Ion chromatography peak boundary extraction method based on fitting residual and tangent judgment

Abstract

The application discloses an ion chromatography peak boundary extraction method based on fitting residual and tangent judgment, and belongs to the technical field of ion chromatography analysis. The method comprises the following steps: S1, reading original waveform data sequence, and performing down-sampling according to a set step length; S2, calculating the slope of the down-sampled data by using a difference method, and finding positive and negative slope sections by performing first-order derivation on the down-sampled waveform; S3, excluding noise interference, grouping the slope sections, then setting a chromatographic peak detection limit value, and screening out an effective peak left section and an effective peak right section; S4, combining the effective peak left section and the effective peak right section, and obtaining an effective peak and a preliminary left boundary point and a preliminary right boundary point of the effective peak; S5, positioning an error integral region by fitting a straight line through the preliminary left boundary point and the preliminary right boundary point; and S6, updating optimal boundary points by traversing evaluation indexes of all points in the error integral region. The method has more accurate boundary search range and higher calculation efficiency.

Description

Technical Field

[0001] This invention relates to the field of ion chromatography analysis technology, and in particular to a method for extracting ion chromatography peak boundaries based on fitting residuals and tangent judgment. Background Technology

[0002] In ion chromatography, peak identification and boundary extraction are crucial steps in quantitative analysis, and their accuracy directly affects the calculated concentration. Currently, common chromatographic peak detection and integration methods mainly include:

[0003] (1) First / Second Derivative Method: The first derivative method, also known as the slope method, uses the change in the first derivative of the signal (i.e., the slope) to determine the start and end points of the rising and falling regions, thereby detecting peaks and boundaries. The advantages of this method are its simplicity, speed, and suitability for real-time processing. The disadvantages are its sensitivity to noise, the need for smoothing processing, and the susceptibility to false detections. It is suitable for waveforms with regular peak shapes and high signal-to-noise ratios. In ion chromatography analysis systems, tailing peaks, asymmetric peaks, or overlapping peaks often appear. In ion chromatography detection, peaks are easily affected by noise, leading to boundary misjudgments, especially in trace analysis or under low signal-to-noise ratio conditions where the effect is poor. This method is suitable for symmetrical waveforms and waveforms with obvious structures.

[0004] (2) Smoothing Filtering Combined with Threshold Detection: Smoothing filtering is used to perform a moving average on the original signal (such as Savitzky-Golay filtering) to suppress noise. After smoothing filtering, this method has strong noise resistance. However, judging the peak shape of ion chromatography by threshold is not suitable for scenarios with severe noise, weak waveform signals, or narrow peaks, such as trace element detection, and peaks with multiple dilution concentrations in a set of peak data. In addition, this method is highly dependent on the peak threshold T. In continuous multi-component analysis of ion chromatography, the peak height difference is significant. Setting T too low will result in false detection, while setting it too high will result in missed detection. If continuous noise occurs, it will also be misjudged as a peak. Therefore, its applicability in the practical application of ion chromatographs is poor.

[0005] (3) Model fitting method, Gaussian polynomial function model is used to fit the chromatographic peak, and the boundary position is deduced. This method is theoretically complete and can effectively process part of the asymmetric peak, but in the ion chromatogram with obvious baseline drift or multi-peak overlap, the fitting convergence is difficult, the calculation amount is large, the time delay is high, and it is less used in the engineering application of real-time detection of wave peak, which is difficult to meet the real-time processing demand, and is usually used for offline data analysis. By using the "high proportion truncation" method, the left and right boundaries are deduced by the fitting curve parameters (the truncation threshold is generally set to 5% - 10% of the maximum value in the peak region), the advantages of this method are that it is theoretically strong and can calculate the area, FWHM and other characteristics, but the disadvantage is that it is large from the initial point, not suitable for dense regions with multiple peaks coexisting, especially multi-peak overlap in ion chromatograph. This method is suitable for high-resolution detection and quantitative analysis, such as IC-MS and LC-UV. In addition, the fitting calculation is complex and not suitable for wave peak detection of a large amount of data and real-time processing, and the boundary initial value point depends on the peak value, which can be used for chromatographic analysis and post-processing extraction of spectral characteristic peaks.

