A method and system for ECG signal denoising based on APSO-VMD algorithm

Abstract

This invention discloses an ECG signal denoising method based on the Adaptive Particle Swarm Optimization-Variational Mode Decomposition (APSO-VMD) algorithm. The implementation steps of the APSO-VMD-based ECG signal denoising method include: acquiring the original ECG signal; inputting the ECG signal into the APSO-VMD algorithm to simulate and calculate the combination of (K, α); continuously updating the individual and global extrema by calculating the fitness function to find the optimal solution; and using the found optimal (K, α) into the VMD model to achieve ECG signal denoising. Compared with traditional EMD and wavelet thresholding denoising algorithms, the APSO-VMD algorithm achieves the highest signal-to-noise ratio (SNR) and autocorrelation coefficient (AC), as well as the smallest mean square error (MSE), thus ensuring effective denoising and accurate reconstruction of the ECG signal. Research has confirmed that the ECG signal denoising method based on the APSO-VMD algorithm of this invention has superior denoising performance, and also has advantages such as high signal reconstruction accuracy, simple calculation, and ease of implementation.

Description

Technical Field

[0001] This invention relates to human electrocardiogram (ECG) detection and analysis technology. Specifically, it designs an ECG signal denoising method and system based on the APSO-VMD algorithm to achieve effective denoising and accurate waveform reconstruction of ECG signals. Background Technology

[0002] The heart is composed of countless cardiomyocytes, and its electrical activity is essential for its normal pumping function. An electrocardiogram (ECG) is a comprehensive reflection of the electrophysiological activity of this group of cardiomyocytes. During depolarization and repolarization, cardiomyocytes generate potential changes that are transmitted to the body surface via tissues. The potential difference captured at different locations on the body surface by surface electrodes connected to the limbs or chest is the ECG signal. ECG signals can reflect the physiological condition of various parts of the heart to a certain extent, containing rich diagnostic information, and have been widely used in the clinical diagnosis of arrhythmias, myocardial infarction, various cardiovascular diseases, and autonomic nervous system disorders. The ECG signal is a nonlinear and non-stationary human bioelectrical signal. Under normal circumstances, the amplitude of the ECG signal ranges from 10μV to 4mV, exhibiting weakness, low frequency, and instability. It is very susceptible to various interferences or additional activity from other parts of the heart, which can cause severe distortion of the ECG signal during detection, greatly reducing the accuracy of diagnosing heart diseases. Therefore, noise reduction is the primary issue in ECG signal detection.

[0003] Under normal circumstances, the main frequency range of ECG signals is 0.05Hz to 100Hz. ECG noise mainly includes power frequency interference, electromyography (EMG) interference, and baseline drift. Power frequency interference can generally be well eliminated by a 50 / 60Hz notch filter. However, many scholars hold different opinions and have proposed many methods for eliminating EMG interference and baseline drift, including digital filter methods, wavelet methods, and empirical mode decomposition methods.

[0004] 1. Digital Filter Method

[0005] Digital filters offer advantages such as ease of implementation, high stability, and flexible design, avoiding problems like voltage drift, temperature drift, and noise that analog filters cannot overcome. Therefore, with the development of digital technology, the use of digital technology to implement filter functions has received increasing attention and widespread application. However, traditional digital filtering methods, when processing ECG signals in the frequency domain, may result in the filtering out of many useful signals due to the aliasing of ECG signals with noise signals.

[0006] 2. Wavelet method

[0007] Wavelet Transform (WT) is a time-frequency localization analysis method with multi-resolution characteristics. It overcomes some shortcomings of FFT and its improved algorithms, and is particularly suitable for analyzing abrupt, non-stationary signals. However, its limitations hinder its application: 1) The frequency domain resolution in WT is coarse, and severe frequency aliasing may exist between frequency bands, far from achieving the level of FFT and its improved algorithms. 2) Wavelet functions of different scales interfere with each other in the frequency domain and are easily affected by noise, failing to effectively separate harmonics and interharmonics with similar frequencies. 3) WT's characteristics at abrupt change points allow for qualitative analysis of signals, but directly detecting signal amplitude using WT presents certain difficulties. 4) WT suffers from varying degrees of computational complexity, is not suitable for real-time computation, and is difficult to implement in embedded systems. Furthermore, the wavelet method is highly dependent on the choice of threshold; setting the threshold too low may introduce artificial noise, while setting it too high will damage the ECG signal, and the choice of wavelet basis also affects the denoising effect.

