CN105548234A - Method for nondestructive detection of water and fat contents of yellow croaker - Google Patents

Abstract

The invention provides a method for non-destructive determination of water and fat content of yellow croaker, comprising the steps of: (1) testing of yellow croaker samples: using CPMG sequence to collect transverse relaxation signals to obtain the transverse relaxation spectrum of yellow croaker samples; (2) yellow croaker samples Determination of fish water and fat content, (3) establishment of prediction model for water and fat content of yellow croaker: T2 relaxation spectrum as independent variable, water and fat content as dependent variable to establish a prediction model, (4) signal data analysis and deal with. The method proposed by the present invention studies the correlation between spectra and components, takes the low-field nuclear magnetic resonance relaxation data of yellow croaker as the research object, and uses the moisture and fat content in the yellow croaker as indicators to establish the moisture and fat content of the yellow croaker. The content PCR and PLSR prediction model realize the rapid detection of the water and fat content of yellow croaker.

Description

A non-destructive determination method for moisture and fat content of yellow croaker

technical field

The invention belongs to the field of analysis and measurement of materials, and in particular relates to an analysis and measurement method based on nuclear magnetic resonance.

Background technique

The yellow croaker belongs to the fish class, the croaker family, and is divided into large yellow croaker (Pseudosciaenacrocea) and small yellow croaker (Psendosciaenapolyactis), which are respectively one of the four major marine species in my country. Large yellow croaker is also called big first, golden dragon, cucumber fish, red melon, golden dragon, sweet-scented osmanthus yellow croaker, king fish, large yellow croaker; small yellow croaker is also called plum, plum fish, small king fish, small first, small spring fish, small cucumber fish, thick scale Aberdeen, flower fish. In recent decades, yellow croaker has been an important economic fishery resource in China, Korea and Japan, with an annual output of more than 300,000 tons (Conservation Genetics Resources, 2013, 6, 397-399; Fisheries Research, 2014, 153, 41-47). Yellow croaker is rich in polyunsaturated fatty acids, protein, various trace elements and rich physiologically active substances (Journal of Anshan Normal University, 2009, 11, 32-34), which has a good tonic effect on the human body and is beneficial to the weak and middle-aged. For the elderly, eating yellow croaker will receive a good therapeutic effect, and its nutritional content is the quality parameter that consumers and the seafood industry are most concerned about. The fat and water content of yellow croaker are important indicators to evaluate its quality and safety. The level of fat content not only directly affects the nutritional value of fish meat, but also affects the post-processing of yellow croaker. On the other hand, since yellow croaker contains a lot of water, the level of water content during storage affects the growth of microbial communities, thereby affecting the shelf life of fish meat. Therefore, it is of great significance to detect the fat and water content in yellow croaker.

The traditional method to measure the moisture and fat content of yellow croaker is obtained by direct heating drying and organic solvent extraction respectively. Although these traditional methods can obtain direct and reliable measurement results, they need to destroy the sample, only detect a small part of the representative sample to obtain the average value, cannot guarantee the real-time performance of the measurement data, and are time-consuming and labor-intensive, polluting the environment. Therefore, it is necessary to develop a rapid and non-destructive detection method that can detect the fat and water content of yellow croaker on-line in real time. Near-infrared spectroscopy can be successfully used to detect the content of water and fat in fish, and has a high correlation with the results measured by physical and chemical analysis methods (Chemometrics and Intelligent Laboratory Systems, 1988, 42: 199-207). However, the main disadvantage of NIR spectroscopy is that reflectance spectroscopy only provides information on the surface layer of the sample and cannot detect the internal fat and moisture content of inhomogeneous samples.

As an important modern analysis method, nuclear magnetic resonance has been widely used in various fields. Compared with near-infrared spectroscopy, since low-field nuclear magnetic resonance detects atomic nuclei (mainly hydrogen protons) with fixed magnetic moments, it can measure complete samples without Affected by surface properties, it has significant advantages such as non-invasive, fast, and accurate measurement results. Low-field NMR has been proven to be a versatile analytical method for studying fish quality (Journal of food science and technology 2013, 50, 228-38), water and fat content (Journal of the Science of Food and Agriculture, 2005, 85, 1299-1304). However, studies on the detection of water and fat content in intact yellow croaker using low-field NMR technology have not been reported yet.

