CN-121979075-A - Uniformity online detection and control system and method

CN121979075ACN 121979075 ACN121979075 ACN 121979075ACN-121979075-A

Abstract

The invention discloses a uniformity on-line detecting and controlling system and method, which relates to uniformity detecting and controlling field, comprising a characteristic parameter analyzing module, a risk agglomeration judging module, an adhesion heat degree module and a controlling module, dividing micro-areas by acquiring synchronous multi-source data, respectively extracting data in each micro-area to calculate and obtain physical characteristic parameter sets thereof, constructing evolution hyperbolas based on the physical characteristic parameter sets of each micro-area, and the high-risk aggregate source is determined by searching a local large value of the dynamic trigger intensity in the high potential interval, the impact point is predicted based on the high-risk aggregate source to define a monitoring area, bimodal signals acquired in the area are analyzed, a node adhesion heat distribution map is generated by judging and accumulating the adhesion event intensity, and a response matrix control is executed according to the node adhesion heat distribution map to obtain a control strategy, so that the batch scrapping risk is avoided to the greatest extent.

Inventors

LIU GUANJUN
WANG SHUDONG
LIU YONGHAO

Assignees

烟台卫康动物保健品有限公司

Dates

Publication Date: 20260505
Application Date: 20260207

Claims (10)

