A low-noise biopotential signal acquisition system and processing method

Abstract

The present application belongs to the technical field of signal acquisition and processing, and discloses a low-noise bioelectric potential signal acquisition system and processing method. The original bioelectric signal of a human body is acquired through an electrode array, and is processed by a multi-stage adaptive filter amplification circuit and then subjected to analog-digital conversion. The digital signal is subjected to adaptive wavelet transform denoising and empirical mode decomposition to obtain a multi-level intrinsic mode function set, from which an intrinsic mode function of a characteristic frequency band is extracted and a signal is reconstructed. A double denoising mechanism combining adaptive wavelet transform and empirical mode decomposition is introduced. A signal quality real-time evaluation system is established, and system parameters, including a variable gain amplifier gain and a dynamic band-pass filter bandwidth, are dynamically adjusted through closed-loop feedback. The adaptive optimization of the whole bioelectric potential signal acquisition process is realized, various types of interference are effectively suppressed, signal characteristics are retained, and the signal-to-noise ratio is significantly improved.

Description

Technical Field

[0001] This invention relates to the field of signal acquisition and processing technology, and more specifically, to a low-noise biopotential signal acquisition system and processing method. Background Technology

[0002] Bioelectric potential signals are electrical signals generated by human physiological activities, including electrocardiograms (ECG), electroencephalograms (EEG), and electromyograms (EMG). They are important sources of information for clinical diagnosis, physiological monitoring, and human-computer interaction systems. These signals typically have extremely small amplitudes (microvolts to millivolts), narrow frequency ranges, and are masked by various noises, posing a significant challenge to accurate acquisition and processing.

[0003] Traditional bioelectrical signal acquisition systems mainly consist of electrodes, amplifiers, filters, and data converters. Conventional surface electrodes form an interface with the skin through electrolyte contact. This interface exhibits complex electrochemical characteristics, generating contact impedance and polarization effects, leading to signal attenuation and distortion. Furthermore, the skin-electrode interface impedance changes significantly over time, posing stability issues for long-term monitoring.

[0004] Bioelectrical signals are inherently weak (typically at the microvolt level) and highly susceptible to environmental electromagnetic interference, motion artifacts, and physiological disturbances, leading to unstable acquisition quality. In clinical practice, significant individual patient differences exist, and traditional fixed-parameter systems cannot dynamically adapt to changes in skin impedance and physiological state, often resulting in a significant decline in signal quality during long-term monitoring. In environments with dense medical equipment, power frequency interference and electromagnetic coupling between devices severely affect signal purity. Crosstalk between channels is prominent in multi-channel acquisition, limiting spatial resolution. Existing algorithms perform poorly when processing non-stationary and nonlinear bioelectrical signals, especially in motion or pathological states where signal characteristics are complex and variable; traditional filtering methods often lose key physiological features while removing noise. Furthermore, the lack of real-time quality assessment and feedback mechanisms prevents the system from making timely adjustments to sudden interferences, potentially missing important physiological information in critical clinical scenarios such as emergency care and intensive care, posing risks to medical decisions.

[0005] In view of this, the present invention proposes a low-noise biopotential signal acquisition system and processing method to solve the above problems. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a low-noise biopotential signal acquisition and processing method, comprising:

[0007] Step 1: Collect bioelectric potential signals from the human body through an electrode array to obtain the raw bioelectric signals;

[0008] Step 2: Input the original bioelectric signal into a multi-stage adaptive filtering and amplification circuit to obtain an amplified and filtered analog signal;

[0009] Step 3: Perform analog-to-digital conversion on the amplified and filtered analog signal to obtain a digital bioelectric signal;

[0010] Step 4: Perform adaptive wavelet transform denoising on the digitized bioelectric signal to obtain a primary denoised signal;

[0011] Step 5: Perform empirical mode decomposition on the primary noise-reduced signal to obtain a multi-level intrinsic mode function set;

[0012] Step 6: Extract the intrinsic mode functions of the characteristic frequency bands from the multi-level intrinsic mode function set to obtain the characteristic intrinsic mode function set;

[0013] Step 7: Reconstruct the signal based on the set of characteristic intrinsic mode functions to obtain a pure bioelectric signal;

[0014] Step 8: Perform real-time quality assessment on the pure bioelectric signal to obtain signal quality index; dynamically adjust system parameters based on the signal quality index to optimize the bioelectric signal acquisition process and obtain a high signal-to-noise ratio bioelectric potential signal output.

[0015] Furthermore, the process of acquiring bioelectric potential signals from the human body via an electrode array to obtain raw bioelectric signals includes:

[0016] Nanomaterial modification treatment is applied to the electrode surface to obtain a low-impedance bioelectrode, and the low-impedance bioelectrode is arranged to form an adaptive structure electrode array to obtain a multi-channel acquisition network.

[0017] Automatic measurement of electrode-skin interface impedance is performed on the multi-channel acquisition network to obtain interface contact quality data. Channel selection is then optimized based on the interface contact quality data to obtain an optimized acquisition channel group.

[0018] The optimized acquisition channel group is used to collect changes in bioelectric potential on the human body surface in real time to obtain multi-channel raw signals. Inter-channel interference suppression is performed on the multi-channel raw signals to obtain preliminary isolated signals.

[0019] The initial isolated signal is subjected to differential amplification preprocessing to obtain a differential enhanced signal, and the differential enhanced signal is subjected to preliminary baseline drift correction to obtain a stable baseline signal;

[0020] The stable baseline signal is transmitted to the subsequent processing circuit through a shielded transmission line to obtain the original bioelectric signal.

[0021] Further, the step of inputting the original bioelectric signal into a multi-stage adaptive filtering and amplification circuit to obtain an amplified and filtered analog signal includes:

[0022] The original bioelectric signal is input into the first-stage low-pass filter to obtain a first-stage filtered signal with high-frequency interference removed. The first-stage low-pass filter adopts a second-order Butterworth structure and the cutoff frequency is adjustable.

[0023] The first-stage filtered signal is input into the second-stage variable gain amplifier to obtain a dynamically range-optimized amplified signal. The gain parameter G of the second-stage variable gain amplifier is dynamically adjusted through real-time feedback control.

[0024] The amplified signal with optimized dynamic range is input into the third-stage dynamic bandpass filter to obtain the second-stage filtered signal with optimized frequency band. The center frequency fc and bandwidth parameter B of the third-stage dynamic bandpass filter are adaptively adjusted according to the signal characteristics.

[0025] The frequency band optimized secondary filter signal is subjected to power frequency noise adaptive notch filtering to obtain a filter signal that suppresses power frequency interference. The power frequency noise adaptive notch filtering uses phase-locked loop tracking technology to compensate for power frequency drift.

[0026] The filtered signal used to suppress power frequency interference is output buffered and impedance matched to obtain an amplified and filtered analog signal.

[0027] Further, the step of performing analog-to-digital conversion on the amplified and filtered analog signal to obtain a digital bioelectric signal includes:

[0028] The amplified and filtered analog signal is conditioned to obtain a standardized signal that meets the conversion range, and the standardized signal is sampled and held to obtain a discrete-time sampling point sequence.

[0029] The discrete-time sampling point sequence is quantized and encoded using a high-precision ADC to obtain an initial digital code stream. The sampling rate of the high-precision ADC is dynamically adjusted according to the signal spectrum characteristics, and the resolution is not less than 16 bits.

[0030] The initial digital bitstream is quantized and noise shaped to obtain an optimized digital signal, and the optimized digital signal is then subjected to digital filtering and anti-aliasing processing to obtain an anti-aliasing digital signal.

[0031] The anti-aliasing digital signal is time-calibrated to obtain a synchronous digital signal, and the synchronous digital signal is converted into a data format to obtain standard format digital data.

[0032] The standard format digital data is controlled through a buffer management mechanism to obtain digital bioelectric signals.

[0033] Further, the adaptive wavelet transform denoising process performed on the digitized bioelectric signal to obtain the primary denoised signal includes:

[0034] The digitized bioelectric signal is subjected to signal characteristic analysis to obtain a signal spectrum feature map, and the optimal wavelet basis function is selected based on the signal spectrum feature map to obtain the suitable wavelet basis;

[0035] The adapted wavelet basis is used to perform multi-scale wavelet decomposition on the digitized bioelectric signal to obtain a multi-layer wavelet coefficient set, and threshold analysis is performed on the multi-layer wavelet coefficient set to obtain the noise threshold distribution.

[0036] An adaptive threshold function is designed based on the noise threshold distribution to obtain an optimized threshold strategy. The optimized threshold strategy is then applied to the wavelet coefficients to perform soft thresholding to obtain denoised wavelet coefficients.

[0037] The signal singularity detection algorithm is applied to the denoised wavelet coefficients to obtain the singularity protection map, and the coefficients are corrected according to the singularity protection map to obtain the wavelet coefficient set that preserves the features.

[0038] The inverse wavelet transform is used to reconstruct the primary denoised signal using the wavelet coefficient set that preserves the characteristics.

[0039] Furthermore, the empirical mode decomposition of the primary denoised signal to obtain a multi-level intrinsic mode function set includes:

[0040] Extreme point detection is performed on the primary noise reduction signal to obtain a signal extreme value distribution map, and upper and lower envelopes are constructed based on the signal extreme value distribution map to obtain envelope pairs;

[0041] The mean envelope is calculated based on the envelope pair to obtain the local mean function, and the primary noise-reduced signal is subtracted from the local mean function to obtain the candidate intrinsic mode function;

[0042] The candidate intrinsic mode functions are subjected to intrinsicity testing to obtain intrinsicity evaluation results. If the intrinsicity evaluation results do not meet the IMF conditions, the envelope calculation process is repeated until the IMF conditions are met to obtain the first layer of intrinsic mode functions.

