Drug ultrafine grinding method and system based on airflow grinding principle

Abstract

The present invention discloses a method and system for ultrafine grinding of medicines based on the principle of airflow grinding, and relates to the technical field of medicine grinding. The method comprises: obtaining a grinding task instruction of an airflow grinding device; performing particle size detection according to the medicine to be ground to obtain real-time medicine particle size distribution; performing grinding task division to obtain D-level grinding task information; establishing a D-level grinding control trigger space according to multivariate grinding control factors; activating multi-order grinding optimization constraints and grinding loss sorting functions to generate D-level grinding control optimization results; performing cleaning verification on the airflow grinding device and the medicine to be ground to obtain cleaning verification results, and grinding the medicine to be ground. The present invention solves the technical problems of low medicine grinding precision and high medicine grinding loss in the prior art, and achieves the technical effect of significantly improving medicine grinding precision and effectively reducing medicine grinding loss.

Description

Technical Field

[0001] The invention relates to the technical field of drug pulverization, and in particular to a drug ultrafine pulverization method and system based on the airflow pulverization principle. Background Art

[0002] In the pharmaceutical field, the crushing of drugs is a key production link. Traditional drug crushing methods often make it difficult to accurately control the particle size distribution, resulting in unstable drug quality and affecting drug efficacy and safety. With the continuous development of pharmaceutical technology, the requirements for the accuracy and efficiency of drug crushing are getting higher and higher. In the past, when using airflow crushing equipment, it usually relied on the operator's experience to set the crushing parameters, lacking a systematic task division and precise control strategy. This can easily cause a large deviation between the particle size distribution of the crushed drug and the expectation, which not only affects the performance of the drug, but may also increase production costs and waste resources. In addition, in the drug production process, cleaning validation is an important link to ensure the quality and safety of drugs. If the cleaning of equipment and drugs is not up to standard, cross contamination may occur, seriously affecting the purity of the drug.

[0003] The existing technology has the technical problems of low drug crushing accuracy and high drug crushing loss. Summary of the invention

[0004] The present application provides a method and system for ultrafine grinding of medicines based on the principle of airflow grinding, which is used to solve the technical problems of low medicine grinding accuracy and high medicine grinding loss in the prior art.

[0005] In view of the above problems, the present application provides a method and system for ultrafine grinding of medicines based on the principle of air flow grinding.

[0006] In a first aspect of the present application, a method for ultrafine grinding of medicines based on the principle of airflow grinding is provided, the method comprising:

[0007] Obtain a pulverizing task instruction of an airflow pulverizing device, wherein the pulverizing task instruction includes basic drug information and expected drug particle size distribution of the drug to be pulverized; perform particle size detection according to the drug to be pulverized to obtain real-time drug particle size distribution; perform pulverizing task division according to the basic drug information, the real-time drug particle size distribution and the expected drug particle size distribution to obtain D-level pulverizing task information, wherein D is a positive integer greater than 1; perform pulverizing control trigger feature mining on the D-level pulverizing task information according to the multivariate pulverizing control factor of the airflow pulverizing device, and establish a D-level pulverizing control trigger space; activate multi-order pulverizing optimization constraints and pulverizing loss sorting functions, perform optimization analysis on the D-level pulverizing control trigger space in combination with the D-level pulverizing task information, and generate a D-level pulverizing control optimization result; perform cleaning verification on the airflow pulverizing device and the drug to be pulverized respectively, and obtain a device drug cleaning verification result. When the device drug cleaning verification result is qualified, pulverize the drug to be pulverized according to the D-level pulverizing control optimization result and the airflow pulverizing device.

[0008] The second aspect of the present application provides a drug ultrafine grinding system based on the airflow grinding principle, the system comprising:

[0009] A pulverizing task instruction acquisition module, the pulverizing task instruction acquisition module is used to obtain the pulverizing task instruction of the airflow pulverizing device, wherein the pulverizing task instruction includes the basic information of the medicine to be pulverized and the expected medicine particle size distribution; a medicine particle size distribution acquisition module, the medicine particle size distribution acquisition module is used to perform particle size detection according to the medicine to be pulverized, and obtain the real-time medicine particle size distribution; a D-level pulverizing task information acquisition module, the D-level pulverizing task information acquisition module is used to divide the pulverizing task according to the basic information of the medicine, the real-time medicine particle size distribution and the expected medicine particle size distribution, and obtain the D-level pulverizing task information, wherein D is a positive integer greater than 1; a pulverizing control trigger space establishment module, the pulverizing control trigger space establishment module is used to determine the pulverizing task information according to the airflow pulverizing device. A multivariate crushing control factor is prepared, and crushing control trigger feature mining is performed on the D-level crushing task information to establish a D-level crushing control trigger space; a crushing control optimization result generation module is used to activate multi-order crushing optimization constraints and crushing loss sorting functions, and optimize the D-level crushing control trigger space in combination with the D-level crushing task information to generate a D-level crushing control optimization result; a cleaning verification result acquisition module is used to perform cleaning verification on the airflow crushing equipment and the to-be-crushed drug respectively, and obtain the equipment and drug cleaning verification result. When the equipment and drug cleaning verification result is qualified, the to-be-crushed drug is crushed according to the D-level crushing control optimization result and the airflow crushing equipment.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0011] Obtain the crushing task instruction of the airflow crushing equipment; perform particle size detection according to the medicine to be crushed to obtain the real-time medicine particle size distribution; perform crushing task division to obtain D-level crushing task information; perform crushing control trigger feature mining on the D-level crushing task information according to the multivariate crushing control factor of the airflow crushing equipment, and establish the D-level crushing control trigger space; activate the multi-order crushing optimization constraint and crushing loss sorting function, and perform optimization analysis on the D-level crushing control trigger space in combination with the D-level crushing task information to generate the D-level crushing control optimization result; perform cleaning verification on the airflow crushing equipment and the medicine to be crushed respectively, and obtain the equipment medicine cleaning verification result. When the equipment medicine cleaning verification result is qualified, crush the medicine to be crushed according to the D-level crushing control optimization result and the airflow crushing equipment. The technical effect of significantly improving the crushing accuracy of medicines and effectively reducing the crushing loss of medicines is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0013] Figure 1 A schematic diagram of a process flow of a drug ultrafine grinding method based on the airflow grinding principle provided in an embodiment of the present application;

[0014] Figure 2 A schematic diagram of the structure of a pharmaceutical ultrafine grinding system based on the airflow grinding principle provided in an embodiment of the present application.

[0015] Explanation of the reference numerals: pulverizing task instruction acquisition module 10, drug particle size distribution acquisition module 20, D-level pulverizing task information acquisition module 30, pulverizing control trigger space establishment module 40, pulverizing control optimization result generation module 50, cleaning validation result acquisition module 60. DETAILED DESCRIPTION

[0016] The present application provides a method and system for ultrafine grinding of medicines based on the principle of airflow grinding, which is used to solve the technical problems of low medicine grinding accuracy and high medicine grinding loss in the prior art.

