Image recognition-based powder uniformity control method and application thereof in preparation of small yellow ginger whole powder

Abstract

The present application relates to the field of image recognition, and provides a powder uniformity control method based on image recognition and application thereof in preparation of small yellow ginger whole powder to solve the problem of poor product uniformity caused by powder agglomeration in the crushing process, which comprises the following steps: obtaining the relationship among the inner diameter of the tube, the particle size of the powder and the dispersion degree of the powder after obtaining the image of the powder in the tube, and obtaining the actual inner diameter of the tube; obtaining the relationship among the inlet pressure of the crushing process, the first feeding speed and the particle size distribution in the tube, and obtaining a first array set with a repeated crushing rate ≤ a first threshold value; obtaining the relationship among the second feeding speed, the repeated crushing rate and the over-crushing rate, and obtaining a second array set with an over-crushing rate ≤ a set value; and obtaining the actual process parameters according to the first array set and the second array set. After the process parameters obtained by the control method provided by the present application, not only the crushing efficiency of the whole process can be ensured, but also the over-crushing rate can be reduced, and thus the powder uniformity is improved.

Description

Technical Field

[0001] This invention relates to the field of image recognition, and more specifically, to a method for controlling powder uniformity based on image recognition and its application in the preparation of whole ginger powder. Background Technology

[0002] Low-temperature ultrafine airflow milling is an existing technology that utilizes supersonic airflow to impact, collide, and shear materials in a low-temperature environment to achieve ultrafine grinding. This method effectively avoids the denaturation of heat-sensitive components, such as gingerol and volatile oils, caused by frictional heat generation. It also leverages the increased brittleness of materials at low temperatures to improve grinding efficiency and powder quality. However, when materials are ground to the micron or nanometer scale, the specific surface area increases dramatically, resulting in extremely high surface energy. To reduce energy, particles spontaneously adsorb and aggregate through intermolecular forces, forming secondary or even tertiary aggregates. Furthermore, friction easily leads to the generation and accumulation of a large amount of static charge on the particle surface. These charges attract each other, causing particles to easily aggregate.

[0003] Low-temperature ultrafine airflow pulverization mainly includes three core steps: pre-cooling and embrittlement of the material, airflow acceleration and pulverization, and particle classification. The principle of particle classification is that the pulverized material enters the classification zone with the rising airflow. The classification zone is usually equipped with a high-speed rotating classification wheel, generating a strong centrifugal force field. Fine particles, subjected to airflow drag greater than centrifugal force, are drawn into the center of the classification wheel and collected by the airflow into subsequent cyclone separators and bag filters. Coarse particles that do not meet the particle size requirements are thrown against the perimeter of the classification zone due to the greater centrifugal force and fall back into the pulverization chamber under gravity, re-participating in the pulverization process, forming a closed-loop cycle until the particle size is met. As can be seen from the above, if particle agglomeration occurs, it may return to the pulverization chamber during the classification process, potentially leading to over-pulverization of particles on the surface of the agglomerates, resulting in a wider particle size distribution and poorer uniformity in the final product. Summary of the Invention

[0004] The purpose of this invention is to provide a powder uniformity control method based on image recognition and its application in the preparation of whole ginger powder, so as to solve the problem of poor product uniformity caused by powder agglomeration during the pulverization process in the prior art.

[0005] The embodiments of the present invention are achieved through the following technical solutions:

[0006] The present invention has at least the following beneficial effects:

[0007] A powder uniformity control method based on image recognition includes:

[0008] A pipe is installed between the crushing process and the grading process;

[0009] After obtaining the powder image inside the tube, the relationship between the inner diameter of the tube, the powder particle size, and the powder dispersion is obtained.

[0010] The actual inner diameter of the tube is obtained based on the preset powder particle size and dispersion threshold.

[0011] Based on the actual inner diameter of the tube, the relationship between the air inlet pressure, the first feed rate, and the particle size distribution inside the tube during the crushing process is obtained; the first feed rate is the feed rate of the material to be crushed.

[0012] Based on the set rotation speed of the classifying wheel and the particle size distribution inside the tube, a first set of arrays with repeatable crushing rate ≤ first threshold is obtained;

[0013] Based on the first set of data, the relationship between the second feed rate, the repeated crushing rate, and the over-crushing rate is obtained to obtain the second set of data where the over-crushing rate is ≤ a set value; the second feed rate is the feed rate of the surface modifier.