[0043] The first layer of intrinsic mode functions is subtracted from the primary noise-reduced signal to obtain the first residual signal, and the empirical mode decomposition process is repeated on the first residual signal to obtain the subsequent intrinsic mode functions.

[0044] By combining all the intrinsic mode functions obtained from the decomposition with the final residual signal, a complete multi-level intrinsic mode function set is obtained.

[0045] Further, the step of extracting eigenmode functions of characteristic frequency bands from the multi-level eigenmode function set to obtain a characteristic eigenmode function set includes:

[0046] Hilbert transform is performed on each IMF in the multi-level intrinsic mode function set to obtain the instantaneous frequency distribution, and the spectral center and bandwidth of each IMF are calculated based on the instantaneous frequency distribution to obtain the IMF spectral characteristic table.

[0047] Based on the IMF spectral characteristic table, the characteristic frequency band of the biopotential signal is analyzed to obtain the target frequency band range, and relevant IMFs are screened according to the target frequency band range to obtain a preliminary IMF set;

[0048] Cross-correlation analysis is performed on the initially selected IMF set to obtain the IMF correlation matrix. Redundant components are then removed based on the IMF correlation matrix to obtain a simplified IMF set.

[0049] The energy contribution of each IMF in the simplified IMF set is calculated to obtain the information content assessment result. The importance is then ranked according to the information content assessment result to obtain the ranked IMF set.

[0050] Adaptive weight allocation is performed based on the sorted IMF set to obtain a weighted IMF combination scheme, and a feature intrinsic mode function set is constructed according to the weighted IMF combination scheme.

[0051] Further, the step of reconstructing the signal based on the set of characteristic intrinsic mode functions to obtain a pure bioelectric signal includes:

[0052] Nonlinear phase correction is performed on the feature intrinsic mode function set to obtain a phase-aligned IMF group, and an optimal reconstruction weight vector is designed for the phase-aligned IMF group to obtain an adaptive fusion strategy.

[0053] The IMF is weighted and superimposed based on the adaptive fusion strategy to obtain a preliminary reconstructed signal, and the preliminary reconstructed signal is enhanced with morphological features to obtain a bioelectric signal with prominent features.

[0054] The bioelectric signals with prominent features are subjected to physiological model constraint processing to obtain signals that conform to physiological characteristics, and the signals that conform to physiological characteristics are subjected to baseline drift final correction to obtain stable zero baseline signals.

[0055] Abnormal waveform detection and repair are performed on the stable zero baseline signal to obtain a corrected signal, and the bandwidth of the corrected signal is optimized and smoothed to obtain the final smooth signal.

[0056] The final smooth signal was then compared with a standard bioelectric template to verify its morphological matching, resulting in a pure bioelectric signal.

[0057] Further, the real-time quality assessment of the purified bioelectric signal to obtain a signal quality index, and the dynamic adjustment of system parameters based on the signal quality index to optimize the bioelectric signal acquisition process and obtain a high signal-to-noise ratio bioelectric potential signal output, includes:

[0058] The signal-to-noise ratio, effective bandwidth, and waveform distortion of the pure bioelectric signal are calculated to obtain a multidimensional quality assessment index. A comprehensive quality function is then constructed based on the multidimensional quality assessment index to obtain the signal quality index.

[0059] Establish the mapping relationship between the signal quality index and the key system parameters to obtain the parameter optimization model, and design a feedback control algorithm based on the parameter optimization model to obtain the system parameter adjustment strategy;

[0060] Based on the system parameter adjustment strategy, the gain parameter G of the second-stage variable gain amplifier is updated in real time to obtain dynamic optimized gain control, and the bandwidth parameter B of the third-stage dynamic bandpass filter is adaptively adjusted according to the signal characteristics to obtain optimized frequency band settings.

[0061] The dynamic optimization gain control and the optimized frequency band setting are fed back to the hardware circuit to obtain a closed-loop self-optimizing structure. Based on the closed-loop self-optimizing structure, the electrode array channel configuration is dynamically adjusted to obtain an optimized acquisition path.

[0062] By combining the synergistic effect of the closed-loop self-optimizing structure and the optimized acquisition path, adaptive optimization of the entire bioelectric potential signal acquisition process is achieved, resulting in a high signal-to-noise ratio bioelectric potential signal output.

[0063] A low-noise biopotential signal acquisition system is used to implement the aforementioned low-noise biopotential signal acquisition and processing method, comprising: an electrode acquisition module, used to acquire biopotential signals from the human body through an electrode array to obtain raw biopotential signals;

[0064] A multi-stage adaptive filtering and amplification module is used to input the original bioelectric signal into a multi-stage adaptive filtering and amplification circuit to obtain an amplified and filtered analog signal.

[0065] An analog-to-digital conversion module is used to perform analog-to-digital conversion on the amplified and filtered analog signal to obtain a digital bioelectric signal.

[0066] The wavelet transform denoising module is used to perform adaptive wavelet transform denoising processing on the digitized bioelectric signal to obtain a primary denoised signal.

[0067] The empirical mode decomposition module is used to perform empirical mode decomposition on the primary noise-reduced signal to obtain a multi-level intrinsic mode function set;

[0068] The feature extraction module is used to extract the intrinsic mode functions of the feature frequency bands from the multi-level intrinsic mode function set to obtain the feature intrinsic mode function set;

[0069] The signal reconstruction module is used to reconstruct the signal based on the set of characteristic intrinsic mode functions to obtain a pure bioelectric signal.

[0070] The quality assessment and parameter optimization module is used to perform real-time quality assessment on the pure bioelectric signal to obtain signal quality indicators; based on the signal quality indicators, the system parameters are dynamically adjusted to optimize the bioelectric signal acquisition process and obtain a high signal-to-noise ratio bioelectric potential signal output.

[0071] The technical effects and advantages of the low-noise biopotential signal acquisition system and processing method of this invention are as follows:

[0072] This invention improves the signal-to-noise ratio and fidelity of signals, enabling the precise capture of weak bioelectrical signals in complex environments. It effectively overcomes the problems of signal instability, feature loss, and poor adaptability in traditional techniques, achieving high-quality processing of non-stationary biological signals. The closed-loop self-optimization mechanism established by the method can adjust system parameters in real time according to signal quality, giving the system excellent environmental adaptability and individual compatibility, providing a stable and reliable signal foundation for different application scenarios. By preserving key physiological characteristics of the signal while effectively suppressing various interferences, the accuracy and reliability of bioelectrical signal analysis are significantly improved. Attached Figure Description

[0073] Figure 1 This is a schematic diagram of a low-noise biopotential signal acquisition and processing method according to the present invention;

[0074] Figure 2 This is a detailed flowchart illustrating the implementation steps of step 1 of the present invention;

[0075] Figure 3 This is a schematic diagram of a low-noise biopotential signal acquisition system according to the present invention. Detailed Implementation

[0076] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0077] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0078] This application provides a low-noise bioelectric potential signal acquisition system and processing method. The execution entities of the system and processing method include, but are not limited to, medical monitoring equipment, bioelectric signal processing platforms, medical data analysis servers, wearable health monitoring devices, etc., which can be considered as general computing nodes in this application. The data processing platform includes, but is not limited to, at least one of a bioelectric signal acquisition system, a medical diagnostic information system, and an electrocardiogram monitoring data management system.

[0079] Please see Figure 1 This invention provides a method for acquiring and processing low-noise biopotential signals, comprising the following steps:

[0080] Step 1: Collect bioelectric potential signals from the human body through an electrode array to obtain the raw bioelectric signals;

[0081] Step 2: Input the original bioelectric signal into a multi-stage adaptive filtering and amplification circuit to obtain an amplified and filtered analog signal;

[0082] Step 3: Perform analog-to-digital conversion on the amplified and filtered analog signal to obtain a digital bioelectric signal;

[0083] Step 4: Perform adaptive wavelet transform denoising on the digitized bioelectric signal to obtain a primary denoised signal;

[0084] Step 5: Perform empirical mode decomposition on the primary noise-reduced signal to obtain a multi-level intrinsic mode function set;

[0085] Step 6: Extract the intrinsic mode functions of the characteristic frequency bands from the multi-level intrinsic mode function set to obtain the characteristic intrinsic mode function set;

[0086] Step 7: Reconstruct the signal based on the set of characteristic intrinsic mode functions to obtain a pure bioelectric signal;

[0087] Step 8: Perform real-time quality assessment on the pure bioelectric signal to obtain signal quality indicators; dynamically adjust system parameters based on the signal quality indicators to optimize the bioelectric signal acquisition process and obtain a high signal-to-noise ratio bioelectric potential signal output.

[0088] This invention reduces the electrode-skin interface impedance and improves signal acquisition quality by modifying the electrode surface with nanomaterials; it effectively suppresses various interferences by employing a multi-stage adaptive filtering and amplification circuit; it achieves efficient noise reduction and feature extraction of the signal by combining adaptive wavelet transform and empirical mode decomposition techniques; and it dynamically optimizes system parameters through real-time quality assessment and closed-loop feedback control, ultimately obtaining a high signal-to-noise ratio biopotential signal output.

