Methods for determining wheat milling yield, B / M ratio, semolina production rate, milling score, and the brightness or redness of wheat flour.

Abstract

This invention provides a quantitative method that allows for the simple evaluation of milling properties and flour properties from wheat grains without milling. [Solution] For each of the several types of wheat grains used as calibration curve samples, the optical spectrum is obtained by transmission using light with wavelengths including the near-infrared band. For each type of wheat grain, the milling yield, B / M ratio, semolina production rate, milling score, and flour brightness or redness are quantified using known quantitative methods. The acquired optical spectrum information and the quantitative information obtained by known quantitative methods are analyzed using PLS regression analysis to set up a regression equation that shows the relationship between the optical spectrum information and the quantitative information obtained by known quantitative methods. The optical spectrum is obtained for the wheat grains by transmission using light with wavelengths including the near-infrared band, and the milling yield, B / M ratio, semolina production rate, milling score, and flour brightness or redness are calculated based on the set regression equation.

Description

[Technical Field]

[0001] This invention relates to a method for quantifying the milling characteristics of wheat grains, or the characteristics of wheat flour obtained by milling those wheat grains, using spectroscopic methods. [Background technology]

[0002] Wheat is milled into flour and processed into various foods. Flour quality, including its brightness and redness, affects the quality of flour products such as bread and noodles. Therefore, these analytical values ​​are crucial data for breeders and end-users. These characteristics are measured after the wheat grains have been milled to produce flour.

[0003] Furthermore, the characteristics related to the milling of wheat grains are also important in evaluating the value of wheat. These milling characteristics include milling yield, B / M ratio, semolina production rate, and milling score. These values ​​indicate how efficiently the desired flour can be obtained from the wheat grains during milling. Milling companies use these analytical values ​​as indicators when selecting the varieties to use, making them important not only for milling companies but also for breeders. Obtaining these values ​​requires actually performing milling.

[0004] Non-patent document 1 and patent document 1 describe methods for estimating milling yield without performing milling. Non-patent document 1 describes a method for estimating milling yield by acquiring an optical spectrum in a broad wavelength range of 400 to 2500 nm, including the entire near-infrared region, using wheat seeds with a reflective near-infrared spectrometer. Patent document 1 describes a method for acquiring a multispectral image using halved wheat seeds and estimating milling yield from the obtained image data.

[0005] Other examples of analysis using near-infrared spectroscopy include Non-Patent Document 2, Patent Document 2, and Patent Document 3. Non-Patent Document 2 describes the estimation of α-amylase activity and gelatinization characteristics of wheat using near-infrared spectroscopy, but it is shown that estimation is difficult using only light spectral data. Patent Documents 2 and 3 describe the estimation of fatty acid ester content and frying oil composition from the near-infrared light spectrum of oils and fats.

[0006] Non-Patent Document 3 is a fundamental paper concerning non-destructive measurement using near-infrared spectroscopy. Non-Patent Document 3 suggests the possibility of new spectroscopic analysis methods, such as component analysis in grains, by using near-infrared spectroscopy to statistically process the complexity of the spectrum, unlike conventional spectral analysis which involves "spectral measurement → band assignment → qualitative and quantitative analysis". Non-Patent Document 4 discloses that when 36 samples of brown rice with known moisture, protein, fatty acid content, and amylose were measured using a transmission near-infrared spectrometer pre-calibrated by the instrument manufacturer using the PLS method, the measured values ​​for moisture and protein were accurate enough to be used as a substitute for reference analysis, but the measured values ​​for fatty acid content and amylose were not accurate enough to be used as a substitute for reference analysis. Furthermore, Patent Document 4 describes a method for quantifying easily digestible or indigestible proteins in rice grains using near-infrared spectroscopy. [Prior art documents] [Patent Documents]

[0007] [Patent Document 1] Japanese Patent Publication No. 2024-60661 [Patent Document 2] Japanese Patent Publication No. 2020-112478 [Patent Document 3] Japanese Patent Publication No. 2018-205226 [Patent Document 4] Japanese Patent Publication No. 2023-114058 [Non-patent literature]

[0008] [Non-Patent Document 1] Walker et al. 2023, Near-infrared spectroscopy enables quality selection in wheat breeding, Cereal Chemistry, vol.100, 1347-1356, DOI: 10.1002 / cche.10717 [Non-Patent Document 2] Natsuga 1999 "Quality Measurement of Grains by Near-Infrared Spectroscopy: (Part 4) Quality Measurement of Grains by Near-Infrared Spectroscopy," Journal of the Faculty of Agriculture, Hokkaido University, Vol. 22, No. 3, 127-168, http: / / hdl.handle.net / 2115 / 12188 [Non-Patent Document 3] Ozaki et al., "Non-destructive Measurement by Near-Infrared Spectroscopy," 1995, Journal of the Japan Society for Infrared Science, Vol. 5, No. 2, pp. 78-88. (Japanese Patent Publication No. 56-501215, "Near-Infrared Quantitative Analyzer") [Non-Patent Document 4] Kawamura et al., "Accuracy of Rice Component Measurement Using Near-Infrared Transmission Analyzers and its Improvement," 2002, Journal of the Japanese Society of Agricultural Machinery, Vol. 64, No. 1, pp. 120-126. [Overview of the project] [Problems that the invention aims to solve]

[0009] Analyzing the characteristics of flour milling using a test mill, and measuring the characteristics of flour obtained through test milling, is time-consuming and labor-intensive. There are two methods for milling using test mills: the large Bühler test mill (product name) and the small Bravender Jr. test mill (product name). When using the Bühler test mill, approximately 1.5 kg of sample is required, and it takes 1.5 hours for two people to mill one sample. Similarly, when using the Bravender Jr. test mill, approximately 150 g of sample is required, and it takes 45 minutes for one person to mill one sample. In either method, milling incurs costs such as time and labor, so there is a need for an analytical method that can easily obtain milling characteristic values ​​from wheat grains without milling.

[0010] In the method for estimating the milling yield described in Non-Patent Document 1, white-grained hard wheat is used as a test material. An optical spectrum in a wide wavelength range of 400 to 2500 nm including the entire near-infrared region is obtained by the reflection method. In Japan, red-grained soft and hard wheat is generally cultivated, and the seed coat color and endosperm color of white-grained hard wheat are significantly different. Therefore, the absorbance and spectrum are also different, and it is difficult to directly use the calibration curve of Non-Patent Document 1 for wheat varieties including those grown and cultivated in Japan.

[0011] In the method for estimating the milling yield described in Patent Document 1, an expensive device for acquiring a multi-spectral image is required, and there are costs for pretreatment such as cutting the seeds in half and fixing them.

[0012] There is no reported example of a technique that can estimate the B / M ratio, semolina production rate, milling score, lightness of wheat flour, or redness of wheat flour for wheat grains by near-infrared spectroscopy.