[0014] The actual process parameters are obtained based on the first set of arrays and the second set of arrays.

[0015] Preferably, the powder uniformity control method includes: using a first array as a reference, obtaining the relationship between the second feeding speed and the process defect degree to obtain the actual second feeding speed. Process defect degree = m1 * repeated crushing rate + m2 * over-crushing rate; m1 is a coefficient, 0 < m1 < 1, m2 is a coefficient, 0 < m2 < 1.

[0016] Preferably, the powder uniformity control method includes: after stopping the addition of the raw material to be crushed, obtaining the relationship between the rate of change of the second feeding speed and the over-crushing rate, and obtaining the actual rate of change.

[0017] Preferably, the powder uniformity control method includes: the raw material to be pulverized and the surface modifier have different colors;

[0018] After stopping the feeding of the raw material to be crushed, obtain the set of change rates of the second feed rate that meets the over-crushing rate requirement;

[0019] Using the rate of change set as a standard, real-time images of powder inside the tube are acquired, and the effective coating rate is obtained through the powder images; when the effective coating rate is less than the set threshold for the first time, the time corresponding to the previous image is acquired;

[0020] Using the time corresponding to the previous image as the secondary control node, the relationship between the second rate of change and the effective coverage rate is obtained, thus yielding the second rate of change.

[0021] Preferably, the effective coverage rate is the average coverage rate of all images within a set time range.

[0022] Preferably, the powder uniformity control method includes: using the rate of change set as a standard, acquiring powder images inside the tube in real time, and obtaining the relationship between effective coating rate and time;

[0023] After obtaining the secondary control node, the rate of change of the effective coverage rate is used to obtain the control period; the rate of change of the effective coverage rate within the control period is greater than the rate of change threshold.

[0024] The second rate of change is obtained based on the control period.

[0025] Preferably, the powder uniformity control method includes: obtaining the relationship between the droplet escape rate in the tube and the second feeding speed, obtaining the maximum second feeding speed, and the actual second feeding speed being ≤ the maximum second feeding speed.

[0026] The application of the powder uniformity control method in the preparation of whole ginger powder.

[0027] The present invention has at least the following beneficial effects:

[0028] This invention prioritizes obtaining a suitable inner diameter of the tube so that when using image recognition technology to obtain particle size distribution information within the tube, the image features can better display the particle size distribution. Then, using the particle size distribution information within the tube, the repeated crushing rate is obtained. The repeated crushing rate not only affects the overall crushing efficiency of the process but also affects the over-crushing rate to a certain extent. This invention first ensures the overall crushing efficiency through a first threshold. Finally, under the premise of ensuring crushing efficiency, the relationship between the second feed rate, the repeated crushing rate, and the over-crushing rate is obtained, thereby obtaining the optimal second feed rate. After using the process parameters obtained by the control method provided by this invention, not only can the overall crushing efficiency of the process be ensured, but the over-crushing rate can also be reduced, thereby improving the powder uniformity.

[0029] When the powder uniformity control method is applied to the preparation of whole ginger powder, the effective coating rate can be obtained simply by the color difference between the whole ginger powder and the surface modifier, thereby reducing the amount of surface modifier used and reducing production costs. Detailed Implementation

[0030] To make the objectives, methods, and advantages of the embodiments of the present invention clearer, the methods in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0031] Example 1: A powder uniformity control method based on image recognition, comprising:

[0032] A pipe is installed between the crushing process and the grading process;

[0033] After obtaining the powder image inside the tube, the relationship between the inner diameter of the tube, the powder particle size, and the powder dispersion is obtained.

[0034] The actual inner diameter of the tube is obtained based on the preset powder particle size and dispersion threshold.

[0035] Based on the actual inner diameter of the tube, the relationship between the air inlet pressure, the first feed rate, and the particle size distribution inside the tube during the crushing process is obtained; the first feed rate is the feed rate of the material to be crushed.

[0036] Based on the set rotation speed of the classifying wheel and the particle size distribution inside the tube, a first set of arrays with repeatable crushing rate ≤ first threshold is obtained;

[0037] Based on the first set of data, the relationship between the second feed rate, the repeated crushing rate, and the over-crushing rate is obtained to obtain the second set of data where the over-crushing rate is ≤ a set value; the second feed rate is the feed rate of the surface modifier.

