香蕉频蕉app

1. Material and method
1.1 Samples

The 415 samples of soybeans collected for the experiment came from all over the world. All samples were cleaned and packed in plastic sealed pockets and stored in a 4°C freezer.

1.2 Instruments
The near-infrared spectrum analysis of the experimental samples was carried out using a new analyzer SupNIR-2700 developed by Focused Photonics Inc.(FPI). The analyzer adopts a new generation of holographic scanning technology to detect light absorption signals at various wavelengths by TEC-cooled Ex-In-GaAs sensor. The analyzer has a wavelength range of 1000-2500 nm. The resolution is 10 nm, the wavelength accuracy is 0.2 nm, the wavelength repeatability is less than 0.05 nm, and the noise level is less than 50 microAU (RMS). The built-in wavelength correction module of the analyzer is used to detect the performance of the instrument, and can diagnose and correct the drift of the spectral signal to ensure stable and reliable performance. The sample is analyzed rotationally to ensure that the sample spectrum collected is representative. The measuring accessories contain different size of sample plates to be selected according to the sample.
The analyzer software system integrates analysis, instrument management, model management and remote control. English operation interface, database to store and manage various data, support user-defined reports, printout detection results, model management with one-click operation, easy to learn and use. The software can remotely manage the near-infrared analyzer in the network via the Internet to monitor and maintain the instrument in real time.

The sample moisture was dried by GW-1AT electric blast drying oven, the crude protein was analyzed by KDN-04A semi-automatic Kjeldahl azotometer, the crude fat was extracted by Soxhlet extractor; the amino acid was detected by S-433D automatic amino acid analyzer (Secam, Germany); the electronic balance model was FA2104S.

1.3 Method
1.3.1 Spectrum Analysis
Soybean and soybean meal samples were taken out from the freezer and naturally warmed to room temperature, and then spectral analysis was performed using a SupNIR-2700 analyzer. Each sample was repeatedly loaded twice to acquire a near-infrared absorption spectrum.
1.3.2 Determination of Reference Values of Each Component in The Sample
The moisture was determined with reference to “GB/T 14489.1 2008 Determination of moisture and volatile matter content of oil”. The soybean crude protein was determined with reference to “GB/T 14489.2 2008 Determination of crude protein in vegetable oils”. The soybean crude fat was determined with reference to “GB/T 14488.1 2008 Determination of oil content of vegetable oil”. The soybean meal crude fat was determined with reference to “GB/T 10359 2008 Determination of Oil Content of Oil Cakes Part 1: Hexane (or Petroleum Ether) Extraction Method". The amino acid was detected by S-433D automatic amino acid analyzer (Secam, Germany). Some samples were detected for one component information, and some samples were detected for multiple component information.
1.3.3 Model Establishment

A database is established by aligning the sample spectrum with the reference value obtained by the national standard detection. In order to eliminate the influence of baseline drift and sample particle size difference on the spectrum, each sample spectrum was pretreated with Savitzky-Golay smoothing, first order reciprocal and multiple scattering correction (MSC). The pre-treated sample spectrum is correlated with its corresponding characters. The model is established by partial least squares (PLS) method. The 80% sample forms a calibration set to establish a model, and the remaining 20% sample forms a testing set to test the model to determine the final result.

2. Results and discussion

Table 1 shows the model results for some characters of soybean and soybean meal, such as moisture, crude protein, crude fat and amino acid content. Figure 1 and Figure 2 show the related curve between the reference value and the near-infrared predicted value of the soybean and soybean meal.

Tab.1 NIR Model Parameters of Soybean and Soybean Meal

Fig.1 Related curve between reference value and NIR predicted value for soybean moisture (A), crude protein (B) and crude fat (C)

Fig.2 Related Curve Between Reference Value and NIR Predicted Value for 

Soybean Meal Moisture (A), Crude Protein (B), Crude Fat (C), Lysine (D) and Methionine (E)

It can be concluded from Table 1, Figure 1 and Figure 2 that the test values obtained by near-infrared analysis have a good correlation with the reference values of the national standard analysis method. It can be seen from Table 1 that the standard deviation of the accuracy of the near-infrared detected moisture index is about 0.25%, the standard deviation of the detected crude protein is about 0.45%, the standard deviation of the accuracy of the detected crude fat is about 0.35%.

Tab.2 Predicted Value of Soybean and Soybean Meal by NIR Analysis

Table 2 shows the absolute error analysis results of the near-infrared predicted value and the reference value of some test samples. It can be seen that the absolute error of moisture of most test samples is less than 0.2%; the absolute error of crude protein is less than 0.4%; the absolute error of crude fat is mostly less than 0.3%. The above results show that the SupNIR-2700 NIR analyzer has good accuracy.

3. Conclusion
Soybean and soybean meal were analyzed by the near-infrared analyzer (SupNIR-2700) independently developed by FPI. The predicted results were comparable to the reference values, which can meet the requirements of rapid quality inspection. It also shows that the near-infrared analyzer has a good application prospect in the field of grain and oil processing.

Related Products：

SupNIR-2700 Near Infra-red Analyzer

//oa.rlxzx.com/en/product_cont_8252.html