CN-122019959-A - Efficient electrochemical signal data processing method and system

CN122019959ACN 122019959 ACN122019959 ACN 122019959ACN-122019959-A

Abstract

The invention provides a high-efficiency electrochemical signal data processing method, which effectively inhibits noise interference by reasonably selecting wavelet basis functions and decomposition layer numbers, adopting methods such as threshold processing and the like, improves the processing quality of electrochemical signals and realizes accurate identification of real peaks in the electrochemical signals according to the steps such as acquisition, pretreatment, wavelet decomposition, signal denoising, peak characteristic extraction, signal reconstruction and the like, and meanwhile, the invention also develops and designs an embedded system according to the method.

Inventors

ZHANG JIFEI
GAO XIN
AN YINGYING
ZHANG QING
ZHAO MENGDI
YANG XIAOBING
ZHENG RULAN

Assignees

中国海关科学技术研究中心
中国质量检验检测科学研究院

Dates

Publication Date: 20260512
Application Date: 20251105

Claims (10)

1. A high-efficiency electrochemical signal data processing method comprises the following steps: s1) electrochemical signal acquisition Detecting a solution containing a substance with a specific concentration by adopting a cyclic voltammetry, and acquiring cyclic voltammetry signals through a three-electrode system consisting of a working electrode, a reference electrode and an auxiliary electrode; S2) the cyclic voltammetric signal pretreatment Preprocessing the cyclic voltammetry signal by adopting a mean value removing method, namely subtracting a mean value from the acquired cyclic voltammetry signal data value to obtain a zero-mean value signal; S3) the zero-mean signal wavelet decomposition And performing third-order wavelet decomposition on the zero-mean signal by adopting a Mallat algorithm based on a Daubechies wavelet basis function, wherein the structural formula of the decomposed signal is as follows: x(n)=A3+D1+D2+D3 wherein: A3 is a low-frequency approximate signal (overall trend of the signal), the scale is 6, and the resolution is the lowest; d1 is a high frequency detail signal (noise and sharp peaks), scale 2, resolution highest; D2 is a medium-high frequency detail signal (core effective feature), scale 4, resolution ratio is moderate; d3 is the mid-low frequency detail signal (between baseline and peak); S4) signal denoising Noise suppression is performed based on thresholding of wavelet coefficients, specifically, a wavelet threshold is determined first, then the wavelet threshold is utilized to remove high frequency noise, and thresholding is applied to detail coefficients, wherein the thresholding is formulated as follows: wherein: D' j [ k ], processing the j-th layer wavelet detail coefficient (k value), denoising and outputting; d j [ k ] is the detail coefficient (kth value) of the j th layer wavelet before processing, and is input after decomposition; a wavelet threshold value, which is a critical value for distinguishing noise from signals; sigma, noise standard deviation, reflecting noise intensity; N, the total sampling points of the original signals are counted, and the threshold value is ensured to be matched with the data quantity; S5) extracting signal peak characteristics The peak value is expressed as an extreme point of a coefficient in the wavelet domain, corresponds to a local maximum point, and further screens the position and the amplitude of the peak value in the denoised coefficient; S6) Signal reconstruction Performing inverse wavelet transformation on the wavelet coefficient subjected to noise processing and peak feature extraction processing, and reconstructing a processed electrochemical signal by using a third-order reconstruction algorithm, wherein the reconstruction formula of the electrochemical signal is as follows: wherein: a j (k) low frequency approximation coefficients of the j-th scale after reconstruction (final output signal core); a j-1 (k) low frequency approximation coefficients of the j-1 th scale (input fine scale trend); d j-1 (k) high frequency approximation coefficients of the j-1 th scale (details of the fine scale after denoising); h j (k-2 n) reconstructing the filter coefficients at low frequency to recover the low frequency components; g j (k-2 n) high-frequency reconstruction filter coefficients, recovering high-frequency components; 2n, up-sampling characteristic of the binary wavelet, and reconstructed inverse down-sampling operation; and k-2n, namely, convolution position, and representing the sliding weighting relation between the filter and the extension signal.
2. The method for processing electrochemical signal data with high efficiency according to claim 1, wherein the calculation formula of the mean value removing method in the step S2 is as follows, x′(t)=x(t)-μ Wherein: x' (t) zero-averaged signal (target output); x (t) original signal (input); μ: the mean value of the original signal x (t) (reflecting baseline drift level); N is the number of sampling points (total length of data) of the original signal x (t); x (i) is the i-th sample value (discretized data point) of the original signal.
3. The method for efficient electrochemical signal data processing according to claim 1, wherein the Mallat algorithm in step S3 is formulated as follows, Wherein: a j (k) is a j-th scale low-frequency approximation coefficient, and input coarse-scale low-frequency data; a j-1 (k),d j-1 (k) is the near low frequency similar coefficient and the high frequency detail coefficient of the j-1 scale respectively; h j (k-2n),g j (k-2 n) are low-pass filter coefficients and high-pass filter coefficients, respectively.
4. The method for processing electrochemical signal data with high efficiency according to claim 1, wherein the algorithm of the third order wavelet decomposition in the step S3 is as follows, Wherein: a j . J low frequency approximation coefficients of the scale; a j-1 ,a j-2 ,a j-3 th, j-1, j-2, j-3 low frequency approximation coefficients; d j-1 ,d j-2 ,d j-3 , high-frequency detail coefficients of all scales; g', high-pass filter coefficient, is used for extracting the high-frequency detail component of the signal; h', low-pass filter coefficients are used for extracting low-frequency approximate components of the signals; And (5) double downsampling.
5. The method for efficient electrochemical signal data processing according to claim 1, wherein the peak point in step S5 is found by satisfying the following formula, d j (k)＞d j-1 (k-1),d j (k)＜d j+1 (k-1) Wherein: D j (k), under the j-th scale (level), the wavelet detail coefficient with index k is obtained, and the core judgment object usually takes j to correspond to the D2 layer (middle-high frequency detail); d j-1 (k-1) the wavelet detail coefficient with index k-1 at the j-1 scale (1 level coarser than the current scale j) ensures that the peak has relative saliency at the coarse scale. D j+1 (k-1) the wavelet detail coefficient with index of k-1 under the j+1th scale (1 level finer than the current scale j) ensures that the peak value accords with the peak value transmission rule under the fine scale.
6. The method for processing electrochemical signal data with high efficiency as set forth in claim 1, wherein the third-order reconstruction algorithm in the step S6 is formulated as follows, Wherein: a j-3 . J-3 th scale low frequency approximation coefficients; d j-1 ,d j-2 ,d j-3 , high-frequency detail coefficients of all scales; a j , reconstructing a coarse-scale low-frequency approximation coefficient; g k , a high-pass reconstruction filter for extracting high-frequency detail components of the signal; h k , a low-pass reconstruction filter, which is used for extracting low-frequency approximate components of the signals; and double up-sampling.
7. A high-efficiency electrochemical signal data processing system designed to implement the method of any one of claims 1 to 6.
8. The system of claim 7, wherein the system comprises a signal acquisition module, a signal processing module and a signal output module.
9. The efficient electrochemical signal data processing system of claim 8, wherein the signal processing module comprises an electrochemical signal preprocessing module, a wavelet decomposition module, a thresholding module, and a wavelet reconstruction module.
10. The efficient electrochemical signal data processing system of claim 9, wherein the efficient electrochemical signal data processing system is an embedded system that is further integrated as a miniaturized portable device.

