Method for Improving Detection Accuracy of Principle of Matching between Characteristic Spectrum Lines of Light Source and Component Absorption

Abstract

The present invention discloses a method for improving the detection accuracy of the principle of matching between the characteristic spectrum of a light source and component absorption. The present invention includes the following steps: First, determine the main components of the substance to be measured, and then determine its molecular structure, functional group composition, and spectrum-structure according to the main components to obtain the characteristic absorption of the main components in the substance to be measured; Next, select a target light source according to the characteristic absorption of the main components in the substance to be measured; Finally, use the target light source to obtain the intensity spectrum of the substance to be measured, and input the intensity spectrum of the substance to be measured into the trained prediction model to output the content detection result of the substance to be measured. The present invention can improve the signal-to-noise ratio, enhance the detection sensitivity, and provide a higher detection accuracy than traditional spectral detection methods; In addition, the present invention is not limited to specific light sources and specific substances to be measured, and is applicable to light sources and substances to be measured that meet the matching conditions of characteristic spectrum-component characteristic absorption, and has strong universality.

Description

Technical Field

[0001] The present invention belongs to a method for detecting the content of substances in the field of spectroscopy, and particularly relates to a method for improving the detection accuracy based on the principle of matching the characteristic spectral lines of a light source with component absorption. Background Art

[0002] The field of spectroscopy is a scientific field that studies the interaction between substances (solids, liquids, gases) and radiation. The detection of substance content is one of the important research directions in the field of spectroscopy, which determines the composition or concentration of substances by analyzing the interaction of light with different wavelengths in substances. The detection of substance content is important in many fields. First, in environmental monitoring, the detection of substance content can help monitor harmful substances in the atmosphere, water bodies, and soil, enabling rapid and accurate detection of pollutants. Second, in the industrial production process, the detection of substance content can monitor the composition of raw materials, intermediate products, and final products in real time, ensuring product quality and production process stability. In addition, the detection of substance content also has important applications in the fields of medicine, food safety, agriculture, etc., helping to meet the requirements of drug quality control, food composition analysis, crop fertilization, etc.

[0003] Existing spectral detection methods for substance content mainly include visible / near-infrared spectroscopy, hyperspectral imaging, multispectral, fluorescence spectroscopy, Raman spectroscopy, terahertz spectroscopy, ultraspectral imaging, and fiber optic spectroscopy. Although these methods can all determine the content of unknown substances by comparing with the spectral characteristics of known substances and achieve the detection of substance content, almost all of these methods have limited accuracy in detecting substance content. The fundamental reason is that the key information characterizing the substance characteristics cannot be accurately captured. Summary of the Invention

[0004] In order to make up for the deficiency of the limited accuracy of existing spectral detection methods for substance content, the present invention proposes a method for improving the detection accuracy based on the principle of matching the characteristic spectral lines of a light source with component absorption. The present invention can break through the limitations of existing spectral detection methods for substance content, provide higher detection accuracy, and is not limited to specific light sources and specific substances to be measured, with strong universality.

[0005] The technical solution of the present invention is as follows:

[0006] I. A method for improving the detection accuracy based on the principle of matching the characteristic spectral lines of a light source with component absorption

[0007] S1: Determine the main components of the substance to be measured, and then determine its molecular structure, functional group composition, and spectrum-structure according to the main components, so as to obtain the characteristic absorption of the main components in the substance to be measured;

[0008] S2: Select a target light source according to the characteristic absorption of the main components in the substance to be measured;

[0009] S3: Obtain the intensity spectrum of the substance to be measured using the target light source, input the intensity spectrum of the substance to be measured into the trained prediction model, and output the content detection result of the substance to be measured.

[0010] Specifically, S1 is as follows:

[0011] S1.1: Determine the main components of the substance to be measured according to the existing substance component composition theory and detection methods, combine the molecular composition research of the compound and spectroscopy knowledge to determine the molecular structure and functional group composition corresponding to each main component, and use the spectral analysis method to determine the spectrum-structure of each main component;

[0012] S1.2: Determine the characteristic absorption of the main components in the substance to be measured by comparing the functional group composition, spectrum-structure with the known characteristic wavelengths.

