Method for regulating and controlling technological parameters of redrying machine based on color Lab color difference value change of colors before and after tobacco leaf curing

Abstract

The invention belongs to the technical field of threshing and re-drying processes, and particularly relates to a method for regulating and controlling process parameters of a re-drying machine based on color Lab color difference value changes of colors before and after tobacco leaf drying, which is mainly used for determining the influence of re-drying strength on the color of tobacco lamina according to a clustering analysis and discriminant analysis method, finding out the relationship among the influence and the relationship, and classifying according to the color of the tobacco lamina. According to the influence of different redrying intensities on the color change of the tobacco lamina, the effective judgment on the redrying processing intensity according to the color change of the tobacco lamina is realized, so that the aroma keeping capability in the redrying process is improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of threshing and redrying process, and mainly determines the influence of redrying intensity on tobacco sheet color according to the method of cluster analysis and discriminant analysis, and finds out the relationship and classifies according to the tobacco sheet color. According to the influence of different redrying intensity on the color change of tobacco sheet, the effective judgment of redrying processing intensity is realized according to the color change of tobacco sheet, so as to improve the aroma retention ability in the redrying process. BACKGROUND

[0002] In recent years, with the implementation of the concept of "big process" in the tobacco industry, the trend of moving the primary processing to the threshing and redrying process is more and more obvious, and the tobacco enterprises have increasingly high requirements on the internal quality stability and purity of the tobacco sheet. At present, the internal quality of threshing and redrying is mainly realized according to the sensory evaluation of tobacco leaves, which has the phenomenon of feedback lag and cannot directly feedback to the operator. The color change of tobacco sheet can well make up for this deficiency, and the aroma retention control ability of the threshing and redrying process can be improved according to the color change of the tobacco sheet in the redrying process. SUMMARY

[0003] The purpose of the present application is to solve the above technical problems. On the basis of comparing the color difference of tobacco leaves and tobacco sheet, the color difference of tobacco leaves is classified and determined, and a method for improving the intelligence of the threshing and redrying machine is established according to the color difference of tobacco leaves before and after the redrying process.

[0004] The purpose of the present application can be realized by the following technical measures:

[0005] A method for adjusting the process parameters of a redrying machine based on the color Lab difference value change of tobacco leaves before and after curing, which detects the color of tobacco leaves after selecting leaves, and detects the color of tobacco sheet after redrying processing, and determines the color difference of tobacco leaves. The specific steps are as follows:

[0006] (1) The color Lab difference value of the raw tobacco leaves before redrying processing is measured by using a Lab color difference instrument. Randomly select not less than 30 tobacco leaves, and repeat the measurement of each tobacco leaf for 3 times. The average value of 3 times of measurement is recorded as the color Lab difference value (L*, a*, b*) of the batch of tobacco leaves;

[0007] (2) The Euclidean distance of the color Lab difference value of any two tobacco leaves measured in step (1) is calculated according to formula 1,

[0008]

[0009] Wherein: i represents the i tobacco leaf, j represents the j tobacco leaf, L represents the brightness of the tobacco leaf, a represents the red-green tone of the tobacco leaf, b represents the yellow-blue tone of the tobacco leaf, x iL , x ia , xib respectively represent the brightness, red-green hue and yellow-blue hue of the color of the i-th tobacco leaf; jL ja jb respectively represent the brightness, red-green hue and yellow-blue hue of the color of the j-th tobacco leaf;

[0010] (3) According to the Euclidean distance, the tobacco leaf samples with the smallest distance are sequentially merged to draw the distribution table and distribution curve of the classification number in Table 1. Through Table 1, the aggregation coefficient corresponding to different classification numbers is observed. When the classification number is less than 4, the aggregation coefficient changes greatly. When the classification number is greater than 4, the aggregation coefficient changes less,

[0011] Table 1 Classification number distribution table

[0012]

[0013] (4) According to the distribution curve, the elbow principle (also known as the elbow rule) is used to determine that the optimal classification number is 4 classes, which is the point at which the change of the aggregation coefficient of the adjacent two items begins to slow down. The cumulative proportion of tobacco leaves in each class is counted;

