A method and system for multi-component analysis of complex system materials based on raman signals

Abstract

The application discloses a kind of complex system material multi-component analysis method and system based on Raman signal, the multi-component analysis method based on Raman signal of complex system material proposed in the application, the present application carries out uniform feature extraction to the spectrum collected and known spectrum, can avoid the feature information extracted due to different parameters settings is different, the material composition of mixed system is pre-judged by permutation and combination form, the characteristic peak of different material can also be identified under the condition of high coupling, has resolution ability to chemical structure analog, improve the accuracy of material identification.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Raman signal processing, and in particular to a method and system for analyzing multiple components of complex system substances based on Raman signals. BACKGROUND

[0002] The Raman spectrum signal under a mixture system contains characteristic information of multiple substance components, that is, the spectrum is coupled by the characteristic information of multiple substances, so that the substance components of the mixture system can be identified according to the spectrum. The conventional method is to extract the peak position, peak width and other characteristic information of the spectrum of the measured substance, load the above characteristic information into a database for retrieval, and if the peak position is adjacent and the peak width and the number of peaks are equal, it is considered that the mixture system contains the substance component. This process has three problems, namely, peak position error, peak width range and identification error. Among them, the peak position deviation is caused by different parameter settings of the extracted characteristic information. The peak position data of each spectrum in the database and the peak position data of the measured substance need to maintain the same method and parameters (such as signal-to-noise ratio). The database spectrum is the Raman signal of a single substance, while the collected Raman signal is under a mixture system, the signals of each substance interfere and superimpose each other, and the peak width problem needs to be considered when calculating the similar peak shape. The peak width range directly affects the calculation coefficient. The conventional method can screen substances from the database, but the substance is still a possible substance, which needs to be verified by combination. SUMMARY

[0003] To solve the technical problems in the background art, the present application provides a method and system for analyzing multiple components of complex system substances based on Raman signals.

[0004] The method for analyzing multiple components of complex system substances based on Raman signals provided by the present application comprises the following steps:

[0005] S1, performing data processing on the collected Raman spectrum data to obtain the collected spectrum signal and the collected spectrum peak position;

[0006] S2, performing the data processing method of S1 on the known Raman spectrum data to obtain the known spectrum signal and the known spectrum peak position, and establishing a known spectrum database;

[0007] S3, performing peak position matching in the known spectrum database based on the collected spectrum peak position to screen the known spectrum signals with similar peak positions;

[0008] S4, arranging and combining the similar known spectrum signals, and performing the data processing method of S1 on the spectrum data of each combination type to compare the collected spectrum signal of S1 with the possible spectrum signal, and identifying multiple components.

[0009] Preferably, in S1, the data processing specifically comprises: filtering out Gaussian white noise by using an S-G filter; performing baseline correction on the spectrum by using airPLS; and supplementing data to the required number of data points by using a multiple spline interpolation method.

[0010] Preferably, in S1, the data processing further comprises: performing wavelet transform on the collected Raman spectrum data to obtain WTf(a, τ) and solving extreme values to obtain a peak position.

[0011] Preferably, the wavelet transform on the collected Raman spectrum data to obtain WTf(a, τ) and solving extreme values specifically comprises:

[0012] The wavelet transform on the collected Raman spectrum data to obtain WTf(a, τ) and solving extreme values to obtain a minimum value X1 and a first maximum value X2; solving extreme values on the collected Raman spectrum data to obtain a second maximum value X3; comparing X2 and X3, if the absolute value of the difference is less than a preset parameter, then the peak position accurate value X4 is X3; and a peak width PW = X1(i+1)-X1i, wherein X1i is the ith minimum value.

[0013] Preferably, in S2, the known spectrum database comprises known spectrum signals, peak positions, peak widths, and peak strengths.

[0014] Preferably, in S3, after the peak position matching in the known spectrum database based on the collected spectrum peak position, a waveform calculation is performed on each matching peak, and possible substances are screened out according to a waveform similarity coefficient; wherein, the waveform similarity coefficient calculation formula is as follows:

[0015]

[0016] Wherein, f1k is the kth matching peak data of the known spectrum, f1 is the matching peak mean value of the known spectrum, f2k is the kth matching peak data of the collected spectrum, and f2 is the matching peak mean value of the collected spectrum.

