Method for detecting contents of total sugar, crude protein and total flavone in folium cortex eucommiae by near infrared spectroscopy
Through near-infrared spectroscopy technology and modeling methods, the problem of time-consuming and complex operation of the components of Eucommia ulmoidea leaves in the prior art is solved, and a fast, lossless and accurate detection effect is achieved.
Patent Information
- Application Number
- CN202311772391.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-21
- Publication Date
- 2025-06-24
AI Technical Summary
The prior art is difficult to quickly and non-destructively detect the content of total sugar, crude protein and total flavonoids in Eucommia ulmoide leaves, and the detection process takes a long time and is complex in operation.
Using near-infrared spectroscopy technology, quantitative analysis model of crude protein, total sugar and total flavonoids of Eucommia ulmoide leaves was established through the combination of spectral pretreatment, characteristic band selection and different modeling methods.
The rapid, non-destructive and accurate detection of the content of crude protein, total sugar and total flavonoids in Eucommia ulmoide leaves is achieved, which simplifies the detection process and improves the detection efficiency and accuracy.
Smart Images
Figure BDA0004621568900000031 
Figure BDA0004621568900000032 
Figure BDA0004621568900000041
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of non-destructive testing, and specifically relates to a method for rapidly and non-destructively detecting the contents of total sugar, crude protein, and total flavonoids in Eucommia ulmoides leaves by using near-infrared spectroscopy technology. Background Art
[0002] Eucommia ulmoides Oliver leaves contain rich active ingredients and nutritional components, and are widely used in the food, medicinal material, and feed industries. At present, the determination of active ingredients in Eucommia ulmoides leaves is time-consuming and complex. Establishing a rapid and non-destructive analysis method for Eucommia ulmoides leaves based on near-infrared technology, which is green, environmentally friendly, efficient, accurate, and low-cost, can improve the technical level of the Eucommia ulmoides industry. Summary of the Invention
[0003] The purpose of the present invention is to provide a method for rapidly detecting the components of Eucommia ulmoides leaves by near-infrared spectroscopy, which can rapidly detect the contents of total sugar, total flavonoids, and crude protein in raspberry, realize the rapid identification of the quality of Eucommia ulmoides leaves, with strong purpose and being rapid and efficient. The method established by the present invention includes the following steps:
[0004] ⑴ A total of 175 Eucommia ulmoides leaf samples were collected. The crude protein content in Eucommia ulmoides leaves was determined by the Kjeldahl method, the total sugar content in Eucommia ulmoides leaves was determined by the anthrone-sulfuric acid method, and the total flavonoid content in Eucommia ulmoides leaves was determined by the sodium nitrite-aluminum nitrate colorimetric method;
[0005] ⑵ The near-infrared spectra of 175 Eucommia ulmoides leaf samples were collected, and spectral preprocessing, characteristic band optimization were carried out, and quantitative analysis models for crude protein, total flavonoids, and total sugar in Eucommia ulmoides were established and optimized by using different modeling methods.
[0006] ⑶ The near-infrared spectrum of an unknown Eucommia ulmoides leaf sample was imported into the model to detect the accuracy of the model.
[0007] The Eucommia ulmoides leaf samples described in step ⑴ were collected from 6 varieties (Qinzhong No. 1, Qinzhong No. 2, Qinzhong No. 3, Qinzhong No. 4, Huazhong No. 5, and Huazhong No. 8) in 3 seasons (spring, summer, and autumn) from the same domestic production area (Lueyang County, Hanzhong City, Shaanxi Province). To ensure the accuracy and anti-interference ability of the model, the content range of the target component in the samples for establishing the calibration model can cover the content range of this component in the samples to be analyzed in the future. The present invention collected Eucommia ulmoides leaf samples of six varieties and three seasons in the country, which can fully represent all the background information of the samples.
[0008] The scanning range of the near-infrared spectrum of the sample in step ⑵ is 12,000–4,000 cm -1; The preprocessing method of the near-infrared spectrum is selected from one or more of the following methods: smoothing, baseline correction, and normalization. The band selection method of the near-infrared spectrum is selected from one or more of the following methods: comparison with reference literature, comparison with reference standards, and genetic algorithm. The modeling method is selected from one of the following methods: partial least squares method, principal component regression method, random forest, decision tree, BP neural network, and support vector machine.
