A cross-scale dynamic regulation method and system for traditional Chinese medicine extraction process

Abstract

This invention belongs to the field of traditional Chinese medicine (TCM) extraction technology, and particularly relates to a method and system for cross-scale dynamic control of TCM extraction processes. The method involves collecting process parameters and corresponding quality attributes of several batches of TCM preparations under small-scale, pilot-scale, and large-scale production conditions during extraction. A multivariate analysis model, a dynamic control standard identification model, a cross-scale nonlinear prediction model, and a multi-scale coupled dynamic simulation model are then trained sequentially. Real-time process parameters are input into the trained cross-scale nonlinear prediction model and multi-scale coupled dynamic simulation model to obtain predicted real-time key quality attributes. This invention achieves real-time prediction and dynamic control of key quality attributes through the coupling of the multivariate analysis model, the dynamic control standard identification model, the cross-scale nonlinear prediction model, and the multi-scale coupled dynamic simulation model, enabling timely adjustment of process parameters to stabilize production and ensure the stability of TCM extraction processes at large-scale production levels.

Description

Technical Field

[0001] This invention belongs to the field of traditional Chinese medicine extraction technology, and in particular relates to a cross-scale dynamic control method and system for traditional Chinese medicine extraction processes. Background Technology

[0002] The efficacy of traditional Chinese medicine (TCM) relies on a complex system of chemical components, and the stable transfer of critical quality attributes (CQAs) is the core of ensuring the clinical efficacy and safety of TCM preparations. The scale-up process from small-scale to pilot-scale to large-scale production in TCM extraction technology is a key link in the industrialization of TCM. This process involves the dynamic coupling of multiple factors such as raw material characteristics, process parameters, and equipment structure, which directly affects the consistency of critical quality attributes such as the dissolution of characteristic components and the yield of extract.

[0003] Current scale-up methods for traditional Chinese medicine extraction processes mainly rely on traditional experience, which has several drawbacks: 1. Optimized process parameters from small-scale trials are difficult to apply directly to pilot-scale and large-scale production due to differences in mass and heat transfer efficiencies, leading to significant fluctuations in key quality attributes and poor batch-to-batch consistency; 2. The coupling effect of multiple factors—raw materials, processes, and equipment—is not fully considered, making it difficult to quantify the nonlinear impact of equipment flow field characteristics and process parameter gradient changes on key quality attributes; 3. A comprehensive simulation model covering multiple physical fields such as fluid dynamics, mass and heat transfer during the extraction process is lacking, making dynamic control of the process impossible, resulting in long R&D cycles and high production costs. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a cross-scale dynamic control method and system for the extraction process of traditional Chinese medicine, aiming to solve the problems mentioned in the background art.

[0005] In a first aspect, the present invention provides a method for cross-scale dynamic control of traditional Chinese medicine extraction processes, comprising the following steps: Collect process parameters and corresponding quality attributes of several batches of traditional Chinese medicine preparations under small-scale, pilot-scale and large-scale production conditions in the extraction process; Construct a multivariate analysis model, a dynamic control standard identification model, a process cross-scale nonlinear prediction model, and a multi-scale coupled dynamic simulation model; train the multivariate analysis model, the dynamic control standard identification model, the process cross-scale nonlinear prediction model, and the multi-scale coupled dynamic simulation model sequentially using process parameters and their corresponding quality attributes from several batches. The real-time process parameters of the extraction process of traditional Chinese medicine preparations in large-scale production are input into the trained process cross-scale nonlinear prediction model and multi-scale coupled dynamic simulation model to obtain the predicted real-time key quality attributes. The predicted real-time critical quality attributes are compared with the dynamic control range of the critical quality attributes. If the predicted real-time critical quality attributes are within the dynamic control range of the critical quality attributes, production is stabilized. If the predicted real-time critical quality attributes exceed the dynamic control range of the critical quality attributes, the real-time process parameters are adjusted until the predicted real-time critical quality attributes are within the dynamic control range of the critical quality attributes, and then production is stabilized.

[0006] Furthermore, multivariate analysis models are used to screen key chemical quality attributes; dynamic control standard identification models are used to obtain the dynamic control range of key quality attributes; process cross-scale nonlinear prediction models are used to realize the dynamic correlation between process parameters and key quality attributes; and multi-scale coupled dynamic simulation models are used to establish the mapping relationship between process parameters and key quality attributes.