[0017] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0018] Embodiment 1

[0019] like Figure 1 As shown, the present application provides a method for ultrafine grinding of medicines based on the principle of airflow grinding, the method comprising:

[0020] Step S100: obtaining a pulverizing task instruction of an airflow pulverizing device, wherein the pulverizing task instruction includes basic information of the medicine to be pulverized and a desired particle size distribution of the medicine.

[0021] Specifically, in industrial production, when airflow crushing equipment is needed to process drugs, the crushing task instruction must be obtained first. This instruction is the starting basis and key guidance for the entire crushing process. The basic information of the drugs to be crushed includes: the type of drug, such as chemical synthetic drugs, Chinese herbal medicine extracts or biological preparations; the chemical composition of the drug, including the main components and their content ratio; the physical properties of the drug, such as density, hardness, etc. These detailed basic information are crucial for selecting appropriate crushing parameters and processes. The expected drug particle size distribution is a specific requirement for the crushing result, expressed in a specific numerical range, such as the expected drug particle size between 5-10 microns accounts for 70%. Clear and accurate goals are set for the entire crushing process, avoiding the blindness of crushing and improving the pertinence and effectiveness of crushing.

[0022] Step S200: performing particle size detection on the medicine to be crushed to obtain real-time medicine particle size distribution.

[0023] Specifically, the particle size of the drug to be crushed is tested. Advanced particle size analysis techniques and equipment, such as laser particle size analyzers and sieving methods, are used in the testing process. Through these testing methods, the size of the drug particles is accurately measured and counted. The real-time drug particle size distribution obtained reflects the actual particle size of the drug to be crushed before it is crushed. This distribution is usually presented in the form of a data chart, including key parameters such as the proportion of particles in different particle size ranges, the average particle size, and the mode particle size. Real-time understanding of the particle size distribution helps to promptly identify problems during the crushing process, such as particles that are too large or unevenly distributed, so as to adjust the process in a timely manner to ensure that the final product meets quality requirements.

[0024] Step S300: dividing the crushing task according to the basic drug information, the real-time drug particle size distribution and the expected drug particle size distribution to obtain D-level crushing task information, where D is a positive integer greater than 1.

[0025] Specifically, the basic information of the drug (such as the composition, properties, hardness, etc. of the drug), the real-time drug particle size distribution (the actual state of the current drug particle size obtained through previous testing), and the expected drug particle size distribution (the ideal state of the drug particle size that is ultimately desired) are comprehensively considered. Based on these various information, the pulverization task is carefully divided into D levels, each with specific goals and requirements. For example, if the expected drug particle size distribution is very fine and the real-time drug particle size distribution is far from the expected, more levels will be divided, such as level 5 or level 6. In the first level pulverization task, the focus is on initially reducing the size of the drug particles to lay the foundation for subsequent finer pulverization. In subsequent levels, the particle size of the drug particles is gradually adjusted to be closer to the expected distribution range. Different pulverization parameters are set for different levels, such as air flow velocity, pulverization time, pressure, etc. Assume that the drug is a fragile substance with a large initial particle size, and a very small and uniform particle size distribution is desired. The first stage of crushing is to crush at a lower airflow speed for a shorter time to prevent over-crushing; the second stage is to appropriately increase the airflow speed and extend the crushing time; the third stage is to further fine-tune the parameters until the desired particle size distribution is achieved. Through this multi-level division, the transition from the current drug state to the desired particle size distribution can be achieved more orderly and accurately, ensuring the controllability of the crushing process and the quality of the final product.

[0026] Step S400: According to the multivariate crushing control factors of the airflow crushing equipment, the crushing control trigger feature mining is performed on the D-level crushing task information to establish a D-level crushing control trigger space.

[0027] Specifically, it is necessary to consider the multivariate crushing control factors of the airflow crushing equipment, which include airflow velocity, pressure, feed rate, temperature and humidity inside the equipment and other parameters that affect the crushing effect. Based on these multivariate control factors, the previously divided D-level crushing task information is deeply analyzed to dig out the features that can trigger effective crushing control at different task levels. For example, in a certain level of crushing task, it is found that when the airflow velocity reaches a specific value, the pressure is within a certain range, and the feed rate remains stable, the ideal crushing effect can be achieved. By mining and integrating these features, a D-level crushing control trigger space is established. This space can be understood as a set of various possible control parameter combinations, each of which corresponds to a specific level of crushing task and can trigger effective crushing control. For example, if D is divided into 3 levels, for the first level crushing task, it may be found that when the airflow velocity is 10m / s, the pressure is 2MPa, and the feed rate is 5kg / min, this level of crushing control can be effectively started. For the second and third level tasks, there will also be corresponding different parameter combinations that will be mined and included in the crushing control trigger space. Establishing such a trigger space helps to quickly and accurately select appropriate control parameters according to different task levels during the actual crushing process, thereby improving crushing efficiency and quality.

[0028] Step S500: activating the multi-order crushing optimization constraint and crushing loss sorting function, performing optimization analysis on the D-level crushing control trigger space in combination with the D-level crushing task information, and generating a D-level crushing control optimization result.

[0029] Specifically, the multi-order pulverization optimization constraints and pulverization loss sorting function are first activated. The multi-order pulverization optimization constraints include various restrictions on different pulverization stages, such as the upper limit of energy consumption, the limit of pulverization time, the minimum standard of product quality, etc. These constraints ensure that the acceptable range will not be exceeded when searching for the optimal pulverization control solution. The pulverization loss sorting function is used to evaluate the various losses generated during the pulverization process, such as material loss, equipment wear, etc. Then, combined with the previously obtained D-level pulverization task information, a comprehensive optimization analysis is performed on the established D-level pulverization control trigger space. This analysis process will comprehensively consider the multi-order pulverization optimization constraints and the pulverization loss sorting function, and explore various possible control parameter combinations in the D-level pulverization control trigger space. Through complex calculations and comparisons, the D-level pulverization control optimization result is finally generated. This result is a control parameter combination that can achieve the optimal pulverization effect while satisfying various constraints and minimizing losses. For example, assuming that the D-level is divided into 3 levels, after optimization analysis, it is concluded that in the first-level crushing task, the best control parameters are air flow velocity of 8m / s and pressure of 1.5MPa; in the second level, the air flow velocity is 12m / s and the pressure is 2.2MPa, etc. The generated D-level crushing control optimization results will provide the most optimized guidance for the actual crushing operation to achieve a crushing effect that is efficient, low-loss and meets quality requirements.

[0030] Step S600: Perform cleaning verification on the airflow pulverizing equipment and the medicine to be pulverized respectively to obtain equipment and medicine cleaning verification results. When the equipment and medicine cleaning verification results are qualified, pulverize the medicine to be pulverized according to the D-level pulverizing control optimization result and the airflow pulverizing equipment.