[0006] (4) Wavelet transform, this method is mainly used in medical imaging field, using multi-scale wavelet function to extract local features of signal in different scales, such as mutation point, slope change, wave peak, etc. When detecting the boundary, continuous wavelet transform (CWT) is used to transform the signal, and the zero-crossing point of wavelet coefficient is found in different scales (the zero point is the boundary candidate), and the reliable boundary is determined by repeating the zero-crossing point in multiple scales. Wavelet transform method can identify overlapping peaks and shoulder peaks. However, the wavelet transform algorithm is complex, which needs continuous traversal and loop calculation, and the wavelet function and scale range need to be selected manually, which has strong dependence on parameters and wave function. It is suitable for biomedical signals (EEG, ECG), and its practicability is greatly limited in ion chromatographic analysis field.

[0007] In addition, there are other peak boundary extraction methods, such as band-pass filtering and envelope detection, EM algorithm peak splitting, Z-score anomaly detection method, and adaptive baseline subtraction method, etc. These methods mainly focus on peak center positioning or peak fitting, and lack accurate strategy for boundary extraction, especially when the peak base is dragged, co-flowed or interfered by low signal-to-noise ratio, the traditional method often appears peak area missing calculation, negative peak area calculation, boundary extension deficiency or overwidth, etc., which affects the accuracy of quantitative analysis.

[0008] Ion chromatography often faces the following special challenges: baseline drift caused by changes in the pH of the mobile phase or the experimental environment; peak shape asymmetry caused by retention time drift; and weak peaks of low-concentration components that are easily obscured by adjacent large peaks. Existing methods often have boundary positioning deviations in the above scenarios, leading to distorted integral areas and affecting quantitative repeatability and accuracy. Therefore, it is urgent to develop a chromatographic peak boundary extraction method that takes into account noise resistance, adaptability, and computational efficiency, which can be embedded in an ion chromatograph data processing system to achieve accurate integration under complex conditions. SUMMARY

[0009] The present application aims to overcome the shortcomings of the prior art and provide an ion chromatography peak boundary extraction method based on fitting residual and tangent judgment.

[0010] The purpose of the present application is achieved by the following technical solution: an ion chromatography peak boundary extraction method based on fitting residual and tangent judgment, comprising the following steps:

[0011] S1: Collecting sample signals by an ion chromatograph to obtain an original ion flow data sequence, and performing down-sampling on the original data according to a preset sampling step to reduce the data volume and preserve the chromatographic peak shape characteristics to obtain chromatographic data;

[0012] S2: Calculating the slope of each data point using the difference method for the down-sampled chromatographic data, and identifying the positive slope section and the negative slope section by first-order differentiation of the signal;

[0013] S3: Combining the signal-to-noise ratio characteristics of the ion chromatograph, excluding interference slope sections caused by electrical noise or baseline drift, grouping the remaining slope sections, and filtering out the left side section of the effective chromatographic peak and the right side section of the effective chromatographic peak according to a preset peak height or area threshold;

[0014] S4: Combining and matching the filtered left side section of the effective chromatographic peak and the right side section of the effective chromatographic peak to preliminarily determine the vertex position of each chromatographic peak and its corresponding left boundary candidate point and right boundary candidate point;

[0015] S5: Constructing a linear fitting baseline based on the preliminarily determined left boundary candidate point and right boundary candidate point l , and then calculating the residual value between the chromatographic peak signal and the linear fitting baseline l , thereby locating the erroneous integration area that may have integration errors;

[0016] S6: Traversing all data points in the erroneous integration area and determining the optimal boundary point through a set evaluation index to complete the integration boundary determination of the chromatographic peak.

[0017] Preferably, the preset sampling step is 20.

[0018] Preferably, S5 further comprises the following steps:

[0019] connecting each group of preliminary boundary pairs (left boundary candidate point b_l , right boundary candidate point b_r ) to form a linear fitting baseline l , if the target chromatographic peak is asymmetric or tailing, the linear fitting baseline l intersects the chromatographic peak curve at three points b_l、 b_r and b_c ; wherein b_c is the alternating point of the normal integration region and the error integration region on the peak curve, if the left boundary candidate point b_l of the peak is higher than the right boundary candidate point b_r， , that is y b_l > y b_r , point b_c will appear at the end of the normal integration region and the beginning of the error integration region, and b_max is recorded as the left boundary candidate point b_l position; if the left boundary candidate point b_l of the peak is lower than the right boundary candidate point b_r， , that is y b_l< y b_r , point b_c will appear at the end of the error integration region and the beginning of the correct integration region, and b_max is recorded as the right boundary candidate point b_r position, that is, the point with the maximum measured concentration of the two endpoints is recorded as: ; point b_max satisfies the following conditions: ;

[0020] The linear fitting baseline l intersects the peak curve at at least two boundary points, and the first-order expression of the linear fitting baseline l is:

