CN104596982A - Method for measuring pectin of paper-making reconstituted tobacco by near-infrared diffuse reflection spectrum technology - Google Patents

Abstract

The invention relates to a method for measuring the pectin content of paper-making reconstituted tobacco leaves by near-infrared diffuse reflectance spectroscopy, and belongs to the technical field of paper-making reconstituted tobacco leaves. The method is mainly realized through the following steps: (1) collection of samples; (2) collection of original spectra; (3) determination of sample reference values; (4) selection and preprocessing of calibration sample sets and verification sample sets; (5) establishment of PLS model; (6) Model validation. The PLS model for determining the pectin content of reconstituted tobacco leaf products was established by near-infrared technology, which has the advantages of fast detection speed, high accuracy, and good reproducibility. important supporting role.

Description

Determination of Pectin in Papermaking Reconstituted Tobacco Leaf by Near Infrared Diffuse Reflectance Spectroscopy

technical field

The invention belongs to the field of reconstituted tobacco leaves by a papermaking method, and in particular relates to a method for measuring pectin in reconstituted tobacco leaves by a papermaking method using a near-infrared diffuse reflection spectrum technique.

Background technique

Papermaking reconstituted tobacco is an important raw material in cigarette production. Adding an appropriate amount of reconstituted tobacco to shredded tobacco can not only improve the physical properties and chemical components of cigarettes, improve the smoking safety of cigarettes, but also make full use of tobacco resources and reduce costs. At present, the reconstituted tobacco leaves of the papermaking method have problems such as high irritation and heavy woody gas, which affect its addition amount and use effect in cigarette products. The pectin content accounts for about 10%~13% of the total mass of reconstituted tobacco leaves. During pyrolysis, aldehydes and ketones with low boiling points will be produced, which will cause irritating coughing when burning and smoking. This is the cause of the above defects in reconstituted tobacco leaves. One of the main reasons. Therefore, it is of great significance to accurately and rapidly determine the pectin content in reconstituted tobacco leaves.

At present, the methods commonly used in the world to determine pectin in tobacco mainly include gravimetric method, carbazole colorimetric method, 3,5-dinitrosalicylic acid method (DNS), high performance liquid chromatography (HPLC), and enzymatic hydrolysis method. , flow analysis and gas chromatography (GC). Tobacco industry standard YC/T 346-2010 "Ion Chromatography for the Determination of Pectin in Tobacco and Tobacco Products" is based on the strong specificity of pectinase, using pectinase to pre-treat samples to generate galactose Alkyduronic acid simplifies the process of sample pretreatment. At the same time, according to the reductive characteristics of galacturonic acid, the pectin content in tobacco is determined by potentiometric titration. These methods are all time-consuming, and the longer measurement time not only reduces the efficiency, but also is difficult to meet the research work of testing personnel to improve the quality of reconstituted tobacco leaves by degrading the pectin content in reconstituted tobacco leaves. In view of this, it is necessary to develop a new rapid method for the determination of pectin content to provide a certain reference for the analysis and evaluation of pectin content in reconstituted tobacco leaves.

After the near-infrared light irradiates the material, several interaction forms such as absorption, transmission, total reflection, and diffuse reflection will occur. There are three main ways to collect near-infrared spectra: transmission, diffuse reflection, and transmission-diffusion.

For liquid samples with good light permeability, near-infrared light can penetrate the entire sample, and the transmission method is often used for spectral scanning, and the measured data is more accurate. The near-infrared light cannot completely penetrate the reconstituted tobacco leaves by the papermaking method, so the diffuse reflection method is used for spectral scanning. After the near-infrared diffuse reflection light enters the sample, it undergoes countless reflections, refraction, diffraction, and absorption, and then returns to the incident surface. This kind of analysis light loads the structure and composition information of the sample, and is a fast and environmentally friendly method. Detection technology, the application of near-infrared diffuse reflectance spectroscopy to determine the conventional chemical composition, style, uniformity and other aspects of tobacco has been reported, but the method for the determination of pectin content in reconstituted tobacco leaves by papermaking method has not been reported.

Contents of the invention

The purpose of the present invention is to provide a simple, easy and rapid method for determining pectin in reconstituted tobacco leaves. The invention adopts the near-infrared spectrum non-destructive detection technology, which not only improves the analysis efficiency and saves the cost, but also has obvious practicability for accurately measuring the reconstituted tobacco leaf pectin, improving the detection efficiency, and objectively reflecting the product quality. For the stable control of the intrinsic quality of reconstituted tobacco products by papermaking method, the use of reconstituted tobacco leaves by papermaking method has obvious beneficial effects on stabilizing the quality of cigarette products and shaping the style of cigarettes.

