CN-122012820-A - Traditional Chinese medicine fermentation method and system

CN122012820ACN 122012820 ACN122012820 ACN 122012820ACN-122012820-A

Abstract

A method and a system for fermenting traditional Chinese medicine belong to the technical field of traditional Chinese medicine fermentation, and the method comprises the following steps of obtaining real-time fermentation data; the real-time fermentation data comprises sensor time sequence data, microorganism image data and process text data, the real-time fermentation data is analyzed through a multi-mode large model to obtain fermentation quality score and parameter adjustment advice, and the traditional Chinese medicine fermentation process of the fermentation tank is adjusted according to the fermentation quality score and the parameter adjustment advice. According to the application, the real-time fermentation data is analyzed through the multi-mode large model, so that the fermentation quality score and parameter adjustment suggestion can be quickly and accurately obtained, and the problems of low accuracy of the traditional Chinese medicine fermentation quality prediction and low content of active ingredients of the traditional Chinese medicine after fermentation are solved.

Inventors

JIANG ZHENGHUA
XU DEMING
ZHU XINHUA
WANG CHENGJIA
JIANG MENGLIN

Assignees

杭州铮铭信息科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260123

Claims (10)

1. A traditional Chinese medicine fermentation method is characterized by comprising the following steps: Acquiring real-time fermentation data, wherein the real-time fermentation data comprises sensor time sequence data, microorganism image data and process text data; analyzing real-time fermentation data through a multi-mode large model to obtain fermentation quality scores and parameter adjustment suggestions; and adjusting the fermentation process of the traditional Chinese medicine in the fermentation tank according to the fermentation quality score and the parameter adjustment suggestion.
2. The method of claim 1, wherein the analyzing real-time fermentation data by multi-modal large model comprises the steps of: Performing numerical coding, visual coding and text coding on the real-time fermentation data to obtain coding feature data, wherein the coding feature data comprises a time sequence feature vector, an image feature matrix and a text semantic vector; calculating the characteristic association weight of the coded characteristic data through an attention mechanism, and outputting a multi-mode characteristic tensor according to the characteristic association weight; reasoning the multi-mode characteristic tensor by combining a tree structure algorithm with knowledge rules of the fermentation field to obtain an intermediate reasoning result, wherein the intermediate reasoning result comprises the current fermentation quality state and the parameter optimization direction; and converting the intermediate reasoning result into fermentation quality score and parameter adjustment advice.
3. The method of claim 2, wherein the multimodal mass model comprises Qwen model and Vision Transformer model; The visual code is based on Vision Transformer model, the text code is based on Qwen model; The tree structure algorithm is a decision tree algorithm, and the knowledge rule in the fermentation field is a knowledge graph in the fermentation field.
4. The method of claim 2, wherein the multimodal mass model comprises LLaMA model and Swin transducer model; The visual code is based on a Swin transform model, and the text code is based on a LLaMA model; the attention mechanism is a cross attention mechanism, and the feature association weight is fine granularity association among different modal features; the tree structure algorithm is a random forest algorithm, and the fermentation field knowledge rule is a fermentation field rule base.
5. The method of claim 4, further comprising, prior to outputting the fermentation quality score: and comparing whether the text semantic vector extracted by the LLaMA model is matched with the image feature matrix extracted by the Swin transducer model.
6. The method of claim 3-5, wherein the training of the multimodal mass model comprises the steps of: Collecting multiple batches of fermentation data of the traditional Chinese medicinal materials, wherein each batch of fermentation data comprises sensor time sequence data, microorganism image data, process text data and final effective component content detection results; cleaning the multi-batch fermentation data; Marking fermentation quality scores for each batch of fermentation data according to the final detection result of the content of the effective components; dividing the multi-batch fermentation data into a training set, a verification set and a test set; And training the multi-mode large model according to the training set, the verification set and the test set.
7. The method for fermenting traditional Chinese medicine according to claim 6, further comprising, after acquiring the real-time fermentation data: Preprocessing sensor time sequence data, microorganism image data and process text data; Wherein the pretreatment comprises the following steps: Removing abnormal values in the data; filling in missing values in the data.
8. A traditional Chinese medicine fermentation system comprising computer means comprising a memory, a processor and a computer program stored on the memory, characterized in that the processor executes the computer program to carry out the steps of the method of any one of claims 1-7.
9. The traditional Chinese medicine fermentation system of claim 8, further comprising: A fermentation tank; The Internet of things sensor is connected with the fermentation tank, is in communication connection with the computer device and is used for acquiring sensor time sequence data; the optical detection module is in communication connection with the computer device and is used for acquiring microorganism image data; The temperature control system is arranged outside the fermentation tank and is in communication connection with the computer device and used for adjusting the fermentation temperature; The pH adjusting system is connected with the fermentation tank, is in communication connection with the computer device and is used for adjusting the fermentation pH; and the stirring system is arranged in the fermentation tank, is in communication connection with the computer device and is used for adjusting the stirring speed.
10. A computer program product comprising a computer program, characterized in that the computer program, when being executed by a processor, realizes the steps of the method according to any one of claims 1-7.

