CN101655454A - Rapid determination method for evaluation of storage quality of grain - Google Patents

Abstract

The invention relates to the technical field of quality detection and provides a rapid determination method for the evaluation of the storage quality of grain, which comprises: S1: grain samples having different storage years are acquired according to the grain samples to be evaluated, thus obtaining a modeling sample set; S2: a near-infrared spectrum of the modeling sample set is obtained by means of acquisition and is pretreated; S3: a fatty acid value of the modeling sample set is obtained by means of measurement; S4: a correction relation for the modeling sample set is established according to the near-infrared spectrum and the fatty acid value, thus obtaining a correction model; and S5: the storage quality of the grain samples to be evaluated is evaluated according to the correction model. On the basis of near-infrared spectrum technique, the rapid determination method is in no need of the sample pretreatment, fast in detection, great in convenience and simplicity and suitable forthe determination of the fatty acid value in kernel of grain crop such as paddy rice, wheat, corn, soybean and the like, thus technically supporting the real-time detection and surveillance of the storage quality of grain.

Description

A Rapid Determination Method for Grain Storage Quality Judgment

technical field

The invention relates to the technical field of agricultural product quality detection, relates to grain storage quality evaluation, and in particular to a rapid determination method for judging grain storage quality based on near-infrared spectroscopy.

Background technique

With the rapid development of the economy and the continuous improvement of people's living standards, the quality and safety of agricultural products has received widespread attention. The decline in the quality of grain storage is a malignant change that occurs during the grain storage process. It is not suitable for processing finished grains for human consumption, involving multiple links in grain storage, transportation, sales, and processing. Due to its industrial nature and concealment, the quality of grain storage is not easy to be noticed by consumers, and it is also difficult for processors to detect. Grain storage quality that is severely unsuitable for storage or even slightly unsuitable for storage can accumulate through the food chain and endanger human health. Therefore, its monitoring Work is especially important.

Grain is affected by temperature, moisture and enzymes during storage, and its lipids undergo hydrolysis and oxidation reactions. When the grain is moldy, the fatty acid enzyme produced by the mold can promote the hydrolysis of the grain. Hydrolysis increases the content of free fatty acids in grains, which has adverse effects on the seed quality and eating quality of grains. Grain fatty acid value (mgKOH/100g), that is, the number of milligrams of potassium hydroxide required to neutralize free fatty acids in 100g of grain samples, is one of the main indicators for judging the quality of grain storage. "Rules for Judgment of Rice Storage Quality" GB/T20569-2006, "Rules for Judgment of Corn Storage Quality" GB/T20570-2006, "Rules for Judgment of Wheat Storage Quality" GB/T20571-2006 for the storage quality control indicators of rice, corn and wheat respectively Regulations were made, among which the fatty acid value is one of the important indicators. Therefore, it is particularly important to determine the fatty acid value of grains. The determination of grain fatty acid value usually adopts national standard methods, including titration, colorimetry and chromatographic analysis. These methods are often carried out in the laboratory for the determination of fatty acid value, and the analysis accuracy is relatively high. However, the procedures in the sample pretreatment and determination process are cumbersome, time-consuming, and costly. It is more difficult to analyze a large number of grain samples.

Contents of the invention

Aiming at the deficiencies of the prior art, the present invention provides a rapid determination method for judging grain storage quality based on near-infrared spectroscopy, which can perform real-time, online, and rapid evaluation of the storage quality of a large amount of grain.

The object of the present invention is to provide a kind of grain storage quality judgment fast assay method, described method comprises:

S1: Collect grain samples with different storage years according to the grain samples to be determined, and obtain a calibration sample set;

S2: collecting and preprocessing the near-infrared spectrum of the calibration sample set;

S3: measuring and obtaining the fatty acid value of the calibration sample set;

S4: Establish a calibration relationship of the calibration sample set according to the near-infrared spectrum and the fatty acid value, and obtain a calibration model;

S5: Determine the storage quality of the grain sample to be determined according to the calibration model.

Wherein, the correction relationship described in step S4 is: y _i =A _i x+a ₀ , where x is a near-infrared spectrum vector, and each element in the vector is an absorbance value at a different wavelength; y _i is a predicted value of fatty acid; A _i is a coefficient vector, its length value is the same as the number of independent variables of the spectrum x, and a ₀ is a constant term; A _i and a ₀ can be output by the partial least squares program.

