Signal noise suppression method and device, electronic equipment and storage medium

Abstract

The invention discloses a signal noise suppression method and device, electronic equipment and a storage medium, and relates to the field of noise suppression, and the suppression method comprises the steps: collecting a target signal, carrying out the Fourier transform of the target signal, obtaining a signal frequency spectrum of the target signal, determining a frequency spectrum threshold value based on the standard deviation of a background noise frequency spectrum, and carrying out the calculation of the frequency spectrum threshold value. The method comprises the steps of obtaining a frequency spectrum threshold value, removing a noise component in the signal frequency spectrum based on the frequency spectrum threshold value, performing Fourier inversion on the signal frequency spectrum with the noise component removed to obtain an initial noise reduction signal, positioning a pulse noise signal in the target signal based on the initial noise reduction signal and the target signal, and performing smoothing processing on the pulse noise signal to obtain a target noise reduction signal. According to the invention, the technical problem that signal noise cannot be effectively suppressed in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of noise suppression, and in particular to a method for suppressing signal noise, a device thereof, an electronic device and a storage medium. Background Art

[0002] In addition to the fundamental frequency component and broadband component, the power system signal also contains noise components, which can be divided into background noise and impulse noise according to their amplitude. Background noise is caused by random fluctuations in the load, and its amplitude is small and usually follows a Gaussian distribution. Impulse noise is usually caused by the switching of high-power electrical equipment and nonlinear loads. Its amplitude is greater than the background noise, and it has the characteristics of strong randomness and short duration. The noise component seriously affects the accuracy of the synchronous phasor measurement algorithm and may even cause bad data in the measurement results. In order to achieve the goal of high-precision measurement, a signal noise suppression method is urgently needed to ensure good denoising effect at noise levels of different contents and intensities.

[0003] In related technologies, linear filtering, nonlinear filtering, and transform domain denoising are generally used to suppress signal noise. The details are as follows:

[0004] (1) Linear filtering: The noise in the signal can be filtered out by designing a filter to convolve with the signal. The filter can be divided into two categories: finite impulse response filter (FIR) and infinite impulse response filter (IIR). The advantages of FIR filter are simple structure, only signal input value is needed in the filtering process, and strong stability, but a higher order is required under the same index. IIR filter has a relatively complex structure, and the output depends not only on the signal input value but also on the previous signal output value, but the order required under the same index is low. In addition, the best linear estimation of the signal can be achieved by utilizing the statistical characteristics of the signal and noise. This method can still effectively filter out noise when the noise and effective signal bands overlap, but it requires prior knowledge of the signal and noise, which is difficult to apply in practice. Linear filtering is equivalent to filtering only in the frequency domain, but background noise and random impulse noise are difficult to distinguish in the frequency domain, so this type of method lacks robustness to impulse noise.

[0005] (2) Nonlinear filtering: Remove noise by changing the signal shape in the time domain. Common nonlinear filtering methods include median filtering and its improved algorithm, robust local regression smoothing, etc. The median filtering method has a strong suppression effect on pulse noise, but the filtering effect is easily affected by the window length. When the signal changes rapidly, it is easy to produce distortion and suppress broadband components, affecting the accuracy of broadband measurement. The robust local regression smoothing method requires iterative fitting of all data in the window, and outlier judgment after fitting and smoothing. If outliers are judged, the correction coefficient needs to be updated and the fitting matrix needs to be recalculated, which involves multiple matrix inversions and multiplications. The calculation complexity is high and it is difficult to apply online.

[0006] (3) Transform domain denoising: transform the signal to other domains, and remove the noise by setting thresholds according to the different characteristics of the signal and noise, such as wavelet denoising, singular value decomposition (SVD), empirical mode decomposition (EMD) and other methods. Wavelet denoising is to transform the noisy signal to the wavelet domain, and process the corresponding wavelet coefficients by setting thresholds to remove the noise. This method is very effective in processing sudden signal, but it is less effective in suppressing noise in the low frequency band (within the measurement band); SVD and EMD both decompose the signal, separate the effective signal and noise, and reconstruct the signal to achieve the purpose of denoising. When the noise band and the effective signal band overlap, these two methods are difficult to distinguish between the effective signal and noise, and the denoising effect is poor.

[0007] Therefore, the signal noise suppression method in the related art has the following problems: (1) lack of robustness to pulse noise; (2) involving a large number of matrix inversions and multiplications, with high computational complexity; (3) unable to suppress noise within the measurement frequency band.

[0008] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0009] The embodiments of the present invention provide a method for suppressing signal noise and a device thereof, an electronic device and a storage medium, so as to at least solve the technical problem that signal noise cannot be effectively suppressed in the related art.

[0010] According to one aspect of an embodiment of the present invention, a method for suppressing signal noise is provided, comprising: acquiring a target signal, and performing Fourier transform on the target signal to obtain a signal spectrum of the target signal, wherein the signal spectrum at least includes: a background noise spectrum; determining a spectrum threshold based on a standard deviation of the background noise spectrum, and removing noise components in the signal spectrum based on the spectrum threshold; performing inverse Fourier transform on the signal spectrum from which the noise components have been removed to obtain an initial noise reduction signal; locating an impulse noise signal in the target signal based on the initial noise reduction signal and the target signal, and smoothing the impulse noise signal to obtain a target noise reduction signal.

[0011] Furthermore, before determining the spectrum threshold based on the standard deviation of the background noise spectrum, it also includes: determining the real part of the noise spectrum and the imaginary part of the noise spectrum of the background noise spectrum, and determining the power spectrum of the background noise and the amplitude spectrum of the background noise based on the real part of the noise spectrum and the imaginary part of the noise spectrum; normalizing the real part of the noise spectrum and the imaginary part of the noise spectrum respectively to obtain the real part of the standard normal noise spectrum and the imaginary part of the standard normal noise spectrum; determining the normalized power spectrum based on the real part of the quasi-normal noise spectrum and the imaginary part of the standard normal noise spectrum, wherein the normalized power spectrum has a chi-square distribution characteristic, and the chi-square distribution characteristic means that the normalized power spectrum obeys a chi-square distribution with a preset degree of freedom.

[0012] Furthermore, the step of determining a spectrum threshold based on the standard deviation of the background noise spectrum includes: determining a preset confidence, and determining a noise reduction threshold based on the preset confidence; determining the standard deviation of the background noise spectrum based on the chi-square distribution characteristic of the normalized power spectrum; and determining the spectrum threshold based on the standard deviation and the noise reduction threshold.

[0013] Furthermore, based on the chi-square distribution characteristics of the normalized power spectrum, the step of determining the standard deviation of the background noise spectrum includes: determining the real part of the signal spectrum and the imaginary part of the signal spectrum, and constructing a signal spectrum vector based on the real part of the signal spectrum and the imaginary part of the signal spectrum; determining the vector standard deviation of the signal spectrum vector; determining the normalized signal power spectrum of the target signal based on the real part of the signal spectrum, the imaginary part of the signal spectrum and the vector standard deviation, and determining the signal power spectrum standard deviation of the normalized signal power spectrum; when the signal power spectrum standard deviation is greater than the preset degree of freedom, eliminating the maximum component in the signal spectrum; based on the signal spectrum with the maximum component eliminated, re-determining the signal power spectrum standard deviation until the signal power spectrum standard deviation is less than or equal to the preset degree of freedom, obtaining the target signal power spectrum standard deviation; characterizing the target vector standard deviation used when determining the target signal power spectrum standard deviation as the standard deviation of the background noise spectrum.