[0009] In the present invention, the basis for judging the model accuracy described in step (3) is that when RPD < 1.4, that is, R 2 is less than 0.4898, the established model is unreliable; when RPD > 2.0, that is, R 2 is greater than 0.75, the established model has relatively high reliability and can be used for routine analysis; when R 2 is greater than 0.88, it can be applied to practical applications.
[0010] In this study, the Kjeldahl method was used to measure the crude protein content of Eucommia ulmoides leaves of 6 varieties including Qinzhong No. 1, 2, 3, 4, Huazhong No. 5 and 8 in spring, summer and autumn seasons. A total of 175 samples showed that the crude protein content in autumn was 9.35%-13.64%
[0011] (w / w), 16.73%-24.40% in spring, and 9.25%-13.98% in summer; the anthrone method was used to measure the total sugar content of Eucommia ulmoides leaves, which was 8.21%-21.35% in autumn, 5.99%-20.72% in spring, and 13.39%-24.49% in summer; the sodium nitrite method was used to measure the total flavonoid content of Eucommia ulmoides leaves, which was 5.91%-12.61% in autumn, 4.06%-8.47% in spring, and 7.33%-14.24% in summer. Based on the near-infrared spectrum data and the chemical determination values, through data optimization, optimal band selection, and comparison of modeling methods, quantitative models for crude protein, total sugar, and total flavonoids in Eucommia ulmoides leaves were established. After preprocessing the spectral data by spectral smoothing, baseline correction, and normalization, the modeling bands 5203.27 - 4751.98 cm -1 and 7000.69 - 6499.26 cm -1 were selected. The PCR method was used to establish a crude protein quantitative model, and the correlation coefficient R 2 = 0.9897; after preprocessing by smoothing, baseline correction, and normalization, the genetic algorithm was used to select the modeling band, and the PLS method was used to establish a total flavonoid quantitative model, and the correlation coefficient R 2 = 0.9571; after preprocessing by smoothing, baseline correction, and normalization, the genetic algorithm was used to select the modeling band, and the random forest algorithm was used to establish a total sugar quantitative model, and the correlation coefficient R 2 = 0.8913.
[0012] The present invention can meet the actual production needs, thereby providing technical support and theoretical basis for the precise production and management of Eucommia ulmoides leaves for leaf use.
[0013] The present invention provides a brand-new method for rapid, non-destructive and random detection of Eucommia ulmoides leaves, thereby providing technical support and theoretical basis for the precise production, management and cultivation of Eucommia ulmoides leaves. Brief Description of the Drawings
[0014] Figure 1 is a flow schematic diagram of this detection method;
[0015] Figure 2a is the near-infrared spectrogram of 175 Eucommia ulmoides leaves;
[0016] Figure 2b is the near-infrared spectrogram of 175 Eucommia ulmoides leaves after pretreatment;
[0017] Figure 3 is the comparison between the predicted value and the actual value of the verification set of the crude protein quantitative analysis model
[0018] Figure 4a Near-infrared spectrogram of the modeling band selected by the genetic algorithm for the total sugar quantitative analysis model
[0019] Figure 4b is the comparison between the predicted value and the actual value of the verification set of the total sugar quantitative analysis model
[0020] Figure 5a Near-infrared spectrogram of the modeling band selected by the genetic algorithm for the total flavonoid quantitative analysis model
[0021] Figure 5b is the comparison between the predicted value and the actual value of the verification set of the total flavonoid quantitative analysis model Detailed Embodiments
[0022] The present invention optimizes and establishes a quantitative analysis model of Eucommia ulmoides leaves through spectral pretreatment, selection of reference with standard products and selection of modeling bands by genetic algorithm, as well as different modeling methods. This research provides theoretical and technical references for the rapid detection of the contents of crude protein, total sugar and total flavonoids in Eucommia ulmoides, and has important significance for the industrialized development of Eucommia ulmoides leaves.