[0038] The actual process parameters are obtained based on the first set of arrays and the second set of arrays.

[0039] In practice, low-temperature ultrafine airflow pulverization is an existing technology that utilizes supersonic airflow to impact, collide, and shear materials in a low-temperature environment to achieve ultrafine pulverization. This method effectively avoids the denaturation of heat-sensitive components, such as gingerol and volatile oils, caused by frictional heat generation. It also leverages the increased brittleness of materials at low temperatures to improve pulverization efficiency and powder quality. However, when materials are pulverized to the micron or nanometer scale, the specific surface area increases dramatically, resulting in extremely high surface energy. To reduce energy, particles spontaneously adsorb and aggregate through intermolecular forces, forming secondary or even tertiary aggregates. Furthermore, friction easily leads to the generation and accumulation of a large amount of static charge on the particle surface. These charges attract each other, causing particles to easily aggregate. High humidity in the air can also lead to particle aggregation. Dehumidified airflow can be used for pulverization.

[0040] Low-temperature ultrafine airflow pulverization mainly includes three core steps: pre-cooling and embrittlement of the material, airflow acceleration and pulverization, and particle classification. The principle of particle classification is that the pulverized material enters the classification zone with the rising airflow. The classification zone is usually equipped with a high-speed rotating classification wheel, generating a strong centrifugal force field. Fine particles, subjected to airflow drag greater than centrifugal force, are drawn into the center of the classification wheel and collected by the subsequent cyclone separator and bag filter. Coarse particles that do not meet the particle size requirements are thrown against the perimeter of the classification zone due to the greater centrifugal force and fall back into the pulverization chamber under gravity, re-participating in the pulverization process, forming a closed-loop cycle until the particle size is met. As can be seen from the above, if particle agglomeration occurs, it may return to the pulverization chamber during the classification process, potentially leading to over-pulverization of particles on the surface of the agglomerates, resulting in a wider particle size distribution in the final product. Furthermore, agglomeration can also cause blockage of the pipes.

[0041] To reduce particle agglomeration, liquid surface modifiers can be sprayed simultaneously inside the grinding chamber. The atomized surface modifiers, combined with the airflow, promote coating of the particle surfaces. Surface modifiers can be selected based on the type of material being ground; for example, cyclodextrin solutions, sucrose fatty acid esters, ethanol, and food-grade surfactants can be used. The aforementioned process of reducing particle agglomeration is a dynamic equilibrium process, namely, a balance between agglomeration and separation rates. When the difference between the separation tendency and the agglomeration tendency increases, the overall agglomeration phenomenon will weaken.

[0042] To address the above situation, the applicant aims to maximize the uniformity of the powder product, and thus conceived the technical solution provided in this embodiment. Exemplarily, the tube body is provided with a visible section, which can be made of a transparent material, such as glass or acrylic. A camera outside the tube body can acquire an image of the powder inside the tube, and the powder dispersion can then be obtained from this image. Exemplarily, the visible section is detachably connected, facilitating tube body cleaning and reducing or avoiding the influence of powder adhering to the inner wall of the tube on the powder dispersion value.

[0043] Image recognition and analysis are existing technologies, and powder dispersion can be directly obtained using these technologies. As an example, the distance from each particle to its nearest neighbor is obtained, and the average of these distances is calculated; this average value is used to characterize the powder dispersion. As an example, the image is divided into several grids, and the number of particles or the ratio of particle number to area within each grid cell is calculated. The smaller the variance of the number of particles or the corresponding ratio within each grid, the better the dispersion. Several images can be acquired, and the average image dispersion can be used as the final powder dispersion. As an example, existing watershed algorithms are used to segment overlapping or sticky particles, thus ensuring that the powder dispersion meets requirements even when there is little or no agglomeration.

[0044] Since this embodiment requires the use of image recognition technology to obtain particle size distribution information within the tube, a suitable inner diameter of the tube is prioritized in order to better display the particle size distribution in the image features. The reasons are at least as follows:

[0045] 1. When the amount of raw material is constant, if the inner diameter of the tube is too small, the distance between the powder particles will be small. In the acquired image, the distance between the powder particles will be small, which may lead to multiple powder particles being identified as a single agglomerated powder particle. This increases the difficulty of image recognition technology. Therefore, the inner diameter of the tube needs to have a certain length.