Contents of the invention

In view of the deficiencies in the prior art, the purpose of the present invention is to provide a fast and non-destructive method for measuring the moisture and fat content of yellow croaker based on low-field nuclear magnetic resonance technology. This method can not only measure the moisture and fat content of yellow croaker simultaneously, but also It does not destroy the sample and is not affected by the surface properties of the sample, which is suitable for the quality control of yellow croaker.

The technical scheme that realizes the object of the present invention is:

A non-destructive determination method for moisture and fat content of yellow croaker, comprising steps:

(1) Yellow croaker sample test: put the complete yellow croaker sample in the center of the permanent magnetic field radio frequency coil of the nuclear magnetic resonance imaging analyzer, use the CPMG sequence to collect the transverse relaxation signal, and collect the signal 1 to 5 times each time, and take the average value , to obtain the transverse relaxation spectrum of the yellow croaker sample by multi-exponential fitting;

(2) Determination of water and fat content of yellow croaker: dry the yellow croaker sample with constant weight to obtain the water content in the yellow croaker sample; the yellow croaker sample uses petroleum ether as an extractant, and obtains the fat content in the yellow croaker sample by Soxhlet extraction. fat content;

(3) Establishment of prediction model for water and fat content of yellow croaker: according to the transverse relaxation data T2 of yellow croaker and the water and fat content data measured in step (2), process NMR relaxation data and water and fat with regression analysis method Content data, with T2 relaxation spectrum as the independent variable, water and fat content as the dependent variable, combined with chemometric principal component regression (PCR) and/or partial least squares regression (PLSR), to establish the moisture content of yellow croaker. and fat content prediction models;

(4) Analysis and processing of signal data: the established model can pass the correlation coefficient (Rcal2) of the calibration set, the root mean square error (RMSEC) of the calibration, the correlation coefficient (Rcv2) of the cross-validation, and the root mean square error (RMSEC) of the cross-validation ( RMSECV) and residual prediction deviation (RPD) are evaluated by one or more methods.

In the non-destructive measurement method of the present invention, specifically, the conditions for collecting transverse relaxation signals with a nuclear magnetic resonance imaging analyzer are: 90-degree pulse width P1: 13 μs, 180-degree pulse width P2: 26 μs, repeated sampling waiting time Tw: 2000 -10000ms, analog gain RG1: 10 to 20, (all integers), digital gain DRG1: 2 to 5 (all integers), preamp gain PRG: 1 to 3 (all integers), NS: 4, 8 , 16, NECH: 2000-10000, receiver bandwidth SW: 100, 200, 300KHz, control parameter RFD of start sampling time: 0.002-0.05ms, delay DL1: 0.1-0.5ms.

Wherein, the measurement parameters of the transverse relaxation data T2 in the step (1) are set as: number of sampling points TD: 200000-600000.

Further, the determination of moisture in the step (2): directly dry the chopped yellow croaker sample in a blast drying oven at 40-80°C to obtain the moisture content in the yellow croaker sample;

Determination of fat: Put the chopped yellow croaker samples into a vacuum freeze dryer for 24-72h. After being taken out from the vacuum freeze-drying instrument, the dried yellow croaker sample is pulverized with a pulverizer, and the pulverized yellow croaker sample is wrapped with filter paper, put into a Soxhlet extractor, then add 50-150ml sherwood oil in a round bottom flask, Extract at 90°C for 6-12 hours, remove the petroleum ether by rotary evaporation, and then vacuum-dry for 1-3 hours to completely volatilize the petroleum ether remaining in the oil, and obtain the fat content in the yellow croaker sample.