[0007] Furthermore, it also includes: collecting the process parameters and corresponding quality attributes of the extraction process of traditional Chinese medicine preparations during stable production, and periodically retraining the multivariate analysis model, dynamic control standard identification model, process cross-scale nonlinear prediction model and multi-scale coupled dynamic simulation model to achieve dynamic adjustment and parameter optimization of the multivariate analysis model, dynamic control standard identification model, process cross-scale nonlinear prediction model and multi-scale coupled dynamic simulation model.

[0008] Furthermore, multivariate analysis models are used to screen key chemical quality attributes, specifically: The multivariate analysis model consists of a data preprocessing module and a key parameter screening module; Several batches of process parameters and their corresponding quality attributes are input into the data preprocessing module for standardization, resulting in standardized process parameters and their corresponding quality attributes. Then, they are input into the key parameter screening module, where principal component analysis is used to reduce the dimensionality of the standardized process parameters, and partial least squares regression is used to perform correlation analysis on the standardized process parameters and their corresponding quality attributes to determine the key chemical quality attributes.

[0009] Furthermore, the dynamic control standard identification model is used to obtain the dynamic control range of key quality attributes, specifically: The dynamic control standard identification model consists of a weight analysis module and a statistics module; Process parameter fluctuations, batch differences, and key chemical quality attributes are input into the weighting analysis module to assign weights to the key chemical quality attributes; then, they are input into the statistics module, where statistical process control, tolerance interval estimation, or analysis of variance are used to obtain the dynamic control range of the key chemical quality attributes.

[0010] Furthermore, a cross-scale nonlinear prediction model for the process is used to achieve a dynamic correlation between process parameters and key quality attributes, specifically: The process cross-scale nonlinear prediction model consists of a fast prediction module and a local verification and correction module. Several batches of process parameters and their corresponding key chemical quality attributes are input into the rapid prediction module, where a backpropagation neural network is used for mapping calculation to obtain the prediction results. Then, the results are input into the local verification and correction module, where the local approximation capability of the radial basis function neural network is used to correct the local error of the prediction results, thereby realizing the dynamic correlation between process parameters and key quality attributes and obtaining the predicted values ​​of key quality attributes during process scale-up.

[0011] Furthermore, a multi-scale coupled dynamic simulation model is used to establish the mapping relationship between process parameters and key quality attributes, specifically: The multi-scale coupled dynamic simulation model consists of a dynamics extraction module, a fluid dynamics calculation module, and a machine learning proxy module. The extraction kinetics module quantitatively describes the extraction process of traditional Chinese medicine components, while the fluid dynamics calculation module numerically solves the flow field, temperature field, and concentration field during the extraction process. The machine learning proxy module uses machine learning algorithms to fit the mapping relationship between the numerical solution results, process parameters, and predicted values ​​of key quality attributes, thereby achieving real-time visualization of process parameters.

[0012] Secondly, the present invention provides a cross-scale dynamic control system for traditional Chinese medicine extraction processes, comprising: The multivariate analysis model consists of a data preprocessing module and a key parameter screening module. Several batches of process parameters and their corresponding quality attributes are input into the data preprocessing module for standardization to obtain standardized process parameters and their corresponding quality attributes. Then, they are input into the key parameter screening module, where principal component analysis is used to reduce the dimensionality of the standardized process parameters, and partial least squares regression is used to perform correlation analysis on the standardized process parameters and their corresponding quality attributes to screen and determine key chemical quality attributes. The dynamic control standard identification model consists of a weighting analysis module and a statistics module. Process parameter fluctuations, batch differences, and key chemical quality attributes are input into the weighting analysis module to assign weights to the key chemical quality attributes. Then, the data is input into the statistics module, where statistical process control, tolerance interval estimation, or analysis of variance is used to obtain the dynamic control range of the key chemical quality attributes. The process cross-scale nonlinear prediction model consists of a rapid prediction module and a local verification and correction module. Several batches of process parameters and their corresponding key chemical quality attributes are input into the rapid prediction module, and a backpropagation neural network is used for mapping calculation to obtain the prediction results. Then, the results are input into the local verification and correction module, and the local approximation capability of the radial basis function neural network is used to correct the local error of the prediction results, thereby realizing the dynamic correlation between process parameters and key quality attributes and obtaining the predicted values ​​of key quality attributes during process scale-up. The multi-scale coupled dynamic simulation model consists of an extraction kinetics module, a fluid dynamics calculation module, and a machine learning proxy module. The extraction kinetics module quantitatively describes the extraction process of traditional Chinese medicine components, the fluid dynamics calculation module numerically solves the flow field, temperature field, and concentration field during the extraction process, and the machine learning proxy module uses machine learning algorithms to fit the mapping relationship between the numerical solution results, process parameters, and predicted values ​​of key quality attributes, thereby achieving real-time visualization of process parameters.