[0021] , wherein , represents the slope of the linear fitting baseline l ; ; f ( x ) is the vertical coordinate value of the boundary fitting straight line corresponding to the uniform horizontal coordinate point; ; is the horizontal coordinate of the left boundary candidate point , and the horizontal coordinate of the right boundary candidate point

[0022] The residual of each data point in a set of chromatographic peaks to : :

[0023] ; wherein p ( i ) is the ordinate value of the corresponding peak when the abscissa x takes i, i.e. the peak curve expression is p = p ( x ); f ( i ) is the ordinate value of the linearly fitted baseline l when the abscissa x takes i, passing through the boundary point;

[0024] All data points with residual < 0 are included in the candidate set, and the corresponding peak curve points are recorded in the boundary record sequence, to obtain the error integration region where the true boundary points of the peak exist.

[0025] Preferably, S6 further comprises the following steps:

[0026] The abscissa and ordinate of the left boundary point of the error integration region are defined as x e_l , y e_l , and the abscissa and ordinate of the right boundary point of the error integration region are defined as x e_r , y e_r ;

[0027] An evaluation index S t is defined for judging the optimal boundary value:

[0028] ; wherein j is the abscissa of a point in the error integration region, p ( j ) is the ordinate value of the point on the peak curve; is the abscissa value of the point with larger ordinate among the two points of the left and right boundaries of the peak; is the ordinate value of the point with larger ordinate among the two points of the left and right boundaries of the peak;

[0029] The evaluation index S t is used to traverse and calculate all points in the error integration region, and the minimum value of S t is updated to obtain the optimal boundary point b_end : ( x b_end , y b_end) ， This optimal boundary point b_end Occur in the error integration interval, that is, the abscissa of the optimal boundary point satisfies x b_end ∈( x e_l , x e_r ).

[0030] The beneficial effects of the present application are:

[0031] 1) More accurate boundary search range and higher calculation efficiency, the present application constructs a linear fitting baseline based on the rough boundary point, and uses residual analysis to accurately lock the edge area of the target peak, which greatly reduces the boundary search range from the full data sequence to the actual signal change interval. This strategy effectively reduces redundant calculations, while ensuring accuracy and significantly improving processing speed. Actual tests show that compared with traditional sliding window or derivative sign change method, the efficiency of the present application in ion chromatogram peak boundary positioning is improved by more than 30%, which is particularly suitable for real-time spectrum processing of million-level data points commonly used in ion chromatography, meeting the high-throughput detection demand.

[0032] 2) Strong anti-interference ability and robustness, for the baseline drift, noise disturbance and tailing phenomenon commonly seen in ion chromatogram, the present application analyzes the distribution characteristics of negative residual points below the fitting baseline, and selects the optimal boundary combining with geometric evaluation index, effectively suppressing the false judgment caused by local signal fluctuation. This method still maintains stable boundary recognition performance in low signal-to-noise ratio, weak peak or complex matrix sample, which is significantly better than traditional threshold or derivative method.

[0033] 3) Good adaptability to non-ideal peak shape, traditional methods mostly rely on symmetric peak shape assumption, such as Gaussian fitting, which often cannot accurately deal with shoulder peak, wide peak or irregular peak commonly seen in ion chromatography. The present application uses a linear fitting strategy without model distribution, which gets rid of the priori restriction on peak shape, and can effectively process non-ideal peak shapes such as front peak, tailing peak, shoulder peak and part of overlapping peak commonly seen in ion chromatography, with good universality and adaptability.

[0034] 4) Improve the positioning accuracy of boundary points and improve the integration accuracy. Accurate positioning of boundary points is the key to chromatographic integration accuracy, the present application optimizes boundary determination to make the integration area more consistent with the actual signal start and end range, thereby reducing the area calculation error. Experimental data show that compared with the first derivative method or fixed proportion truncation method, the present application can reduce the integration error by an average of 12%–18% in typical ion chromatography samples, and the effect is particularly significant in trace analysis.

