CN118243654A

CN118243654A - Powder mixing uniformity detection method

Info

Publication number: CN118243654A
Application number: CN202410661785.0A
Authority: CN
Inventors: 程健博; 王帅; 王石; 朱业伟; 谢阳; 王征; 孙建国; 张博涵; 李凯; 冷友斌
Original assignee: Beijing Great Tech Technology Co ltd; Heilongjiang Feihe Dairy Co Ltd
Current assignee: Beijing Great Tech Technology Co ltd; Heilongjiang Feihe Dairy Co Ltd
Priority date: 2024-05-27
Filing date: 2024-05-27
Publication date: 2024-06-25
Anticipated expiration: 2044-05-27
Also published as: CN118243654B

Abstract

The application relates to the technical field of powder mixing uniformity detection, in particular to a powder mixing uniformity detection method, which comprises the following steps: placing the first powder and the second powder in a mixing device and activating the mixing device; collecting real-time spectrum data of the first powder and the second powder in the mixing device by adopting an infrared spectrometer; analyzing the real-time spectrum data by a first test method, and obtaining a first analysis result; analyzing the real-time spectrum data by a second test method, and obtaining a second analysis result; obtaining a uniformity result of powder mixing according to a first analysis result and a second analysis result, wherein the first test method is a moving window standard difference method, and the second test method is an F test method; or the first test method is a moving window standard deviation method, and the second test method is a principal component analysis method. The technical scheme of the application effectively solves the problem of poor accuracy of powder mixing uniformity detection in the related technology.

Description

Powder mixing uniformity detection method

Technical Field

The invention relates to the technical field of powder mixing uniformity detection, in particular to a powder mixing uniformity detection method.

Background

Powder mixing means that two or more different components are stirred and mixed in a mixing device, so that the different components are uniformly dispersed and mutually permeated, and various properties and functions of the mixed materials are improved.

Specifically, in the field of milk powder, formula milk powder is food produced by adding various nutritional and functional components such as vitamins, folic acid, lutein and the like on the basis of milk powder. The added nutrition and functional components enable the milk powder to provide more comprehensive nutrition and more perfect functions for consumers, thereby being beneficial to the diet health of consumers. In general, the quality of the compound nutritional functional components added into the milk powder is obviously smaller than that of the milk powder, so that the formula milk powder is very necessary to fully mix the added compound nutritional functional components in the production and processing process. The purpose of full mixing is to fully distribute a small amount of compound nutritional functional components in a large amount of milk powder, so that the quality stability of the milk powder product is ensured. However, in practical operation, the mixing process is often accompanied by a centrifugal phenomenon, that is, after the mixing operation is performed to a certain extent, the components to be dispersed are recombined with excessive progress of the mixing operation, resulting in non-rising uniformity. Therefore, it is important to detect the distribution uniformity of various nutritional functional components added in the milk powder on line, rapidly, timely, efficiently and accurately.

In the related art, the traditional method needs to control the time of mixing operation by experience in the process of mixing milk powder, adopts a shutdown sampling mode to carry out offline analysis on the compound nutritional functional components in the milk powder, calculates relative standard Deviation (RELATIVE STANDARD devition, RSD) on the content detection values of various nutritional functional components after sampling for a plurality of times and detecting the sampled products by adopting a national standard method, and evaluates the distribution uniformity of the functional nutritional components by the size of the RSD value so as to judge whether the materials are uniformly mixed. The traditional detection method needs multiple times of machine halt and multiple points of sampling. Therefore, the analysis method has long analysis time, high detection cost, low working efficiency and low detection accuracy, and is difficult to adapt to the requirements of modern high-efficiency and high-quality production.

Disclosure of Invention

The invention mainly aims to provide a powder mixing uniformity detection method for solving the problem of poor accuracy of powder mixing uniformity detection in the related art.

In order to achieve the above object, the present invention provides a powder mixing uniformity detecting method, including:

placing the first powder and the second powder in a mixing device and activating the mixing device;

Collecting real-time spectrum data of the first powder and the second powder in the mixing device by adopting an infrared spectrometer;

Analyzing the real-time spectrum data by a first test method, and obtaining a first analysis result;

analyzing the real-time spectrum data by a second test method, and obtaining a second analysis result;

obtaining a uniformity result of powder mixing according to the first analysis result and the second analysis result,

The first test method is a moving window standard deviation method, and the second test method is an F test method; or the first test method is a moving window standard deviation method, and the second test method is a principal component analysis method; or the first test method is an F test method, and the second test method is a principal component analysis method.

Further, the step of analyzing the real-time spectrum data by the second test method and obtaining the second analysis result further includes:

analyzing the real-time spectrum data by a third test method, and obtaining a third analysis result;

Obtaining a uniformity result of powder mixing according to the first analysis result, the second analysis result and the third analysis result,

The third test method is one of a moving window standard deviation method, an F test method and a principal component analysis method, and is different from the first test method and the second test method.

Further, the step of analyzing the real-time spectral data by the moving window standard deviation method includes:

calculating a plurality of groups of real-time spectrum data to obtain MBSD values of each group of real-time spectrum data;

And obtaining a moving window standard deviation analysis result according to the relation between MBSD values and MBSD threshold values.

Further, the step of calculating the plurality of sets of real-time spectral data to obtain MBSD values for each set of real-time spectral data includes:

by the formula The standard deviation of the spectral data for the individual wavelengths in each set of real-time spectral data is calculated,

Wherein,Standard deviation of spectral data of a j-th wavelength representing the i-th set of spectral data, i representing the i-th set of spectral data, j representing the j-th wavelength, k representing the k-th spectrum, N representing the number of spectral bars in each set of real-time spectral data,/>Absorbance value of j-th wavelength representing k-th spectrum,/>An average value of absorbance at a j-th wavelength of the N-th spectrum in the i-th set of spectrum data;

by the formula Calculating a moving window standard deviation value in each set of real-time spectrum data,

Wherein,Calculated value representing the moving window standard deviation of the ith set of spectral data,/>The standard deviation of the spectral data of the j-th wavelength of the i-th set of spectral data is represented, j represents the j-th wavelength, and P represents the number of wavelengths per spectrum.

