AI prediction guided gamma-pga fermentation medium intelligent formula system and method

Abstract

The application discloses an AI prediction guided gamma-PGA fermentation medium intelligent formula system and method, which comprises the following steps: a data acquisition module is used for collecting various data related to gamma-PGA fermentation, including but not limited to historical fermentation medium formula data, fermentation product yield data under different formula, and fermentation process environmental parameter data; a data preprocessing module is used for cleaning and normalizing the collected data to eliminate noise and abnormal values in the data, and unifying the data to a specific numerical range; and a feature extraction module is used for extracting key features related to the gamma-PGA fermentation medium formula from the preprocessed data. The application uses AI technology and optimization algorithm to search for the optimal formula combination in the solution space, improves the fermentation yield and quality stability of gamma-PGA, can quickly and accurately generate a fermentation medium intelligent formula meeting the target yield, greatly shortens the formula design time, and improves the research and development efficiency.

Description

Technical Field

[0001] This invention relates to the field of bio-fermentation technology, and in particular to an AI-predictive guided intelligent formulation system and method for γ-PGA fermentation medium. Background Technology

[0002] γ-Polyglutamic acid (γ-PGA) is a water-soluble polyamino acid synthesized by microorganisms. It possesses excellent biocompatibility, biodegradability, and moisturizing properties, and has broad application prospects in various fields such as food, medicine, cosmetics, and agriculture. In the fermentation production of γ-PGA, the formulation of the fermentation medium is one of the key factors affecting the yield and quality of γ-PGA.

[0003] Traditional fermentation medium formulation design methods rely heavily on experience and extensive experimental screening, which has significant drawbacks: the formulation design cycle is as long as 30-60 days, resulting in extremely low R&D efficiency; and it can only screen a limited number of formulation combinations, making it difficult to find the globally optimal solution, leading to fluctuations in γ-PGA fermentation yield of ±15% and persistently high production costs. With the development of artificial intelligence technology, using AI to guide fermentation medium formulation design has become a new trend, but existing related technologies still have many shortcomings: they do not consider the availability of raw materials, and the generated formulations often cannot be implemented due to raw material shortages; and they do not introduce cost evaluation indicators, resulting in poor formulation economics and failing to meet the comprehensive requirements of actual production for 'accurate prediction, economic feasibility, and easy availability of raw materials'. Summary of the Invention

[0004] This invention provides an AI-predictive guided intelligent formulation system and method for γ-PGA fermentation medium to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: AI-predictive guided intelligent formulation system for γ-PGA fermentation media includes: The data acquisition module is used to collect various data related to γ-PGA fermentation, including historical fermentation medium formulation data, fermentation product yield data under different formulations, and fermentation process environmental parameter data. This module is connected to the NCBI fermentation database and the CNKI academic database to obtain publicly available γ-PGA fermentation-related research data. The historical fermentation medium formulation data includes the types and contents of carbon sources, nitrogen sources, and inorganic salts. The fermentation process environmental parameter data includes temperature 25-37℃, pH 5.5-7.5, dissolved oxygen 3-8mg / L, and stirring speed 100-500r / min. The data preprocessing module cleans and removes outliers that deviate from the mean by more than three standard deviations, and normalizes the data to a range of [0,1] to eliminate noise and outliers. The feature extraction module extracts key features related to the γ-PGA fermentation medium formulation from the preprocessed data. These key features include the types and contents of carbon sources (glucose and sucrose), the types and contents of nitrogen sources (peptone and yeast extract), and the types and contents of inorganic salts (potassium dihydrogen phosphate and magnesium sulfate). Principal component analysis is used for feature extraction, retaining principal components with a cumulative contribution rate of ≥90%. The AI ​​prediction model construction module, based on extracted feature data, uses a deep neural network algorithm to construct an AI prediction model for predicting the yield of γ-PGA fermentation products. The deep neural network includes an input layer, 3-5 hidden layers, and an output layer. The number of neurons in the input layer equals the key feature dimension, with 32-128 neurons per layer. The output layer has one neuron, corresponding to the predicted yield value. The activation function is the ReLU function, and the loss function is the mean squared error function. The intelligent formula generation module, based on the set target γ-PGA yield (5-30 g / L), combines the AI ​​prediction model with a genetic algorithm to generate an intelligent formula for the fermentation medium that meets the target yield. The genetic algorithm has a population size of 50-100, 50-100 generations, a crossover probability of 0.6-0.8, and a mutation probability of 0.01-0.05. The formula evaluation and optimization module further evaluates the generated intelligent formula and optimizes and adjusts the formula based on the evaluation results. The evaluation indicators include cost indicators and raw material availability indicators, with the cost indicator having a weight of 0.4 and the raw material availability indicator having a weight of 0.6. The formula with a comprehensive evaluation score of ≥80 is the optimal formula.