[0089] In this embodiment of the invention, the steps of the low-noise biopotential signal acquisition and processing method include:

[0090] Step 1: Collect bioelectric potential signals from the human body through an electrode array to obtain the raw bioelectric signals;

[0091] In this embodiment, the electrode surface is first modified with nanomaterials to obtain low-impedance bioelectrodes. These electrodes are then arranged into an adaptive electrode array to form a multi-channel acquisition network. Automatic impedance measurement is performed at the electrode-skin interface to obtain interface contact quality data, which is used to optimize the selection of acquisition channels. The optimized acquisition channel group is used to acquire real-time changes in biopotential on the human body surface, obtaining multi-channel raw signals. These signals undergo inter-channel interference suppression processing to obtain initially isolated signals. Differential amplification preprocessing is performed on the initially isolated signals to enhance the useful signals, followed by preliminary baseline drift correction to obtain a stable baseline signal. Finally, the signals are transmitted to subsequent processing circuits through shielded transmission lines to obtain the raw bioelectrical signals.

[0092] Step 2: Input the original bioelectric signal into a multi-stage adaptive filtering and amplification circuit to obtain an amplified and filtered analog signal;

[0093] In this embodiment, the original bioelectric signal is first input to a first-stage low-pass filter. This filter employs a second-order Butterworth structure with an adjustable cutoff frequency to remove high-frequency interference, resulting in a first-stage filtered signal. The first-stage filtered signal is then input to a second-stage variable-gain amplifier, whose gain parameter G is dynamically adjusted through real-time feedback control to optimize the signal's dynamic range, resulting in a dynamically range-optimized amplified signal. Next, the amplified signal is input to a third-stage dynamic bandpass filter. The center frequency fc and bandwidth parameter B of this filter are adaptively adjusted according to the signal characteristics, resulting in a frequency band-optimized second-stage filtered signal. The second-stage filtered signal undergoes adaptive notch filtering for power frequency noise, and phase-locked loop tracking technology is used to compensate for power frequency drift, effectively suppressing power frequency interference. Finally, the power frequency interference-suppressed filtered signal is output buffered and impedance matched to obtain the amplified and filtered analog signal.

[0094] Step 3: Perform analog-to-digital conversion on the amplified and filtered analog signal to obtain a digital bioelectric signal;

[0095] In this embodiment, the amplified and filtered analog signal is first conditioned to meet the conversion range requirements, resulting in a standardized signal. This standardized signal is then sampled and held to obtain a discrete-time sampling point sequence. Next, a high-precision ADC is used to quantize and encode the discrete-time sampling point sequence, yielding an initial digital bitstream. The sampling rate of this ADC is dynamically adjusted based on the signal's spectral characteristics, with a resolution of at least 16 bits. The initial digital bitstream undergoes quantization noise shaping to obtain an optimized digital signal, followed by digital filtering and anti-aliasing processing to obtain an anti-aliasing digital signal. The anti-aliasing digital signal is then time-calibrated to obtain a synchronized digital signal, and its data format is converted to obtain standard format digital data. Finally, a buffer management mechanism is used for data flow control to obtain a digitized bioelectrical signal.

[0096] Step 4: Perform adaptive wavelet transform denoising on the digitized bioelectric signal to obtain a primary denoised signal;

[0097] In this embodiment, the signal characteristics of the digitized bioelectric signal are first analyzed to obtain a signal spectrum feature map, and the optimal wavelet basis function is selected accordingly to obtain an adapted wavelet basis. Multi-scale wavelet decomposition of the digitized bioelectric signal is then performed using the adapted wavelet basis to obtain a multi-level wavelet coefficient set. Threshold analysis is then performed on these coefficients to obtain a noise threshold distribution. An adaptive threshold function is designed based on the noise threshold distribution to obtain an optimized threshold strategy, which is then applied to perform soft thresholding on the wavelet coefficients to obtain denoised wavelet coefficients. A signal singularity detection algorithm is applied to the denoised wavelet coefficients to obtain a singularity protection map, and the coefficients are corrected accordingly to obtain a feature-preserving wavelet coefficient set. Finally, the feature-preserving wavelet coefficient set is used for inverse wavelet transform reconstruction to obtain the primary denoised signal.

[0098] Step 5: Perform empirical mode decomposition on the primary noise-reduced signal to obtain a multi-level intrinsic mode function set;

[0099] In this embodiment, firstly, extreme points are detected in the primary denoised signal to obtain a signal extreme point distribution map. Based on this distribution map, upper and lower envelopes are constructed to obtain envelope pairs. The mean envelope is calculated based on the envelope pairs to obtain a local mean function. The primary denoised signal is then subtracted from the local mean function to obtain candidate intrinsic mode functions (IMFs). The IMFs are then subjected to an intrinsicity test to obtain an intrinsicity evaluation result. If the evaluation result does not meet the IMF condition, the envelope calculation process is repeated until the IMF condition is met, resulting in the first layer of IMFs. The primary denoised signal is then subtracted from the first layer of IMFs to obtain the first residual signal. The empirical mode decomposition process is then repeated on the first residual signal to obtain subsequent IMFs. Finally, all the decomposed IMFs and the final residual signal are combined to obtain a complete multi-level IMF set.

[0100] Step 6: Extract the intrinsic mode functions of the characteristic frequency bands from the multi-level intrinsic mode function set to obtain the characteristic intrinsic mode function set;

[0101] In this embodiment, firstly, Hilbert transform is performed on each IMF in the multi-level intrinsic mode function set to obtain the instantaneous frequency distribution. Based on the instantaneous frequency distribution, the spectral center and bandwidth of each IMF are calculated to obtain an IMF spectral characteristic table. Based on the IMF spectral characteristic table, characteristic frequency band analysis of the biopotential signal is performed to obtain the target frequency band range. Relevant IMFs are then screened according to the target frequency band range to obtain a preliminary IMF set. Cross-correlation analysis is performed on the preliminary IMF set to obtain the IMF correlation matrix. Redundant components are removed based on the correlation matrix to obtain a simplified IMF set. The energy contribution of each IMF in the simplified IMF set is calculated to obtain the information content evaluation result. Based on the evaluation result, the IMFs are ranked by importance to obtain a ranked IMF set. Adaptive weight allocation is performed on the ranked IMF set to obtain a weighted IMF combination scheme. A characteristic intrinsic mode function set is then constructed based on the combination scheme.

[0102] Step 7: Reconstruct the signal based on the set of characteristic intrinsic mode functions to obtain a pure bioelectric signal;

[0103] In this embodiment, nonlinear phase correction is first performed on the characteristic intrinsic mode function set to obtain a phase-aligned IMF group. An optimal reconstruction weight vector is then designed for the phase-aligned IMF group to obtain an adaptive fusion strategy. Based on the adaptive fusion strategy, the IMFs are weighted and superimposed to obtain a preliminary reconstructed signal. Morphological feature enhancement is then applied to the preliminary reconstructed signal to obtain a bioelectrical signal with prominent features. Physiological model constraints are applied to the bioelectrical signal with prominent features to obtain a signal that conforms to physiological characteristics. Baseline drift is then finally corrected on the signal that conforms to physiological characteristics to obtain a stable zero-baseline signal. Abnormal waveform detection and repair are performed on the stable zero-baseline signal to obtain a corrected signal. The corrected signal is then bandwidth optimized and smoothed to obtain a final smooth signal. The final smooth signal is then morphologically matched with a standard bioelectrical template to obtain a pure bioelectrical signal.

[0104] Step 8: Perform real-time quality assessment on the pure bioelectric signal to obtain signal quality indicators; dynamically adjust system parameters based on the signal quality indicators to optimize the bioelectric signal acquisition process and obtain a high signal-to-noise ratio bioelectric potential signal output.

[0105] In this embodiment, the signal-to-noise ratio, effective bandwidth, and waveform distortion of the pure bioelectric signal are first calculated to obtain multi-dimensional quality assessment indicators. A comprehensive quality function is then constructed based on these indicators to obtain the signal quality index. A mapping relationship between the signal quality index and key system parameters is established to obtain a parameter optimization model. A feedback control algorithm is designed based on this model to obtain a system parameter adjustment strategy. The gain parameter G of the second-stage variable gain amplifier is updated in real time based on the system parameter adjustment strategy to obtain dynamic optimized gain control. The bandwidth parameter B of the third-stage dynamic bandpass filter is adaptively adjusted according to the signal characteristics to obtain an optimized bandwidth setting. The dynamic optimized gain control and optimized bandwidth setting are fed back to the hardware circuit to obtain a closed-loop self-optimizing structure. Based on this structure, the electrode array channel configuration is dynamically adjusted to obtain an optimized acquisition path. By combining the synergistic effect of the closed-loop self-optimizing structure and the optimized acquisition path, adaptive optimization of the entire bioelectric potential signal acquisition process is achieved, resulting in a high signal-to-noise ratio bioelectric potential signal output.

[0106] See Figure 2 The diagram below illustrates the detailed implementation steps of step 1. In this embodiment, the detailed implementation steps of step 1 include:

[0107] Nanomaterial modification treatment is applied to the electrode surface to obtain a low-impedance bioelectrode, and the low-impedance bioelectrode is arranged to form an adaptive structure electrode array to obtain a multi-channel acquisition network.

[0108] Automatic measurement of electrode-skin interface impedance is performed on the multi-channel acquisition network to obtain interface contact quality data. Channel selection is then optimized based on the interface contact quality data to obtain an optimized acquisition channel group.

[0109] The optimized acquisition channel group is used to collect changes in bioelectric potential on the human body surface in real time to obtain multi-channel raw signals. Inter-channel interference suppression is performed on the multi-channel raw signals to obtain preliminary isolated signals.