[0035] 5) Meet the real-time processing requirements of the instrument, facilitate engineering deployment, the overall process of the method includes five parts of downsampling preprocessing, coarse boundary preliminary screening, linear fitting, residual judgment and geometric optimization, each part adopts a lightweight mathematical model (such as first-order least squares straight line, simple difference calculation, sorting screening), which has lower time complexity and memory occupation. In a typical chromatographic workstation or embedded industrial computer (such as Intel J4125 + 8G memory), the boundary extraction of a single chromatogram can be completed within 30 milliseconds, fully meeting the performance requirements of online detection, real-time integration and continuous analysis of ion chromatograph, and having good engineering application prospect. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 is a flow chart of the method of the present application;

[0037] Figure 2 is an error integration area schematic diagram in an embodiment;

[0038] Figure 3 is an error integration area schematic diagram in another embodiment;

[0039] Figure 4 is a peak boundary point effect schematic diagram of detection before and after optimization in an embodiment;

[0040] Figure 5 is a peak boundary point effect schematic diagram of detection before and after optimization in another embodiment. DETAILED DESCRIPTION

[0041] The technical solutions of the present application will be described in detail below with reference to the embodiments, obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor belong to the scope of protection of the present application.

[0042] Reference Figures 1-5 , the present application provides a technical solution: ion chromatographic peak boundary extraction method based on fitting residual and tangent judgment, comprising the following steps:

[0043] S1: acquire sample signal through ion chromatograph, obtain original ion flow data sequence, and perform downsampling on original data according to preset sampling step to reduce data amount and keep chromatographic peak shape feature to obtain chromatographic data;

[0044] S2: calculate the slope of each data point using difference method on the chromatographic data after downsampling, and identify the positive slope section and negative slope section by first-order difference of the signal;

[0045] S3: Excluding the interference slope section caused by electrical noise or baseline drift according to the signal-to-noise ratio characteristics of the ion chromatograph, grouping the remaining slope sections, and screening the left side section of the effective chromatographic peak and the right side section of the effective chromatographic peak according to a preset peak height or area threshold value;

[0046] S4: Combining and matching the screened left side section of the effective chromatographic peak and the right side section of the effective chromatographic peak, and preliminarily determining the vertex position of each chromatographic wave peak and the corresponding left boundary candidate point and right boundary candidate point;

[0047] S5: Constructing a linear fitting baseline based on the preliminarily determined left boundary candidate point and right boundary candidate point l , and then calculating the residual value between the chromatographic wave peak signal and the linear fitting baseline l , so as to locate the error integration area where the integral error may exist;

[0048] S6: Traversing all data points in the error integration area, determining the optimal boundary point through a set evaluation index, and completing the integration boundary determination of the chromatographic wave peak.

[0049] In this embodiment, the original waveform data sequence is down-sampled to reduce the data processing amount. The number of original data points is large (usually several hundred thousand for a complete group of waveform data), and the waveform of the curve is obtained by sampling at an interval of Space = 20. The calculation complexity is reduced, which is used for subsequent rapid judgment of slope change area; the slope is calculated by using the difference method, and the positive and negative slope sections are found by completing the first-order derivative (K value) on the down-sampled waveform. The continuous positive slope region is the left side of the wave peak; and the continuous negative slope region is the right side of the wave peak.

[0050] Excluding short noise interference, finding effective wave peaks: grouping the slope region, and then setting a wave peak detection limit det. If the first-order derivative of the wave peak curve is greater than 0 for det consecutive points, it is considered that the left side of the wave peak is continuous; and if the first-order derivative of the wave peak curve is less than 0 for det consecutive points, it is considered that the right side of the wave peak is continuous.

[0051] Combining the left side section of the effective chromatographic peak and the right side section of the effective chromatographic peak, considering it as a wave peak region, and defining the left boundary point with >0 as the left boundary candidate point, <0 as the right boundary candidate point, so as to form the left boundary candidate point and the right boundary candidate point of the complete effective wave peak. Figure 2 and Figure 3 There are multiple groups of wave peaks in and it can be seen that the first-order and second-order derivative positioning peak method is not suitable for positioning all peak shapes. For left-right asymmetric wave peaks, the one-stage wave peak boundary found by the continuous change of the derivative will have a part of the peak area with negative value, which makes the integral of this part of the peak area negative, and finally leads to the error detection of the instrument.​​

[0052] The detection boundary accuracy of the application is significantly improved, the signal swing or tailing can be automatically avoided, the "signal end point" instead of the "threshold position" can be captured, the peak area calculation accuracy is improved, the more real integral area is obtained through more accurate searching of the peak boundary, and the detection accuracy of the automatic instrument is improved; the boundary positioning of various irregular peaks and weak signal peaks, such as shoulder peaks and tailing peaks, can be applied, and the algorithm of the application can still identify reasonable boundaries when the existing method has certain limitations in processing this type of peaks and cannot be universally applied to instrument algorithms; the algorithm is simple and has high execution efficiency, can process million-level big data waveforms on a low-configuration computer, and has good real-time performance; it is easy to reproduce, convenient for engineering landing, can be directly used in automatic instruments, intelligently judges the peak and integral, and does not need manual boundary intervention and manual integration.