[0031] Specifically, strict cleaning verification is carried out for the airflow pulverizing equipment and the drugs to be pulverized. For airflow pulverizing equipment, cleaning verification includes checking whether there is any residual drug powder, impurities or contaminants inside the equipment, and ensuring that all parts of the equipment, such as the pulverizing chamber, pipes, filters, etc., meet the cleaning standards. The cleaning verification of the drugs to be pulverized is mainly to confirm that the drugs themselves are not contaminated, such as no other foreign matter or harmful substances are mixed in. The results of the equipment drug cleaning verification are obtained through a series of testing and evaluation methods. Only when this result is judged to be qualified, will the airflow pulverizing equipment be used to pulverize the drugs to be pulverized based on the previously generated D-level pulverization control optimization results. The importance of this link lies in ensuring the hygiene and safety of the pulverization process, preventing cross-contamination, improving the purity of drugs, and ensuring that the produced drugs meet high quality standards.

[0032] In a possible implementation, step S300 further includes:

[0033] Step S310: Loading the medicine crushing task division record set.

[0034] Step S320: performing data cleaning according to the drug crushing task division record set to obtain a drug crushing task division sample set.

[0035] Step S330: constructing a pulverizing task division model according to the drug pulverizing task division sample set.

[0036] Step S340: Based on the basic drug information, the real-time drug particle size distribution and the expected drug particle size distribution, the D-level pulverizing task information is generated according to the pulverizing task division model.

[0037] Specifically, loading the drug pulverization task division record set is to obtain the relevant record data generated when the drug pulverization task division was performed in the past from the storage device or database. These record sets are stored in the form of structured data tables, containing detailed information on the pulverization task division of various drugs in the past. For example, each record covers the name, composition, initial particle size, expected particle size, pulverization process used, and divided task level of a specific drug. The loading process involves communication with the storage system, data reading and parsing, so as to load these records into the current processing environment in the form of a usable data structure, in preparation for subsequent data analysis and processing steps.

[0038] Perform in-depth data cleaning operations on the previously loaded drug crushing task division record set. Check the missing values ​​in the data. For example, some records lack key ingredient information of the drug, or some parameters about the crushing process are not recorded. For these missing values, appropriate methods will be used to process them according to the characteristics and rules of the data, such as filling with the mean value or median. Secondly, outliers will be processed. Outliers are obviously unreasonable data points caused by data entry errors or special circumstances. For example, the particle size record of a certain drug is far beyond the normal range. These outliers need to be verified and corrected. If they cannot be corrected, they may be deleted. Then, duplicate data records will be removed to ensure that each record is unique and valid. Through the above series of data cleaning operations, accurate, clean, and valid data are screened and sorted out from the original drug crushing task division record set, thereby obtaining a drug crushing task division sample set.

[0039] Take the decision tree algorithm as an example to explain the process of building a crushing task partitioning model. The drug crushing task partitioning sample set is divided into a training set, a validation set, and a test set according to a certain ratio (for example, 70%, 15%, and 15%). Use the training set data to build a decision tree. The construction process of the decision tree is as follows: Select an optimal feature as the root node. The selection of this feature is usually based on indicators such as information gain and Gini coefficient. The purpose is to make the subset after division the highest purity. For each subset, repeatedly select the best feature for further division until the stopping condition is met, such as the purity of the subset is high enough, the predetermined tree depth is reached, or the number of samples in the subset is too small. In the process of building a decision tree, by constantly adjusting parameters such as the depth of the tree, the minimum number of leaf node samples, etc., the validation set is used to evaluate the performance of the model under different parameters. For example, if the depth of the tree is too large, it will lead to overfitting, and the model will perform poorly on the validation set; if the depth of the tree is too small, it will lead to underfitting, and the model will not be able to fully learn the patterns in the data. After many trials and adjustments, the model parameters that perform best on the validation set are found. Finally, the test set is used to evaluate the generalization ability and accuracy of the finalized decision tree model.

[0040] First, the basic information of the current drugs to be processed is obtained, including the composition and properties of the drugs; at the same time, the real-time drug particle size distribution data obtained through detection and the set expected drug particle size distribution requirements are obtained. Then, this information is input into the previously constructed crushing task division model, which has learned the relationship between different drug characteristics and reasonable crushing task division through a large amount of sample data. After receiving the input drug-related information, the model will perform internal calculations and reasoning, and analyze and judge the current drug situation based on the patterns and laws learned by the model. Finally, the model outputs D-level crushing task information. This information clearly subdivides the entire crushing task into D levels, each of which contains different crushing parameters, target particle size ranges, operation time and other specific task requirements. In this way, accurate multi-level crushing task planning can be generated according to the characteristics and needs of specific drugs, providing clear and effective guidance for subsequent actual crushing operations.

[0041] In a possible implementation, step S400 further includes:

[0042] Step S410: The multi-element pulverizing control factors include air flow velocity, air flow temperature, air flow humidity, air flow pressure, air flow rate, air flow composition, control time and classification wheel control index.

[0043] Step S420: extracting the dth crushing task information according to the D-level crushing task information, where d is a positive integer and d belongs to D.

[0044] Step S430: Associating the pulverizing control samples of the airflow pulverizing device based on the d th pulverizing task information to generate a d th associated pulverizing control sample set.

[0045] Step S440: performing trigger interval identification on the dth associated pulverization control sample set according to the multivariate pulverization control factor to generate a dth pulverization control trigger domain.

[0046] Step S450: collecting the pulverizing control constraint parameters of the airflow pulverizing device according to the multivariate pulverizing control factor to obtain the dth pulverizing control constraint domain.

[0047] Step S460: performing intersection analysis according to the d pulverizing control trigger domain and the d pulverizing control constraint domain to generate a d pulverizing control trigger space, and adding the d pulverizing control trigger space to the D-level pulverizing control trigger space.

[0048] Specifically, multiple crushing control factors are defined. The airflow velocity refers to the speed of the airflow in the crushing equipment, which directly affects the collision intensity and frequency between the material particles, thereby affecting the crushing effect. The airflow temperature will change the physical properties of the material, such as hardness, brittleness, etc., and thus affect the difficulty and efficiency of crushing. The airflow humidity affects the adhesion and fluidity of the material. Too high or too low humidity may cause the material to agglomerate or disperse unevenly during the crushing process. The airflow pressure determines the energy of the airflow. The greater the pressure, the more sufficient the energy provided to the material crushing. The airflow flow rate reflects the amount of airflow passing through the crushing equipment per unit time, affecting the material transportation and processing volume. The airflow composition, such as the type and purity of the gas, will react chemically with the material or affect the surface properties of the material, thereby indirectly affecting the crushing effect. The control time refers to the duration of the crushing operation, which determines the length of time the material is subjected to the crushing effect. The control index of the classifying wheel involves parameters such as the speed and angle of the classifying wheel, the inner and outer diameter ratio of the classifying wheel, and the number of blades of the classifying wheel, which are used to control the classification and screening of the material particles after crushing. These multi-factor pulverization control factors interact with each other and jointly affect the pulverization performance of the airflow pulverizing equipment and the final pulverization results. Comprehensive consideration and precise control of them are the key to achieving efficient and high-quality pulverization.

[0049] The entire D-level crushing task information has been obtained. The D-level crushing task information actually subdivides a larger crushing task into multiple subtasks of different levels. Then, we need to extract the specific d-th crushing task information from this series of subtasks. Here d is a positive integer, and d is an element in the set D. For example, if the entire crushing task is divided into 5 levels, that is, D={1, 2, 3, 4, 5}, then when d=3, the 3rd level crushing task information will be extracted from the D-level crushing task information. The purpose of this operation is to be able to analyze and process each specific subtask separately, so as to more accurately determine the control parameters and operating conditions related to it.