Further, by the formulaThe step of calculating the standard deviation of the moving window in each set of real-time spectrum data comprises the following steps:

Calculate MBSD values for the first set of spectral data, noted as ；

Deleting a first one of the first set of spectral data, and aggregating one spectral data after the first set of spectral data and the first set of spectral data deleted the first set of spectral data to form a second set of spectral data;

calculate MBSD values for the second set of spectral data, noted as ；

Deleting the first spectral data in the ith-1 group of spectral data, and forming an ith group of spectrum by integrating one spectral data after the (i+N-1) th spectral data and the ith-1 group of spectral data deleted the first spectral data;

Calculate MBSD values for the ith set of spectral data, noted as 。

Further, MBSD values were calculated for the ith set of spectral data, noted asThe steps of (a) include:

setting MBSD threshold value as when N is between 18 and 30 ，/>The value is 9.0 multiplied by 10 ^-5 ~ 10.9×10^-5;

Respectively compare Value to/>Value and/>；

When (when)Value to/>The continuous 27 of the values satisfy less than/>When the first judgment condition is met;

setting MBSD threshold value as when N is between 10 and 17 ，/>The value is 11.0 multiplied by 10 ^-5 ~ 13.9×10^-5;

Respectively compare Value to/>Value and/>；

When (when)Value to/>The consecutive 21 of the values are less than/>When the second judgment condition is satisfied;

when the first judging condition and the second judging condition are both met, judging that the analysis result of the moving window standard deviation method meets a first preset condition.

Further, the step of analyzing the real-time spectral data by the F-test method includes:

calculating a plurality of groups of real-time spectrum data to obtain F test values of each group of real-time spectrum data;

and obtaining an analysis result of the F test method according to the relation between the F test value and the F test threshold value.

Further, the step of calculating the plurality of sets of real-time spectral data to obtain an F-test value for each set of real-time spectral data includes:

by the formula The variance of the spectral data for the individual wavelengths in each set of real-time spectral data is calculated,

Wherein,Represents the average variance of the ith set of spectral data, P represents the number of wavelengths per set of spectral data, j represents the jth wavelength, k represents the kth spectrum, N represents the number of spectral strips per set of spectral data,/>Absorbance value of j-th wavelength representing k-th spectrum,/>An average value of absorbance at a j-th wavelength of the N-th spectrum in the i-th set of spectrum data;

A plurality of variances is obtained.

Further, the step of obtaining the plurality of variances comprises:

Calculating a first set of spectral data A value, wherein the number of spectrum bars is N;

Collecting the (n+1) -th to (2N) -th spectral data into a second set of spectral data, and calculating the second set of spectral data ；

The (i+1) th spectral data is integrated into the (i+1) th spectral data to form the (i) th spectral data, and the (i) th spectral data is calculated。

Further, the step of obtaining the plurality of variances further comprises:

variance values of two adjacent groups of spectrum data are calculated according to the formula Performing a calculation, wherein/>Numerical value is greater thanA numerical value;

Obtaining a plurality of F test values;

comparing each F-test value with a F-test threshold value, wherein the F-test threshold value ；

When 21 consecutive F test values are smaller thanAnd when the analysis result of the F test method meets the second preset condition.

Further, the step of analyzing the real-time spectral data by principal component analysis includes:

Standard normal transformation and first-order derivation are carried out on a plurality of groups of real-time spectrum data;

obtaining a first principal component score, a second principal component score, and a third principal component score for each set of real-time spectral data;

The first principal component score is used as a horizontal axis value of a plane rectangular coordinate system, the second principal component score is used as a vertical axis value of the plane rectangular coordinate system, and a first scatter diagram of the first principal component score and the second principal component score of N pieces of spectrum data is drawn on the plane rectangular coordinate system;

the first principal component score is used as a horizontal axis value of a plane rectangular coordinate system, the third principal component score is used as a vertical axis value of the plane rectangular coordinate system, and a second scatter diagram of the first principal component score and the third principal component score of the N pieces of spectrum data is drawn on the plane rectangular coordinate system;

When the first scatter diagrams of the real-time spectrum data with the number not less than the preset group number are all located in the first ellipse, and when the second scatter diagrams of the real-time spectrum data with the number not less than the preset group number are all located in the second ellipse, judging that the analysis result of the principal component analysis method meets a third preset condition.

Further, the step of placing the first powder and the second powder in the mixing device and activating the mixing device comprises:

mixing a first weight of the first powder and a second weight of the second powder and forming a premix;

The mixing device mixes for a first preset period of time at a first speed, wherein the first speed is between 15% and 25% of the maximum rotational speed of the mixing device;

Increasing the mixing speed of the mixing device to a second speed within a second preset time period, and maintaining a third preset time period, wherein the second speed is between 55% and 65% of the highest rotating speed of the mixing device;

Mixing the premix and a third weight of the first powder to obtain a mixture;

the mixing device mixes the mixture at a third speed, and the mixture is mixed and kept for a fourth preset time period, wherein the third speed is between 25% and 35% of the highest rotating speed of the mixing device;

and in the fifth preset time period, the mixing speed of the mixing device is increased to a fourth speed, and the sixth preset time period is kept, wherein the fourth speed is the highest rotating speed of the mixing device.

Further, the first preset duration is between 10 seconds and 30 seconds, the second preset duration is between 1 minute and 3 minutes, the third preset duration is between 50 seconds and 70 seconds, the fourth preset duration is between 20 seconds and 40 seconds, the fifth preset duration is between 4 minutes and 6 minutes, and the sixth preset duration is between 120 seconds and 150 seconds.

Further, the density of the first powder is less than the density of the second powder.

Further, the particle size of the first powder is larger than the particle size of the second powder.