Usually, the chopped size is centimeter-sized, and the pulverized size is micron-sized powder, for example, sample powder below 100 microns.

Wherein, in the step (3), the 1000 CPMG echo peak point data of the modeling set sample are used as the independent variable X, and the dependent variable Y is the moisture or fat content of the yellow croaker, based on TheUnscrambler software, establish X by PCR and PLSR For the correlation model with Y, the full interactive validation is used to test whether the model has overfitting phenomenon.

Further, in step (4), the content prediction model established in step (3) is evaluated with the model determination coefficient R2 and the root mean square error (RMSE). The larger the R2, the smaller the RMSE, and the better the effect of the obtained model . Based on the evaluation in step (4), it can be determined which model established by the two regression methods is better.

More preferably, in step (3), when the principal component regression analysis (PCR) is used to establish the prediction model, the factor numbers of water and fat are determined to be 1 and 8 through residual variance analysis.

Wherein, in step (3), when the partial least squares regression method (PLSR) establishes the prediction model, the factor numbers of water and fat are determined to be 1 and 7 through residual variance analysis.

The method of the present invention also includes the step of: using a nuclear magnetic resonance imaging analyzer for the yellow croaker sample to be tested, adopting the same method as step (1) to collect transverse relaxation signals, based on the regression model obtained in step (3). Determine the water and fat content in the yellow croaker sample to be tested.

The beneficial effects of the present invention are:

The method that the present invention proposes utilizes Principal Component Regression (PCR) and Partial Least Squares Regression (PLSR) to study the correlation between spectra and components, taking the low-field nuclear magnetic resonance relaxation data of yellow croaker as research object, Taking the water and fat content in yellow croaker as indicators, the PCR and PLSR prediction models of water and fat content in yellow croaker were established, and the PCR and PLSR prediction models were compared to realize the rapid detection of water and fat content in yellow croaker. A rapid and non-destructive method for the determination of water and fat content of yellow croaker based on low-field nuclear magnetic resonance technology was established.

In the model established by the method of the present invention, the PCR prediction model of moisture, the correlation coefficients Rcal2 and Rcv2 of the correction set and the interactive verification set are all greater than 0.98, the root mean square error RMSEC and RMSECV are all less than 0.72, and the RPD value is 9.1835, which is greater than 3. The correlation coefficients Rcal2 and Rcv2 of the PCR prediction model for fat were both greater than 0.88, the root mean square error RMSEC and RMSECV were both less than 0.16, and the RPD value was 3.2015, greater than 3.

For the PLSR prediction model of moisture, the correlation coefficients Rcal2 and Rcv2 of the calibration set and the interactive verification set are both greater than 0.98, the root mean square error RMSEC and RMSECV are both less than 0.72, and the RPD value is 9.2360, which is greater than 3. The correlation coefficients Rcal2 and Rcv2 of the fat PLSR prediction model were both greater than 0.89, the root mean square error RMSEC and RMSECV were both less than 0.16, and the RPD value was 3.3730, greater than 3.

Description of drawings

Figure 1 shows the CPMG decay curve (A) and T2 transverse relaxation spectrum (B) of the yellow croaker sample.

Figure 2 is the residual variance and principal component score relationship diagram (A) and the predicted scatter distribution diagram (B) of the PCR model of water content of yellow croaker.

Figure 3 is the residual variance and principal component score relationship diagram (A) and the predicted scatter distribution diagram (B) of the PCR model of fat content of yellow croaker.

Figure 4 is the residual variance and principal component score relationship diagram (A) and the predicted scatter distribution diagram (B) of the PLSR model for water content of yellow croaker.

Figure 5 is the relationship diagram (A) between the residual variance and the principal component score of the PLSR model of the fat content of yellow croaker, and the predicted scatter distribution diagram (B).

detailed description

The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

Unless otherwise specified, the means adopted in the examples are conventional technical means in the art.

