Mid-infrared Rapid Batch Detection Method for Saturated Fatty Acids in Milk

Abstract

The present invention belongs to the field of dairy cow performance measurement and milk quality detection, and discloses a rapid batch detection method for saturated fatty acids in milk by mid-infrared spectroscopy. In terms of the selection of characteristic bands, instead of using algorithms commonly used to screen features, a method of manual selection + multiple traversals is used. Finally, the characteristic bands for modeling are selected, especially the absorption region containing part of water is screened out, and it is proved that adding part of the water absorption band can improve the accuracy of the model. The optimal pretreatment and algorithm combination for establishing the saturated fatty acid model are selected, the optimal parameters are determined, and the accuracy of the model is improved. The rapid, accurate and low-cost detection of the saturated fatty acid content in raw milk is realized, and rapid batch detection is achieved.

Description

Technical Field

[0001] The present invention belongs to the field of dairy cow performance measurement and milk quality detection, and particularly relates to a rapid batch detection method for saturated fatty acids in milk by mid-infrared spectroscopy. Background Art

[0002] Milk has always been regarded as one of the most perfect foods naturally formed in nature. Milk is one of the natural sources of fatty acids, rich in linoleic acid, linolenic acid, especially conjugated linoleic acid, which is beneficial to human health and has an important impact on human health. It has now become one of the most important nutritional sources for Chinese people.

[0003] Saturated fatty acids (SFAs) have no unsaturated double bonds in their carbon chains and are important components of cell membranes. Some studies have found that SFAs are the main factors affecting plasma cholesterol. Therefore, some hypotheses suggest that reducing the intake of SFAs will reduce the risk of coronary heart disease. [1-2] . However, a comprehensive analysis of other studies found no direct link between SFA intake and heart disease. [3] . A large number of studies have found that not all SFAs can increase cholesterol levels, and some SFAs also have various beneficial biological functions, such as anti-cancer, anti-inflammatory, antibacterial, etc. [4-7] .

[0004] Short-chain saturated fatty acids (SC-SFAs) are mainly present in milk and dietary fiber. Although dietary fiber does not contain SC-SFAs, the fermentation of microorganisms in the cecum and distal colon produces SC-SFAs including butyrate. [8] Medium-chain saturated fatty acids (MC-SFAs) and long-chain saturated fatty acids (LC-SFAs) are mainly present in red meat, milk fat, and vegetable oils. [9] Very-long-chain saturated fatty acids (VLC-SFAs) are mainly present in mustard seed oil, peanut oil, and rapeseed oil.

[0005] Saturated fatty acids in milk are the most important component of fatty acids, and their content can account for more than 60% of the total fatty acids in milk

[10] . There is currently a national standard for the determination method of saturated fatty acids in milk, namely the national food safety standard - determination of fatty acids in foods (GB 5009.168-2016). In China, it is stipulated that gas chromatography is used to detect fatty acids in milk and dairy products (GB 5009.168—2016).

[11] , This method can quantitatively detect the types and contents of fatty acids, but the sample pretreatment process is complex, time-consuming, and costly. Researchers have tried to improve this method []. The main methods for detecting milk fat at home and abroad include gas chromatography, ultraviolet spectrophotometry, liquid chromatography, etc., but they all have problems such as high cost and low efficiency, and it is difficult to be used quickly and batchwise in production practice. Mid-infrared spectroscopy (MIR) technology is a very economical and efficient detection tool. By the different frequencies and corresponding peaks of the absorption of specific chemical bonds in molecules by mid-infrared light, the properties or contents of the substances and traits being studied can be predicted, and thus information such as the physiological state of organisms can be obtained. In the livestock industry, the use of MIR to detect the nutrition, physiology, etc. of dairy cows has gradually emerged in China. Currently, it is already possible to detect indicators such as the total protein and total fat, but there is currently no rapid batch detection model with very high precision for the content of saturated fatty acids.