[0008] 3. Empirical Mode Decomposition Method

[0009] Empirical Mode Decomposition (EMD) decomposes signals based on the inherent time-scale characteristics of the data, without requiring any pre-defined basis functions. This is fundamentally different from Fourier and wavelet decomposition methods, which rely on a priori harmonic and wavelet basis functions. Due to this characteristic, EMD can theoretically be applied to the decomposition of any type of signal, thus offering significant advantages in processing non-stationary and nonlinear data. It is suitable for analyzing nonlinear and non-stationary signal sequences and boasts a high signal-to-noise ratio. The key to this method is empirical mode decomposition, which decomposes complex signals into a finite number of intrinsic mode functions (IMFs). Each IMF component contains local characteristic signals of the original signal at different time scales. Therefore, EMD can adaptively decompose noisy ECG signals into a series of Intrinsic Mode Function (IMF) components, discard the IMF components that are judged as noise, and reconstruct the remaining IMF components to obtain the denoised ECG signal. However, there is usually mode aliasing between IMF components, which affects the denoising effect. Summary of the Invention

[0010] The technical problem this invention aims to solve is to provide an ECG signal denoising method and system based on the APSO-VMD algorithm, addressing the aforementioned problems in existing technologies. This invention proposes that the VMD algorithm can effectively alleviate the IMF mode aliasing problem in the EMD algorithm. This invention cleverly utilizes the Adaptive Particle Swarm Optimization (APSO) algorithm to optimize the selection of the mode decomposition number K and penalty factor α in the VMD algorithm, overcoming the shortcomings of traditional VMD algorithms, such as slow speed and poor generalization due to manual selection of (K, α) parameters. The proposed APSO-VMD denoising algorithm achieves the lowest MSE, highest SNR, and highest AC, indicating that the APSO-VMD denoising algorithm has strong denoising capabilities; it also preserves the original waveform characteristics of the ECG signal relatively completely, thus ensuring effective denoising and accurate reconstruction of the ECG signal.

[0011] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0012] A method for denoising ECG signals based on the APSO-VMD algorithm, characterized by the following implementation steps:

[0013] 1) Acquisition of the raw ECG signal;

[0014] 2) Input the ECG signal into the APSO-VMD algorithm to simulate and calculate the combination of (K, α);

[0015] 3) By continuously updating the individual and global extrema through the fitness function, the optimal solution is eventually found;

[0016] 4) Use the found optimal (K,α) into the VMD model to achieve ECG signal denoising.

[0017] Optionally, the detailed steps of step 1) include:

[0018] Under normal circumstances, the main frequency range of ECG signals is 0.05Hz to 100Hz. ECG noise mainly includes power frequency interference, electromyography interference, and baseline drift.

[0019] This invention uses record 103 from the MIT Arrhythmia Database as a pure ECG signal and adds 20dB Gaussian white noise and 0.3Hz baseline drift to simulate ECG signals under real-world conditions.

[0020] Optionally, the detailed steps of step 2) include:

[0021] 2.1) Variational Mode Decomposition

[0022] Variational mode decomposition (VMD) is an adaptive, fully non-recursive method for mode variation and signal processing. VMD can decompose a real signal f(t) into K discrete modes u. k (k=1,2,3,…,K), unlike the EMD algorithm, u k The bandwidth in the frequency domain possesses specific sparsity properties, making it an amplitude-frequency modulation (AM-FM) signal. This effectively suppresses mode aliasing and endpoint effects in EMD and has a more robust mathematical theoretical foundation. Each u in VMD... k Tightly around the corresponding center frequency ω k Furthermore, its bandwidth is obtained through Gaussian smoothing demodulation. The constrained variational problem in the VMD process is shown in the following equation:

[0023]

[0024]

[0025] In the above formula, {u k}={u1,u2,u3,…,u k} and {ω k}={ω1,ω2,ω3,…,ω k} represents the set of K decomposed modes and their corresponding center frequencies.