[0014] (5) According to formula 2, the distance between the tobacco leaf color Lab color difference value measured in step (1) and the origin (L* = 0, a* = 0, b* = 0) is calculated, and the tobacco leaf color Lab color difference value is sequentially sorted from small to large. Combine the class group number 4 obtained in step (4); divide the sorted results according to formula 3; the median in each division is recorded as the initial clustering center of the division, as shown in Table 2;

[0015]

[0016] wherein: x jL ja jb x j is the Lab color difference value of the tobacco leaf color (Lab color value is the total of three values, including L value, a value, and b value, i.e. brightness, red-green hue, and yellow-blue hue)

[0017]

[0018] wherein: N is the total number of tobacco leaf samples

[0019] Table 2 Initial clustering center

[0020] Initial cluster centers Leaf color lab color difference value ranking number L* a* b* Class 1 (1+n) / 2 X 1L ]]> X 1a ]]> X 1b ]]> Class 2 (n+1+2n) / 2 X 2L ]]> X 2a ]]> X 2b ]]> Class 3 (2n+1+3n) / 2 X 3L ]]> X 3a ]]> X 3b ]]> Class 4 (3n+1+4n) / 2 X 4L ]]> X 4a ]]> <![CDATA[X 4b ]]>

[0021] (6) For the tobacco leaf color Lab color difference value measured in step (1), the distance of each color Lab color difference value to each initial clustering center in Table 2 is calculated according to formula 1, and then it is assigned to the class corresponding to the nearest initial clustering center; ​​​​

[0022] (7) Recalculate the average of the points that have been assigned as the new cluster center according to formula 4;

[0023]

[0024] Wherein: n k is the number of samples contained in the kth class, k ranges from (1, 4); x lL , x la , x lb is x l Lab color difference value of tobacco leaf, x l is the sample contained in the kth class, l ranges from (1, n k );

[0025] (8) Repeat step (6) and step (7) until the cluster center no longer changes or changes very small, record the final cluster center, and perform statistics on the clustering results, and calculate the proportion of the number of tobacco leaves in each cluster to the total number of sampling tobacco leaves in step (1) according to formula 5;

[0026] P i = n i / N formula 5

[0027] Wherein: n is the number of tobacco leaves in the ith cluster, and N is the total number of sampling tobacco leaves in step (1);

[0028] (9) Calculate the center point u

[0029] of all tobacco leaf samples according to formula 4;

[0030]

[0031] Wherein: u is the center point of the whole sample point set; T in the formula represents the matrix transpose symbol,

[0032] (11) Calculate the intra-class scatter matrix according to formula 7 and formula 8;

[0033]

[0034] ∑j=∑x∈x i (x-u i )*(x-u i ) T Formula 8

[0035] Wherein: u i is the center point of each class point set;

[0036] (12) Calculate the inter-class scatter matrix according to formula 9;

[0037]

[0038] wherein: u i is the cluster center of each class, N represents the number of classes, m i is the number of samples of the i-th class;

[0039] (13) The optimal solution of the discriminant function is calculated according to formula 10:

[0040]

[0041] wherein: W is the characteristic vector to be solved, W T is the transposed vector of the characteristic vector (Explanation: find a projection direction (or a group of directions) W, so that when we project the data onto this direction, the center points of different categories are as far apart as possible (maximize the numerator), and the data points within each category are as close as possible (minimize the denominator), thereby maximizing their ratio (i.e. maximizing the ratio of inter-class difference to intra-class difference).

[0042] (14) Three groups of discriminant functions are calculated by solving, as shown below:

[0043] Function 1: Y1 = -21.762 + 0.840*L - 0.509*a - 0.521*b

[0044] Function 2: Y2 = -50.006 + 0.255*L + 0.804*a + 0.633*b

[0045] Function 3: Y3 = -16.114 + 0.534*L + 1.011*a - 0.797*b

[0046] (15) The color Lab color difference values of the re-dried tobacco sheets are determined;

[0047] (16) The tobacco leaf color Lab color difference values determined in step (15) are classified according to the model of step (14), and the proportion of each class of tobacco leaf is calculated according to formula 5;

[0048] (17) By comparing the proportion of each class of tobacco leaf in step (16) and step (4), the curing machine parameters are adjusted according to the degree of change in the results. The specific adjustment scheme is as follows: a, when the change in the proportion of tobacco leaf in the first class is greater than 10% and less than 20%, the curing machine setting temperature should be reduced by 2℃; b, when the change in the proportion of tobacco leaf in the first class is greater than 20%, the curing machine setting temperature should be reduced by 5℃.