By applying the technical scheme of the application, the first powder and the second powder are firstly placed in the mixing device and the mixing device is started so as to mix the first powder and the second powder. The method comprises the steps of collecting real-time spectrum data of mixed powder of first powder and second powder in a mixing device by an infrared spectrometer, analyzing the real-time spectrum data by a first test method to obtain a first analysis result, and analyzing the real-time spectrum data by a second test method to obtain a second analysis result. And obtaining a uniformity result of powder mixing according to the first analysis result and the second analysis result. The first test method and the second test method are two of a moving window standard deviation method, an F test method and a principal component analysis method. Through the arrangement, the real-time detection is realized through a plurality of methods, and thus mutual verification is carried out through a plurality of methods, and the detection accuracy is better, so that the technical scheme of the application effectively solves the problem of poor accuracy of powder mixing uniformity detection in the related technology.

Drawings

The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the application. In the drawings:

fig. 1 shows a flow diagram of an embodiment of a powder mixing uniformity detection method according to the present invention;

FIG. 2 is a flow chart of a third test method of the powder mixing uniformity detection method of FIG. 1;

fig. 3 is a schematic diagram showing a specific flow of step S30 of the powder mixing uniformity detection method of fig. 1;

Fig. 4 shows a specific flow diagram of step S31 of fig. 3;

Fig. 5 is a schematic diagram showing a specific flow of step S40 of the powder mixing uniformity detection method of fig. 1;

fig. 6 is a schematic diagram showing a specific flow of step S60 of the powder mixing uniformity detection method of fig. 2;

fig. 7 is a schematic diagram showing a specific flow of step S10 of the powder mixing uniformity detection method of fig. 1;

FIG. 8 shows a near infrared spectrum of a first powder of the powder mixing uniformity detection method of FIG. 1;

FIG. 9 shows a near infrared spectrum of vitamin C of the powder mixing uniformity detection method of FIG. 1;

FIG. 10 shows a near infrared spectrum of vitamin D of the powder mixing uniformity detection method of FIG. 1;

FIG. 11 shows a near infrared spectrum of folic acid of the powder mixing uniformity detection method of FIG. 1;

FIG. 12 shows a near infrared spectrum of lutein of the powder mixing uniformity detection method of FIG. 1;

FIG. 13 is a graph showing a first set MBSD of calculated values of the moving window standard deviation method as a function of data set for the powder mixing uniformity detection method of FIG. 1;

FIG. 14 is a graph showing a second set MBSD of calculated values of the moving window standard deviation method versus data set for the powder mixing uniformity detection method of FIG. 1;

FIG. 15 is a graph showing the change of F calculated value with the number of F test method of the powder mixing uniformity detecting method of FIG. 1;

Fig. 16 shows a principal component analysis first and second principal component dispersion scattergram of the powder mixing uniformity detection method of fig. 1;

Fig. 17 shows a principal component analysis first and third principal component dispersion scattergram of the powder mixing uniformity detection method of fig. 1.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. The following description of at least one exemplary embodiment is merely exemplary in nature and is in no way intended to limit the invention, its application, or uses. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the present application. As used herein, the singular is also intended to include the plural unless the context clearly indicates otherwise, and furthermore, it is to be understood that the terms "comprises" and/or "comprising" when used in this specification are taken to specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof.

The relative arrangement of the components and steps, numerical expressions and numerical values set forth in these embodiments do not limit the scope of the present invention unless it is specifically stated otherwise. Meanwhile, it should be understood that the sizes of the respective parts shown in the drawings are not drawn in actual scale for convenience of description. Techniques, methods, and apparatus known to one of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, the techniques, methods, and apparatus should be considered part of the specification. In all examples shown and discussed herein, any specific values should be construed as merely illustrative, and not a limitation. Thus, other examples of the exemplary embodiments may have different values. It should be noted that: like reference numerals and letters denote like items in the following figures, and thus once an item is defined in one figure, no further discussion thereof is necessary in subsequent figures.

In recent years, with the development of chemometrics, the progress of computer technology and the rising of micro-electromechanical manufacturing technology, near infrared spectrum technology is widely applied in various fields of industrial and agricultural production due to its nondestructive, rapid, efficient and environment-friendly analysis characteristics. The near infrared spectrum technology can complete spectrum acquisition, data operation and result report in 1 second or even less, so that a technical solution can be provided for real-time online monitoring of materials and providing rapid, efficient, timely and accurate monitoring results. However, the existing near infrared spectrum technology needs to predict the content of the component to be analyzed based on the correction model so as to judge whether the mixing uniformity degree meets the production requirement, and the establishment of the correction model needs to put in much manpower, material resources and financial resources, and the model to be modeled needs to be frequently maintained so as to ensure the prediction accuracy of the model to be built.

In summary, for the mixing process of the formula milk powder, a milk powder mixing process and a mixing uniformity detection method which can adapt to the requirements of automatic continuous and efficient production, are based on spectral data analysis and do not need to establish a correction model are developed, so that the on-line, quick, timely, efficient and accurate judgment of the mixing uniformity of the compound nutritional functional components in the milk powder in the production process of the formula milk powder is realized, the production process is further optimized, the quality stability of the milk powder product is ensured, and the problem to be solved currently is solved.

As shown in fig. 1, in the present embodiment, the powder mixing uniformity detecting method includes:

Step S10: placing the first powder and the second powder in a mixing device and activating the mixing device;

step S20: collecting real-time spectrum data of the first powder and the second powder in the mixing device by adopting an infrared spectrometer;

step S30: analyzing the real-time spectrum data by a first test method, and obtaining a first analysis result;

step S40: analyzing the real-time spectrum data by a second test method, and obtaining a second analysis result;

step S50: obtaining a uniformity result of powder mixing according to the first analysis result and the second analysis result,

By applying the technical scheme of the embodiment, first, the first powder and the second powder are placed in a mixing device and the mixing device is started, so that the first powder and the second powder are mixed. The method comprises the steps of collecting real-time spectrum data of mixed powder of first powder and second powder in a mixing device by an infrared spectrometer, analyzing the real-time spectrum data by a first test method to obtain a first analysis result, and analyzing the real-time spectrum data by a second test method to obtain a second analysis result. And obtaining a uniformity result of powder mixing according to the first analysis result and the second analysis result. The first test method and the second test method are two of a moving window standard deviation method, an F test method and a principal component analysis method. Through the arrangement, the real-time detection is realized through a plurality of methods, and thus mutual verification is carried out through a plurality of methods, and the detection accuracy is better, so that the problem of poor accuracy of powder mixing uniformity detection in the related technology is effectively solved by the technical scheme of the embodiment.