[0023] The establishment and application of the near-infrared spectral model described in the present invention are as shown in the attached Figure 1 , specifically as follows:
[0024] 1. Instrument conditions and sample treatment:
[0025] In Hanfeng Village, Jinjiahe Town, Lueyang County, Hanzhong City, Shaanxi Province, ten Eucommia ulmoides trees with strong growth trends and varieties of Qinzhong No. 1, Qinzhong No. 2, Qinzhong No. 3, Qinzhong No. 4, Huazhong No. 5, and Huazhong No. 8 were selected, numbered 1-10 respectively, with two trees spaced between each two numbers. In November 2021 (autumn), March 2022 (spring), and July 2022 (summer), a number of tree tip leaves were picked, totaling 175 Eucommia ulmoides leaves. They were taken back to the laboratory on the same day, dried, crushed, passed through a 40-mesh sieve, and stored in a -20°C refrigerator for later use.
[0026] Instrument: The instrument used was the MPA II type multifunctional Fourier transform near-infrared spectrometer of BRUKER Company in Germany. The Eucommia ulmoides sample powder passed through a 40-mesh sieve was poured into a quartz sample cup, evenly spread out, and near-infrared spectral images were collected using an integrating sphere diffuse reflection with air as the reference. The collection conditions were a resolution of 8 cm -1 , and the spectral collection band was 12,000 - 4,000 cm -1 . The number of scans was 64 times. The temperature was (25 ± 0.5)°C, and the relative humidity was 35% - 45%. OPUS 6.0 software was used to collect spectral data. Each Eucommia ulmoides sample was collected 3 times repeatedly to obtain the average spectrum. As shown in the attachment Figure 2a , it is the original data of the near-infrared spectrum of Eucommia ulmoides leaves.
[0027] 2. Pretreatment of near-infrared spectral images
[0028] As Figure 2b shown, it is the near-infrared spectral image after smoothing, baseline correction, and normalization pretreatment.
[0029] 3. Reasonable division of the calibration set and validation set of the crude protein model samples
[0030] For the 175 Eucommia ulmoides samples, the calibration set and validation set samples were set to 140 and 35 respectively. The built-in function Randperm in MATLAB software was used to generate random numbers to disrupt the inherent order of the samples themselves and enhance the applicability of the model. The number of samples in the calibration set and validation set for each variety, the maximum value, minimum value, average value, and relative standard deviation (RSD) of the crude protein content are shown in Table 1-1. The modeling requirement is that the sample concentration in the validation set is included in the calibration set concentration. In this experiment, the sample concentration of the validation set randomly selected by MATLAB software is included in the calibration set concentration, meeting the modeling requirements.
[0031] Table 1-1 Division of the calibration set and validation set samples of the crude protein model
[0032]
[0033] 4. Establishment of the crude protein quantitative analysis model
[0034] As shown in Table 1-2, the quantitative analysis model R established based on the crude protein content values of Eucommia ulmoides leaves determined by near-infrared spectroscopy data and the Kjeldahl method 2 values, respectively studied and compared the effects of different modeling methods, spectral preprocessing, and modeling bands on model R 2 Finally, a quantitative analysis model for crude protein was established. The modeling conditions were as follows: baseline correction and smoothing preprocessing were adopted, and the modeling bands were selected as 7000.68 - 6499.26 cm -1 and 5203.27 - 4751.98 cm -1 . The quantitative analysis model for crude protein was established by PCR. The relative deviations between the actual values and predicted values of the validation set samples were all less than 0.1%, and the absolute deviations were all between -1% and 1%, indicating that the established model had good performance and accurate prediction results
[0035] Table 1-2 Model R values based on different modeling methods and bands 2 values
[0036]
[0037] 5. Reasonably divide the calibration set and validation set of the total sugar model samples
[0038] For 175 Eucommia ulmoides samples, 140 and 35 samples were set as the calibration set and validation set respectively. The built-in function Randperm in MATLAB software was used to generate random numbers to disrupt the inherent order of the samples and enhance the applicability of the model. The numbers of the calibration set and validation set for each variety, the maximum, minimum, average values, and relative standard deviation (RSD) of the total sugar content are shown in Table 1-3. The modeling requires that the concentration of the validation set samples be included in the concentration of the calibration set. In this experiment, the concentration of the validation set samples randomly selected by MATLAB software was included in the concentration of the calibration set, meeting the modeling requirements
[0039] Table 1-3 Division of the calibration set and validation set samples of the total sugar model
[0040]
[0041] 6. Establishment of the total sugar quantitative analysis model
[0042] As shown in Table 1-4, the quantitative analysis model R established based on the total sugar content values of Eucommia ulmoides leaves determined by near-infrared spectroscopy data and the anthrone-sulfuric acid method 2 values, respectively studied and compared the effects of different modeling methods, spectral preprocessing, and modeling bands on model R 2 . Finally, a quantitative analysis model for total sugar was established. The modeling conditions were as follows: baseline correction and smoothing preprocessing were adopted, and the genetic algorithm was used to select the modeling bands as Figure 4a, the modeling method is the total sugar quantitative analysis model established by random forest. The relative deviation between the actual value and the predicted value of the validation set samples is less than 0.3%, and the absolute deviation is between -3% and 3%, indicating that the established model has good performance and accurate prediction results.