The magnetic resonance relaxation spectrum of embodiment 1 yellow croaker sample

Take a commercially available fresh yellow croaker sample, set three parallel samples, the mass is 25.24±9.29g, rinse with clean water repeatedly, and dry the surface moisture with filter paper. Put the yellow croaker samples in the center of the permanent magnetic field radio frequency coil of the nuclear magnetic resonance imaging analyzer (the model is MiniMR-Rat), and use the Carr-Purcell-Meiboom-Gill (CPMG) sequence to measure, and the sampling parameters are set to: repeat sampling while waiting Tw: 3000ms, analog gain RG1: 15, digital gain DRG1: 3, preamplifier gain PRG: 1, cumulative times NS: 8, number of echoes NECH: 1000, delay DL1: 0.5ms, number of sampling points TD: 208496 , the signal was collected three times for each sample, and the average value was taken. Finally, MultiExpInvAnalysis software was used for multi-exponential fitting to obtain the transverse relaxation spectrum of the yellow croaker sample.

Figure 1A is the CPMG relaxation curve of the complete yellow croaker sample, and Figure 1B is the transverse relaxation spectrum of the yellow croaker sample through multi-exponential fitting, as can be seen from Figure 1 due to the difference in water and fat content in each sample, The corresponding CPMG relaxation curves and transverse relaxation maps are also different. However, the water and fat transverse relaxation signals of yellow croaker samples overlap, so the quantitative analysis of water and fat content cannot be carried out through the transverse relaxation spectrum of yellow croaker samples. Analysis of fat content.

Embodiment 2: the mensuration of yellow croaker moisture and fat content (direct method)

Determination of moisture: Mince the yellow croaker sample, then directly dry it in a blast drying oven for 48 hours to reach a constant weight, weigh the weight difference before and after drying the yellow croaker sample, and calculate the moisture content.

Determination of fat: Mince the yellow croaker sample and put it in a vacuum freeze dryer to dry for 48 hours, then use a pulverizer to crush the dried yellow croaker sample to micron-sized powder, wrap the crushed yellow croaker sample with filter paper, and put it into the Soxhlet Then add 150ml of petroleum ether to the round bottom flask, extract at 90°C for 6h, remove the petroleum ether by rotary evaporation, and then vacuum dry for 2h to completely volatilize the residual petroleum ether in the oil to obtain the fat in the yellow croaker sample content.

Table 1 lists the maximum value, minimum value, average value, standard deviation and coefficient of variation of water and fat (fat) content in yellow croaker samples. It can be seen from the table that the water content of the yellow croaker sample is between 8.38-27.02g, the fat content is between 0.43-1.72g, and the average value and standard deviation of the water and fat content are 16.63±6.58g and 1.11± 0.51g. The coefficients of variation of moisture and fat content were 39.57% and 45.95%, respectively.

Table 1 Moisture and fat content of yellow croaker samples

Example 3: Establishment of prediction model for moisture and fat content of yellow croaker and model evaluation

Using Principal Component Regression (PCR) and Partial Least Squares Regression (PLSR) modeling, the data of 1000 CPMG echo peak points collected from the sample is used as the independent variable X, and the dependent variable Y is the water or fat content of the yellow croaker. Based on TheUnscrambler The software uses PCR and PLSR to establish a correlation model between X and Y.

Figure 2 and Figure 3 are the relationship diagram (A) and the predicted scatter distribution diagram (B) of the PCR model residual variance and principal component scores of water and fat in yellow croaker samples, respectively. The optimal number of factors for establishing a prediction model can be determined through the residual variance scatter diagram. It can be seen from Figure 2A and Figure 3A that when PCR is used to establish a prediction model, the optimal factor number of water and fat can be obtained by residual variance analysis of water and fat content in yellow croaker. The number of factors was 1 and 8, respectively, so that the scatter distribution diagrams of the measured values and predicted values of the water and fat low-field NMR prediction model of yellow croaker samples were obtained (Fig. 2B and Fig. 3B). The abscissa is the moisture and fat content measured by physical and chemical methods, and the ordinate is the predicted moisture and fat content. It can be seen that the scatter diagram of water is relatively uniform, and the scatter diagram of fat is relatively scattered.