[0014] Furthermore, the step of locating the impulse noise signal in the target signal based on the initial denoised signal and the target signal includes: determining the signal residual based on the initial denoised signal and the target signal; determining the residual mean and residual variance of the signal residual, and determining an initial threshold value based on the residual mean, the residual variance and a preset adjustment coefficient; removing elements in the signal residual that are greater than the initial threshold value, and re-determining the initial threshold value based on the signal residual after removing the elements, until there are no elements in the signal residual that are greater than the initial threshold value, and obtaining a target threshold value; locating the impulse noise signal in the target signal based on the target threshold value.

[0015] Furthermore, the step of smoothing the impulse noise signal to obtain a target noise reduction signal includes: based on the position index of the located impulse noise signal, locally sampling the target signal to obtain a local sampling sequence; based on a fitting parameter matrix and the local sampling sequence, determining a smoothed sampling sequence; replacing the local sampling sequence in the target signal with the smoothed sampling sequence to obtain a smoothed noise reduction signal; and based on the spectrum threshold, denoising the smoothed noise reduction signal to obtain the target noise reduction signal.

[0016] Furthermore, after locally sampling the target signal based on the position index of the located pulse noise signal to obtain a local sampling sequence, it also includes: performing polynomial fitting on the local sampling sequence to obtain a position parameter matrix; determining an initial fitting parameter matrix based on the position parameter matrix and the signal sampling value matrix, wherein the initial fitting parameter matrix includes a weighting coefficient diagonal matrix and a fitting weight diagonal matrix; correcting the weighting coefficient diagonal matrix to obtain a target weighting coefficient diagonal matrix; and determining the fitting parameter matrix based on the target weighting coefficient diagonal matrix.

[0017] According to another aspect of an embodiment of the present invention, a signal noise suppression device is also provided, including: a transformation unit, used to collect a target signal, and perform Fourier transform on the target signal to obtain a signal spectrum of the target signal, wherein the signal spectrum at least includes: a background noise spectrum; a determination unit, used to determine a spectrum threshold based on a standard deviation of the background noise spectrum, and remove noise components in the signal spectrum based on the spectrum threshold; an inverse transformation unit, used to perform inverse Fourier transform on the signal spectrum from which the noise component has been removed to obtain an initial noise reduction signal; and a positioning unit, used to locate an impulse noise signal in the target signal based on the initial noise reduction signal and the target signal, and smooth the impulse noise signal to obtain a target noise reduction signal.

[0018] Furthermore, the suppression device also includes: a first determination module, used to determine the real part of the noise spectrum and the imaginary part of the noise spectrum of the background noise spectrum before determining the spectrum threshold based on the standard deviation of the background noise spectrum, and determine the power spectrum of the background noise and the amplitude spectrum of the background noise based on the real part of the noise spectrum and the imaginary part of the noise spectrum; a first processing module, used to normalize the real part of the noise spectrum and the imaginary part of the noise spectrum respectively to obtain the real part of the standard normal noise spectrum and the imaginary part of the standard normal noise spectrum; a second determination module, used to determine the normalized power spectrum based on the real part of the quasi-normal noise spectrum and the imaginary part of the standard normal noise spectrum, wherein the normalized power spectrum has a chi-square distribution characteristic, and the chi-square distribution characteristic means that the normalized power spectrum obeys the chi-square distribution with a preset degree of freedom.

[0019] Furthermore, the determination unit includes: a third determination module, used to determine a preset confidence and determine a noise reduction threshold based on the preset confidence; a fourth determination module, used to determine the standard deviation of the background noise spectrum based on the chi-square distribution characteristics of the normalized power spectrum; and a fifth determination module, used to determine the spectrum threshold based on the standard deviation and the noise reduction threshold.

[0020] Furthermore, the fourth determination module includes: a first determination submodule, used to determine the real part of the signal spectrum and the imaginary part of the signal spectrum, and construct a signal spectrum vector based on the real part of the signal spectrum and the imaginary part of the signal spectrum; a second determination submodule, used to determine the vector standard deviation of the signal spectrum vector; a third determination submodule, used to determine the normalized signal power spectrum of the target signal based on the real part of the signal spectrum, the imaginary part of the signal spectrum and the vector standard deviation, and determine the signal power spectrum standard deviation of the normalized signal power spectrum; a first elimination submodule, used to eliminate the maximum component in the signal spectrum when the signal power spectrum standard deviation is greater than the preset degree of freedom; a fourth determination submodule, used to re-determine the signal power spectrum standard deviation based on the signal spectrum with the maximum component eliminated, until the signal power spectrum standard deviation is less than or equal to the preset degree of freedom, to obtain the target signal power spectrum standard deviation; a first characterization submodule, used to characterize the target vector standard deviation used when determining the target signal power spectrum standard deviation as the standard deviation of the background noise spectrum.

[0021] Furthermore, the positioning unit includes: a sixth determination module, used to determine the signal residual based on the initial noise reduction signal and the target signal; a seventh determination module, used to determine the residual mean and residual variance of the signal residual, and determine the initial threshold value based on the residual mean, the residual variance and a preset adjustment coefficient; a first removal module, used to remove elements in the signal residual that are greater than the initial threshold value, and re-determine the initial threshold value based on the signal residual after the elements are removed, until there are no elements in the signal residual that are greater than the initial threshold value, and the target threshold value is obtained; a first positioning module, used to locate the pulse noise signal in the target signal based on the target threshold value.

[0022] Furthermore, the positioning unit also includes: a first sampling module, used to perform local sampling on the target signal based on the position index of the located impulse noise signal to obtain a local sampling sequence; an eighth determination module, used to determine a smoothed sampling sequence based on a fitting parameter matrix and the local sampling sequence; a first replacement module, used to replace the local sampling sequence in the target signal with the smoothed sampling sequence to obtain a smoothed denoised signal; and a first denoising module, used to denoise the smoothed denoised signal based on the spectrum threshold to obtain the target denoised signal.

[0023] Furthermore, the suppression device also includes: a first fitting module, which is used to perform local sampling of the target signal based on the position index of the located pulse noise signal, and after obtaining the local sampling sequence, perform polynomial fitting on the local sampling sequence to obtain a position parameter matrix; a ninth determination module, which is used to determine the initial fitting parameter matrix based on the position parameter matrix and the signal sampling value matrix, wherein the initial fitting parameter matrix includes a weighting coefficient diagonal matrix and a fitting weight diagonal matrix; a first correction module, which is used to correct the weighting coefficient diagonal matrix to obtain a target weighting coefficient diagonal matrix; and a tenth determination module, which is used to determine the fitting parameter matrix based on the target weighting coefficient diagonal matrix.

[0024] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is further provided, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned signal noise suppression methods.

[0025] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any one of the above-mentioned signal noise suppression methods.

[0026] In the present invention, a target signal is collected and Fourier transformed to obtain a signal spectrum of the target signal. A spectrum threshold is determined based on the standard deviation of a background noise spectrum. Based on the spectrum threshold, noise components in the signal spectrum are removed. An inverse Fourier transform is performed on the signal spectrum from which the noise components are removed to obtain an initial noise reduction signal. Based on the initial noise reduction signal and the target signal, an impulse noise signal in the target signal is located. The impulse noise signal is smoothed to obtain a target noise reduction signal.