The technical scheme of this embodiment can solve the following technical problems:

(1) And (5) stopping the machine for sampling. The traditional mixing link operation needs to be stopped for sampling, has low automation degree, thereby destroying the continuity of production, increasing the risk of pollution to materials and being difficult to adapt to the requirements of modern high-efficiency production.

(2) And (5) offline analysis. The traditional mixing link needs to perform off-line analysis of multiple quality indexes on the sampled products, and the adopted analysis method has long time, high cost, low efficiency, poor synchronism of analysis results and actual conditions, and is difficult to adapt to the requirements of modern high-efficiency and high-quality production.

(3) The model is difficult. The existing near infrared spectrum technology relies on a correction model to predict the content of each component to be analyzed so as to judge whether the mixing uniformity degree meets the production requirement, and the production cost of enterprises is seriously increased by establishing and maintaining the correction model.

The densities of the first powder and the second powder are different, and the particle diameters of the first powder and the second powder are different. Further, the density of the first powder is less than the density of the second powder, and the particle size of the first powder is greater than the particle size of the second powder.

Specifically, in this embodiment, the first powder is a milk powder-based powder, the second powder is a compound functional nutritional powder, the particle size of the milk powder-based powder is larger than the particle size of the compound functional nutritional powder, and the density of the milk powder-based powder is smaller than the density of the compound functional nutritional powder. Of course, in other embodiments, the first powder and the second powder may be powders in the pharmaceutical field, or powders in the food field, or powders in the battery field.

As shown in fig. 1 and 7, in the present embodiment, the compound functional nutritional powder contains at least one of vitamin C, vitamin D, folic acid, lutein, protein, fat, ash, niacin, vitamin B ₆, iron, zinc, calcium, and choline. As shown in fig. 2, in the present embodiment, step S40: the step of analyzing the real-time spectrum data by the second test method and obtaining a second analysis result further comprises the following steps:

step S60: analyzing the real-time spectrum data by a third test method, and obtaining a third analysis result;

step S70: and obtaining a uniformity result of powder mixing according to the first analysis result, the second analysis result and the third analysis result.

The first powder and the second powder are detected through the first test method, the second test method and the third test method respectively, and when the uniformity result of powder mixing obtained through the first test method and the uniformity result of powder mixing obtained through the second test method meet the requirements, the powder mixing qualification can be judged.

As shown in fig. 1, 3 and 13 and 14, in the present embodiment, step S30: the step of analyzing the real-time spectral data by the moving window standard deviation method comprises:

Step S31: calculating a plurality of groups of real-time spectrum data to obtain MBSD values of each group of real-time spectrum data;

step S32: and obtaining a moving window standard deviation analysis result according to the relation between MBSD values and MBSD threshold values.

By the method, the analysis result of the moving window standard deviation method can be obtained through the moving window standard deviation method, and further the results of the uniformity of the first powder and the second powder can be obtained.

Step S31: the step of computing a plurality of sets of real-time spectral data to obtain MBSD values for each set of real-time spectral data comprises:

As shown in fig. 1,3 and 4, in the present embodiment, step S311: by the formula The standard deviation of the spectral data for the individual wavelengths in each set of real-time spectral data is calculated,

step S312: by the formula Calculating a moving window standard deviation value in each set of real-time spectrum data,

Wherein,Calculated value representing the moving window standard deviation of the ith set of spectral data,/>The standard deviation of the spectral data of the j-th wavelength of the i-th set of spectral data is represented, j represents the j-th wavelength, and P represents the number of wavelengths per spectrum. Through the two formulas, the standard deviation of the spectrum data of each wavelength in each group of real-time spectrum data is calculated, and then the standard deviation is brought into the second formula, so that the standard deviation value of the moving window in each group of real-time spectrum data can be calculated.

As shown in fig. 1,3 and 4, in the present embodiment, step S312: by the formulaThe step of calculating the standard deviation of the moving window in each set of real-time spectrum data comprises the following steps:

Step S3121: calculate MBSD values for the first set of spectral data, noted as ；

Step S3122: deleting a first one of the first set of spectral data, and aggregating one spectral data after the first set of spectral data and the first set of spectral data deleted the first set of spectral data to form a second set of spectral data;

step S3123: calculate MBSD values for the second set of spectral data, noted as ；

Step S3124: deleting the first spectral data in the ith-1 group of spectral data, and forming an ith group of spectrum by integrating one spectral data after the (i+N-1) th spectral data and the ith-1 group of spectral data deleted the first spectral data;

Step S3125: calculate MBSD values for the ith set of spectral data, noted as 。

Specifically, the first set of spectral data includes spectrum 1, spectrum 2, …, spectrum 20, the second set of spectral data includes spectrum 2, spectrum 3, …, spectrum 21, … …, and the thirty-six sets of spectral data includes spectrum 36, spectrum 37, …, spectrum 55. This in turn calculates MBSD values for each group.

As shown in fig. 1, 3 and 4, in the present embodiment, step S3125: calculate MBSD values for the ith set of spectral data, noted asThe steps of (a) include:

Respectively comparing MBSD ₁ values to Value and/>；

When MBSD ₁ is up toThe continuous 27 of the values satisfy less than/>When the first judgment condition is met;

In particular, the method comprises the steps of, The value is 9.5 multiplied by 10 ^-5. Setting the number of spectral data of each group, n=20, and selecting a spectral wavelength range: 1316 nm-1651 nm, spectral data preprocessing method: standard normal variable transforms and first derivatives (second order polynomial & convolution window width set to 3).

Standard normal variable transformation:

Since the scanned material is solid powder, the same material may have spectrum differences due to different physical properties such as granularity, crystal form and the like, and the differences are caused by different optical paths and absorption degrees of scattered light entering the solid, which are called scattering effect. Therefore, standard normal transformation (SNV) is typically used to eliminate scattering.