[0006] As a further improvement to this technical solution: the machine learning algorithm used in the AI ​​prediction model construction module is a deep neural network algorithm. The deep neural network includes an input layer, 3-5 hidden layers, and an output layer. The number of neurons in the input layer is equal to the key feature dimension, i.e., the number of principal components extracted by PCA. The number of neurons in each hidden layer is 32-128, and the output layer has 1 neuron, corresponding to the predicted value of γ-PGA fermentation yield. The activation function is the ReLU function, the loss function is the mean squared error function, the optimizer is the Adam optimizer, and the learning rate is 0.001-0.01. After training with 5-fold cross-validation, the model prediction accuracy is ≥92%. This algorithm performs complex nonlinear mapping on the input feature data through a multi-layer neuron structure to accurately predict the yield of γ-PGA fermentation products.

[0007] As a further improvement to this technical solution: the intelligent formula generation module uses a genetic algorithm for optimization. The parameters of the genetic algorithm are set as follows: population size 50-100, number of iterations 50-100 generations, crossover probability 0.6-0.8, mutation probability 0.01-0.05, and fitness function is the reciprocal of the absolute value of the difference between the predicted γ-PGA yield and the target yield. By simulating the natural selection and genetic mechanism in the process of biological evolution, the optimal fermentation medium formula that satisfies the target yield is searched in the solution space, and the formula generation time is ≤2 hours.

[0008] As a further improvement to this technical solution: the cost index is calculated using the following formula: ; Where C represents the total cost of the fermentation medium. This represents the unit price of the i-th raw material. This indicates the amount of the i-th ingredient used in the formula. This indicates the total number of types of ingredients in the formula.

[0009] As a further improvement to this technical solution: the raw material availability index is calculated using the following formula: ; Where A represents the raw material availability index value, This represents the current inventory level of the i-th raw material. This represents the average demand for the i-th raw material over a certain period of time. This indicates the total number of types of ingredients in the formula.

[0010] As a further improvement to this technical solution: the data acquisition module is also connected to an external database to obtain publicly available research data related to γ-PGA fermentation, so as to enrich the data sources and improve the accuracy and generalization ability of the model.

[0011] As a further improvement to this technical solution, the system also includes a visualization module, which is used to visualize the generated intelligent formula, evaluation results, and fermentation process-related data, making it convenient for users to understand and analyze them intuitively.