[0006] Currently, foreign countries have already started research on detecting the components of milk by mid-infrared, but there are problems such as low precision and inaccurate characteristic bands. There is no relevant report in China. Compared with other countries, due to the influence of domestic climate, geographical environment, and feeding conditions, Chinese dairy cows are quite different from foreign dairy cows, and the milk quality also has its own characteristics. Foreign detection models may not necessarily be suitable for Chinese dairy cows. Therefore, it is necessary to quickly establish a detection method for saturated fat content suitable for Chinese dairy cow milk with independent intellectual property rights in China. This method can not only be used for rapid, efficient, accurate, and non-invasive determination of the content of saturated fatty acids, an important performance indicator in the lactation traits of dairy cows, providing phenotypic data and theoretical basis for dairy cow genetic breeding, but also can detect and analyze the content of saturated fatty acids in raw milk and dairy products, providing reference for the dairy processing industry and consumers. The purpose of the present invention is to solve the existing problems and establish a rapid detection technology for saturated fatty acids in milk, providing technical support for the development of the Chinese dairy industry. Summary of the Invention

[0007] The purpose of the present invention is to provide a rapid batch detection method for saturated fatty acids in milk by mid-infrared spectroscopy, which has the advantages of simplicity, rapidity, and batch processing.

[0008] Another purpose of the present invention is to provide the application of the rapid batch detection method for saturated fatty acids in milk by mid-infrared spectroscopy.

[0009] To achieve the above object, the present invention takes the following technical measures:

[0010] A rapid batch detection method for saturated fatty acids in milk by mid-infrared spectroscopy, comprising the following steps:

[0011] 1. Collect the MIR data in the characteristic bands in the infrared spectrum of the milk sample as: 1412.03 cm -1 -1489.19 cm -1 、1558.63 cm -1 -1720.67 cm -1 、2106.47 cm -1 -2229.92 cm -1 、2445.97 cm -1 -2689.03 cm -1 、3016.96 cm -1 -3468.34 cm -1 、3638.09 cm -1 -3746.12 cm -1 、4101.05 cm -1 -4170.50 cm -1 、4212.94 cm -1 -4510.00 cm -1 、4618.03 cm -1 -5007.68 cm -1 in the MIR data;

[0012] 2. Substitute the obtained MIR data into the diff2 + diff2 + PLSR(n_component = 5) model, and the predicted result of the saturated fatty acid content can be output.

[0013] The diff2 mentioned above is the second-order difference, and diff1 is the first-order difference.

[0014] In the above method, preferably, there is a difference of two wave points allowed before and after each band in step 1.

[0015] The protection content of the present invention also includes: using the above detection method to detect the content of saturated fatty acids in milk.

[0016] Compared with the prior art, the advantages of the present invention are:

[0017] 1. In terms of the selection of characteristic bands, instead of using the commonly used algorithm to screen features, a method of manual selection + multiple traversals is used. Finally, the characteristic bands used for modeling are selected. In particular, the absorption regions containing part of water are screened out, and it is proved that adding some water absorption bands can improve the accuracy of the model.

[0018] 2. The optimal pretreatment and algorithm combination for establishing the saturated fatty acid model are selected, the optimal parameters are determined, and the accuracy of the model is improved.

[0019] 3. The rapid, accurate and low-cost detection of the saturated fatty acid content in raw milk is realized. Rapid batch detection is achieved. The determination time for each sample is only 10 - 15 seconds, which improves the detection efficiency. It has strong practicability and will be widely applied to the determination of dairy cow performance and the detection of milk quality. Description of the Drawings

[0020] Figure 1 FIG. (a) shows the infrared spectrogram and FIG. (b) shows the average spectrogram of the untreated milk sample.

[0021] Figure 2 FIG. shows the infrared spectrogram of the milk sample pretreated with diff2 + diff2.

[0022] Figure 3 FIG. (a) shows the total spectral map of the nine selected characteristic bands and FIG. (b) shows the enlarged view of each characteristic band.

[0023] Figure 4 FIG. shows the correlation and fitting straight line graph between the true value and the predicted value of the model milk data. Detailed Embodiments

[0024] The technical solutions described in the present invention are all conventional solutions in the art unless otherwise specified; the reagents or materials are all from commercial channels unless otherwise specified.

[0025] 1. Experimental Materials

[0026] The test materials are from 266 Chinese Holstein cows in 36 dairy farms in 11 provinces of 5 regions in China. One milk sample is collected from each cow. The milk sample collection is completed by an automatic milking device. First, wipe the milking room with a disinfected towel, then disinfect the udder with an iodoglycerol mixed solution. After squeezing out the first three handfuls of milk, collect the milk samples throughout the milking process. Each milk sample is 40 ml and is filled into a new cylindrical sampling bottle with a diameter of 3.5 cm and a height of 9 cm. They are numbered in sequence, and 2-bromo-2-nitro-1,3-propanediol preservative is immediately added to each sampling bottle and slowly shaken to dissolve it fully. Ice packs (2 - 4 °C) are placed around the milk samples during transportation to prevent deterioration. After the samples reach the laboratory, spectral collection is carried out immediately.

