RU2489705C1 - Method of determining degree of homogenisation of heterogeneous mixtures based on visual information on surface thereof - Google Patents

Abstract

FIELD: chemistry.

SUBSTANCE: method involves remote scanning of the surface of a mixture in a mixing apparatus accompanied by LED illumination of the surface of the mixture with white light with a stabilised power supply, processing the obtained image using computer RGB colour models with colour resolution of each point of the sample of the surface of the mixture into three components and calculating entropy of visual information, the value of which is used to determine the degree of homogeneity of the mixed heterogeneous mixture.

EFFECT: invention provides rapid inspection of the degree of homogeneity of the heterogeneous mixture, high accuracy of said information and simple inspection method.

9 dwg, 3 tbl

Description

The invention relates to the production technology of multicomponent heterogeneous mixtures and can be used in the paint and varnish, pharmaceutical industry in the analysis of the degree of homogeneity of both the finished multicomponent heterogeneous composition and its semi-finished products.

A known method for determining the quality of mixing bulk materials, including the introduction into the mixed mass of short-lived radioactive isotopes (AS USSR No. 347070, class B01F 3/18, G01N 23/00, publ. 16.03.71).

The disadvantage of this method is the difficulty of working with short-lived isotopes, the need for special equipment for measuring radioactivity in the volume of the mixture and the inevitable contamination of the mixture with extraneous impurities.

A known method for determining the uniformity of a granular mixture, including sampling operations, measuring light scattering using three photometric devices and converting the optical density into an electrical signal, the magnitude of which is used to judge the degree of homogeneity (see application No. 2002106464 of 03/12/02).

The disadvantage of this method is the complexity of the hardware design, insufficient sensitivity and accuracy of determining the degree of homogeneity of the mixture. This is especially true for mixtures in which one of the components is not a macro component, but an additive in an amount measured in units and tenths of a percent. In addition, photometry is difficult in gray mixtures, i.e. not having a pronounced color.

The closest method of the same purpose to the claimed invention according to a combination of features is a method for determining the time of mixing bulk materials. The method uses a RGB computer color model to process a scanned image of the surface of the mixture samples taken from the mixing device at certain intervals of mixing and molded in the form of a flat cylinder (tablets). The achievement of the highest possible degree of homogenization is judged by the minimum on the graph of the dependence of the brightness of the color of the image on the time of mixing materials (see RF patent No. 2267117, class G01N 21/85, 2004)

Signs of the prototype, which are common with the claimed invention, scanning the surface of the mixture; processing the obtained image using computer RGB color models with the decomposition of the color of each point of the image of the surface of the mixture into three components; determination of the degree of homogeneity of the mixture.

The disadvantages of the known method adopted for the prototype is the low accuracy of controlling the degree of homogeneity of a heterogeneous mixture due to the fact that it is impossible to obtain a numerical estimate of the degree of homogenization using the optoleptic information used in the known method. Another significant drawback is the need to form a sample in the form of a flat cylinder (tablet), which increases the mixing time of materials and excludes the possibility of operational control and determination of the degree of homogenization of a number of heterogeneous compositions, in particular, the phases "solid-liquid", "liquid-liquid" and "gaseous “liquid” and makes it impossible to automate the process of controlling the degree of homogenization of a mixed heterogeneous mixture. In addition, third-party software (for example, Adobe Photoshop) is used to process brightness information, which leads to an increase in the cost of the control system.

The objective of the invention is to provide operational control of the degree of homogeneity of a heterogeneous mixture, increasing its accuracy and simplifying the control method.

The problem is solved due to the fact that in the known method, including scanning the surface of the mixture, processing the received image using computer RGB color models with decomposing the color of each point of the image of the surface of the mixture into three components and determining the degree of uniformity of the mixture, scanning the surface of the mixture is carried out in the mixing apparatus remotely, while scanning is accompanied by LED illumination of the white mixture surface with a stabilized power source, after processing the resulting image calculate the entropy of optoleptic information, the value of which determines the degree of homogeneity of the mixed heterogeneous mixture.

The features of the proposed method, distinctive from the prototype - carry out remote scanning of the surface of the mixture in the mixing apparatus; scanning is accompanied by LED illumination of the white mixture surface with a stabilized power source; after processing the received image, the entropy of optoleptic information is calculated, the value of which determines the degree of homogeneity of the mixed heterogeneous mixture.