[0017] Preferably, in S4, the comparison between the collected spectrum signal of S2 and the possible spectrum signal specifically comprises: calculating the error coefficient of the two by using a deviation variance method, and calculating by using the following formula:

[0018]

[0019] Wherein, Ri is the waveform similarity coefficient of the ith matching peak, and R is the waveform similarity coefficient mean value of the ith matching peak.

[0020] In the present application, the proposed multi-component analysis method of complex system substances based on Raman signals can uniformly extract features from collected spectra and known spectra, can avoid different extracted feature information caused by different parameter settings, can predict the components of the mixed system through permutation and combination, can identify the characteristic peaks of different substances under high coupling, has the ability to distinguish similar chemical structures, and improves the accuracy of substance identification.

[0021] The present application also provides a multi-component identification system for analyzing the components of complex system substances, comprising:

[0022] a processor;

[0023] a memory; and

[0024] one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the programs comprise a program for executing the above-mentioned multi-component analysis method of complex system substances based on Raman signals.

[0025] In the present application, the proposed multi-component analysis system of complex system substances based on Raman signals has similar technical effects to the above-mentioned method, and thus will not be described again. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 The Raman signal of an ethanol and ethylene glycol mixed system in one embodiment of the multi-component analysis method of complex system substances based on Raman signals provided by the present application.

[0027] Figure 2 The Raman signal of a solid mixture system in one embodiment of the multi-component analysis method of complex system substances based on Raman signals provided by the present application. DETAILED DESCRIPTION

[0028] As shown in Figure 1 and 2 , Figure 1 The Raman signal of an ethanol and ethylene glycol mixed system in one embodiment of the multi-component analysis method of complex system substances based on Raman signals provided by the present application, Figure 2 The Raman signal of a solid mixture system in one embodiment of the multi-component analysis method of complex system substances based on Raman signals provided by the present application.

[0029] The present application provides a multi-component analysis method of complex system substances based on Raman signals, comprising the following steps:

[0030] S1, performing data processing on the collected Raman spectrum data to obtain collected spectrum signals and collected spectrum peak positions;

[0031] Specifically, the data processing specifically includes: filtering out Gaussian white noise by using an S-G filter; performing baseline correction on the spectrum by using airPLS; and supplementing data to the required number of data points by using a multiple spline interpolation method. In actual processing, the S-G filter noise smoothing is based on a window width of 5, the Raman signal is corrected based on an adaptive iterative penalty weighted least squares method, the Raman shift conversion is performed on the data wavenumber, the data range is intercepted at 200cm-1-3000cm-1, and finally the data is obtained by performing three times spline interpolation.

[0032] Then, the peak position of the collected spectrum is calculated according to the above data. Specifically, the data processing further includes: performing wavelet transform on the collected Raman spectrum data to obtain WTf(a, τ) and extreme value solving to obtain the peak position. First, the collected Raman spectrum data is wavelet transformed to obtain WTf(a, τ), and extreme value solving is performed to obtain a minimum value X1 and a first maximum value X2. Then, the collected Raman spectrum data is subjected to extreme value solving to obtain a second maximum value X3. Comparing X2 and X3, if the absolute value of the difference is less than a preset parameter, then the accurate value X4 of the peak position is X3. The peak width is determined by the difference between the adjacent two minimum values. The peak width PW = X1(i+1)-X1i, wherein X1i is the i-th minimum value.

[0033] S2, performing the data processing method of S1 on the known Raman spectrum data to obtain the known spectrum signal and the known spectrum peak position, and establishing a known spectrum database;

[0034] Specifically, the known spectrum database includes known spectrum signals, peak positions, peak widths, and peak strengths.

[0035] S3, performing peak position matching in the known spectrum database based on the collected spectrum peak position, and screening known spectrum signals similar in peak position;

[0036] Specifically, after performing peak position matching in the known spectrum database based on the collected spectrum peak position, the waveform of each matching peak is calculated, and possible substances are screened out according to the waveform similarity coefficient. The waveform similarity coefficient calculation formula is as follows:

[0037]

[0038] Wherein, f1k is the k-th matching peak data of the known spectrum, f1 is the matching peak mean value of the known spectrum, f2k is the k-th matching peak data of the collected spectrum, and f2 is the matching peak mean value of the collected spectrum.

[0039] S4, performing arrangement and combination on the similar known spectrum signals, and performing the data processing method of S1 on the spectrum data of each combination type, comparing the collected spectrum signal of S1 with the possible spectrum signal, and performing multi-component identification.