[0053] In some embodiments, the preset sampling step is 20, which is suitable for the sampling frequency and peak width characteristics of typical ion chromatography signals.

[0054] In some embodiments, S5 further comprises the following steps:

[0055] connecting each group of preliminary boundary pairs (left boundary candidate point b_l , right boundary candidate point b_r ) to form a linear fitting baseline l , if the target chromatographic peak is asymmetric or tailing, the linear fitting baseline l intersects the chromatographic peak curve at three points b_l、 b_r and b_c ; wherein b_c is the alternating point of the normal integral region and the error integral region on the peak curve, if the left boundary candidate point b_l of the peak is higher than the right boundary candidate point b_r， , that is y b_l > y b_r , point b_c will appear at the position where the normal integral region ends and the error integral region starts, and b_max is recorded as the left boundary candidate point b_l position; if the left boundary candidate point b_l of the peak is lower than the right boundary candidate point b_r， , that is y b_l< y b_r , point b_c will appear at the position where the error integral region ends and the correct integral region starts, and b_max is recorded as the right boundary candidate point b_rPosition, that is, the point with the maximum measured concentration in the two end points is recorded as: ; point b_max satisfies the following conditions: ;

[0056] Linearly fitting the baseline l Intersecting the peak curve at at least two boundary points, linearly fitting the baseline l The first-order expression of the baseline is:

[0057] , wherein represents the slope of the linearly fitted baseline l ; ; f ( x ) is the ordinate value of the boundary fitting straight line corresponding to the uniform abscissa point; ; is the abscissa of the left boundary candidate point, is the abscissa of the right boundary candidate point;

[0058] The residual of each data point in a group of chromatographic peak to is calculated:

[0059] ; wherein p ( i ) is the ordinate value of the corresponding peak when the abscissa x is i, that is, the peak curve expression is p = p ( x ); f ( i ) is the ordinate value of the linearly fitted baseline l passing through the boundary point when the abscissa x is i;

[0060] All data points with residual <0 are included in the candidate set, and the corresponding peak curve points are recorded in the boundary record sequence to obtain the error integration area where the real boundary points of the peak exist.

[0061] In this embodiment, the error integration area is Figure 2 the peak shape area of b_l 1 to b_c 1 , b_c 2 to b_r 2 , Figure 3 the peak shape area of b_l 1 to b_c 1 , b_c2 to b_r 2 section and b_c 3 to b_r 3 section, which is the interval where the true boundary points exist on the side of the peak.

[0062] In some embodiments, the S6 further comprises the following steps:

[0063] define the horizontal and vertical coordinates of the left boundary point of the error integration region as (xL, yL) x e_l , y e_l define the horizontal and vertical coordinates of the right boundary point of the error integration region as (xR, yR) x e_r , y e_r ;

[0064] define the evaluation index S t for determining the optimal boundary value:

[0065] ; wherein j x is the horizontal coordinate of a point in the error integration interval, p (y) is the vertical coordinate value of the point on the peak curve; j xL is the horizontal coordinate value of the point with the larger vertical coordinate among the two points on the left and right boundaries of the peak; yL is the vertical coordinate value of the point with the larger vertical coordinate among the two points on the left and right boundaries of the peak;

[0066] through the evaluation index S t iterate through all points in the error integration region and update the minimum value of S t to obtain the optimal boundary point b_end : x b_end , y b_end ) ， This optimal boundary point b_end occurs within the error integration interval, i.e., the horizontal coordinate of the optimal boundary point satisfies x b_end ∈( x e_l , x e_r ).

[0067] In this embodiment, the optimal boundary is updated by tangent optimization, and the conclusions derived from the above and Figure 2 , Figure 3 ​It is not difficult to see that the one-stage peak boundary pair obtained by using the derivative method will only have the wrong integration interval of the high and low peaks at the low end of the peak boundary value.

[0068] The above description is merely that of preferred embodiments of the application, and it is understood that the application is not limited in its application to the details set forth in the description above. Rather, the application is capable of various modifications and alternative constructions from that which is explicitly shown and described above. It is understood that all such modifications and changes are intended to fall within the true spirit and scope of the application. Accordingly, the particular embodiments of the present application are not to be taken in a limiting sense, as the scope of the present application is defined by the appended claims.