Distinctive features, in combination with the known ones, provide operational control of the degree of homogeneity of a heterogeneous mixture, increase its accuracy and simplify the control method.

Remote scanning of the surface of the mixture, accompanied by a white LED backlight with a stabilized power source, allows you to provide operational control of the homogenization process in the entire free space of the mixing apparatus.

The white LED backlight with a stabilized power source eliminates the need to pre-prepare the mixture sample, as it allows remote scanning of the mixture surface directly in the mixing apparatus, which simplifies the control method.

Determining the degree of homogeneity of the mixed heterogeneous mixture from the calculated value of the entropy of the optoleptic information provides an increase in the control accuracy.

The proposed method is illustrated by the drawings shown in figures 1-9.

Figure 1 presents a sample with a composition of low uniformity.

Figure 2 - the scatter of the values of the color components of the sample with low uniformity of the coating.

Figure 3 - sample with a composition of medium uniformity.

Figure 4 - the scatter of the values of the color components of the sample with an average uniformity of the coating.

Figure 5 - sample with a composition of high uniformity.

Figure 6 - scatter of the values of the color components of the sample with high uniformity of coating.

Figure 7 presents a histogram of the distribution of the levels of color components of the R-, G-, B-components within the same sample.

On Fig and 9 shows the experimental dependence of the entropy of information depending on the degree of homogenization, characterized by the number of revolutions of the mixer in absolute (bits) and reduced units.

A method for determining the degree of homogenization of heterogeneous mixtures from optoleptic information about their surface is as follows.

Using a light-scanning device, the image of the mixture surface in a raster-digital form is remotely captured. For this, the mixing apparatus is equipped with a white LED backlight with a stabilized power source. The received information is transmitted to the optoleptic information processing device.

According to the study, matrices O _R , O _G , O _{B of the} display of color components are formed. Empirical laws of the distribution of the intensity of color components P _R (ξ), P _G (ξ), P _B (ξ) are constructed, for which the number of points with the same state of intensity is calculated, where ξ is the state of the light intensity of a surface point. The entropy of opteleptic information for each color layer is determined, and then the total entropy, by the value of which the degree of homogeneity of the mixed heterogeneous mixture is determined.

The solution to this problem is achieved by using a remote photosensitive scanning mixture for sample surfaces, accompanied by white LED backlight with a stabilized power source in the entire volume of the device, using an additive RGB color model with 24 bits of color representation, i.e. obtaining optoleptic information about the state of the mixture.

The most common form of representing optoleptic information in digital form is a two-dimensional matrix, each element of which describes the brightness or color of the image element with the corresponding coordinates. The light reflected from the object is incident on a photosensitive matrix, each element of which produces an electrical signal proportional to the strength of the incident light. For each cell of the photosensitive matrix, three signals are generated that are proportional to different components of the radiation. The most commonly used decomposition is in three components: red (R), green (G) and blue (B). The surface of the sample can be represented as an image:

, l=1, 2, 3…, $$O_{i,j}^k\in \left[0l\right]$$

, l = 1, 2, 3 ...,

- отклик элемента светочувствительной матрицы для k-го цветового слоя;Where $$O_{i,j}^k$$

- response of the element of the photosensitive matrix for the k-th color layer;

k is the number of colors transmitted by the matrix;

i, j are the element numbers in the row and column of the matrix, respectively;

n, m is the number of elements of the photosensitive matrix in height and width, respectively;

l is the number of signal levels transmitted by each cell of the matrix.

The main characteristics of photosensitive conversion will be the resolution and color rendering depth. Resolution is defined as the number of image elements (dots) per unit length. The color rendering depth determines how many discrete levels each of the coordinates X _r , X _g and X _{b can have} .

A device for processing optoleptic information is a computational element of program-logical control, which can use any modern PC or program-logic controller.

Also, this computing element is simultaneously an information and software tool for controlling the mixing time and performs the following operations: obtaining information about the surface of the mixed mixture; calculation of the entropy of optoleptic information; the formation of a team to stop mixing.

The surface of the mixture is illuminated in the entire free space of the device using white LEDs, powered from a stabilized power source.

The entropy of optoleptic information about the surface of the mixture, which is automatically calculated, is used as a quantitative assessment of the homogeneity of the mixture, which makes it easier to control the degree of homogeneity of the heterogeneous composition and determine the optimal time for homogenization.