The technical scheme that the present invention adopts is as follows:

The method for determining the pectin of the reconstituted tobacco leaf by the papermaking method by near-infrared diffuse reflectance spectroscopy comprises the following steps:

Step (1), collection of samples: collect and prepare a batch of representative reconstituted tobacco leaf products by papermaking method;

Step (2), collect the original spectrum: take the reconstituted tobacco leaf products collected by the papermaking method in step (1), crush each product into powder and mix it evenly, put it into the sample cup, press lightly to make it flat, the thickness of the sample is ≥ 10mm, every Take 3 parallel samples of each product, use near-infrared spectroscopy to scan and collect spectra in a diffuse reflection manner, scan the spectrum once for each parallel sample, and each product corresponds to 3 parallel spectra, and then average the 3 parallel spectra to obtain a Original spectrum, each reconstituted tobacco leaf product is scanned and collected in the same way in turn to obtain the original spectrum corresponding to each product;

Step (3), measure the reference value of the sample: use the standard method to measure the pectin of the reconstituted tobacco leaf products collected in step (1) one by one, and obtain the reference value of the sample;

Step (4), selection and preprocessing of the calibration sample set and verification sample set: use the standard GB/T 29858-2013 method to select the calibration sample set and verification sample set from the original spectrum obtained in step (2); The spectra in the sample set and verification sample set are preprocessed to eliminate the influence of noise and baseline drift;

Step (5), establish the PLS model: make a one-to-one correspondence between the calibration sample set processed in step (4) and the sample reference value obtained in step (3), and use the partial least squares method to compare the spectral data with the corresponding pectin measurement data Fitting, establishing a quantitative model, detecting and removing outliers in the process, and performing a one-to-one correspondence with the obtained sample reference values on the remaining spectral values after removing outliers, and using the partial least squares method to measure the spectral data and the corresponding pectin The data is fitted, and the quantitative model for obtaining the pectin index is established;

Step (6), model verification: use the verification sample set processed in step (4) to perform external prediction on the quantitative model of the pectin index established in step (6).

Further, the spectral collection conditions in step (2) are: scanning range: 4000cm ^-1 ~10000cm ^-1 ; resolution: 8cm ^-1 ; scanning times not less than 72 times.

Step (3) Determination of sample reference value is to use a standard method to measure the pectin of the collected samples one by one, and the standard method to obtain the reference value is the tobacco industry standard YC/T 346-2010 "Tobacco and Tobacco Products Determination of Pectin Ion Chromatography".

Further, the preprocessing in step (4) refers to performing multivariate scattering correction, first-order derivative and Norris Derivative filter preprocessing on the original spectrum, that is, performing MSC+first-order+Norris preprocessing on the original spectrum.

The detection and elimination of abnormal values in the step (5) described in the above technical solution is to use the method of detecting the leverage value, specifically through the following steps:

Calculate the sample leverage value according to the following formula:

Among them, H _i is the sample leverage value, t _i is the factor vector of sample i , T ^T T is the factor score matrix of the modeling set, is the transpose of t _i ;

When the sample leverage value is greater than 3k/n, where k is the main component number and n is the number of samples, its spectrum has a significant impact on the regression and should be eliminated.

Further, the model verification in the step (6) is to use the t-test method to determine whether the predicted value obtained by the quantitative model of the verification sample set input step (6) has a statistical deviation from the corresponding sample reference value measured in step 4 : That is, compare the predicted value of the quantitative model of the pectin index established in step (6) with the reference value t of the sample in step (3) and the critical value t _(a,dv-1) of the degree of freedom d _v- 1 For comparison, take the significant level a=0.05. When |t|<t _(a,dv-1) and the probability P>0.05, it means that there is no significant difference in the detection results of the two methods, and the model verification is successful. This model can be used for Determination of pectin in paper-making reconstituted tobacco leaves.

Determination of pectin: the original spectrum of the reconstituted tobacco leaf sample to be tested is input into the quantitative model of the pectin index established in step (5), that is, the pectin is obtained by measurement.

The PLS model is a development of the principal component regression correction method (PCR). In principal component regression analysis, the spectral matrix is decomposed by a certain number of principal factors to achieve the purpose of data dimensionality reduction and elimination of useless information (noise). In the PLS regression analysis, in addition to decomposing the spectral matrix, the concentration matrix is also decomposed and reduced in dimension, and mutual information is introduced. The principle is as follows:

(1) The model of matrix decomposition is:

A _m×p =T _m×k P _k×p +E _A

C _m×n =U _m×k Q _k×n +E _C

Among them, A is the spectral matrix; C is the concentration or property matrix; T is the score matrix of the spectral matrix; U is the score matrix of the concentration matrix; P is the loading matrix of the spectral matrix-principal component matrix; Q is the loading matrix of the concentration matrix- Principal component matrix; E _A is the spectral error matrix introduced by applying PLS for regression analysis; E _C is the concentration error matrix introduced by applying PLS for regression analysis; m is the number of samples; p is the number of wavelengths; k is the number of principal components.