[0013] In Non-Patent Document 2, the α-amylase activity and gelatinization characteristics of wheat were estimated by near-infrared spectroscopy, but it was shown that it was difficult to make such an estimation only with optical spectrum data. In Non-Patent Document 4, it was disclosed that the fatty acid value and amylose were not accurate enough to be used in place of the reference analysis. As shown in Non-Patent Document 2 and Non-Patent Document 4, since it cannot always be expected that analysis can be performed from the combination of the analysis target component and the analyte, it is considered difficult to directly use the calibration curve of Non-Patent Document 1 for wheat varieties including those grown and cultivated in Japan.

[0014] In the invention described in Patent Document 4, near-infrared spectroscopy is used, but the target is the protein of rice. Rice is completely different from wheat in terms of components, grain shape, surface properties, etc., and the absorbance and spectrum are significantly different.

[0015] An object of this invention is to provide a quantitative method that can simply evaluate the characteristics related to milling and the characteristics of wheat flour from wheat grains without milling. [Means for solving the problem]

[0016] To solve the above problems, the method for quantifying wheat milling yield, B / M ratio, semolina production rate, milling score, lightness of wheat flour, or redness of wheat flour according to this invention involves obtaining the optical spectrum of each type of wheat grain, which is a sample for creating a calibration curve, using light with wavelengths including the near-infrared band, by transmission method. For each type of wheat grain in the sample used to create the calibration curve, the milling yield, B / M ratio, semolina production rate, milling score, and the brightness or redness of the flour were quantified using known quantitative methods. The acquired optical spectrum information and the quantitative information obtained using known quantitative methods were analyzed using PLS regression analysis to set up a regression equation that shows the relationship between the optical spectrum information and the quantitative information obtained using known quantitative methods. For the wheat grain sample being measured, the optical spectrum is obtained using a transmission method with wavelengths including the near-infrared band, and the milling yield, B / M ratio, semolina production rate, milling score, and the brightness or redness of the flour are calculated based on a set regression equation.

[0017] For each type of wheat grain used in the calibration curve creation sample, the milling yield, B / M ratio, semolina production rate, milling score, and the brightness and redness of the flour obtained from those wheat grains were all quantified using known quantitative methods, and regression equations corresponding to each indicator were set. The optical spectrum of the wheat grain sample to be measured may be obtained by transmission using light with wavelengths including the near-infrared band, and the milling yield, B / M ratio, semolina production rate, milling score, and the brightness and redness of the flour obtained from the wheat grain may all be calculated.

[0018] It is preferable to perform PLS regression analysis using optical spectrum information from wheat grains used as calibration curve samples and wheat grains used as the measurement sample, with a lower limit of 800 nm to 900 nm and an upper limit of 1050 nm to 1150 nm. [Effects of the Invention]

[0019] According to the method for quantifying the milling yield, B / M ratio, semolina production rate, milling score, and the brightness and redness of wheat flour obtained from the wheat grains of this invention, these indicators can be quantified simply and quickly. [Brief explanation of the drawing]

[0020] [Figure 1] This flowchart outlines the milling yield, B / M ratio, semolina production rate, and milling score of wheat grains, as well as the methods for quantifying the brightness and redness of the flour obtained from those wheat grains. [Figure 2] This is a conceptual diagram showing the powder obtained by milling using a Bühler test mill. [Figure 3] This is a conceptual diagram showing the powder obtained by milling using the Brabender Jr. Test Mill. [Figure 4] This is a conceptual diagram showing the optical spectrum used in PLS regression analysis. [Figure 5] This graph shows an example of the optical spectrum of a wheat grain sample used in the analysis of milling yield. [Figure 6] This graph shows an example of the sample spectrum of wheat grains used in the analysis of milling yield after Detrend treatment, SNV treatment, and first derivative analysis. [Figure 7] This is a scatter plot comparing the calculated milling yield based on a calibration curve obtained by performing PLS regression on all samples with the measured values ​​obtained from milling. [Figure 8] This graph shows the regression coefficients of the calibration curve obtained by performing PLS regression on all samples regarding milling yield. [Figure 9] This is a scatter plot comparing the calculated milling yield after cross-validation with the measured value obtained from milling. [Figure 10] This scatter plot compares the calculated milling yield based on the calibration curve obtained by performing PLS regression on a calibration curve creation set with the measured values ​​obtained from milling. [Figure 11]This graph shows the regression coefficients of the calibration curve obtained by performing PLS regression on a calibration curve creation set for milling yield. [Figure 12] This is a scatter plot comparing the calculated milling yield values ​​from a calibration curve verification set based on a calibration curve with the measured values ​​obtained from milling. [Figure 13] This graph shows an example of the optical spectrum of a wheat grain sample used for B / M ratio analysis. [Figure 14] This graph shows an example of the sample spectrum of wheat grains used in the analysis of the B / M ratio using first derivative. [Figure 15] This is a scatter plot comparing the calculated B / M ratio based on a calibration curve obtained by performing PLS regression on all samples with the measured value obtained from milling. [Figure 16] This graph shows the regression coefficients of the calibration curve obtained by performing PLS regression on all samples for the B / M ratio. [Figure 17] This is a scatter plot comparing the calculated B / M ratio obtained using cross-validation with the measured value obtained through milling. [Figure 18] This is a scatter plot comparing the calculated B / M ratio based on the calibration curve obtained by performing PLS regression on a calibration curve creation set with the measured value obtained from milling. [Figure 19] This graph shows the regression coefficients of the calibration curve obtained by performing PLS regression on a calibration curve creation set for the B / M ratio. [Figure 20] This is a scatter plot comparing the calculated B / M ratio values ​​from a calibration curve verification set based on a calibration curve with the measured values ​​obtained through milling. [Figure 21] This graph shows an example of the optical spectrum of a wheat grain sample used for the analysis of semolina formation rate. [Figure 22] This graph shows an example of the sample spectrum of wheat grains used in the analysis of semolina production rate using first derivative. [Figure 23] This is a scatter plot comparing the calculated semolina production rate based on a calibration curve obtained by performing PLS regression on all samples with the measured value obtained from milling. [Figure 24]This graph shows the regression coefficients of the calibration curve obtained by performing PLS regression on all samples regarding semolina formation rate. [Figure 25] This is a scatter plot comparing the calculated semolina production rate obtained using cross-validation with the measured value obtained through milling. [Figure 26] This scatter plot compares the calculated semolina production rate based on a calibration curve obtained by performing PLS regression on a calibration curve creation set with the measured value obtained from milling. [Figure 27] This graph shows the regression coefficients of the calibration curve obtained by performing PLS regression on a calibration curve creation set for semolina formation rate. [Figure 28] This is a scatter plot comparing the calculated semolina production rate of a calibration curve verification set based on a calibration curve with the measured value obtained from milling. [Figure 29] This graph shows an example of the optical spectrum of a wheat grain sample used for milling score analysis. [Figure 30] This graph shows an example of the sample spectrum of wheat grains used in the analysis of milling scores after SNV processing, Detrend processing, and first derivative analysis. [Figure 31] This scatter plot compares the calculated milling score, based on a calibration curve obtained by performing PLS regression on all samples, with the measured value obtained from milling. [Figure 32] This graph shows the regression coefficients of the calibration curve obtained by performing PLS regression on all samples for the milling score. [Figure 33] This is a scatter plot comparing the calculated milling score obtained using cross-validation with the measured value obtained through milling. [Figure 34] This scatter plot compares the calculated milling score, based on a calibration curve obtained by performing PLS regression on a calibration curve creation set, with the measured value obtained from milling. [Figure 35] This graph shows the regression coefficients of the calibration curve obtained by performing PLS regression on a calibration curve creation set for the milling score. [Figure 36]This is a scatter plot comparing the calculated milling score of a calibration curve verification set based on a calibration curve with the measured value obtained through milling. [Figure 37] This graph shows an example of the optical spectrum of a wheat grain sample used for the analysis of wheat flour lightness (L*). [Figure 38] This graph shows an example of the sample spectrum of wheat grains used in the analysis of flour lightness (L*) by second differentiation. [Figure 39] This scatter plot compares the calculated values ​​of flour lightness (L*) based on a calibration curve obtained by performing PLS regression on all samples with the measured values ​​obtained from milling and analysis. [Figure 40] This graph shows the regression coefficients of the calibration curve obtained by performing PLS regression on all samples for wheat flour lightness (L*). [Figure 41] This is a scatter plot comparing the calculated lightness (L*) values ​​of wheat flour obtained through cross-validation with the measured values ​​obtained from milling and analysis. [Figure 42] This scatter plot compares the calculated values ​​of flour lightness (L*) based on a calibration curve obtained by performing PLS regression on a calibration curve creation set, with the measured values ​​obtained by milling and analyzing the flour. [Figure 43] This graph shows the regression coefficients of the calibration curve obtained by performing PLS regression on a calibration curve creation set for flour lightness (L*). [Figure 44] This is a scatter plot comparing the calculated values ​​of flour lightness (L*) from a calibration curve verification set based on a calibration curve with the measured values ​​obtained from milling and analysis. [Figure 45] This graph shows an example of the optical spectrum of a wheat grain sample used for the analysis of the reddish color (a*) of wheat flour. [Figure 46] This graph shows an example of the sample spectrum of wheat grains used in the analysis of the redness (a*) of wheat flour by second differentiation. [Figure 47] This scatter plot compares the calculated redness (a*) of wheat flour, based on a calibration curve obtained by performing PLS regression on all samples, with the measured values ​​obtained from milling and analysis. [Figure 48]This graph shows the regression coefficients of the calibration curve obtained by performing PLS regression on all samples for the redness (a*) of wheat flour. [Figure 49] This is a scatter plot comparing the calculated redness (a*) of wheat flour obtained through cross-validation with the measured values ​​obtained from milling and analysis. [Figure 50] This scatter plot compares the calculated redness (a*) of wheat flour, based on a calibration curve obtained by performing PLS regression on a calibration curve creation set, with the measured values ​​obtained from milling and analysis. [Figure 51] This graph shows the regression coefficients of the calibration curve obtained by performing PLS regression on a calibration curve creation set for the redness (a*) of wheat flour. [Figure 52] This is a scatter plot comparing the calculated values ​​of redness (a*) of wheat flour from a calibration curve verification set based on a calibration curve with the measured values ​​obtained from milling and analysis. [Modes for carrying out the invention]