[0050] Based on the extracted information of the dth crushing task, the past operation data of the airflow crushing equipment is analyzed. First, all historical crushing operation records that are similar to the current dth crushing task in some key features are searched. These key features include the properties of the material to be crushed, the initial particle size, the desired final particle size, and other conditions and requirements related to the crushing task. Then, these similar historical operation records are screened out to form a specific sample set, namely the dth associated crushing control sample set, which contains various control parameters and operation methods used in similar crushing tasks in the past, as well as the corresponding crushing results. By generating such an associated sample set, it can provide valuable reference and basis for the subsequent determination of the optimal control strategy for the current dth crushing task.

[0051] Using the previously determined multivariate pulverization control factors, such as air flow velocity, air flow temperature, air flow humidity, etc., an in-depth analysis of the d-th associated pulverization control sample set is conducted, and the values ​​of the multivariate pulverization control factors corresponding to each sample in the sample set are checked one by one. Through the observation and statistical analysis of a large amount of sample data, the effective value range or interval of each control factor under a specific pulverization effect is found. For example, for the control factor of air flow velocity, it is found that when the pulverization effect reaches the expected level, the air flow velocity is mostly between 20-30m / s. For air flow temperature, better results can be achieved between 40-60°C. Taking into account the effective value ranges of all multivariate pulverization control factors, a parameter combination range that can effectively trigger the ideal pulverization effect is finally determined. This is the d-th pulverization control trigger domain. Assume that in the d-th associated crushing control sample set, after analysis it is found that when the airflow velocity is 15-25m / s, the airflow pressure is 0.5-0.8MPa, and the airflow temperature is 50-70°C and other conditions are met at the same time, the desired crushing effect can be stably achieved. Then the combination of these parameter ranges constitutes the d-th crushing control trigger domain. The generated d-th crushing control trigger domain provides the key parameter basis for the subsequent precise control of the airflow crushing equipment to perform the d-th level crushing task.

[0052] Based on the multivariate pulverization control factors, the characteristics and limitations of the airflow pulverization equipment itself are evaluated and parameters are collected. These limitations come from multiple aspects. For example, the mechanical structure of the equipment has upper or lower limits on airflow velocity, pressure, etc. to avoid excessive wear or damage; the energy supply of the equipment limits the maximum value that certain control factors can reach; or for safety reasons, specific ranges are set for airflow temperature, humidity, etc. By collecting and analyzing the limitations and constraints related to these equipment, the feasible value range of each multivariate pulverization control factor on the equipment is determined. Combining the value ranges of the control factors under all these constraints, the dth pulverization control constraint domain is obtained. The obtained dth pulverization control constraint domain ensures that when performing the pulverization operation, the setting of the control parameters can meet the safety and performance requirements of the equipment without exceeding the capacity of the equipment.

[0053] The dth pulverizing control trigger domain is the parameter range that can effectively trigger the ideal pulverizing effect obtained by analyzing the dth associated pulverizing control sample set. The dth pulverizing control constraint domain is the feasible range of control parameters collected based on the characteristics and restrictions of the airflow pulverizing device itself. By performing an intersection analysis on these two domains, we find their overlapping parts. This overlapping part is the parameter range that can effectively achieve the pulverizing effect and meet the device restrictions and constraints. We define it as the dth pulverizing control trigger space. For example, assuming that the effective range of airflow velocity in the dth pulverizing control trigger domain is 15-25m / s, and the feasible range of airflow velocity in the dth pulverizing control constraint domain is 10-30m / s, then their intersection, 15-25m / s, is the part of the airflow velocity in the dth pulverizing control trigger space. After generating this dth pulverizing control trigger space, add it to the overall D-level pulverizing control trigger space. The D-level pulverizing control trigger space is composed of pulverizing control trigger spaces of all levels (including the dth level). The purpose of this is to gradually improve the entire D-level crushing control trigger space and provide a comprehensive and reliable parameter range for subsequent accurate and effective crushing control at different levels.

[0054] In a possible implementation, step S430 further includes:

[0055] Step S431: performing global device association according to the airflow crushing device to determine a global crushing device, wherein the global crushing device includes the airflow crushing device and a plurality of airflow crushing devices of the same model corresponding to the airflow crushing device.

[0056] Step S432: performing shredding record retrieval according to the global shredding device to obtain a plurality of shredding record groups.

[0057] Step S433: traverse the multiple crushing record groups and extract the first crushing record group, wherein the first crushing record group includes the first sample crushing task information and the first sample crushing control scheme.

[0058] Step S434: performing similarity identification based on the dth crushing task information and the first sample crushing task information to determine the first crushing task similarity coefficient.

[0059] Step S435: Determine whether the first crushing task similarity coefficient is greater than / equal to a preset crushing task similarity coefficient.

[0060] Step S436: If the first crushing task similarity coefficient is greater than / equal to the preset crushing task similarity coefficient, the first sample crushing control scheme is added to the dth associated crushing control sample set.

[0061] Specifically, the association operation is based on the airflow crushing equipment currently in use, and other airflow crushing equipment of the same model will be searched in the database or equipment management system through the key identification information such as the model and specification of the equipment. These equipment of the same model together with the current equipment constitute the global crushing equipment. The purpose of this is to expand the scope of data collection, not only limited to the historical records of the current equipment, but also to integrate the operation and crushing related data of multiple equipment of the same model.

[0062] Using the previously determined global pulverizing equipment, pulverizing records are retrieved from the relevant database or storage system. These pulverizing records usually contain detailed information about each device's past pulverizing operations, such as the type of material pulverized, the initial and final particle size of the material, the control parameters used (such as air flow velocity, temperature, pressure, etc.), the pulverizing time, the quality of the output, etc. Since the global pulverizing equipment includes multiple devices of the same model, the search results will obtain multiple pulverizing record groups, each of which corresponds to a series of pulverizing operation records for one device. By obtaining multiple pulverizing record groups, the operation of multiple devices can be comprehensively analyzed, and more universal and effective laws and patterns can be extracted, providing a rich data basis for subsequent processing steps.

[0063] Multiple crushing record groups will be checked and processed one by one. Traversal means checking these record groups one by one in a certain order. When one of the record groups is processed, it is defined as the first crushing record group. This first crushing record group contains two important pieces of information: the first sample crushing task information and the first sample crushing control plan. The first sample crushing task information describes the specific task requirements of the crushing operation at that time, such as the characteristics of the material to be crushed, the expected crushing particle size, etc. The first sample crushing control plan records the specific control parameters and operating methods used to complete the crushing task, such as specific air flow speed settings, temperature adjustment, pressure control, etc. By extracting such a first crushing record group, preparations can be made for subsequent similarity comparison and analysis with the current dth crushing task.

