Background signal extraction method and device

Abstract

The present invention provides a method and device for extracting background signals, which are applied to the field of high-precision measurement. The method includes: obtaining multiple groups of measured signal data of various samples to be measured, where each group of the measured signal data is collected based on the same measurement condition; respectively performing parametric equation fitting on each group of the measured signal data to obtain the parametric numerical solutions of each group of the measured signal data, and the parametric numerical solutions include the height information, width information, position information, and shape information of each peak; determining the peaks of the background signal originating from the measurement condition according to the parametric numerical solutions of each group; and obtaining the background signal of the measurement condition based on the determined peaks of the background signal originating from the measurement condition. The present invention improves the accuracy of background signal extraction.

Description

Technical Field

[0001] The present invention relates to the field of high-precision measurement, and particularly to a method and device for extracting background signals. Background Art

[0002] Signal analysis is an important means in modern scientific research and industrial production, and is widely used in fields such as chemical analysis, materials science, environmental monitoring, etc. The measured signal plays a crucial role in signal analysis. As the basic data for analysis, it usually covers three major components: the true signal of the sample, background error, and random error. The true signal of the sample reflects the unique signal characteristics of the substance and is the main target of signal analysis. However, in the actual measurement process, the introduction of background error and random error will affect the accuracy and reliability of the signal data, and also affect the extraction and fitting of the true signal.

[0003] In the prior art, although the important impact of background error on the accuracy of signal analysis has been recognized, in actual operation, there are still many challenges in extracting the signal distribution pattern of background error. Traditional processing methods often have difficulty in accurately distinguishing the true signal from the background error, resulting in the loss of some true signal information while eliminating the error, thus reducing the accuracy and reliability of signal analysis. Therefore, there is an urgent need and important significance to develop a method that can accurately extract the signal distribution pattern of background error for improving the overall accuracy and reliability of signal analysis. Summary of the Invention

[0004] In view of this, on the one hand, the present invention provides a method for extracting background signals, including:

[0005] Obtaining multiple groups of measured signal data of multiple samples to be measured, and each group of the measured signal data is collected based on the same measurement condition;

[0006] Performing parametric equation fitting on each group of measured signal data respectively to obtain the parametric numerical solutions of each group of measured signal data, and the parametric numerical solutions include the height information, width information, position information, and shape information of each peak;

[0007] Determining the peaks of the background signal originating from the measurement condition according to the parametric numerical solutions of each group;

[0008] Obtaining the background signal of the measurement condition based on the determined peaks of the background signal originating from the measurement condition.

[0009] Optionally, the determining the peaks of the background signal originating from the measurement condition according to the parametric numerical solutions of each group includes:

[0010] For each set of measured signal data, determine whether any peak in the measured signal data exists as a similar peak in any of the other measured signal data;

[0011] If similar peaks exist in any of the other measured signal data, determine that the peak is a peak of the background signal resulting from the measurement conditions.

[0012] Optionally, determining the peak of the background signal resulting from the measurement conditions according to the numerical solutions of the parameters for each group includes:

[0013] Perform clustering analysis on the numerical solutions of the parameters of each peak to determine whether at least one set of peaks with similar parameters is obtained;

[0014] If no set of peaks with similar parameters is obtained, the background signal is not recognized;

[0015] If at least one set of peaks with similar parameters is obtained, each peak in each set is a peak of the background signal resulting from the measurement conditions.

[0016] Optionally, obtaining the background signal of the measurement conditions based on the determined peak of the background signal resulting from the measurement conditions includes:

[0017] Select any one of the sets of measured signal data;

[0018] Determine the peak of the background signal resulting from the measurement conditions in the selected measured signal data, and obtain the background signal of the measurement conditions by using the corresponding numerical solution of the parameters.

[0019] Optionally, obtaining the background signal of the measurement conditions based on the determined peak of the background signal resulting from the measurement conditions includes:

[0020] Based on the peak of the background signal resulting from the measurement conditions, extract the background signal from each of the sets of measured signal data;

[0021] Perform a comprehensive operation on the multiple sets of background signals to obtain the background signal of the measurement conditions, and the comprehensive operation includes taking the average or median.

[0022] Optionally, obtaining the background signal of the measurement conditions based on the determined peak of the background signal resulting from the measurement conditions includes:

[0023] Calculate the clustering center for at least one frame of the peak of the background signal resulting from the measurement conditions to obtain the background signal of the measurement conditions.

[0024] The second aspect of the present invention also provides a method for extracting a background signal, including:

[0025] Obtain at least one set of measured signal data without placing the sample to be measured, and each set of the measured signal data is collected based on the same measurement condition;

[0026] Perform parametric equation fitting on each set of measured signal data respectively to obtain the parametric numerical solutions of each set of measured signal data, and the parametric numerical solutions include the height information, width information, position information and shape information of each peak;

[0027] Based on the parametric numerical solutions of each set of measured signal data, obtain the background signal of the measurement condition.

[0028] Optionally, the obtaining the background signal of the measurement condition based on the parametric numerical solutions of each set of measured signal data includes:

[0029] Select any one of the measured signal data from multiple sets of measured signal data;

[0030] Use the parametric numerical solutions of the selected measured signal data to obtain the background signal of the measurement condition.

[0031] Optionally, the obtaining the background signal of the measurement condition based on the parametric numerical solutions of each set of measured signal data includes:

[0032] Use the parametric numerical solutions of multiple sets of measured signal data to obtain the corresponding background signal of the measurement condition;

[0033] Perform a comprehensive operation on multiple sets of background signals to obtain the background signal of the measurement condition, and the comprehensive operation includes taking the average or median.

[0034] Optionally, the obtaining the background signal of the measurement condition based on the parametric numerical solutions of each set of measured signal data includes:

[0035] Calculate the clustering center of the parametric numerical solutions of multiple sets of measured signal data to obtain the background signal of the measurement condition.