[0021] The embodiments for carrying out the present invention will now be described. In the quantitative method of the present invention, as will be described later, a regression equation is set that shows the relationship between optical spectrum information and the quantitative determination of various indicators related to the characteristics of wheat grains, and calculations are performed based on that regression equation. Therefore, it is preferable to have an analytical device equipped with a computing device (CPU) and memory, and capable of installing and executing a program that performs calculations based on the determined regression equation. Such spectroscopic analyzers are commercially available. Here, an example will be given using the Inframatic 9500 (PerkinElmer) as the spectroscopic analyzer.

[0022] Figure 1 is a flowchart outlining the process of setting up regression equations for determining the milling yield, B / M ratio, semolina production rate, milling score, and the brightness and redness of flour obtained from wheat grains. First, prepare a sufficient number of samples, i.e., test materials, to create a calibration curve, i.e., to set up the regression equation. These samples are wheat grains.

[0023] The study consists of 749 samples of common wheat varieties cultivated from Hokkaido to Kyushu (such as Norin 61, Fukusaya, Chikugoizumi, Shirogane Komugi, and Setokirara) and breeding lines derived from their crosses. These samples were cultivated in test fields between 2017 and 2021. All samples were stored in airtight containers at 4-8°C until each analysis was performed. All samples were dried and prepared raw wheat samples after threshing, and no pretreatment such as milling or polishing was required for use in near-infrared spectroscopy. The number of sets used varied depending on the evaluation item, so the number of sets used for each analysis item is shown separately. In this example, samples were prepared for calibration curve creation and for calibration curve verification.

[0024] For each type of sample in the calibration curve creation set, the optical spectrum is acquired, and further quantitative analysis is performed using a known analytical method other than the near-infrared spectroscopy of the present invention (measurement of conventional analytical values), and the optical spectrum information and the conventional analytical values ​​are stored in association. For each type of sample, the sample may be divided into two parts, with the optical spectrum acquired using one part and the conventional quantitative analysis performed using the other part. Also, since the acquisition of the optical spectrum can be performed non-destructively, the conventional quantitative analysis may be performed using the same sample after the optical spectrum has been acquired.

[0025] Similarly, optical spectrum acquisition and conventional analytical value measurement are performed for the calibration curve verification set. Alternatively, the samples for the calibration curve creation set and the calibration curve verification set may be separated beforehand, and then optical spectrum acquisition and quantitative analysis using methods other than spectroscopic analysis may be performed. Alternatively, both samples may be treated identically at this stage, and then the results may be used to separate the data for calibration curve creation and the data for calibration curve verification.

[0026] Quantitative analysis using conventional methods other than near-infrared spectroscopy will utilize already established analytical methods. Any method that can accurately measure the target indicator (evaluation item) is acceptable, and it may be a time-consuming or costly method. The evaluation items for wheat grains to which this invention applies are a total of six items: four items related to milling characteristics (milling yield, B / M ratio, semolina production rate, and milling score) and two items related to flour characteristics (flour brightness (L*) and flour redness (a*)).

[0027] Milling yield is the ratio of wheat flour (straight flour) obtained by milling to the amount of wheat flour + bran. A higher milling yield is considered better because it means that more wheat flour is obtained from the wheat grains. Milling yield can be obtained using either the Bühler Test Mill (product name) or the Bravender Jr. Test Mill (product name).

