Electrocardiogram signal noise suppression method and system based on singular spectrum analysis mean strategy

Abstract

The present application relates to the technical field of signal processing, in particular to a kind of ECG signal noise suppression method and system based on singular spectrum analysis mean strategy, method includes: obtaining ECG noise signal sequence, ECG noise signal sequence is converted to frequency domain and obtains ECG signal spectrum;Frequency response matrix is constructed based on ECG noise signal spectrum, target function is established based on frequency response matrix with the orientation of inhibiting spectral leakage, the target vector is obtained by solving target function, the optimal window is determined according to target vector;Multiple noise suppression signals are generated based on ECG noise signal sequence and optimal window, noise suppression ECG signal is generated based on multiple noise suppression signals;The present application can suppress noise low-frequency component, to improve noise suppression effect.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of signal processing, in particular to an electrocardiosignal noise suppression method and system based on singular spectrum analysis mean strategy. BACKGROUND

[0002] Gaussian white noise is ubiquitous, so the signal obtained from the actual system will inevitably be disturbed by additive Gaussian white noise in the time domain representation. Because the potential trend and time-frequency characteristics of the observed signal are often disturbed by Gaussian white noise, removing Gaussian white noise is a classic problem in digital signal processing. In many instruments and measurement systems such as power measurement systems, biomedical measurement systems and partial discharge measurement systems, it is an urgent technical problem to suppress noise and maintain the important structure of the original signal as much as possible. SUMMARY

[0003] The present application aims to provide an electrocardiosignal noise suppression method and system based on singular spectrum analysis mean strategy, which aims to suppress the low-frequency components of noise and improve the noise suppression effect.

[0004] In order to achieve the above-mentioned purpose, the present application provides the following technical scheme:

[0005] In a first aspect, the present application provides an electrocardiosignal noise suppression method based on singular spectrum analysis mean strategy, which comprises the following steps:

[0006] S100, obtaining an electrocardiosignal noise sequence, converting the electrocardiosignal noise sequence to the frequency domain to obtain an electrocardiosignal spectrum;

[0007] S200, constructing a frequency response matrix based on the electrocardiosignal noise spectrum, establishing a target function oriented to suppress spectral leakage based on the frequency response matrix, solving the target function to obtain a target vector, and determining an optimal window according to the target vector;

[0008] S300, generating a plurality of noise suppression signals based on the electrocardiosignal noise sequence and the optimal window, and generating a noise suppression electrocardiosignal based on the plurality of noise suppression signals.

[0009] Optionally, in S200, the frequency response matrix is constructed based on the electrocardiosignal noise spectrum, the target function oriented to suppress spectral leakage is established based on the frequency response matrix, the target function is solved to obtain the target vector, and the optimal window is determined according to the target vector, which comprises:

[0010] S210, obtaining the electrocardiosignal noise spectrum, and finding a plurality of maximum points in the electrocardiosignal noise spectrum;

[0011] S220, a plurality of frequency bands are adaptively divided according to a plurality of maximum value points, a frequency response corresponding to each frequency band is determined, and a frequency response matrix is constructed;

[0012] S230, a target function oriented to suppress spectral leakage is established based on the frequency response matrix, the target function is solved to obtain a target matrix, and an estimated matrix is obtained by interpolating and estimating column vectors of the target matrix;

[0013] S240, the electrocardio noise signal sequence is converted into a trajectory matrix, and an optimal window is obtained according to the trajectory matrix and an estimated vector in the estimated matrix.

[0014] Optionally, the frequency response formula is:

[0015] ;

[0016] wherein, the constructor is represented, the independent variable of the constructor is represented, when , when , when , ;

[0017] satisfy:

[0018] ;

[0019] wherein, the frequency response of the kth maximum value point is represented, the number of maximum value points is represented, and k represents the index of the maximum value point, ; , the kth frequency point and the k+1th frequency point are represented respectively, and the kth transition band interval and the k+1th transition band interval are represented respectively.