[0013] Specifically, S2 is as follows:

[0014] S2.1: Use a spectrometer to measure the light source, obtain the intensity spectrum of the light source, and then determine the characteristic spectral lines of the light source;

[0015] S2.2: Under the action of the light source with characteristic spectral lines, use a spectrometer to collect the absorption spectra of the main components of the substance to be measured, and then determine the light source characteristic spectral line-characteristic absorption of the main components in the substance to be measured;

[0016] S2.3: Calculate the matching result between the characteristic absorption of the main components in the substance to be measured and the light source characteristic spectral line-characteristic absorption. If the matching result meets the conditions, use the current light source as the target light source; otherwise, adjust the light source according to the matching result until the matching result meets the conditions to obtain the target light source.

[0017] Specifically, S3 is as follows:

[0018] S3.1: Under the action of the light source with characteristic spectral lines, use a spectrometer to collect the intensity spectrum of the substance to be measured;

[0019] S3.2: Input the intensity spectrum of the substance to be measured into the trained prediction model, and output the content detection result of the substance to be measured.

[0020] The spectrometer is a spectrometer whose wavelength range can cover the wavelength range of the light source.

[0021] In S3.2, the prediction model is a model established by chemometrics, machine learning or deep learning methods.

[0022] II. A computer device

[0023] The computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the method are implemented.

[0024] III. A computer-readable storage medium

[0025] A computer program is stored on the medium, and when the computer program is executed by a processor, the steps of the method are implemented.

[0026] IV. A computer program product

[0027] The product includes a computer program / instructions, and when the computer program / instructions are executed by a processor, the steps of the method are implemented.

[0028] The beneficial effects of the present invention are as follows:

[0029] 1) The characteristic spectral line light source used in the present invention, which matches the characteristic absorption of the main components in the substance to be measured, can improve the signal-to-noise ratio and enhance the detection sensitivity.

[0030] 2) The present invention can break through the limitations existing in the existing spectral detection methods for substance content and provide higher detection accuracy.

[0031] 3) The present invention is not limited to a specific light source and a specific substance to be measured, and is applicable to light sources and substances to be measured that meet the condition of characteristic spectral line-component characteristic absorption matching, with strong universality. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 is a flowchart of the present invention.

[0033] Figure 2 is a molecular structure diagram of the main components glucose, fructose, and sucrose of the sugar content of the substance to be measured in the embodiment of the present invention.

[0034] Figure 3 is a functional group composition and spectrum-structure diagram of the main components glucose, fructose, and sucrose of the sugar content of the substance to be measured in the embodiment of the present invention.

[0035] Figure 4 is the intensity spectrum of the xenon lamp used as the light source in the embodiment of the present invention.

[0036] Figure 5 is the absorption spectrum of the main components glucose, fructose, and sucrose of the sugar content of the substance to be measured under the action of the xenon lamp used as the light source in the embodiment of the present invention.

[0037] Figure 6 is the sugar content intensity spectrum of cherry tomatoes used in the embodiment of the present invention.

[0038] Figure 7 is the detection result of the PLSR model for the sugar content of cherry tomatoes using the traditional spectral detection method in the embodiment of the present invention.

[0039] Figure 8It is the detection result of the PLSR model for the fructose content of cherry tomatoes using this method in the embodiments of the present invention.

[0040] Figure 9 It is the detection result of the FNN model for the fructose content of cherry tomatoes using the traditional spectral detection method in the embodiments of the present invention.

[0041] Figure 10 It is the detection result of the FNN model for the fructose content of cherry tomatoes using this method in the embodiments of the present invention.

[0042] Figure 11 It is the detection result of the CNN model for the fructose content of cherry tomatoes using the traditional spectral detection method in the embodiments of the present invention.