[0028] The B / M ratio represents the ratio of the amount of brake (B) flour and the amount of middle (M) flour, which are two types of flour obtained in the Bühler test mill. It is calculated as "amount of brake (B) flour ÷ amount of middle (M) flour × 100". The B / M ratio is a reference for the hardness of the grain; hard wheat has a low B / M ratio, and soft wheat has a high B / M ratio. The B / M ratio can only be obtained through milling using the Bühler test mill.

[0029] The semolina production rate indicates the rate at which semolina, a mass of endosperm, is produced. In the Bühler test mill, the grain is coarsely ground by the brake rolls (B) to produce semolina, and then further ground by the reduction rolls (M), with the powder that passes through the sieve being collected. The semolina production rate indicates the proportion of the total fraction of the milling process that is sent to the reduction rolls. A higher semolina production rate is desirable because a higher semolina yield improves the efficiency of wheat flour production. The semolina production rate can only be obtained from milling using the Bühler test mill.

[0030] The milling score is an evaluation value of milling performance that takes ash content into account. Since ash content reduces the color of flour, a lower ash content is desirable. Milling yield can be increased by milling down to the tissue at the base of the seed coat, but since the base of the seed coat contains a lot of ash, the resulting flour will have a high ash content. Even if the milling yield is high, if the resulting flour has a high ash content, it cannot be said that the milling performance is high. Therefore, the milling score, which takes into account the milling yield and ash content, is calculated using the following formula. Milling score = 100 - {(80 - milling yield) + 50 (straight flour ash content - 0.30)} Here, "straight flour" refers to all wheat flour obtained through milling, and the ash content value of straight flour is the value obtained by near-infrared spectroscopy analysis using straight flour. This near-infrared spectroscopy analysis of ash content is performed using a procedure similar to that shown in the flowchart in Figure 1, and it has been confirmed that there is a good correspondence with values ​​obtained by already established and known analytical methods. A higher milling score is desirable because it indicates that more high-quality wheat flour can be obtained. In this invention, the milling score was calculated only for milling using the Bühler test mill.

[0031] Two characteristics of wheat flour are its lightness (L*) and redness (a*). These indicators are analytical values ​​related to the color of wheat flour, and flour with high lightness and low redness is preferred. Lightness and redness greatly affect the color of wheat products such as bread and noodles, so these analytical values ​​are considered important not only by milling companies but also by secondary processors. Furthermore, since lightness and redness are traits that depend not only on the milling method but also on the variety, they are also important analytical values ​​for breeders. To analyze lightness and redness, it is necessary to actually mill the flour, just like with milling characteristics. In the case of lightness and redness analysis, it is also necessary to suspend the flour in water after milling, place it in a petri dish, and measure it with a colorimeter, so the analytical cost is increased in addition to the milling cost. Therefore, there is a need for an analytical method that can easily obtain lightness and redness values ​​from wheat seeds without milling. The lightness (L*) and redness (a*) of wheat flour can be obtained by milling using either the Bühler test mill or the Brabender Jr. test mill.

[0032] Further details will be provided regarding the indicators for wheat grain milling. Conventional analysis will utilize established analytical methods. Any method that can accurately measure the results is acceptable, even if it is time-consuming or costly. Here, milling and analysis were performed based on the official methods established by the AACC (American Association of Cereal Chemists) (AACC International (2015) Approved Methods of Analysis (11th Ed.): 26-10.02 Experimental Milling: Introduction, Equipment, Sample Preparation, and Tempering; 26-21.02 Experimental Milling-Buhler Method for Hard Wheat; 26-31.01 Experimental Milling-Buhler Method for Soft Wheat Straight-Grade Flour; 26-50.01 Brabender Quadrumat Jr. (Quadruplex) Method).

[0033] First, we will describe the milling and analysis method using the Bühler test mill. Figure 2 is a conceptual diagram showing the powder obtained by milling using the Bühler test mill. To increase milling efficiency, 1.5 to 2.0 kg of wheat grains were subjected to a 24-hour water treatment (tempering treatment). The target moisture content at this time was 14.5% for soft wheat, 15% for medium wheat, and 16% for hard wheat. After tempering, the wheat grains were fed into the Bühler test mill by an auto-feeder at a flow rate of approximately 90 g / min and milled. During milling, the indoor environment was adjusted to a temperature of 23°C and a humidity of 60%. In milling using the Bühler test mill, two types of flour are obtained: from the brake (B) side and the middle (M) side. Each of these flours is further classified, yielding a total of six types of flour and two types of bran: three types (1B-3B) and large-sized bran (large bran) from the brake (B) side, and three types (1M-3M) and small-sized bran (small bran) from the middle (M) side. The weight of each type of flour and bran obtained was measured. The obtained flours were mixed starting from the fraction with the least bran contamination (1B-1M → 2B-2M → 3B-3M, in order of decreasing contamination) until a milling yield of 60% was achieved, and 60% flour was prepared and sieved. Straight flour, a mixture of all types of flour, was also prepared. Furthermore, the ash content of the straight flour was measured by near-infrared spectroscopy. The milling characteristics analysis values ​​were calculated from the weight and ash content of each type of wheat flour and bran using the following formula.

[0034] Milling yield = (1B flour weight + 2B flour weight + 3B flour weight + 1M flour weight + 2M flour weight + 3M flour weight) / (1B flour weight + 2B flour weight + 3B flour weight + 1M flour weight + 2M flour weight + 3M flour weight + large bran weight + small bran weight) x 100

[0035] B / M ratio = (1B powder weight + 2B powder weight + 3B powder weight) / (1M powder weight + 2M powder weight + 3M powder weight) × 100

[0036] Semolina production rate = (Weight of 1M flour + Weight of 2M flour + Weight of 3M flour + Weight of small bran) / (Weight of 1B flour + Weight of 2B flour + Weight of 3B flour + Weight of 1M flour + Weight of 2M flour + Weight of 3M flour + Weight of large bran + Weight of small bran) × 100

[0037] Milling score = 100 - {(80 - milling yield) + 50 (straight flour ash content - 0.30)}

[0038] Next, we will describe the milling and analysis method using the Bravender Jr. test mill. Figure 3 is a conceptual diagram showing the powder obtained by milling using the Bravender Jr. test mill. 150g of wheat grains were tempered as described above. After tempering, the wheat seeds were fed into the Bravender Jr. test mill by an auto-feeder at a flow rate of approximately 22g / min and milled. During milling, the indoor environment was adjusted to a temperature of 23°C and a humidity of 60%. Milling with the Bravender Jr. test mill yielded three types of wheat flour (flour 1 to flour 3) and one type of bran. The weight of each type of wheat flour and bran was measured. The obtained wheat flour was mixed starting from the fraction with the least bran contamination (flour 1 to flour 3, in order of least contamination) until a milling yield of 60% was achieved, and the 60% flour was sieved. The milling yield was calculated from the weight of each type of wheat flour and bran as follows. Milling yield = (1 weight of flour + 2 weight of flour + 3 weight of flour) / (1 weight of flour + 2 weight of flour + 3 weight of flour + bran weight) × 100

[0039] Next, we will explain in detail the indicators related to the properties of wheat flour. Conventional analysis methods utilize already established analytical techniques. Any method that can accurately measure the results is acceptable, even if it is time-consuming or costly. Here, the analysis was performed using a spectrophotometer (Minolta, CM-3500d). Analysis was performed using 60% powder obtained by milling with a Bühler test mill or a Brabender Jr. test mill. 6g of 60% powder was measured into a mortar, 10ml of distilled water was added, and the mixture was stirred for 90-120 seconds by tapping the sample in the mortar. After stirring, the resulting slurry was placed in a Petri dish-type cell, and it was confirmed that no air bubbles were present in the cell. Immediately thereafter, the cell was set in a spectrophotometer, and two measurements were taken with the specular reflection removal mode and measurement diameter φ30mm to determine the sample average values ​​for lightness L* and redness a*.