[0036] Optionally, performing parametric equation fitting on each set of measured signal data respectively includes:

[0037] Obtain a mathematical model adapted to the measured signal data, and the mathematical model includes a multi-pseudo Voigt model or a mathematical model combining multi-pseudo Voigt and a straight line;

[0038] Use the mathematical model and the measured signal data to obtain the fitting equation of the measured signal data.

[0039] Optionally, if the mathematical model adapted to the measured signal data is a combined mathematical model of multiple pseudo Voigt functions and a straight line, then a straight line segment and the first peak segment and the second peak segment closest to both sides of the straight line segment are intercepted from the measured signal data;

[0040] The first peak segment and the second peak segment are respectively subjected to single pseudo Voigt peak fitting to obtain a first pseudo Voigt function and a second pseudo Voigt function;

[0041] The vertices of the first pseudo Voigt function and the second pseudo Voigt function are obtained, and a line is drawn between the two vertices;

[0042] A sum function is calculated according to the first pseudo Voigt function and the second pseudo Voigt function;

[0043] A line is drawn again using the coordinates of multiple intersection points of the line and the sum function to obtain a target straight line, and it is determined whether the target straight line is the tangent of the peak point of the sum function;

[0044] If the target straight line is the tangent of the peak point of the sum function, two peak segments of the sum function are intercepted using the target abscissas of multiple intersection points, and the target abscissas are obtained by averaging the abscissas of multiple intersection points;

[0045] A fitting equation of the background signal is calculated according to the intercepted data and the target straight line.

[0046] Optionally, drawing a line again using the coordinates of multiple intersection points of the line and the sum function to obtain a target straight line, and determining whether the target straight line is the tangent of the peak point of the sum function includes:

[0047] Multiple target intersection points are obtained according to the target abscissas and their corresponding ordinates;

[0048] A line is drawn through multiple target intersection points to obtain a target straight line;

[0049] The error between the ordinate of the target straight line and the sum function at a certain abscissa is calculated and an error judgment is made. If the error is less than a preset error, the target straight line is regarded as the tangent of the peak point of the sum function;

[0050] If the error is greater than the preset error, the connection is continued to be iterated according to the intersection points of the target straight line and the sum function until the error between the ordinate of the iterated target straight line and the sum function at a certain abscissa is less than the preset error, and the target straight line is regarded as the tangent of the peak point of the sum function.

[0051] The third aspect of the present invention provides a signal processing method, including:

[0052] Obtaining a background signal according to the background signal extraction method described in any one of the above;

[0053] Subtracting the background signal from the measured signal corresponding to the sample to be measured collected to obtain the true signal of the sample to be measured.

[0054] The fourth aspect of the present invention provides a background signal extraction device, which includes: a processor and a memory connected to the processor; wherein, the memory stores instructions executable by the processor, and the instructions are executed by the processor to enable the processor to execute the above background signal extraction method.

[0055] The fifth aspect of the present invention provides a signal processing device, which includes: a processor and a memory connected to the processor; wherein, the memory stores instructions executable by the processor, and the instructions are executed by the processor to enable the processor to execute the above signal processing method.

[0056] The present invention obtains multiple groups of measured signal data of different samples to be measured under the same measurement conditions in the data acquisition stage, ensuring that the data is both diverse and representative, providing a rich and reliable data basis for subsequent analysis. Since the measured signal data includes the true signal of the sample, background error, and random error, it is necessary to extract the background signal subsequently. In the parametric equation fitting step, by fitting each group of measured signal data, a numerical solution of parameters including the height, width, position, and shape information of each peak is obtained, which can express the signal characteristics in detail. Then, based on the numerical solution of the parameters, the peak information of the background signal corresponding to the error that does not change with the sample is found, and the characteristics of the background error can be accurately located. Finally, the background signal is obtained according to the peak information of the background signal, realizing the accuracy of the background error extraction, creating conditions for subsequent elimination of the background error and more accurate identification and processing of the true signal of the sample, thereby significantly improving the accuracy and reliability of signal analysis.

[0057] The present invention directly obtains background signal data without placing a sample, collects at least one set of signal data under the same measurement conditions, then performs parametric equation fitting, and obtains the background signal according to the solution of the obtained parametric equation. Directly obtaining signal data without placing a sample can exclude the interference of the sample signal on the background signal. The obtained signal data purely reflects the background information generated by the measurement conditions. Moreover, directly obtaining the signal data without placing a sample, and only performing parametric equation fitting on these data can obtain the background signal, which greatly simplifies the extraction process of the background signal and improves work efficiency. In addition, obtaining at least one set of measured signal data ensures that the background signal can be effectively obtained even with limited data. When obtaining multiple sets of measured signal data, the obtained background signal can be made smoother. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0059] Figure 1 It is a flowchart of a method for extracting background signals in an embodiment of the present invention;

[0060] Figure 2 It is a spectrogram obtained after performing parametric equation fitting on each set of measured spectral data in an embodiment of the present invention;

[0061] Figure 3 It is a peak center extraction diagram in an embodiment of the present invention;

[0062] Figure 4 It is a clustering set diagram in an embodiment of the present invention;

[0063] Figure 5 It is a three-pseudo Voigt model diagram in an embodiment of the present invention;

[0064] Figure 6 It is a measured signal diagram of a double-pseudo Voigt peak plus a linear equation in an embodiment of the present invention;

[0065] Figure 7 It is a flowchart of a signal processing method in an embodiment of the present invention;

[0066] Figure 8 It is a flowchart of another method for extracting background signals in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0067] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0068] In addition, the technical features involved in different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other. Embodiment 1

[0069] As Figure 1 shown, the embodiment of the present invention provides a background signal extraction method. This method extracts the background signal with the help of samples, and this method is executed by an electronic device such as a computer or a server. Specifically, it includes:

[0070] S11, obtaining multiple groups of measured signal data of multiple samples to be measured, and each group of measured signal data is collected based on the same measurement condition.