[0013] Thirdly, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements a cross-scale dynamic control method for traditional Chinese medicine extraction processes.

[0014] Fourthly, the present invention provides an electronic device including at least one processor and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform a cross-scale dynamic control method for a traditional Chinese medicine extraction process.

[0015] The present invention has the following beneficial effects: (1) By integrating multivariate analysis, machine learning algorithms and multiphysics simulation technology, a cross-scale dynamic control method covering small-scale, pilot-scale and large-scale production of traditional Chinese medicine extraction process is constructed to reduce the uncertainty and resource consumption in the scale-up process of traditional Chinese medicine extraction process.

[0016] (2) By coupling the multivariate analysis model, the dynamic control standard identification model, the process cross-scale nonlinear prediction model and the multi-scale coupled dynamic simulation model, the real-time prediction and dynamic control of key quality attributes can be realized, the process parameters can be adjusted in a timely manner to stabilize production, and the stability of the Chinese medicine extraction process under large-scale production can be ensured. Attached Figure Description

[0017] Exemplary embodiments of the present invention can be more fully understood by referring to the following figures: Figure 1 This is a flowchart of a cross-scale dynamic control method for a traditional Chinese medicine extraction process according to the present invention. Detailed Implementation

[0018] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention.

[0020] This invention provides a method for cross-scale dynamic control of traditional Chinese medicine extraction processes, comprising the following steps: Collect process parameters and corresponding quality attributes of several batches of traditional Chinese medicine preparations under small-scale, pilot-scale and large-scale production conditions in the extraction process; Construct a multivariate analysis model, a dynamic control standard identification model, a process cross-scale nonlinear prediction model, and a multi-scale coupled dynamic simulation model; train the multivariate analysis model, the dynamic control standard identification model, the process cross-scale nonlinear prediction model, and the multi-scale coupled dynamic simulation model sequentially using process parameters and their corresponding quality attributes from several batches. The real-time process parameters of the extraction process of traditional Chinese medicine preparations in large-scale production are input into the trained process cross-scale nonlinear prediction model and multi-scale coupled dynamic simulation model to obtain the predicted real-time key quality attributes. The predicted real-time critical quality attributes are compared with the dynamic control range of the critical quality attributes. If the predicted real-time critical quality attributes are within the dynamic control range of the critical quality attributes, production is stabilized. If the predicted real-time critical quality attributes exceed the dynamic control range of the critical quality attributes, the real-time process parameters are adjusted until the predicted real-time critical quality attributes are within the dynamic control range of the critical quality attributes, and then production is stabilized.

[0021] It is understandable that process parameters such as extraction temperature, material-to-liquid ratio, extraction time, stirring rate, number of extractions, and concentration temperature are important. Quality attributes such as the content of characteristic components, yield, impurity content, moisture content of the extract, and characteristic chromatograms; Key quality attributes such as the content of characteristic ingredients and the yield of ointment.

[0022] In some embodiments, a multivariate analysis model is used to screen key chemical quality attributes; a dynamic control standard identification model is used to obtain the dynamic control range of key quality attributes; a process cross-scale nonlinear prediction model is used to realize the dynamic correlation between process parameters and key quality attributes; and a multi-scale coupled dynamic simulation model is used to establish the mapping relationship between process parameters and key quality attributes.

[0023] In some embodiments, the method further includes: collecting process parameters and corresponding quality attributes of the extraction process of traditional Chinese medicine preparations during stable production; and periodically retraining the multivariate analysis model, dynamic control standard identification model, process cross-scale nonlinear prediction model, and multi-scale coupled dynamic simulation model to achieve dynamic adjustment and parameter optimization of the multivariate analysis model, dynamic control standard identification model, process cross-scale nonlinear prediction model, and multi-scale coupled dynamic simulation model.

[0024] In some embodiments, the multivariate analysis model is used to screen key chemical quality attributes, specifically: The multivariate analysis model consists of a data preprocessing module and a key parameter screening module; Several batches of process parameters and their corresponding quality attributes are input into the data preprocessing module for standardization, resulting in standardized process parameters and their corresponding quality attributes. Then, they are input into the key parameter screening module, where principal component analysis is used to reduce the dimensionality of the standardized process parameters, and partial least squares regression is used to perform correlation analysis on the standardized process parameters and their corresponding quality attributes to determine the key chemical quality attributes.