[0027] 2. Mid-infrared spectroscopy determination and acquisition

[0028] Pour the sample into a cylindrical sample tube with a diameter of 3.5 cm and a height of 9 cm, and water bath in a 42 °C water bath for 15 - 20 min. Use the MilkoScan TM 7RM milk composition detector of FOSS company. Stretch the solid optical fiber probe into the liquid, mix the sample evenly and then scan.

[0029] 3. Detection method for the true (reference) value of saturated fatty acids in milk

[0030] 3.1 Instruments, equipment and reagents

[0031] Electric thermostatic water bath (Wuhan Yiheng Sujing Scientific Instruments Co., Ltd.); Agilent gas chromatograph, including automatic sampler, column oven, sample vial, vortex oscillator, syringe filter, 0.22 μm nylon filter membrane, 100 m × 0.25 mm × 0.2 μm, fused silica capillary column, helium gas.

[0032] Methyl undecanoate C11:0 FAME (CAS No. 1731 - 86 - 8): ≥99%, 94118 - 1ML, purchased from Supelco; Tristearin C13:0 TAG (CAS No. 26536 - 12 - 9): ≥99%, T3882 - 500MG purchased from Sigma - Aldrich; FAME mixed standard (37 fatty acids):, 1269119 - 100MG, purchased from Sigma - Aldrich; Other reagents are all domestic chromatographically pure.

[0033] 3.2 Experimental method

[0034] 3.2.1 Acquisition of mid - infrared spectrum

[0035] Use MilkoScanTM FT+ to collect spectra. The specific collection steps are as follows: Put the milk samples into an electric thermostatic water bath at 45 °C in batches and preheat for 5 min. Put the preheated milk samples on the detection rack and shake them up and down several times to mix the milk colloidal solution evenly. Put the detection rack on the detection conveyor belt, open the bottle caps, and perform detections in sequence. After collecting the spectra, the milk samples are stored frozen at - 20 °C for subsequent determination of the content of saturated fatty acids.

[0036] 3.2.2 Determination of saturated fatty acid content by gas chromatography technology

[0037] (1) Fat extraction: Dichloromethanol method (commonly used at home and abroad, more convenient) [12 -13]

[0038] (15 ml centrifuge tube) Add 1 ml of milk, 1.8 ml of water, and 8 ml of CH2Cl2 / CH3OH (2:1, v / v). Vortex for 10 min, centrifuge at 8000 rmp for 10 min, and take the lower layer solution. Add 4 ml of CH2Cl2 / CH3OH (2:1, v / v) to the remaining solution and extract again. Combine the extracts and blow dry with nitrogen.

[0039] (2) Fatty acid methylation: KOH methanol solution method

[0040] Add 50 mg of fat to 1 mL of n-hexane, add 50 μL of KOH-MeOH solution (2M) for methylation. Vortex for 30 s, centrifuge at 8000 rmp for 5 min, and take the supernatant for membrane filtration and GC-MS analysis

[14]

[0041] (3) Gas chromatography instrument parameters

[0042] Split injection mode

[0043] Chromatographic column: 100 m × 0.25 mm × 0.2 μm, fused silica capillary column Stationary phase: cyanopropyl-polysiloxane

[0044] Carrier gas: helium

[0045] Column head pressure: 225 kPa (175 kPa - 225 kPa)

[0046] Total flow rate: 25.5 ml / min

[0047] Split ratio: 10:1

[0048] Injector temperature: 250 °C

[0049] Detector temperature: 275 °C

[0050] Injection volume: 1 μl.

[0051] Programmed temperature rise:

[0052] Time / min Temperature / °C Retention time / min Rate / °C / min 0 60 5 5 165 1 15 10 225 20 2

[0053] (4) Sum up the measured results of the 17 saturated fatty acids, which is the content of saturated fatty acids

[0054] 4. Selection of effective samples

[0055] Among the 266 samples, eliminate the invalid data caused by sample deterioration, loss, abnormal mid-infrared spectrum measurement, abnormal sample reference value measurement, etc. A total of 34 abnormal samples were eliminated, and 232 samples were selected for model establishment and optimization.