[0026] The constrained variational problem above is solved by introducing a quadratic penalty parameter α and a Lagrange multiplier operator λ(t). The augmented Lagrange function is shown in the following equation:

[0027]

[0028] The saddle point of the above equation is obtained using the alternating direction multiplier method, and u is iteratively updated in the frequency domain. k ω k and λ.

[0029] The specific steps of VMD in decomposing a signal into K modal components are as follows:

[0030] (1) Initialization λ 1 And n is 0;

[0031] (2) Modal function update. This can be achieved by adding a Wiener filter in the frequency domain:

[0032]

[0033] (3) Center frequency update.

[0034]

[0035] In the above formula, This represents the center frequency of the corresponding IMF power spectrum.

[0036] (4) Lagrange multiplier update. Update using the following formula.

[0037]

[0038] In the above formula, τ is the noise tolerance parameter. When the signal contains strong noise, τ can be set to 0 in order to obtain a good noise reduction effect.

[0039] (5) Repeat steps 2, 3 and 4 until the iteration termination condition is met.

[0040]

[0041] In the above formula, ε represents the discrimination precision, ε>0

[0042] (6) Output the results and obtain K modal components.

[0043] The number of decomposition layers K and the penalty factor α have the greatest impact on the VMD decomposition results, directly affecting the denoising effect. Typically, the values ​​of (K, α) need to be manually selected before performing variational mode decomposition, but manual selection cannot achieve the best results. Therefore, finding a simple and direct way to select the parameters (K, α) is a problem that urgently needs to be solved.

[0044] 2.2) Constructing a variational modal model

[0045] First, the particle swarm in the APSO algorithm is initialized, the particle swarm size is determined, and the particle swarm velocity matrix V and position matrix P are randomly generated. A mapping is established between the particle swarm position dimension and the variational mode decomposition level K and penalty factor α. The mapping relationship is that the values ​​of the decomposition level K and penalty factor α correspond one-to-one with the corresponding particle swarm velocity matrix V and position matrix P.

[0046] Optionally, the detailed steps of step 3) include:

[0047] The signal dispersion entropy value of variational mode decomposition is used as the fitness function of APSO. The autocorrelation coefficient of the threshold method is selected as the useful component of IMF. Finally, the useful IMF components are accumulated, and the dispersion entropy of each combination is calculated as the fitness value. The fitness function is a measure of the quality of a particle's spatial position. The smaller the fitness function, the better the particle's position. Its definition is shown in the following formula:

[0048]

[0049] In the above formula, m is the signal dimension, c is the number of categories; p(i) represents the sum of all values ​​c. mRelative frequency, x i This refers to the dispersion pattern corresponding to the embedded signal.

[0050] 3.2) Iterative update to find the optimal solution

[0051] The individual extreme value Pbest and the global extreme value Gbest of the particle swarm are continuously updated through a fitness function, and each particle continuously updates its own velocity as shown in the following equation:

[0052]

[0053]

[0054] In the above formula, t represents the current iteration number; w represents the inertia factor; c1 and c2 represent the learning factors; and r1 and r2 represent random numbers in the interval [0,1].

[0055] To achieve nonlinear over-optimization and accelerate convergence, this paper adopts an adaptive inertia factor adjustment method, as shown in the following formula:

[0056]

[0057] In the above formula, ζ represents the current fitness value of each particle; min ζ represents the minimum fitness value of all current particles; avg This represents the average fitness value of all particles. The adaptive adjustment method incorporates the current fitness value of each particle as a variable into the adjustment strategy. By continuously and dynamically adjusting the inertia factor w, the global and local optimization performance can be adjusted, thereby improving the convergence speed and facilitating the rapid acquisition of the global optimum.

[0058] Optionally, the detailed steps in step 4) are as follows:

[0059] When the fitness function F is minimized, the iteration stops. The global extremum Gbest of the particle swarm is the optimal solution of the optimized variational mode. The optimal decomposition level K and penalty factor α found through iteration are substituted into the VMD model to achieve effective denoising and accurate reconstruction of the ECG signal.