[0110] The initial isolated signal is subjected to differential amplification preprocessing to obtain a differential enhanced signal, and the differential enhanced signal is subjected to preliminary baseline drift correction to obtain a stable baseline signal;

[0111] The stable baseline signal is transmitted to the subsequent processing circuit through a shielded transmission line to obtain the original bioelectric signal.

[0112] In this embodiment, nanomaterials (such as nano-silver, nano-gold, graphene, etc.) are used to modify the surface of traditional bioelectrodes, significantly reducing the impedance at the electrode-skin interface and improving signal acquisition quality. The nanomaterials are fixed to the electrode surface through physical deposition or chemical modification, forming a porous nanostructure that increases the effective surface area of ​​the electrode while improving conductivity and biocompatibility. The low-impedance bioelectrodes are arranged into an adaptive electrode array according to human anatomical characteristics and the target bioelectrical signal characteristics, forming a multi-channel acquisition network that improves spatial resolution and anti-interference capability.

[0113] The system utilizes a built-in impedance measurement circuit to automatically measure the impedance at the electrode-skin interface in real time, acquiring data on the interface contact quality. The measurement principle is based on injecting a weak AC signal (typically 1-10kHz, <10μA) and analyzing the voltage response to calculate the impedance value. Based on the interface contact quality data, an optimization algorithm (such as greedy search or genetic algorithm) is used to optimize channel selection, identifying the electrode combination with the best signal quality to form an optimized acquisition channel group.

[0114] By optimizing the acquisition channel group, changes in the bioelectric potential of the human body surface are acquired in real time to obtain multi-channel raw signals. During the acquisition process, Drive Right Leg Circuit (DRL) technology is applied to suppress common-mode interference and improve signal quality. Spatial filtering algorithms (such as Laplace filtering or independent component analysis) are applied to the multi-channel raw signals to effectively suppress crosstalk and interference between channels, obtaining preliminary isolated signals.

[0115] The initial isolated signal is pre-amplified differentially using an instrumentation amplifier with high input impedance (>10GΩ) and low noise (<1μVrms) to enhance the useful signal and further suppress common-mode interference, resulting in a differentially enhanced signal. A high-pass filter (cutoff frequency 0.05-0.1Hz) is then applied to the differentially enhanced signal for initial baseline drift correction, eliminating low-frequency drift caused by factors such as breathing and body movement, thus obtaining a stable baseline signal.

[0116] A stable baseline signal is transmitted to subsequent processing circuits via a multi-layered shielded transmission line, with the shielding layer grounded to effectively isolate external electromagnetic interference. Differential transmission is employed to further enhance anti-interference capabilities. Signal integrity is maintained throughout transmission, ultimately yielding the original bioelectrical signal.

[0117] In this embodiment, the detailed implementation steps of step 2 include:

[0118] The original bioelectric signal is input into the first-stage low-pass filter to obtain a first-stage filtered signal with high-frequency interference removed. The first-stage low-pass filter adopts a second-order Butterworth structure and the cutoff frequency is adjustable.

[0119] The first-stage filtered signal is input into the second-stage variable gain amplifier to obtain a dynamically range-optimized amplified signal. The gain parameter G of the second-stage variable gain amplifier is dynamically adjusted through real-time feedback control.

[0120] The amplified signal with optimized dynamic range is input into the third-stage dynamic bandpass filter to obtain the second-stage filtered signal with optimized frequency band. The center frequency fc and bandwidth parameter B of the third-stage dynamic bandpass filter are adaptively adjusted according to the signal characteristics.

[0121] The frequency band optimized secondary filter signal is subjected to power frequency noise adaptive notch filtering to obtain a filter signal that suppresses power frequency interference. The power frequency noise adaptive notch filtering uses phase-locked loop tracking technology to compensate for power frequency drift.

[0122] The filtered signal used to suppress power frequency interference is output buffered and impedance matched to obtain an amplified and filtered analog signal.

[0123] In this embodiment, the original bioelectrical signal is input into a first-stage low-pass filter. This filter employs a second-order Butterworth structure, featuring a flat passband response and a relatively steep transition band, effectively suppressing high-frequency interference. The filter's cutoff frequency can be adjusted within the range of 1-500Hz to adapt to the frequency characteristics of different types of bioelectrical signals (such as ECG, EMG, EEG, etc.). The low-pass filter is implemented using a precision operational amplifier and high-precision passive components, ensuring the stability and consistency of filtering performance, resulting in a first-stage filtered signal free of high-frequency interference.

[0124] The first-stage filtered signal is input to a second-stage variable gain amplifier, which is based on a precision instrumentation amplifier and features a high common-mode rejection ratio (>100dB) and low noise characteristics (<0.5μVrms). The amplifier's gain parameter G is continuously adjustable within the range of 10-10000, achieved through a digitally controlled potentiometer or a digitally controlled gain amplifier. The gain parameter G is dynamically adjusted by a real-time feedback control system, automatically optimizing the gain setting based on the signal amplitude to avoid signal truncation or insufficient dynamic range utilization, thus obtaining a dynamically optimized amplified signal.

[0125] The dynamically range-optimized amplified signal is input to a third-stage dynamic bandpass filter. This filter employs a state-variable filter structure with independently adjustable center frequency fc and bandwidth parameter B. The center frequency fc can be adjusted within the range of 0.5-200Hz, and the bandwidth parameter B can be adjusted within the range of 0.1-50Hz, adaptively setting the optimal parameters according to the spectral characteristics of the target bioelectric signal. The filter uses a high-quality Q factor design to ensure in-band signal passage while effectively suppressing out-of-band interference, resulting in a frequency band-optimized second-stage filtered signal.

[0126] The frequency band-optimized secondary filtered signal undergoes adaptive notch filtering for power frequency noise, effectively suppressing 50 / 60Hz power frequency interference and its harmonics. The notch filter employs phase-locked loop (PLL) tracking technology to detect the actual frequency of the power frequency signal in real time (considering the ±0.5Hz drift of the power grid frequency), and dynamically adjusts the notch filter's center frequency accordingly, ensuring that the notch filter is always aligned with the actual power frequency to maximize the suppression effect. The notch filter adopts a high-Q, narrow-band design to minimize the impact on the useful signal, resulting in a filtered signal that suppresses power frequency interference.

[0127] Output buffering and impedance matching are performed on the filtered signal to suppress power frequency interference. A buffer amplifier with low output impedance (<50Ω) is used to ensure that the signal is not affected by the load during transmission. Simultaneously, impedance matching is performed to match the output impedance with the input impedance of the subsequent analog-to-digital conversion circuit, maximizing signal transmission efficiency and reducing reflection and distortion. The buffer circuit has sufficient bandwidth and drive capability to ensure complete signal transmission, resulting in the amplified and filtered analog signal.

[0128] In this embodiment, the detailed implementation steps of step 3 include:

[0129] The amplified and filtered analog signal is conditioned to obtain a standardized signal that meets the conversion range, and the standardized signal is sampled and held to obtain a discrete-time sampling point sequence.

[0130] The discrete-time sampling point sequence is quantized and encoded using a high-precision ADC to obtain an initial digital code stream. The sampling rate of the high-precision ADC is dynamically adjusted according to the signal spectrum characteristics, and the resolution is not less than 16 bits.

[0131] The initial digital bitstream is quantized and noise shaped to obtain an optimized digital signal, and the optimized digital signal is then subjected to digital filtering and anti-aliasing processing to obtain an anti-aliasing digital signal.

[0132] The anti-aliasing digital signal is time-calibrated to obtain a synchronous digital signal, and the synchronous digital signal is converted into a data format to obtain standard format digital data.

[0133] The standard format digital data is controlled through a buffer management mechanism to obtain digital bioelectric signals.

[0134] In this embodiment, the amplified and filtered analog signal is conditioned to ensure that its amplitude range matches the input range of the ADC (e.g., ±2.5V), avoiding truncation and low quantization efficiency. The conditioning circuit includes a precision amplifier and a level bias circuit, which can perform automatic gain control and zero-point calibration according to the signal characteristics to obtain a standardized signal that meets the conversion range. The standardized signal is sampled using a high-performance sample-and-hold circuit, which has a fast acquisition time (<100ns) and low voltage drop characteristics (<0.01%), ensuring sampling accuracy. The sampling frequency is dynamically adjusted according to the signal bandwidth, following the Nyquist sampling theorem, and is generally set to 2.5-5 times the highest frequency of the signal to obtain a discrete-time sampling point sequence.

[0135] A high-precision ADC is used to quantize and encode the discrete-time sampling point sequence, converting the analog signal into a digital representation. The ADC employs a 16-24 bit high-resolution architecture (such as successive approximation or Sigma-Delta type) to ensure accurate capture of minute changes in the bioelectrical signal. The ADC's sampling rate is dynamically adjusted according to the signal's spectral characteristics, typically within the range of 1kHz-10kHz, ensuring sufficient time resolution. The ADC features low noise (<1μVrms), high linearity (INL<0.5LSB), and low power consumption (<10mW), making it suitable for portable bioelectrical signal acquisition devices. The quantization process generates an initial digital code stream, representing the sampled signal amplitude in binary form.

[0136] Digital signal processing techniques are applied to the initial digital bitstream for quantization noise shaping, redistributing the quantization noise power spectrum and pushing the noise out of the signal band, thereby improving the signal-to-noise ratio within the signal band. Common techniques include oversampling and Sigma-Delta modulation, which can increase the effective bit depth by 1-4 bits. After quantization noise shaping, an optimized digital signal is obtained, which is then subjected to anti-aliasing processing using a digital filter to remove aliasing components that may have been introduced during sampling. The digital filter employs an FIR structure, possessing linear phase characteristics to ensure that the signal waveform remains undistorted, resulting in an anti-aliasing digital signal.

