Method for measuring concentration of mixture solution based on principal component analysis

Abstract

On the basis of a spectral linear superposition theory and a mathematical analysis method, spectral decomposition and reconstruction of the low-concentration mixture solution are realized by extracting a characteristic Raman spectrum of a single solute solution, and a measurement method capable of being used for rapidly detecting the low-concentration mixture solution is obtained. The method comprises the following steps: firstly, decomposing and reconstructing Raman spectrums of single-solute solutions with different concentrations, and extracting a solute Raman signal and a characteristic spectrum of background noise; performing orthogonalization treatment on the characteristic spectra, and performing projection decomposition and group transformation on a measurement spectrum of the low-concentration mixture solution to obtain an intensity coefficient of a solute spectrum; and finally, obtaining the concentration information of each solute according to the linear relationship between the intensity coefficient of the solute spectrum and the concentration. The method can effectively overcome the dependence of a traditional method on the solute spectrum characteristic peak, and can still play a role even when the solute concentration is low and the characteristic peak is submerged by background noise.

Description

(1) Technical field

[0001] The present invention relates to a method for measuring the concentration of a mixture solution based on principal component analysis, which can be used to perform component analysis and concentration measurement on a mixture solution of unknown components, and belongs to the technical field of Raman spectral analysis and spectral imaging detection. (2) Background technology

[0002] Raman spectroscopy is based on the Raman scattering effect. Using a laser of a fixed wavelength as a light source, it analyzes the spectrum of scattered light at frequencies different from the incident light, revealing the vibrational and rotational characteristics of molecular bonds, thereby analyzing the composition of a substance. Raman spectroscopy allows for simultaneous detection of all components in a sample using a single laser source, resulting in rapid detection, high sensitivity, and, due to its non-contact measurement principle, maintenance-free on-site measurements. Molecular bond rotation varies across different substances, and these differences are reflected in the Raman spectrum.

[0003] Principal component analysis (PCA) is an effective data simplification method that transforms original correlated variables into linearly independent principal component variables through orthogonal transformation. This method projects the data into a new feature space, where the direction of the first principal component corresponds to the largest variance, the direction of the second principal component corresponds to the second largest variance, and so on. As a widely used dimensionality reduction tool, PCA can optimize the data structure by extracting orthogonal eigenvalues ​​while maximally preserving the original information. Therefore, PCA is applied to Raman spectroscopy, decomposing a set of Raman spectral data into different principal components and studying the principal components and their principal components. As an optimal orthogonal transformation method based on the minimum mean square error criterion, PCA transforms the original high-dimensional data into an uncorrelated low-dimensional representation through linear projection. The core of this method lies in finding a new orthogonal basis that maximizes the data variance by rotating the coordinate axes. This effectively extracts key eigenvalues ​​and achieves dimensionality reduction while eliminating linear correlations in the original data.

[0004] Based on the above background technology, the present invention proposes a method for measuring the concentration of a mixture solution based on principal component analysis. By performing principal component analysis and deorthogonalization on methanol solutions and ethanol solutions with different concentration gradients, a linear relationship between the concentration gradients of methanol and ethanol and the principal component intensity coefficients is established. The Raman spectral data of the three substances, water, ethanol, and methanol, are determined by principal component analysis and deorthogonalization, and combined to form a new basis vector. The Raman spectra of methanol-ethanol mixture solutions with different concentrations and proportions are measured, and the Raman spectra of the mixture solutions are projected onto the basis vectors to obtain the principal component intensity coefficients of methanol and ethanol in the mixture solutions. The concentrations of methanol and ethanol in the mixture solutions are confirmed based on the linear relationship between the principal component intensity coefficients and the concentrations. This paper successfully realizes the extraction of Raman signals from mixture solutions and establishes the relationship between the concentration and principal component intensity coefficients between mixture solutions and single substance solutions. After the data model is established, the concentrations of the components of the mixture solution can be quickly confirmed based on a single spectral measurement. (3) Summary of the invention

[0005] The purpose of the present invention is to provide a method for measuring the concentration of a mixture solution based on principal component analysis. By performing principal component analysis on the spectral data of the mixture solution, spectral signal interference is eliminated, and the concentration of each component in the mixture solution can be quickly and accurately measured.

[0006] To achieve the above objectives, the technical solutions adopted by the present invention are as follows:

[0007] Step S1: taking out samples of a single substance with different concentrations and placing them in a glass dish to be tested, and using a Raman spectrometer to collect Raman spectrum information of different groups of samples.