[0071] The sample to be measured can be any substance, and the signal can be any one-dimensional, single-peak or multi-peak data such as spectrum, chromatography, energy spectrum, nuclear magnetic resonance, etc. The measurement device is used to collect signals from the samples to be measured placed on the measurement table. Specifically, different samples are placed under the same measurement condition, and multiple groups of measured signal data are collected. For example, when the signal is a spectrum, the measurement device is a spectrometer that can collect spectra. This measurement condition covers all customizable input parameters such as specific light source intensity, wavelength range, measurement angle, and environmental temperature, humidity, and resolution to ensure the diversity and representativeness of the data. The measured signal data is a signal with the abscissa being the wavelength or pixel and the ordinate being the light intensity.

[0072] The measured signal data contains random errors, background errors, and the true signal of the sample. The background error refers to the error that appears every time of measurement and does not change with the measurement conditions and samples; the random error refers to the error that appears accidentally during the measurement process.

[0073] S12, respectively performing parametric equation fitting on each group of measured signal data to obtain the parametric numerical solutions of each group of measured signal data. The parametric numerical solutions include the height information, width information, position information, and shape information of each peak.

[0074] Parametric equation fitting can be implemented using existing mathematical processing programs, such as the scipy library in Python and mathematical processing programs with parametric equation fitting capabilities like Originlab. By performing parametric equation fitting on each set of measured signal data, the numerical solutions of the parameters for a set of measured signal data can contain various information about multiple peaks, such as the height, width, position, and shape of the peaks. Since random errors are usually high-frequency signals, which are significantly different from background errors and real signals, and based on the characteristics of parametric equation fitting, certain random errors can be eliminated in this step.

[0075] S13. Determine the peaks of the background signal originating from the measurement conditions based on the numerical solutions of the parameters for each group.

[0076] Since the numerical solutions of the parameters include the height information, width information, position information, and shape information of each peak, and each set of measured signal data is collected based on the same measurement conditions, this ensures the relative stability of the background error, that is, the background signal belonging to the background error is relatively stable. While the samples to be measured are different, the numerical solutions of the parameters corresponding to the real signals of different samples to be measured vary greatly. Therefore, it is possible to analyze from the numerical solutions of the parameters which are the relatively stable numerical solutions of the parameters belonging to the background signal and which are the numerical solution signals that do not belong to the background signal.

[0077] S14. Based on the determined peaks of the background signal originating from the measurement conditions, obtain the background signal of the measurement conditions.

[0078] Since the background signal is composed of the peaks belonging to the background error in the signal, therefore, having obtained the peaks of the background signal originating from the measurement conditions, the background signal can be obtained according to the peak information belonging to the background signal.

[0079] In this embodiment, by obtaining multiple sets of measured signal data of different samples to be measured under the same measurement conditions in the data acquisition stage, it is ensured that the data is both diverse and representative, providing a rich and reliable data basis for subsequent analysis. Since the measured signal data contains the real signal of the sample, background error, and random error, it is necessary to extract the background signal subsequently. In the parametric equation fitting step, by fitting each set of measured signal data, numerical solutions of the parameters containing the height, width, position, and shape information of each peak are obtained, which can express the signal characteristics in detail. Then, based on the numerical solutions of the parameters, the peak information of the background signal corresponding to the error that does not change with the sample is found, which can accurately locate the characteristics of the background error. Finally, the background signal is obtained according to the peak information of the background signal, achieving the accuracy of extracting the background error, creating conditions for subsequent elimination of the background error and more accurate identification and processing of the real signal of the sample, thereby significantly improving the accuracy and reliability of signal analysis.

[0080] In one embodiment, step S13 includes:

[0081] S131a. For each set of measured signal data, determine whether any peak in the measured signal data exists as a similar peak in all other measured signal data.

[0082] S132a. If a similar peak exists in all other measured signal data, determine that the peak is a peak of the background signal resulting from the measurement conditions.

[0083] Since the background error has the characteristic of not changing with the measurement conditions and the sample, this indicates that its corresponding signal will show consistency in multiple sets of measured signal data, that is, it appears as the same peak. Therefore, by finding the invariant peaks in all measured signals, the peaks of the background signal resulting from the measurement conditions can be accurately located.

[0084] This embodiment makes full use of the characteristic that the background error does not change with the measurement conditions and the sample. By judging the similarity of peaks in each set of measured signal data in other measured signal data, the peaks corresponding to the background error can be accurately screened out, which helps to accurately distinguish the true signal of the sample from the background error subsequently, avoid the interference of the background error on the true signal of the sample, and thus significantly improve the accuracy of the subsequent analysis of the true signal of the sample.

[0085] In another embodiment, step S13 further includes:

[0086] S131b. Perform a clustering analysis on the parameter numerical solutions of each peak to determine whether at least one set of peaks with similar parameters is obtained.

[0087] S132b. If no set of peaks with similar parameters is obtained, the background signal is not recognized.

[0088] S133b. If at least one set of peaks with similar parameters is obtained, each peak in each set results from the background signal of the measurement conditions.

[0089] There is also a method for extracting the peaks of the background signal, that is, by performing a clustering analysis on the parameter numerical solutions of each peak and using the clustering characteristics of the data to identify the peaks of the background signal. Since the characteristics of peaks in actual signal data are often complex and variable, there may be a certain similarity but not exactly the same between different peaks. Clustering analysis can comprehensively consider the similarity between peaks based on multiple parameters (such as peak height, peak width, peak position, peak shape, etc.), and classify the peaks with similar characteristics into one category to form a set. Even if the parameters of the peaks are not exactly the same, as long as they are similar within a certain range, they can be clustered together, so as to more comprehensively and accurately identify the set of peaks with similar characteristics.

[0090] Such as Figure 2After performing parametric equation fitting on each set of measured spectral data to obtain a spectrogram, with the abscissa being the wavelength and the ordinate being the maximum signal intensity, clustering analysis is then carried out based on the spectral information. That is, first, the peak centers are extracted according to the numerical solutions of the parameters (such as Figure 3 shown), and then the similarity between the extracted peak centers is analyzed. Peaks with similar characteristics are grouped into one category, and thus three sets can be obtained as shown in Figure 4 .