[0043] Figure 12 It is the detection result of the CNN model for the fructose content of cherry tomatoes using this method in the embodiments of the present invention. Detailed implementation manners

[0044] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] As Figure 1 shown, the present invention includes the following steps:

[0046] S1: Determine the fructose content of the substance to be measured, and then determine its molecular structure, functional group composition, and spectrum-structure according to the main components to obtain the characteristic absorptions of glucose, fructose, and sucrose in the fructose content; the main components are usually components with a content higher than 10% in the total amount of the substance.

[0047] Specifically, S1 is as follows:

[0048] S1.1: Determine glucose, fructose, and sucrose in the substance to be measured according to the existing theories and detection methods of substance component compositions, and determine the molecular structures and functional group compositions corresponding to each main component in combination with the molecular composition research and spectroscopy knowledge of glucose, fructose, and sucrose, and use spectroscopic analysis methods to determine the spectrum-structures of glucose, fructose, and sucrose;

[0049] S1.2: Determine the characteristic absorptions of glucose, fructose, and sucrose in the substance to be measured by comparing the functional group composition and spectrum-structure with their known characteristic wavelengths.

[0050] S2: Select a target light source according to the characteristic absorptions of glucose, fructose, and sucrose in the substance to be measured;

[0051] Specifically, S2 is as follows:

[0052] S2.1: The spectrometer is a normally operating spectrometer whose wavelength range can cover the wavelength range of the light source. In a specific implementation, a Maya2000 Pro spectrometer is used to measure the light source, obtaining the intensity spectrum of the light source, and then determining the characteristic spectral lines of the light source. The characteristic wavelengths or peaks in the intensity spectrum are the characteristic spectral lines;

[0053] S2.2: Under the action of a xenon lamp with characteristic spectral lines, the absorption spectra of the substances to be measured, namely glucose, fructose, and sucrose, are collected using a spectrometer, and then the characteristic absorption of the light source spectral lines - characteristic absorption of glucose, fructose, and sucrose in the substances to be measured is determined by analyzing the characteristic peaks in the absorption spectra;

[0054] S2.3: Calculate the matching result between the characteristic absorption of glucose, fructose, and sucrose in the substances to be measured and the characteristic absorption of the corresponding light source spectral lines - characteristic absorption. If the matching result meets the condition, that is, the matching result is greater than or equal to the preset matching threshold, which is the matching rate of the characteristic absorption of the main components of the substances to be measured and the characteristic absorption of the light source spectral lines - characteristic absorption. In a specific implementation, the preset matching threshold is 50%. Then the current light source is used as the target light source. Otherwise, the light source is adjusted according to the matching result until the matching result meets the condition to obtain the target light source.

[0055] S3: Use the target light source to obtain the intensity spectrum of the substance to be measured, and input the intensity spectrum of the substance to be measured into the trained prediction model to output the detection result of the sugar content of the substance to be measured.

[0056] S3 specifically is:

[0057] S3.1: Under the action of a light source with characteristic spectral lines, the intensity spectrum of the substance to be measured is collected using a spectrometer;

[0058] S3.2: Input the intensity spectrum of the substance to be measured into the trained prediction model to output the detection result of the sugar content of the substance to be measured.

[0059] The prediction model is a model established by chemometrics, machine learning, or deep learning methods.

[0060] Among them, the specific process of establishing a model by chemometrics is as follows:

[0061] Under the action of a xenon lamp with characteristic spectral lines, the intensity spectra of the fructose content of 208 cherry tomatoes were collected using a Maya2000 Pro spectrometer, and the content values of the fructose content were obtained using a PAL-1 refractometer by refractometry. Prediction models of the fructose content under chemometrics, machine learning, and deep learning were established by combining the partial least squares regression (PLSR) method, feedforward neural network (FNN), and convolutional neural network (CNN), respectively.