[0137] Timing calibration is performed on the anti-aliasing digital signals to compensate for time delays introduced by various components in the system, ensuring accurate phase relationships between multi-channel signals. The calibration process employs correlation analysis and interpolation techniques to precisely align signal timing, resulting in synchronized digital signals. The synchronized digital signals undergo data format conversion, transforming the raw binary data into a standardized format (such as IEEE 754 floating-point or fixed-point format) for easier subsequent processing and storage. Simultaneously, unit conversion is performed, converting the digital encoding output by the ADC into actual physical quantity units (such as μV), yielding standard format digital data.

[0138] Standard format digital data is managed through a buffer system to control data flow, balancing the difference between data generation and processing rates to prevent data loss or processing delays. The buffer employs a FIFO (First-In, First-Out) structure with overflow protection and underload handling mechanisms to ensure the continuity and integrity of the data flow. The buffer size is dynamically adjusted based on system processing capabilities and data generation rate, typically ranging from 1 to 32KB, ultimately resulting in a stable and reliable digital bioelectric signal.

[0139] In this embodiment, the detailed implementation steps of step 4 include:

[0140] The digitized bioelectric signal is subjected to signal characteristic analysis to obtain a signal spectrum feature map, and the optimal wavelet basis function is selected based on the signal spectrum feature map to obtain the suitable wavelet basis;

[0141] The adapted wavelet basis is used to perform multi-scale wavelet decomposition on the digitized bioelectric signal to obtain a multi-layer wavelet coefficient set, and threshold analysis is performed on the multi-layer wavelet coefficient set to obtain the noise threshold distribution.

[0142] An adaptive threshold function is designed based on the noise threshold distribution to obtain an optimized threshold strategy. The optimized threshold strategy is then applied to the wavelet coefficients to perform soft thresholding to obtain denoised wavelet coefficients.

[0143] The signal singularity detection algorithm is applied to the denoised wavelet coefficients to obtain the singularity protection map, and the coefficients are corrected according to the singularity protection map to obtain the wavelet coefficient set that preserves the features.

[0144] The inverse wavelet transform is used to reconstruct the primary denoised signal using the wavelet coefficient set that preserves the characteristics.

[0145] In this embodiment, time-frequency analysis is first performed on the digitized bioelectric signal, including preliminary analysis using short-time Fourier transform (STFT) and wavelet transform, to obtain the signal's time-frequency characteristics. Characteristic parameters such as energy distribution, spectral center, bandwidth, and peak frequency are calculated to construct a signal spectral feature map. Based on the signal spectral feature map, the optimal wavelet basis function is selected from a candidate wavelet library (including wavelet families such as Daubechies, Symlet, Coiflet, and Biorthogonal). The selection criteria are based on indicators such as the similarity between the wavelet basis and the signal morphology, energy concentration, and entropy minimization, ultimately determining the most suitable wavelet basis for the current bioelectric signal characteristics.

[0146] Multi-scale wavelet decomposition of digitized bioelectrical signals is performed using an adapted wavelet basis, decomposing the signal into wavelet coefficients at different frequency scales. The number of decomposition levels is determined based on the signal's spectral characteristics, typically 3-8 levels, to ensure sufficient separation of the signal's main frequency components. The decomposition process is implemented using a fast wavelet transform algorithm, resulting in a multi-level wavelet coefficient set containing approximation and detail coefficients. Statistical analysis is performed on the detail coefficients at each level of the multi-level wavelet coefficient set to estimate the distribution characteristics of noise components. The noise level is estimated based on the MAD (Median Absolute Deviation) method, and the noise threshold at each scale is calculated to obtain the noise threshold distribution.

[0147] Based on the noise threshold distribution, an adaptive threshold function is designed, which dynamically adjusts the threshold size according to different decomposition scales and local signal characteristics. The threshold function adopts a scale-dependent design, using a smaller threshold to preserve signal structure at low-frequency scales and a larger threshold to enhance noise reduction at high-frequency scales. Simultaneously, a local signal energy factor is introduced to lower the threshold in regions of concentrated signal energy, protecting useful information, resulting in an optimized threshold strategy. This optimized threshold strategy is applied to soft-threshold wavelet coefficients; the soft-threshold function is defined as:

[0148] η(d, λ) = sign(d)·max(0, |d|-λ); where d is the wavelet coefficient and λ is the threshold. Soft thresholding can effectively reduce the discontinuity of the processed signal compared to hard thresholding, resulting in denoised wavelet coefficients.

[0149] A signal singularity detection algorithm is applied to the denoised wavelet coefficients to identify important feature points in the signal (such as spikes, abrupt changes, and other key waveforms in bioelectrical signals). The detection algorithm, based on Lipschitz exponent analysis, calculates the attenuation characteristics of the wavelet coefficients at different scales to determine whether a coefficient corresponds to a signal singularity. A singularity protection map is constructed based on the detection results, marking the locations of wavelet coefficients requiring special protection. The denoised wavelet coefficients are then corrected according to the singularity protection map; coefficients corresponding to singularities are reduced or thresholding is removed to ensure that important features are not overly smoothed, resulting in a set of wavelet coefficients that preserves their characteristics.

[0150] The time-domain signal is reconstructed using an inverse wavelet transform based on a feature-preserving set of wavelet coefficients. The reconstruction process employs the same wavelet basis functions as the decomposition process, ensuring signal integrity and consistency. The reconstruction algorithm utilizes a fast inverse wavelet transform, offering high computational efficiency suitable for real-time processing. Boundary processing techniques are applied during reconstruction to reduce the impact of boundary effects on the signal. The reconstructed signal retains the main characteristics of the original bioelectrical signal while effectively suppressing noise components, forming a primary denoised signal.

[0151] In this embodiment, the detailed implementation steps of step 5 include:

[0152] Extreme point detection is performed on the primary noise reduction signal to obtain a signal extreme value distribution map, and upper and lower envelopes are constructed based on the signal extreme value distribution map to obtain envelope pairs;

[0153] The mean envelope is calculated based on the envelope pair to obtain the local mean function, and the primary noise-reduced signal is subtracted from the local mean function to obtain the candidate intrinsic mode function;

[0154] The candidate intrinsic mode functions are subjected to intrinsicity testing to obtain intrinsicity evaluation results. If the intrinsicity evaluation results do not meet the IMF conditions, the envelope calculation process is repeated until the IMF conditions are met to obtain the first layer of intrinsic mode functions.

[0155] The first layer of intrinsic mode functions is subtracted from the primary noise-reduced signal to obtain the first residual signal, and the empirical mode decomposition process is repeated on the first residual signal to obtain the subsequent intrinsic mode functions.

[0156] By combining all the intrinsic mode functions obtained from the decomposition with the final residual signal, a complete multi-level intrinsic mode function set is obtained.

[0157] In this embodiment, the primary noise-reduced signal is first subjected to extremum detection to identify all local maxima and minima in the signal. The detection algorithm is based on derivative analysis and uses a three-point comparison method to identify extrema, while applying smoothing preprocessing to reduce the influence of false extrema. The detected extrema are arranged in chronological order to form a signal extremum distribution map, which reflects the oscillation characteristics of the signal. Based on the signal extremum distribution map, the upper envelope (connecting all maxima) and lower envelope (connecting all minima) of the signal are constructed using interpolation methods. Cubic spline interpolation is used to ensure the smoothness and accuracy of the envelopes, resulting in envelope pairs.

[0158] The arithmetic mean of the upper and lower envelopes is calculated to obtain the mean envelope, which represents the local mean function of the signal. The local mean function reflects the low-frequency trend components of the signal. Subtracting the local mean function from the primary noise-reduced signal yields candidate intrinsic mode functions (IMFs). Theoretically, a candidate IMF should be a zero-mean oscillatory function, but in practice, further processing may be needed to satisfy the strict definition of the IMF.

[0159] The eigenfunctions of candidate intrinsic modes are tested for eigenvalues ​​to assess whether they satisfy the two conditions of the IMF:

[0160] 1. The difference between the number of extreme points and the number of zero-crossing points does not exceed 1;

[0161] 2. The local mean function approaches zero. Calculate the standard deviation ratio (SD) of candidate IMFs:

[0162] Where h(k-1)(t) and hk(t) are the candidate IMFs in the previous and next iterations, respectively. When the SD is less than a preset threshold (usually 0.2-0.3), the candidate IMF is considered to satisfy the intrinsic condition. If the intrinsic condition is not met, the candidate IMF is used as a new input signal, and the envelope construction and mean function calculation process is repeated for screening iterations until the IMF condition is met or the maximum number of iterations (usually 10) is reached, thus obtaining the first layer of intrinsic mode functions.

[0163] Subtracting the first layer of intrinsic mode functions (IMFs) from the primary denoised signal yields the first residual signal. This residual signal contains the remaining low-frequency components of the signal. The empirical mode decomposition (EMF) process is repeated on this residual signal to obtain the second layer of IMFs. This process continues, extracting IMFs layer by layer, until the residual signal becomes a monotonic function or its energy falls below a preset threshold, making it impossible to extract meaningful IMFs. All the decomposed IMFs (arranged from high to low frequency) and the final residual signal (representing the signal trend) are combined to form a complete multi-level IMF set, which fully expresses the oscillation modes of the original signal at different feature scales.