[0091] In this embodiment, by performing clustering analysis on the numerical solutions of the parameters of each peak, the relationship between peaks in the signal data can be deeply explored. If there is a set of peaks with similar parameters, it indicates that these peaks have similar characteristics and are very likely to be generated by the same background error factor. The clustering analysis has good adaptability. It can perform clustering dynamically according to the actual distribution of the data. Even if there are certain fluctuations in the parameters of the background signal peaks, they can be effectively clustered together, improving the stability and comprehensiveness of background signal recognition.

[0092] Furthermore, there are various ways to obtain the background signal of the measurement conditions in step S14, specifically as follows:

[0093] As the first implementation method, step S14 includes:

[0094] S141a, Select any one of the measured signal data from multiple sets of measured signal data.

[0095] S142a, Determine the peaks of the background signal originating from the measurement conditions in the selected measured signal data, and obtain the background signal of the measurement conditions using the corresponding numerical solutions of the parameters.

[0096] Since the multiple sets of measured signals contain background signals, through the method of this embodiment, it is possible to conveniently and effectively select any one of the measured signal data from multiple sets of measured signal data, and obtain the background signal according to the numerical solutions of the parameters of this set of measured signals. In this process, only need to arbitrarily select one set of measured signal data from multiple sets of measured signal data, and then based on the identified peaks of the background signal originating from the measurement conditions and their numerical solutions of the parameters, the background signal can be successfully obtained. Among multiple sets of measured signal data, it is not necessary to process the peaks of the background signal for each set of measured signals. Only need to arbitrarily select one set of measured signal data, and the background signal can be determined based on the identified peaks of the background signal and their numerical solutions of the parameters. This greatly reduces the workload and time cost of data processing, especially when dealing with a large amount of signal data, it can significantly improve the efficiency of obtaining the background signal.

[0097] As the second implementation method, step S14 includes:

[0098] S141b, Based on the peaks of the background signal originating from the measurement conditions, extract the background signals from multiple sets of measured signal data respectively.

[0099] S142b, performing comprehensive operations on multiple groups of background signals to obtain background signals under measurement conditions, wherein the comprehensive operations include taking an average or a median.

[0100] Specifically, multiple groups of background signals are obtained based on the peak data of the background signal. For all background signals, all light intensity values ​​corresponding to the same wavelength are comprehensively calculated, and the average light intensity value or median light intensity of multiple groups of background signals at each wavelength point is obtained by taking the average and median. The final background signal is obtained based on the average light intensity value or median light intensity and the corresponding wavelength.

[0101] In this embodiment, the background signal extracted from the measured signal data of different samples under the same measurement conditions may have random errors caused by differences in sample characteristics, accidental measurement factors, etc. Each set of data may be interfered by random errors to varying degrees during the measurement process. By performing comprehensive operations on the peaks of multiple sets of background signals (taking the average or median), these random errors will offset each other to a certain extent. The quality of the background signal can be significantly improved, random noise can be effectively reduced, measurement errors can be balanced, and interference from outliers can be weakened, making the background signal closer to the actual situation.

[0102] As a third implementation, step S14 includes:

[0103] S141c, performing cluster center calculation on the peak of at least one frame of background signal originating from the measurement condition to obtain the background signal of the measurement condition.

[0104] This embodiment only processes the peak data of one or more frames of background signals, and classifies peaks with similar characteristics into different clusters through cluster analysis, finds the typical characteristics of each cluster (cluster center), and thus fits the background signal more accurately. This embodiment only selects part of the peak data, which reduces the complexity and improves the processing rate. In addition, compared with the single processing of a frame of data, cluster analysis can mine the inherent structure and pattern of the data. Single processing is easily affected by accidental factors in the data, and this embodiment can integrate multiple groups of data information through clustering, improve the anti-interference ability and stability of the results, and obtain a more accurate background signal.

[0105] In one embodiment, in step S12, parameter equation fitting is performed for each set of measured signal data, specifically including:

[0106] S121, obtaining a mathematical model adapted to the measured signal data, wherein the mathematical model includes a multi-pseudoVoigt model or a multi-pseudoVoigt and straight line combined mathematical model;

[0107] S122, using a mathematical model and the measured signal data to obtain a fitting equation for the measured signal data.

[0108] Since the measured signal data obtained contains random noise, the prior art usually uses methods such as multi-frame averaging, multi-frame median, moving average smoothing, filtering and noise reduction to remove random noise. However, considering the influence of intermediate frequency information such as power supply ripple, the prior art cannot eliminate the intermediate frequency noise in the measured signal, and although smoothing and noise reduction can reduce the noise of the data, it will also cause the smoothed result to deviate from the original data.

[0109] Therefore, in this embodiment, it is matched in the way of parametric equation fitting, and a suitable model is selected according to the specific display form of the measured signal. For example, if the measured signal contains multiple pseudo Voigt peaks and random noise, a multi-pseudo Voigt model is selected; if the measured signal contains multiple pseudo Voigt peaks and a signal of a linear equation, a mathematical model combining multi-pseudo Voigt and a straight line is selected. In this way, the high-frequency noise and intermediate frequency noise are completely screened out.

[0110] Among them, each pseudo Voigt represents a peak, and the expression of each pseudo Voigt is:

[0111] ;

[0112] Among them, represents the intensity of the signal, represents the maximum value of the signal intensity, represents the proportion of the Gaussian signal in the signal, represents the wavelength of the signal, represents the symmetry center of the signal, represents the width parameter of the signal.

[0113] Multi-pseudo Voigt is the superposition of the expressions of multiple pseudo Voigts.

[0114] The expression of the linear equation is:

[0115] Y = a*X + b;

[0116] Among them, Y represents the intensity of the signal, X represents the wavelength of the signal, and a and b are constants.