Description

Traditional Chinese medicine fermentation method and system Technical Field The invention belongs to the technical field of traditional Chinese medicine fermentation, and particularly relates to a traditional Chinese medicine fermentation method and system. Background The traditional Chinese medicine fermentation tank is core equipment specially used for the traditional Chinese medicine fermentation process, the key functions of the traditional Chinese medicine fermentation tank cover the accurate regulation and control of the temperature and the humidity in the fermentation environment, efficient sterilization treatment and automatic operation, and various application scenes from laboratory-level small-scale tests to industrialized large-scale production can be widely supported. At present, the traditional Chinese medicine fermentation technology generally adopts preset fixed technological parameters, and manual adjustment is mainly carried out according to experience of operators in the actual production process. The system can only monitor basic environmental parameters such as temperature, pH value and the like, and lacks quality prediction and process dynamic optimization capacity based on artificial intelligence. The key decision is still mainly manually judged, so that the error rate of judgment is high, and the accuracy and stability of fermentation process control are obviously insufficient. Therefore, development of a method and a system for fermenting traditional Chinese medicine is needed to solve the problems in the prior art. Disclosure of Invention The invention aims to provide a traditional Chinese medicine fermentation method and system, which can quickly and accurately obtain fermentation quality scores and parameter adjustment suggestions by analyzing real-time fermentation data through a multi-mode large model, and adjust the traditional Chinese medicine fermentation process in real time according to the fermentation quality scores and the parameter adjustment suggestions, so that the problems of low accuracy of traditional Chinese medicine fermentation quality prediction and low content of active ingredients of the traditional Chinese medicine after fermentation are solved. In order to solve the technical problems, the specific technical scheme of the invention is as follows: A traditional Chinese medicine fermentation method comprises the following steps: Acquiring real-time fermentation data, wherein the real-time fermentation data comprises sensor time sequence data, microorganism image data and process text data; analyzing real-time fermentation data through a multi-mode large model to obtain fermentation quality scores and parameter adjustment suggestions; and adjusting the fermentation process of the traditional Chinese medicine in the fermentation tank according to the fermentation quality score and the parameter adjustment suggestion. Further, the real-time fermentation data is analyzed by the multi-modal large model, comprising the following steps: Performing numerical coding, visual coding and text coding on the real-time fermentation data to obtain coding feature data, wherein the coding feature data comprises a time sequence feature vector, an image feature matrix and a text semantic vector; calculating the characteristic association weight of the coded characteristic data through an attention mechanism, and outputting a multi-mode characteristic tensor according to the characteristic association weight; reasoning the multi-mode characteristic tensor by combining a tree structure algorithm with knowledge rules of the fermentation field to obtain an intermediate reasoning result, wherein the intermediate reasoning result comprises the current fermentation quality state and the parameter optimization direction; and converting the intermediate reasoning result into fermentation quality score and parameter adjustment advice. Further, the multi-modal large model includes Qwen models and Vision Transformer models; The visual code is based on Vision Transformer model, the text code is based on Qwen model; The tree structure algorithm is a decision tree algorithm, and the knowledge rule in the fermentation field is a knowledge graph in the fermentation field. Further, the multi-modal large model includes LLaMA model and Swin transducer model; The visual code is based on a Swin transform model, and the text code is based on a LLaMA model; the attention mechanism is a cross attention mechanism, and the feature association weight is fine granularity association among different modal features; the tree structure algorithm is a random forest algorithm, and the fermentation field knowledge rule is a fermentation field rule base. Further, before outputting the fermentation quality score, the method further comprises: and comparing whether the text semantic vector extracted by the LLaMA model is matched with the image feature matrix extracted by the Swin transducer model. Further, the train