(2) Regression analysis of the score matrix T and U to obtain the correlation coefficient matrix B

U _m×k =T _m×f B _k×k

(3) In predictive analysis, according to A _m×p =T _{m×k P k×p} +E _A , the load matrix P obtained by decomposing the unknown spectral matrix A of the sample is obtained to obtain the _unknown score matrix T of the sample. Calculate the concentration of the unknown sample according to C _unknown = T _unknown BQ.

Compared with the prior art, the present invention has the beneficial effects as follows: the method of the present invention has the advantages of fast detection speed, no pollution, environmental protection, high accuracy, good reproducibility, etc., and is suitable for on-site analysis of pectin in reconstituted tobacco leaf products , rapid detection and online monitoring of product quality fluctuations. For stabilizing and controlling the intrinsic quality of reconstituted tobacco products by papermaking method, it has obvious beneficial effects to make use of reconstituted tobacco leaves by papermaking method to stabilize the quality of cigarette products and shape the style of cigarettes.

a. Fast analysis speed: the measurement of a sample is completed within 3 minutes, and the pectin in the reconstituted tobacco leaf product can be quickly determined through the established pectin model, and the fluctuation of the sample quality can be preliminarily judged;

b. It is a non-destructive analysis technology: the sample is not damaged during the near-infrared spectroscopy measurement process, and it will not affect the sample from the appearance to the inside;

c. Low analysis cost and no pollution: the sample itself is not consumed during the sample analysis process, no chemical reagents are used, the cost is reduced, and it does not cause any pollution to the environment;

d. Good test reproducibility: the spectral measurement is less disturbed by human factors, and the measurement is stable;

e. Easy to realize online analysis: near-infrared spectrum has good transmission characteristics in optical fiber, which can realize online analysis and remote monitoring.

The present invention provides a method for measuring pectin in reconstituted tobacco leaves by a papermaking process using near-infrared diffuse reflectance spectroscopy, which is simple, easy and fast. The use of near-infrared spectrum non-destructive detection technology can not only improve analysis efficiency, but also save costs. For accurate determination of pectin in reconstituted tobacco leaves, Improve detection efficiency, objectively reflect product quality, and have obvious practicability.

Description of drawings

Figure 1 is a data fitting diagram of the quantitative model of the pectin index established in step (5).

Detailed ways

The present invention will be further described in detail below in conjunction with the examples.

Those skilled in the art will understand that the following examples are only for illustrating the present invention and should not be considered as limiting the scope of the present invention. If no specific technique or condition is indicated in the examples, it shall be carried out according to the technique or condition described in the literature in this field or according to the product specification. The reagents or instruments used were not indicated by the manufacturer, and they were all conventional products that could be purchased.

The method for measuring the pectin of reconstituted tobacco leaves by the papermaking method by using near-infrared diffuse reflectance spectroscopy comprises the following steps:

Step (1), sample collection: collect 250-300 representative papermaking reconstituted tobacco leaf samples;

Step (2), take the reconstituted tobacco leaf products collected by the papermaking method in step (1), crush each product into powder and mix evenly, put it into the sample cup, press lightly to make it flat, the thickness of the sample is ≥ 10mm, and take 3 pieces of each product Each parallel sample is scanned and collected by diffuse reflectance using near-infrared spectroscopy. Each parallel sample scans the spectrum once, and each product corresponds to 3 parallel spectra. Then the 3 parallel spectra are averaged to obtain an original spectrum. Use the same method to scan and collect spectra for each reconstituted tobacco leaf product to obtain the original spectrum corresponding to each product; the spectral collection conditions are: scanning range: 4000cm ^-1 ~10000cm ^-1 ; resolution: 8cm ^-1 ; The number of scans is not less than 72 times.