[0027] In the present invention, by establishing a normalized power spectrum chi-square distribution model of background noise, the standard deviation of the background noise spectrum can be obtained, and then the spectrum threshold is determined, so that the noise component can be removed according to the spectrum threshold, thereby effectively suppressing the noise in the measurement band. In addition, the impulse noise signal in the target signal can be located according to the initial noise reduction signal and the target signal after the noise component is removed, so as to smooth the impulse noise signal and finally obtain the target noise reduction signal, which not only reduces the amount of calculation, but also can effectively suppress the noise in the signal, thereby solving the technical problem that the signal noise cannot be effectively suppressed in the related art. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0029] Figure 1 is a flow chart of an optional signal noise suppression method according to an embodiment of the present invention;

[0030] Figure 2 is a schematic diagram of an optional signal noise suppression process according to an embodiment of the present invention;

[0031] Figure 3 is a schematic diagram of an optional distribution network voltage recording waveform diagram according to an embodiment of the present invention;

[0032] Figure 4 is a schematic diagram of an optional amplitude measurement result according to an embodiment of the present invention;

[0033] Figure 5 is a schematic diagram of an optional frequency measurement result according to an embodiment of the present invention;

[0034] Figure 6 is a schematic diagram of an optional signal noise suppression device according to an embodiment of the present invention;

[0035] Figure 7 The present invention is a hardware structure block diagram of an electronic device (or mobile device) for a method for suppressing signal noise according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0037] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0038] It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) collected and involved in the present invention are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data are in compliance with the relevant laws, regulations and standards of the relevant regions, necessary confidentiality measures are taken, and public order and good customs are not violated, and corresponding operation entrances are provided for users to choose to authorize or refuse. For example, an interface is set between the system and the relevant users or organizations. Before obtaining relevant information, it is necessary to send an acquisition request to the aforementioned user or organization through the interface, and obtain relevant information after receiving the consent information fed back by the aforementioned user or organization.

[0039] In the present invention, by establishing a normalized power spectrum chi-square distribution model of background noise, based on its distribution characteristics and according to the characteristic that the spectrum value of the effective signal is greater than the noise spectrum value, the standard deviation of the background noise can be obtained by continuously eliminating the component with the largest spectrum value, and then the spectrum threshold is obtained, wherein the spectrum threshold can adaptively adjust the threshold for noises with different noise levels, and can effectively suppress the noise in the measurement band. Moreover, according to the characteristic that the amplitude of the pulse noise is much greater than the background noise, the accurate distinction between the background noise and the pulse noise and the location of the pulse noise can be achieved based on the residual and the threshold value. In addition, based on the idea of ​​nonlinear fitting, a pulse noise suppression method is proposed, by setting a fixed fitting parameter matrix, so that only one matrix multiplication is required to suppress a group of pulse noises, which can reduce the amount of calculation.

[0040] The present invention is described in detail below in conjunction with various embodiments.

[0041] Embodiment 1

[0042] According to an embodiment of the present invention, an embodiment of a method for suppressing signal noise is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0043] Figure 1 is a flow chart of an optional signal noise suppression method according to an embodiment of the present invention, such as Figure 1 As shown, the method comprises the following steps:

[0044] Step S101, collecting a target signal, and performing Fourier transform on the target signal to obtain a signal spectrum of the target signal, wherein the signal spectrum at least includes: a background noise spectrum.

[0045] Step S102: determining a spectrum threshold based on the standard deviation of the background noise spectrum, and removing noise components in the signal spectrum based on the spectrum threshold.

[0046] Step S103, performing inverse Fourier transform on the signal spectrum from which the noise component has been removed to obtain an initial noise reduction signal.

[0047] Step S104: based on the initial noise reduction signal and the target signal, locate the impulse noise signal in the target signal, and perform smoothing on the impulse noise signal to obtain a target noise reduction signal.

[0048] Through the above steps, the target signal can be collected, and the target signal is Fourier transformed to obtain the signal spectrum of the target signal. Based on the standard deviation of the background noise spectrum, the spectrum threshold is determined, and based on the spectrum threshold, the noise component in the signal spectrum is removed, and the signal spectrum with the noise component removed is Fourier inverse transformed to obtain the initial noise reduction signal. Based on the initial noise reduction signal and the target signal, the impulse noise signal in the target signal is located, and the impulse noise signal is smoothed to obtain the target noise reduction signal. In the embodiment of the present invention, by establishing a normalized power spectrum chi-square distribution model of the background noise, the standard deviation of the background noise spectrum can be obtained, and then the spectrum threshold is determined to remove the noise component according to the spectrum threshold, thereby effectively suppressing the noise in the measurement band. In addition, the impulse noise signal in the target signal can also be located according to the initial noise reduction signal with the noise component removed and the target signal, so as to smooth the impulse noise signal, and finally obtain the target noise reduction signal, which not only reduces the amount of calculation, but also can effectively suppress the noise in the signal, thereby solving the technical problem that the signal noise cannot be effectively suppressed in the related art.

[0049] The embodiment of the present invention is described in detail below in combination with the above steps.

[0050] Step S101, collecting a target signal, and performing Fourier transform on the target signal to obtain a signal spectrum of the target signal, wherein the signal spectrum at least includes: a background noise spectrum.

[0051] In the embodiment of the present invention, the target signal may be a power signal, and the target signal y(t) may be characterized as:

[0052]

[0053] Among them, x(t) is the fundamental signal, η(t) represents the interference signal including background noise and harmonics, I(t) represents the impulse noise, a(t) and represents the instantaneous amplitude and phase angle of the fundamental wave, f0 represents the fundamental wave frequency, and t represents a continuous signal.

[0054] In the embodiment of the present invention, the target signal may be processed by FFT (Fast Fourier Transform) to obtain a signal spectrum Y(k) of the target signal, which may be expressed as:

[0055]

[0056] Wherein, y(n) is the sampling value of the power signal, M is the number of sampling points in the window, j represents the imaginary unit, and n and k represent each sampling point.

[0057] Step S102: determining a spectrum threshold based on the standard deviation of the background noise spectrum, and removing noise components in the signal spectrum based on the spectrum threshold.

[0058] In the embodiment of the present invention, the signal denoising can be performed by the spectrum threshold method. The core of the spectrum threshold method denoising is FFT, which has a small amount of calculation. In addition, since the noise content and size of the actual power signal are unknown, the focus of the spectrum threshold method denoising is how to adaptively set the spectrum threshold so as to accurately distinguish the effective signal component from the noise component.

[0059] In step S102, a spectrum threshold may be determined according to the standard deviation of the background noise spectrum to adaptively set the spectrum threshold, thereby removing the noise component in the signal spectrum according to the spectrum threshold.

[0060] In the embodiment of the present invention, the amplitude of the effective signal component (baseband and broadband component) in the power signal spectrum (ie, signal spectrum) is much larger than the amplitude of the noise component. Therefore, the noise component can be suppressed according to the determined spectrum threshold.

[0061] As shown in formula (3):

[0062]

[0063] Among them, Y'(k) represents the signal spectrum after removing the noise component, N th Represents the spectrum threshold. Components with spectrum values ​​greater than the spectrum threshold are considered valid signal components and are retained, while components with spectrum values ​​less than the spectrum threshold are considered noise components and are set to 0.

[0064] Optionally, before determining the spectrum threshold based on the standard deviation of the background noise spectrum, the method also includes: determining the real part of the noise spectrum and the imaginary part of the noise spectrum of the background noise spectrum, and determining the power spectrum of the background noise and the amplitude spectrum of the background noise based on the real part of the noise spectrum and the imaginary part of the noise spectrum; normalizing the real part of the noise spectrum and the imaginary part of the noise spectrum respectively to obtain the real part of the standard normal noise spectrum and the imaginary part of the standard normal noise spectrum; determining the normalized power spectrum based on the real part of the quasi-normal noise spectrum and the imaginary part of the standard normal noise spectrum, wherein the normalized power spectrum has a chi-square distribution characteristic, and the chi-square distribution characteristic means that the normalized power spectrum obeys a chi-square distribution with a preset degree of freedom.

[0065] In the embodiment of the present invention, the background noise is usually Gaussian white noise, which obeys a normal distribution with a mean of 0. Assuming that n(k) is the background noise, then n(k)~N(0,σ 2 ), where σ is the standard deviation of the background noise.