[0046] 2. The transport of powder between the crushing and grading processes depends on airflow, and the inner diameter of the tube will affect the airflow velocity or air pressure. Therefore, it is necessary to determine the inner diameter of the tube before adjusting the inlet pressure.

[0047] The preset powder particle size is the desired product particle size. The dispersibility threshold can be set according to actual conditions, such as the equipment used for specific heat image recognition. As an example, the dispersibility threshold is 3.5 times the preset powder particle size. The actual inner diameter of the tube can be selected as the minimum inner diameter that meets the dispersibility threshold, in order to reduce the pressure required for the airflow to lift the powder, thereby reducing energy consumption.

[0048] Existing image recognition technology is used to obtain the particle size distribution within the tube. The speed of the classifying wheel is directly related to the preset powder particle size. After obtaining the particle size distribution, the proportion of particles larger than the preset powder particle size to the total number of particles can be easily obtained, thus yielding the repeatability rate. As an example, particles are separated from the background using a grayscale threshold to generate a binary image. Edge detection is used to identify the contour of each particle, and then the equivalent diameter of the identified independent particles is obtained. After data aggregation, a particle size or volume distribution histogram can be easily obtained, thus yielding the repeatability rate. The equivalent diameter is the diameter of a circle with the same projected area as the particle. When obtaining the particle size distribution, since it is necessary to identify the particle size of individual particles or aggregates, the watershed algorithm is no longer used to segment overlapping or sticky particles. As an example, two-dimensional images of the particle dispersion system are captured from multiple angles, and then a three-dimensional image is reconstructed using computer algorithms. The MART algorithm can be used for the three-dimensional reconstruction.

[0049] The rate of repeated grinding not only affects the overall grinding efficiency of the process, but also influences the over-grinding rate to some extent. In this embodiment, a first threshold is used to ensure the overall grinding efficiency. The first threshold can be set according to the actual situation, for example, 15%. The first array set is a set of pairs, which consist of the air inlet pressure and the first feed rate. The rate of repeated grinding corresponding to the first array set is greater than 0.

[0050] Increased pulverization efficiency may lead to an increase in over-pulverization rate. Therefore, this embodiment, while ensuring pulverization efficiency, obtained the relationship between the second feed rate, the re-pulverization rate, and the over-pulverization rate. No surface modifier was added to the system when obtaining the first set of data. When obtaining the second set of data, due to the addition of the surface modifier, the agglomeration of powder particles was reduced under the same first feed rate and air pressure. Within a certain range of the second feed rate, the re-pulverization rate decreased with the increase of the second feed rate, and consequently, the over-pulverization rate decreased. However, when the second feed rate was too fast, the over-pulverization rate increased, usually accompanied by an increase in the re-pulverization rate. The applicant hypothesizes that the surface modifier is a liquid, and excessive dosage can easily lead to excessively high surface humidity of the powder, thereby increasing the possibility and stability of agglomeration.

[0051] As an example, the over-grinding rate can be obtained directly through product testing. For instance, if the expected product particle size range is a ± 5%a, the over-grinding rate can be obtained by measuring the percentage of powder particles with a particle size smaller than a ± 5%a to the total mass of the powder.

[0052] As an example, when the powder particle size is too fine, such as less than 30 μm, the particle size distribution of the product can be measured using a laser particle size analyzer or a sedimentation particle size analyzer. The over-pulverization rate can be easily obtained from the distribution data.

[0053] The over-grinding rate can be set as needed. As an example, the setting is 3%.

[0054] The actual process parameters include: inner diameter of the tube, inlet pressure, first feed rate and second feed rate.

[0055] As an example, the second feed rate corresponding to the lowest pulverization rate is selected to ensure optimal powder uniformity.

[0056] As an example, the second feed rate with the smallest value in the second array is selected as the actual parameter, thereby reducing the amount of surface modifier used and lowering costs.

[0057] Other parameters not included in the control range, such as temperature or feed particle size, can be maintained at fixed values ​​using existing technologies during the implementation of the control method. Temperature settings are typically directly related to the heat sensitivity of the raw material components and can therefore be set according to the type of raw material. Feed particle size can be set according to equipment requirements; for example, a feed particle size of 2 mm is used.