[0043] Table 1-4 Model R based on different modeling methods and bands 2 Value
[0044]
[0045] 7. Reasonably divide the calibration set and validation set of the total flavonoid model samples
[0046] For 175 Eucommia samples, set the calibration set and validation set samples to 140 and 35 respectively. Use the built-in function Randperm in MATLAB software to generate random numbers to disrupt the inherent order of the samples and enhance the applicability of the model. The number of calibration set and validation set for each variety, the maximum value, minimum value, average value and relative standard deviation (RSD) of the total sugar content are shown in Table 1-5. The modeling requires that the concentration of the validation set samples is included in the concentration of the calibration set. In this experiment, the concentration of the validation set samples randomly selected by MATLAB software is included in the concentration of the calibration set, meeting the modeling requirements.
[0047] Table 1-5 Division of calibration set and validation set samples of the total flavonoid model
[0048]
[0049] 8. Establishment of the total flavonoid quantitative analysis model
[0050] As shown in Table 1-6 is the quantitative analysis model R established based on the near-infrared spectral data and the value of the total flavonoid content in Eucommia ulmoides leaves determined by the sodium nitrite method 2 Value, respectively studied and compared the effects of different modeling methods, spectral preprocessing and modeling bands on model R 2 Finally, a total flavonoid quantitative analysis model was established. The modeling conditions are: baseline correction and smoothing preprocessing are adopted, and the genetic algorithm is used to select the modeling band such as Figure 5a , the modeling method is the total sugar quantitative analysis model established by partial least squares method. The relative deviation between the actual value and the predicted value of the validation set samples is less than 0.2%, and the absolute deviation is between -2% and 2%, indicating that the established model has good performance and accurate prediction results.
[0051] Table 1-6 Model R based on different modeling methods and bands 2 Value
[0052]
[0053] The quantitative analysis models of crude protein, total sugar and total flavonoids in Eucommia ulmoides leaves established by the present invention can accurately, quickly and non-destructively detect the contents of crude protein, total sugar and total flavonoids in the corresponding Eucommia ulmoides leaves. It provides a new method for rapid, non-destructive and random detection of Eucommia ulmoides leaves for leaf use, thus providing technical support and theoretical basis for the precise production, management and cultivation of Eucommia ulmoides leaves for leaf use.
Claims
1. Method for detecting contents of total sugar, crude protein and total flavonoid compounds in Eucommia ulmoides leaves by near-infrared spectroscopy, comprising the following steps: ① Collect Eucommia ulmoides leaf samples, select a sufficient amount of representative calibration set samples, determine the crude protein content of Eucommia ulmoides leaves by the Kjeldahl method, determine the total sugar content of Eucommia ulmoides leaves by the anthrone-sulfuric acid method, and determine the total flavonoid content of Eucommia ulmoides leaves by the sodium nitrite-aluminum nitrate colorimetric method; ②Measure the near-infrared spectrogram of the calibration set samples, preprocess the spectral data through smoothing, baseline correction, and normalization, and select the modeling bands 5203.27 - 4751.98 cm -1 and 7000.69 - 6499.26 cm -1 , establish a crude protein quantitative model by combining the principal component regression method; select the modeling bands by genetic algorithm and establish a total flavonoid quantitative model by combining the partial least squares modeling method; select the modeling bands by genetic algorithm and establish a total sugar quantitative model by combining the random forest algorithm ③ Measure the near-infrared spectrogram of the unknown Eucommia ulmoides leaf sample, process the obtained spectrogram data by using the same pretreatment method as in step ②, and transfer the pretreated spectrum into the calibration model in step ② to obtain the contents of total sugar, crude protein and total flavonoid compounds in the Eucommia ulmoides leaf sample to be measured.