[0040] Specifically, the collected spectrum signal of S2 is compared with the possible spectrum signal, specifically: the error coefficient of the two is calculated by the deviation variance method, and the following formula is used to calculate:

[0041]

[0042] Wherein, Ri is the waveform similarity coefficient of the ith matching peak, and Ri is the mean value of the waveform similarity coefficient of the ith matching peak.

[0043] The Raman signal-based complex system substance multi-component analysis method of the embodiment is described in detail through two examples.

[0044] Example 1

[0045] The Raman signal of the mixture system of the easily-made-toxic chemical ethanol and ethylene glycol is taken as an example to further illustrate the technical scheme of the present application.

[0046] Sample preparation: the standard samples of ethanol and ethylene glycol are mixed in any ratio (the ratio range is not more than 1:5), and a glass sample bottle is used for loading.

[0047] Detection instrument: portable Raman spectrometer (CASA18T1, Anhui Zhongke Saifeier Technology Co., Ltd.), laser power 250mW, integration time 5000mS, laser focal length 7±0.5cm, Raman detection method, signal-to-noise ratio parameter setting 800.

[0048] The steps of the method of the present example include:

[0049] Step 1: Place the mixed system under the Raman instrument to collect spectrum data f.

[0050] Step 2: Use S-G filter to filter out Gaussian white noise and smooth the spectrum; use airPLS to correct the baseline of the spectrum; use multiple spline interpolation method to supplement the data to the required data points f1.

[0051] Step 3: Perform wavelet transform on the Raman spectrum signal f1 to get WTf(a, τ), and perform extreme value solving, the minimum value is X1, and the maximum value is X2. Perform extreme value solving on the Raman spectrum signal f1, and record it as X3. Compare the two maximum value sets X2 and X3, and the absolute value of the element difference is less than the parameter a, then the elements in the X3 set are the accurate values of the Raman signal peak X4. Peak width start and end position judgment. For the condition X1i≤X4j≤X1(i+1), the peak width is recorded as PW=X1(i+1)-X1i.

[0052] Step 4: All the spectral data are subjected to step 2 to obtain standard spectral data, which is added to the database, and then the standard spectral data are subjected to step 3 to extract the peak position, peak width, peak intensity and other characteristics of each spectrum and save them to the database.

[0053] Step 5: According to the peak position X4 of the to-be-tested substance, the spectrum library is searched to realize the first matching of possible substances. This process cannot accurately exclude, and the waveform calculation of each matching peak is required to further exclude possible substances according to the waveform similarity coefficient.

[0054] Step 6: The possible substances are arranged and combined, each combination form is compared, and then steps 2, 3, 4 and 5 are performed on the combined spectrum to calculate the error coefficient with the original spectrum by the deviation variance method.

[0055] The experimental results are shown in Figure 1 This method can analyze the substance components of alcohol mixed samples and list the corresponding Raman signals of the substances, which are the Raman signals of ethanol and ethylene glycol. The characteristic peak positions of the mixed sample and the standard sample in the figure match, indicating that this method can accurately identify the substance components, and the signal resolution reaches 5 Raman shifts.

[0056] Example 2

[0057] The SERS signal of the mixed system of the easily-made-toxic chemical propionic acid and succinic acid is taken as an example to further illustrate the technical scheme of the present application.

[0058] Sample preparation: Take any proportion of propionic acid and succinic acid standard sample in a mortar, mix thoroughly, ensure uniform mixing of the mixture (the proportion range is not more than 1:5), and use a glass sample bottle to hold the sample.

[0059] Detection instrument: portable Raman spectrometer (CASA18T1, Anhui Zhongke Saifeier Technology Co., Ltd.), laser power 250mW, integration time 5000mS, laser focal length 7±0.5cm, select Raman detection method, and the signal-to-noise ratio parameter is set to 800.

[0060] The steps of the method of the present example include:

[0061] Step 1: Place the mixed system under the Raman instrument, collect 10 Raman spectra at different sites, and select the Raman spectrum with the strongest characteristic peak intensity as the algorithm processing data f.

[0062] Step 2: Use S-G filter to filter out Gaussian white noise and smooth the spectrum; use airPLS to correct the baseline of the spectrum; use multiple spline interpolation method to supplement the data to the required data point number f1.