[0066] The spectrum of the background noise (i.e., the background noise spectrum) is obtained through DFT (Discrete Fourier Transform) / FFT. The spectrum of the background noise is a series of complex numbers and can be expressed as (i.e., determining the real part of the noise spectrum and the imaginary part of the noise spectrum of the background noise):

[0067] N(k)=R(k)+jI(k) (4);

[0068] Where N(k) is the background noise spectrum, R(k) and I(k) are the real and imaginary parts of the spectrum.

[0069] According to the properties of Fourier transform, N(k), R(k), and I(k) all obey the normal distribution, and their mean and standard deviation are the same as n(k). Therefore, R(k)~(0,σ R 2 ),I(k)~(0,σ I 2 ),σ R =σ I =σ, where σ R and σ I denote the standard deviation of the real and imaginary parts respectively.

[0070] The power spectrum and amplitude spectrum of the background noise can be expressed as follows (that is, the power spectrum of the background noise and the amplitude spectrum of the background noise are determined based on the real part of the noise spectrum and the imaginary part of the noise spectrum):

[0071]

[0072] Wherein, P(k) represents the power spectrum of the background noise, and X(k) represents the amplitude spectrum of the background noise.

[0073] Divide the real and imaginary parts of the background noise spectrum by the standard deviation respectively, and normalize them to a standard normal distribution with a variance of 1 (that is, normalize the real part of the noise spectrum and the imaginary part of the noise spectrum respectively to obtain the real part R'(k) and the imaginary part I'(k) of the standard normal noise spectrum):

[0074]

[0075] At this time, the power spectrum obeys the chi-square distribution with a degree of freedom of 2 (that is, based on the real part of the quasi-normal noise spectrum and the imaginary part of the standard normal noise spectrum, the normalized power spectrum P'(k) is determined, and the normalized power spectrum has a chi-square distribution characteristic. The chi-square distribution characteristic means that the normalized power spectrum obeys a chi-square distribution with a preset degree of freedom (for example, 2)):

[0076]

[0077] In an embodiment of the present invention, a probability density curve of a normalized power spectrum of 30dB to 80dB background noise can be simulated by MATLAB (i.e., a programming language and mathematical software) to verify the validity of the above analysis. In the present embodiment, by simulating a signal comprising different background noises (e.g., 30dB, 40dB, 50dB, 60dB, 70dB, 80dB background noise), a normalized power spectrum probability density curve can be obtained, from which it can be seen that the normalized power spectrum of background noise obeys a chi-square distribution with a degree of freedom of 2.

[0078] Optionally, the step of determining a spectrum threshold based on the standard deviation of the background noise spectrum includes: determining a preset confidence and determining a noise reduction threshold based on the preset confidence; determining the standard deviation of the background noise spectrum based on the chi-square distribution characteristics of the normalized power spectrum; and determining the spectrum threshold based on the standard deviation and the noise reduction threshold.

[0079] In the embodiment of the present invention, it is assumed that there is only background noise in the power signal and no pulse noise is included. According to the chi-square distribution characteristics of the background noise, a method for determining the spectrum threshold is proposed. Assuming k is the noise reduction threshold of the normalized power spectrum, according to the characteristics of the chi-square distribution, it can be obtained:

[0080]

[0081] Wherein, P represents the probability that the spectrum value of the noise component in the normalized power spectrum is greater than the noise reduction threshold k, and α represents the confidence level. In this embodiment, when α is smaller, the noise components higher than the noise reduction threshold k are fewer, and the denoising effect is better. k and α are one-to-one corresponding, and can be obtained according to the critical value table of the chi-square distribution (that is, first determine the preset confidence α, and then determine the noise reduction threshold k based on the preset confidence α).

[0082] In the embodiment of the present invention, when the confidence level α is 0.01, the denoising effect is better. At this time, according to the critical value table of the chi-square distribution, it can be obtained that:

[0083] When k is determined, the spectrum threshold can be obtained according to formulas (8) and (5) (that is, based on the chi-square distribution characteristics of the normalized power spectrum, the standard deviation of the background noise spectrum is determined, and then the spectrum threshold is determined based on the standard deviation and the noise reduction threshold):

[0084]

[0085] Therefore, we only need to find the standard deviation σ of the background noise to find the spectrum threshold N. th .

[0086] Optionally, based on the chi-square distribution characteristics of the normalized power spectrum, the step of determining the standard deviation of the background noise spectrum includes: determining the real part of the signal spectrum and the imaginary part of the signal spectrum, and constructing a signal spectrum vector based on the real part of the signal spectrum and the imaginary part of the signal spectrum; determining the vector standard deviation of the signal spectrum vector; determining the normalized signal power spectrum of the target signal based on the real part of the signal spectrum, the imaginary part of the signal spectrum and the vector standard deviation, and determining the signal power spectrum standard deviation of the normalized signal power spectrum; when the signal power spectrum standard deviation is greater than a preset degree of freedom, eliminating the maximum component in the signal spectrum; based on the signal spectrum with the maximum component eliminated, re-determining the signal power spectrum standard deviation until the signal power spectrum standard deviation is less than or equal to the preset degree of freedom, obtaining the target signal power spectrum standard deviation; characterizing the target vector standard deviation used when determining the target signal power spectrum standard deviation as the standard deviation of the background noise spectrum.

[0087] In the embodiment of the present invention, according to the property of the chi-square distribution, the standard deviation of the normalized power spectrum of the background noise satisfies the following formula:

[0088] std(P'(k))=2(11);

[0089] Among them, std(·) represents the standard deviation. Based on this feature, the standard deviation σ of the background noise can be calculated by iteration:

[0090] (1) Obtain the spectrum Y(k) of the power signal according to formula (2), and take the real part and imaginary part of the spectrum (i.e. determine the real part of the signal spectrum and the imaginary part of the signal spectrum):

[0091] R(k)=real[Y(k)], I(k)=imag[Y(k)] (12);

[0092] Among them, R(k) and I(k) are the real and imaginary parts of the spectrum respectively.

[0093] (2) Calculate the normalized power spectrum of the power signal (based on the real part of the signal spectrum and the imaginary part of the signal spectrum, construct the signal spectrum vector H, and then determine the vector standard deviation σ of the signal spectrum vector Y , then determine the normalized signal power spectrum P(k) of the target signal based on the real part of the signal spectrum, the imaginary part of the signal spectrum and the vector standard deviation:

[0094] H=[R(0),…,R(M-1),I(0),…,I(M-1)] (13);

[0095] σ Y =std(H) (14);

[0096]

[0097] Where H is the vector composed of the real and imaginary parts of the signal spectrum, σ Y is the standard deviation of the vector H, and P(k) is the normalized power spectrum of the power signal.

[0098] (3) Calculate the standard deviation of the normalized power spectrum (i.e. determine the standard deviation of the signal power spectrum σ P ):

[0099] σ P =std(P(k)) (16);

[0100] Among them, σ P is the standard deviation of the normalized power spectrum of the signal.

[0101] In the embodiment of the present invention, according to the properties of the chi-square distribution, when the spectrum contains only background noise components, σ P It should be equal to 2. If it is not equal to 2, it means that there is a valid signal component in the spectrum. According to the characteristic that the spectrum value of the valid signal is greater than the noise spectrum value, the component with the largest spectrum value is eliminated (that is, when the standard deviation of the signal power spectrum is greater than the preset degrees of freedom, the maximum component in the signal spectrum is eliminated).

[0102] (4) Return to step (2) until σ P Less than or equal to 2, this indicates that the spectrum contains only background noise components, take σP σ when it is closest to 2 Y As the standard deviation of the background noise, the spectrum threshold N is obtained according to formula (10): th (That is, based on the signal spectrum with the maximum component removed, the signal power spectrum standard deviation is re-determined until the signal power spectrum standard deviation is less than or equal to the preset degrees of freedom, thereby obtaining the target signal power spectrum standard deviation; the target vector standard deviation used in determining the target signal power spectrum standard deviation is represented as the standard deviation of the background noise spectrum).