[0040] Near-infrared spectral data was acquired using the PerkinElmer Inframatic 9500 (IM9500) near-infrared spectrometer. By feeding raw wheat into the sample inlet of this instrument (for samples of approximately 200g or more), or by filling a dedicated cuvette with raw wheat and inserting it into the sample inlet (for samples of only a few tens of grams), near-infrared spectral data can be obtained in approximately 30 seconds. Since this instrument acquires near-infrared spectral data non-destructively, the sample can be recovered without modification. Furthermore, because this instrument is a transmission type, it can acquire near-infrared spectral data from within the raw wheat grain. Since wheat flour quality is largely influenced by the endosperm rather than the outer husk, near-infrared spectral data from within the raw wheat grain is crucial for assessing wheat flour quality.

[0041] This section explains how calibration curves were determined. For all evaluation items, two calibration curves were created: one using all sample data for calibration curve creation, and another using 3 / 4 of the sample data for calibration curve creation and the remaining 1 / 4 for calibration curve evaluation. After sorting the sample data by analytical value, data picked every four points were designated as calibration curve validation sets, and the remaining data were designated as calibration curve creation sets. After performing the necessary preprocessing for each item on the spectra obtained by Inframatic 9500, calibration curves were created using PLS (Partial least squares) regression with the required wavelength range from 570nm to 1100nm spectral data obtained by Inframatic 9500. When all sample data was used for calibration curve creation, evaluation was performed using cross-validation. When the sample data was split into calibration curve creation and calibration curve evaluation sets, the calibration curve created using the calibration curve creation set was fitted with the calibration curve evaluation data for evaluation.

[0042] Figure 4 is a conceptual diagram showing the optical spectrum used for PLS regression analysis. The near-infrared wavelength range is generally considered to be 700 to 2500 nm (General Rules for Near-Infrared Spectroscopic Analysis JIS K 0134), but in this invention, a relatively short wavelength band in the near-infrared is used in the optical spectrum information used for PLS regression analysis, and it is preferable that the upper limit is 1050 nm to 1150 nm. In particular, 1100 nm is preferred. Furthermore, light in the region close to visible light is also included, and it is preferable that the lower limit is 800 nm to 900 nm. It has been found that by using optical spectrum information in this band, good calculations with a high correlation to measured values ​​can be performed. Regarding the lower limit, 850 nm is particularly preferred for milling yield, B / M ratio, semolina production rate, and milling score, while particularly favorable correlations were obtained with 820 nm for the brightness of wheat flour and 825 nm for the redness of wheat flour.

[0043] Examples of each evaluation item are shown below.

[0044] An example of evaluating milling yield is described. In this example, data obtained from milling using a Bühler test mill and a Brabender Jr. test mill were used. First, data for creating a calibration curve is selected. In this example, spectrum acquisition and analysis by conventional measurement methods were performed on 749 samples. To ensure a wide distribution of milling yield values, data from 562 samples were selected for calibration curve creation, and the remaining 187 samples were used for calibration curve verification. Figure 5 is a graph showing an example of the optical spectrum of wheat grain samples, showing all absorbance spectra of the samples used for calibration curve creation and calibration curve verification. Here, the optical spectrum of the calibration curve creation set is shown in gray, and the optical spectrum of the calibration curve verification set is shown in black. The near-infrared spectrum collected by the Inframatic 9500 has 1061 wavelength points at 0.5 nm intervals in the wavelength range of 570 to 1100 nm.

[0045] This section describes the case where all data is used without splitting the calibration curve creation set and the calibration curve validation set. Using the near-infrared spectra of wheat collected with Inframatic 9500 and the milling yield values ​​obtained from actual milling, a calibration curve was created to determine the milling yield from the spectra, and evaluation was performed using cross-validation. After detrend processing, SNV processing, and first derivative analysis were performed on the spectra obtained with Inframatic 9500, a calibration curve was created using PLS regression in the wavelength range of 850 nm to 1100 nm (Figure 6).

[0046] Figure 7 is a scatter plot comparing the calculated milling yield (IM9500 predicted value) based on the calibration curve, which was determined by performing PLS regression on all 749 samples without separating the calibration curve creation and calibration curve verification, with the measured value obtained from milling. The accuracy during calibration curve creation was R 2 The coefficient of determination was 0.7506, and the standard error (SEC) was 2.83. Figure 8 is a graph showing the calibration curve at this time. It shows the regression coefficients (vertical axis) for each wavelength (horizontal axis) in the obtained calibration curve.

[0047] Figure 9 is a scatter plot comparing the calculated milling yield (IM9500 predicted value) after cross-validation (tolerance verification) with the measured value obtained from milling. The accuracy in cross-validation is R 2 The result was 0.7238, and the SECV (standard error) was 2.98.

[0048] Next, we will explain the case where the samples are divided into two groups: a calibration curve creation set (562 points) and a calibration curve verification set (187 points). First, a calibration curve is created using the calibration curve creation set, and then the calibration curve verification set is evaluated. The method of dividing the sample set was to rearrange the sample data in the order of the already obtained milling yield values, and then extract data from every four samples to create a group of 187 samples for calibration curve verification. The remaining 562 samples were then used as the calibration curve creation group.

[0049] Similar to the example shown in Figure 6, after performing Detrend processing, SNV processing, and first derivative analysis on the near-infrared spectrum, a calibration curve was created using PLS regression in the 850nm to 1100nm region. Figure 10 is a scatter plot showing the correlation between IM9500 predicted values ​​(vertical axis) and measured values ​​(horizontal axis) when calibration curves were created for 562 samples used for calibration curve creation. The accuracy at this time was R 2 The result was 0.7632 at SEC and 2.76 at SEC. Figure 11 is a graph showing the calibration curve at that time.

[0050] Next, we will explain the results of evaluating the 187 points used for calibration curve verification using the obtained calibration curve. Figure 12 is a scatter plot comparing the calculated milling yield (IM9500 predicted value) based on the calibration curve with the measured value obtained from milling. 2 The result was 0.7336, and the SEP (Standard Error) was 2.92.