[0058] Example 2: To further balance efficiency and over-grinding rate, improvements were made based on Example 1. In this example, using the first array set as a benchmark, the relationship between the second feeding speed and the process defect rate was obtained to obtain the actual second feeding speed. Process defect rate = m1 * repeated grinding rate + m2 * over-grinding rate; m1 is a coefficient, 0 < m1 < 1, m2 is a coefficient, 0 < m2 < 1.

[0059] In the specific implementation process, m1 + m2 = 1. The repeated crushing rate and the over-crushing rate do not rise or fall completely synchronously, so the weight allocation can be made according to actual needs. This embodiment improves the process of obtaining the second data set in Embodiment 1 by using the first data set as a benchmark to obtain the relationship between the second feeding speed and the process defect degree, and then obtains the relationship between the second feeding speed and the process defect degree.

[0060] As an example, when ensuring overall process efficiency is required, m1 can be set to 0.7 and m2 to 0.3. As an example, when ensuring stable product particle size is required, m2 can be set to 0.7 and m1 to 0.7. The second feed rate corresponding to the lowest process defect rate should be selected as the actual process parameter.

[0061] Example 3: After stopping the addition of raw materials to be crushed, the relationship between the rate of change of the second feeding speed and the over-crushing rate was obtained to obtain the actual rate of change.

[0062] In the specific implementation process, during the feeding process, the amount of raw material and the particle size distribution in the crushing chamber are in a dynamic equilibrium process, that is, the area of ​​newly formed surface per unit time is relatively stable. At this time, a fixed second feeding speed can be selected. However, after the feeding is completed, no new raw material enters the crushing chamber. During the crushing process, the increase in the newly formed surface of existing raw material will gradually decrease. At this time, the amount of surface modifier required to cover the newly formed surface will also decrease. Therefore, this embodiment additionally obtains the rate of change of the second feeding speed when feeding stops, ensuring powder uniformity while reducing costs. The rate of change can be characterized by the decrease in the amount of surface modifier added per unit time.

[0063] As an example, the change in the second feed rate can be a continuous and uniform decrease, or it can be a segmented change. For example, a uniform feed rate V1 can be maintained in the first time period, a uniform feed rate V2 in the second time period, a uniform feed rate V3 in the third time period, and so on. V1 - V2 = V2 - V3. The length of each time period can be set according to the situation. For example, under normal circumstances, if the system needs to run for 5 minutes after the raw material to be crushed is fed, the length of each time period can be set to 1 minute.

[0064] As an example, there may be multiple rates of change that meet the over-grinding requirement. The highest rate of change can be selected as the actual rate of change to reduce the amount of surface modifier used.

[0065] Example 4: In order to further improve uniformity and reduce costs, an improvement was made based on Example 3. In this example, the raw material to be crushed and the surface modifier are different in color.

[0066] After stopping the feeding of the raw material to be crushed, obtain the set of change rates of the second feed rate that meets the over-crushing rate requirement;

[0067] Using the rate of change set as a standard, real-time images of powder inside the tube are acquired, and the effective coating rate is obtained through the powder images; when the effective coating rate is less than the set threshold for the first time, the time corresponding to the previous image is acquired;

[0068] Using the time corresponding to the previous image as the secondary control node, the relationship between the second rate of change and the effective coverage rate is obtained, thus yielding the second rate of change.

[0069] In specific implementation, when the raw material to be crushed and the surface modifier have different colors, the effective coating rate can be characterized by the color ratio in the image. During the feeding process, the second feeding speed remains constant, and the relationship between it and the over-crushing rate has already indirectly confirmed the coating effect to some extent. The purpose of this embodiment is to improve the second feeding speed after the addition of the raw material to be crushed is stopped. After the addition of the raw material to be crushed is stopped, the area of ​​the newly formed surface does not increase at a uniform rate; it usually increases faster in the initial stage. Using a uniformly reduced second feeding speed may still lead to an increase in the amount of surface modifier used, and even when the amount of surface modifier is significantly higher than the amount required for the newly formed surface, it may lead to an increase in agglomeration. Therefore, this embodiment utilizes the characteristic that the raw material to be crushed and the surface modifier have different colors. After the surfactant covers the raw material to be crushed, it completely or partially covers the color of the raw material to be crushed, resulting in a color change or color fading. The change in the covering effect can then be obtained through image recognition technology, thereby controlling the change process of the second feeding speed.

