise signal from the original signal of the sample as the target signal, and perform a Fourier transform on the target signal; The original signal of the sample is subjected to spectral subtraction using over-subtraction to obtain the denoised estimated signal; In the oversubtraction method, the energy spectrum is estimated as follows: in, The original signal of the sample. for The energy spectrum, This is the frequency domain form of the noise signal of the sample. for The energy spectrum, and For the adjustable coefficient in the over-subtraction, when This is equivalent to a decrease in the energy spectrum. This is equivalent to a decrease in the amplitude spectrum. The larger the value of , the greater the noise reduction and the greater the impact on the original signal.

2. The micro-vibration data processing method based on Wiener filter according to claim 1, characterized in that, The original micro-vibration signal is reconstructed using wavelet decomposition algorithm to obtain an effective signal after removing zero-point drift, specifically including: Determine the initial fluctuation position of the original micro-vibration signal; Discrete wavelet decomposition is performed on the signal after the initial fluctuation position in the original micro-vibration signal according to the preset number of decomposition layers. Based on the results of wavelet decomposition, it is determined that the original micro-vibration signal has zero-point drift. If so, the original micro-vibration signal is reconstructed to obtain an effective signal after removing the zero-point drift.

3. The micro-vibration data processing method based on Wiener filter according to claim 2, characterized in that, Based on the wavelet decomposition results, it was determined that the original micro-vibration signal exhibited zero-point drift, specifically including: Calculate the mean of the approximate values ​​of the decomposed signal. If the mean is greater than a threshold, it is determined that the original micro-vibration signal has zero-point drift.

4. The micro-vibration data processing method based on Wiener filter according to claim 2, characterized in that, The original micro-vibration signal is reconstructed to obtain an effective signal after removing zero-point drift, and then the process further includes: The drift effect is judged based on the generated impact response spectrum.

5. A micro-vibration data processing device based on a Wiener filter, used to implement the method as described in any one of claims 1-4, characterized in that, The device includes: The signal acquisition unit is used to acquire the original micro-vibration signal to be processed. The signal processing unit is used to input the original micro-vibration signal into a pre-trained Wiener filter to obtain a denoised signal output by the Wiener filter; and, in the case that the original micro-vibration signal has the risk of zero-point drift, to reconstruct the original micro-vibration signal using a wavelet decomposition algorithm to obtain an effective signal after removing the zero-point drift. The Wiener filter is trained using the original signal of the sample and the estimated signal obtained after denoising the original signal of the sample.

6. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method as described in any one of claims 1-4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-4.

Images