[0064] Different weights are assigned to the various comparison factors (such as material type, initial particle size, final particle size, material properties, etc.) in the dth crushing task information and the first sample crushing task information. The weight assignment can be determined according to the degree of influence of these factors on the crushing task similarity. Assume that the weight of the material type is w1, the weight of the initial particle size is w2, the weight of the final particle size is w3, the weight of the material property is w4, and w1+w2+w3+w4=1. Then, for each comparison factor, its similarity score is calculated separately. For example, the similarity score of the material type is s1 (the value range is 0 to 1, 1 means exactly the same, 0 means completely different), the similarity score of the initial particle size is s2, the similarity score of the final particle size is s3, and the similarity score of the material property is s4. Next, the weighted formula is used to calculate the similarity coefficient of the first crushing task: the similarity coefficient of the first crushing task = w1×s1+w2×s2+w3×s3+w4×s4, and a value between 0 and 1 is obtained as the similarity coefficient of the first crushing task. The closer the value is to 1, the more similar the dth crushing task is to the first sample crushing task; the closer the value is to 0, the lower the similarity is.

[0065] When the calculated similarity coefficient of the first crushing task is greater than or equal to the preset crushing task similarity coefficient, the operation of adding the first sample crushing control scheme to the dth associated crushing control sample set will be executed. The preset crushing task similarity coefficient is a threshold set in advance, which is used to judge whether the similarity between the two crushing tasks is high enough to determine whether the corresponding sample crushing control scheme has reference value. When the similarity coefficient of the first crushing task meets this threshold condition, it means that the first sample crushing control scheme has a high similarity and reference value with the current dth crushing task. Adding it to the dth associated crushing control sample set can enrich the content of the sample set and provide more reference data and scheme selection for the subsequent generation of a more accurate and effective dth level crushing control strategy. Doing so can make full use of historical experience and control schemes for similar tasks, and improve the accuracy and efficiency of setting the control parameters of the current crushing task.

[0066] In a possible implementation, step S500 further includes:

[0067] Step S510: The multi-stage pulverization optimization constraints include pulverization granularity accuracy constraints, granularity optimization capacity constraints and pulverization loss optimization times threshold.

[0068] Step S520: making a pulverizing control decision according to the d pulverizing control triggering space to obtain a d pulverizing control decision space.

[0069] Step S530: performing granular optimization on the d pulverizing control decision space according to the pulverizing granularity accuracy constraint and the granularity optimization capacity constraint and in combination with the d pulverizing task information to generate a d granularity optimization pulverizing control decision space.

[0070] Step S540: performing iterative optimization of the crushing loss on the d-th particle size optimization crushing control decision space according to the crushing loss sorting function, until the number of iterative optimization of the crushing loss meets the crushing loss optimization number threshold, and obtaining the d-th crushing control optimization result.

[0071] Step S550: adding the d-th pulverizing control optimization result to the D-level pulverizing control optimization result.

[0072] Specifically, the constraints involved in the multi-stage pulverization optimization process are clarified. The pulverization particle size accuracy constraint refers to the requirement for the accuracy of the particle size of the material after final pulverization. For example, it is stipulated that the particle size of the material after pulverization must be within a certain range, and the deviation cannot exceed a certain limit to meet specific production standards or product quality requirements. The particle size optimization capacity constraint limits the adjustable range when searching for the optimal particle size, which includes restrictions on the variable range of pulverization equipment parameters, the upper and lower limits of material processing volume, etc., to ensure that the optimization process is carried out within a feasible operating space. The pulverization loss optimization number threshold sets the maximum number of calculations and optimizations for pulverization loss when seeking the best pulverization control solution. This is to avoid excessive consumption of computing resources and excessive optimization time, and also balances the accuracy and efficiency of optimization to a certain extent. These constraints work together to provide clear rules and restrictions for the multi-stage pulverization optimization process, ensuring that the optimal pulverization control solution is found within a reasonable range.

[0073] The decision on crushing control is made based on the d crushing control trigger space obtained previously. The d crushing control trigger space is a set of parameter ranges that may achieve effective crushing. When making a crushing control decision, selections and combinations will be made within this parameter range to determine the specific control strategy. For example, the range of airflow velocity in the d crushing control trigger space is 20-30 m / s, and the range of crushing time is 10-20 minutes. Then when making a decision, an airflow velocity of 25 m / s and a crushing time of 15 minutes will be selected. Such a combination of specific parameter values ​​constitutes the d crushing control decision space. In this way, a preliminary and feasible control scheme can be determined from many possible parameter combinations, providing a basis for subsequent optimization and adjustment.

[0074] The crushing particle size precision constraint specifies the degree of accuracy that the material particle size needs to achieve after crushing, such as requiring the particle size to be within a certain numerical range and the deviation cannot exceed a specific value. The particle size optimization capacity constraint limits the amplitude and range that can be adjusted when optimizing the particle size, such as setting upper and lower limits for certain equipment parameters or operating conditions. The dth crushing task information contains specific requirements for this crushing task, such as the characteristics of the material to be crushed, the expected crushing effect, etc. The dth crushing control decision space is the range of the crushing control scheme that was preliminarily determined before. When performing particle size optimization, the above-mentioned crushing particle size precision constraint, particle size optimization capacity constraint and the dth crushing task information are comprehensively considered. In the dth crushing control decision space, various possible control parameter combinations are screened and evaluated according to the particle size precision requirements and the optimization capacity constraints. Based on these conditions, exploration and calculation are carried out in the dth crushing control decision space, and the parameter combination range that meets the particle size requirements is screened according to the crushing particle size precision constraint. Then, within this range, combined with the granular optimization capacity constraint, the feasible parameter combinations are further narrowed down. At the same time, the special requirements in the d-th crushing task information are always considered to ensure that the optimized solution matches it. By continuously analyzing and comparing different parameter combinations, the optimal granularity parameter combination is finally determined under the premise of meeting various constraints and task requirements. The space formed by these parameter combinations is the d-th granular optimization crushing control decision space.

[0075] The crushing loss sorting function is defined as pursuing the smaller the comprehensive crushing loss coefficient is, the better, which means that a smaller comprehensive crushing loss coefficient represents a better crushing effect and lower cost. For the d-th particle size optimization crushing control decision space, this function will be used for repeated calculations and optimizations. In each iteration, the corresponding crushing loss comprehensive coefficient will be calculated based on the current parameter combination and compared with the previous results. If the current coefficient is smaller, the current parameter combination will be retained as the new optimal solution; if not, continue to adjust the parameters for the next iteration. Repeat this process until the number of iterative optimization reaches the pre-set crushing loss optimization number threshold. When this threshold is reached, the optimal parameter combination obtained at this time and its corresponding crushing loss comprehensive coefficient are determined as the d-th crushing control optimization result.

[0076] After obtaining the d-th crushing control optimization result, add it to the set of the entire D-level crushing control optimization results. This means that the optimal control solution found for the d-th level crushing task is integrated with the crushing control optimization results of other levels (which may be level 1, level 2... until level d-1) that have been obtained before. By continuously adding the optimization results of each level, a complete system of D-level crushing control optimization results is gradually constructed. The purpose of this is to comprehensively consider all the optimization results of the entire D level in subsequent operations, so as to formulate a more comprehensive, systematic and effective crushing control strategy.