[0103] In the embodiment of the present invention, the method for determining the spectrum threshold can adaptively adjust the threshold for noises with different noise levels, and can effectively suppress the noise within the measurement frequency band.

[0104] Step S103, performing inverse Fourier transform on the signal spectrum from which the noise component has been removed to obtain an initial noise reduction signal.

[0105] In step S103, the signal spectrum Y'(k) with the noise component removed may be subjected to an inverse Fourier transform to obtain an initial noise reduction signal y'(n), that is, the processed signal spectrum is subjected to an inverse Fourier transform to obtain a noise reduction signal:

[0106]

[0107] Where y'(n) represents the signal after noise reduction, real(·) represents the real part, M is the number of sampling points in the window, and n and k represent the serial numbers of the sampling points.

[0108] Step S104: based on the initial noise reduction signal and the target signal, locate the impulse noise signal in the target signal, and perform smoothing on the impulse noise signal to obtain a target noise reduction signal.

[0109] In the embodiment of the present invention, when the signal contains both background noise and impulse noise, the spectrum of the impulse noise is superimposed on the spectrum of the background noise, and the normalized power spectrum of the noise does not obey the chi-square distribution characteristic, which makes the estimated threshold unreasonable and the denoising effect is not ideal. In order to suppress the influence of impulse noise on the denoising effect, an impulse noise suppression method is proposed to effectively suppress the impulse noise signal in the target signal, and then perform denoising processing through a determined spectrum threshold to obtain the final denoised signal (i.e., the target noise reduction signal).

[0110] In step S104, the impulse noise signal in the target signal may be located according to the initial noise reduction signal and the target signal, and then the impulse noise signal may be smoothed to obtain the target noise reduction signal.

[0111] Optionally, the step of locating the impulse noise signal in the target signal based on the initial denoised signal and the target signal includes: determining the signal residual based on the initial denoised signal and the target signal; determining the residual mean and residual variance of the signal residual, and determining the initial threshold value based on the residual mean, the residual variance and a preset adjustment coefficient; removing elements in the signal residual that are greater than the initial threshold value, and re-determining the initial threshold value based on the signal residual after the elements are removed, until there are no elements in the signal residual that are greater than the initial threshold value, and obtaining the target threshold value; locating the impulse noise signal in the target signal based on the target threshold value.

[0112] In the embodiment of the present invention, it is assumed that the power signal y(n) contains both background noise and impulse noise, and y(n) is denoised using a determined spectrum threshold to obtain a denoised signal (i.e., an initial denoised signal) y'(n), and then the residual Z(n) is calculated (i.e., the signal residual is determined based on the initial denoised signal and the target signal):

[0113] Z(n)=|y(n)-y'(n)|, 0≤n≤M-1 (18);

[0114] Among them, Z(n) represents the removed noise component. According to the characteristics of the pulse noise amplitude being larger than the background noise amplitude, the pulse noise is detected and located by setting an adaptive threshold ξ:

[0115] ξ= μ +βσ (19);

[0116] Among them, μ and σ are the mean and variance of Z(n) respectively, and β is the adjustment coefficient.

[0117] In this embodiment, when β=6, the probability of misjudgment of impulse noise is only 2×10 -9 , so β=6 is taken. The pulse noise positioning process is as follows: the mean and variance of Z(n) can be calculated first to obtain the initial value of ξ (that is, the residual mean and residual variance of the signal residual are determined, and the initial threshold value is determined based on the residual mean, residual variance and preset adjustment coefficient). Then, the elements higher than ξ in Z(n) are removed, and the threshold value ξ is recalculated. The above process is repeated until there are no elements higher than ξ in Z(n), and the final ξ is obtained (that is, the elements greater than the initial threshold value in the signal residual are removed, and the initial threshold value is re-determined based on the signal residual of the removed elements, until there are no elements greater than the initial threshold value in the signal residual, and the target threshold value is obtained). The points higher than ξ in Z(n) are taken and expressed as (that is, based on the target threshold value, the impulse noise signal in the target signal is positioned):

[0118] I=[i h ]h=1,2,…,H (20);

[0119] Among them, i h is the position of the hth group of impulse noise, and H is the number of groups.

[0120] Optionally, the step of smoothing the impulse noise signal to obtain a target noise reduction signal includes: based on the position index of the located impulse noise signal, locally sampling the target signal to obtain a local sampling sequence; based on the fitting parameter matrix and the local sampling sequence, determining a smoothed sampling sequence; replacing the local sampling sequence in the target signal with the smoothed sampling sequence to obtain a smoothed noise reduction signal; and based on a spectrum threshold, denoising the smoothed noise reduction signal to obtain a target noise reduction signal.

[0121] Another optional embodiment, after locally sampling the target signal based on the position index of the located pulse noise signal to obtain a local sampling sequence, also includes: performing polynomial fitting on the local sampling sequence to obtain a position parameter matrix; determining an initial fitting parameter matrix based on the position parameter matrix and the signal sampling value matrix, wherein the initial fitting parameter matrix includes a weighting coefficient diagonal matrix and a fitting weight diagonal matrix; correcting the weighting coefficient diagonal matrix to obtain a target weighting coefficient diagonal matrix; and determining the fitting parameter matrix based on the target weighting coefficient diagonal matrix.

[0122] In the embodiment of the present invention, it is assumed that D = {(x n ,y n ), n = 1, 2, ..., N} is the local sampling sequence of y(n) (i.e., based on the position index of the located impulse noise signal, the target signal is locally sampled to obtain a local sampling sequence), where x n is the position parameter, y n For x n The signal sampling value at , N is the number of local sampling sequences, polynomial fitting is performed on D, and the fitting order is set to M, which is expressed as (that is, polynomial fitting is performed on the local sampling sequence to obtain the position parameter matrix X):

[0123] Y = Xα (21);

[0124] Where X∈R N×(M+1) is the position parameter matrix, as shown in the following formula (22), Y∈R N×1 is the signal sampling value matrix, α∈R (M+1)×1 To fit the parameter matrix, the weighted least squares method is used to solve equation (21) and obtain α WLS (That is, based on the position parameter matrix and the signal sampling value matrix, the initial fitting parameter matrix is ​​determined, where the initial fitting parameter matrix includes a weighting coefficient diagonal matrix and a fitting weight diagonal matrix), as shown in formula (23), according to α WLS The local sampling sequence after fitting and smoothing is obtained, that is, the local sampling sequence after removing the impulse noise.

[0125]

[0126] α WLS =(X T WΔX) -1 X T WΔY (23);

[0127] Where Δ∈R N×N is the diagonal matrix of weight coefficients, W∈R N×N is the fitting weight diagonal matrix, the fitting weight ω of the kth point k As shown in the formula:

[0128]

[0129] Among them, x0 is the location parameter of the midpoint of the sampling window, x k is the position parameter of the kth point, and d(x0) represents the farthest distance from x0 to other sampling points in the window, that is:

[0130] d(x0)=max{|x0-x k |}k=1,2,…,N (25);

[0131] Substituting equation (23) into equation (21) yields the smoothed local sampling sequence Y WLS :

[0132] Y WLS =Xα WLS (26);

[0133] Formula (26) can be written as formula (27):

[0134] Y WLS =Ψ WLS Y (27);

[0135] Among them, WLS ∈R N×N is the fitting parameter matrix, as shown in formula (28):

[0136] Ψ WLS =X(X T WX) -1 X T WΔ (28);

[0137] In this embodiment, the correction coefficient δ of the kth point in Δ is k Set to:

[0138]

[0139] That is, the weighting coefficients of the N0 sampling points in the middle of the sampling window are set to 0 to suppress the impulse noise in this area. When N (i.e., the number of local sampling sequences) and N0 (i.e., the number of sampling points with a fitting weight of 0, i.e., the number of sampling points affected by impulse noise) are determined, Δ remains unchanged, and the fitting parameter matrix Ψ WLS The weighting coefficient diagonal matrix is ​​also unchanged (that is, the weighting coefficient diagonal matrix is ​​modified to obtain the target weighting coefficient diagonal matrix, and the fitting parameter matrix is ​​determined based on the target weighting coefficient diagonal matrix).