[0025] Specifically, in the data preprocessing module, the original process parameters and quality attribute data are standardized to eliminate the influence of dimensions. This involves constructing a data matrix containing n samples (batches) and m variables (process parameters + quality attributes), and then standardizing the data. The formula is expressed as: In the formula, x ij This represents the original value of the j-th variable in the i-th sample; and denoted as the mean and standard deviation of the j-th variable, respectively.

[0026] In some embodiments, the dynamic control standard identification model is used to obtain the dynamic control range of key quality attributes, specifically: The dynamic control standard identification model consists of a weight analysis module and a statistics module; Process parameter fluctuations, batch differences, and key chemical quality attributes are input into the weighting analysis module to assign weights to the key chemical quality attributes; then, they are input into the statistics module, where statistical process control, tolerance interval estimation, or analysis of variance are used to obtain the dynamic control range of the key chemical quality attributes.

[0027] In some embodiments, the process cross-scale nonlinear prediction model is used to achieve a dynamic correlation between process parameters and key quality attributes, specifically: The process cross-scale nonlinear prediction model consists of a fast prediction module and a local verification and correction module. Several batches of process parameters and their corresponding key chemical quality attributes are input into the rapid prediction module, where a backpropagation neural network is used for mapping calculation to obtain the prediction results. Then, the results are input into the local verification and correction module, where the local approximation capability of the radial basis function neural network is used to correct the local error of the prediction results, thereby realizing the dynamic correlation between process parameters and key quality attributes and obtaining the predicted values ​​of key quality attributes during process scale-up.

[0028] In some embodiments, the multi-scale coupled dynamic simulation model is used to establish the mapping relationship between process parameters and key quality attributes, specifically: The multi-scale coupled dynamic simulation model consists of a dynamics extraction module, a fluid dynamics calculation module, and a machine learning proxy module. The extraction kinetics module quantitatively describes the extraction process of traditional Chinese medicine components, while the fluid dynamics calculation module numerically solves the flow field, temperature field, and concentration field during the extraction process. The machine learning agent (XGBoost) module uses machine learning algorithms to fit the mapping relationship between the numerical solution results, process parameters, and predicted values ​​of key quality attributes, thereby achieving real-time visualization of process parameters.

[0029] Specifically, in the extraction kinetics module, at least one mathematical model among first-order kinetics, Weibull, and Higuchi is used to quantitatively describe the extraction process of traditional Chinese medicine components, depending on the characteristics of the extraction system and the required fitting accuracy. The first-order kinetic mathematical model is a classical kinetic model based on reaction rate theory, which is suitable for simple extraction processes with relatively fast dissolution rates. The Weibull mathematical model is a semi-empirical model based on probability statistics, which has good adaptability to the extraction process of complex traditional Chinese medicine systems. The Higuchi mathematical model, derived based on Fick's diffusion law, is suitable for describing extraction processes where diffusion is the rate-limiting step. In the fluid dynamics calculation module, the flow field, temperature field, and concentration field inside the extraction tank are numerically solved. Specifically, the flow field distribution is obtained by solving the Navier-Stokes momentum equation, the temperature field distribution is obtained by solving the energy conservation equation, and the concentration field distribution is obtained by solving the convection-diffusion component transport equation, thus realizing the coupled numerical solution of multiple physics fields inside the extraction tank. In the machine learning agent module, multiple decision trees are integrated to fit complex mapping relationships. The objective function includes a loss function and a regularization term.