[0008] Step S2: performing principal component analysis on the Raman spectrum information of different groups of samples to obtain information on different material components of the spectra of different groups of samples, and performing corrections based on their respective spectral characteristics.

[0009] Step S3, combining the correction information with the information of different components to establish a linear model between the principal component score coefficient of the single substance component in the mixed solution of different concentrations and the solution concentration, performing principal component analysis on the corrected single substance component spectrum, and combining the new vectors after analysis to form the basic vector of the mixture solution projection.

[0010] In step S4, the Raman spectrum information measured for the mixture solution is projected onto the basis vector to obtain the principal component score coefficient of each component in each group of mixture solutions on the basis vector. The principal component score coefficient of each component in each group of mixture solutions on the single substance component is obtained through the principal component decomposition relationship. The constructed linear model is used to detect the mixed solution of unknown concentration to quickly determine the current concentration of different components.

[0011] In order to better and more conveniently predict the concentration of the mixture solution, the measured sample spectrum is directly projected onto the orthogonal basis vector to reduce the number of operation steps.

[0012] Each set of mixture solution spectra is regarded as a linear superposition of spectra of different material components, and the separation and reconstruction of different material components are completed through principal component analysis.

[0013] After obtaining the Raman signals of different components of substances through the Raman spectrometer, the characteristic peaks of the Raman signals are compared with those of the standard signals for reference. The measured Raman signals (i.e., the principal components separated after principal component analysis) are corrected according to the spectral characteristics of the standard signals of different material components, and substituted into the original principal component analysis expression to obtain the corrected principal component score coefficient. The corrected information of different components is combined with the concentration to construct a linear model. (IV) Description of the accompanying drawings

[0014] Figure 1 The present invention provides a flow chart of a method for measuring the concentration of a mixture solution based on principal component analysis.

[0015] Figure 2 Flowchart for establishing the linear relationship between the principal component coefficients and ethanol concentration. (a) shows six sets of ethanol spectra at different concentrations, (b) shows the first principal component (water) after PCA, (c) shows the second principal component (ethanol) after PCA, (d) shows the first principal component after reconstruction, (e) shows the second principal component after reconstruction, and (f) shows the linear relationship between the principal component coefficients and ethanol concentration.

[0016] Figure 3 The linear fitting relationship between methanol and ethanol concentrations and the reconstructed principal components is shown in Figure 2. (a) is the methanol fitting relationship and (b) is the ethanol fitting relationship.

[0017] Figure 4 The linear relationships between the concentrations of methanol and ethanol in eight solutions are shown in Figure 2. (a) is the fitted relationship for methanol, and (b) is the fitted relationship for ethanol. (V) Specific implementation methods

[0018] The technical solution of the present invention will be further described below with reference to specific embodiments:

[0019] The experimental instrument used in this experiment used an excitation light wavelength of 785 nm, an optical power of approximately 800 mW, and an integration time of 2 s for each spectrum. The reagents used were methanol (analytical grade, Tianjin Damao Chemical Reagent Factory), ethanol (analytical grade, Sichuan Xilong Technology Co., Ltd.), and distilled water. The actual concentration was confirmed by preparing a solution of the desired concentration.

[0020] In step S1, a set of ethanol samples with concentrations of 15%, 10%, 5%, 2%, 1%, and 0.5% are prepared by mixing water and anhydrous ethanol. The samples of different concentrations are placed in a glass dish to be tested, and Raman spectra of the different sets of samples are collected using a Raman spectrometer.

[0021] Step S2, performing principal component analysis on the Raman spectrum information of different groups of samples to obtain information on different material components of the spectra of different groups of samples, and correcting them according to their respective spectral characteristics. Principal component analysis is performed on the Raman spectrum measurement results of the six solutions to obtain six principal components. Among the six principal components of the spectrum, only the first principal component and the second principal component contain valid information. The first principal component mainly corresponds to the background noise of water, the second principal component mainly corresponds to the Raman spectrum of ethanol, and the other third to sixth principal components correspond to random noise during measurement. Due to the influence of the strict orthogonality of the principal components of the spectrum, although the second principal component corresponds to the Raman spectrum of ethanol, it is also mixed with the background noise of water, resulting in baseline bending (such as Figure 2 (c)). In order to further correct the influence of the orthogonality of the principal components, the first principal component and the second principal component are deorthogonalized to obtain the reconstructed water background noise ( Figure 2 (d)) and ethanol characteristic Raman spectrum ( Figure 2 (e)). The first principal component represents the Raman spectrum of water, i.e. Figure 2 (b), the background noise after deorthogonalization is as follows Figure 2 (d) The ethanol concentration of the measured solution is low, and the characteristic peak of the measured spectrum is submerged in the background noise. The strong background noise is separated from the weak spectral signal, thereby extracting the characteristic spectrum of the low-concentration solution; the second principal component corresponds to the Raman spectrum of ethanol, that is, Figure 2 (c), ethanol after deorthogonalization Figure 2 (e); The third to sixth principal components correspond to random noise.