Table 2 is the evaluation results of the prediction model for water and fat content of yellow croaker samples based on transverse relaxation signals. It can be seen that the PCR prediction model of moisture has achieved very good results. The results of the calibration set and the interactive verification set are similar. The correlation coefficients Rcal2 and Rcv2 are 0.9891 and 0.9876, both greater than 0.98. The root mean square error RMSEC and RMSECV are 0.6427 and 0.6427 respectively. 0.7165, both were small, and the RPD value was 9.1835, greater than 3, indicating that low-field NMR combined with PCR can accurately predict the water content of yellow croaker. The correlation coefficients Rcal2 and Rcv2 of the PCR prediction model for fat were 0.9714 and 0.8950, the root mean square error RMSEC and RMSECV were 0.0797 and 0.1593, and the RPD value was 3.2015, which was greater than 3. The correlation coefficient of fat was relatively low.

Table 2 The parameters of the PCR model for water and fat content of yellow croaker samples

Figure 4 and Figure 5 are the relationship diagram (A) and the predicted scatter distribution diagram (B) of the PLSR model residual variance and principal component scores of water and fat in yellow croaker samples, respectively. It can be seen from Figure 4A and Figure 5A that when PLSR establishes the prediction model, the optimal factor numbers of water and fat are obtained by the analysis of residual variance of water and fat content of yellow croaker, which are 1 and 7, respectively, so that the water and fat of yellow croaker samples are low. The scatter plots of the measured and predicted values of the field NMR technique PLSR prediction model (Fig. 4B and Fig. 5B). The abscissa is the moisture and fat content measured by physical and chemical methods, and the ordinate is the predicted moisture and fat content. From Figure 4B and Figure 5B, it can be clearly found that the scatter diagram of water is more evenly distributed, and the scatter diagram of fat is more dispersed.

Table 3 shows the evaluation results of the PLSR prediction model for water and fat content of yellow croaker samples based on transverse relaxation signals. The PLSR prediction model of moisture has achieved very good results. The results of the calibration set and the interactive validation set are similar. The correlation coefficients Rcal2 and Rcv2 are 0.9892 and 0.9877, both greater than 0.98. The root mean square error RMSEC and RMSECV are 0.6397 and 0.7132, respectively. Both were small, and the RPD value was 9.2360, greater than 3, indicating that low-field NMR combined with PLSR could accurately predict the water content of yellow croaker. The correlation coefficients Rcal2 and Rcv2 of the PLSR prediction model for fat were 0.9714 and 0.8950, the root mean square error RMSEC and RMSECV were 0.0325 and 0.1512, and the RPD value was 3.3730, which was greater than 3. The correlation coefficient of fat was relatively low.

Table 3 The parameters of the PLSR model for water and fat content of yellow croaker samples

Using low-field nuclear magnetic resonance technology combined with PCR and PLSR to quickly and non-destructively detect the fat and water content of yellow croaker. The experimental results show that the R2 is greater than 0.89, the RMSE is small, and the RPD value is greater than 3, indicating that the PCR and PLSR prediction models can effectively predict the water and fat content of the yellow croaker sample, and the water prediction model is better than the fat prediction model , the PLSR prediction model is slightly better than the PCR prediction model.

The above-disclosed or claimed embodiments can be made or implemented within the scope of the experimental means not exceeding the present disclosure. All products and/or methods described in the preferred embodiments of the present invention clearly refer to those that do not violate the concept, scope and spirit of the present invention and can be used for the products and/or experimental methods and subsequent steps. All changes and improvements to the technical means in the described process belong to the concept, scope and spirit defined by the claims of the present invention.