[0117] For the multi-pseudo Voigt model, it fits the measured signal data with a parametric equation, and appropriately adjusts the parameter range and the number of peaks to achieve a lower root mean square error, improve the accuracy of the fitting equation, and better reflect the law of the measured signal data. Considering the robustness, generality, and overfitting risk of the model, the number of peaks should not be too large. Select different numbers of peaks for parametric equation fitting and solve the RMSE. If the RMSE does not increase significantly after the number of peaks reaches a specific value, then the increase in the number of peaks should be stopped at this time. As can be seen from the experiment, for example Figure 5 As shown, when the number of peaks = 3, the fitting equation (red curve) can already better reflect the law of the spectral data (blue curve), while when the number of peaks = 4, there is only a slight improvement. At this time, a three-peak equation can be selected for fitting, that is, the three-pseudo Voigt model is the optimal one.

[0118] Furthermore, if the mathematical model adapted to the measured signal data is a combined mathematical model of multi-pseudo Voigt and a straight line, then intercept the straight line segment and the first peak segment and the second peak segment closest to both sides of the straight line segment from the measured signal data;

[0119] Perform single-pseudo Voigt peak fitting on the first peak segment and the second peak segment respectively to obtain the first pseudo Voigt function and the second pseudo Voigt function;

[0120] Obtain the vertices of the first pseudo Voigt function and the second pseudo Voigt function, and connect the two vertices;

[0121] Calculate the sum function according to the first pseudo Voigt function and the second pseudo Voigt function;

[0122] Use the coordinates of multiple intersection points of the connecting line and the sum function to connect again to obtain the target straight line, and judge whether the target straight line is the tangent line of the peak point of the sum function;

[0123] If the target straight line is the tangent line of the peak point of the sum function, then use the target abscissas of multiple intersection points to intercept the two peak segments of the sum function, and the target abscissas are obtained by averaging the abscissas of multiple intersection points;

[0124] Calculate the fitting equation of the background signal according to the intercepted data and the target straight line.

[0125] Specifically, taking the measured signal as the form of a double-pseudo Voigt peak plus a straight line equation as an example, such as Figure 6For the blue curve shown, first intercept the first peak segment A, the straight-line segment B, and the second peak segment C. Then, perform single pseudo-Voigt peak fitting on the first peak segment A and the second peak segment C respectively to obtain the first pseudo-Voigt function and the second pseudo-Voigt function. Calculate the sum function of the two functions, and connect the vertices of the two pseudo-Voigt functions. As long as there is a difference in the height of the pseudo-Voigt functions, the connecting line will surely pass through the sum function of the two pseudo-Voigt functions 4 times. Connect the line again according to the abscissa of the intersection point and the corresponding ordinate of the sum function to obtain the target connecting line. Only need to judge whether the target connecting line is the tangent of the peak point of the sum function. If it is a tangent, take the abscissas of the 4 intersection points, which are x1, x2, x3, and x4 respectively. Take x5 = (x1 + x2) / 2 and x6 = (x3 + x4) / 2 to obtain the target abscissas x5 and x6. Intercept the peak segments of the sum function according to x5 and x6, that is, take a peak segment in the range of x ≤ x5 and a peak segment in the range of x ≥ x6. And the target straight-line range is x5 < x < x6. Finally, process the two peak segments and the target straight line to obtain the fitting equation of the background signal. The actual background signal is shown as Figure 6 the orange curve in

[0126] Further, connect the lines again using the coordinates of multiple intersection points of the connecting line and the sum function to obtain the target straight line, and judge whether the target straight line is the tangent of the peak point of the sum function, including:

[0127] Obtain multiple target intersection points according to the target abscissa and its corresponding ordinate;

[0128] Connect the multiple target intersection points to obtain the target straight line;

[0129] Calculate the error between the ordinate of the target straight line and the sum function at a certain abscissa and perform error judgment. If the error is less than the preset error, regard the target straight line as the tangent of the peak point of the sum function;

[0130] If the error is greater than the preset error, continue to connect the lines and iterate according to the intersection points of the target straight line and the sum function until the error between the ordinate of the target straight line after iteration and the sum function at a certain abscissa is less than the preset error, and regard the target straight line as the tangent of the peak point of the sum function.

[0131] Specifically, when determining whether the target line is the tangent line of the sum function peak point, the corresponding vertical coordinates y values are taken as y1 and y2 in the sum function according to the target abscissas x5 and x6, and a new line is obtained by connecting (x5, y1) and (x6, y2). The new line still has 4 intersections with the sum function, but at this time the line is closer to the tangent line of the double pseudo Voigt function. Repeat this process continuously until the absolute error between the y obtained from the line equation and the y obtained from the sum function is less than the preset error, then the iteration can be stopped to obtain the final target line. The abscissas of the two points of the final target line are x5' and x6'. Take a peak segment in the range of x ≤ x5', take a peak segment in the range of x ≥ x6', and the range of the target line is x5' < x < x6'. Finally, the two peak segments and the target line are processed to obtain the fitting equation of the background signal. The actual background signal is shown as Figure 6 the orange curve in

[0132] In this embodiment, equation fitting is performed on the measured signal combined with multiple pseudo Voigt and a straight line. By performing segmented processing on the measured signal and fitting the peak segment and the straight line segment respectively, the characteristics of the peak can be focused on, thereby improving the accuracy of the entire signal fitting. In the process of finding the target tangent line, an iterative method is used to update the straight line equation multiple times until it becomes the common tangent line of the double peak or multiple peaks and meets the accuracy requirements. This iterative process can effectively overcome the error of the initial estimation and make the finally obtained target line more in line with the actual situation. Even when there are certain fluctuations or noises in the measured signal data, relatively stable and accurate results can be obtained through multiple iterations, improving the accuracy of the parameter numerical solution.

[0133] As Figure 7 shown, an embodiment of the present invention further provides a signal processing method, which is executed by an electronic device such as a computer or a server, and specifically includes:

[0134] S21, obtaining the background signal according to the background signal extraction method of any one of the above.

[0135] S22, subtracting the background signal from the measured signal corresponding to the sample to be measured collected to obtain the true signal of the sample to be measured.

[0136] After obtaining the accurate background signal when the sample is not placed, there is also a signal processing method, that is, eliminating the background signal in the measured signal, and the true signal of the sample to be measured can be obtained, which greatly improves the quality of the signal data and provides accurate data for the subsequent analysis of the sample.