[0062] The present invention is applied to the detection of the fructose content of cherry tomatoes, and the specific results are as follows:

[0063] The cherry tomatoes used were of the "Millennium" variety, purchased from a local supermarket in Hangzhou, Zhejiang Province, China, with a total of 208 samples. All the samples were mature fruits, without damage and with a uniform appearance. Their diameter range was 25.82 - 32.66 mm, and the weight range was 11.22 - 19.81 g. The spectral acquisition mode was the transmission mode, the light source was a xenon lamp, and a Maya2000 Pro spectrometer was used to collect spectral data. The spectral band range was 650 - 1100 nm. Prediction models were established for the sample set using the PLSR, FNN, and CNN methods, respectively.

[0064] The characteristic absorptions of glucose, fructose, and sucrose mainly concentrated at 762 nm, 767 nm, 773 nm, 930 nm, 996 nm, 1004 nm, and 1065 nm, as Figure 3 shown. Figure 4 is the intensity spectrum of the xenon lamp as the light source, and its characteristic spectral lines mainly concentrated at 768 nm, 828 nm, 887 nm, 898 - 921 nm, 942 - 954 nm, 982 - 996 nm, and 1016 nm. Under the action of the xenon lamp as the light source, obvious characteristic absorption peaks appeared near 923 nm and 1032 nm in the collected glucose spectrum, near 920 nm and 1000 nm in the collected fructose spectrum, and near 920 nm and 985 nm in the collected sucrose spectrum, as Figure 5 shown, that is, the xenon lamp characteristic spectral lines - characteristic absorptions of glucose, fructose, and sucrose were located at 923 nm and 1032 nm, 757 nm, 920 nm and 1000 nm, and 920 nm and 985 nm, respectively. By comparing the characteristic absorptions of glucose, fructose, and sucrose with the xenon lamp characteristic spectral lines - characteristic absorptions, it can be seen that the two were close near 760 nm, 920 nm, 990 nm, and 1000 nm, indicating that the matching of the two characteristic absorptions was good. Figure 6It is the sugar degree intensity spectrum of cherry tomatoes. It can be seen that its characteristic absorption is quite similar to the characteristic spectral lines of the xenon lamp light source.

[0065] The determination coefficient of the calibration set of the PLSR prediction model established for the sugar degree of cherry tomatoes using the traditional spectral detection method is 0.9401, the root mean square error of the calibration set RMSEC is 0.2617%, the determination coefficient of the prediction set is 0.8127, the root mean square error of the prediction set RMSEP is 0.3114%, and the residual prediction deviation RPD is 2.2832, as Figure 7 shown; while the determination coefficient of the calibration set of the PLSR prediction model established for the sugar degree of cherry tomatoes by the method proposed in the present invention is 0.9595, the root mean square error of the calibration set RMSEC is 0.2126%, the determination coefficient of the prediction set is 0.8960, the root mean square error of the prediction set RMSEP is 0.2489%, and the residual prediction deviation RPD is 3.0631, as Figure 8 shown.

[0066] The determination coefficient of the calibration set of the FNN prediction model established for the sugar degree of cherry tomatoes using the traditional spectral detection method is 0.9349, the root mean square error of the calibration set RMSEC is 0.2609%, the determination coefficient of the prediction set is 0.6867, the root mean square error of the prediction set RMSEP is 0.3766%, and the residual prediction deviation RPD is 1.8082, as Figure 9 shown; while the determination coefficient of the calibration set of the FNN prediction model established for the sugar degree of cherry tomatoes by the method proposed in the present invention is 0.9617, the root mean square error of the calibration set RMSEC is 0.2032%, the determination coefficient of the prediction set is 0.8147, the root mean square error of the prediction set RMSEP is 0.3019%, and the residual prediction deviation RPD is 2.3514, as Figure 10 shown.