Wherein, the correction relationship is established by using one or more chemometric methods selected from multiple linear regression, partial least squares regression, artificial neural network, and support vector machine.

Wherein, the diffuse reflectance near-infrared spectrum within the wavenumber range of 4000-12500 cm ⁻¹ is selected as the region for establishing the correction relationship.

Wherein, said step S5 further includes the following steps:

S501: Collect and preprocess the near-infrared spectrum of the grain samples to be determined;

S502: Obtain the fatty acid value of the grain sample to be tested according to the near-infrared spectrum and the calibration relationship;

S503: Determine the storage quality of the grain sample to be determined according to the fatty acid value according to national standards.

Wherein, the step S5 also includes the process of performing error correction and repeated optimization on the calibration model.

Wherein, the error correction and iterative optimization process includes the following steps:

Step S701: Establish a verification sample set according to the grain samples to be judged;

Step S702: collecting and preprocessing the near-infrared spectrum of the verification sample set;

Step S703: Obtain the predicted fatty acid value of the verification sample set according to the calibration model;

Step S704: measuring and obtaining the chemical value of the fatty acid value of the grain samples in the verification sample set;

Step S705: Correcting the error value by comparing the predicted value with the chemical value;

Steps S701-S705 are repeated to optimize the correction model.

Wherein, the near-infrared spectrum preprocessing method is selected from one or more of centering, standard variable transformation, additional scatter correction, orthogonal signal correction, smoothing, wavelet denoising, derivative transformation and genetic algorithm wavelength optimization.

Wherein, the measurement of the fatty acid value is realized by KOH titration.

The method of the invention does not require sample pretreatment, and the detection is fast and simple, and is suitable for the determination of fatty acid values in grains of major food crops such as rice, corn, and wheat, and provides technical support for real-time detection and monitoring of grain storage quality.

Compared with the prior art, the present invention has the advantages of:

The sample pretreatment is simple, and the grain samples only need simple extraction, purification and crushing;

Quick check. After the model is established, the near-infrared spectrum of the sample can be collected to calculate the fatty acid value of the sample through the calibration model;

It can provide technical support for online real-time detection. The connection between the near-infrared spectrometer and the computer can realize online detection, which is of positive significance for the detection of a large number of samples and the processing of detection data.

Description of drawings

Fig. 1 is the near-infrared spectrum of rice sample in the embodiment of the present invention;

Fig. 2 is a prediction scatter diagram of the fatty acid value determination model in rice samples established by using characteristic bands in the embodiment of the present invention.

Detailed ways

The rapid determination method for judging grain storage quality proposed by the present invention is described as follows in conjunction with the accompanying drawings and examples.

S101: Collect representative grain samples to form a modeling sample set;

Specifically, according to the scope of application of the calibration model, collect grain samples of different storage years to form a modeling sample set;

S102: collecting the near-infrared spectrum of the sample and performing preprocessing;

Specifically, the preprocessing method can be selected from one or more of centering, standard variable transformation, additional scatter correction, orthogonal signal correction, smoothing, wavelet denoising, derivation transformation, and genetic algorithm wavelength optimization; The purpose of the processing is to remove the influence of systematic noise, random noise and wavelengths not related to fatty acids in the spectrum, and to correct the light scattering caused by the uneven grain size. In order to obtain a spectrum with high signal-to-noise ratio, large amount of effective information, and low interference;

S103: Determination of the fatty acid value in the grain sample according to the standard analysis method;

Specifically, it can be achieved by KOH titration;

S104: Establish a correction model between the near-infrared spectrum and the fatty acid value of the grain;

Specifically, the diffuse reflectance near-infrared spectrum within the wavenumber range of 4000-12500cm ^-1 is selected as the area for establishing the correction relationship;

In the specific implementation process, the correction relationship can be: y _i =A _i x+a ₀ , where x is the near-infrared spectrum vector, and each element in the vector is the absorbance value at different wavelengths; y _i is the predicted value of fatty acid ; A _i is a coefficient vector, its length value is the same as the number of independent variables of the spectrum x, and a ₀ is a constant item; A _i and a ₀ can be output by the partial least squares method program;

The correction relationship can be established by one or more chemometric methods selected from multiple linear regression, partial least squares regression, artificial neural network, support vector machine; wherein multiple linear regression and partial least squares method can be A linear model between the spectrum and fatty acids was established, using multiple linear regression when there were fewer independent variables in the spectrum, and partial least squares when there were more independent variables. If there may be a nonlinear relationship between spectra and fatty acids, use artificial neural networks and support vector machines to establish corrective relationships;