[0060] Furthermore, this invention also provides an ECG signal denoising system based on the APSO-VMD algorithm, comprising:

[0061] Signal generation unit, used to generate ECG signals;

[0062] The signal processing unit is used to input the acquired ECG signal into the APSO-VMD algorithm to find the optimal decomposition level K and penalty factor α;

[0063] The VMD program unit uses the found optimal decomposition level K and penalty factor α to perform signal reconstruction in the VMD algorithm.

[0064] The GUI display unit is used to display the waveform of the signal reconstruction performed by the VMD algorithm.

[0065] Furthermore, the present invention also provides an ECG signal denoising system based on the APSO-VMD algorithm, comprising a PC device, characterized in that the PC device is programmed or configured to perform the steps of the ECG signal denoising method based on the APSO-VMD algorithm as described in any one of claims 1 to 5.

[0066] Furthermore, the present invention also provides an ECG signal denoising system based on the APSO-VMD algorithm, comprising a PC device, characterized in that the PC device has an embedded program stored in its memory that is programmed or configured to execute the ECG signal denoising system based on the APSO-VMD algorithm as described in any one of claims 1 to 5.

[0067] Furthermore, the present invention also provides a digital signal readable storage medium, characterized in that the digital signal readable storage medium stores an embedded program programmed or configured to execute the ECG signal denoising method based on the APSO-VMD algorithm as described in any one of claims 1 to 5.

[0068] Furthermore, the present invention also provides an ECG signal denoising method and system based on the APSO-VMD algorithm, including a power supply module, an analog-to-digital converter, and a PC device. The output terminal of the power supply module is connected to the analog-to-digital converter, the analog-to-digital converter is connected to the PC device, and the PC device is programmed or configured to execute the steps of the ECG signal denoising method based on the APSO-VMD algorithm.

[0069] Compared with existing technologies, this invention has the following advantages: In acquiring the filtered digital signal of the measured voltage, this invention proposes the VMD algorithm to effectively alleviate the IMF mode aliasing problem in the EMD algorithm. It utilizes the fast convergence speed and global optimization characteristics of the APSO algorithm to optimize the selection of the mode decomposition number K and penalty factor α in the VMD algorithm, overcoming the shortcomings of traditional VMD algorithms, such as slow speed due to manual selection of (K, α) parameters and poor generalization. The proposed APSO-VMD denoising algorithm achieves the lowest MSE, the highest SNR, and AC, indicating that the APSO-VMD denoising algorithm has strong denoising capabilities; it also preserves the original waveform characteristics of the ECG signal relatively completely. This invention can achieve effective denoising and accurate reconstruction of ECG signals and has the advantages of simple computation and easy embedded implementation. Attached Figure Description

[0070] Figure 1 This is a schematic diagram of the basic process of the method in an embodiment of the present invention.

[0071] Figure 2 This embodiment of the invention extracts useful modal component maps based on the APSO-VMD algorithm.

[0072] Figure 3 This is a diagram showing the ECG signal denoising result based on the APSO-VMD algorithm in an embodiment of the present invention.

[0073] Figure 4 This is a schematic diagram of the ECG signal denoising method and system structure based on the APSO-VMD algorithm in an embodiment of the present invention. Detailed Implementation

[0074] like Figure 1 The implementation steps of the ECG signal denoising method based on the APSO-VMD algorithm in this embodiment include:

[0075] 1) Acquisition of the raw ECG signal;

[0076] 2) Input the ECG signal into the APSO-VMD algorithm to simulate and calculate the combination of (K, α);

[0077] 3) By continuously updating the individual and global extrema through the fitness function, the optimal solution is eventually found;

[0078] 4) Use the found optimal (K,α) into the VMD model to achieve ECG signal denoising.

[0079] Optionally, the detailed steps of step 1) include:

[0080] Under normal circumstances, the main frequency range of ECG signals is 0.05Hz to 100Hz. ECG noise mainly includes power frequency interference, electromyography interference, and baseline drift.

[0081] This invention uses record 103 from the MIT Arrhythmia Database as a pure ECG signal and adds 20dB Gaussian white noise and 0.3Hz baseline drift to simulate ECG signals under real-world conditions.