[0030] In some embodiments, the present invention provides a cross-scale dynamic control system for traditional Chinese medicine extraction processes, comprising: The multivariate analysis model consists of a data preprocessing module and a key parameter screening module. Several batches of process parameters and their corresponding quality attributes are input into the data preprocessing module for standardization to obtain standardized process parameters and their corresponding quality attributes. Then, they are input into the key parameter screening module, where principal component analysis is used to reduce the dimensionality of the standardized process parameters, and partial least squares regression is used to perform correlation analysis on the standardized process parameters and their corresponding quality attributes to screen and determine key chemical quality attributes. The dynamic control standard identification model consists of a weighting analysis module and a statistics module. Process parameter fluctuations, batch differences, and key chemical quality attributes are input into the weighting analysis module to assign weights to the key chemical quality attributes. Then, the data is input into the statistics module, where statistical process control, tolerance interval estimation, or analysis of variance is used to obtain the dynamic control range of the key chemical quality attributes. The process cross-scale nonlinear prediction model consists of a rapid prediction module and a local verification and correction module. Several batches of process parameters and their corresponding key chemical quality attributes are input into the rapid prediction module, and a backpropagation neural network is used for mapping calculation to obtain the prediction results. Then, the results are input into the local verification and correction module, and the local approximation capability of the radial basis function neural network is used to correct the local error of the prediction results, thereby realizing the dynamic correlation between process parameters and key quality attributes and obtaining the predicted values ​​of key quality attributes during process scale-up. The multi-scale coupled dynamic simulation model consists of an extraction kinetics module, a fluid dynamics calculation module, and a machine learning proxy module. The extraction kinetics module quantitatively describes the extraction process of traditional Chinese medicine components, the fluid dynamics calculation module numerically solves the flow field, temperature field, and concentration field during the extraction process, and the machine learning proxy module uses machine learning algorithms to fit the mapping relationship between the numerical solution results, process parameters, and predicted values ​​of key quality attributes, thereby achieving real-time visualization of process parameters.

[0031] In some embodiments, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements a cross-scale dynamic control method for traditional Chinese medicine extraction processes.

[0032] In some embodiments, the present invention provides an electronic device including at least one processor and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform a cross-scale dynamic control method for a traditional Chinese medicine extraction process.

[0033] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for cross-scale dynamic control of traditional Chinese medicine extraction process, characterized in that, Includes the following steps: Collect process parameters and corresponding quality attributes of several batches of traditional Chinese medicine preparations under small-scale, pilot-scale and large-scale production conditions in the extraction process; Construct a multivariate analysis model, a dynamic control standard identification model, a process cross-scale nonlinear prediction model, and a multi-scale coupled dynamic simulation model; train the multivariate analysis model, the dynamic control standard identification model, the process cross-scale nonlinear prediction model, and the multi-scale coupled dynamic simulation model sequentially using process parameters and their corresponding quality attributes from several batches. The real-time process parameters of the extraction process of traditional Chinese medicine preparations in large-scale production are input into the trained process cross-scale nonlinear prediction model and multi-scale coupled dynamic simulation model to obtain the predicted real-time key quality attributes. The predicted real-time critical quality attributes are compared with the dynamic control range of the critical quality attributes. If the predicted real-time critical quality attributes are within the dynamic control range of the critical quality attributes, production is stabilized. If the predicted real-time critical quality attributes exceed the dynamic control range of the critical quality attributes, the real-time process parameters are adjusted until the predicted real-time critical quality attributes are within the dynamic control range of the critical quality attributes, and then production is stabilized.

2. The method for cross-scale dynamic control of traditional Chinese medicine extraction process as described in claim 1, characterized in that, Multivariate analysis models are used to screen key chemical quality attributes; dynamic control standard identification models are used to obtain the dynamic control range of key quality attributes; and process cross-scale nonlinear prediction models are used to realize the dynamic correlation between process parameters and key quality attributes. Multi-scale coupled dynamic simulation models are used to establish the mapping relationship between process parameters and key quality attributes.

3. The method for cross-scale dynamic control of traditional Chinese medicine extraction process as described in claim 2, characterized in that, Also includes: The process parameters and corresponding quality attributes of the extraction process of traditional Chinese medicine preparations during stable production are collected. The multivariate analysis model, dynamic control standard identification model, process cross-scale nonlinear prediction model and multi-scale coupled dynamic simulation model are retrained regularly to achieve dynamic adjustment and parameter optimization of the multivariate analysis model, dynamic control standard identification model, process cross-scale nonlinear prediction model and multi-scale coupled dynamic simulation model.

4. The method for cross-scale dynamic control of traditional Chinese medicine extraction process as described in claim 3, characterized in that, Multivariate analysis models are used to screen key chemical quality attributes, specifically: The multivariate analysis model consists of a data preprocessing module and a key parameter screening module; Several batches of process parameters and their corresponding quality attributes are input into the data preprocessing module for standardization, resulting in standardized process parameters and their corresponding quality attributes. Then, they are input into the key parameter screening module, where principal component analysis is used to reduce the dimensionality of the standardized process parameters, and partial least squares regression is used to perform correlation analysis on the standardized process parameters and their corresponding quality attributes to determine the key chemical quality attributes.