Step (3), determine the reference value of the sample: use the standard method (tobacco industry standard YC/T 346-2010 "Ion Chromatography for the Determination of Pectin in Tobacco and Tobacco Products") to reconstitute the reconstituted tobacco products collected in step (1) one by one Carry out the determination of pectin, obtain sample reference value;

Step (4), select the calibration sample set and verification sample set and pretreatment: use the standard GB/T 29858-2013 method to select the calibration sample set and verification sample set from the original spectrum obtained in step (2); then the calibration sample set Perform MSC + first-order + Norris (3, 5) preprocessing on the spectrum of the verification sample set to eliminate the influence of noise and baseline drift;

Step (5), establish the PLS model: select the calibration set samples in the full spectral range to perform PLS regression and complete cross-validation. When the number of principal components of the model reaches 2, the root mean square cross-validation error RMSECV of the model is the smallest, and the maximum value of the model is selected. The appropriate number of principal components is 2.

One-to-one correspondence between the calibration sample set processed in step (4) and the sample reference value obtained in step (3), statistical fitting of the spectral data and the corresponding pectin measurement data by using the partial least squares method, and the establishment of a quantitative model, the process To detect and eliminate outliers in

Calculate the sample leverage value according to the following formula:

H _i is the sample leverage value, t _i is the factor vector of sample i , T ^T T is the factor score matrix of the modeling set, is the transpose of t _i .

When the sample leverage value is greater than 3k/n, where k is the main component number and n is the number of samples, its spectrum has a significant impact on the regression and should be eliminated; all calibration sample sets and verification sample sets in step (3) are obtained by calculation The spectral lever values are all less than 3k/n, and no abnormal samples need to be eliminated;

For the remaining spectral values after removing outliers, one-to-one correspondence with the obtained sample reference values was carried out again, and the spectral data were fitted with the corresponding pectin measurement data by using the partial least squares method to establish a quantitative model for obtaining pectin indicators (see Fig. 1), the correlation coefficient of the quantitative model is 0.855, and the RMSECV is 0.924. There is a significant linear relationship between the visible spectrum data and the quantitative index of the sample, indicating that the near-infrared spectrum of the sample contains information closely related to the quantitative index;

Step (6), model verification: Use the 50 products in the verification sample set processed in step (4) to perform external prediction on the quantitative model of the pectin index established in step (5), and calculate the root mean square prediction error according to the following formula RMSEP:

 

Among them, D _i f _i =xi- _{y i} is the difference between the near-infrared measured value x _i of the i-th sample and the reference value y _i of the sample as the analysis standard. The root mean square prediction error (RMSEP) calculated by the formula is 0.5397, and the model established by PLS has high prediction accuracy and prediction stability. Table 1 shows the prediction results of the PLS model prediction set. Use the t test method to determine whether the predicted value obtained by the quantitative model of the verification sample set input step (5) has a statistical deviation from the corresponding sample reference value measured in step (3): that is, the established step (5) Compare the predicted value of the quantitative model of the pectin index with the sample reference value t value in step (3) and the critical value t _(a,dv-1) of the degree of freedom d _v-1 , and take the significance level a=0.05. In this embodiment, carry out paired t test by the quantitative correction model prediction verification sample result of establishment and standard measurement method, t value is found out according to 95% confidence interval, t _0.05,49 =2.009, the quantitative model prediction result built by PLS and The measured value of the standard reference method |t|<t _0.05,49 , P>0.05, there is no significant difference in the detection results of the two methods, and the prediction results of the quantitative calibration model built by PLS are reliable.

Table 1 Prediction results of PLS model prediction set

Determination of pectin: input the original spectrum of 10 samples of reconstituted tobacco leaves to be tested into the quantitative model of the pectin index established in step (5), that is, to obtain pectin; predict 5 times for each sample, and apply step (3) at the same time The standard method was used to measure the sample pectin, and the results are shown in Table 2 below. It can be seen from Table 2 that it is completely feasible to apply near-infrared diffuse reflectance spectroscopy to the quantitative detection of reconstituted tobacco leaf products in production practice.

Table 2 Measurement results of actual samples

This model can be used to quickly and accurately determine the pectin of reconstituted tobacco products, which is of great significance for realizing on-site analysis and monitoring the quality stability and quality fluctuation of reconstituted tobacco products.

In addition, the reason why the preprocessing of step (4) adopts MSC + first-order + Norris (3, 5) preprocessing is to correct the PLS model results of the sample set under different spectral preprocessing, as shown in Table 3. Table 3 shows the results of different spectral preprocessing Process the PLS model results of the lower calibration set; it can be seen from Table 1 that different spectral preprocessing methods have different effects on the PLS modeling results, and MSC+first-order+Norris (3,5) has a better effect.

Table 3 Results of PLS model under different spectral preprocessing

Note: k is the number of principal components, RMSECV is root mean square cross-validation error, RMSEC is root mean square correction error, MSC is multiple scattering correction, SNV is vector normalization, Norris (a, b) is Norris Derivative filter, a is the segment length, b is the segment spacing, Savitzky-Golay (a, b) is a polynomial smoothing method, a is the number of smoothed data points, b is the number of polynomial powers; the first order is the first order derivative; the second order is the second order derivative.