Claims (9)

1. A non-destructive assay method for moisture and fat content of yellow croaker, comprising steps: (1) Yellow croaker sample test: put the complete yellow croaker sample in the center of the permanent magnetic field radio frequency coil of the nuclear magnetic resonance imaging analyzer, use the CPMG sequence to collect the transverse relaxation signal, and collect the signal 1 to 5 times each time, and take the average value , to obtain the transverse relaxation spectrum of the yellow croaker sample by multi-exponential fitting; (2) Determination of water and fat content of yellow croaker: dry the yellow croaker sample with constant weight to obtain the water content in the yellow croaker sample; the yellow croaker sample uses petroleum ether as an extractant, and obtains the fat content in the yellow croaker sample by Soxhlet extraction. fat content; (3) Establishment of prediction model for water and fat content of yellow croaker: according to the transverse relaxation data T2 of yellow croaker and the water and fat content data measured in step (2), process NMR relaxation data and water and fat with regression analysis method Content data, with T2 relaxation spectrum as an independent variable, water and fat content as dependent variables, combined with the principal component regression method and/or partial least squares regression method of chemometrics, the water and fat content prediction model of yellow croaker was established, (4) Analysis and processing of signal data: the established model can be one or There are many ways to evaluate. 2. The non-destructive measurement method according to claim 1, wherein the condition for collecting transverse relaxation signals with a nuclear magnetic resonance imaging analyzer is: 90-degree pulse width P1: 13 μs, 180-degree pulse width P2: 26 μs, repeated sampling Waiting time Tw: 2000-10000ms, analog gain RG1: 10 to 20, digital gain DRG1: 2 to 5, preamp gain PRG: 1 to 3, NS: 4, 8, 16, NECH: 2000-10000, receiving Machine bandwidth SW: 100, 200, 300KHz, control parameter RFD of start sampling time: 0.002-0.05ms, delay DL1: 0.1-0.5ms. 3. The non-destructive measurement method according to claim 1, characterized in that the measurement parameters of the transverse relaxation data T2 in the step (1) are set as: the number of sampling points TD200000-600000. 4. The non-destructive assay method according to claim 1, characterized in that, the determination of moisture in the step (2): the chopped yellow croaker sample is directly dried to constant weight in a 40-80°C blast drying oven , to obtain the water content in the yellow croaker sample; Determination of fat: Put the chopped yellow croaker sample into the vacuum freeze-drying apparatus for 24-72h, after taking it out from the vacuum freeze-drying apparatus, use a pulverizer to crush the dried yellow croaker sample, wrap the crushed yellow croaker sample with filter paper, Put it into a Soxhlet extractor, then add 50-150ml of petroleum ether to the round bottom flask, extract at 90°C for 6-12h, remove the petroleum ether by rotary evaporation, and then vacuum dry for 1-3h to make the residual petroleum ether in the oil Thoroughly volatilize to obtain the fat content in the yellow croaker sample. 5. The non-destructive measurement method according to claim 1, characterized in that, in the step (3), the 1000 CPMG echo peak point data of the modeling set sample are used as the independent variable X, and the dependent variable Y is the yellow croaker. For water or fat content, the correlation model between X and Y is established by principal component regression analysis and partial least squares regression. 6. non-destructive testing method according to claim 5, is characterized in that, in step (4), evaluates the content prediction model that step (3) sets up with model determination coefficient R2 and root mean square error (RMSE), The larger the R2, the smaller the RMSE, and the better the obtained model. 7. The non-destructive measurement method according to claim 1, characterized in that, in step (3), when the principal component regression analysis method sets up the prediction model, the factor numbers of moisture and fat are determined to be 1 and 8 by residual variance analysis. 8. The non-destructive measurement method according to claim 1, characterized in that, in step (3), when the partial least squares regression method sets up the predictive model, the factor numbers for determining moisture and fat are 1 and 7 by residual analysis of variance . 9. The non-destructive measurement method according to any one of claims 1 to 8, characterized in that, the method further comprises the step of using a nuclear magnetic resonance imaging analyzer for the yellow croaker sample to be tested, and adopting the same method as step (1) to collect Transverse relaxation signal, based on the regression model obtained in step (3), determines the water and fat content in the yellow croaker sample to be tested.