[0051] An example of evaluating the B / M ratio is described below. In this example, data obtained from milling using a Bühler test mill was used. First, data for creating a calibration curve is selected. In this example, spectrum acquisition and measurement analysis were performed on 272 samples. To ensure a wide distribution of B / M ratio values, data from 204 samples were selected for calibration curve creation, and the remaining 68 samples were used for calibration curve verification. Figure 13 is a graph showing an example of the optical spectrum of wheat grain samples, showing all the optical spectra of the samples used for calibration curve creation and calibration curve verification. Here, the optical spectrum of the calibration curve creation set is shown in gray, and the optical spectrum of the calibration curve verification set is shown in black. The near-infrared spectrum collected by the Inframatic 9500 has 1061 wavelength points at 0.5 nm intervals in the wavelength range of 570 to 1100 nm.

[0052] This section describes the case where all data is used without splitting it into a calibration curve creation set and a calibration curve validation set. Using the near-infrared spectra of wheat collected with Inframatic 9500 and the B / M ratio values ​​obtained from actual milling, a calibration curve was created to determine the B / M ratio from the spectra, and evaluation was performed using cross-validation.

[0053] After performing the first derivative on the spectrum obtained with Inframatic 9500, a calibration curve was created using PLS regression in the wavelength range of 850 nm to 1100 nm (Figure 14). Figure 15 is a scatter plot comparing the calculated B / M ratio (IM9500 predicted value) based on a calibration curve determined by performing PLS regression on all 272 samples without separating the calibration curve creation set and the calibration curve verification set, with the measured value obtained from milling. The accuracy during calibration curve creation is R 2 The values ​​obtained were 0.8149 for SEC and 2.09 for SEC. Figure 16 is a graph showing the calibration curve at this time. It shows the regression coefficients (vertical axis) for each wavelength (horizontal axis) in the obtained calibration curve.

[0054] Figure 17 is a scatter plot comparing the calculated B / M ratio (IM9500 predicted value) after cross-validation with the measured value obtained from milling. The accuracy in cross-validation is R 2 The result was 0.7937, and 2.21 for SECV.

[0055] Next, we will explain the case where the samples are divided into two groups: a calibration curve creation set (204 points) and a calibration curve verification set (68 points). First, a calibration curve is created using the calibration curve creation set, and then the calibration curve verification set is used for evaluation. The method of dividing the sample set is as follows: the sample data is sorted according to the B / M ratio values ​​already obtained, and then data is extracted every four points to form a group of 68 samples for calibration curve verification. The remaining 204 samples are then used as the calibration curve creation group. Similar to the example shown in Figure 14, after performing the first derivative on the near-infrared spectrum, a calibration curve was created using PLS regression in the 850 nm to 1100 nm region.

[0056] Figure 18 is a scatter plot showing the correlation between IM9500 predicted values ​​(vertical axis) and measured values ​​(horizontal axis) when a calibration curve was created for 204 sample points used for calibration curve creation. The accuracy at this time is R 2 The result was 0.8118 at SEC and 2.11 at SEC. Figure 19 is a graph showing the calibration curve at that time.

[0057] Next, we will explain the results of evaluating the 68 points used for calibration curve verification using the obtained calibration curve. Figure 20 is a scatter plot comparing the calculated B / M ratio (IM9500 predicted value) based on the calibration curve with the measured value obtained by milling. 2 The result was 0.8269, and 2.04 in SEP.

[0058] The evaluation of semolina production rate will be described. In this example, the data obtained by milling using a Bühler test mill was used. First, the data for calibration curve creation was selected. In this example, although the analysis by spectrum acquisition and actual measurement was performed for 272 samples, the data of 204 samples was selected for calibration curve creation so that the semolina production rate values were widely distributed, and the data of the remaining 68 samples was used for calibration curve verification. Fig. 21 is a graph showing an example of the optical spectrum of a wheat grain sample, and shows all the optical spectra of the samples for calibration curve creation and the samples for calibration curve verification. Here, the optical spectra of the calibration curve creation set are shown in gray, and the optical spectra of the calibration curve verification set are shown in black. The near-infrared spectrum collected by Inframatic 9500 has values at 1061 wavelength points at 0.5 nm intervals in the wavelength range of 570 to 1100 nm.

[0059] The case of using all the data without dividing the calibration curve creation set and the calibration curve verification set will be described. Using the near-infrared spectrum of wheat collected by Inframatic 9500 and the value of the semolina production rate actually obtained by milling, a calibration curve for determining semolina production from the spectrum was created and evaluated by cross-validation. After performing the first derivative on the spectrum obtained by Inframatic 9�00, a calibration curve was created by PLS regression using the wavelength range of 850 nm to 1100 nm (Fig. 22).

[0060] Fig. 23 is a scatter diagram comparing the calculated value of the semolina production rate (IM9500 predicted value) based on this and the actually measured value obtained by milling, with PLS regression performed on all 272 samples without dividing them into the calibration curve creation set and the calibration curve verification set. The accuracy at the time of calibration curve creation was 0.8090 in terms of R 2 and 1.22 in terms of SEC. Also, Fig. 24 is a graph showing the calibration curve at this time. It shows the regression coefficient (vertical axis) for each wavelength (horizontal axis) in the obtained calibration curve.

[0061] <000036�>Figure 25 is a scatter plot comparing the calculated semolina production rate (IM9500 predicted value) after cross-validation with the measured value obtained from milling. The accuracy in cross-validation is R 2 The result was 0.7775, and 1.32 for SECV.

[0062] Next, we will explain the case where the samples are divided into two groups: a calibration curve creation set (204 points) and a calibration curve verification set (68 points). First, a calibration curve is created using the calibration curve creation set, and then the calibration curve verification set is evaluated. The method for dividing the sample set is as follows: the sample data is sorted according to the already obtained semolina production rate values, and then data is extracted every four samples to form a group of 68 samples for calibration curve verification. The remaining 204 samples are then used as the calibration curve creation group. Similar to the example shown in Figure 22, after performing the first derivative on the near-infrared spectrum, a calibration curve was created using PLS regression in the 850 nm to 1100 nm region.

[0063] Figure 26 is a scatter plot showing the correlation between IM9500 predicted values ​​(vertical axis) and measured values ​​(horizontal axis) when a calibration curve was created for 204 sample points used for calibration curve creation. The accuracy at this time is R 2 The result was 0.8306 at SEC and 1.15 at SEC. Figure 27 is a graph showing the calibration curve at that time.

[0064] Next, we will explain the results of evaluating the 68 points used for calibration curve verification using the obtained calibration curve. Figure 28 is a scatter plot comparing the calculated semolina production rate based on the calibration curve (IM9500 predicted value) with the measured value obtained by milling. 2 The result was 0.7306, and 1.48 in SEP.