In the middle ofIs the average value of absorbance of the ith spectrum, n is the number of variables of the content in the spectrum,/>Standard normal transformation of the jth wavelength representing the ith spectrum,/>Represents the j-th wavelength of the i-th spectrum.

First derivative:

Because the near infrared spectrum has weaker signal, the derivative is used for preprocessing the spectrum of the material, so that the resolution of low-resolution components in the complex spectrum can be increased, the characteristic information of the band is enhanced, and the drift irrelevant to the wavelength can be eliminated, namely, the drift index has large change. But the derivative process also amplifies the noise information. It is therefore often considered to choose the First Derivative (FD) for analysis of the spectral data.

And (3) spectrum differentiation calculation:

The pretreated spectra were averaged first, the spectra of one material were averaged to 1, and then the standard deviation of the spectra was calculated point by point. The standard deviation is large, namely the difference of materials in the band range is large, and the band can be used as a characteristic band for near infrared detection of powder mixing uniformity.

The standard deviation is calculated by the following formula:

Wherein: s is the standard deviation; n is the number of classes of samples; Is the average value of n samples at a single wavelength point,/> Indicating the ith spectrum.

Table 1 shows the data sets and their corresponding MBSD values calculated according to the moving window standard deviation method "first determination condition". As can be seen from Table 1, the MBSD values of the 32 consecutive calculations starting from the fifth set of spectral data are all less than。

TABLE 1 moving Window standard deviation data-first set

Setting MBSD threshold value as when N is between 10 and 17，/>The value is 11.0 multiplied by 10 ^-5~ 13.9×10^-5;

Respectively comparing MBSD ₁ values to Value and/>；

The moving window width, i.e. the number of spectral bars per group, is set to n=15, spectral wavelength range: 988 nm-1657 nm, spectral data preprocessing method: standard normal variable transform + detrending + second derivative (second order polynomial & convolution window width set to 3),The value is 11.5 multiplied by 10 ^-5.

Table 2 is a table of data sets and their corresponding MBSD values calculated according to the moving window standard deviation method "second determination condition". As can be seen from table 2, when the 5 th data set is calculated, MBSD calculated values start to be smaller than the threshold value 11.5×10 ^-5 of the "second determination condition", and 32 consecutive calculations MBSD data from the 5 th data set all are smaller than the threshold value 11.5×10 ^-5 of the "second determination condition".

The broken line in fig. 13 represents a threshold value of the first determination condition; the broken line in fig. 14 represents the threshold value of the second determination condition.

TABLE 2 moving Window standard deviation data-second group

By combining the above analysis with the data in tables 1 and 2, the above spectral data were satisfied with both the "first determination condition" and the "second determination condition", and the moving window standard deviation method was considered to pass.

As shown in fig. 1, 5 and 15, in the present embodiment, step S40: the step of analyzing the real-time spectral data by the F-test method comprises:

Step S41: calculating a plurality of groups of real-time spectrum data to obtain F test values of each group of real-time spectrum data;

step S42: and obtaining an analysis result of the F test method according to the relation between the F test value and the F test threshold value.

The dashed line in fig. 15 represents the F-test threshold of the F-test method.

As shown in fig. 1 and 5, in the present embodiment, step S41: the step of calculating the plurality of sets of real-time spectral data to obtain F-test values for each set of real-time spectral data includes:

Step S411: by the formula The variance of the spectral data for the individual wavelengths in each set of real-time spectral data is calculated,

Step S412: a plurality of variances is obtained.

The number of spectral bars of each set of spectral data was set to 16, and the variance of the spectral data of each wavelength was calculated for one set of spectral data according to the above formula.

As shown in fig. 1 and 5, in the present embodiment, step S412: the step of deriving the plurality of variances comprises:

The above-mentioned N is 16. At the beginning of the calculation, first, the 1 st set of spectral data (spectrum 1, spectrum 2, …, spectrum 16) is calculatedA value; calculation of/>, complete group 1 spectral dataAfter the values, the/>, was calculated for the 2 nd set of spectral data (spectrum 17, spectrum 18, …, spectrum 32); … … Calculate/>, for group 30 data (spectrum 481, spectrum 482, …, spectrum 496); Calculation/>, for group 31 data (spectra 497, 498, …, 512); From this calculation process a series/>Calculated values:。

Step S412: the step of deriving the plurality of variances further comprises:

variance values of two adjacent groups of spectrum data are calculated according to the formula

A calculation is performed, wherein,Numerical value is greater than/>A numerical value;

Obtaining a plurality of F test values;

the F value can be calculated by the above formula: 。

Each group of the spectrum numbers n=16, the degree of freedom of each group of the calculated data spectrum numbers is 15, and the spectrum wavelength selection range is as follows: 1124 nm-1651 nm, spectral data preprocessing method: multiple scatter correction+first derivative (second order polynomial & convolution window width set to 5), F check threshold(Significance level α=0.05); when the F test calculated value is continuously not less than 21 times to meet/>。

The 21 consecutive F test values are smaller thanThe judgment condition is F test.

Table 3 shows that according to the F test method "F test determination condition", each group of the spectral numbers n=16, the degree of freedom of the calculated data spectral numbers is 15, and the spectral wavelength selection range is: 1124 nm-1651 nm, spectral data preprocessing method: the multivariate scattering correction+first derivative (second order polynomial & convolution window width set to 5), each F value calculated. As can be seen from Table 3, when the 5 th data set is calculated, the F calculated value begins to be less than the threshold value(Significance level α=0.05) and the F values of 26 consecutive calculations starting from data set 5 are all less than the threshold/>(Significance level α=0.05).

Table 3F calculated values

The above spectral data are analyzed by combining the data in table 3, and the F test condition is satisfied for the above spectral data, and the F test is determined to pass.