[0056] Example 1:

[0057] Selection of the prediction model algorithm for saturated fatty acids:

[0058] The purpose of this application is to establish a quantitative determination model for saturated fatty acids. Therefore, the modeling algorithm used is the regression algorithm. There are many types of regression algorithms. In this embodiment, the Ridge regression and Partial Least Squares Regression (PLSR) algorithms are mainly used for model establishment and comparison. The reasons are as follows:

[0059] Ridge regression is a type of linear regression. When establishing the regression equation by the algorithm, Ridge regression adds a regularization constraint to achieve the effect of solving overfitting. There are two types of regularization, namely l1 regularization and l2 regularization. The advantages of l2 regularization compared to l1 regularization are: (1) It can perform cross-validation; (2) It realizes stochastic gradient descent. Ridge regression is a linear regression model with l2 regularization added, retaining the advantages of linear regression, meeting the requirements of model establishment, and having relatively stable results. It is one of the commonly used basic algorithms. Therefore, this algorithm is selected as the candidate algorithm in this embodiment.

[0060] The Partial Least Squares Regression algorithm is one of the very effective algorithms for multi-feature samples. In mid-infrared spectral data, each sample corresponds to 1060 wave points, which is representative of multi-feature samples. At the same time, the Partial Least Squares Regression algorithm rarely shows overfitting. Therefore, many researchers of mid-infrared spectra choose to use the Partial Least Squares Regression algorithm for model establishment. Therefore, this algorithm is selected as the candidate algorithm in this embodiment.

[0061] Example 2:

[0062] Screening of the mid-infrared spectral preliminary model algorithm and its optimal pretreatment combination:

[0063] Effective feature screening is a basic operation for processing spectral data, aiming to denoise and lay a good foundation for feature extraction. Effective feature screening mainly includes three types: feature extraction, feature preprocessing, and feature dimensionality reduction. In this embodiment, five processing methods, namely SG (convolution smoothing), MSC (multiplicative scatter correction), SNV (standard normal variate transformation), diff1 (first-order difference), and diff2 (second-order difference), are mainly used to perform feature preprocessing on the spectral data.

[0064] Each sample in this embodiment corresponds to a piece of MIR spectral data. The full spectral band is substituted for modeling, and the accuracy of the model is compared and analyzed. Two algorithms and six methods (including no pretreatment) are paired one by one to determine the preliminary optimal algorithm and pretreatment combination of the model. The results are as follows in the table:

[0065] The comparison results are as follows:

[0066]

[0067] After comparison, it is found that the combination of partial least squares regression and second-order difference (PLSR+diff2) has a better overall effect, and the overfitting situation is weaker compared with the Ridge algorithm. Therefore, the preliminary optimal algorithm and preprocessing combination of the model are finally selected: PLSR+diff2.

[0068] Example 3:

[0069] Establishment of a method for detecting the content of saturated fatty acids in milk by mid-infrared spectroscopy:

[0070] 1. Division of the modeling data set

[0071]

[0072] In the division of the modeling data set in this example, 80% is the training set and 20% is the test set. The ratio of the training set to the test set is 4:1. At the same time, the training set is also called the cross-validation set, and 10-fold cross-validation is performed during the process of training the model.

[0073] 2. Manual selection process and determination of the modeling feature bands

[0074] There are many methods for selecting feature bands, mainly including two types: algorithm-based feature selection and manual feature selection. The principle of algorithm-based feature selection mainly comes from the correlation between each wave point and the reference value. Its advantages are fast speed and high efficiency, but the disadvantage is that it ignores the synergistic effect between adjacent wave points, and the idea is relatively single; the advantage of manual feature selection is that the role of bands (i.e., adjacent wave points) can be strengthened during the selection process, and at the same time, more of the original information state of the spectrum can be retained during the process of improving the model, with stronger inclusiveness and generalization ability, and accurate band selection. The disadvantage is slow selection speed and low efficiency.

[0075] In this example, the method of manual selection is used to select the feature bands, and the selection steps are as follows:

[0076] (1) Determine the basic algorithm and preprocessing combination. As can be seen from Example 2, the overall effect of the PLSR+diff2 combination is better. Therefore, the PLSR+diff2 combination is finally selected as the preliminary model algorithm and preprocessing combination for saturated fatty acids.

[0077] (2) Determine the optimal second preprocessing. During the modeling process, sometimes a single preprocessing may limit the generalization ability and optimization potential of the model. In this embodiment, it is found during the modeling process that performing a second preprocessing can better optimize the model. Therefore, the results of the combination of the preliminary model algorithm selection and preprocessing are subjected to a second preprocessing to select the optimal second preprocessing method. After comparison, it is found that using diff2 for the second preprocessing significantly improves the training set effect, and the test set effect does not decrease severely, indicating great optimization potential. Therefore, diff2 is finally selected as the optimal second preprocessing method (the results are shown in the following table).