[0020] Optionally, in S230, the target function oriented to suppress spectral leakage is established based on the frequency response matrix, and the target function is solved to obtain the target matrix; the column vectors of the target matrix are interpolated and estimated to obtain the estimated matrix, comprising:

[0021] the target function is established based on the frequency response matrix:

[0022] ;

[0023] the frequency response matrix is obtained by solving the target function: ; wherein, the frequency response matrix is represented. is a target matrix; is a multiplier matrix composed of Lagrange multipliers; indicates an index of a maximum point, i indicates a row number of an element in the multiplier matrix, j indicates a column number of the element in the multiplier matrix, i = 0, 1,..., L-1, j = 0, 1,..., L-1; is a discrete-time impulse function;

[0024] singular value decomposition is performed on the frequency response matrix to obtain a left singular matrix, a right singular matrix and a diagonal matrix, and a transposed matrix of the left singular matrix and the right singular matrix are multiplied to obtain the target matrix.

[0025] Optionally, in S240, the ECG noise signal sequence is converted into a trajectory matrix, and an optimal window is obtained according to the trajectory matrix and an estimation vector in an estimation matrix, including:

[0026] The ECG noise signal sequence is converted into a first trajectory matrix;

[0027] A first autocorrelation matrix is obtained based on the first trajectory matrix, and singular value decomposition is performed on the first autocorrelation matrix to obtain a first singular matrix:

[0028] A first singular vector of the first singular matrix is determined, a ratio of an L2 norm of the estimation vector to an L2 norm of the first singular vector is taken as a window coefficient of the optimal window, and thus the optimal window is obtained.

[0029] Optionally, in S300, the ECG noise signal sequence and the optimal window are used to generate a plurality of noise suppression signals, and a noise suppression ECG signal is generated based on the plurality of noise suppression signals, including:

[0030] In S310, a plurality of different windows are generated based on a Euclidean metric space with the optimal window as a center:

[0031] In S320, the ECG noise signal sequence is decomposed using singular spectrum analysis based on the generated windows to obtain a preliminary noise suppression signal;

[0032] In S330, a maximum correlation is set, and it is determined whether a correlation between the preliminary noise suppression signal and the noise ECG signal sequence is less than the maximum correlation, if yes, a value of the correlation is assigned to the maximum correlation, and the preliminary noise suppression signal is taken as the ECG noise signal sequence to continue S320; if the correlation is equal to the maximum correlation, a noise suppression signal of the window is obtained;

[0033] In S340, the noise suppression signals of the plurality of windows are subjected to mean value processing to obtain a final noise suppression ECG signal.

[0034] Optionally, in S320, the electrocardio noise signal sequence is decomposed using singular spectrum analysis based on the generated window to obtain a preliminary noise suppression signal, including:

[0035] In S321, the electrocardio noise signal sequence is segmented into a plurality of segmented signals, and each segmented signal is respectively multiplied with each window to obtain a plurality of window vectors.

[0036] In S322, a second trajectory matrix is generated based on the plurality of window vectors, the second trajectory matrix is multiplied with its eigenvector matrix to obtain a second autocorrelation matrix, and singular value decomposition is performed on the second autocorrelation matrix to obtain a second diagonal matrix and a second singular matrix.

[0037] In S323, a noise suppression matrix is generated based on the second trajectory matrix, the second diagonal matrix and the second singular matrix, and a weighted diagonal average is performed on the noise suppression matrix to obtain a preliminary noise suppression signal of the window.

[0038] In a second aspect, an embodiment of the present application provides an electrocardio signal noise suppression system based on a singular spectrum analysis mean value strategy, and the system comprises:

[0039] at least one processor;

[0040] at least one memory for storing at least one program;

[0041] When the at least one program is executed by the at least one processor, the at least one processor implements the method according to any one of the above.