[0140] For example, taking the hth group of impulse noise as an example, the process of fitting and smoothing is as follows: after obtaining the index i of the hth group of impulse noise h After that, take i h The local sampling sequence Y with length N as the center h , as shown in formula (30): The sampling sequence after fitting smoothing It can be obtained by formula (27), and Y0 is shown in formula (31) (that is, based on the fitting parameter matrix and the local sampling sequence, the smoothed sampling sequence is determined, and the smoothed sampling sequence replaces the local sampling sequence in the target signal to obtain a smoothed denoised signal).

[0141]

[0142] in, To obtain WLS The submatrix obtained by the middle N0 row replaces the corresponding local sampling sequence in y(n) with Y0, eliminating the waveform distortion caused by the impulse noise in y(n).

[0143] In this embodiment, after the fitting parameters are determined, It remains unchanged and can be calculated in advance after determining the fitting parameters. Therefore, suppressing a set of impulse noises only involves one matrix multiplication. Compared with other methods based on nonlinear fitting, such as robust local smoothing, it can greatly reduce the amount of calculation.

[0144] In the embodiment of the present invention, the smoothed noise reduction signal may be denoised according to the determined spectrum threshold to obtain a target noise reduction signal.

[0145] Another optional specific implementation is described in detail below.

[0146] Figure 2 is a schematic diagram of an optional signal noise suppression process according to an embodiment of the present invention, such as Figure 2 As shown, the process includes the following: first obtain the power signal spectrum Y(k), calculate the normalized power spectrum P(k) of the power signal, and then calculate the power spectrum standard deviation σ P , and then judge whether it is enough P≤2, if not, then after eliminating the maximum power spectrum, recalculate the normalized power spectrum P(k) of the power signal, if yes, then obtain the noise standard deviation σ and the noise reduction threshold N th , and then calculate the initial denoised signal y'(n) and residual Z(n), and set the adaptive threshold ξ=μ+βσ, take the maximum point in Z(n) higher than ξ as the impulse noise position point, obtain the position set I of the impulse signal in y(n), and fit and smooth the impulse signal in y(n), and then use the denoising threshold N th Get the signal after noise reduction.

[0147] In the embodiment of the present invention, the specific steps of signal noise suppression are as follows:

[0148] (1) Using the spectrum threshold determination method to obtain the spectrum threshold N th , and preliminarily denoise the power signal y(n);

[0149] (2) Based on the residual Z(n), the impulse noise position set I is obtained, and the impulse noise in y(n) is suppressed by applying the impulse noise suppression method to obtain y”(n);

[0150] (3) Using N th De-noising is performed on y”(n) to obtain the final denoised signal.

[0151] In this embodiment, because the threshold N th The determination is based on the premise that the signal contains only background noise. If the signal also contains impulse noise, the denoising effect will be poor. Therefore, the threshold N is first used. th Perform preliminary noise reduction on the signal, then use the residual to locate the impulse noise, and then remove the impulse noise in the signal to obtain y" (n). Therefore, the current y" (n) only contains background noise. Finally, use the threshold N th De-noising y”(n) is performed to obtain a de-noised signal. Furthermore, when de-noising y”(n) is performed, it is not necessary to recalculate the threshold N. th , directly use the previously determined threshold N th This is because the threshold N th It is obtained when there is impulse noise in the signal. Due to the superposition of the spectrum of the impulse noise and the spectrum of the background noise, the overall spectrum value becomes larger, which makes the obtained threshold value larger. Therefore, when this threshold is used to denoise y" (n), the denoising effect is better and the threshold value can be avoided from being calculated again, thereby reducing the amount of calculation.

[0152] The signal noise suppression method is described in detail below through simulation tests and field recording data verification.

[0153] Assume that the parameters in the signal noise suppression method are set to: N = 27, N0 = 5, the fitting order is 2nd order, and the window length is 4s. The sampling rate of the signal during the simulation test is set to f s =1kHz.

[0154] (1) Simulation test:

[0155] This test illustrates the denoising effect of the signal noise suppression method by comparing the phasor measurement accuracy before and after denoising. The test signal parameters are shown in Table 1. The phasor measurement algorithm used is a phasor measurement algorithm based on a complex coefficient bandpass filter. In the test signal, 40dB background noise and random pulse noise are added. The amplitude of the random pulse noise in the voltage signal is randomly distributed in the range of 50V-80V, and the amplitude of the random pulse noise in the current signal is randomly distributed in the range of 0.5A-1A, with a coverage rate of 1%.

[0156] Table 1 Standard test parameters

[0157]

[0158]

[0159] The test results of the current signal are shown in Table 2, and the test results of the voltage signal are shown in Table 3 and Table 4. Since the Phasor Measurement Unit (PMU) does not need to calculate the frequency and frequency change rate of the current signal, the corresponding measurement results are not given in the table. In addition, since the PMU standard has not yet specified the error requirements for noise testing, the evaluation indicators are not given in the table.

[0160] Through the analysis of the measurement results, it can be seen that after the signal noise suppression method is used for noise reduction, the amplitude and phase angle accuracy are generally improved by 8-10 times, and the frequency and frequency change rate accuracy are improved by more than 10 times. Due to the influence of DFT / FFT frequency leakage, the denoising effect is relatively poor during the frequency ramp test (frequency linear change), but the measurement accuracy is also improved by more than 7 times. Therefore, the signal noise suppression method proposed in this embodiment has a good denoising effect and can significantly improve the accuracy of phasor measurement.

[0161] Table 2 Current signal test results

[0162]

[0163] Table 3 Voltage signal amplitude and phase angle test results

[0164]

[0165] Table 4 Voltage signal frequency and frequency change rate test results

[0166]

[0167] (2) Field recording data verification:

[0168] The effectiveness of the signal noise suppression method proposed in this embodiment is further verified by field recording data. Figure 3 is a schematic diagram of an optional distribution network voltage recording waveform diagram according to an embodiment of the present invention, such as Figure 3 As shown, a voltage recording waveform of a distribution network is displayed (with time / s as the horizontal axis and amplitude / V as the vertical axis). Figure 4 is a schematic diagram of an optional amplitude measurement result according to an embodiment of the present invention, such as Figure 4 As shown, the amplitude measurement results before denoising and after denoising by the signal noise suppression method are shown (with time / s as the horizontal axis and amplitude / V as the vertical axis). Figure 5 is a schematic diagram of an optional frequency measurement result according to an embodiment of the present invention, such as Figure 5 As shown, the frequency measurement results before denoising and after denoising by the signal noise suppression method are shown (with time / s as the horizontal axis and amplitude / V as the vertical axis).

[0169] from Figure 4 and Figure 5 It can be seen that due to the influence of noise, the un-denoised measurement results have many glitches in both amplitude and frequency, and there is obvious jitter. After denoising by the signal noise suppression method, the amplitude and frequency measurement results have no obvious jitter and glitches, and the waveform is relatively smooth. It can be seen that the signal noise suppression method in this embodiment has a good denoising effect and can effectively suppress the influence of field signal noise.