[0067] The determination coefficient of the calibration set of the CNN prediction model established for the sugar degree of cherry tomatoes using the traditional spectral detection method is 0.9347, the root mean square error of the calibration set RMSEC is 0.2731%, the determination coefficient of the prediction set is 0.7569, the root mean square error of the prediction set RMSEP is 0.3505%, and the residual prediction deviation RPD is 2.0282, as Figure 11 shown; while the determination coefficient of the calibration set of the CNN prediction model established for the sugar degree of cherry tomatoes by the method proposed in the present invention is 0.9089, the root mean square error of calibration set RMSEC is 0.3187%, and the determination coefficient of prediction set is 0.8302, the root mean square error of prediction set RMSEP is 0.3142%, and the residual prediction deviation RPD is 2.4265, as Figure 12 shown.

[0068] It can be seen from this embodiment that the performance and detection accuracy provided by the three prediction models established by the method proposed by the present invention for the fructose content of cherry tomatoes are better than those of the traditional spectral detection method, which reflects that the present invention has more excellent capabilities in the detection of substance content.

[0069] In summary, the method for improving the detection accuracy based on the principle of matching the characteristic spectral lines of the light source and the component absorption proposed by the present invention can break through the limitations of the existing spectral detection methods for substance content, provide higher detection accuracy, and is not limited to specific light sources and specific substances to be measured, with strong universality.

[0070] It should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of them. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for improving detection accuracy based on the principle of light source characteristic spectrum line-component absorption matching, characterized in that: The following steps are involved: S1: Determine the main components of the substance to be tested, and then determine its molecular structure, functional group composition and spectrum-structure based on the main components to obtain the characteristic absorption of the main components in the substance to be tested; S2: Select the target light source according to the characteristic absorption of the main components in the substance to be tested; The S2 is specifically: S2.1: Use a spectrometer to measure the light source, obtain the intensity spectrum of the light source, and then determine the characteristic spectrum line of the light source; S2.2: Under the action of a light source with characteristic spectral lines, use a spectrometer to collect the absorption spectra of the main components of the substance to be tested, and then determine the light source characteristic spectral lines - characteristic absorption of the main components in the substance to be tested; S2.3: Calculate the matching result between the characteristic absorption of the main components in the substance to be tested and the characteristic spectrum line-characteristic absorption of the light source. If the matching result meets the conditions, the current light source is used as the target light source. Otherwise, the light source is adjusted according to the matching result until the matching result meets the conditions and the target light source is obtained. S3: Use the target light source to obtain the intensity spectrum of the substance to be tested, input the intensity spectrum of the substance to be tested into the trained prediction model, and output the content detection result of the substance to be tested.

2. The method for improving detection accuracy based on the principle of light source characteristic spectrum line-component absorption matching according to claim 1, characterized in that: The S1 is specifically: S1.1: Determine the main components of the substance to be tested based on the existing material component composition theory and detection methods, determine the molecular structure and functional group composition of each main component by combining the molecular composition research of the compound with spectroscopy knowledge, and determine the spectrum-structure of each main component by using spectral analysis methods; S1.2: Determine the characteristic absorption of the main components in the substance to be tested by comparing the functional group composition, spectrum-structure and known characteristic wavelength.

3. The method for improving detection accuracy based on the principle of light source characteristic spectrum line-component absorption matching according to claim 1, characterized in that: The S3 is specifically: S3.1: The intensity spectrum of the substance to be tested is collected by using a spectrometer under the action of a light source with characteristic spectral lines; S3.2: Input the intensity spectrum of the substance to be tested into the trained prediction model, and output the content detection result of the substance to be tested.

4. The method for improving detection accuracy based on the principle of light source characteristic spectrum line-component absorption matching according to claim 3, characterized in that: The spectrometer is a spectrometer whose wavelength range can cover the wavelength range of the light source.

5. The method for improving detection accuracy based on the principle of light source characteristic spectrum line-component absorption matching according to claim 1, characterized in that: In S3.2, the prediction model is a model established by chemometrics, machine learning or deep learning methods.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images