[0164] In this embodiment, the detailed implementation steps of step 6 include:

[0165] Hilbert transform is performed on each IMF in the multi-level intrinsic mode function set to obtain the instantaneous frequency distribution, and the spectral center and bandwidth of each IMF are calculated based on the instantaneous frequency distribution to obtain the IMF spectral characteristic table.

[0166] Based on the IMF spectral characteristic table, the characteristic frequency band of the biopotential signal is analyzed to obtain the target frequency band range, and relevant IMFs are screened according to the target frequency band range to obtain a preliminary IMF set;

[0167] Cross-correlation analysis is performed on the initially selected IMF set to obtain the IMF correlation matrix. Redundant components are then removed based on the IMF correlation matrix to obtain a simplified IMF set.

[0168] The energy contribution of each IMF in the simplified IMF set is calculated to obtain the information content assessment result. The importance is then ranked according to the information content assessment result to obtain the ranked IMF set.

[0169] Adaptive weight allocation is performed based on the sorted IMF set to obtain a weighted IMF combination scheme, and a feature intrinsic mode function set is constructed according to the weighted IMF combination scheme.

[0170] In this embodiment, Hilbert transform is first applied to each IMF in the multi-level intrinsic mode function set to convert the real signal into an analytic signal, obtaining the instantaneous amplitude and instantaneous phase of the signal. The instantaneous frequency distribution is obtained by calculating the derivative of the instantaneous phase. Based on the instantaneous frequency distribution, the spectral center (average instantaneous frequency) and bandwidth (standard deviation of instantaneous frequency) of each IMF are calculated, and an IMF spectral characteristic table is constructed, which comprehensively describes the frequency characteristics of each IMF.

[0171] Based on the IMF spectral characteristic table and the known frequency characteristics of the target bioelectrical potential signal (e.g., the main frequency of ECG signals is 0.5-40Hz, the main frequency of EEG signals is 0.5-30Hz, and the main frequency of EMG signals is 20-500Hz), characteristic frequency band analysis is performed to determine the target frequency band range containing useful information. According to the target frequency band range, IMFs whose spectral centers fall within this range are selected to form a preliminary IMF set. The selection process considers frequency overlap to ensure complete coverage of important frequency bands, while excluding IMF components that significantly exceed the target range.

[0172] Perform cross-correlation analysis on each pair of IMFs (IMF_i and IMF_j) in the initial IMF set, and calculate their cross-correlation coefficient ρ_ij:

[0173] Where μ_i and μ_j are the means of IMF_i and IMF_j, respectively, and σ_i and σ_j are their standard deviations. All cross-correlation coefficients are organized into an n×n matrix (n is the number of initially selected IMFs) to obtain the IMF correlation matrix. Based on the correlation matrix, highly correlated IMF pairs (absolute correlation coefficient > 0.7) are identified. Through methods such as principal component analysis or information entropy comparison, the pair with the larger information content is retained, and redundant components are removed to obtain a simplified IMF set.

[0174] Calculate the energy contribution of each IMF (IMF_i) in the simplified IMF set to assess the amount of effective information it contains. The formula for calculating the energy contribution is:

[0175] Where x(t) represents the original signal. Simultaneously, by combining indicators such as the correlation between the IMF and the target bioelectrical signal template, and the signal-to-noise ratio, the information value of each IMF is comprehensively evaluated, yielding an information content assessment result. Based on the information content assessment result, the IMFs in the simplified IMF set are ranked by importance, arranged from largest to smallest information contribution, resulting in a ranked IMF set.

[0176] Based on a sorted IMF set, an adaptive weight allocation strategy is designed to assign reconstruction weights to different IMFs. The weight allocation considers factors such as the energy contribution of the IMF, the matching degree between its frequency characteristics and the target signal, and the signal-to-noise ratio. An optimization algorithm (such as a genetic algorithm or gradient descent) is used to solve for the optimal weight combination. After obtaining the weighted IMF combination scheme, key components from the sorted IMF set are selected and integrated according to this scheme to construct a set of characteristic intrinsic mode functions. This set contains the main characteristic components of the biopotential signal, providing a foundation for subsequent signal reconstruction.

[0177] In this embodiment, the detailed implementation steps of step 7 include:

[0178] Nonlinear phase correction is performed on the feature intrinsic mode function set to obtain a phase-aligned IMF group, and an optimal reconstruction weight vector is designed for the phase-aligned IMF group to obtain an adaptive fusion strategy.

[0179] The IMF is weighted and superimposed based on the adaptive fusion strategy to obtain a preliminary reconstructed signal, and the preliminary reconstructed signal is enhanced with morphological features to obtain a bioelectric signal with prominent features.

[0180] The bioelectric signals with prominent features are subjected to physiological model constraint processing to obtain signals that conform to physiological characteristics, and the signals that conform to physiological characteristics are subjected to baseline drift final correction to obtain stable zero baseline signals.

[0181] Abnormal waveform detection and repair are performed on the stable zero baseline signal to obtain a corrected signal, and the bandwidth of the corrected signal is optimized and smoothed to obtain the final smooth signal.

[0182] The final smooth signal was then compared with a standard bioelectric template to verify its morphological matching, resulting in a pure bioelectric signal.

[0183] In this embodiment, nonlinear phase correction is first performed on each IMF in the characteristic intrinsic mode function set to address mode aliasing and phase distortion problems that may be introduced during EMD decomposition. Phase correction employs group delay compensation technology. A phase correction filter is designed based on the frequency characteristics of each IMF to adjust the phase relationship between IMFs, ensuring correct alignment during reconstruction, resulting in a phase-aligned IMF group. An optimal reconstruction weight vector is designed for the phase-aligned IMF group, determining the contribution ratio of each IMF in the reconstruction process. The weight design uses an objective function optimization method to minimize the error between the reconstructed signal and the ideal bioelectrical signal template, while considering smoothness and physiological rationality constraints, resulting in an adaptive fusion strategy.

[0184] Based on an adaptive fusion strategy, the IMF is weighted and superimposed, and the reconstructed formula is:

[0185] x1(t) = ∑[w_i·IMF_i(t)] + rn(t); where w_i is the IMF weight and rn(t) is the residual signal. During the weighted superposition process, window function techniques are applied to handle boundary effects, ensuring the stability of the reconstructed signal across the entire time domain, resulting in a preliminary reconstructed signal. Morphological algorithms are then applied to the preliminary reconstructed signal for feature enhancement, highlighting key waveform features of the bioelectrical signal (such as the QRS complex in ECG signals and the alpha wave in EEG signals). The enhancement process employs an adaptive morphological operator, dynamically adjusting the structural elements according to signal characteristics, preserving features while avoiding the introduction of artifacts, resulting in a bioelectrical signal with prominent features.

[0186] Physiological model constraints are applied to bioelectrical signals with prominent characteristics to ensure that the signals conform to bioelectrical physiological properties. Constraint processing is based on bioelectrical signal models constructed from prior knowledge (such as the Pan-Tompkins model for ECG signals and the ARMA model for EEG signals), which are used to verify and adjust the rationality of parameters such as waveform, amplitude range, and temporal relationships. A Bayesian inference framework is employed to correct abnormal components that do not conform to physiological laws while maintaining the main characteristics of the signal, resulting in a signal that conforms to physiological characteristics. Baseline drift final correction is then performed on the physiologically consistent signal to eliminate residual low-frequency drift components. The correction uses a combination of high-pass filtering and polynomial fitting to achieve zero-phase filtering, avoiding the introduction of phase distortion and obtaining a stable zero-baseline signal.

[0187] Anomaly detection and repair are performed on stable zero-baseline signals to identify and correct potential sudden anomalies (such as spikes, steps, etc.). The detection algorithm is based on the statistical characteristics and morphological features of anomalies, employing machine learning methods (such as support vector machines or random forests) for anomaly waveform classification. For the identified anomaly waveforms, adaptive interpolation or template replacement methods are used for repair, maintaining signal continuity and consistency to obtain the corrected signal. The corrected signal undergoes bandwidth optimization and smoothing to precisely control its frequency band characteristics and remove residual high-frequency noise and irregular fluctuations. A zero-phase FIR filter is used for processing, with filter parameters optimized based on the spectral characteristics of the target bioelectrical signal to ensure effective noise suppression while maximizing the retention of useful information, resulting in a final smooth signal.

[0188] The final smoothed signal was morphologically matched with a standard bioelectric template to verify its authenticity and accuracy. The matching process employed a Dynamic Time Warping (DTW) algorithm to address potential timescale variations between different signals and accurately assess waveform similarity. Simultaneously, the positional and amplitude errors of key feature points (such as peaks, troughs, and zero-crossings) were calculated to comprehensively evaluate signal quality. Signals that passed verification were confirmed as high-quality, pure bioelectric signals, suitable for subsequent physiological parameter extraction and medical diagnostic analysis.

[0189] In this embodiment, the detailed implementation steps of step 8 include:

[0190] The signal-to-noise ratio, effective bandwidth, and waveform distortion of the pure bioelectric signal are calculated to obtain a multidimensional quality assessment index. A comprehensive quality function is then constructed based on the multidimensional quality assessment index to obtain the signal quality index.

[0191] Establish the mapping relationship between the signal quality index and the key system parameters to obtain the parameter optimization model, and design a feedback control algorithm based on the parameter optimization model to obtain the system parameter adjustment strategy;

[0192] Based on the system parameter adjustment strategy, the gain parameter G of the second-stage variable gain amplifier is updated in real time to obtain dynamic optimized gain control, and the bandwidth parameter B of the third-stage dynamic bandpass filter is adaptively adjusted according to the signal characteristics to obtain optimized frequency band settings.