[0042] The present application has the following beneficial effects: the present application finds an optimal window to suppress the spectral leakage of a noise electrocardio signal sequence, so as to improve the separability of electrocardio information and noise; then, a noise sample is generated based on the optimal window to realize mean value noise suppression. The present application can suppress noise low-frequency components, thereby improving the noise suppression effect. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments. Obviously, the drawings described below are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0044] Figure 1 is a flowchart of an electrocardio signal noise suppression method based on a singular spectrum analysis mean value strategy in an embodiment of the present application;

[0045] Figure 2 is a structural diagram of an electrocardio signal noise suppression system based on a singular spectrum analysis mean value strategy in an embodiment of the present application. DETAILED DESCRIPTION

[0046] The concept, specific structure and generated technical effects of the present application will be described clearly and completely in combination with the embodiments and drawings below, so as to fully understand the purpose, scheme and effects of the present application. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0047] In the related art, since the screening process in empirical mode decomposition is an adaptive iterative process of extracting each intrinsic time scale of the original signal from high-frequency components to low-frequency components, the empirical mode decomposition can be used for noise estimation of many measurement systems and can also be used for applications requiring nonlinear and non-stationary signal processing. The traditional denoising method based on empirical mode always discards the first intrinsic mode function with the highest frequency. However, the noise is not only dominant in the first intrinsic mode function, but also dominant in other high-frequency intrinsic mode functions. In order to obtain better denoising effect, some methods first find the boundary of the noise-dominant intrinsic mode function and the information-dominant intrinsic mode function by some strategies. Then, the information-dominant intrinsic mode function is used to construct the denoised signal.

[0048] On the other hand, singular spectrum analysis can reveal the frequency spectrum demarcation point between the noise part and the information part inside the signal by adaptively decomposing the observed signal, so the singular spectrum analysis is widely used for noise suppression in signal processing. However, the signal denoising based on singular spectrum analysis is also limited by the spectral leakage of singular spectrum analysis, so the separability of signal and noise needs to be improved. Afterwards, many works are devoted to the spectral and statistical differences between signal and noise to suppress the negative effects of spectral leakage.

[0049] Time domain de-noising of signal in Gaussian white noise is the most frequent and basic application in signal processing field, and signal adaptive decomposition separates noise and electrocardio information by decomposing noise electrocardio signal sequence into multiple decomposition components. Since energy characteristics of electrocardio signal are mainly concentrated in low frequency region, components in high frequency region are mainly noise. Based on this point, existing noise suppression methods separate decomposition components into high frequency components and low frequency components through some strategies. High frequency components are considered as noise, and low frequency components are considered as useful information. However, the existing noise suppression methods still have shortcomings. The separability of noise and electrocardio signal characteristics of the existing noise suppression method is seriously restricted by spectral confusion, that is, different frequency band components appear mutual confusion of spectral leakage phenomenon in the frequency spectrum, resulting in the energy distribution of the generated decomposition component extending to the adjacent frequency band, thereby reducing the separability of signal adaptive decomposition. On the other hand, Gaussian white noise is a full communication signal in frequency domain, that is, noise has both high frequency components and low frequency components. The low frequency components of Gaussian white noise overlap with the signal in the frequency spectrum, which means that the existing methods based on spectral distribution difference cannot remove the low frequency components of noise from the observed signal. Although some noise suppression algorithms can handle noise low frequency components, these methods use specific signal characteristics in specific scenarios to distinguish signal and noise, so they lack universality.

[0050] In view of the problem of low separability of decomposition components caused by spectral confusion, the present application generalizes the traditional singular spectrum analysis based on rectangular window to generalized singular spectrum analysis and adaptively finds the optimal singular spectrum analysis according to the characteristics of the signal to suppress spectral leakage. As for the problem that the low frequency components of Gaussian white noise overlap with the signal in the frequency spectrum, most of the noise in the signal is independent and identically distributed, and according to the central limit theorem, the spectral energy of a single noise sample is greater than the spectral energy of the mean of multiple noise samples. The present application will select a suitable range of windows and corresponding singular spectrum analysis based on the Euclidean metric space centered on the optimal singular spectrum analysis to utilize the above noise statistical characteristics to suppress the low frequency components of noise, thereby improving the noise suppression effect.

[0051] Referring to Figure 1 The present application provides a method for electrocardio signal noise suppression based on singular spectrum analysis mean strategy, which comprises the following steps:

[0052] S100, obtaining an electrocardio noise signal sequence, and converting the electrocardio noise signal sequence to frequency domain to obtain an electrocardio signal spectrum;

[0053] S200, constructing a frequency response matrix based on the electrocardio noise signal spectrum, establishing a target function oriented to suppress spectral leakage based on the frequency response matrix, solving the target function to obtain a target vector, and determining an optimal window according to the target vector;

[0054] S300, generating a plurality of noise suppression signals based on the electrocardio noise signal sequence and the optimal window, and generating a noise suppression electrocardio signal based on the plurality of noise suppression signals.