[0012] A smart formulation method for γ-PGA fermentation medium includes the following steps: S1 collects various data related to γ-PGA fermentation, including historical fermentation medium formulation data, fermentation product yield data under different formulations, and fermentation process environmental parameter data. This module is connected to the NCBI fermentation database and the CNKI academic database to obtain publicly available γ-PGA fermentation-related research data, with a data sample size of ≥1000 groups. Historical fermentation medium formulation data includes the types and contents of carbon sources, nitrogen sources, and inorganic salts. Fermentation process environmental parameter data includes temperature 25-37℃, pH 5.5-7.5, dissolved oxygen 3-8mg / L, and stirring speed 100-500r / min. S2, clean the collected data, delete outliers that deviate from the mean by 3 times the standard deviation, normalize the data, and use the min-max standardization method to unify the data to the [0,1] range, thereby eliminating noise and outliers in the data; S3. Key features related to the γ-PGA fermentation medium formulation were extracted from the preprocessed data, and principal component analysis was used to extract principal components with a cumulative contribution rate of ≥90% as model input features. S4. Based on the extracted feature data, an AI prediction model for predicting the yield of γ-PGA fermentation products was constructed using a deep neural network algorithm. The model was trained and validated using a 5-fold cross-validation method, and the model prediction accuracy was ≥92%. S5. Based on the target yield of 5-30 g / L of γ-PGA, combined with the AI ​​prediction model, a genetic algorithm is used to generate an intelligent formula for fermentation medium that meets the target yield. The population size is 50-100, the number of iterations is 50-100 generations, the crossover probability is 0.6-0.8, and the mutation probability is 0.01-0.05. S6, Evaluate the generated smart formula. Evaluation indicators include cost indicators, according to the formula. Calculations and raw material availability indicators are based on the formula. Calculation, where The optimal formula is the one with a comprehensive evaluation score of ≥80 points, which is the average demand over 30 days. If the score is lower than 80 points, the types and contents of raw materials are adjusted and the formula is regenerated until the requirements are met, thus obtaining the final intelligent formula for the fermentation medium.

[0013] As a further improvement to this technical solution: when building the AI ​​prediction model, a 5-fold cross-validation method is used to train and validate the model. That is, the dataset is randomly divided into 5 equal parts, and 4 parts are used as the training set and 1 part is used as the validation set. This is repeated 5 times, and the average accuracy of the 5 validations is taken as the final performance index of the model, so as to improve the stability and accuracy of the model. The model prediction accuracy is ≥92%.

[0014] As a further improvement to this technical solution: after generating the intelligent formula, the formula information is stored in a database for subsequent querying and analysis, and the formula data in the database is updated and optimized according to the actual fermentation production situation.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention, through a weighted evaluation of cost and raw material availability (cost weight 0.4, availability weight 0.6), generates a fermentation medium formula that, compared to existing AI formula systems, reduces raw material procurement costs by 10%-15% and increases raw material availability to over 95%, completely solving the problems of poor economic efficiency and difficulty in obtaining raw materials in existing technologies, and significantly reducing fermentation production costs. 2. This invention adopts a combination of 'deep neural network + genetic algorithm', which shortens the formula design time from 30-60 days to 2-3 days compared with traditional experience screening methods, and improves the R&D efficiency by more than 90%. Compared with existing shallow AI models, the fermentation yield of γ-PGA is increased by 15%-20%, and the yield fluctuation range is reduced to within ±3%, which greatly improves the fermentation yield and quality stability. 3. The data acquisition module of this invention integrates multi-source data, with a model training sample size of ≥1000 groups. Combined with 5-fold cross-validation, the model prediction accuracy is ≥92%, which is significantly improved compared to existing AI formulation systems (prediction accuracy of 80%-85%), ensuring that the generated formulation can stably achieve the target yield. 4. The formulation evaluation and optimization module of this invention forms a closed-loop adjustment mechanism. The success rate of the optimal formulation with a comprehensive evaluation score of ≥80 points is ≥98%, which greatly improves the practical application value of the formulation compared to existing technologies (success rate of 70%-80%).

[0016] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the steps of the AI ​​prediction-guided intelligent formulation method for γ-PGA fermentation medium proposed in this invention; Figure 2 This is a schematic diagram of the AI ​​prediction-guided intelligent formulation system for γ-PGA fermentation medium proposed in this invention. Detailed Implementation

[0018] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are for illustrative purposes only and are not intended to limit the scope of the invention. The invention is described more specifically in the following paragraphs by way of example with reference to the accompanying drawings. It should be noted that the drawings are in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the present invention.