As shown in fig. 1, 6 and 16 and 17, in the present embodiment, step S60: the step of analyzing the real-time spectral data by principal component analysis includes:

step S61: standard normal transformation and first-order derivation are carried out on a plurality of groups of real-time spectrum data;

Step S62: obtaining a first principal component score, a second principal component score, and a third principal component score for each set of real-time spectral data;

step S63: the first principal component score is used as a horizontal axis value of a plane rectangular coordinate system, the second principal component score is used as a vertical axis value of the plane rectangular coordinate system, and a first scatter diagram of the first principal component score and the second principal component score of N pieces of spectrum data is drawn on the plane rectangular coordinate system;

Step S64: the first principal component score is used as a horizontal axis value of a plane rectangular coordinate system, the third principal component score is used as a vertical axis value of the plane rectangular coordinate system, and a second scatter diagram of the first principal component score and the third principal component score of the N pieces of spectrum data is drawn on the plane rectangular coordinate system;

Step S65: when the first scatter diagrams of the real-time spectrum data with the number not less than the preset group number are all located in the first ellipse, and when the second scatter diagrams of the real-time spectrum data with the number not less than the preset group number are all located in the second ellipse, judging that the analysis result of the principal component analysis method meets a third preset condition.

The ellipses in fig. 16 represent the ellipses of the first and second principal component score distributions of the PCA method determination condition, and the ellipses in fig. 17 represent the ellipses of the first and third principal component score distributions of the PCA method determination condition.

The first principal component score and the second principal component score points of principal component analysis having consecutive not less than 15 sets of spectral data are all within the first elliptical range. The principal component analysis corresponding first principal component score and third principal component score points having consecutive no less than 15 sets of spectral data are all within the second elliptical perimeter.

The PCA method described above is a principal component analysis method.

As shown in fig. 1 and 6, in the present embodiment, the first ellipse uses the origin of the rectangular planar coordinate system as the ellipse center, the major axis of the ellipse is parallel to the transverse axis of the rectangular planar coordinate system, the semi-major axis is 0.0200, and the semi-minor axis is 0.0139; the second ellipse takes the origin of the rectangular plane coordinate system as the center of the ellipse, the major axis of the ellipse is parallel to the longitudinal axis of the rectangular plane coordinate system, the semi-major axis is 0.0150, and the semi-minor axis is 0.0107.

When the first scatter diagrams of the real-time spectrum data with the number not less than the preset group number are all located in the first ellipse, determining a condition A by a PCA method;

and when the second scatter diagrams of the real-time spectrum data with the number not less than the preset group number are all located in the second ellipse, judging the condition B by the PCA method.

In the mixing process, 10 spectra are collected as a group, 15 groups of spectrum data are continuously collected, and the spectrum wavelength is selected in the range: 920 nm-1670 nm, preprocessing the standard normal variable transformation+first derivative (the width of a second order polynomial & convolution window is set to be 3) of each group of spectrum data, performing principal component analysis, and extracting the scores of the first, second and third principal components of each spectrum. Dividing the first principal component score into a horizontal axis of a plane rectangular coordinate system, dividing the second principal component score into a vertical axis of the plane rectangular coordinate system, and drawing a first principal component score map and a second principal component score map; and drawing a first principal component dispersion point diagram and a third principal component dispersion point diagram by dividing the first principal component score into a horizontal axis of a plane rectangular coordinate system and dividing the third principal component score into a vertical axis of the plane rectangular coordinate system. First and second principal component dispersion maps of 15 sets of spectral data collected in succession are shown in fig. 16, and first and third principal component dispersion maps of 15 sets of spectral data collected in succession are shown in fig. 17.

As shown in fig. 16, the first and second principal components of 15 sets of spectrum data collected continuously obtain scattered points which are all distributed in an elliptical range with an origin (0, 0) of a rectangular planar coordinate system as an ellipse center, an elliptical long axis parallel to a horizontal axis of the coordinate system, a semi-long axis of 0.0200 and a semi-short axis of 0.0139, so as to satisfy the "PCA method judgment condition a"; meanwhile, as can be seen from fig. 17, the dispersing points of the first and third main components of the 15 sets of spectrum data collected continuously are all distributed in an elliptical range with the origin (0, 0) of the rectangular planar coordinate system as the ellipse center, the major axis of the ellipse being parallel to the longitudinal axis of the coordinate system, the semi-major axis being 0.0150 and the semi-minor axis being 0.0107, so as to satisfy the "PCA method judgment condition B".

As is clear from the above analysis, the above spectral data has satisfied both the "PCA judgment condition a" and the "PCA judgment condition B", and the principal component analysis method is considered to pass.

As shown in fig. 1 and 7, in the present embodiment, step S10: the step of placing the first powder and the second powder in the mixing device and activating the mixing device comprises:

step S11: mixing a first weight of the first powder and a second weight of the second powder and forming a premix;

Step S12: the mixing device mixes for a first preset period of time at a first speed, wherein the first speed is between 15% and 25% of the maximum rotational speed of the mixing device;

Step S13: increasing the mixing speed of the mixing device to a second speed within a second preset time period, and maintaining a third preset time period, wherein the second speed is between 55% and 65% of the highest rotating speed of the mixing device;

Step S14: mixing the premix and a third weight of the first powder to obtain a mixture;

Step S15: the mixing device mixes the mixture at a third speed, and the mixture is mixed and kept for a fourth preset time period, wherein the third speed is between 25% and 35% of the highest rotating speed of the mixing device;

step S16: and in the fifth preset time period, the mixing speed of the mixing device is increased to a fourth speed, and the sixth preset time period is kept, wherein the fourth speed is the highest rotating speed of the mixing device.

Specifically, in the present embodiment, the maximum rotation speed of the mixing device is 70rpm.

As shown in fig. 1 and 7, in the present embodiment, the first preset time period is between 10 seconds and 30 seconds, the second preset time period is between 1 minute and 3 minutes, the third preset time period is between 50 seconds and 70 seconds, the fourth preset time period is between 20 seconds and 40 seconds, the fifth preset time period is between 4 minutes and 6 minutes, and the sixth preset time period is between 120 seconds and 150 seconds.