[0055] In the embodiments provided by the present application, firstly, the optimal window is searched to suppress the spectral leakage of the noise electrocardio signal sequence, so as to improve the separability of the electrocardio information and the noise; and then, the noise sample is generated based on the optimal window to realize the mean noise suppression. The present application can suppress the low-frequency component of the noise, so as to improve the noise suppression effect.

[0056] In some embodiments, in S200, the frequency response matrix is constructed based on the electrocardio noise signal spectrum, the target function oriented to suppress the spectral leakage is established based on the frequency response matrix, the target vector is obtained by solving the target function, and the optimal window is determined according to the target vector, which includes:

[0057] S210, obtaining the electrocardio noise signal spectrum, and searching for a plurality of maximum points in the electrocardio noise signal spectrum;

[0058] S220, adaptively dividing a plurality of frequency bands according to the plurality of maximum points, determining the frequency response corresponding to each frequency band, and then constructing the frequency response matrix;

[0059] S230, establishing the target function oriented to suppress the spectral leakage based on the frequency response matrix, obtaining the target matrix by solving the target function, and obtaining the estimation matrix by interpolating the column vectors of the target matrix;

[0060] S240, converting the electrocardio noise signal sequence into a trajectory matrix, and obtaining the optimal window according to the estimation vector in the trajectory matrix and the estimation matrix.

[0061] The specific steps are as follows:

[0062] 1) The electrocardio noise signal sequence is transformed into the frequency domain to obtain the electrocardio noise signal spectrum ; ;

[0063] 2) A plurality of maximum points are searched for in the electrocardio noise signal spectrum, and the frequency bands are adaptively divided according to the plurality of maximum points in the spectrum ; In the entire scheme, because the electrocardio signal and the noise two components need to be decomposed by singular spectrum analysis, the ;

[0064] 3) The corresponding frequency response is designed in each frequency band ;

[0065] In some embodiments, the frequency response formula is: ​​

[0066] ;

[0067] wherein, represents a constructor, is an argument of the constructor, when , ; when , ; when , ;

[0068] To prevent spectral leakage, it is necessary to satisfy:

[0069] ;

[0070] wherein, represents the frequency response of the kth maximum point, represents the number of maximum points, and k represents the index of the maximum point, ; , respectively represent the kth frequency point and the k+1th frequency point, and respectively represent the kth transition band interval and the k+1th transition band interval.

[0071] Frequency bands adaptively divided according to extreme points in signal spectrum , is and the transition band interval between frequency bands, is and the transition band interval between frequency bands, and ; () represents a constructor.

[0072] Specifically, let , then when , ; when , ; when , .

[0073] 4) Construct a frequency response matrix , design an optimization problem oriented to suppress spectral leakage, and solve the optimization problem to obtain a target matrix .

[0074] 5) The column vector of the target matrix can be obtained, and the kth frequency point can be estimated according to the following formula: The coefficient of the maximum point ;

[0075] ;

[0076] According to the coefficient , and then an estimation matrix is obtained. ; wherein, , , , , ; The coefficient of the maximum point of the kth frequency point.

[0077] 6) According to the estimation vector , the corresponding optimal window is obtained. .

[0078] In some embodiments, in S230, the target function oriented to suppress spectral leakage is established based on the frequency response matrix, and the target matrix is obtained by solving the target function; and the estimation matrix is obtained by performing interpolation estimation on the column vectors of the target matrix, including:

[0079] The target function is established based on the frequency response matrix:

[0080] ;

[0081] The frequency response matrix is obtained by solving the target function: ; wherein, is the frequency response matrix; is the target matrix; is a multiplier matrix composed of Lagrange multipliers; indicates the index of the maximum point, i indicates the number of rows of elements in the multiplier matrix, j indicates the number of columns of elements in the multiplier matrix, i=0, 1,..., L-1, j=0, 1,..., L-1; is a discrete-time impulse function;

[0082] The singular value decomposition of the frequency response matrix is performed to obtain a left singular matrix, a right singular matrix and a diagonal matrix, and the transposed matrix of the left singular matrix and the right singular matrix is multiplied to obtain the target matrix.