[0070] First, referring to the technical solution provided in Example 3, a set of change rates that meet the over-pulverization rate requirement is obtained, i.e., a change rate set. A certain change rate in the set is taken as the first change rate. Then, the powder image inside the tube is acquired in real time to obtain the effective coating rate. When the effective coating rate decreases below a set threshold, the time of the previous image is acquired, and then the second change rate is acquired to ensure effective powder coating. This process is repeated to continuously acquire the third, fourth, and fifth change rates, etc. Both the first and second change rates can be selected from the maximum values ​​in the set of change rates that meet the requirements. To simplify the process parameter control, only the first and second change rates can be acquired.

[0071] As an example, obtain the grayscale value A1 of the product color and the grayscale value A2 of the raw material, where A2 > A1. Set an effective grayscale value threshold based on these two values, for example, grayscale threshold = A1 + n% * (A2 - A1). The value of n can be set as needed, for example, n = 10. After the image is converted to grayscale, it is divided into several patches. Patches with grayscale values ​​greater than the threshold are designated as abnormal patches, and the rest are normal patches. The effective coverage rate is the percentage between the number of normal patches and the total number of patches. The threshold can be set as needed, for example, to 85%.

[0072] Example 5: In order to reduce the impact of system bias, an improvement was made based on Example 4. In this example, the effective coverage rate is the average coverage rate of all images within a set time range.

[0073] In practice, if the coverage rate of an image at a certain moment is used as the effective coverage rate, it is easily affected by system bias. Subsequent time periods may still meet the effective coverage requirement without adjusting the rate of change. Therefore, this embodiment sets a length range, such as 10 seconds, and continuously iterates the average value of the coverage rate of the images within the most recent 10 seconds. This average value is used as the effective coverage rate, which can reduce the impact of system bias.

[0074] As an example, when the effective coverage rate of the image between 49s and 59s is greater than or equal to a set threshold, and the effective coverage rate of the image between 50s and 60s is less than the set threshold, the relationship between the second rate of change and the effective coverage rate is obtained starting from 59s to obtain the maximum second rate of change that satisfies the condition that the effective coverage rate is greater than or equal to the set threshold.

[0075] Example 6: In order to simplify the process parameter control process, an improvement was made based on Example 4. In this example, the change rate set is used as the standard to obtain the powder image inside the tube in real time and obtain the relationship between the effective coating rate and time.

[0076] After obtaining the secondary control node, the rate of change of the effective coverage rate is used to obtain the control period; the rate of change of the effective coverage rate within the control period is greater than the rate of change threshold.

[0077] The second rate of change is obtained based on the control period.

[0078] In practice, using the first rate of change as a standard, the effective coating rate is obtained as a function of time, and the slope of the two-dimensional curve is used as the rate of change of the effective coating rate. If the rate of change is low, the value of rate of change control is low; therefore, a time period with a rate of change greater than the threshold can be selected as the control period. Based on the control period, a new rate of change is obtained to ensure that the second feed rate meets the requirements for effective coating. Typically, the effective coating rate decreases gradually. Near the end of the crushing process, the area of ​​the newly formed surface changes little, and the rate of change of the effective coating rate becomes extremely small, even achieving effective coating at the first rate of change. Therefore, the control period is usually only one segment. The new rate of change is usually relatively lower, meaning the rate of decrease in the second feed rate slows down. If effective coating is achieved within the control period, the coating effect in subsequent time periods will also improve to some extent.

[0079] Example 7: In order to improve the stability of the system, an improvement was made based on Examples 1-6. In this example, the relationship between the escape rate of droplets in the tube and the second feed rate was obtained to obtain the maximum second feed rate. The actual second feed rate is less than or equal to the maximum second feed rate.

[0080] In practice, since the surface modifier is mainly used to coat the particle surface, it has a good coating effect. However, if too much is added, the surface modifier may escape from the grinding chamber into subsequent areas such as the classifying wheel, affecting the stability of those areas. For example, if the surface modifier coats the surface of the classifying wheel, it will affect the classification accuracy of the classifying wheel.

[0081] As an example, the droplet escape rate can be obtained by detecting a humidity sensor. The humidity D1 of the airflow before entering the pulverizing chamber and the humidity D2 of the airflow inside the tube are measured. The droplet escape rate Y = 100% * (D2 - D1) / D2. The humidity can be the average of multiple measured values. Due to the influence of detection accuracy, if the calculated result of Y is negative, it is taken as 0.