To stop the homogenization of the mixture, the control action generated by the threshold value of the entropy of the optoleptic information about its surface can be fed to the control device of the agitator drive, which can be an automatic key, a contact relay of an electronic key, or a discrete output of the USO of a program-logic controller.

The implementation of the proposed method is shown in a specific example.

For the full-scale experiment, the following parameters were selected for digitizing the image. Horizontal and vertical resolution 75 dpi (2.95 dpi / mm). The area of the coloring composition is selected so that 200 points fit vertically, 100 points horizontally. We can say that the surface image is presented in the form of a matrix

; i=1…100; y=1…200; k=1…3. $$O_{i,j}^k\in \left[0...255\right]$$

; i = 1 ... 100; y = 1 ... 200; k = 1 ... 3.

An assessment of the heterogeneity of the composition can be obtained by evaluating the degree of randomness of the information about the appearance of the mixture obtained using photosensitive scanning. Figures 1, 3, 5 show the results of photosensitive scanning of surface samples of the investigated mixed composition with paint of various degrees of homogenization, and Figures 2, 4, 6 show the coordinates of the points of the image of surfaces O in the spaces R, G and B-color components of the corresponding these surface samples. Samples obtained from the same composition taken from the mixer at different points in time. The scatter of points on the graphs characterizing the color components of the RGB model of the image of the surface of the samples, significant for samples with poor homogeneity, decreases for surfaces with greater uniformity, which makes it possible to use the numerical characteristic of the scatter to assess the uniformity of the mixture. The degree of scatter of points can be estimated using the entropy of optoleptic information.

The entropy of information is determined by the expression

H _nat = lnM,

where H _nat is the entropy of information of the system in natural units (nat);

M is the number of values that a random variable ξ can take - the state of one image point.

For practical presentation, it is more convenient to use the following expression for entropy:

H _nat = -lnP (ξ),

where P (ξ) is the probability of the event ξ.

The optoleptic information presented in digital form is discrete. The image of the mixture surface is characterized by some array of points on the O plane. Therefore, it is more expedient to express the entropy in bits as follows:

$${\text{H}}_{\text{nat}}=\text{-one}\text{44}\cdot {\displaystyle \sum _{i=\mathrm{one}}^k{p}_i}\cdot {\log }_2\cdot p_i,$$

where k is the number of system state levels;

P _i - probability of occurrence of the i-th state of the system, the estimate of which is calculated by the formula

, $$p_i\left(\xi \right)\cong N_i^{*}/N$$

,

- количество точек с одинаковым цветом;Where $$N_i^{*}$$

- the number of points with the same color;

ξ _i is the state of surface information;

N is the total number of points.

A histogram view of the distribution of the level of color components is presented in Fig.7.

Since the value of entropy for different batches of products will differ, in practice it is advisable to use the entropy of optoleptic information obtained when measuring during mixing with a certain time interval Δτ, reduced to the maximum value of entropy, which is the value of the entropy of optoleptic information at the initial time

, $$H_i^{*}=H_i/H_{\max }$$

,

where H _max is the entropy of the information of the sample at zero moment (for a non-homogenized mixture);

H _i - the entropy of optoleptic information at time τ _i .

For each component R, G and B, the entropy values H _R H _G and H _B are calculated, then their sum is calculated

H = H _R + H _B + H _G.

To evaluate the performance of the algorithm, a full-scale experiment was conducted with several different pigment compositions. The results of calculating the entropy of information on the state of the surface of the sample obtained using photosensitive scanning are shown in table 1.

The analysis of the dependence of the entropy of optoleptic information on the degree of mixing of the mixture. To determine the type of dependence of the estimation of the entropy of information on the surface of the mixture on the degree of mixing, expressed in the number of revolutions of the mixer of the mixer, the experimental values of the entropy of the optoleptic information were approximated by a function of the form

F (n) = a ₁ / exp (n · a ₂ ) + a ₃ .

To do this, the function was minimized

$$R\left(a_{\mathrm{one}},a_2,a_3\right)={\displaystyle \sum _{i=\mathrm{one}}^p{\displaystyle \sum _{j=\mathrm{one}}^m{\left(F\left(n_i,a_{\mathrm{one}},a_2,a_3\right)H_{ij}^{*}\right)}^2\to \underset{\overrightarrow{a}}{\min }}},$$

- значения энтропии i-му отбору пробы, i=1…p; р - количество точек отбора пробы, j - номер образца в каждом отборе, i=1…m, m - количество образцов в каждом отборе (принималось m=5).where a ₁ , a ₂ , and ₃ are the parameters of the function F (n); $$H_i^{*}$$

- entropy values for the i-th sampling, i = 1 ... p; p is the number of sampling points, j is the number of the sample in each sampling, i = 1 ... m, m is the number of samples in each sampling (taken m = 5).