[0083] It should be noted that, in order to solve the phenomenon of spectral leakage of the noise electrocardio signal in the frequency domain, the optimal window design of the generalized singular spectrum analysis oriented to suppress spectral leakage is realized by solving the optimization design problem with orthogonal constraint conditions, and then the separability of the noise high-frequency component and the electrocardio signal component in the noise electrocardio signal is improved. ​

[0084] The specific process is as follows:

[0085] (1) Define the multiplier matrix composed of Lagrange multipliers as , and define as a discrete-time impulse function. In addition, define the column vector of the target matrix as , where i represents the row number of the element in the multiplier matrix, j represents the column number of the element in the multiplier matrix, i=0, 1,..., L-1, j=0, 1,..., L-1, then .

[0086] (2) Convert the above optimization problem into the following unconstrained optimization problem:

[0087] ;

[0088] (3) Convert the unconstrained optimization problem into the objective function:

[0089] ;

[0090] (4) Take the derivative of the objective function with respect to , and obtain:

[0091]

[0092] (5) According to the definition of the derivation rule, let the above formula be 0, and obtain:

[0093]

[0094] And further expand it to matrix form, and obtain:

[0095]

[0096] (6) Singular value decomposition is performed on the frequency response matrix , and after decomposition, the left singular matrix , the right singular matrix , and the diagonal matrix are obtained. , , . At the same time, according to the singular value decomposition, the symmetric matrix is regarded as the result of the interaction between the Hermitian matrix and the diagonal matrix , then the above formula can be expressed as:

[0097] ;

[0098] From the above formula, the target matrix can be derived.​

[0099] In some embodiments, in S240, the converting the electrocardio noise signal sequence into a trajectory matrix, and obtaining the optimal window according to the trajectory matrix and an estimation vector in the estimation matrix, comprises:

[0100] converting the electrocardio noise signal sequence into a first trajectory matrix;

[0101] obtaining a first autocorrelation matrix based on the first trajectory matrix, and performing singular value decomposition on the first autocorrelation matrix to obtain a first singular matrix:

[0102] determining a first singular vector of the first singular matrix, taking a ratio of an L2 norm of the estimation vector and an L2 norm of the first singular vector as a window coefficient, thereby obtaining the optimal window.

[0103] The specific process is as follows:

[0104] converting an electrocardio noise signal sequence with a length of into a first trajectory matrix ; , i.e.

[0105] ;

[0106] thereby obtaining a first autocorrelation matrix , , performing singular value decomposition on the first autocorrelation matrix through to obtain a first singular matrix ; , , , denotes a first singular vector, k=0, 1, …, L-1, and a first diagonal matrix =diag , , wherein diag( ) is a diagonal matrix operator, is the kth first singular value.

[0107] Finally, the window coefficient is estimated through the formula , thereby obtaining the optimal window .

[0108] In some embodiments, in S300, the generating a plurality of noise suppression signals based on the electrocardio noise signal sequence and the optimal window, and generating a noise suppression electrocardio signal based on the plurality of noise suppression signals, comprises:

[0109] S310, based on the Euclidean metric space, generating a plurality of different windows centered on the optimal window:​

[0110] S320, based on the generated window, uses singular spectrum analysis to decompose the ECG noise signal sequence to obtain a preliminary noise suppression signal;

[0111] S330: Set the maximum correlation. Determine whether the correlation between the preliminary noise suppression signal and the noisy ECG signal sequence is less than the maximum correlation. If so, assign the correlation value to the maximum correlation and continue with S320 using the preliminary noise suppression signal as the ECG noise signal sequence. If the correlation is equal to the maximum correlation, the noise suppression signal of this window is obtained.

[0112] S340 performs averaging on the noise-suppressed signals from multiple windows to obtain the final noise-suppressed ECG signal.