5. The method for cross-scale dynamic control of traditional Chinese medicine extraction process as described in claim 4, characterized in that, The dynamic control standard identification model is used to obtain the dynamic control range of key quality attributes, specifically: The dynamic control standard identification model consists of a weight analysis module and a statistics module; Input process parameter fluctuations, batch differences, and key chemical quality attributes into the weight analysis module, and assign weights to the key chemical quality attributes. Then, by inputting the data into the statistics module, statistical process control, tolerance interval estimation, or analysis of variance can be used to obtain the dynamic control range of key chemical quality attributes.

6. The method for cross-scale dynamic control of traditional Chinese medicine extraction process as described in claim 5, characterized in that, A cross-scale nonlinear prediction model for processes is used to achieve a dynamic correlation between process parameters and key quality attributes, specifically: The process cross-scale nonlinear prediction model consists of a fast prediction module and a local verification and correction module. Several batches of process parameters and their corresponding key chemical quality attributes are input into the rapid prediction module, where a backpropagation neural network is used for mapping calculation to obtain the prediction results. Then, the results are input into the local verification and correction module, where the local approximation capability of the radial basis function neural network is used to correct the local error of the prediction results, thereby realizing the dynamic correlation between process parameters and key quality attributes and obtaining the predicted values ​​of key quality attributes during process scale-up.

7. The method for cross-scale dynamic control of traditional Chinese medicine extraction process as described in claim 6, characterized in that, Multi-scale coupled dynamic simulation models are used to establish the mapping relationship between process parameters and key quality attributes, specifically: The multi-scale coupled dynamic simulation model consists of a dynamics extraction module, a fluid dynamics calculation module, and a machine learning proxy module. The extraction kinetics module quantitatively describes the extraction process of traditional Chinese medicine components, while the fluid dynamics calculation module numerically solves the flow field, temperature field, and concentration field during the extraction process. The machine learning agent module uses machine learning algorithms to fit a mapping relationship between the numerical solution results, process parameters, and predicted values ​​of key quality attributes, thereby enabling real-time visualization of process parameters.

8. A cross-scale dynamic control system for traditional Chinese medicine extraction processes, characterized in that, include: The multivariate analysis model consists of a data preprocessing module and a key parameter screening module. Several batches of process parameters and their corresponding quality attributes are input into the data preprocessing module for standardization to obtain standardized process parameters and their corresponding quality attributes. Then, they are input into the key parameter screening module, where principal component analysis is used to reduce the dimensionality of the standardized process parameters, and partial least squares regression is used to perform correlation analysis on the standardized process parameters and their corresponding quality attributes to screen and determine key chemical quality attributes. The dynamic control standard identification model consists of a weight analysis module and a statistics module. Process parameter fluctuations, batch differences, and key chemical quality attributes are input into the weight analysis module, and weights are assigned to the key chemical quality attributes. Then, input the data into the statistics module and use statistical process control, tolerance interval estimation, or analysis of variance to obtain the dynamic control range of key chemical quality attributes. The process cross-scale nonlinear prediction model consists of a rapid prediction module and a local verification and correction module. Several batches of process parameters and their corresponding key chemical quality attributes are input into the rapid prediction module, and a backpropagation neural network is used for mapping calculation to obtain the prediction results. Then, the results are input into the local verification and correction module, and the local approximation capability of the radial basis function neural network is used to correct the local error of the prediction results, thereby realizing the dynamic correlation between process parameters and key quality attributes and obtaining the predicted values ​​of key quality attributes during process scale-up. The multi-scale coupled dynamic simulation model consists of an extraction kinetics module, a fluid dynamics calculation module, and a machine learning proxy module. The extraction kinetics module quantitatively describes the extraction process of traditional Chinese medicine components, while the fluid dynamics calculation module numerically solves the flow field, temperature field, and concentration field during the extraction process. The machine learning agent module uses machine learning algorithms to fit a mapping relationship between the numerical solution results, process parameters, and predicted values ​​of key quality attributes, thereby enabling real-time visualization of process parameters.

9. A computer-readable storage medium, characterized in that: The device contains a computer program that, when executed by a processor, implements a cross-scale dynamic control method for a traditional Chinese medicine extraction process as described in any one of claims 1 to 7.

10. An electronic device, characterized in that: It includes at least one processor and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform a cross-scale dynamic control method for a traditional Chinese medicine extraction process as described in any one of claims 1 to 7.

Images