Specifically, 10 kg of the first powder and 1 kg of the second powder are premixed to form 11 kg of premix, the mixture is mixed for 20 seconds (sec) at 20% of the highest rotation speed, and then 60 sec is mixed at a constant speed up to 60% of the highest rotation speed in 2 min; secondly, adding 289 kg first powder into the premix for mixing, wherein the total speed of the powder mixture is 300: 300 kg, mixing for 30 seconds at 30% of the maximum speed, accelerating to the maximum speed at a constant speed within 5 minutes, and mixing at the maximum speed of 138: 138 sec; and in the period of the highest rotating speed, a near infrared spectrometer is adopted to collect the near infrared spectrum of the powder mixture in the mixing device and record spectrum data.

The mixing device is a double-paddle mixing device.

The ambient temperature T of the mixing operation of the mixing device is between 22 ℃ and 33 ℃ and the relative humidity RH is between 20% -25%.

The near infrared spectrometer adopts a linear gradient spectroscopic near infrared spectrometer, the whole wavelength range of the linear gradient spectroscopic near infrared spectrometer is 908 nm-1676 nm, the spectrum center resolution of the linear gradient spectroscopic near infrared spectrometer is 10 nm, the single spectrum integration time of the linear gradient spectroscopic near infrared spectrometer is 5000 microseconds, the accumulation times of the linear gradient spectroscopic near infrared spectrometer are 20 times, and the linear gradient spectroscopic near infrared spectrometer acquires 10 pieces of spectrum data for powder mixture in the mixing device every 1 second.

Specifically, in the present embodiment, the second powder contains vitamin C, vitamin D, folic acid, and lutein.

The linear gradient spectroscopic near infrared spectra of the first powder, vitamin C, vitamin D, folic acid and lutein are shown in figure 8, figure 9, figure 10, figure 11 and figure 12 respectively.

To further confirm the above analysis conclusion, the mixture having reached the mixing end was sampled 24 times, and the contents of vitamin C, vitamin D, folic acid, and lutein in the 24 samples were detected, respectively, and the detection results are shown in table 4. The above-mentioned mixing end point means that the mixing meets the requirements.

TABLE 4 sampling test results

Sample volumes, minimum values, maximum values, average values, sample standard deviations, extremely poor values, and relative standard deviations (RELATIVE STANDARD devices, RSD) were calculated for the detection results of vitamin C, vitamin D, folic acid, and lutein, respectively, according to the data of table 4, as shown in table 5.

TABLE 5 sample detection results statistics

In table 5, RSD characterizes the degree of variability of the sample, i.e., characterizes the uniformity of dispersion of the target component in the sample. As can be seen from the data in Table 5, the process and method of the present embodiment are adopted to mix the first powder and the second powder, and only through spectral data analysis, when the mixing endpoint is reached, the RSD of the 24-time sampling analysis detection results of vitamin C, vitamin D, folic acid and lutein are all less than 10%, so that the requirements of industrial production on the uniformity of powder mixing can be satisfied, further the requirements of the above 4 target components on the uniformity of dispersion in the mixed materials are further illustrated, and the produced products have better quality stability.

Comparative example:

adopting a traditional method, namely combining a direct stirring method with shutdown sampling for 24 times; the contents of vitamin C, vitamin D, folic acid and lutein in 24 samples were measured respectively, and the measurement results are shown in Table 6.

Specifically, the steps of the conventional method described above are as follows: mixing the first powder and the second powder in a mixing device, starting the mixing device, stopping the mixing device after every preset time, and performing sampling detection. From the foregoing, it can be seen that the prior art is to stop for detection, which results in lower detection efficiency, and the sampling is performed at preset intervals, which easily results in excessive mixing.

TABLE 6 sample detection results using conventional methods

Sample volumes, minimum values, maximum values, average values, sample standard deviations, extremely poor values, and relative standard deviations (RELATIVE STANDARD devices, RSD) were calculated for the detection results of vitamin C, vitamin D, folic acid, and lutein, respectively, according to the data of table 6, as shown in table 7.

TABLE 7 statistics of sample detection results using conventional methods

In table 7, RSD characterizes the degree of variability of the sample, i.e., characterizes the uniformity of dispersion of the target component in the sample. As can be seen from the data in Table 7, although the conventional method is adopted for stopping sampling, the RSD of the 24 sampling analysis detection results of vitamin C, vitamin D, folic acid and lutein is not more than 10%, compared with the data in Table 5, the RSD values of the detection results of the conventional method are obviously larger than the RSD values of the sampling detection results based on the process and the method of the invention, so that the 4 target components are lower in dispersion uniformity in the formula and the quality stability of the produced formula is poorer.

The powder mixing uniformity detection method of the embodiment has the following beneficial effects:

(1) Continuous non-stop operation can be realized. By adopting the powder mixing uniformity detection method, continuous operation can be realized, shutdown sampling is not needed, the automation degree of the production process is high, the continuity of production can be ensured, and the risk of pollution to materials can be effectively reduced, so that the requirements of modern high-efficiency production can be met.

(2) The synchronous degree of the judging result of the mixing end point and the actual condition of the materials is high. By adopting the powder mixing uniformity detection method, the mixed materials can be monitored in real time on line, whether the mixing uniformity degree of the materials reaches the production requirement can be judged by only analyzing spectral data, off-line analysis is not needed, the detection cost is greatly reduced, and the real-time judgment result and the actual situation given by adopting the detection method of the embodiment have high synchronism.

(3) Complex modeling and model maintenance work can be avoided. By adopting the powder mixing uniformity detection method, whether the uniformity degree of material mixing reaches the production requirement can be judged by only analyzing the optical data, and a correction model is not required to be established and maintained, so that the production cost of enterprises can be greatly reduced.

In conclusion, by adopting the powder mixing uniformity detection method, the online, rapid, timely, efficient, accurate and automatic analysis of the powder mixing process can be realized, so that the problems of shutdown sampling, offline analysis, long detection time, high modeling and model maintenance cost of the mixed materials in the prior art are solved, the synchronization degree of the mixing end point judgment result and the actual condition of the materials is high, the production automation degree of the formula milk powder product can be improved, the continuity of the production process of the formula milk powder product can be enhanced, and the production efficiency of the formula milk powder can be greatly improved; provides a technical solution for guaranteeing the quality stability of the formula milk powder product.