[0082] Optionally, the detailed steps of step 2) include:

[0083] 2.1) Variational Mode Decomposition

[0084] Variational mode decomposition (VMD) is an adaptive, fully non-recursive method for mode variation and signal processing. VMD can decompose a real signal f(t) into K discrete modes u. k (k=1,2,3,…,K), unlike the EMD algorithm, u kThe bandwidth in the frequency domain possesses specific sparsity properties, making it an amplitude-frequency modulation (AM-FM) signal. This effectively suppresses mode aliasing and endpoint effects in EMD and has a more robust mathematical theoretical foundation. Each u in VMD... k Tightly around the corresponding center frequency ω k Furthermore, its bandwidth is obtained through Gaussian smoothing demodulation. The constrained variational problem in the VMD process is shown in the following equation:

[0085]

[0086]

[0087] In the above formula, {u k}={u1,u2,u3,…,u k} and {ω k}={ω1,ω2,ω3,…,ω k} represents the set of K decomposed modes and their corresponding center frequencies.

[0088] The constrained variational problem above is solved by introducing a quadratic penalty parameter α and a Lagrange multiplier operator λ(t). The augmented Lagrange function is shown in the following equation:

[0089]

[0090] The saddle point of the above equation is obtained using the alternating direction multiplier method, and u is iteratively updated in the frequency domain. k ω k and λ.

[0091] The specific steps of VMD in decomposing a signal into K modal components are as follows:

[0092] (1) Initialization λ 1 And n is 0;

[0093] (2) Modal function update. This can be achieved by adding a Wiener filter in the frequency domain:

[0094]

[0095] (3) Center frequency update.

[0096]

[0097] In the above formula, This represents the center frequency of the corresponding IMF power spectrum.

[0098] (4) Lagrange multiplier update. Update using the following formula.

[0099]

[0100] In the above formula, τ is the noise tolerance parameter. When the signal contains strong noise, τ can be set to 0 in order to obtain a good noise reduction effect.

[0101] (5) Repeat steps 2, 3 and 4 until the iteration termination condition is met.

[0102]

[0103] In the above formula, ε represents the discrimination precision, ε>0

[0104] (6) Output the results and obtain K modal components.

[0105] The number of decomposition layers K and the penalty factor α have the greatest impact on the VMD decomposition results, directly affecting the denoising effect. Typically, the values ​​of (K, α) need to be manually selected before performing variational mode decomposition, but manual selection cannot achieve the best results. Therefore, finding a simple and direct way to select the parameters (K, α) is a problem that urgently needs to be solved.

[0106] 2.2) Constructing a variational modal model

[0107] First, the particle swarm in the APSO algorithm is initialized, the particle swarm size is determined, and the particle swarm velocity matrix V and position matrix P are randomly generated. A mapping is established between the particle swarm position dimension and the variational mode decomposition level K and penalty factor α. The mapping relationship is that the values ​​of the decomposition level K and penalty factor α correspond one-to-one with the corresponding particle swarm velocity matrix V and position matrix P.

[0108] Optionally, the detailed steps of step 3) include:

[0109] The signal dispersion entropy value of variational mode decomposition is used as the fitness function of APSO. The autocorrelation coefficient of the threshold method is selected as the useful component of IMF. Finally, the useful IMF components are accumulated, and the dispersion entropy of each combination is calculated as the fitness value. The fitness function is a measure of the quality of a particle's spatial position. The smaller the fitness function, the better the particle's position. Its definition is shown in the following formula:

[0110]

[0111] In the above formula, m is the signal dimension, c is the number of categories; p(i) represents the sum of all values ​​c. m Relative frequency, x i This refers to the dispersion pattern corresponding to the embedded signal.

[0112] 3.2) Iterative update to find the optimal solution

[0113] The individual extreme value Pbest and the global extreme value Gbest of the particle swarm are continuously updated through a fitness function, and each particle continuously updates its own velocity as shown in the following equation:

[0114]

[0115]

[0116] In the above formula, t represents the current iteration number; w represents the inertia factor; c1 and c2 represent the learning factors; and r1 and r2 represent random numbers in the interval [0,1].