[0170] In the embodiment of the present invention, by establishing a normalized power spectrum chi-square distribution model of background noise, based on its distribution characteristics and according to the characteristic that the spectrum value of the effective signal is greater than the spectrum value of the noise, the standard deviation of the background noise can be obtained by continuously eliminating the component with the largest spectrum value, and then the spectrum threshold is obtained, wherein the spectrum threshold can adaptively adjust the threshold for noises with different noise levels, and can effectively suppress the noise in the measurement band. Moreover, according to the characteristic that the amplitude of the pulse noise is much greater than the background noise, the background noise and the pulse noise can be accurately distinguished and the pulse noise can be located based on the residual and the threshold value. In addition, based on the idea of ​​nonlinear fitting, a pulse noise suppression method is proposed, by setting a fixed fitting parameter matrix, so that only one matrix multiplication is required to suppress a group of pulse noises, which can reduce the amount of calculation.

[0171] The following is a detailed description in conjunction with another embodiment.

[0172] Embodiment 2

[0173] A signal noise suppression device provided in this embodiment includes multiple implementation units, each implementation unit corresponds to each implementation step in the above-mentioned embodiment 1.

[0174] Figure 6 is a schematic diagram of an optional signal noise suppression device according to an embodiment of the present invention, such as Figure 6 As shown, the suppression device may include: a transformation unit 60, a determination unit 61, an inverse transformation unit 62, and a positioning unit 63.

[0175] The transform unit 60 is used to collect the target signal and perform Fourier transform on the target signal to obtain a signal spectrum of the target signal, wherein the signal spectrum at least includes: a background noise spectrum;

[0176] A determination unit 61, configured to determine a spectrum threshold based on a standard deviation of a background noise spectrum, and remove noise components from a signal spectrum based on the spectrum threshold;

[0177] An inverse transform unit 62 is used to perform an inverse Fourier transform on the signal spectrum after the noise component is removed to obtain an initial noise reduction signal;

[0178] The positioning unit 63 is used to locate the impulse noise signal in the target signal based on the initial noise reduction signal and the target signal, and to smooth the impulse noise signal to obtain the target noise reduction signal.

[0179] The above-mentioned suppression device can collect the target signal through the transformation unit 60, and perform Fourier transformation on the target signal to obtain the signal spectrum of the target signal, determine the spectrum threshold based on the standard deviation of the background noise spectrum through the determination unit 61, and remove the noise component in the signal spectrum based on the spectrum threshold, perform Fourier inverse transformation on the signal spectrum after removing the noise component through the inverse transformation unit 62 to obtain the initial noise reduction signal, locate the impulse noise signal in the target signal based on the initial noise reduction signal and the target signal through the positioning unit 63, and smooth the impulse noise signal to obtain the target noise reduction signal. In the embodiment of the present invention, by establishing a normalized power spectrum chi-square distribution model of the background noise, the standard deviation of the background noise spectrum can be obtained, and then the spectrum threshold is determined to remove the noise component according to the spectrum threshold, so as to effectively suppress the noise in the measurement band. In addition, the impulse noise signal in the target signal can be located according to the initial noise reduction signal after removing the noise component and the target signal, so as to smooth the impulse noise signal, and finally obtain the target noise reduction signal, which not only reduces the amount of calculation, but also can effectively suppress the noise in the signal, thereby solving the technical problem that the signal noise cannot be effectively suppressed in the related art.

[0180] Optionally, the suppression device also includes: a first determination module, used to determine the real part of the noise spectrum and the imaginary part of the noise spectrum of the background noise spectrum before determining the spectrum threshold based on the standard deviation of the background noise spectrum, and determine the power spectrum of the background noise and the amplitude spectrum of the background noise based on the real part of the noise spectrum and the imaginary part of the noise spectrum; a first processing module, used to normalize the real part of the noise spectrum and the imaginary part of the noise spectrum, respectively, to obtain the real part of the standard normal noise spectrum and the imaginary part of the standard normal noise spectrum; a second determination module, used to determine the normalized power spectrum based on the real part of the quasi-normal noise spectrum and the imaginary part of the standard normal noise spectrum, wherein the normalized power spectrum has a chi-square distribution characteristic, and the chi-square distribution characteristic means that the normalized power spectrum obeys a chi-square distribution with a preset degree of freedom.

[0181] Optionally, the determination unit includes: a third determination module, used to determine a preset confidence and determine a noise reduction threshold based on the preset confidence; a fourth determination module, used to determine the standard deviation of the background noise spectrum based on the chi-square distribution characteristics of the normalized power spectrum; and a fifth determination module, used to determine the spectrum threshold based on the standard deviation and the noise reduction threshold.

[0182] Optionally, the fourth determination module includes: a first determination submodule, used to determine the real part of the signal spectrum and the imaginary part of the signal spectrum, and construct a signal spectrum vector based on the real part of the signal spectrum and the imaginary part of the signal spectrum; a second determination submodule, used to determine the vector standard deviation of the signal spectrum vector; a third determination submodule, used to determine the normalized signal power spectrum of the target signal based on the real part of the signal spectrum, the imaginary part of the signal spectrum and the vector standard deviation, and determine the signal power spectrum standard deviation of the normalized signal power spectrum; a first elimination submodule, used to eliminate the maximum component in the signal spectrum when the signal power spectrum standard deviation is greater than a preset degree of freedom; a fourth determination submodule, used to re-determine the signal power spectrum standard deviation based on the signal spectrum with the maximum component eliminated, until the signal power spectrum standard deviation is less than or equal to the preset degree of freedom, to obtain the target signal power spectrum standard deviation; a first characterization submodule, used to characterize the target vector standard deviation used when determining the target signal power spectrum standard deviation as the standard deviation of the background noise spectrum.

[0183] Optionally, the positioning unit includes: a sixth determination module, used to determine the signal residual based on the initial noise reduction signal and the target signal; a seventh determination module, used to determine the residual mean and residual variance of the signal residual, and determine the initial threshold value based on the residual mean, the residual variance and a preset adjustment coefficient; a first removal module, used to remove elements in the signal residual that are greater than the initial threshold value, and re-determine the initial threshold value based on the signal residual of the removed elements, until there are no elements in the signal residual that are greater than the initial threshold value, and the target threshold value is obtained; a first positioning module, used to locate the pulse noise signal in the target signal based on the target threshold value.

[0184] Optionally, the positioning unit also includes: a first sampling module, used to perform local sampling of the target signal based on the position index of the located pulse noise signal to obtain a local sampling sequence; an eighth determination module, used to determine a smoothed sampling sequence based on a fitting parameter matrix and a local sampling sequence; a first replacement module, used to replace the local sampling sequence in the target signal with the smoothed sampling sequence to obtain a smoothed noise-reduced signal; and a first denoising module, used to denoise the smoothed noise-reduced signal based on a spectrum threshold to obtain a target noise-reduced signal.

[0185] Optionally, the suppression device also includes: a first fitting module, which is used to perform local sampling of the target signal based on the position index of the located pulse noise signal, and after obtaining the local sampling sequence, perform polynomial fitting on the local sampling sequence to obtain a position parameter matrix; a ninth determination module, which is used to determine the initial fitting parameter matrix based on the position parameter matrix and the signal sampling value matrix, wherein the initial fitting parameter matrix includes a weighting coefficient diagonal matrix and a fitting weight diagonal matrix; a first correction module, which is used to correct the weighting coefficient diagonal matrix to obtain a target weighting coefficient diagonal matrix; and a tenth determination module, which is used to determine the fitting parameter matrix based on the target weighting coefficient diagonal matrix.

[0186] The above-mentioned suppression device may also include a processor and a memory. The above-mentioned transformation unit 60, determination unit 61, inverse transformation unit 62, positioning unit 63, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions.

[0187] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the kernel parameters are adjusted to locate the impulse noise signal in the target signal based on the initial noise reduction signal and the target signal, and the impulse noise signal is smoothed to obtain the target noise reduction signal.