[0193] The dynamic optimization gain control and the optimized frequency band setting are fed back to the hardware circuit to obtain a closed-loop self-optimizing structure. Based on the closed-loop self-optimizing structure, the electrode array channel configuration is dynamically adjusted to obtain an optimized acquisition path.

[0194] By combining the synergistic effect of the closed-loop self-optimizing structure and the optimized acquisition path, adaptive optimization of the entire bioelectric potential signal acquisition process is achieved, resulting in a high signal-to-noise ratio bioelectric potential signal output.

[0195] In this embodiment, multiple quality assessment indicators are first calculated for the pure bioelectric signal. The signal-to-noise ratio (SNR) is calculated using the ratio of signal energy to noise energy, estimated through spectral analysis or comparison with a reference signal. The effective bandwidth is determined by calculating the frequency range where the signal energy is concentrated, defined as the minimum bandwidth containing 90% of the signal energy. Waveform distortion is quantified by calculating the root mean square error (RMSE) between the signal and an ideal template or the structural similarity index (SSIM). These indicators are combined to form a multidimensional quality assessment index, comprehensively reflecting different aspects of signal quality. A comprehensive quality function is constructed based on the multidimensional quality assessment index. Using weighted summation or fuzzy logic methods, the multidimensional indicators are mapped to a single comprehensive evaluation value, resulting in the Signal Quality Index (SQI), typically ranging from 0 to 1, with higher values ​​indicating better signal quality.

[0196] A mapping relationship between the signal quality index and key system parameters is established, and the impact of different parameter settings on signal quality is analyzed. Key parameters include amplifier gain G, filter bandwidth B, center frequency fc, and electrode configuration. The mapping relationship is expressed using a mathematical model, which can be a regression model based on experimental data or an analytical model based on theoretical analysis. The model can predict the impact of parameter changes on signal quality, providing a theoretical basis for parameter optimization, thus obtaining a parameter optimization model. Based on the parameter optimization model, a feedback control algorithm is designed. This algorithm can calculate the optimal parameter adjustment direction and magnitude based on the current signal quality index. The control algorithm employs methods such as PID control, model predictive control, or reinforcement learning, exhibiting fast convergence and stable control characteristics, thus obtaining a system parameter adjustment strategy.

[0197] The gain parameter G of the second-stage variable gain amplifier is updated in real time based on the system parameter adjustment strategy. The gain adjustment formula is:

[0198] G(t+1) = G(t) + ΔG; where G(t) is the gain value at the current time t, G(t+1) is the gain value at the next time, and ΔG is calculated according to the control algorithm, generally within the range of ±1dB to ±10dB. Gain adjustment is achieved through a digitally controlled potentiometer or a programmable gain amplifier, with a response time of <10ms, ensuring that the system can quickly adapt to changes in signal amplitude and achieve dynamically optimized gain control. Similarly, the bandwidth parameter B and center frequency fc of the third-stage dynamic bandpass filter are adaptively adjusted according to the signal characteristics to optimize the filter's frequency response characteristics. The bandwidth adjustment range is generally between ±5Hz and ±20Hz, and the center frequency adjustment range is between ±2Hz and ±10Hz. Adjustment is achieved through digitally controlled capacitors or switched capacitor filter technology to obtain optimized frequency band settings.

[0199] The parameters for dynamically optimized gain control and optimized frequency band settings are fed back to the hardware circuit in real time via a digital control interface, enabling dynamic adjustment of system parameters and forming a closed-loop self-optimizing structure. The update frequency of the feedback control loop is 10-50Hz, meeting real-time processing requirements. Based on the closed-loop self-optimizing structure, the dynamic adjustment of the electrode array channel configuration is further realized. The optimal acquisition electrode combination is automatically selected based on the signal quality of each channel, and the signal path is dynamically switched to obtain the optimized acquisition path. The channel optimization algorithm considers electrode-skin contact impedance, environmental interference distribution, and target signal characteristics, comprehensively evaluating the signal quality of each channel and selecting the optimal channel combination.

[0200] By combining the synergistic effect of a closed-loop self-optimizing structure and an optimized acquisition path, the system achieves adaptive optimization throughout the entire bioelectric potential signal acquisition process. The system can adjust parameters at each stage in real time based on changes in the environment, the patient's condition, and signal characteristics, maintaining optimal operating conditions. The optimization process considers a multi-objective balance, ensuring system stability and power efficiency while improving the signal-to-noise ratio (SNR). Ultimately, a high SNR bioelectric potential signal output is obtained, with an SNR improvement of 10-20 dB, effective bandwidth precisely matching the target signal characteristics, and waveform distortion reduced by more than 50%, providing a high-quality data foundation for subsequent bioelectric signal analysis and medical diagnosis.

[0201] The low-noise biopotential signal acquisition and processing method in the embodiments of this application has been described above. The low-noise biopotential signal acquisition system in the embodiments of this application is described below. Please refer to [link / reference]. Figure 3 One embodiment of the low-noise biopotential signal acquisition system in this application includes:

[0202] The electrode acquisition module is used to acquire bioelectric potential signals from the human body through an electrode array to obtain raw bioelectric signals;

[0203] A multi-stage adaptive filtering and amplification module is used to input the original bioelectric signal into a multi-stage adaptive filtering and amplification circuit to obtain an amplified and filtered analog signal.

[0204] An analog-to-digital conversion module is used to perform analog-to-digital conversion on the amplified and filtered analog signal to obtain a digital bioelectric signal.

[0205] The wavelet transform denoising module is used to perform adaptive wavelet transform denoising processing on the digitized bioelectric signal to obtain a primary denoised signal.

[0206] The empirical mode decomposition module is used to perform empirical mode decomposition on the primary noise-reduced signal to obtain a multi-level intrinsic mode function set;

[0207] The feature extraction module is used to extract the intrinsic mode functions of the feature frequency bands from the multi-level intrinsic mode function set to obtain the feature intrinsic mode function set;

[0208] The signal reconstruction module is used to reconstruct the signal based on the set of characteristic intrinsic mode functions to obtain a pure bioelectric signal.

[0209] The quality assessment and parameter optimization module is used to perform real-time quality assessment on the pure bioelectric signal to obtain signal quality indicators; based on the signal quality indicators, the system parameters are dynamically adjusted to optimize the bioelectric signal acquisition process and obtain a high signal-to-noise ratio bioelectric potential signal output.

[0210] This invention improves the signal-to-noise ratio and fidelity of signals, enabling the precise capture of weak bioelectrical signals in complex environments. It effectively overcomes the problems of signal instability, feature loss, and poor adaptability in traditional techniques, achieving high-quality processing of non-stationary biological signals. The closed-loop self-optimization mechanism established by the method can adjust system parameters in real time according to signal quality, giving the system excellent environmental adaptability and individual compatibility, providing a stable and reliable signal foundation for different application scenarios. By preserving key physiological characteristics of the signal while effectively suppressing various interferences, the accuracy and reliability of bioelectrical signal analysis are significantly improved.

[0211] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0212] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0213] In the description of this invention, it should be understood that the terms "first," "second," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0214] In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0215] In the description of this invention, "several" means one or more, and "a large number" means two or more.

[0216] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0217] All formulas in this manual are dimensionless and calculated numerically. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0218] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for acquiring and processing low-noise biopotential signals, characterized in that, include: Step 1: Collect bioelectric potential signals from the human body through an electrode array to obtain the raw bioelectric signals; Step 2: Input the original bioelectric signal into a multi-stage adaptive filtering and amplification circuit to obtain an amplified and filtered analog signal; Step 3: Perform analog-to-digital conversion on the amplified and filtered analog signal to obtain a digital bioelectric signal; Step 4: Perform adaptive wavelet transform denoising on the digitized bioelectric signal to obtain a primary denoised signal; Step 5: Perform empirical mode decomposition on the primary noise-reduced signal to obtain a multi-level intrinsic mode function set; Step 6: Extract the intrinsic mode functions of the characteristic frequency bands from the multi-level intrinsic mode function set to obtain the characteristic intrinsic mode function set; Step 7: Reconstruct the signal based on the set of characteristic intrinsic mode functions to obtain a pure bioelectric signal; Step 8: Perform real-time quality assessment on the purified bioelectric signal to obtain a signal quality index; dynamically adjust system parameters based on the signal quality index to optimize the bioelectric signal acquisition process and obtain a high signal-to-noise ratio bioelectric potential signal output, including: The signal-to-noise ratio, effective bandwidth, and waveform distortion of the pure bioelectric signal are calculated to obtain a multidimensional quality assessment index. A comprehensive quality function is then constructed based on the multidimensional quality assessment index to obtain the signal quality index. Establish the mapping relationship between the signal quality index and the key system parameters to obtain the parameter optimization model, and design a feedback control algorithm based on the parameter optimization model to obtain the system parameter adjustment strategy; Based on the system parameter adjustment strategy, the gain parameter G of the second-stage variable gain amplifier is updated in real time to obtain dynamic optimized gain control, and the bandwidth parameter B of the third-stage dynamic bandpass filter is adaptively adjusted according to the signal characteristics to obtain optimized frequency band settings. The dynamic optimization gain control and the optimized frequency band setting are fed back to the hardware circuit to obtain a closed-loop self-optimizing structure. Based on the closed-loop self-optimizing structure, the electrode array channel configuration is dynamically adjusted to obtain an optimized acquisition path. By combining the synergistic effect of the closed-loop self-optimizing structure and the optimized acquisition path, adaptive optimization of the entire bioelectric potential signal acquisition process is achieved, resulting in a high signal-to-noise ratio bioelectric potential signal output.