[0117] To achieve nonlinear over-optimization and accelerate convergence, this paper adopts an adaptive inertia factor adjustment method, as shown in the following formula:

[0118]

[0119] In the above formula, ζ represents the current fitness value of each particle; min ζ represents the minimum fitness value of all current particles; avg This represents the average fitness value of all particles. The adaptive adjustment method incorporates the current fitness value of each particle as a variable into the adjustment strategy. By continuously and dynamically adjusting the inertia factor w, the global and local optimization performance can be adjusted, thereby improving the convergence speed and facilitating the rapid acquisition of the global optimum.

[0120] Optionally, the detailed steps in step 4) are as follows:

[0121] When the fitness function F is minimized, the iteration stops. The global extremum Gbest of the particle swarm is the optimal solution of the optimized variational mode. The optimal decomposition level K and penalty factor α found through iteration are substituted into the VMD model to achieve effective denoising and accurate reconstruction of the ECG signal.

[0122] The following section will further simulate the ECG signal denoising method based on the APSO-VMD algorithm in this embodiment. In the ECG denoising experiment, wavelet method, EMD method and APSO-VMD method were used respectively, and the experimental results are shown in Table 1.

[0123] Table 1. Denoising results of different algorithms on ECG signals

[0124]

[0125] As shown in Table 1, among the three algorithms mentioned above, the signal-to-noise ratio and autocorrelation coefficient of the ECG signal processed by the wavelet method are the smallest, while the mean square error is the largest. The signal obtained by processing the ECG signal using the method proposed in this invention has the highest signal-to-noise ratio and autocorrelation coefficient, and the smallest mean square error.

[0126] Therefore, simulation results show that this method has the characteristics of good denoising effect and good robustness, and can accurately and effectively reconstruct the original ECG signal. Compared with existing detection methods, the method in this paper has the advantages of good denoising effect on ECG signals, simple calculation, and easy embedded implementation.

[0127] Furthermore, this invention also provides an ECG signal denoising system based on the APSO-VMD algorithm, comprising:

[0128] Signal generation unit, used to generate ECG signals;

[0129] The signal processing unit is used to input the acquired ECG signal into the APSO-VMD algorithm to find the optimal decomposition level K and penalty factor α;

[0130] The VMD program unit uses the found optimal decomposition level K and penalty factor α to perform signal reconstruction in the VMD algorithm.

[0131] The GUI display unit is used to display the waveform of the signal reconstruction performed by the VMD algorithm.

[0132] Furthermore, the present invention also provides an ECG signal denoising system based on the APSO-VMD algorithm, comprising a PC device, characterized in that the PC device is programmed or configured to perform the steps of the ECG signal denoising method based on the APSO-VMD algorithm as described in any one of claims 1 to 5.

[0133] Furthermore, the present invention also provides an ECG signal denoising system based on the APSO-VMD algorithm, comprising a PC device, characterized in that the PC device has an embedded program stored in its memory that is programmed or configured to execute the ECG signal denoising system based on the APSO-VMD algorithm as described in any one of claims 1 to 5.

[0134] Furthermore, the present invention also provides a digital signal readable storage medium, characterized in that the digital signal readable storage medium stores an embedded program programmed or configured to execute the ECG signal denoising method based on the APSO-VMD algorithm as described in any one of claims 1 to 5.

[0135] like Figure 2 As shown, the ECG signal input is used to filter useful components in the VMD decomposition after finding the optimal decomposition level K and penalty factor α based on the APSO-VMD algorithm.

[0136] refer to Figure 3 It can be seen that by integrating the useful modal components into a denoised signal and comparing it with a noisy signal without baseline shift, from... Figure 3It is quite obvious that the waveform shape and trend remain unchanged and interference noise is effectively filtered out.

[0137] refer to Figure 4 It can be seen that the power supply module 1 is used to power the analog-to-digital converter 2 and the Adalm1000 A / D acquisition 3. The ECG signal is converted into a digital signal by the analog-to-digital converter 2 and then sent to the PC4 by the Adalm1000 A / D acquisition 3. The PC4 performs ECG signal denoising processing based on the APSO-VMD algorithm and displays the denoised ECG waveform.