[0019] It should be noted that when a component is described as "fixed to" another component, it can be directly on the other component or may have a component in between. When a component is considered "connected to" another component, it can be directly connected to the other component or may have a component in between. When a component is considered "set on" another component, it can be directly set on the other component or may have a component in between. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0020] Please see Figures 1-2 In this embodiment of the invention, the AI ​​prediction-guided intelligent formulation system for γ-PGA fermentation medium mainly includes a data acquisition module, a data preprocessing module, a feature extraction module, an AI prediction model construction module, an intelligent formulation generation module, and a formulation evaluation and optimization module.

[0021] The data acquisition module is responsible for collecting various data related to γ-PGA fermentation, including historical fermentation medium formulation data, fermentation product yield data under different formulations, and environmental parameter data of the fermentation process. Simultaneously, this module connects to external databases to obtain publicly available research data related to γ-PGA fermentation, enriching data sources and improving the model's accuracy and generalization ability.

[0022] Data preprocessing module: Cleans the collected data, removes noise and outliers, and then performs normalization to unify the data to a specific numerical range, such as [0,1], for subsequent feature extraction and model building.

[0023] Feature extraction module: Extracts key features related to the γ-PGA fermentation medium formulation from the preprocessed data. These features include the types and contents of carbon sources, nitrogen sources, and inorganic salts. For example, carbon source features can be extracted as glucose content and sucrose content; nitrogen source features can be extracted as peptone content and yeast extract content.

[0024] AI Prediction Model Construction Module: Based on extracted feature data, this module utilizes machine learning algorithms to construct an AI prediction model for predicting the yield of γ-PGA fermentation products. This invention employs a deep neural network algorithm, which uses a multi-layered neuron structure to perform complex nonlinear mapping on the input feature data, enabling accurate prediction of γ-PGA fermentation product yield. During model construction, cross-validation is used for training and validation to improve the model's stability and accuracy.

[0025] Intelligent formulation generation module: Based on the set target yield of γ-PGA, and combined with an AI prediction model, an optimization algorithm is used to generate an intelligent formulation of fermentation medium that meets the target yield. This invention employs a genetic algorithm, which searches the solution space for the optimal fermentation medium formulation that meets the target yield by simulating the natural selection and genetic mechanisms in the process of biological evolution.

[0026] Formula Evaluation and Optimization Module: Further evaluates the generated smart formulas, with evaluation indicators including cost indicators and raw material availability indicators.

[0027] Cost metrics: Calculated using the following formula:

[0028] Where C represents the total cost of the fermentation medium. This represents the unit price of the i-th raw material. This indicates the amount of the i-th ingredient used in the formula. This indicates the total number of types of ingredients in the formula.

[0029] Specifically, in a fermentation medium formulation, the unit price of glucose is 5 yuan / kg, and the usage is 10 kg; the unit price of peptone is 20 yuan / kg, and the usage is 2 kg; the unit price of the inorganic salt mixture is 10 yuan / kg, and the usage is 1 kg. Therefore, the total cost of this formulation is C = 5 × 10 + 20 × 2 + 10 × 1 = 50 + 40 + 10 = 100 yuan.

[0030] Raw material availability index: calculated using the following formula:

[0031] Where A represents the raw material availability index value, This represents the current inventory level of the i-th raw material. This represents the average demand for the i-th raw material over a certain period of time. This indicates the total number of types of ingredients in the formula.

[0032] Specifically, in a formula, there are three raw materials: glucose, with a current inventory of 100 kg and an average daily demand of 20 kg; peptone, with a current inventory of 50 kg and an average daily demand of 10 kg; and an inorganic salt mixture, with a current inventory of 30 kg and an average daily demand of 5 kg. What is the raw material availability index value? .

[0033] Based on the evaluation results, the formula can be optimized and adjusted. For example, if the cost is too high, you can try to find cheaper alternative raw materials; if the raw material availability index is low, you can adjust the amount of raw materials used or find other raw materials with better availability.