[0077] In a possible implementation, step S530 further includes:

[0078] Step S531: extracting a first pulverizing control decision according to the dth pulverizing control decision space.

[0079] Step S532: activating a crushing particle size accuracy prediction channel, performing crushing particle size accuracy prediction based on the dth crushing task information and the first crushing control decision, and generating a first predicted crushing particle size accuracy.

[0080] Step S533: Determine whether the first predicted crushing particle size accuracy meets the crushing particle size accuracy constraint.

[0081] Step S534: If the first predicted pulverization particle size accuracy satisfies the pulverization particle size accuracy constraint, the first pulverization control decision is added to the d-th particle size optimal pulverization control decision space.

[0082] Step S535: Based on the pulverization particle size accuracy prediction channel, continue to iteratively optimize the d pulverization control decision space according to the pulverization particle size accuracy constraint and the d pulverization task information until the decision capacity of the d pulverization control decision space meets the granularity optimization capacity constraint.

[0083] Specifically, a specific control decision scheme is selected from the d-th pulverization control decision space and marked as the first pulverization control decision. The d-th pulverization control decision space contains multiple possible pulverization control decision options based on different parameter combinations or control strategies. Extracting the first pulverization control decision can be based on some preset rules, such as random selection, selection in a specific order, or selecting the scheme that looks most promising based on some preliminary evaluation indicators.

[0084] Taking the random forest algorithm as an example, how to construct the crushing particle size accuracy prediction channel is described in detail: First, a large amount of historical data needs to be prepared to build the prediction channel. These data should include different crushing task information (such as material type, initial particle size, hardness, etc.) and corresponding crushing control decisions (such as air flow velocity, pressure, time, etc.) and the actual crushing particle size accuracy. Then, these historical data are used to train the random forest model. Random forest is an integrated learning algorithm composed of multiple decision trees. During the training process, for each decision tree, a part of the samples (with replacement sampling) are randomly selected from the historical data as the training set. A part of the features are randomly selected from all the features (i.e., the relevant parameters of the crushing task information and control decisions). Based on the selected training set and features, a decision tree is constructed. Each node of the decision tree is split according to a certain threshold of the selected feature, so that the purity of the node after the split is higher (i.e., the category of the sample in the node is more consistent). By repeating the above process to construct multiple decision trees, a random forest is formed, and the crushing particle size accuracy prediction channel is completed. When a new crushing particle size accuracy prediction is to be made, this prediction channel is activated, and the dth crushing task information and the first crushing control decision are used as input features. The input features are fed into each decision tree in the random forest, and each decision tree will give a predicted crushing particle size accuracy. The prediction results of all decision trees are combined, and the final first predicted crushing particle size accuracy is generated by the average value or majority voting method. The accuracy and reliability of the prediction are improved by continuously updating and optimizing the random forest model with new data.

[0085] Clarify the specific requirements of the crushing particle size accuracy constraint, including the specified particle size range, the allowed degree of deviation, or other relevant accuracy indicators. Then, compare the first predicted crushing particle size accuracy with these constraints. If the first predicted crushing particle size accuracy falls completely within the range specified by the crushing particle size accuracy constraint, or within the allowed deviation range, then it can be judged that the constraint is met. For example, the crushing particle size accuracy constraint requires the final particle size to be between 40-50 microns. If the first predicted crushing particle size accuracy is 45 microns, then the constraint is met.

[0086] When it is determined that the first predicted crushing particle size accuracy meets the crushing particle size accuracy constraint, the operation of adding the first crushing control decision to the d-th particle size optimization crushing control decision space will be executed. This means that the first crushing control decision is considered to be a feasible control scheme that is expected to achieve the ideal crushing particle size accuracy, so it is included in the d-th particle size optimization crushing control decision space. The purpose of this is to gradually enrich and improve this decision space and provide more options for subsequent further optimization and screening.

[0087] Based on the constructed pulverization particle size accuracy prediction channel, the iterative granular optimization operation is continuously carried out on the d-th pulverization control decision space according to the pulverization particle size accuracy constraint and the specific information of the d-th pulverization task. In each iteration, a new control decision is selected from the d-th pulverization control decision space for prediction and evaluation to determine whether it meets the pulverization particle size accuracy constraint. If it meets the constraint, it is added to the d-th granular optimization pulverization control decision space. This process is repeated continuously to continuously explore and screen better control decisions until the number or scale of decisions contained in the d-th granular optimization pulverization control decision space reaches the upper limit set by the granular optimization capacity constraint.

[0088] In a possible implementation, step S532 further includes:

[0089] ;

[0090] Among them, CLY represents the comprehensive coefficient of crushing loss, CLF represents the predicted crushing thermal degradation loss coefficient, CLV represents the predicted crushing adsorption loss coefficient, α represents the thermal degradation loss weight, and β represents the adsorption loss weight.

[0091] Specifically, a comprehensive coefficient CLY for measuring crushing loss is defined here. It consists of two parts, namely the predicted crushing thermal degradation loss coefficient CLF and the predicted crushing adsorption loss coefficient CLV, and these two parts are multiplied by the corresponding weights α and β respectively during the comprehensive calculation. CLF represents the predicted value of the loss caused by thermal degradation during the crushing process. By performing crushing operations in a controlled experimental environment, monitoring the temperature changes during the crushing process and the changes in the chemical properties of the materials, using professional instruments and analysis methods to determine the amount of material loss caused by thermal degradation, and calculating the thermal degradation loss coefficient based on experimental data. CLV represents the predicted value of the loss caused by adsorption. By using surface analysis techniques such as X-ray photoelectron spectroscopy (XPS) and Fourier transform infrared spectroscopy (FTIR), the chemical composition and functional group changes on the surface of the material before and after crushing are studied, and the adsorption loss coefficient is estimated in combination with adsorption theory. The weights α and β are used to reflect the relative importance of thermal degradation loss and adsorption loss in the overall crushing loss. Through this calculation method, the influence of multiple factors on crushing loss can be comprehensively considered, so as to more comprehensively and accurately evaluate the loss in the crushing process. The smaller the comprehensive crushing loss coefficient, the better.

[0092] Embodiment 2

[0093] Based on the same inventive concept as the drug ultrafine grinding method based on the airflow grinding principle in the above-mentioned embodiment, Figure 2As shown, the present application provides a drug ultrafine grinding system based on the airflow grinding principle, and the system and method embodiments in the present application are based on the same inventive concept. The system includes:

[0094] The pulverizing task instruction acquisition module 10 is used to obtain the pulverizing task instruction of the airflow pulverizing device, wherein the pulverizing task instruction includes the basic information of the medicine to be pulverized and the expected particle size distribution of the medicine.

[0095] The drug particle size distribution acquisition module 20 is used to perform particle size detection based on the drug to be crushed to obtain real-time drug particle size distribution.

[0096] The D-level crushing task information acquisition module 30 is used to divide the crushing tasks according to the basic information of the medicine, the real-time medicine particle size distribution and the expected medicine particle size distribution, and obtain the D-level crushing task information, wherein D is a positive integer greater than 1.

[0097] A crushing control trigger space establishment module 40 is used to mine crushing control trigger features of the D-level crushing task information according to the multivariate crushing control factors of the airflow crushing equipment, and establish a D-level crushing control trigger space.