[0138] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for denoising ECG signals based on the APSO-VMD algorithm, characterized in that... The implementation steps include: S1: Acquire the raw ECG signal; S2: Input the ECG signal into the APSO-VMD algorithm to simulate and calculate the combination of (K, α); the detailed steps of step S2 include: S2.1: Variational mode decomposition; S2.2: Constructing the variational mode model: Initialize the particle swarm in the adaptive particle swarm (APSO) algorithm, determine the particle swarm size, randomly generate the particle swarm velocity matrix V and position matrix P, and establish the mapping between the particle swarm position dimension and the variational mode decomposition level K and the penalty factor α. The mapping relationship is that the values ​​of the decomposition level K and the penalty factor α correspond one-to-one with the corresponding particle swarm velocity matrix V and position matrix P. S3: By continuously updating the individual and global extrema through the fitness function, the optimal solution is eventually found; the detailed steps of step S3 include: S3.1: Calculate the fitness function: Use the signal dispersion entropy value of the variational mode decomposition as the fitness function of APSO, select the autocorrelation coefficient of the threshold method as the selection of the useful IMF component, and finally accumulate the useful IMF components to calculate the dispersion entropy of each combination as the fitness value. S3.2: Iterative update to find the optimal solution: The individual extreme value Pbest and the global extreme value Gbest of the particle swarm position are continuously updated through the fitness function; an adaptive inertia factor adjustment method is adopted, incorporating the current fitness value of each particle as a variable into the adjustment strategy. By continuously and dynamically adjusting the inertia factor, the global and local optimization performance is adjusted, the convergence speed is improved, and the global optimal solution is obtained quickly. S4: Substitute the found optimal (K,α) into the VMD model to achieve ECG signal denoising.

2. The ECG signal denoising method based on the APSO-VMD algorithm according to claim 1, characterized in that, The detailed steps of step S1 include: Record 103 from the MIT Arrhythmia Database was used as the pure ECG signal, and 20dB Gaussian white noise and 0.3Hz baseline drift were added to simulate the ECG signal under real-world conditions.

3. The ECG signal denoising method based on the APSO-VMD algorithm according to claim 1, characterized in that, The detailed steps in step S4 are as follows: When the fitness function F is minimized, the iteration stops. The global extremum Gbest of the particle swarm is the optimal solution of the optimized variational mode. The optimal decomposition level K and penalty factor α found through iteration are substituted into the VMD model to achieve effective denoising and accurate reconstruction of the ECG signal.

4. An ECG signal denoising system based on the APSO-VMD algorithm, the system being used to perform the steps of the ECG signal denoising method based on the APSO-VMD algorithm as described in any one of claims 1 to 3, specifically including: Signal generation unit, used to generate ECG signals; The signal processing unit is used to input the acquired ECG signal into the APSO-VMD algorithm to find the optimal number of decomposition layers K and the penalty factor α; The VMD program unit uses the found optimal decomposition level K and penalty factor α to perform signal reconstruction in the VMD algorithm. The GUI display unit is used to display the waveform of the signal reconstruction performed by the VMD algorithm.

5. An ECG signal denoising system based on the APSO-VMD algorithm, comprising a PC device, characterized in that, The PC device is programmed or configured to perform the steps of the ECG signal denoising method based on the APSO-VMD algorithm as described in any one of claims 1 to 3.

6. An ECG signal denoising system based on the APSO-VMD algorithm, comprising a PC device, characterized in that, The PC device has an embedded program stored in its memory that is programmed or configured to execute the ECG signal denoising method based on the APSO-VMD algorithm as described in any one of claims 1 to 3.

7. A digital signal readable storage medium, characterized in that, The digital signal readable storage medium stores an embedded program that is programmed or configured to perform the ECG signal denoising method based on the APSO-VMD algorithm as described in any one of claims 1 to 3.

8. An ECG signal denoising system based on the APSO-VMD algorithm, characterized in that, The device includes a power supply module (1), an analog-to-digital converter (2), an Adalm1000 A / D acquisition unit (3), and a PC (4). The output of the power supply module (1) is electrically connected to the analog-to-digital converter (2) and the Adalm1000 A / D acquisition unit (3), respectively. The output of the analog-to-digital converter (2) is connected to the PC (4) through the Adalm1000 A / D acquisition unit (3). The PC (4) is programmed or configured to perform the steps of the ECG signal denoising method based on the APSO-VMD algorithm as described in any one of claims 1 to 3.

Images