[0137] The background signal extraction method and signal processing method provided in this embodiment can be applied to the field of semiconductor material analysis. For example, it can be specifically used for thickness measurement of wafer films. The thickness measurement system collects the reflected light beams of the wafer film for a multi-wavelength light source to obtain the spectral relationship between the wavelength and reflectivity of the reflected light beams (the measured sample is the wafer film, and the measured signal data is the spectrum of the reflected light beam). Using this method, the background spectrum (background signal) of the measurement conditions where this system is located can be obtained. Subtracting the background spectrum from the spectrum of the reflected light beam of the wafer film can obtain the true spectrum of the wafer film, and then the thickness information of the wafer film can be obtained based on this true spectrum, which can improve the accuracy of wafer film thickness measurement.

[0138] Embodiment 2

[0139] As Figure 8 shown, the embodiment of the present invention also provides a background signal extraction method. This method extracts the background signal in the absence of a sample, and this method is executed by an electronic device such as a computer or a server. Specifically, it includes:

[0140] S31, obtain at least one set of measured signal data without the measured sample placed, and each set of measured signal data is collected based on the same measurement conditions.

[0141] S32, perform parametric equation fitting for each set of measured signal data respectively to obtain the parametric numerical solutions of each set of measured signal data. The parametric numerical solutions include the height information, width information, position information, and shape information of each peak.

[0142] S33, obtain the background signal of the measurement conditions based on the parametric numerical solutions of each set of measured signal data.

[0143] This method directly obtains the background signal data without placing a sample, and collects at least one set of signal data under the same measurement conditions, and then performs parametric equation fitting. According to the obtained parametric equation solutions, the background signal is obtained. Obtaining signal data directly without placing a sample can exclude the interference of the sample signal on the background signal. The obtained signal data reflects the background information generated by the measurement conditions. Moreover, by directly obtaining the signal data without placing a sample, only these data need to be subjected to parametric equation fitting to obtain the background signal, which greatly simplifies the background signal extraction process and improves work efficiency. In addition, obtaining at least one set of measured signal data ensures that the background signal can be effectively obtained even with limited data. When obtaining multiple sets of measured signal data, the obtained background signal can be made smoother.

[0144] Further, there are various ways to obtain the background signal of the measurement conditions in step S33. Specifically, as follows:

[0145] As the first implementation method, step S33 includes:

[0146] S331a. Select any one of the measured signal data from multiple groups of measured signal data.

[0147] S332a. Use the parameter numerical solution of the selected measured signal data to obtain the background signal of the measurement condition.

[0148] Through the method of this embodiment, any measured signal can be conveniently and effectively selected from multiple groups of measured signal data, and the background signal can be obtained according to the parameter numerical solution of this group of measured signals. In this process, only one group of measured signal data needs to be selected from multiple groups of measured signal data for parameter equation fitting to obtain the parameter numerical solution, and then the background signal can be obtained. There is no need to process each group of measured signals, and only any one group of measured signal data needs to be selected. This greatly reduces the workload and time cost of data processing, especially when dealing with a large amount of signal data, and can significantly improve the efficiency of obtaining the background signal.

[0149] As the second implementation method, step S33 includes:

[0150] S331b. Use the parameter numerical solutions of multiple groups of measured signal data to obtain the background signals of the corresponding measurement conditions.

[0151] S332b. Perform a comprehensive operation on multiple groups of background signals to obtain the background signal of the measurement condition, and the comprehensive operation includes taking the average or median.

[0152] The specific comprehensive operation method is the same as the step of placing the sample to be measured, and will not be elaborated here.

[0153] Since each group of data may be affected by different degrees of random error interference during the measurement process, it is necessary to process the background signals obtained from the parameter numerical solutions of all groups, that is, through the comprehensive operation (taking the average or median) of the background signals, and these random errors will cancel each other out to a certain extent. It can significantly improve the quality of the background signal, effectively reduce random noise, balance measurement errors, and weaken the interference of outliers, making the background signal closer to the real situation.

[0154] As the third implementation method, step S33 includes:

[0155] S331c. Calculate the cluster center of the parameter numerical solutions of multiple groups of measured signal data to obtain the background signal of the measurement condition.

[0156] In this embodiment, through cluster analysis of the parameter numerical solutions of multiple groups of background signals, the peaks with similar characteristics are grouped into different clusters, and the typical characteristics (cluster centers) of each cluster are found, so as to more accurately fit the background signal. Through clustering, this embodiment can synthesize the information of multiple groups of data, improve the anti-interference ability and stability of the result, and obtain a more accurate background signal.

[0157] In one embodiment, for each group of measured signal data in step S32, parameter equation fitting is performed respectively, specifically including:

[0158] S321, obtain a mathematical model adapted to the measured signal data, where the mathematical model includes a multi-pseudoVoigt model or a mathematical model combining multi-pseudo Voigt and a straight line;

[0159] S322, use the mathematical model and the measured signal data to obtain the fitting equation of the measured signal data.

[0160] Further, if the mathematical model adapted to the background signal is a mathematical model combining multi-pseudo Voigt and a straight line, then intercept a straight line segment and the first peak segment and the second peak segment closest to both sides of the straight line segment from the measured signal data;

[0161] Perform single-pseudo Voigt peak fitting on the first peak segment and the second peak segment respectively to obtain a first pseudoVoigt function and a second pseudo Voigt function;

[0162] Obtain the vertices of the first pseudo Voigt function and the second pseudo Voigt function, and connect the two vertices;

[0163] Calculate the sum function according to the first pseudo Voigt function and the second pseudo Voigt function;

[0164] Use the coordinates of multiple intersection points of the connection line and the sum function to connect again to obtain a target straight line, and determine whether the target straight line is the tangent of the peak point of the sum function;

[0165] If the target straight line is the tangent of the peak point of the sum function, then intercept two peak segments of the sum function using the target abscissas of multiple intersection points, and the target abscissas are obtained by averaging the abscissas of multiple intersection points;

[0166] Calculate the fitting equation of the background signal according to the intercepted data and the target straight line.