In the description of the present invention, it should be understood that the azimuth or positional relationships indicated by the azimuth terms such as "front, rear, upper, lower, left, right", "lateral, vertical, horizontal", and "top, bottom", etc., are generally based on the azimuth or positional relationships shown in the drawings, merely to facilitate description of the present invention and simplify the description, and these azimuth terms do not indicate and imply that the apparatus or elements referred to must have a specific azimuth or be constructed and operated in a specific azimuth, and thus should not be construed as limiting the scope of protection of the present invention; the orientation word "inner and outer" refers to inner and outer relative to the contour of the respective component itself.

Spatially relative terms, such as "above … …," "above … …," "upper surface on … …," "above," and the like, may be used herein for ease of description to describe one device or feature's spatial location relative to another device or feature as illustrated in the figures. It will be understood that the spatially relative terms are intended to encompass different orientations in use or operation in addition to the orientation depicted in the figures. For example, if the device in the figures is turned over, elements described as "above" or "over" other devices or structures would then be oriented "below" or "beneath" the other devices or structures. Thus, the exemplary term "above … …" may include both orientations "above … …" and "below … …". The device may also be positioned in other different ways (rotated 90 degrees or at other orientations) and the spatially relative descriptors used herein interpreted accordingly.

In addition, the terms "first", "second", etc. are used to define the components, and are only for convenience of distinguishing the corresponding components, and the terms have no special meaning unless otherwise stated, and therefore should not be construed as limiting the scope of the present invention.

The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, but various modifications and variations can be made to the present invention by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for detecting uniformity of powder mixing, comprising:

obtaining a uniformity result of powder mixing according to the first analysis result and the second analysis result;

2. The method of claim 1, wherein the step of analyzing the real-time spectral data by the second test method and obtaining the second analysis result further comprises:

obtaining a uniformity result of powder mixing according to the first analysis result, the second analysis result and the third analysis result;

3. The method of claim 1, wherein the step of analyzing the real-time spectral data by a moving window standard deviation method comprises:

4. The method of claim 3, wherein the step of calculating a plurality of sets of real-time spectral data to obtain MBSD values for each set of real-time spectral data comprises:

by the formula Calculating standard deviation of spectrum data of each wavelength in each set of real-time spectrum data;

Wherein, Standard deviation of spectral data of a j-th wavelength representing the i-th set of spectral data, i representing the i-th set of spectral data, j representing the j-th wavelength, k representing the k-th spectrum, N representing the number of spectral bars in each set of real-time spectral data,/>Absorbance value of j-th wavelength representing k-th spectrum,/>An average value of absorbance at a j-th wavelength of the N-th spectrum in the i-th set of spectrum data;

by the formula Calculating a standard deviation value of a moving window in each group of real-time spectrum data;

Wherein, Calculated value representing the moving window standard deviation of the ith set of spectral data,/>The standard deviation of the spectral data of the j-th wavelength of the i-th set of spectral data is represented, j represents the j-th wavelength, and P represents the number of wavelengths per spectrum.

5. The method for detecting uniformity of powder mixture according to claim 4, wherein the formula isThe step of calculating the standard deviation of the moving window in each set of real-time spectrum data comprises the following steps:

Calculate MBSD values for the first set of spectral data, noted as ；

Deleting a first one of the first set of spectral data, forming a second set of spectral data from a set of spectral data subsequent to the first set of spectral data and the first set of spectral data deleted of the first set of spectral data;

calculate MBSD values for the second set of spectral data, noted as ；

Deleting the first spectral data in the i-1 th set of spectral data, and forming an i-th set of spectrum by integrating one spectral data after the i+N-1 th spectral data and the i-1 th set of spectral data deleted from the first spectral data;

The MBSD values were calculated for the i-th set of spectral data, noted MBSD _i.

6. The method of claim 5, wherein the calculation MBSD of the i-th set of spectral data is recorded asThe steps of (a) include:

Respectively compare Value to/>Value and/>；

7. The method of claim 1, wherein the step of analyzing the real-time spectral data by the F-test method comprises:

8. The method of claim 7, wherein the step of calculating the plurality of sets of real-time spectral data to obtain the F-test value for each set of real-time spectral data comprises:

by the formula Calculating variances of the spectrum data of the wavelengths in each set of real-time spectrum data;

Wherein, Represents the average variance of the ith set of spectral data, P represents the number of wavelengths per set of spectral data, j represents the jth wavelength, k represents the kth spectrum, N represents the number of spectral strips per set of spectral data,/>Absorbance value of j-th wavelength representing k-th spectrum,/>An average value of absorbance at a j-th wavelength of the N-th spectrum in the i-th set of spectrum data;

A plurality of variances is obtained.

9. The method of claim 8, wherein the step of obtaining a plurality of variances comprises:

10. The method of claim 9, wherein the step of obtaining a plurality of variances further comprises:

variance values of two adjacent groups of spectrum data are calculated according to the formula Performing a calculation, wherein/>Numerical value is greater than/>A numerical value;

Obtaining a plurality of F test values;

11. The powder mixing uniformity detection method according to claim 2, wherein the step of analyzing the real-time spectral data by a principal component analysis method comprises:

12. The method of claim 1, wherein the steps of placing the first powder and the second powder in the mixing device and activating the mixing device comprise:

Mixing the premix and a third weight of the first powder to obtain a mixture;

13. The method of claim 12, wherein the first preset time period is between 10 seconds and 30 seconds, the second preset time period is between 1 minute and 3 minutes, the third preset time period is between 50 seconds and 70 seconds, the fourth preset time period is between 20 seconds and 40 seconds, the fifth preset time period is between 4 minutes and 6 minutes, and the sixth preset time period is between 120 seconds and 150 seconds.

14. The method of claim 1, wherein the first powder has a density less than a density of the second powder.

15. The method of claim 1, wherein the first powder has a particle size greater than a particle size of the second powder.