[0049] The most prominent advantage of the present application is that the tobacco leaf color difference is classified and determined on the basis of the color contrast difference of the tobacco raw material and the cut tobacco, and a method for improving the intelligence of the threshing and redrying machine is established according to the color difference of the tobacco before and after the redrying process. The method mainly determines the influence of the redrying intensity on the color of the cut tobacco according to the clustering analysis and discriminant analysis method, and finds the relationship between them according to the color of the cut tobacco. According to the influence of different redrying intensities on the color change of the cut tobacco, the redrying processing intensity is effectively judged according to the color change of the cut tobacco, so as to improve the aroma retention capacity of the redrying process. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 The clustering classification curve in the method of the present application. DETAILED DESCRIPTION

[0051] The present application is further described by the following specific examples, but is not limited to the present application.

[0052] Example 1

[0053] In the present case, the flue-cured tobacco raw material is Fujian Yunyan 87 and Cui Bi No. 1, and the clustering analysis and discriminant analysis method is used to determine the color Lab color difference value of the tobacco before redrying and form a clustering model and a discriminant model. The color Lab color difference value of the tobacco after redrying is detected, and the clustering model and the discriminant model before redrying are used to determine whether the color Lab color difference value of the tobacco after redrying is suitable for the redrying model to determine whether the redrying machine parameters are suitable.

[0054] (1) The Lab color difference instrument is used to detect the tobacco raw material of Fujian Yunyan 87 and Cui Bi No. 1 (in order to ensure the universality and feasibility of the method, the raw material is selected as Fujian Yunyan 87 and Cui Bi No. 1, Yunyan 87 is a commonly planted flue-cured tobacco variety, and Cui Bi No. 1 is a new variety, the two raw materials are mixed, and the sample is randomly selected from the mixed tobacco to ensure the credibility of the test).

[0055] (2) The color Lab color difference value of the raw material is analyzed by system clustering method, and the best classification number is determined by the elbow principle as 4 classes. Therefore, the classification number of 4 is selected for clustering analysis.

[0056] (3) The K-Means clustering method with a classification number of 4 is used to cluster analyze the color Lab color difference value of the raw tobacco, and the final clustering center value is shown in Table 3.

[0057] Table 3 Final clustering center

[0058]

[0059] (4) The discriminant model is established according to formula 6, formula 7, formula 8, formula 9 and formula 10 for subsequent classification determination of the tobacco after redrying.

[0060] (5) Detecting the color Lab color difference value of the cured tobacco leaves.

[0061] (6) Using the model established in (4) to classify the cured tobacco leaves.

[0062] Comparing the classification results of the cured tobacco leaves and the pre-cured tobacco leaves, and adjusting the curing machine process parameters according to the comparison results. The specific adjustment scheme is as follows: a. When the change of the proportion of the first type of tobacco leaves is greater than 10% and less than 20%, the curing machine setting temperature should be reduced by 2℃; b. When the change of the proportion of the first type of tobacco leaves is greater than 20%, the curing machine setting temperature should be reduced by 5℃.