1. The on-line uniformity detection and control method is characterized by comprising the following steps: Synchronous multi-source data are acquired, micro areas are divided through droplet density analysis, and data in each micro area are extracted respectively to calculate and obtain a physical characteristic parameter set of the micro areas; constructing an evolution hyperbola based on the physical characteristic parameter set of each micro-region, and determining a high-risk aggregate source by searching a local large value of dynamic trigger intensity in a high-potential interval; Predicting impact points based on high-risk aggregate sources to define a monitoring area, analyzing bimodal signals acquired in the monitoring area, and generating a node adhesion heat distribution map by judging and accumulating adhesion event intensities; And executing response matrix control according to the node adhesion heat distribution map to obtain a control strategy.
2. The on-line uniformity detecting and controlling method according to claim 1, wherein the physical characteristic parameter set analyzing step comprises the steps of: Carrying out texture structure feature and information entropy analysis based on the data subset of the three-dimensional space information to obtain liquefaction entropy of each micro-region; Performing liquid drop identification and binarization according to the data subset of the three-dimensional space information to obtain a binarized three-dimensional image; Performing momentum moment contribution tensor accumulation based on the three-dimensional velocity field data and the data subset of the three-dimensional space information to obtain total momentum moment of all liquid drops in each micro-area; Extracting time sequence data of internal charge peak values attenuated along with time according to a friction charge distribution image sequence and a data subset of three-dimensional space information, fitting the time sequence data with an exponential decay model through a nonlinear least square method, and solving a relaxation time constant of each micro-region; the liquefaction entropy, initial drop volume estimate, total moment of momentum of all drops, and relaxation time constant for each micro-region are collectively referred to as the physical feature parameter set for each micro-region.
3. The on-line uniformity detection and control method according to claim 2, wherein the data subset analysis step of the three-dimensional space information of each micro-area is as follows: Converting the continuous probability density field into a three-dimensional density curved surface, calculating a Riemann tensor of the three-dimensional density curved surface, and calculating to obtain the arc length square of differential displacement on the curved surface based on the Riemann tensor; Initializing a group of generating elements on the three-dimensional density curved surface based on the geodesic distance, and performing iterative optimization through a Laue algorithm to determine a final uniform irregular micro-region of the droplet density; And finally matching and extracting data and areas of the three-dimensional gray image data according to the uniform irregular micro-areas of the droplet density to obtain a data subset of the three-dimensional space information of each micro-area.
4. The on-line uniformity detection and control method according to claim 3, wherein said continuous probability density field analysis steps are as follows: Based on the three-dimensional gray image data obtained by laser-induced fluorescence scanning, obtaining initial droplet spreading three-dimensional point cloud data by adopting three-dimensional gray image data segmentation and centroid calculation, registering and synchronizing the initial droplet spreading three-dimensional point cloud data with three-dimensional speed field data obtained by a particle image velocimetry system and a friction charge distribution image sequence obtained by an imaging sensor array in space and time; projecting the initial liquid drop scattered three-dimensional point cloud data to a target plane to obtain a two-dimensional discrete point set; And converting the two-dimensional discrete point set into a continuous probability density field by using a nuclear density estimation method.
5. The on-line uniformity detection and control method according to claim 4, wherein said high risk agglomerate source analysis steps are as follows: calculating the structural inoculation potential value of each micro-region based on the liquefaction entropy of each micro-region and the initial liquid drop volume estimation value; Calculating a dynamic trigger intensity value of each micro-region based on the total momentum moment and the relaxation time constant of each micro-region; constructing an evolution hyperbola of each micro-region based on the structure inoculation potential value and the dynamic trigger intensity value of each micro-region, wherein the evolution hyperbola comprises a first structure inoculation potential curve and a second dynamic trigger intensity curve; Locking a high potential interval in the inoculation potential curve of the first structure, searching a local large value of the second dynamic trigger intensity curve based on the high potential interval to judge that resonance amplification effect exists, and acquiring a micro-area corresponding to the local large value to obtain a high-risk aggregate source.
6. The method for on-line uniformity detection and control according to claim 1, wherein said node adhesion heat distribution map analysis steps are as follows: detecting the transient temperature change signal by adopting a baseline difference and local minimum value detection method to obtain a negative pulse amplitude; the stress wave signal is subjected to high-pass filtering and combined with a signal energy integration method to obtain high-frequency energy burst strength; and for each monitoring node, when each monitoring node detects an effective adhesion event, the intensity of the effective adhesion event is accumulated to the existing accumulated intensity value of the corresponding node, and the adhesion heat level is divided to obtain a node adhesion heat distribution map.
7. The on-line uniformity detection and control method according to claim 6, wherein said transient temperature change signal and stress wave signal analyzing steps are as follows: calculating maximum deviation based on the average impact position centroid of each cluster to obtain the farthest distance of each cluster from the average impact position centroid, taking the farthest distance as a basic radius and additionally adding a safety coefficient to obtain an uncertainty radius Based on uncertainty radius Defining clusters Combining the circular areas serving as the radius circular areas to obtain an inner wall monitoring area; the method comprises the steps of carrying out discretization on a continuous monitoring surface by adopting a homogenization grid strategy based on an inner wall monitoring area to obtain a bimodal sensor arrangement set, and collecting instantaneous temperature change signals and stress wave signals based on the bimodal sensor arrangement set.
8. The on-line uniformity detection and control method according to claim 7, wherein the average impact position centroid analysis of each cluster comprises the steps of: Obtaining the geometric center of a high-risk aggregate source as an initial position and obtaining the total momentum moment of the high-risk aggregate source to construct a momentum tensor, decomposing the momentum tensor by using a characteristic value to determine a characteristic vector corresponding to a maximum characteristic value as a main motion direction, emitting rays from the initial position along the main motion direction, calculating the intersection point of the rays and the inner wall of the equipment to serve as a predicted inner wall impact point, and finally collecting the predicted inner wall impact points calculated by all the high-risk aggregate sources to form a predicted inner wall impact point set; And obtaining the average impact position centroid of each cluster through data clustering and centroid calculation based on the predicted inner wall impact point set.
9. The on-line uniformity detection and control method according to claim 1, wherein the control strategy analysis steps are as follows: Performing one-time targeted physical state instant verification based on the node adhesion heat distribution map to obtain an instant state vector; And performing two-dimensional response matrix control execution based on the instant state vector to obtain a control strategy.
10. An on-line uniformity detection and control system for implementing the on-line uniformity detection and control method according to any one of claims 1-9, characterized in that the system comprises: the characteristic parameter analysis module is used for acquiring synchronous multi-source data, dividing micro-areas through droplet density analysis, and respectively extracting data in each micro-area to calculate and obtain a physical characteristic parameter set of the micro-areas; the risk agglomeration judging module is used for constructing an evolution hyperbola based on the physical characteristic parameter set of each micro-region and determining a high risk agglomeration source by searching a local large value of dynamic trigger intensity in a high potential region; The adhesion heat degree module predicts impact points based on high-risk aggregate sources to define a monitoring area, analyzes bimodal signals acquired in the monitoring area, and generates a node adhesion heat degree distribution map by judging and accumulating adhesion event intensities; And the control module is used for executing response matrix control according to the node adhesion heat distribution diagram to obtain a control strategy.