The basic principles and main features of the present invention and the advantages of the present invention have been shown and described above. Those skilled in the industry should understand that the present invention is not limited by the above-mentioned embodiments. What are described in the above-mentioned embodiments and the description only illustrate the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will also have Variations and improvements are possible, which fall within the scope of the claimed invention. The protection scope of the present invention is defined by the appended claims and their equivalents .

Claims (5)

1. the method for measuring papermaking process reconstituted tobacco leaf pectin by near-infrared diffuse reflectance spectroscopy, is characterized in that, comprises the steps: Step (1), collection of samples: collect and prepare a batch of representative reconstituted tobacco leaf products by papermaking method; Step (2), collect the original spectrum: take the reconstituted tobacco leaf products collected by the papermaking method in step (1), crush each product into powder and mix it evenly, put it into the sample cup, press lightly to make it flat, the thickness of the sample is ≥ 10mm, every Take 3 parallel samples of each product, use near-infrared spectroscopy to scan and collect spectra in a diffuse reflection manner, scan the spectrum once for each parallel sample, and each product corresponds to 3 parallel spectra, and then average the 3 parallel spectra to obtain a Original spectrum, each reconstituted tobacco leaf product is scanned and collected in the same way in turn to obtain the original spectrum corresponding to each product; Step (3), measure the reference value of the sample: use the standard method to measure the pectin of the reconstituted tobacco leaf products collected in step (1) one by one, and obtain the reference value of the sample; Step (4), selection and preprocessing of the calibration sample set and verification sample set: use the standard GB/T 29858-2013 method to select the calibration sample set and verification sample set from the original spectrum obtained in step (2); The spectra in the sample set and verification sample set are preprocessed to eliminate the influence of noise and baseline drift; Step (5), establish the PLS model: make a one-to-one correspondence between the calibration sample set processed in step (4) and the sample reference value obtained in step (3), and use the partial least squares method to compare the spectral data with the corresponding pectin measurement data Fitting, establishing a quantitative model, detecting and removing outliers in the process, and performing a one-to-one correspondence with the obtained sample reference values on the remaining spectral values after removing outliers, and using the partial least squares method to measure the spectral data and the corresponding pectin The data is fitted, and the quantitative model for obtaining the pectin index is established; Step (6), model verification: use the verification sample set processed in step (4) to perform external prediction on the quantitative model of the pectin index established in step (6). 2. The method for measuring pectin in reconstituted tobacco leaves by the papermaking method according to claim 1, characterized in that the spectrum collection conditions in step (2) are: scanning range: 4000cm ^-1 ~ 10000cm ^{- 1} ; Resolution: 8cm ^-1 ; The number of scans is not less than 72 times. 3. The method for measuring the pectin of reconstituted tobacco leaves by the papermaking method according to the near-infrared diffuse reflectance spectroscopy technique according to claim 1, characterized in that, the pretreatment in step (4) refers to performing multivariate scattering correction and first-order derivative to the original spectrum. And Norris Derivative filter preprocessing. 4. The method for measuring the pectin of reconstituted tobacco leaves by the papermaking method according to the near-infrared diffuse reflectance spectroscopy technique according to claim 1, characterized in that: in the step (5), detection and elimination of abnormal values is the method of detecting lever values, specifically Go through the following steps: Calculate the sample leverage value according to the following formula: Among them, H _i is the sample leverage value, t _i is the factor vector of sample i , T ^T T is the factor score matrix of the modeling set, is the transpose of t _i ; When the sample leverage value is greater than 3k/n, where k is the main component number and n is the number of samples, its spectrum has a significant impact on the regression and should be eliminated. 5. The method for measuring the pectin of reconstituted tobacco leaves by the papermaking method according to the near-infrared diffuse reflectance spectroscopy technique according to claim 1, characterized in that: the model verification of the step (6) is to use the t-test method to determine the input step of the verification sample set ( 6) Whether there is a statistical deviation between the predicted value obtained by the quantitative model and the corresponding sample reference value measured in step 4: that is, the predicted value of the quantitative model of the pectin index established in step (6) and the step ( 3) Compare the reference value t of the sample with the critical value t _(a,dv-1) of the degree of freedom d _v-1 , take the significant level a=0.05, when |t|<t _(a,dv-1) , When the probability P>0.05, it means that there is no significant difference in the detection results of the two methods, and the model verification is successful, and the model can be used to determine the pectin in reconstituted tobacco leaves by papermaking method.