The approximation results are shown in table 2.

table 2 Recipe Number Odds
ent one 2 3 four 5 a ₁ 9.005 4.842 10.369 12.336 4.285 a ₂ 0.075 0.114 0.074 0.108 0.116 a ₃ 12.039 8.748 10.325 9.182 4.519

On Fig and 9 shows the experimental dependence of the entropy of information depending on the degree of homogenization, characterized by the number of revolutions of the mixer in absolute (bits) and reduced units. The experimental values correspond to points: 2 - composition 1; 4 - composition 2; 6 - composition 3; 8 - composition 4; 10 - composition 5. Approximating dependences: 1 - for composition 1; 3 - for composition 2; 5 - for composition 3; 7 - for composition 4; 9 - for composition 5.

относительно альтернативной гипотезы $$H_1:{\sigma }_{общ}^2>{\sigma }^2$$

с помощью выборочной статистикиThe analysis of the regression correspondence with experimental data was carried out by testing the null hypothesis $$H_0:{\sigma }_{\mathrm{about}\mathrm{<figure-callout\; id="2"\; label="b\; u"\; filenames="00000026.png"\; state="\{\{state\}\}">b\; </figure-callout>}u}^2={\mathrm{<figure-callout\; id="2"\; label="\sigma "\; filenames="00000026.png"\; state="\{\{state\}\}">\sigma </figure-callout>}}^2$$

regarding an alternative hypothesis $$H_{\mathrm{one}}:{\sigma }_{\mathrm{about}\mathrm{<figure-callout\; id="2"\; label="b\; u"\; filenames="00000026.png"\; state="\{\{state\}\}">b\; </figure-callout>}u}^2>{\mathrm{<figure-callout\; id="2"\; label="\sigma "\; filenames="00000026.png"\; state="\{\{state\}\}">\sigma </figure-callout>}}^2$$

using sample statistics

, $$F=S_{\mathrm{about}bu}^2\left\{H\right\}/S_{\mathrm{about}\mathrm{from}\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="00000026.png"\; state="\{\{state\}\}">t</figure-callout>}}^2$$

,

- общая дисперсия, соответствующая экспериментальным значениям; $$S_{ост}^2$$

- остаточная дисперсия.Where $$S_{\mathrm{about}bu}^2\left\{H\right\}$$

- total variance corresponding to the experimental values; $$S_{\mathrm{about}\mathrm{from}t}^2$$

- residual dispersion.

для числа степеней свободы числителя v₁, знаменателя v₂ и уровня значимости q.The value of F statistics calculated by the formula (11) was compared with tabular critical values of the F - Fisher distribution $$F_{v_{\mathrm{one}}{v}_2}\left(\mathrm{one}-q\right)$$

for the number of degrees of freedom of the numerator v ₁ , the denominator v ₂ and the significance level q.

, то гипотеза Н₀ отбрасывается. Результаты оценки соответствия уравнения регрессии экспериментальным данным приведены в таблице 3.If $$F>F_{v_{\mathrm{one}}{v}_2}\left(\mathrm{one}-q\right)$$

, then the hypothesis H _{0 is} discarded. The results of assessing the compliance of the regression equation with experimental data are shown in table 3.

Table 3 Structure one 2 3 four 5 S _total 0.564 0.578 0.627 0.716 0.5 S _ost 0.183 0.131 0.215 0.151 0.245 F 3.087 4.415 2.914 4.74 2.041

The advantage of the proposed method is that it increases the accuracy of operational control of the degree of homogeneity of a heterogeneous mixture and simplifies the control method.

Claims (1)

A method for determining the degree of homogenization of heterogeneous mixtures from optoleptic information about their surface, including scanning the surface of the mixture, processing the received image using RGB computer color models with decomposing the color of each point of the image of the mixture surface into three components and determining the degree of homogeneity of the mixture, characterized in that the surface scan mixtures are carried out remotely in the mixing apparatus, while scanning is accompanied by LED illumination of the mixture surface. of the color with a stabilized power source, after processing the resulting image, the entropy of optoleptic information is calculated, the value of which determines the degree of homogeneity of the mixed heterogeneous mixture.

Images