S105: Verify the calibration model;

Collect common grain samples, collect and preprocess their near-infrared spectra according to step S102, predict them according to the calibration model established in step S104, and obtain predicted values; at the same time, measure their fatty acid values according to step S103; compare the fatty acid values of the grain samples and predicted values, and repeatedly optimize the calibration model according to the error requirements in actual production;

It should be noted that in the actual implementation process, this step can be omitted according to the specific conditions of the calibration model; in actual modeling, it is often difficult to determine in advance what kind of calibration relationship exists between the spectrum and fatty acids, so it is necessary to use the above multiple The two chemometric methods were used to build models respectively, and then compared their errors to finally determine the optimal correction relationship;

Specifically, the optimization method mainly includes the optimization of the near-infrared spectrum, the elimination of corresponding outliers, and the optimization of the model algorithm; the optimization goal is to make the predicted value of the model closer to the chemical value of the sample, that is, to make the calibration model correct The prediction of the verification sample set is better in parameter performance;

S106: Collect and preprocess the near-infrared spectrum of the grain sample to be tested, and use the verified calibration model to quantitatively detect the fatty acid value;

S107: Determine the storage quality of the grain based on the determination of the fatty acid value of the grain by near infrared.

Specifically, it is judged in accordance with GB/T20569-2006 "Judgment Rules for Rice Storage Quality", "GB/T20570-2006 Judgment Rules for Corn Storage Quality" and "GB/T20571-2006 Judgment Rules for Wheat Storage Quality".

The near-infrared spectroscopy technology used in the present invention has the characteristics of simple and fast use, and does not require sample pretreatment, and the measurement time of each sample only needs 1-3 minutes.

The method of the present invention will be further described in detail by citing specific examples below.

In this embodiment, the establishment and application range of the calibration model is located at rice samples with a storage period of 1-10 years.

Collection and preparation of samples: 200 rice samples with a storage period of 1-10 years were collected from grain depots in Beijing, Heilongjiang, and Jilin, respectively, and used for modeling sample sets and model inspection sample sets respectively; the samples were treated with impurities removal and purification Dry moderately in a dark and ventilated place; take 200g of each sample and grind it with a FOSS knife mill and put it in a ziplock bag for later use; measure the spectrum within 1h, and measure the chemical value within 2h;

Sample near-infrared spectrum collection: BuChi N-200 Fourier near-infrared spectrometer was used to collect diffuse reflectance infrared spectra of rice samples. The working wavenumber range of the instrument was 4000-10000cm ^-1 , and the resolution was 2cm ^-1 . The average spectrum was obtained after repeated scanning 3 times. The spectrometer is equipped with a quartz cup, and each time the sample is filled, it is smoothed with a glass piece to avoid the influence of the sample volume on the spectral collection effect; the collected near-infrared spectrum is shown in Figure 1.

Standard chemical method to analyze the fatty acid value in the rice sample: measure according to the KOH solution titration method in the national standard GB/T15684-1995 "Grain Grinding Products-Determination of Fatty Acid";

Establishment of calibration model: A calibration model of fatty acid values was established for 58 rice samples. When modeling, first remove the samples with obvious abnormalities, and then use the concentration gradient method to divide the calibration set and validation set.

The obvious abnormality in this example refers to the sample data whose fatty acid value is significantly higher than that of other samples in the grain samples with the same storage period, which can be eliminated to make the model representative.

The calibration set samples are used to build the model, and the validation set samples are used to evaluate the model. Smoothing and wavelet denoising were performed on the infrared spectrum of the sample, and the second-order derivative transformation pretreatment was performed at the same time, and the calibration model between the near-infrared spectrum and the standard chemical value was established by the partial least squares regression method.

Verification of the calibration model: for the established calibration model, the verification set samples (20 samples in this embodiment) are used for verification. The fatty acid value of the sample is predicted by the correction model to obtain the predicted value, and the standard chemical value of the sample is determined by KOH solution titration, and the predicted value is compared with the chemical value. The results are shown in Figure 2, and the predicted value has a higher correlation coefficient with the true value ( are greater than 0.8), indicating that the established model can accurately predict the fatty acid value of rice samples. In addition, as shown in Table 1, the cross-validation standard deviation of the model is relatively close to the predicted standard deviation, and the error between the predicted value and the standard chemical value is small, indicating that the model works well and the distribution of the validation set samples can be compared. Good measure of model performance. The above modeling results show that near-infrared spectroscopy can quickly and accurately determine the fatty acid value in rice samples.