[0098] The pulverizing control optimization result generation module 50 is used to activate the multi-order pulverizing optimization constraints and pulverizing loss sorting function, perform optimization analysis on the D-level pulverizing control trigger space in combination with the D-level pulverizing task information, and generate the D-level pulverizing control optimization result.

[0099] The cleaning validation result acquisition module 60 is used to perform cleaning validation on the airflow pulverizing equipment and the medicine to be pulverized respectively, and obtain the equipment and medicine cleaning validation result. When the equipment and medicine cleaning validation result is qualified, the medicine to be pulverized is pulverized according to the D-level pulverizing control optimization result and the airflow pulverizing equipment.

[0100] Furthermore, the D-level crushing task information acquisition module 30 also includes:

[0101] A partition record set loading unit, wherein the partition record set loading unit is used to load a drug crushing task partition record set.

[0102] A partition sample set acquisition unit is used to perform data cleaning according to the drug crushing task partition record set to obtain a drug crushing task partition sample set.

[0103] A pulverizing task division model construction unit is used to construct a pulverizing task division model according to the drug pulverizing task division sample set.

[0104] A D-level pulverizing task information generation unit generates the D-level pulverizing task information based on the basic drug information, the real-time drug particle size distribution and the expected drug particle size distribution according to the pulverizing task division model.

[0105] Furthermore, the crushing control trigger space establishing module 40 further includes:

[0106] The multi-element pulverization control factors include air flow velocity, air flow temperature, air flow humidity, air flow pressure, air flow rate, air flow composition, control time and classification wheel control index.

[0107] The d-th crushing task information extraction unit is used to extract the d-th crushing task information according to the D-level crushing task information, wherein d is a positive integer and d belongs to D.

[0108] The dth associated pulverizing control sample set generating unit performs pulverizing control sample association on the airflow pulverizing device based on the dth pulverizing task information to generate a dth associated pulverizing control sample set.

[0109] The dth pulverization control trigger domain generation unit is used to identify the trigger interval of the dth associated pulverization control sample set according to the multivariate pulverization control factor to generate the dth pulverization control trigger domain.

[0110] The dth pulverizing control constraint domain acquisition unit is used to collect pulverizing control constraint parameters of the airflow pulverizing device according to the multivariate pulverizing control factor to obtain the dth pulverizing control constraint domain.

[0111] The dth pulverizing control trigger space generation unit is used to perform intersection analysis based on the dth pulverizing control trigger domain and the dth pulverizing control constraint domain to generate the dth pulverizing control trigger space, and add the dth pulverizing control trigger space to the D-level pulverizing control trigger space.

[0112] Furthermore, the dth associated crushing control sample set generating unit further includes:

[0113] A global comminution device determination unit is used to perform global device association according to the airflow comminution device to determine the global comminution device, wherein the global comminution device includes the airflow comminution device and multiple airflow comminution devices of the same model corresponding to the airflow comminution device.

[0114] The shredding record group acquisition unit is used to perform shredding record retrieval according to the global shredding device to obtain multiple shredding record groups.

[0115] The first shredding record group extraction unit is used to traverse the multiple shredding record groups and extract a first shredding record group, wherein the first shredding record group includes first sample shredding task information and a first sample shredding control plan.

[0116] The first crushing task similarity coefficient determining unit is used to perform similarity identification based on the dth crushing task information and the first sample crushing task information to determine the first crushing task similarity coefficient.

[0117] The first crushing task similarity coefficient judgment unit is used to judge whether the first crushing task similarity coefficient is greater than / equal to a preset crushing task similarity coefficient.

[0118] The first sample crushing control scheme adding unit is used to add the first sample crushing control scheme to the dth associated crushing control sample set if the first crushing task similarity coefficient is greater than / equal to the preset crushing task similarity coefficient.

[0119] Furthermore, the pulverizing control optimization result generation module 50 also includes:

[0120] The multi-stage pulverization optimization constraints include pulverization particle size accuracy constraints, particle size optimization capacity constraints and pulverization loss optimization times threshold.

[0121] A control decision space acquisition unit, wherein the d comminution control decision space acquisition unit is used to make a comminution control decision according to the d comminution control trigger space to obtain a d comminution control decision space.

[0122] The optimal pulverizing control decision space generation unit is used to perform granular optimization on the d pulverizing control decision space according to the pulverizing granularity accuracy constraint and the granularity optimization capacity constraint, combined with the d pulverizing task information, to generate a d granularity optimal pulverizing control decision space.

[0123] A control optimization result acquisition unit is used to perform iterative optimization of the crushing loss on the d-th particle size optimization crushing control decision space according to the crushing loss sorting function, until the number of iterative optimization of the crushing loss meets the crushing loss optimization number threshold, thereby obtaining the d-th crushing control optimization result.

[0124] A control optimization result adding unit is used to add the d-th crushing control optimization result to the D-level crushing control optimization result.

[0125] Furthermore, the optimization crushing control decision space generation unit further includes:

[0126] According to the dth pulverizing control decision space, a first pulverizing control decision is extracted.

[0127] The first predicted crushing particle size accuracy generation unit is used to activate the crushing particle size accuracy prediction channel, combine the dth crushing task information and the first crushing control decision to predict the crushing particle size accuracy, and generate a first predicted crushing particle size accuracy.

[0128] The first predicted pulverization particle size accuracy judgment unit is used to judge whether the first predicted pulverization particle size accuracy meets the pulverization particle size accuracy constraint.

[0129] The first pulverizing control decision adding unit is used to add the first pulverizing control decision to the d-th particle size optimization pulverizing control decision space if the first predicted pulverizing particle size accuracy meets the pulverizing particle size accuracy constraint.

[0130] An iterative granularity optimization unit, wherein the iterative granularity optimization unit continues to iteratively optimize the d pulverizing control decision space based on the pulverizing granularity accuracy prediction channel and according to the pulverizing granularity accuracy constraint and the d pulverizing task information until the decision capacity of the d granularity optimized pulverizing control decision space meets the granularity optimization capacity constraint.

[0131] Furthermore, the pulverizing control optimization result generation module 50 also includes:

[0132] ;

[0133] Among them, CLY represents the comprehensive coefficient of crushing loss, CLF represents the predicted crushing thermal degradation loss coefficient, CLV represents the predicted crushing adsorption loss coefficient, α represents the thermal degradation loss weight, and β represents the adsorption loss weight.