Claims

1. A method for extracting the peak boundary of ion chromatogram based on fitting residual and tangent judgment, characterized in that: The method comprises the following steps: S1: Collecting a sample signal by an ion chromatograph, obtaining a raw ion flow data sequence, and performing down-sampling on the raw data according to a preset sampling step length to reduce the data amount and keep the chromatographic peak shape characteristics to obtain chromatographic data; S2: Calculating the slope of each data point by using a difference method on the chromatographic data after down-sampling, and identifying the positive slope section and the negative slope section by performing first-order difference on the signal; S3: Excluding the interference slope section caused by electrical noise or baseline drift in combination with the signal-to-noise ratio characteristics of the ion chromatograph, grouping the remaining slope sections, and screening the left side section of the effective chromatographic peak and the right side section of the effective chromatographic peak according to a preset peak height or area threshold; S4: Combining and matching the screened left side section of the effective chromatographic peak and the right side section of the effective chromatographic peak, and preliminarily determining the vertex position of each chromatographic wave peak and the corresponding left boundary candidate point and right boundary candidate point; S5: Construct a linear fitting baseline based on the initially determined left and right boundary candidate points. l Then, the chromatographic peak signal and the linear fitting baseline were calculated. l The residual values ​​between them are used to locate the erroneous integration region where integration error may exist; S6: Traversing all data points in the erroneous integration area, determining the optimal boundary point through a set evaluation index, and completing the integration boundary determination of the chromatographic wave peak.

2. The ion chromatography peak boundary extraction method based on fitting residual and tangent judgment according to claim 1, characterized in that: The preset sampling step length is 20.

3. The method for ion chromatography peak boundary extraction based on fitting residual and tangent judgment according to claim 1, characterized in that: The S5 further comprises the following steps: connecting each set of preliminary boundary pairs (left boundary candidate point b_l , right boundary candidate point b_r ) to form a linear fitting baseline l , if the target chromatographic peak is asymmetric or tailing, the linear fitting baseline l intersects the chromatographic peak curve at three points b_l, b_r and b_c ; wherein b_c is the alternating point of the normal integration region and the error integration region on the peak curve, if the left boundary candidate point b_l of the peak is higher than the right boundary candidate point b_r， , that is y b_l > y b_r , point b_c will appear at the position where the normal integration region ends and the error integration region begins, and b_max is recorded as the left boundary candidate point b_l position; if the left boundary candidate point b_l of the peak is lower than the right boundary candidate point b_r， , that is y b_l < y b_r , point b_c will appear at the position where the error integration region ends and the correct integration region begins, and b_max is recorded as the right boundary candidate point b_r position, that is, the point with the maximum measured concentration of the two endpoints is recorded as: ; point b_max satisfies the following condition: ; linearly fitted baseline l linearly fitted baseline l the first order expression of which is , wherein represents the slope of the linear fit baseline l ; ; f ( y ) is the ordinate value of the boundary fitting straight line corresponding to the uniform abscissa point; x ; ; is the abscissa of the left boundary candidate point, is the abscissa of the right boundary candidate point; The residuals of each data point in a set of chromatographic peaks to the mean : ;in p ( i Let x be the horizontal coordinate of the peak when x takes the value i. The expression for the peak curve is: p = p ( x ); f ( i (x) represents the linear fitting baseline that crosses the boundary points when the x-coordinate is i. l The ordinate value; All residuals Data points with <0 are included in the candidate set, and the corresponding peak curve points are recorded in the boundary record sequence to obtain the error integral area where the true boundary points of the peak exist.

4. The method for ion chromatography peak boundary extraction based on fitting residual and tangent judgment according to claim 1, characterized in that: The S6 further comprises the following steps: Define the x and y coordinates of the left boundary point of the erroneous integration region as ( x e_l , y e_l The x and y coordinates of the right boundary point of the erroneous integration region are ( x e_r , y e_r ); Definition of evaluation index S t For determining the optimal boundary value: ; wherein j is the abscissa of a point in the error integration interval, p j is the ordinate value of this point on the peak curve; is the abscissa of the point with the greater ordinate value of the two points on the left and right boundaries of the peak; is the ordinate value of the point with the greater ordinate value of the two points on the left and right boundaries of the peak;​ By evaluating the index S t Traverse all points in the error integral region and update S t The minimum value of the optimal boundary point b_end x b_end , y b_end ， This optimal boundary point b_end occurs in the error integral region, that is, the abscissa of the optimal boundary point satisfies x b_end ∈( x e_l , x e_r ).​​

Images