2. The low-noise biopotential signal acquisition and processing method according to claim 1, characterized in that, The process of acquiring bioelectric potential signals from the human body via an electrode array to obtain raw bioelectric signals includes: Nanomaterial modification treatment is applied to the electrode surface to obtain a low-impedance bioelectrode, and the low-impedance bioelectrode is arranged to form an adaptive structure electrode array to obtain a multi-channel acquisition network. Automatic measurement of electrode-skin interface impedance is performed on the multi-channel acquisition network to obtain interface contact quality data. Channel selection is then optimized based on the interface contact quality data to obtain an optimized acquisition channel group. The optimized acquisition channel group is used to collect changes in bioelectric potential on the human body surface in real time to obtain multi-channel raw signals. Inter-channel interference suppression is performed on the multi-channel raw signals to obtain preliminary isolated signals. The initial isolated signal is subjected to differential amplification preprocessing to obtain a differential enhanced signal, and the differential enhanced signal is subjected to preliminary baseline drift correction to obtain a stable baseline signal; The stable baseline signal is transmitted to the subsequent processing circuit through a shielded transmission line to obtain the original bioelectric signal.

3. The low-noise biopotential signal acquisition and processing method according to claim 2, characterized in that, The step of inputting the original bioelectric signal into a multi-stage adaptive filtering and amplification circuit to obtain an amplified and filtered analog signal includes: The original bioelectric signal is input into the first-stage low-pass filter to obtain a first-stage filtered signal with high-frequency interference removed. The first-stage low-pass filter adopts a second-order Butterworth structure and the cutoff frequency is adjustable. The first-stage filtered signal is input into the second-stage variable gain amplifier to obtain a dynamically range-optimized amplified signal. The gain parameter G of the second-stage variable gain amplifier is dynamically adjusted through real-time feedback control. The amplified signal with optimized dynamic range is input into the third-stage dynamic bandpass filter to obtain the second-stage filtered signal with optimized frequency band. The center frequency fc and bandwidth parameter B of the third-stage dynamic bandpass filter are adaptively adjusted according to the signal characteristics. The frequency band optimized secondary filter signal is subjected to power frequency noise adaptive notch filtering to obtain a filter signal that suppresses power frequency interference. The power frequency noise adaptive notch filtering uses phase-locked loop tracking technology to compensate for power frequency drift. The filtered signal used to suppress power frequency interference is output buffered and impedance matched to obtain an amplified and filtered analog signal.

4. The low-noise biopotential signal acquisition and processing method according to claim 3, characterized in that, The step of performing analog-to-digital conversion on the amplified and filtered analog signal to obtain a digital bioelectric signal includes: The amplified and filtered analog signal is conditioned to obtain a standardized signal that meets the conversion range, and the standardized signal is sampled and held to obtain a discrete-time sampling point sequence. The discrete-time sampling point sequence is quantized and encoded using a high-precision ADC to obtain an initial digital code stream. The sampling rate of the high-precision ADC is dynamically adjusted according to the signal spectrum characteristics, and the resolution is not less than 16 bits. The initial digital bitstream is quantized and noise shaped to obtain an optimized digital signal, and the optimized digital signal is then subjected to digital filtering and anti-aliasing processing to obtain an anti-aliasing digital signal. The anti-aliasing digital signal is time-calibrated to obtain a synchronous digital signal, and the synchronous digital signal is converted into a data format to obtain standard format digital data. The standard format digital data is controlled through a buffer management mechanism to obtain digital bioelectric signals.

5. The low-noise biopotential signal acquisition and processing method according to claim 4, characterized in that, The adaptive wavelet transform denoising process performed on the digitized bioelectric signal to obtain a primary denoised signal includes: The digitized bioelectric signal is subjected to signal characteristic analysis to obtain a signal spectrum feature map, and the optimal wavelet basis function is selected based on the signal spectrum feature map to obtain the suitable wavelet basis; The adapted wavelet basis is used to perform multi-scale wavelet decomposition on the digitized bioelectric signal to obtain a multi-layer wavelet coefficient set, and threshold analysis is performed on the multi-layer wavelet coefficient set to obtain the noise threshold distribution. An adaptive threshold function is designed based on the noise threshold distribution to obtain an optimized threshold strategy. The optimized threshold strategy is then applied to the wavelet coefficients to perform soft thresholding to obtain denoised wavelet coefficients. The signal singularity detection algorithm is applied to the denoised wavelet coefficients to obtain the singularity protection map, and the coefficients are corrected according to the singularity protection map to obtain the wavelet coefficient set that preserves the features. The inverse wavelet transform is used to reconstruct the primary denoised signal using the wavelet coefficient set that preserves the characteristics.

6. The low-noise biopotential signal acquisition and processing method according to claim 5, characterized in that, The empirical mode decomposition of the primary noise-reduced signal yields a multi-level intrinsic mode function set, including: Extreme point detection is performed on the primary noise reduction signal to obtain a signal extreme value distribution map, and upper and lower envelopes are constructed based on the signal extreme value distribution map to obtain envelope pairs; The mean envelope is calculated based on the envelope pair to obtain the local mean function, and the primary noise-reduced signal is subtracted from the local mean function to obtain the candidate intrinsic mode function; The candidate intrinsic mode functions are subjected to intrinsicity testing to obtain intrinsicity evaluation results. If the intrinsicity evaluation results do not meet the IMF conditions, the envelope calculation process is repeated until the IMF conditions are met to obtain the first layer of intrinsic mode functions. The first layer of intrinsic mode functions is subtracted from the primary noise-reduced signal to obtain the first residual signal, and the empirical mode decomposition process is repeated on the first residual signal to obtain the subsequent intrinsic mode functions. By combining all the intrinsic mode functions obtained from the decomposition with the final residual signal, a complete multi-level intrinsic mode function set is obtained.

7. The low-noise biopotential signal acquisition and processing method according to claim 6, characterized in that, The step of extracting eigenmode functions of characteristic frequency bands from the multi-level eigenmode function set to obtain a characteristic eigenmode function set includes: Hilbert transform is performed on each IMF in the multi-level intrinsic mode function set to obtain the instantaneous frequency distribution, and the spectral center and bandwidth of each IMF are calculated based on the instantaneous frequency distribution to obtain the IMF spectral characteristic table. Based on the IMF spectral characteristic table, the characteristic frequency band of the biopotential signal is analyzed to obtain the target frequency band range, and relevant IMFs are screened according to the target frequency band range to obtain a preliminary IMF set; Cross-correlation analysis is performed on the initially selected IMF set to obtain the IMF correlation matrix. Redundant components are then removed based on the IMF correlation matrix to obtain a simplified IMF set. The energy contribution of each IMF in the simplified IMF set is calculated to obtain the information content assessment result. The importance is then ranked according to the information content assessment result to obtain the ranked IMF set. Adaptive weight allocation is performed based on the sorted IMF set to obtain a weighted IMF combination scheme, and a feature intrinsic mode function set is constructed according to the weighted IMF combination scheme.

8. The low-noise biopotential signal acquisition and processing method according to claim 7, characterized in that, The process of reconstructing the signal based on the set of characteristic intrinsic mode functions to obtain a pure bioelectric signal includes: Nonlinear phase correction is performed on the feature intrinsic mode function set to obtain a phase-aligned IMF group, and an optimal reconstruction weight vector is designed for the phase-aligned IMF group to obtain an adaptive fusion strategy. The IMF is weighted and superimposed based on the adaptive fusion strategy to obtain a preliminary reconstructed signal, and the preliminary reconstructed signal is enhanced with morphological features to obtain a bioelectric signal with prominent features. The bioelectric signals with prominent features are subjected to physiological model constraint processing to obtain signals that conform to physiological characteristics, and the signals that conform to physiological characteristics are subjected to baseline drift final correction to obtain stable zero baseline signals. Abnormal waveform detection and repair are performed on the stable zero baseline signal to obtain a corrected signal, and the bandwidth of the corrected signal is optimized and smoothed to obtain the final smooth signal. The final smooth signal was then compared with a standard bioelectric template to verify its morphological matching, resulting in a pure bioelectric signal.

9. A low-noise biopotential signal acquisition system, used to implement the low-noise biopotential signal acquisition and processing method according to any one of claims 1 to 8, characterized in that, include: The electrode acquisition module is used to acquire bioelectric potential signals from the human body through an electrode array to obtain raw bioelectric signals; A multi-stage adaptive filtering and amplification module is used to input the original bioelectric signal into a multi-stage adaptive filtering and amplification circuit to obtain an amplified and filtered analog signal. An analog-to-digital conversion module is used to perform analog-to-digital conversion on the amplified and filtered analog signal to obtain a digital bioelectric signal. The wavelet transform denoising module is used to perform adaptive wavelet transform denoising processing on the digitized bioelectric signal to obtain a primary denoised signal. The empirical mode decomposition module is used to perform empirical mode decomposition on the primary noise-reduced signal to obtain a multi-level intrinsic mode function set; The feature extraction module is used to extract the intrinsic mode functions of the feature frequency bands from the multi-level intrinsic mode function set to obtain the feature intrinsic mode function set; The signal reconstruction module is used to reconstruct the signal based on the set of characteristic intrinsic mode functions to obtain a pure bioelectric signal. The quality assessment and parameter optimization module is used to perform real-time quality assessment on the pure bioelectric signal to obtain signal quality indicators; based on the signal quality indicators, the system parameters are dynamically adjusted to optimize the bioelectric signal acquisition process and obtain a high signal-to-noise ratio bioelectric potential signal output.

Images