[0167] Further, using the coordinates of multiple intersection points of the connection line and the sum function to connect again to obtain a target straight line, and determining whether the target straight line is the tangent of the peak point of the sum function includes:

[0168] According to the target abscissas and their corresponding ordinates, obtain multiple target intersection points;

[0169] Connect the multiple target intersection points to obtain a target straight line;

[0170] Calculate the error between the target line and the ordinate of the sum function at a certain abscissa, and perform error judgment. If the error is less than the preset error, the target line is regarded as the tangent line of the peak point of the sum function;

[0171] If the error is greater than the preset error, continue to perform connection iteration according to the intersection points of the target line and the sum function until the error between the target line after iteration and the ordinate of the sum function at a certain abscissa is less than the preset error, and regard this target line as the tangent line of the peak point of the sum function.

[0172] The specific method of performing parametric equation fitting on each group of measured signal data respectively is the same as the parametric fitting equation fitting steps when placing the measured sample, and will not be elaborated here.

[0173] The embodiment of the present invention also provides a signal processing method, which is executed by an electronic device such as a computer or a server, and specifically includes:

[0174] S41, obtain the background signal according to the background signal extraction method described in any one of the above.

[0175] S42, subtract the background signal from the measured signal corresponding to the measured sample collected to obtain the true signal of the measured sample.

[0176] After obtaining the accurate background signal when no sample is placed, there is also a signal processing method, that is, eliminating the background signal in the measured signal, and the true signal of the measured sample can be obtained, which greatly improves the quality of signal data and provides accurate data for the subsequent analysis of the sample.

[0177] The background signal extraction method and the signal processing method provided in this embodiment can be applied to the field of semiconductor material analysis. For example, it can be specifically used for thickness measurement of a wafer thin film. Without placing the wafer thin film (the measured sample), collect the corresponding relationship spectrum of wavelength and reflectivity under this measurement condition (which can be understood as the spectrum of ambient light), and use this method to obtain the background spectrum (background signal) of the measurement condition where this system is located. Subtract the background spectrum from the spectrum of the reflected light beam of the wafer thin film to obtain the true spectrum of the wafer thin film, and then obtain the thickness information of the wafer thin film according to this true spectrum, which can improve the accuracy of wafer thin film thickness measurement.

[0178] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0179] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to produce a machine, such that the instructions executed by the processors of the computer or other programmable data processing devices produce means for implementing the functions specified in one or more flows Figure 1 one or more flows and / or blocks Figure 1 or means for implementing the functions specified in one or more blocks.

[0180] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including instruction means that implement the functions specified in one or more flows Figure 1 one or more flows and / or blocks Figure 1 or means for implementing the functions specified in one or more blocks.

[0181] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operational steps are performed on the computer or other programmable device to produce a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more flows Figure 1 one or more flows and / or blocks Figure 1 or means for implementing the functions specified in one or more blocks.

[0182] Obviously, the above embodiments are only examples for clear illustration and not limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to exhaustively list all implementation manners here. And the obvious changes or modifications derived therefrom are still within the protection scope of the present invention.

Claims

1. A background signal extraction method, characterized in that: include: Acquire multiple groups of measured signal data of multiple samples, each group of measured signal data is collected based on the same measurement conditions; Performing parameter equation fitting for each set of measured signal data to obtain a parameter numerical solution for each set of measured signal data, wherein the parameter numerical solution includes height information, width information, position information and shape information of each peak; Determining the peak of the background signal originating from the measurement condition according to each group of the parameter numerical solutions, including, for each group of measured signal data, determining whether any peak in the measured signal data has similar peaks in other measured signal data, and if similar peaks exist in other measured signal data, determining that the peak is the peak of the background signal originating from the measurement condition; Based on the determined peak of the background signal originating from the measurement condition, the background signal of the measurement condition is obtained.

2. The method according to claim 1, characterized in that Determining the peak of the background signal originating from the measurement condition according to the numerical solution of each group of the parameters includes: Performing cluster analysis on the parameter numerical solution of each peak to determine whether at least one set of peaks with similar parameters is obtained; If a collection of peaks with similar parameters is not obtained, the background signal is not identified; If at least one set of peaks with similar parameters is obtained, each peak in each set originates from the background signal of the measurement condition.

3. The method according to claim 1 or 2, characterized in that: The obtaining the background signal of the measurement condition based on the determined peak of the background signal originating from the measurement condition comprises: Select any one of the multiple groups of measured signal data; The peak of the background signal originating from the measurement condition is determined in the selected measured signal data, and the background signal of the measurement condition is obtained by using a corresponding parameter numerical solution.

4. The method according to claim 1 or 2, characterized in that: The obtaining the background signal of the measurement condition based on the determined peak of the background signal originating from the measurement condition comprises: Based on the peak of the background signal originating from the measurement condition, respectively extracting background signals from a plurality of groups of measured signal data; A comprehensive operation is performed on multiple groups of background signals to obtain the background signals of the measurement conditions, wherein the comprehensive operation includes taking an average or a median.

5. The method according to claim 1 or 2, characterized in that: The obtaining the background signal of the measurement condition based on the determined peak of the background signal originating from the measurement condition comprises: A cluster center calculation is performed on the peak of at least one frame of the background signal originating from the measurement condition to obtain the background signal of the measurement condition.

6. The method according to claim 1 or 2, characterized in that: The performing parameter equation fitting for each set of measured signal data respectively includes: Acquire a mathematical model adapted to the measured signal data, wherein the mathematical model comprises a multi-pseudo Voigt model or a multi-pseudo Voigt and straight line combined mathematical model; The mathematical model and the measured signal data are used to obtain a fitting equation for the measured signal data.