[0063] Step 3: wavelet transform is performed on the Raman spectrum signal f1 to obtain WTf(a, τ), and extreme value solving is performed, and the minimum value is denoted as X1, and the maximum value is denoted as X2. Extreme value solving is performed on the Raman spectrum signal f1, and the result is denoted as X3. The two maximum value sets X2 and X3 are compared, and if the absolute value of the element difference is less than the parameter a, then the element in the X3 set is the accurate Raman signal peak position value X4. Peak width start and end position discrimination. For the condition X1i≤X4j≤X1(i+1), the peak width is denoted as PW=X1(i+1)-X1i.

[0064] Step 4: all spectrum data are subjected to step 2 to obtain standard spectrum data, which is added to a database, and then step 3 is performed on the standard spectrum data to extract peak position, peak width, peak strength and the like characteristics of each spectrum and save to the database.

[0065] Step 5: according to the peak position X4 of the to-be-measured substance, searching is performed in the spectrum library to realize the first possible substance matching. This process cannot accurately exclude, and waveform calculation is required to be performed on each matching peak, and the possible substances are further excluded according to the waveform similarity coefficient.

[0066] Step 6: the possible substances are arranged and combined, each combination form is compared, and then steps 2, 3, 4 and 5 are performed on the combined spectrum, and the error coefficient with the original spectrum is calculated by the deviation variance method.

[0067] The experimental results are shown in Figure 2 The method can analyze the substance components of the solid mixed sample, and list the corresponding Raman signals of the substances, which are the Raman signals of malonic acid and succinic acid. In the figure, the characteristic peak positions of the solid mixed sample and the standard sample are matched, which indicates that the method can accurately identify the substance components, and directly shows the superposition process of the overlapping peaks in the mixed sample from multiple peaks. The Raman spectra are completely matched in peak position and peak number, and the substance analysis is accurate.

[0068] The embodiment also provides a multi-component identification system for analyzing the substance components of a complex system, which comprises:

[0069] The above merely describes the preferred embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can make equivalent replacements or changes to the technical solutions and the inventive concept of the present application within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for multicomponent analysis of complex system substances based on Raman signals, characterized by, The method comprises the following steps: S1, performing data processing on the collected Raman spectrum data to obtain collected spectrum signals and collected spectrum peak positions; S2, performing the data processing of S1 on known Raman spectrum data to obtain known spectrum signals and known spectrum peak positions, and establishing a known spectrum database; S3, performing peak position matching in the known spectrum database based on the collected spectrum peak positions, and screening known spectrum signals similar in peak position; S4, performing arrangement and combination on the similar known spectrum signals, and performing the data processing of S1 on spectrum data of each combination type, comparing the collected spectrum signals of S1 with possible spectrum signals, and performing multi-component identification; Specifically, the error coefficient of the two is calculated by the bias variance method; In S1, the data processing further comprises: performing wavelet transform on the collected Raman spectrum data to obtain WTf(a, τ) and extreme value solving, and obtaining the peak position; In S3, after the peak position matching in the known spectrum database based on the collected spectrum peak positions, the waveform of each matching peak is calculated, and possible substances are screened out according to the waveform similarity coefficient; wherein, the waveform similarity coefficient calculation formula is as follows: wherein f1k is the kth matching peak data of the known spectrum, is the matching peak mean value of the known spectrum, and f2k is the kth matching peak data of the collected spectrum, is the matching peak mean value of the collected spectrum; In S1, the data processing specifically comprises: filtering Gaussian white noise by using an S-G filter; correcting the baseline of the spectrum by using airPLS; and supplementing data to the required data point number by using a multiple spline interpolation method.

2. The method for multicomponent analysis of complex systems of matter based on Raman signals according to claim 1, characterized in that, The wavelet transform of the collected Raman spectrum data to WTf(a, τ) and the extreme value solving are specifically as follows: The wavelet transform of the collected Raman spectrum data to WTf(a, τ) is performed, and the extreme value solving is performed to obtain a minimum value X1 and a first maximum value X2; the extreme value solving is performed on the collected Raman spectrum data to obtain a second maximum value X3; X2 and X3 are compared, if the absolute value of the difference is less than a preset parameter, then the peak accurate value X4 is X3; the peak width PW = X1(i+1)-X1i, wherein X1i is the ith minimum value.

3. The method according to claim 1, wherein In S2, the known spectrum database comprises known spectrum signals, peak positions, peak widths, and peak strengths.

4. A multi-component identification system for analyzing the composition of a complex system material, characterized by, It comprises: a processor; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the programs comprise a method for performing multi-component analysis of complex system substances based on Raman signals according to any one of claims 1-3.

Images