[0082] As an example, the escape rate threshold is set to 0, and the second feed rate when the escape rate is 0 is obtained, which is then used as the maximum second feed rate.

[0083] As an example, an escape rate threshold of 5% is set to obtain the maximum second feed rate when the escape rate is ≤5%.

[0084] Example 8: This example provides an application of the powder uniformity control method described in Examples 1-7 in the preparation of whole ginger powder.

[0085] In practical implementation, for whole ginger powder to achieve good results, it typically needs to be pulverized to an extremely small particle size. Studies have shown that when the particle size of whole ginger powder is less than 15 μm, the effective substances can dissolve rapidly. The surface structure of whole ginger powder contains hydroxyl groups, making it more prone to agglomeration when pulverized to ultrafine particle size, significantly enhancing its surface hydrophilicity, adsorption, and agglomeration tendency. Using the process parameters obtained through the powder uniformity control method provided in this invention, the uniformity of the product particle size is improved, and production costs are effectively reduced. Whole ginger powder is yellow, while most surface modifiers are milky white, thus the effective coating rate can be easily obtained. When the surface modifier is colorless, changes in surface humidity also lead to color differences, which can also be used to determine the effective coating rate. White surface modifiers are preferred, and the aforementioned color difference can be increased by adding food coloring.

[0086] The above are merely preferred embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A powder uniformity control method based on image recognition, characterized in that, include: A pipe is installed between the crushing process and the grading process; After obtaining the powder image inside the tube, the relationship between the inner diameter of the tube, the powder particle size, and the powder dispersion is obtained. The actual inner diameter of the tube is obtained based on the preset powder particle size and dispersion threshold. Based on the actual inner diameter of the tube, the relationship between the air inlet pressure, the first feed rate, and the particle size distribution inside the tube during the crushing process is obtained; the first feed rate is the feed rate of the material to be crushed. Based on the set rotation speed of the classifying wheel and the particle size distribution inside the tube, a first set of arrays with repeatable crushing rate ≤ first threshold is obtained; Based on the first set of data, the relationship between the second feed rate, the repeated crushing rate, and the over-crushing rate is obtained to obtain the second set of data where the over-crushing rate is ≤ a set value; the second feed rate is the feed rate of the surface modifier. The actual process parameters are obtained based on the first set of arrays and the second set of arrays.

2. The powder uniformity control method according to claim 1, characterized in that, include: Using the first set of data as a reference, the relationship between the second feed rate and the process defect rate is obtained to get the actual second feed rate. Process defect rate = m1 * repeated crushing rate + m2 * over-crushing rate; m1 is a coefficient, 0 < m1 < 1, m2 is a coefficient, 0 < m2 < 1.

3. The powder uniformity control method according to claim 1, characterized in that, include: After stopping the feeding of the raw material to be crushed, the relationship between the rate of change of the second feeding speed and the over-crushing rate is obtained to obtain the actual rate of change.

4. The powder uniformity control method according to claim 3, characterized in that, include: The raw material to be crushed is different in color from the surface modifier; After stopping the feeding of the raw material to be crushed, obtain the set of change rates of the second feed rate that meets the over-crushing rate requirement; Using the rate of change set as a standard, real-time images of powder inside the tube are acquired, and the effective coating rate is obtained through the powder images; when the effective coating rate is less than the set threshold for the first time, the time corresponding to the previous image is acquired; Using the time corresponding to the previous image as the secondary control node, the relationship between the second rate of change and the effective coverage rate is obtained, thus yielding the second rate of change.

5. The powder uniformity control method according to claim 4, characterized in that, The effective coverage rate is the average coverage rate of all images within a set time range.

6. The powder uniformity control method according to claim 4, characterized in that, include: Using the rate of change set as a standard, real-time images of powder inside the tube are acquired to obtain the relationship between effective coating rate and time; After obtaining the secondary control node, the rate of change of the effective coverage rate is used to obtain the control period. The rate of change of the effective coverage rate during the control period is greater than the rate of change threshold. The second rate of change is obtained based on the control period.

7. The powder uniformity control method according to any one of claims 1-6, characterized in that, include: The relationship between the droplet escape rate inside the tube and the second feed rate is obtained to determine the maximum second feed rate. The actual second feed rate is less than or equal to the maximum second feed rate.

8. The application of the powder uniformity control method according to any one of claims 1-7 in the preparation of whole ginger powder.