[0134] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0135] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0136] This specification and the drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A drug ultrafine grinding method based on the principle of airflow grinding, characterized in that: The method comprises: Obtaining a pulverizing task instruction of an airflow pulverizing device, wherein the pulverizing task instruction includes basic information of the drug to be pulverized and a desired drug particle size distribution; Performing particle size detection on the medicine to be crushed to obtain real-time particle size distribution of the medicine; According to the basic information of the drug, the real-time drug particle size distribution and the expected drug particle size distribution, the crushing task is divided to obtain D-level crushing task information, where D is a positive integer greater than 1; According to the multivariate crushing control factors of the airflow crushing equipment, the crushing control trigger feature mining is performed on the D-level crushing task information to establish the D-level crushing control trigger space; Activate multi-order crushing optimization constraints and crushing loss sorting functions, perform optimization analysis on the D-level crushing control trigger space in combination with the D-level crushing task information, and generate a D-level crushing control optimization result; Performing cleaning verification on the airflow pulverizing equipment and the medicine to be pulverized respectively, obtaining equipment and medicine cleaning verification results, and when the equipment and medicine cleaning verification results are qualified, pulverizing the medicine to be pulverized according to the D-level pulverizing control optimization result and the airflow pulverizing equipment; The crushing task is divided according to the basic drug information, the real-time drug particle size distribution and the expected drug particle size distribution to obtain the D-level crushing task information, including: Load the drug crushing task division record set; Perform data cleaning according to the drug crushing task division record set to obtain a drug crushing task division sample set; According to the drug crushing task division sample set, a crushing task division model is constructed; Based on the basic drug information, the real-time drug particle size distribution and the expected drug particle size distribution, and according to the pulverization task division model, the D-level pulverization task information is generated; According to the multivariate crushing control factors of the airflow crushing equipment, the crushing control triggering feature mining is performed on the D-level crushing task information to establish the D-level crushing control triggering space, including: The multi-element pulverization control factors include air flow velocity, air flow temperature, air flow humidity, air flow pressure, air flow rate, air flow composition, control time and classification wheel control index; According to the D-level crushing task information, extract the d-th crushing task information, wherein d is a positive integer and d belongs to D; Based on the dth pulverizing task information, the pulverizing control sample association of the airflow pulverizing device is performed to generate a dth associated pulverizing control sample set; Perform trigger interval identification on the dth associated pulverization control sample set according to the multivariate pulverization control factor to generate a dth pulverization control trigger domain; According to the multivariate pulverization control factor, the pulverization control constraint parameters of the airflow pulverization device are collected to obtain the dth pulverization control constraint domain; Perform intersection analysis on the dth pulverizing control trigger domain and the dth pulverizing control constraint domain to generate a dth pulverizing control trigger space, and add the dth pulverizing control trigger space to the D-level pulverizing control trigger space; Among them, activating the multi-order crushing optimization constraint and crushing loss sorting function, combining the D-level crushing task information to perform optimization analysis on the D-level crushing control trigger space, and generating the D-level crushing control optimization result, including: The multi-stage crushing optimization constraints include crushing particle size accuracy constraints, particle size optimization capacity constraints and crushing loss optimization times threshold; Perform a crushing control decision according to the d-th crushing control trigger space to obtain a d-th crushing control decision space; According to the pulverization granularity accuracy constraint and the granularity optimization capacity constraint, the d pulverization control decision space is granularly optimized in combination with the d pulverization task information to generate a d granularity optimization pulverization control decision space; According to the crushing loss sorting function, the crushing loss iterative optimization is performed on the d-th particle size optimization crushing control decision space until the number of crushing loss iterative optimizations meets the crushing loss optimization number threshold, thereby obtaining the d-th crushing control optimization result; Adding the d-th pulverizing control optimization result to the D-level pulverizing control optimization result; According to the pulverization granularity accuracy constraint and the granularity optimization capacity constraint, the d pulverization control decision space is granularly optimized in combination with the d pulverization task information to generate the d granularity optimization pulverization control decision space, including: Extracting a first pulverizing control decision according to the dth pulverizing control decision space; Activate a crushing particle size accuracy prediction channel, perform crushing particle size accuracy prediction based on the dth crushing task information and the first crushing control decision, and generate a first predicted crushing particle size accuracy; Determining whether the first predicted crushing particle size accuracy meets the crushing particle size accuracy constraint; If the first predicted pulverization particle size accuracy satisfies the pulverization particle size accuracy constraint, adding the first pulverization control decision to the d-th particle size optimal pulverization control decision space; Based on the pulverization granularity accuracy prediction channel, the d pulverization control decision space is continuously iterated for granularity optimization according to the pulverization granularity accuracy constraint and the d pulverization task information, until the decision capacity of the d granularity optimization pulverization control decision space satisfies the granularity optimization capacity constraint; The crushing loss sorting function is: ； Among them, CLY represents the comprehensive coefficient of crushing loss, CLF represents the predicted crushing thermal degradation loss coefficient, CLV represents the predicted crushing adsorption loss coefficient, α represents the thermal degradation loss weight, and β represents the adsorption loss weight; The pulverizing control sample association of the airflow pulverizing device is performed based on the d-th pulverizing task information to generate the d-th associated pulverizing control sample set, including: Performing global device association according to the airflow crushing device to determine a global crushing device, wherein the global crushing device includes the airflow crushing device and a plurality of airflow crushing devices of the same model corresponding to the airflow crushing device; Performing shredding record retrieval according to the global shredding device to obtain a plurality of shredding record groups; Traversing the plurality of crushing record groups, extracting a first crushing record group, wherein the first crushing record group includes first sample crushing task information and a first sample crushing control scheme; Perform similarity identification based on the dth crushing task information and the first sample crushing task information to determine a first crushing task similarity coefficient; Determining whether the first crushing task similarity coefficient is greater than / equal to a preset crushing task similarity coefficient; If the first crushing task similarity coefficient is greater than / equal to the preset crushing task similarity coefficient, the first sample crushing control scheme is added to the dth associated crushing control sample set.

2. The drug ultrafine grinding system based on the airflow grinding principle is characterized by: The system is used to implement the drug ultrafine grinding method based on the airflow grinding principle according to claim 1, and the system comprises: A pulverizing task instruction acquisition module, wherein the pulverizing task instruction acquisition module is used to obtain a pulverizing task instruction of an airflow pulverizing device, wherein the pulverizing task instruction includes basic information of the medicine to be pulverized and a desired medicine particle size distribution; A drug particle size distribution acquisition module, the drug particle size distribution acquisition module is used to perform particle size detection according to the drug to be crushed to obtain real-time drug particle size distribution; A D-level crushing task information acquisition module, the D-level crushing task information acquisition module is used to divide the crushing task according to the basic information of the drug, the real-time drug particle size distribution and the expected drug particle size distribution, and obtain D-level crushing task information, wherein D is a positive integer greater than 1; A crushing control trigger space establishment module, the crushing control trigger space establishment module is used to mine crushing control trigger features of the D-level crushing task information according to the multivariate crushing control factors of the airflow crushing equipment, and establish a D-level crushing control trigger space; A pulverizing control optimization result generation module, the pulverizing control optimization result generation module is used to activate the multi-order pulverizing optimization constraint and the pulverizing loss sorting function, perform optimization analysis on the D-level pulverizing control trigger space in combination with the D-level pulverizing task information, and generate a D-level pulverizing control optimization result; A cleaning validation result acquisition module is used to perform cleaning validation on the airflow pulverizing equipment and the medicine to be pulverized respectively, and obtain equipment and medicine cleaning validation results. When the equipment and medicine cleaning validation results are qualified, the medicine to be pulverized is pulverized according to the D-level pulverizing control optimization result and the airflow pulverizing equipment.

Images