[0022] Step S3, combining the correction information with the information of different components to establish a linear model between the principal component score coefficient of the single substance component in the mixed solution of different concentrations and the solution concentration, and performing principal component analysis on the corrected single substance component spectrum. The new vector combination after analysis forms the basic vector of the mixture solution projection. The measured spectra of the six groups of ethanol solutions with different concentrations are decomposed into a linear superposition of water background noise, ethanol characteristic spectrum and random noise, and the intensity coefficient of the ethanol characteristic spectrum is linearly fitted with its concentration to obtain a monotonic function relationship between the two. The fitting results are shown as follows: Figure 3 (b) The same method can also be used to obtain the characteristic Raman spectrum of methanol, as well as the linear relationship between the intensity coefficient and concentration ( Figure 3 (a)).

[0023] Step S4, by projecting the Raman spectrum information measured from the mixture solution onto the basis vector, the principal component score coefficient of each component in each group of mixture solutions on the basis vector can be obtained, and the principal component score coefficient of each component in each group of mixture solutions on the single substance component can be obtained through the principal component decomposition relationship. The mixed solution of unknown concentration is detected using the constructed linear model to quickly determine the current concentrations of different components. A group of ethanol samples with the same volume are prepared at room temperature, and then different volumes of methanol solution and aqueous solution are added to the ethanol samples to form 8 groups of mixture solutions with different concentrations. A linear relationship is established between the concentration calculated using the principal component coefficient obtained by projecting onto the basis vector and the actual concentration. The linear relationship between the two concentrations of methanol and ethanol in the eight groups of solutions is as follows: Figure 4 .

[0024] The above embodiments are intended only to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to specific embodiments, those skilled in the art will appreciate that modifications or equivalent substitutions may be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and such modifications and equivalent substitutions are intended to be encompassed by the claims of the present invention. Any techniques, shapes, and structural components not described in detail herein are generally known.

Claims

1. A method for measuring the concentration of a mixture solution based on principal component analysis, characterized in that The steps include: Step S1: taking out samples of a single substance with different concentrations and placing them in a glass dish to be tested, and using a Raman spectrometer to collect Raman spectrum information of different groups of samples. Step S2: performing principal component analysis on the Raman spectrum information of different groups of samples to obtain information on different material components of the spectra of different groups of samples, and performing corrections based on their respective spectral characteristics. Step S3, combining the correction information with the information of different components to establish a linear model between the principal component score coefficient of the single substance component in the mixed solution of different concentrations and the solution concentration, performing principal component analysis on the corrected single substance component spectrum, and combining the new vectors after analysis to form the basic vector of the mixture solution projection. In step S4, the Raman spectrum information measured for the mixture solution is projected onto the basis vector to obtain the principal component score coefficient of each component in each group of mixture solutions on the basis vector. The principal component score coefficient of each component in each group of mixture solutions on the single substance component is obtained through the principal component decomposition relationship. The constructed linear model is used to detect the mixed solution of unknown concentration to quickly determine the current concentration of different components.

2. The method for measuring the concentration of a mixture solution based on principal component analysis according to claim 1, wherein: In order to better and more conveniently predict the concentration of the mixture solution, the measured sample spectrum is directly projected onto the orthogonal basis vector to reduce the number of operation steps.

3. The method for measuring the concentration of a mixture solution based on principal component analysis according to claim 1, wherein: Each set of mixture solution spectra is regarded as a linear superposition of spectra of different material components, and the separation and reconstruction of different material components are completed through principal component analysis.

4. The method for measuring the concentration of a mixture solution based on principal component analysis according to claim 1, wherein: After obtaining the Raman signals of different components of substances through the Raman spectrometer, the characteristic peaks of the Raman signals are compared with those of the standard signals for reference. The measured Raman signals (i.e., the principal components separated after principal component analysis) are corrected according to the spectral characteristics of the standard signals of different material components, and substituted into the original principal component analysis expression to obtain the corrected principal component score coefficient. The corrected information of different components is combined with the concentration to construct a linear model.

Images