Description

Efficient electrochemical signal data processing method and system Technical Field The present invention relates to a method and a system for processing data, and more particularly, to a method and a system for processing electrochemical signal data. Background The electrochemical signal has the processing difficulties of low signal-to-noise ratio and signal baseline drift, and the measurement accuracy of the electrochemical signal is affected, while the traditional electrochemical signal processing methods such as a sliding average filtering method and a Gaussian fitting method have the defects of poor noise resistance, being affected by the baseline drift and the like, and the singular point of the electrochemical signal cannot be identified. In addition, wavelet analysis is a powerful analysis tool after a Fourier analysis method, has been widely applied to the fields of signal processing, pattern recognition and the like, and compared with traditional wavelet transformation, the resolution analysis characteristic of the rapid wavelet transformation (Mallat algorithm) based on discrete signals is very effective for the analysis of electrochemical non-stationary signals, and can realize noise filtering and peak potential detection of electrochemical signals. Aiming at the problems, the invention combines electrochemical signal characteristics and wavelet analysis, and provides a new way for realizing the purposes of automation, rapidness, accuracy and the like of electrochemical analysis and detection. Disclosure of Invention The invention aims to provide a high-efficiency electrochemical signal data processing method, which mainly comprises the following steps: s1) electrochemical signal acquisition And detecting the solution containing the substance with the specific concentration by adopting a cyclic voltammetry, and acquiring cyclic voltammetry signals through a three-electrode system consisting of a working electrode, a reference electrode and an auxiliary electrode. S2) the cyclic voltammetric signal pretreatment And preprocessing the cyclic voltammetry signal by adopting a mean value removing method, namely subtracting a mean value from the acquired cyclic voltammetry signal data value to obtain a zero-mean value signal. S3) the zero-mean signal wavelet decomposition And performing third-order wavelet decomposition on the zero-mean signal by adopting a Mallat algorithm based on a Daubechies wavelet basis function, wherein the structural formula of the decomposed signal is as follows: x(n)=A3+D1+D2+D3 wherein: A3 is a low-frequency approximate signal (overall trend of the signal), the scale is 6, and the resolution is the lowest; d1 is a high frequency detail signal (noise and sharp peaks), scale 2, resolution highest; D2 is a medium-high frequency detail signal (core effective feature), scale 4, resolution ratio is moderate; d3 is the mid-low frequency detail signal (between baseline and peak). S4) signal denoising Noise suppression is performed based on thresholding of wavelet coefficients, specifically, a wavelet threshold is determined first, then the wavelet threshold is utilized to remove high frequency noise, and thresholding is applied to detail coefficients, wherein the thresholding is formulated as follows: wherein: D' j [ k ], processing the j-th layer wavelet detail coefficient (k value), denoising and outputting; d j [ k ] is the detail coefficient (kth value) of the j th layer wavelet before processing, and is input after decomposition; a wavelet threshold value, which is a critical value for distinguishing noise from signals; sigma, noise standard deviation, reflecting noise intensity; and N, the total sampling point number of the original signal, and ensuring that the threshold value is matched with the data quantity. S5) extracting signal peak characteristics The peak value is expressed as an extreme point of the coefficient in the wavelet domain, corresponds to a local maximum point, and further screens the peak value position and the amplitude in the denoised coefficient. S6) Signal reconstruction Performing inverse wavelet transformation on the wavelet coefficient subjected to noise processing and peak feature extraction processing, and reconstructing a processed electrochemical signal by using a third-order reconstruction algorithm, wherein the reconstruction formula of the electrochemical signal is as follows: wherein: a j (k) low frequency approximation coefficients of the j-th scale after reconstruction (final output signal core); a j-1 (k) low frequency approximation coefficients of the j-1 th scale (input fine scale trend); d j-1 (k) high frequency approximation coefficients of the j-1 th scale (details of the fine scale after denoising); h j (k-2 n) reconstructing the filter coefficients at low frequency to recover the low frequency components; g j (k-2 n) high-frequency reconstruction filter coefficients, recovering high-frequency components; 2n, up-sampling characteristi