[0113] It should be noted that, in response to the limitations of current denoising algorithms, which are constrained by both the spectral leakage of noisy ECG signals and the low-frequency noise components that overlap with the spectrum of information components, this invention focuses on optimal singular spectrum analysis. Based on the Euclidean metric space, it selects an appropriate window range and the corresponding singular spectrum analysis to use a mean strategy to suppress low-frequency noise components, thereby realizing a noisy ECG signal suppression algorithm that covers the entire spectrum.

[0114] The specific process is as follows:

[0115] 1) Definition For a random vector generated by the Matlab function "randn", the following formula is used to optimize the window size. Different windows are generated based on this:

[0116]

[0117] in, For the first One window, , Indicates the number of windows. It is a random vector.

[0118] 2) Based on the generated window Singular spectrum analysis was used to analyze ECG noise signal sequences. The initial noise suppression signal is obtained by decomposition.

[0119] 3) Initialize maximum correlation Calculate the initial noise suppression signal With noisy ECG signal sequence correlation .if Then The value assigned To initially suppress noise signals As the input signal, go to step 2) to obtain the next stage of noise suppression signal, and compare again; if , obtain the noise suppression signal based on the generated window . .

[0120] 4) For different windows , obtain the noise suppression electrocardio signal under different windows by steps 2) and 3) respectively.

[0121] 5) Perform mean value processing on the noise suppression electrocardio signals based on different windows to obtain the final noise suppression electrocardio signal , that is: .

[0122] In some embodiments, in S320, the generated window is used to decompose the electrocardio noise signal sequence by singular spectrum analysis to obtain a preliminary noise suppression signal, including:

[0123] S321, the electrocardio noise signal sequence is segmented into a plurality of segmented signals, and each segmented signal is respectively multiplied with each window to obtain a plurality of window vectors;

[0124] S322, a second trajectory matrix is generated based on the plurality of window vectors, the second trajectory matrix is multiplied with its Hermitian matrix to obtain a second autocorrelation matrix, the second autocorrelation matrix is singular value decomposed to obtain a second diagonal matrix and a second singular matrix;

[0125] S323, a noise suppression matrix is generated based on the second trajectory matrix, the second diagonal matrix and the second singular matrix, and the noise suppression matrix is weighted diagonally averaged to obtain the preliminary noise suppression signal of the window.

[0126] The specific steps are as follows:

[0127] (1) The electrocardio noise signal sequence with a length of is segmented into segmented signals with a length of , and each segmented signal is multiplied with a window to obtain a window vector , , , . Since the electrocardio signal and the noise are decomposed by singular spectrum analysis, the window vector .

[0128] Specifically, The number of windows is represented. Based on each window, the electrocardio signal and the noise are decomposed by performing singular spectrum analysis on the noise electrocardio signal. Therefore, it is known that there are one version of noise suppression signal in the present application, generally the value range of L is between 15 and 30. L represents the embedding dimension of singular spectrum analysis, and L also determines the number of components obtained by decomposition in the process of singular spectrum analysis. Since the present application needs to decompose the electrocardio signal and the noise into two components by singular spectrum analysis, .

[0129] (2) The window vector forms a second trajectory matrix , , then:

[0130] ;

[0131] (3) The second trajectory matrix is multiplied by its autocorrelation matrix to obtain the second autocorrelation matrix , and singular value decomposition is performed on the second autocorrelation matrix to obtain the second singular matrix , , , , the second singular vector is represented by , the second diagonal matrix =diag( ), , wherein diag( ) is a diagonal matrix operator, is the first second singular value.

[0132] (4) The noise suppression matrix is generated based on the second trajectory matrix, the second diagonal matrix and the second singular matrix;

[0133] The element in the noise suppression matrix is defined as , then:

[0134] ;

[0135] The weighted diagonal average is performed on the noise suppression matrix to obtain the preliminary noise suppression signal of the window , , that is:

[0136] ;

[0137] wherein, denotes an element of the noise suppression matrix in the jth row and the ith column of the noise suppression matrix denotes an element of the noise suppression matrix in the jth row and the ith column of the noise suppression matrix denotes an element of the noise suppression matrix in the jth row and the ith column of the noise suppression matrix denotes an element of the noise suppression matrix

[0138] Compared with the related art, the present application has the following advantages:

[0139] In the embodiments provided by the present application, the optimal window is adaptively searched based on the spectrum of the electrocardiosignal to suppress the spectrum leakage phenomenon of the electrocardiosignal, thereby improving the separability of the high-frequency components of the noise and the information components in the noisy electrocardiosignal and avoiding the information loss. In addition, the optimal window is taken as the center, the windows with appropriate ranges are selected based on the Euclidean metric space to generate different noise samples, thereby obtaining different versions of the noise suppression signals to implement the mean strategy to suppress the low-frequency components of the noise, and the information component structure will not be severely damaged in the reordering.