[0065] An example of evaluating milling scores is described below. In this example, data obtained from milling using a Bühler test mill was used. First, data for creating a calibration curve is selected. In this example, spectrum acquisition and measurement analysis were performed on 270 samples. To ensure a wide distribution of milling score values, data from 203 samples were selected for calibration curve creation, and the remaining 67 samples were used for calibration curve verification. Figure 29 is a graph showing an example of the optical spectrum of wheat grain samples, displaying all optical spectra of the samples in the calibration curve creation set and the calibration curve verification set. Here, the optical spectrum of the calibration curve creation set is shown in gray, and the optical spectrum of the calibration curve verification set is shown in black. The near-infrared spectra collected by the Inframatic 9500 have 1061 wavelength points at 0.5 nm intervals in the wavelength range of 570 to 1100 nm.

[0066] This section describes the case where all data was used without splitting it into calibration curve creation and calibration curve validation sets. Using the near-infrared spectra of wheat collected with Inframatic 9500 and the milling scores obtained from actual milling, a calibration curve was created to determine semolina formation from the spectra, and evaluation was performed using cross-validation.

[0067] After performing SNV processing, detrend processing, and first derivative analysis on the spectra obtained with Inframatic 9500, a calibration curve was created using PLS regression in the wavelength range of 850 nm to 1100 nm (Figure 30).

[0068] Figure 31 is a scatter plot comparing the calculated milling score (IM9500 predicted value) based on a calibration curve determined by performing PLS regression on all 270 samples without separating the calibration curve creation and calibration curve validation, with the measured value obtained from milling. The accuracy during calibration curve creation is R 2 The values ​​obtained were 0.5282 for SEC and 2.08 for SEC. Figure 32 is a graph showing the calibration curve at this time. It shows the regression coefficients (vertical axis) for each wavelength (horizontal axis) in the obtained calibration curve.

[0069] Figure 33 is a scatter plot comparing the calculated milling score (IM9500 predicted value) after cross-validation with the measured value obtained from milling. The accuracy in cross-validation is R 2 The result was 0.4575, and 2.24 for SECV.

[0070] Next, we will explain the case where the samples are divided into two groups: a calibration curve creation set (203 points) and a calibration curve verification set (67 points). First, a calibration curve is created using the calibration curve creation set, and then the calibration curve verification set is used for evaluation. The sample set was divided as follows: the sample data was sorted according to the already obtained milling score values, and then data from every four samples was extracted to create a group of 67 samples for calibration curve verification. The remaining 203 samples were then used as the calibration curve creation group.

[0071] Similar to the example shown in Figure 30, after performing SNV processing, Detrend processing, and first derivative analysis on the near-infrared spectrum, a calibration curve was created using PLS regression in the 850 nm to 1100 nm region.

[0072] Figure 34 is a scatter plot showing the correlation between IM9500 predicted values ​​(vertical axis) and measured values ​​(horizontal axis) when a calibration curve was created for 203 sample points used for calibration curve creation. The accuracy in this case is R 2 The result was 0.5348 at SEC and 2.09 at SEC. Figure 35 is a graph showing the calibration curve at that time.

[0073] Next, we will explain the results of evaluating the 67 points used for calibration curve verification using the obtained calibration curve. Figure 36 is a scatter plot comparing the calculated milling score (IM9500 predicted value) based on the calibration curve with the measured value obtained by milling. 2 The result was 0.4669, and 2.16 in SEP.

[0074] This section describes an example of evaluating wheat flour lightness (L*). In this example, data obtained from milling using a Bühler test mill and a Brabender Jr. test mill were used. First, data for calibration curve creation was selected. In this example, spectrum acquisition and measurement analysis were performed on 749 samples. To ensure a wide distribution of wheat flour lightness (L*) values, data from 562 samples were selected for calibration curve creation, and the remaining 187 samples were used for calibration curve verification. Figure 37 is a graph showing an example of the optical spectrum of wheat grain samples, displaying all optical spectra of the samples used for calibration curve creation and calibration curve verification. Here, the optical spectrum of the calibration curve creation set is shown in gray, and the optical spectrum of the calibration curve verification set is shown in black. The near-infrared spectra collected by the Inframatic 9500 have 1061 wavelength points at 0.5 nm intervals in the 570-1100 nm wavelength range.

[0075] This section describes the case where all data is used without splitting it into calibration curve creation sets and calibration curve validation sets. Using the near-infrared spectra of wheat collected with Inframatic 9500 and the lightness (L*) values ​​of wheat flour that was actually milled and analyzed, a calibration curve was created to determine semolina formation from the spectra, and evaluation was performed by cross-validation.

[0076] After performing the second derivative on the spectrum obtained with Inframatic 9500, a calibration curve was created using PLS regression in the wavelength range of 820 nm to 1100 nm (Figure 38). Figure 39 is a scatter plot comparing the calculated values ​​of flour lightness (L*) (IM9500 predicted values) based on the calibration curve determined by performing PLS regression on all 749 samples without separating the calibration curve creation and calibration curve verification, with the measured values ​​obtained from milling and analysis. The accuracy during calibration curve creation is R 2 The values ​​obtained were 0.7855 for SEC and 0.30 for SEC. Figure 40 is a graph showing the calibration curve at this time. It shows the regression coefficients (vertical axis) for each wavelength (horizontal axis) in the obtained calibration curve.

[0077] Figure 41 is a scatter plot comparing the calculated values ​​(IM9500 predicted values) of flour lightness (L*) after cross-validation with the measured values ​​obtained after milling and analysis. The accuracy in cross-validation is R 2 The result was 0.6953 for [the first method] and 0.36 for [the second method].

[0078] Next, we will explain the case where the samples are divided into two groups: a calibration curve creation set (562 points) and a calibration curve verification set (187 points). First, a calibration curve is created using the calibration curve creation set, and then the calibration curve verification set is evaluated. The method for dividing the sample set is as follows: the sample data is sorted in the order of the already obtained flour lightness (L*) values, and then data is extracted every four samples to form the calibration curve verification group of 187 points. The remaining 562 points of the sample are then used as the calibration curve creation group.

[0079] Similar to the example shown in Figure 38, after performing a second derivative on the near-infrared spectrum, a calibration curve was created using PLS regression in the 820 nm to 1100 nm region. Figure 42 is a scatter plot showing the correlation between IM9500 predicted values ​​(vertical axis) and measured values ​​(horizontal axis) when a calibration curve was created for 562 samples used for calibration curve creation. The accuracy at this time was R 2 The result was 0.8096 at SEC and 0.28 at SEC. Figure 43 is a graph showing the calibration curve at that time. Next, we will explain the results of evaluating the 187-point calibration curve validation set using the obtained calibration curve. Figure 44 is a scatter plot comparing the calculated values ​​of flour lightness (L*) based on the calibration curve (IM9500 predicted values) with the measured values ​​obtained from milling and analysis. 2 The result was 0.6465, and 0.39 for SEP.