Claims

1. A method for regulating process parameters of a redrying machine based on the Lab color difference between tobacco leaves before and after curing, characterized by: After the tobacco leaves are sorted, the color of the tobacco leaves is tested. After the redrying process, the color of the tobacco leaves is tested to determine the color difference of the tobacco leaves. The specific steps are as follows: (1) Use a Lab colorimeter to measure the Lab color difference value of the raw tobacco leaves before redrying. Randomly select no less than 30 tobacco leaves, and repeat the measurement three times for each tobacco leaf. The average value of the three measurements is recorded as the Lab color difference value (L*, a*, b*) of the tobacco leaves in this batch. (2) Calculate the Euclidean distance of the Lab color difference values ​​of any two tobacco leaves measured in step (1) according to formula 1, Where: i represents the i-th tobacco leaf, j represents the j-th tobacco leaf, L represents the brightness of the tobacco leaf color, a represents the red-green hue of tobacco leaves, b represents the yellow-blue hue of tobacco leaves, and x iL 、x ia 、x ib They represent the brightness, red-green hue, and yellow-blue hue of the color of the i-th tobacco leaf, respectively. jL 、x ja 、x jb represent the brightness, red-green hue, and yellow-blue hue of the color of the jth tobacco leaf respectively; (3) Based on the Euclidean distance, the tobacco leaf samples with the smallest distance are merged in turn to draw the classification number distribution table and distribution curve in Table 1. The aggregation coefficient corresponding to different classification numbers is observed through Table 1. When the classification number is less than 4, the aggregation coefficient changes greatly. When the classification number is greater than 4, the aggregation coefficient changes less. Table 1 Classification number distribution table (4) According to the distribution curve, the optimal number of classifications is determined to be 4 by the "elbow principle", which is the point where the change of the aggregation coefficient of two adjacent items begins to slow down, and the cumulative proportion of tobacco leaves in each category is counted; (5) Calculate the distance between the Lab color difference value of the tobacco leaf color measured in step (1) and the origin (L*=0, a*=0, b*=0) according to formula 2, and sort the Lab color difference values ​​of the tobacco leaf color from small to large; combine the number of category groups 4 obtained in step (4); divide the sorting results into equal parts according to formula 3; record the median of each equal part as the initial cluster center of the equal part, as shown in Table 2: Where: x iL 、x ia 、x ib is the brightness, red-green hue, and yellow-blue hue of the i-th tobacco leaf Where: N is the total number of tobacco leaf samples Table 2 Initial cluster centers (6) For the Lab color difference values ​​of the tobacco leaves measured in step (1), the distance between each Lab color difference value and each initial cluster center in Table 2 is calculated according to Formula 1, and then assigned to the category corresponding to the nearest initial cluster center; (7) Recalculate the average value of the assigned points according to Formula 4 as the new cluster center; Where: n k is the number of samples contained in the kth class, and the value range of k is (1,4); x lL 、x la 、 x lb is x l Lab color difference value of tobacco leaf color, x l is the sample contained in the kth class, and the value range of l is (1, n k ); (8) Repeat steps (6) and (7) until the cluster center no longer changes or changes very little, record the final cluster center, and perform statistics on the clustering results. According to formula 5, the proportion of the number of tobacco leaves in each cluster to the total number of tobacco leaves sampled in step (1) is calculated; P i =n i / N Formula 5 Where: ni is the number of cigarette pieces in cluster i, N is the total number of cigarette pieces sampled in step (1); (9) Calculate the center point u of all tobacco leaf samples according to formula 4 (10) Calculate the global scatter matrix of tobacco leaf samples according to formula 6; Where: u is the center point of the entire sample point set; T represents the matrix transpose symbol, (11) Calculate the intra-class scatter matrix according to Formula 7 and Formula 8; Σj=Σ x ∈x i (-u i )*(xu i ) T Formula 8 Where: u i is the center point of each category point set; (12) Calculate the inter-class scatter matrix according to Formula 9; Where: u i is the cluster center of each class, N represents the number of classes, m i is the number of samples in the i-th category; (13) Calculate the optimal solution of the discriminant function according to formula 10; Where: W is the eigenvector to be solved, W T is the transposed vector of the eigenvector; (14) After solving and calculating, three sets of discriminant functions are obtained, as shown below: Function 1: Y1 = -21.762 + 0.840*L - 0.509*a - 0.521*b Function 2: Y2 = -50.006 + 0.255*L + 0.804*a + 0.633*b Function 3: Y3 = -16.114 + 0.534*L + 1.011*a - 0.797*b (15) Determine the Lab color difference value of the tobacco leaves after redrying; (16) classifying the Lab color difference values ​​of the tobacco leaves measured in step (15) according to the model of step (14), and calculating the percentage of each type of tobacco leaves according to formula 5; (17) By comparing the results of each category of tobacco leaf proportion in step (16) and step (4), the roasting machine parameters are adjusted according to the degree of change in the results. The specific adjustment scheme is as follows: a. When the change in the proportion of tobacco leaves in the first category is greater than 10% and less than 20%, the roasting machine setting temperature should be reduced by 2°C; b. When the change in the proportion of tobacco leaves in the first category is greater than 20%, the roasting machine setting temperature should be reduced by 5°C.

Images