[0140] Corresponding to the method of the present application, with reference to Figure 1 , the embodiments of the present application provide an electrocardiosignal noise suppression system based on the mean strategy of singular spectrum analysis, comprising: Figure 2 at least one processor;

[0141] at least one memory for storing at least one program;

[0142] when the at least one program is executed by the at least one processor, the at least one processor implements the above-mentioned method.

[0143] It can be seen that the contents in the above-mentioned method embodiments are all applicable to the present system embodiments, the functions specifically implemented by the present system embodiments are the same as those of the above-mentioned method embodiments, and the beneficial effects achieved by the present system embodiments are also the same as those achieved by the above-mentioned method embodiments.

[0144] In addition, the embodiments of the present application also disclose a computer program product or a computer program, which is stored in a computer readable storage medium. The processor of the computer device can read the computer program from the computer readable storage medium, and the processor executes the computer program to make the computer device execute the above-mentioned method. Similarly, the contents in the above-mentioned method embodiments are all applicable to the present storage medium embodiments, the functions specifically implemented by the present storage medium embodiments are the same as those of the above-mentioned method embodiments, and the beneficial effects achieved by the present storage medium embodiments are also the same as those achieved by the above-mentioned method embodiments.

[0145] In addition, the embodiments of the present application also disclose a computer program product or a computer program, which is stored in a computer readable storage medium. The processor of the computer device can read the computer program from the computer readable storage medium, and the processor executes the computer program to make the computer device execute the above-mentioned method. Similarly, the contents in the above-mentioned method embodiments are all applicable to the present storage medium embodiments, the functions specifically implemented by the present storage medium embodiments are the same as those of the above-mentioned method embodiments, and the beneficial effects achieved by the present storage medium embodiments are also the same as those achieved by the above-mentioned method embodiments.

[0146] As will be appreciated by one of ordinary skill in the art, all or a portion of the methods disclosed herein can be embodied in software, firmware, hardware, or any suitable combination thereof. Any of the physical components or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or as hardware, or as an integrated circuit, such as an application- specific integrated circuit. Such software can be distributed on computer readable media, which can comprise computer storage media (or non-transitory media) and communication media (or transitory media). As is well known to those of ordinary skill in the art, the term computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by a computer. Further, as is well known to those of ordinary skill in the art, communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal, such as a carrier wave or other transport mechanism, and includes any information delivery media.

[0147] The above description is that of the preferred embodiments of the present disclosure. However, the present disclosure is not limited to the embodiments described above. It will be appreciated by those skilled in the art that any equivalent or modification of the present disclosure is possible without departing from the spirit of the present disclosure, and such equivalents or modifications are included in the scope of the present disclosure as defined by the appended claims.