[0080] This section describes an example of evaluating the redness (a*) of wheat flour. In this example, data obtained from milling using a Bühler test mill and a Brabender Jr. test mill were used. First, data for calibration curve creation was selected. In this example, spectrum acquisition and measurement analysis were performed on 749 samples. To ensure a wide distribution of the redness (a*) values ​​of the wheat flour, data from 562 samples were selected for calibration curve creation, and the remaining 187 samples were used for calibration curve verification. Figure 45 is a graph showing an example of the optical spectrum of wheat grain samples, displaying all optical spectra of the samples used for calibration curve creation and calibration curve verification. Here, the optical spectrum of the calibration curve creation set is shown in gray, and the optical spectrum of the calibration curve verification set is shown in black. The near-infrared spectra collected by the Inframatic 9500 have 1061 wavelength points at 0.5 nm intervals in the wavelength range of 570 to 1100 nm.

[0081] This section describes the case where all data is used without splitting it into calibration curve creation sets and calibration curve validation sets. Using the near-infrared spectra of wheat collected with Inframatic 9500 and the redness (a*) values ​​of wheat flour that was actually milled and analyzed, a calibration curve was created to determine semolina formation from the spectra, and evaluation was performed by cross-validation.

[0082] After performing the second derivative on the spectrum obtained by Inframatic 9500, a calibration curve was created using PLS regression in the wavelength range of 825 nm to 1100 nm (Figure 46). Figure 47 is a scatter plot comparing the calculated value of the redness (a*) of wheat flour (IM9500 predicted value) based on the calibration curve, which was determined by performing PLS regression on all 749 samples without separating the calibration curve creation and calibration curve verification. The measured value obtained by milling and analyzing the flour is R 2 The values ​​obtained were 0.7670 for SEC and 0.18 for SEC. Figure 48 is a graph showing the calibration curve at this time. It shows the regression coefficients (vertical axis) for each wavelength (horizontal axis) in the obtained calibration curve.

[0083] Figure 49 is a scatter plot comparing the calculated values ​​(IM9500 predicted values) of the redness (a*) of wheat flour after cross-validation with the measured values ​​obtained after milling and analysis. The accuracy in cross-validation is R 2 The result was 0.6924 for [the first method] and 0.20 for [the second method].

[0084] Next, we will explain the case where the samples are divided into two groups: a calibration curve creation set (562 points) and a calibration curve verification set (187 points). First, a calibration curve is created using the calibration curve creation set, and then the calibration curve verification set is evaluated. The method for dividing the sample set is as follows: the sample data is sorted according to the already obtained redness (a*) values ​​of the wheat flour, and then data is extracted every four samples to create a group of 187 samples for calibration curve verification. The remaining 562 samples are then used as the calibration curve creation group.

[0085] Similar to the example shown in Figure 46, after performing a second derivative on the near-infrared spectrum, a calibration curve was created using PLS regression in the 825 nm to 1100 nm region. Figure 50 is a scatter plot showing the correlation between IM9500 predicted values ​​(vertical axis) and measured values ​​(horizontal axis) when calibration curves were created for 562 samples used for calibration curve creation. The accuracy at this time was R 2 The values ​​were 0.7845 and 0.17 at SEC. Figure 51 is a graph showing the calibration curve at that time.

[0086] Next, we will explain the results of evaluating the 187-point calibration curve validation set using the obtained calibration curve. Figure 52 is a scatter plot comparing the calculated value of the redness (a*) of wheat flour based on the calibration curve (IM9500 predicted value) with the measured value obtained by milling and analyzing the flour. 2 The result was 0.6959, and 0.20 for SEP.

[0087] As shown above, the calculated values ​​of wheat grain milling characteristics (milling yield, B / M ratio, semolina production rate, milling score) and flour characteristics (flour brightness (L*) and flour redness (a*)) according to the present invention show correlation with measured values. Using the quantitative method according to the present invention, four wheat grain milling characteristics (milling yield, B / M ratio, semolina production rate, milling score) or two flour characteristics (flour brightness (L*) and flour redness (a*)) can be easily and simultaneously quantified from wheat seeds. Furthermore, the method can be performed non-destructively, the operation is simple, and no special skills are required.

[0088] While the milling yield, B / M ratio, semolina production rate, milling score, flour brightness, and flour redness that can be quantified by the present invention are all important evaluation items, they do not represent all evaluation items related to wheat. Many other evaluation items were also carefully considered. However, for evaluation items such as flour yellowness (b*) and flour gelatinization characteristics (RVA maximum viscosity, RVA minimum viscosity, RVA final viscosity), even using the same method as the present invention, the calibration curve accuracy using all samples was not as high for flour yellowness (b*) as shown in the original invention. 2 =0.4385 and SEC=0.8852, with R at the highest viscosity of R 2 =0.4041, and SEC=31.0578, with R at the lowest viscosity of RVA. 2 = 0.2006 and SEC = 11.4631, R in the final viscosity of RVA 2 The values ​​were 0.4519 and SEC = 20.6041, and it was found that estimating these evaluation values ​​from the optical spectrum is difficult.

Claims

1. For several types of wheat grains used as calibration curves, the optical spectrum was obtained for each type using light with wavelengths including the near-infrared band via transmission. For each type of wheat grain in the sample used to create the calibration curve, one of the following will be quantified using a known quantitative method: milling yield, B / M ratio, semolina production rate, milling score, or the brightness or redness of the flour obtained from that wheat grain. The acquired optical spectrum information and the quantitative information obtained using known quantitative methods are analyzed using PLS regression analysis to set up a regression equation that shows the relationship between the optical spectrum information and the quantitative information obtained using known quantitative methods. A quantitative method for measuring wheat grains, which are the sample to be measured, by obtaining the optical spectrum using light with wavelengths including the near-infrared band via transmission, and calculating milling yield, B / M ratio, semolina production rate, milling score, and lightness or redness of flour based on a set regression equation.

2. For several types of wheat grains used as calibration curves, the optical spectrum was obtained for each type using light with wavelengths including the near-infrared band via transmission. For each type of wheat grain used in the calibration curve creation sample, the milling yield, B / M ratio, semolina production rate, milling score, and the brightness and redness of the flour obtained from those wheat grains were quantified using known quantitative methods. The acquired optical spectrum information and the quantitative information obtained using known quantitative methods are analyzed using PLS regression analysis to set up a regression equation that shows the relationship between the optical spectrum information and the quantitative information obtained using known quantitative methods. A quantitative method for measuring wheat grains, which are the sample to be measured, by acquiring the optical spectrum using light with wavelengths including the near-infrared band via transmission, and calculating milling yield, B / M ratio, semolina production rate, milling score, lightness of flour, and redness of flour based on a set regression equation.

3. The quantitative method according to claim 1 or claim 2, wherein PLS regression analysis is performed on wheat grains, which are samples for calibration curve creation, and wheat grains, which are samples to be measured, using optical spectrum information from light with a lower limit of 800 nm to 900 nm and an upper limit of 1050 nm to 1150 nm.

Images