[0078]

[0079] The manual selection process of the modeling feature bands is as follows:

[0080] (1) Segment the full band into segments with 100 wave points each, and finally divide it into 11 segments. According to the preliminary experiment, it is found that retaining some spectral regions related to water absorption can improve the accuracy of the model. Therefore, in this step, the spectral regions related to water absorption are not removed.

[0081] (2) Taking 50 wave points as a group, using the partial least squares regression algorithm, first increase or decrease a group of wave points at both ends of the critical point of the first band segment to find the optimal effect, and perform a similar operation on the second band segment based on this. Finally, after completing a round of operations on all 11 bands, the first traversal is completed.

[0082] (3) After the first traversal is completed, perform the second, third, or more manual traversals until all wave points no longer change, which is the optimal feature band.

[0083] Finally, after ten rounds of screening, the optimal results are obtained, as shown in the table:

[0084]

[0085] The finally selected feature band results are: 1412.03 cm -1 -1489.19 cm -1 、1558.63 cm -1 -1720.67 cm -1 、2106.47 cm -1 -2229.92 cm -1 、2445.97 cm -1 -2689.03 cm -1 、3016.96 cm -1 -3468.34 cm -1 、3638.09 cm -1 -3746.12 cm-1 、4101.05 cm -1 -4170.50 cm -1 、4212.94 cm -1 -4510.00 cm -1 、4618.03 cm -1 -5007.68 cm -1 There is a tolerance of two wave points before and after each section. It is found that after adding the first section of water absorption and part of the second section of water absorption area to the model, the model can achieve the optimal effect, indicating that the characteristic wave band of saturated fatty acids includes part of the water absorption area.

[0086] 3. Screening and determination of model parameters

[0087] The model parameters include the parameters of the preprocessing method and the parameters of the algorithm. There are no parameters for the diff2 preprocessing method involved in this model; the main parameter is the parameter of the partial least squares regression algorithm: the number of principal components (n_component). The comparison results of parameter selection are as follows (partial):

[0088]

[0089] According to the comparison results, the number of principal components (n_component) is finally selected as 5.

[0090] After comparative analysis, the best regression model for saturated fatty acids is: diff2 (second-order difference) + diff2 (second-order difference) + PLSR (n_component = 5) model. The correlation coefficients of the training set and the test set are 0.8732 and 0.8797 respectively; the root mean square errors of the training set and the test set are 4.7074 and 4.5252 respectively.

[0091] Example 4:

[0092] Application of the rapid batch detection method of mid-infrared spectrum MIR for saturated fatty acids in milk:

[0093] The established best regression model for saturated fatty acids (diff2 (second-order difference) + diff2 (second-order difference) + PLSR (n_component = 5)) is used to predict 5 randomly selected external milk samples (not one of the 232 experimental materials), and the prediction results are compared with the true values.

[0094] 1. The characteristic wave bands in the infrared spectrum of the milk sample are: 1412.03 cm -1 -1489.19 cm -1 、1558.63 cm -1 -1720.67 cm -1 、2106.47 cm-1 -2229.92 cm -1 and 2445.97 cm -1 -2689.03 cm -1 and 3016.96 cm -1 -3468.34 cm -1 and 3638.09 cm -1 -3746.12 cm -1 and 4101.05 cm -1 -4170.50 cm -1 and 4212.94 cm -1 -4510.00 cm -1 and 4618.03 cm -1 -5007.68 cm -1 in the MIR data;

[0095] At the same time, the true values of saturated fatty acids in the same batch of milk were detected by gas chromatography.

[0096] 2. Substitute the obtained MIR data into the diff2 (second-order difference) + diff2 (second-order difference) + PLSR (n_component = 5) model constructed in Example 3, and the predicted results of the saturated fatty acid content can be output.

[0097] The results predicted by this model are very close to the true results (as shown in the following table), so the accuracy of this model is relatively high and can be used to predict the saturated fatty acid content of milk.

[0098]

[0099] References

[0100] [1] KEYS A, ANDERSON J T, GRANDE F. Prediction of serum-cholesterol responses of man to changes in fats in the diet[J]. Lancet, 1957, 273(7003): 959 - 966.