2. The method according to claim 1, characterized in that The scanning range for measuring the near-infrared spectrum of the sample is 12,000–4,000 cm -1 .
3. The method according to claim 1, characterized in that The collection season of Eucommia ulmoides leaf samples for near-infrared scanning is spring, summer and autumn, the collection varieties are Qinzhong No. 1, Qinzhong No. 2, Qinzhong No. 3, Qinzhong No. 4, Huazhong No. 5 and Huazhong No. 8, and the collection address is Lueyang County, Hanzhong City, Shaanxi Province.
4. The method according to claim 1 or 2, characterized in that The method for measuring the near-infrared spectrum diagram uses a Bruker MPA II Fourier transform near-infrared spectrometer, and the integrating sphere diffuse reflection method is adopted. The acquisition conditions are a resolution of 8 cm -1 , a spectral acquisition band of 12,000 - 4,000 cm -1 , and 64 scans. The temperature is (25 ± 0.5) °C, and the relative humidity is 35% - 45%. OPUS 6.0 software is used to collect spectral data. Each Eucommia ulmoides sample is collected 3 times repeatedly, and the average spectrum is obtained.
5. The method according to claim 1, characterized in that The operating conditions of the Kjeldahl method in step ① are: First, add 0.15 g of Eucommia ulmoides leaf powder, 1.5 g of mixed catalyst and 5 mL of concentrated sulfuric acid into a digestion tube (each time, six groups of samples are digested, and two controls are set for each group of samples; two blank tubes are set, and only 1.5 g of mixed catalyst and 5 mL of concentrated sulfuric acid are added into each tube), set the temperature in the digestion furnace to 420 °C, and the digestion time to 180 min. Then, after the digestion tube is cooled, measure the total nitrogen content in a FOSS Kjeldahl nitrogen analyzer respectively, and multiply the total nitrogen content measured by the Kjeldahl nitrogen analyzer by the coefficient 6.25 to record as the crude protein content of Eucommia ulmoides leaves.
6. The method according to claim 1, characterized in that The operating conditions of the anthrone-sulfuric acid method in step ① are: First, add 1 mL of diluted Eucommia ulmoides extract and 5 mL of 0.2% (m / v) C14H10O-H2SO4 solution into a 10 mL test tube. Then, place the test tube in a constant temperature water bath at 95 °C, heat for 15 min, take it out and let it stand for 30 min, and then measure the absorbance of the reaction system at a wavelength of 622 nm using a UV spectrophotometer. TSC is calculated with a glucose calibration curve (R2 = 0.9999), and the result is expressed as the equivalent of glucose (Glucose, G) per gram of dry leaf powder (mg RE / g DW).
7. The method according to claim 1, wherein The operating conditions of the sodium nitrite-aluminum nitrate colorimetric method in step ① are: First, add 1 mL of diluted Eucommia ulmoides extract and 0.5 mL of 5% (m / v) NaNO2 solution into a 10 mL volumetric flask, and react for 7 min. Then, add 0.6 mL of 10% (m / v) Al(NO3)3 solution to the mixed system. After reacting for 7 min, add 3 mL of 1 M NaOH solution to the reaction system, and make up the volume to 10 mL with deionized water. After 16 min, measure the absorbance of the reaction system at a wavelength of 506 nm using a UV spectrophotometer. TFC is calculated with a rutin calibration curve (R2 = 0.9999), and the result is expressed as the equivalent of rutin (Rutin, R) per gram of dry leaf powder (mg RE / g DW).
8. The method according to claim 1, wherein The pretreatment method of the near-infrared spectrogram is selected from one or more of the following methods: smoothing, baseline correction, normalization.
9. The method according to claim 1, characterized in that The method for selecting the band of the near-infrared spectrum diagram is selected from one or more of the following methods: comparison with reference literature, comparison with reference standards, and genetic algorithm.
10. The method according to claim 1, characterized in that The modeling method is selected from one of the following methods: partial least squares method, principal component regression method, random forest, decision tree, BP neural network, and support vector machine.