7. The method according to claim 6, characterized in that If the mathematical model adapted to the measured signal data is a multi-pseudo Voigt and straight line combined mathematical model, a straight line segment and a first peak segment and a second peak segment located closest to both sides of the straight line segment are intercepted from the measured signal data; Performing single pseudo Voigt peak fitting on the first peak segment and the second peak segment respectively to obtain a first pseudo Voigt function and a second pseudo Voigt function; Obtaining vertices of the first pseudo Voigt function and the second pseudo Voigt function, and connecting the two vertices; calculating a sum function according to the first pseudo Voigt function and the second pseudo Voigt function; Using the coordinates of the connecting line and the multiple intersection points of the sum function to connect the line again, a target straight line is obtained, and it is determined whether the target straight line is a tangent line to the peak point of the sum function; If the target straight line is a tangent line of the peak point of the sum function, two peak segments of the sum function are intercepted using target horizontal coordinates of multiple intersection points, where the target horizontal coordinates are calculated by averaging the horizontal coordinates of multiple intersection points; The fitting equation of the background signal is obtained by calculation based on the intercepted data and the target straight line.

8. The method according to claim 7, characterized in that Using the coordinates of the connecting line and the multiple intersection points of the sum function to connect the line again to obtain a target straight line, and determining whether the target straight line is a tangent line of the peak point of the sum function, including: According to the target abscissa and its corresponding ordinate, a plurality of target intersection points are obtained; Connecting a plurality of target intersection points to obtain a target straight line; Calculate the error between the target straight line and the ordinate of the sum function at a certain abscissa and make an error judgment. If the error is less than a preset error, the target straight line is regarded as a tangent to the peak point of the sum function. If the error is greater than the preset error, continue to iterate the connection based on the intersection of the target straight line and the sum function until the error between the vertical coordinate of the iterative target straight line and the sum function at a certain horizontal coordinate is less than the preset error, and the target straight line is regarded as the tangent of the peak point of the sum function.

9. A background signal extraction method, characterized in that: include: Acquire at least one set of measured signal data without placing the measured sample, each set of the measured signal data is collected based on the same measurement conditions; Perform parameter equation fitting for each group of measured signal data to obtain a parameter numerical solution for each group of measured signal data, wherein the parameter numerical solution includes height information, width information, position information and shape information of each peak, and further includes obtaining a mathematical model suitable for the measured signal data, wherein the mathematical model includes a multi-pseudo Voigt model or a multi-pseudo Voigt and straight line combined mathematical model; using the mathematical model and the measured signal data to obtain a fitting equation for the measured signal data, if the mathematical model suitable for the measured signal data is a multi-pseudo Voigt and straight line combined mathematical model, then intercepting a straight line segment and a first peak segment and a second peak segment located closest to both sides of the straight line segment from the measured signal data; performing single pseudo Voigt peak fitting on the first peak segment and the second peak segment respectively to obtain a first pseudo Voigt function and a second pseudo Voigt function; obtaining vertices of the first pseudo Voigt function and the second pseudo Voigt function, and connecting the two vertices; and calculating a fitting equation for the measured signal data according to the first pseudoVoigt function and the second pseudoVoigt function. The Voigt function calculates the sum function; the connecting line is used to connect the coordinates of the multiple intersection points of the sum function again to obtain a target straight line, and it is determined whether the target straight line is a tangent to the peak point of the sum function; if the target straight line is a tangent to the peak point of the sum function, two peak segments of the sum function are intercepted using the target horizontal coordinates of the multiple intersection points, and the target horizontal coordinate is calculated by averaging the horizontal coordinates of the multiple intersection points; the fitting equation of the background signal is calculated based on the intercepted data and the target straight line; Based on the parameter numerical solution of each set of measured signal data, the background signal of the measurement condition is obtained.

10. The method according to claim 9, characterized in that The parameter numerical solution based on each set of measured signal data to obtain the background signal of the measurement condition includes: Select any one of the multiple groups of measured signal data; The background signal of the measurement condition is obtained by using the parameter numerical solution of the selected measured signal data.

11. The method according to claim 9, characterized in that The parameter numerical solution based on each set of measured signal data to obtain the background signal of the measurement condition includes: Using the parameter numerical solutions of the multiple groups of measured signal data, the background signals corresponding to the measurement conditions are obtained; A comprehensive operation is performed on multiple groups of background signals to obtain the background signals of the measurement conditions, wherein the comprehensive operation includes taking an average or a median.

12. The method according to claim 9, characterized in that The parameter numerical solution based on each set of measured signal data to obtain the background signal of the measurement condition includes: Cluster center calculation is performed on parameter numerical solutions of multiple groups of measured signal data to obtain background signals under the measurement conditions.

13. The method according to claim 9, characterized in that Using the coordinates of the connecting line and the multiple intersection points of the sum function to connect the line again to obtain a target straight line, and determining whether the target straight line is a tangent line of the peak point of the sum function, including: According to the target abscissa and its corresponding ordinate, a plurality of target intersection points are obtained; Connecting a plurality of target intersection points to obtain a target straight line; Calculate the error between the target straight line and the ordinate of the sum function at a certain abscissa and make an error judgment. If the error is less than a preset error, the target straight line is regarded as a tangent to the peak point of the sum function. If the error is greater than the preset error, continue to iterate the connection based on the intersection of the target straight line and the sum function until the error between the vertical coordinate of the iterative target straight line and the sum function at a certain horizontal coordinate is less than the preset error, and the target straight line is regarded as the tangent of the peak point of the sum function.

14. A signal processing method, characterized in that: include: Obtaining a background signal according to the background signal extraction method described in any one of claims 1 to 13; The background signal is subtracted from the actual measured signal corresponding to the sample being tested to obtain the real signal of the sample being tested.

15. A background signal extraction device, characterized in that: include: A processor and a memory connected to the processor; wherein the memory stores instructions executable by the processor, and the instructions are executed by the processor so that the processor executes the background signal extraction method as described in any one of claims 1-13.

16. A signal processing device, characterized in that: include: A processor and a memory connected to the processor; wherein the memory stores instructions executable by the processor, and the instructions are executed by the processor so that the processor performs the signal processing method as claimed in claim 14.

Images