[0101] [2] AHRENS E H. The diet-heart question in 1985: has it really been settled[J]. Lancet, 1985, 1(8437): 1085 - 1087.

[0102] [3]HEILESON J L. Dietary saturated fat and heart disease: a narrative review[J]. Nutr Rev, 2020, 78(6): 474 - 485.

[0103] [4]FLOCK M R, KRIS-ETHERTON P M. Diverse physiological effects of long-chain saturated fatty acids: implications for cardiovascular disease[J]. Curr Opin Clin Nutr Metab Care, 2013, 16(2): 133 - 140.

[0104] [5]VERMA P, GHOSH A, RAY M, et al. Lauric Acid Modulates Cancer-Associated microRNA Expression and Inhibits the Growth of the Cancer Cell[J]. Anticancer Agents Med Chem, 2020, 20(7): 834 - 844.

[0105] [6]ZHANG X S, XUE C Y, XU Q, et al. Caprylic acid suppresses inflammation via TLR4 / NF-κB signaling and improves atherosclerosis in ApoE-deficient mice[J]. Nutr Metab(Lond), 2019, 16: 40.

[0106] [7]HUANG W C, TSAI T H, CHUANG L T, et al. Anti-bacterial and anti-inflammatory properties of capric acid against Propionibacterium acnes: a comparative study with lauric acid[J]. J Dermatol Sci, 2014, 73(3): 232 - 240.

[0107] [8] MCNABNEY S M, HENAGAN T M. Short Chain Fatty Acids in the Colon and Peripheral Tissues: A Focus on Butyrate, Colon Cancer, Obesity and Insulin Resistance[J]. Nutrients, 2017, 9(12):1348.

[0108] [9] DE OLIVEIRA OTTO M C, MOZAFFARIAN D, KROMHOUT D, et al. Dietary intake of saturated fat by food source and incident cardiovascular disease: the Multi-Ethnic Study of Atherosclerosis[J]. Am J Clin Nutr, 2012, 96(2):397 - 404.

[0109]

[10] A M O'Donnell-Megaro, D M Barbano, D E Bauman. Survey of the fatty acid composition of retail milk in the United States including regional and seasonal variations[J]. J Dairy Sci,2011, 94:59–65.

[0110]

[11] National Standard of the People's Republic of China. GB 5009.168—2016 National Food Safety Standard Determination of Fatty Acids in Foods[S]. Beijing: Standards Press of China. 2016.

[0111]

[12] Deng L, Zou Q, Liu B, et al. Fatty acid positional distribution in colostrum and mature milk of women living in Inner Mongolia, North Jiangsu and Guangxi of China[J]. Food Funct, 2018, 9(8):4234 - 4245.

[0112]

[13] Ten-Domenech I, Beltran-Iturat E, Herrero-Martinez JM, et al. Triacylglycerol Analysis in Human Milk and Other Mammalian Species: Small-Scale Sample Preparation, Characterization, and Statistical Classification Using HPLC-ELSD Profiles[J]. J Agric Food Chem, 2015, 63(24): 5761-70.

[0113]

[14] Zou X, Huang J, Jin Q, et al. Lipid composition analysis of milk fats from different mammalian species: potential for use as human milk fat substitutes[J]. J Agric Food Chem, 2013, 61(29): 7070-80.

Claims

1. A rapid batch detection method for saturated fatty acids in milk by mid-infrared spectroscopy, comprising the following steps: 1) The characteristic bands in the infrared spectrum of the milk sample are: 1412.03 cm -1 -1489.19 cm -1 、1558.63 cm -1 -1720.67 cm -1 、2106.47 cm -1 -2229.92 cm -1 、2445.97 cm -1 -2689.03 cm -1 、3016.96 cm -1 -3468.34 cm -1 、3638.09 cm -1 -3746.12 cm -1 、4101.05 cm -1 -4170.50 cm -1 、4212.94 cm -1 -4510.00 cm -1 、4618.03 cm -1 -5007.68 cm -1 in the MIR data; 2) Substitute the obtained MIR data into the diff2+diff2+PLSR model to output the predicted result of the saturated fatty acid content; The diff2 is the second-order difference, diff1 is the first-order difference, and PLSR is partial least squares regression, where the main component is 5.

2. The method according to claim 1, characterized in that: In step 1, there is a tolerance of two wave points before and after each band.

3. Use of the method according to claim 1 for detecting the content of saturated fatty acids in milk.

Images