Description

Uniformity online detection and control system and method Technical Field The invention relates to the field of uniformity detection and control, in particular to a uniformity online detection and control system and method. Background In powder mixing production in pharmaceutical, food and chemical industries, it is often necessary to uniformly disperse a very small amount of liquid phase active ingredient, which is costly or functionally critical, into a large amount of powder base material to ensure content uniformity of the final product. However, in the prior art, despite strictly following the set process parameters such as mixing time, rotating speed, etc., the problem of uneven distribution of active ingredients still frequently occurs in the final product, and hot spots with too high local concentration and cold spots with concentration far below the theoretical value are generated, resulting in rejection of the whole batch of products. The traditional quality control method, such as High Performance Liquid Chromatography (HPLC), usually performs offline sampling analysis on the finished product after mixing, and the "post-inspection" method cannot realize real-time monitoring and early warning in the process, and cannot trace the dynamic cause of the problem to perform process optimization. The root cause of such uniformity failure is not simply inadequate mixing, but rather a more complex failure mode of dynamic adhesion and cure. In the initial stage of mixing, the liquid phase additive forms a viscous tiny aggregate rich in active ingredients with part of powder particles. Inside the mixer, the intense friction and collision of the powder with the walls, paddles, etc. of the device can create a large amount of static electricity. These tiny agglomerates with high viscosity and unique charge characteristics are very easily electrostatically adsorbed and selectively adhere to "dead" areas such as the mixer inner wall, blade back, discharge port, etc. As mixing proceeds, these adherends gradually solidify and harden under the action of continuous shearing and air drying, eventually fall off due to strong impact in the later stage of mixing or in the discharging process, and randomly fall into the final product in the form of hard blocks, thereby deteriorating the uniformity of the whole batch. Therefore, there is a strong need in the art for a system and method for monitoring the whole process of adhesion-curing-detachment in real time on line, so as to provide early warning and intervention at the early stage of failure, but there is no technical means for effectively identifying such specific failure modes and performing closed-loop control. In order to solve the above-mentioned defect, a technical scheme is provided. Disclosure of Invention The present invention has been made in order to solve the technical problems set forth in the background art described above. The embodiment of the invention provides a uniformity on-line detection and control system and a uniformity on-line detection and control method. The invention aims at realizing the technical scheme that the uniformity on-line detection and control method comprises the following steps: Synchronous multi-source data are acquired, micro areas are divided through droplet density analysis, and data in each micro area are extracted respectively to calculate and obtain a physical characteristic parameter set of the micro areas; constructing an evolution hyperbola based on the physical characteristic parameter set of each micro-region, and determining a high-risk aggregate source by searching a local large value of dynamic trigger intensity in a high-potential interval; Predicting impact points based on high-risk aggregate sources to define a monitoring area, analyzing bimodal signals acquired in the monitoring area, and generating a node adhesion heat distribution map by judging and accumulating adhesion event intensities; And executing response matrix control according to the node adhesion heat distribution map to obtain a control strategy. Further, the physical characteristic parameter set analysis steps are as follows: Carrying out texture structure feature and information entropy analysis based on the data subset of the three-dimensional space information to obtain liquefaction entropy of each micro-region; Performing liquid drop identification and binarization according to the data subset of the three-dimensional space information to obtain a binarized three-dimensional image; Performing momentum moment contribution tensor accumulation based on the three-dimensional velocity field data and the data subset of the three-dimensional space information to obtain total momentum moment of all liquid drops in each micro-area; Extracting time sequence data of internal charge peak values attenuated along with time according to a friction charge distribution image sequence and a data subset of three-dimensional space information, fitting the time