Table 1

Parameters LV# (principal component score) nc (calibration set sample number) np (Number of samples in the validation set) R (correlation coefficient) SECV (interactive verification standard deviation) SEP (Standard Deviation of Prediction) Fatty acid value 4 58 20 0.9301 12.7766 9.1636

Judgment of storage quality of rice samples: Finally, according to the established calibration model, the fatty acid values of 12 rice samples from different grain depots were measured, and the storage quality was judged according to the "Rules for Judgment of Rice Storage Quality" GB/T20569-2006, The rice storage quality indicators are shown in Table 2:

Table 2

For the japonica rice samples in this example, the judgment results are shown in Table 3.

table 3

Sample serial number 1 2 3 4 5 6 7 8 9 10 11 12 Fatty acid value (mgKOH/100g) 26.88 33.87 34.35 31.32 35.73 62.96 29.79 65.24 59.2 57.21 62.99 113.47 storage quality Slightly unsuitable for storage Slightly unsuitable for storage Slightly unsuitable for storage Slightly unsuitable for storage Severely unsuitable for storage Severely unsuitable for storage Slightly unsuitable for storage Severely unsuitable for storage Severely unsuitable for storage Severely unsuitable for storage Severely unsuitable for storage Severely unsuitable for storage

The above embodiments are only used to illustrate the present invention, but not to limit the present invention. Those of ordinary skill in the relevant technical field can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, all Equivalent technical solutions also belong to the category of the present invention, and the scope of patent protection of the present invention should be defined by the claims.

Claims (9)

1, a kind of rapid assay methods of grain storage quality judgement comprises the following steps:

S1: gather the different grain samples that store the time limit according to grain samples to be determined, obtain the modeling sample collection;

S2: gather and obtain the also near infrared spectrum of the described modeling sample collection of pre-service;

S3: measure the fatty acid value that obtains described modeling sample collection;

S4: set up the correction relationship of described modeling sample collection according to described near infrared spectrum and described fatty acid value, obtain calibration model;

S5: the storage quality of judging grain samples according to described calibration model.

2, the method for claim 1 is characterized in that, correction relationship described in the step S4 is: y _i=A _iX+a ₀, wherein x is the near infrared spectrum vector, each element in the vector is the absorbance under the different wave length; y _iBe the fatty acid predicted value; A _iBe coefficient vector, its length value is identical with the independent variable number of spectrum x, a ₀Be constant term; A _iAnd a ₀Can export by the partial least square method program.

3, the method for claim 1 is characterized in that, adopts one or more the chemometrics method be selected from multiple linear regression, partial least squares regression, artificial neural network, the support vector machine to set up described correction relationship.

4, the method for claim 1 is characterized in that, selects 4000-12500cm ^-1Diffuse reflection near infrared spectrum in the wave-number range is the zone of setting up described correction relationship.

5, the method for claim 1 is characterized in that, described step S5 further comprises the following steps:

S501: gather the also near infrared spectrum of pre-service grain samples to be determined;

S502: the fatty acid value that obtains described grain samples to be determined according to described near infrared spectrum and described correction relationship;

S503: the storage quality of judging described grain samples to be determined according to described fatty acid value according to national standard.

6, the method for claim 1 is characterized in that, described step S5 also comprises the process that described calibration model is carried out error correction and optimizes repeatedly.

7, method as claimed in claim 5 is characterized in that, described error correction and repeatedly optimizing process comprise the following steps:

S701: set up the verification sample collection according to grain samples to be determined;

S702: gather and obtain the also near infrared spectrum of the described verification sample collection of pre-service;

S703: the fatty acid value predicted value that obtains described verification sample collection according to described calibration model;

S704: measure and obtain the fatty acid value chemical score that described verification sample is concentrated grain samples;

S705: by more described predicted value and described chemical score, rectification error value;

Repeating step S701～S705 optimizes described calibration model.

8, as claim 1 or 5 or 7 described methods, it is characterized in that described near infrared spectrum preprocess method is selected from that centralization, canonical variable conversion, additional scatter correction, orthogonal signal are proofreaied and correct, in level and smooth, small echo denoising, differentiate conversion and the genetic algorithm Wavelength optimization one or more.

As claim 1 or 7 described methods, it is characterized in that 9, the measurement of described fatty acid value realizes by the KOH titrimetry.

Images