Claims

1. A method for suppressing ECG signal noise based on a singular spectrum analysis mean strategy, characterized in that, The method includes the following steps: S100: Obtain the ECG noise signal sequence and convert the ECG noise signal sequence to the frequency domain to obtain the ECG signal spectrum; S200: Construct a frequency response matrix based on the spectrum of ECG noise signal, establish an objective function oriented towards suppressing spectral leakage based on the frequency response matrix, solve the objective function to obtain the objective vector, and determine the optimal window based on the objective vector; S300, generates multiple noise suppression signals based on the ECG noise signal sequence and the optimal window, and generates a noise-suppressed ECG signal based on the multiple noise suppression signals; S200 specifically includes: S210, acquire the spectrum of the ECG noise signal, and find multiple maxima points in the ECG noise signal spectrum; S220, multiple frequency bands are obtained by adaptively dividing the multiple maximum points, the frequency response corresponding to each frequency band is determined, and then a frequency response matrix is ​​constructed; S230: Based on the frequency response matrix, establish an objective function aimed at suppressing spectral leakage; solve the objective function to obtain the objective matrix; and perform interpolation estimation on the column vectors of the objective matrix to obtain the estimation matrix. S240 converts the ECG noise signal sequence into a trajectory matrix, and obtains the optimal window based on the trajectory matrix and the estimation vector in the estimation matrix; S230 specifically includes: Establish the objective function based on the frequency response matrix: ； Solving the objective function yields the frequency response matrix: ； in, H(z) is the frequency response matrix; H(z) is the target matrix; Let be the multiplier matrix composed of Lagrange multipliers; l represents the index of the maximum point, i represents the row number of the elements in the multiplier matrix, and j represents the column number of the elements in the multiplier matrix, i=0,1, L-1, j=0,1, L-1; It is a discrete-time impulse function; Perform singular value decomposition on the frequency response matrix to obtain the left singular matrix, the right singular matrix, and the diagonal matrix. Multiply the transposes of the left and right singular matrices to obtain the target matrix.

2. The method according to claim 1, characterized in that, The frequency response formula is: ； in, This represents the constructor function, where t is the argument of the constructor function. hour, When t > 1, When 0 ≤ t ≤ 1, ; satisfy: ； in, Let L represent the frequency response at the k-th maximum point, L represent the number of maximum points, and k represent the index of the maximum point, where k = 0, ..., L-1. , These represent the k-th frequency point and the (k+1)-th frequency point, respectively. and These represent the k-th transition zone interval and the (k+1)-th transition zone interval, respectively.

3. The method according to claim 1, characterized in that, In S240, the step of converting the ECG noise signal sequence into a trajectory matrix and obtaining the optimal window based on the trajectory matrix and the estimation vector in the estimation matrix includes: Transform the ECG noise signal sequence into the first trajectory matrix; The first autocorrelation matrix is ​​obtained based on the first trajectory matrix, and the first singular matrix is ​​obtained by performing singular value decomposition on the first autocorrelation matrix: The first singular vector of the first singular matrix is ​​determined, and the ratio of the L2 norm of the estimated vector to the L2 norm of the first singular vector is used as the window coefficient of the optimal window, thus obtaining the optimal window.

4. The method according to claim 1, characterized in that, In S300, the generation of multiple noise suppression signals based on the ECG noise signal sequence and an optimal window, and the generation of a noise-suppressed ECG signal based on the multiple noise suppression signals, includes: S310 generates multiple different windows based on the Euclidean metric space, centered on the optimal window: S320, based on the generated window, uses singular spectrum analysis to decompose the ECG noise signal sequence to obtain a preliminary noise suppression signal; S330: Set the maximum correlation. Determine whether the correlation between the preliminary noise suppression signal and the noisy ECG signal sequence is less than the maximum correlation. If so, assign the correlation value to the maximum correlation and continue with S320 using the preliminary noise suppression signal as the ECG noise signal sequence. If the correlation is equal to the maximum correlation, the noise suppression signal of this window is obtained. S340 performs averaging on the noise-suppressed signals from multiple windows to obtain the final noise-suppressed ECG signal.

5. The method according to claim 4, characterized in that, In S320, the step of using singular spectrum analysis to decompose the ECG noise signal sequence based on the generated window to obtain a preliminary noise suppression signal includes: S321, the ECG noise signal sequence is divided into multiple segment signals, and each segment signal is multiplied by each window to obtain multiple window vectors; S322, generate a second trajectory matrix based on multiple window vectors, multiply the second trajectory matrix by its chief matrix to obtain a second autocorrelation matrix, and perform singular value decomposition on the second autocorrelation matrix to obtain a second diagonal matrix and a second singular matrix; S323, a noise suppression matrix is ​​generated based on the second trajectory matrix, the second diagonal matrix, and the second singular matrix. The noise suppression matrix is ​​then weighted and averaged diagonally to obtain the initial noise suppression signal for the window.

6. A ECG signal noise suppression system based on a singular spectrum analysis mean strategy, characterized in that, The system includes: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor performs the method as described in any one of claims 1 to 5.

Images