[0188] The above-mentioned memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one storage chip.

[0189] The present invention also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program that is initialized with the following method steps: collecting a target signal and performing Fourier transform on the target signal to obtain a signal spectrum of the target signal, wherein the signal spectrum at least includes: a background noise spectrum; determining a spectrum threshold based on the standard deviation of the background noise spectrum, and removing noise components in the signal spectrum based on the spectrum threshold; performing inverse Fourier transform on the signal spectrum with the noise components removed to obtain an initial noise reduction signal; locating an impulse noise signal in the target signal based on the initial noise reduction signal and the target signal, and smoothing the impulse noise signal to obtain a target noise reduction signal.

[0190] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium including a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned signal noise suppression method.

[0191] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by one or more processors, the one or more processors implement the above-mentioned signal noise suppression method.

[0192] Figure 7 1 is a hardware structure block diagram of an electronic device (or mobile device) for a method for suppressing signal noise according to an embodiment of the present invention. Figure 7 As shown, the electronic device may include one or more processors (e.g., Figure 7 The processor 702a, processor 702b, ..., processor 702n, etc., which may include but are not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 704 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, a keyboard, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 7 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 7 More or fewer components as shown, or with Figure 7 Different configurations shown.

[0193] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0194] The embodiments or examples of the present disclosure are not exhaustive, but are only illustrative of some embodiments or examples, and are not intended to be specific limitations on the scope of protection of the present disclosure. In the absence of contradiction, each step in a certain embodiment or example can be implemented as an independent example, and the steps can be combined arbitrarily. For example, the scheme after removing some steps in a certain embodiment or example can also be implemented as an independent example, and the order of the steps in a certain embodiment or example can be arbitrarily exchanged. In addition, the optional methods or optional examples in a certain embodiment or example can be combined arbitrarily; in addition, the various embodiments or examples can be combined arbitrarily, for example, some or all steps of different embodiments or examples can be combined arbitrarily, and a certain embodiment or example can be combined arbitrarily with the optional methods or optional examples of other embodiments or examples.

[0195] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0196] In the several embodiments provided by the present invention, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0197] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0198] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0199] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0200] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for suppressing signal noise, characterized in that: include: Collecting a target signal, and performing Fourier transform on the target signal to obtain a signal spectrum of the target signal, wherein the signal spectrum at least includes: a background noise spectrum; Determining a spectrum threshold based on a standard deviation of the background noise spectrum, and removing noise components in the signal spectrum based on the spectrum threshold; Performing inverse Fourier transform on the signal spectrum from which the noise component is removed to obtain an initial noise reduction signal; Based on the initial noise reduction signal and the target signal, the impulse noise signal in the target signal is located, and the impulse noise signal is smoothed to obtain a target noise reduction signal.

2. The suppression method according to claim 1, characterized in that: Before determining the spectrum threshold based on the standard deviation of the background noise spectrum, the method further includes: Determine a real part of a noise spectrum and an imaginary part of a noise spectrum of the background noise spectrum, and determine a power spectrum of the background noise and an amplitude spectrum of the background noise based on the real part of the noise spectrum and the imaginary part of the noise spectrum; Normalizing the real part of the noise spectrum and the imaginary part of the noise spectrum respectively to obtain the real part of the standard normal noise spectrum and the imaginary part of the standard normal noise spectrum; Based on the real part of the quasi-normal noise spectrum and the imaginary part of the standard normal noise spectrum, a normalized power spectrum is determined, wherein the normalized power spectrum has a chi-square distribution characteristic, and the chi-square distribution characteristic means that the normalized power spectrum obeys a chi-square distribution with a preset degree of freedom.

3. The suppression method according to claim 2, characterized in that: The step of determining a spectrum threshold based on the standard deviation of the background noise spectrum comprises: Determining a preset confidence level, and determining a noise reduction threshold based on the preset confidence level; Determining the standard deviation of the background noise spectrum based on the chi-square distribution characteristic of the normalized power spectrum; The spectrum threshold is determined based on the standard deviation and the noise reduction threshold.

4. The suppression method according to claim 3, characterized in that: The step of determining the standard deviation of the background noise spectrum based on the chi-square distribution characteristic of the normalized power spectrum comprises: Determine a signal spectrum real part and a signal spectrum imaginary part of the signal spectrum, and construct a signal spectrum vector based on the signal spectrum real part and the signal spectrum imaginary part; Determining a vector standard deviation of the signal spectrum vector; Determine a normalized signal power spectrum of the target signal based on the real part of the signal spectrum, the imaginary part of the signal spectrum and the vector standard deviation, and determine a signal power spectrum standard deviation of the normalized signal power spectrum; When the standard deviation of the signal power spectrum is greater than the preset degree of freedom, removing the maximum component in the signal spectrum; Based on the signal spectrum with the maximum component removed, re-determining the signal power spectrum standard deviation until the signal power spectrum standard deviation is less than or equal to the preset degree of freedom, thereby obtaining a target signal power spectrum standard deviation; The target vector standard deviation used when determining the target signal power spectrum standard deviation is characterized as the standard deviation of the background noise spectrum.

5. The suppression method according to claim 1, characterized in that: The step of locating the impulse noise signal in the target signal based on the initial noise reduction signal and the target signal comprises: Determining a signal residual based on the initial noise reduction signal and the target signal; Determine a residual mean and a residual variance of the signal residual, and determine an initial threshold value based on the residual mean, the residual variance and a preset adjustment coefficient; Removing elements greater than the initial threshold value from the signal residual, and re-determining the initial threshold value based on the signal residual after removing the elements, until no elements greater than the initial threshold value exist in the signal residual, thereby obtaining a target threshold value; The impulse noise signal in the target signal is located based on the target threshold value.

6. The suppression method according to claim 1, characterized in that: The step of smoothing the impulse noise signal to obtain a target noise reduction signal comprises: Based on the located position index of the impulse noise signal, locally sample the target signal to obtain a local sampling sequence; Determine a smooth sampling sequence based on the fitting parameter matrix and the local sampling sequence; Replacing the local sampling sequence in the target signal with the smoothed sampling sequence to obtain a smoothed noise-reduced signal; Based on the spectrum threshold, the smoothed noise reduction signal is denoised to obtain the target noise reduction signal.

7. The suppression method according to claim 6, characterized in that: After locally sampling the target signal based on the located position index of the impulse noise signal to obtain a local sampling sequence, the method further includes: Performing polynomial fitting on the local sampling sequence to obtain a position parameter matrix; Based on the position parameter matrix and the signal sampling value matrix, an initial fitting parameter matrix is ​​determined, wherein the initial fitting parameter matrix includes a weighting coefficient diagonal matrix and a fitting weight diagonal matrix; Modifying the weighting coefficient diagonal matrix to obtain a target weighting coefficient diagonal matrix; The fitting parameter matrix is ​​determined based on the target weighting coefficient diagonal matrix.

8. A signal noise suppression device, characterized in that: include: A transform unit, used for collecting a target signal and performing Fourier transform on the target signal to obtain a signal spectrum of the target signal, wherein the signal spectrum at least includes: a background noise spectrum; A determination unit, configured to determine a spectrum threshold based on a standard deviation of the background noise spectrum, and remove noise components from the signal spectrum based on the spectrum threshold; An inverse transform unit, configured to perform an inverse Fourier transform on the signal spectrum from which the noise component has been removed, to obtain an initial noise reduction signal; A positioning unit is used to locate the impulse noise signal in the target signal based on the initial noise reduction signal and the target signal, and to smooth the impulse noise signal to obtain a target noise reduction signal.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the signal noise suppression method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: It includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the signal noise suppression method described in any one of claims 1 to 7.

Images