[0034] Visualization Module: The system also includes a visualization module, which is used to visualize the generated smart recipes, evaluation results, and fermentation process-related data, making it easier for users to understand and analyze them intuitively.

[0035] The AI ​​prediction-guided intelligent formulation method for γ-PGA fermentation medium of the present invention includes the following steps: S1 collects various data related to γ-PGA fermentation, including historical fermentation medium formulation data, fermentation product yield data under different formulations, fermentation process environmental parameter data, etc., and obtains publicly available research data related to γ-PGA fermentation.

[0036] S2 cleans and normalizes the collected data, eliminating noise and outliers, and unifying the data to a specific numerical range.

[0037] S3, extracting key features related to the γ-PGA fermentation medium formulation from the pretreated data.

[0038] S4. Based on the extracted feature data, an AI prediction model for predicting the yield of γ-PGA fermentation products is constructed using machine learning algorithms, and the model is trained and validated using cross-validation.

[0039] S5, based on the set target yield of γ-PGA, combines an AI prediction model and uses an optimization algorithm to generate an intelligent formula for the fermentation medium that meets the target yield.

[0040] S6 evaluates the generated smart formula and optimizes and adjusts the formula based on the evaluation results to obtain the final smart formula for the fermentation medium.

[0041] S7 stores the formula information in the database for subsequent querying and analysis, and updates and optimizes the formula data in the database based on the actual fermentation production situation.

[0042] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Those skilled in the art can readily implement the present invention based on the accompanying drawings and the above description. However, any modifications, alterations, or variations made by those skilled in the art without departing from the scope of the present invention, utilizing the disclosed technical content, are equivalent embodiments of the present invention. Furthermore, any modifications, alterations, or variations made to the above embodiments based on the essential technology of the present invention are still within the protection scope of the present invention.

Claims

1. An AI-predictive-guided intelligent formulation system for γ-PGA fermentation medium, characterized in that, include: The data acquisition module is used to collect various data related to γ-PGA fermentation, including historical fermentation medium formulation data, fermentation product yield data under different formulations, and fermentation process environmental parameter data. This module is connected to the NCBI fermentation database and the CNKI academic database to obtain publicly available γ-PGA fermentation-related research data. The historical fermentation medium formulation data includes the types and contents of carbon sources, nitrogen sources, and inorganic salts. The fermentation process environmental parameter data includes temperature 25-37℃, pH 5.5-7.5, dissolved oxygen 3-8mg / L, and stirring speed 100-500r / min. The data preprocessing module cleans the collected data, removes outliers that deviate from the mean by 3 times the standard deviation, and normalizes the data to a range of [0,1] to eliminate noise and outliers in the data. The feature extraction module extracts key features related to the γ-PGA fermentation medium formulation from the preprocessed data. These key features include the types and contents of carbon sources (i.e., the proportion of glucose and sucrose), the types and contents of nitrogen sources (i.e., the proportion of peptone and yeast extract), and the types and contents of inorganic salts (i.e., the proportion of potassium dihydrogen phosphate and magnesium sulfate). The feature extraction uses principal component analysis to retain principal components with a cumulative contribution rate of ≥90%. The AI ​​prediction model construction module, based on extracted feature data, uses a deep neural network algorithm to construct an AI prediction model for predicting the yield of γ-PGA fermentation products. The deep neural network includes an input layer, 3-5 hidden layers, and an output layer. The number of neurons in the input layer equals the key feature dimension, with 32-128 neurons per layer. The output layer has one neuron, corresponding to the predicted yield value. The activation function is the ReLU function, and the loss function is the mean squared error function. The intelligent formula generation module, based on the set target γ-PGA yield (5-30 g / L), combines the AI ​​prediction model with a genetic algorithm to generate an intelligent formula for the fermentation medium that meets the target yield. The genetic algorithm has a population size of 50-100, 50-100 generations, a crossover probability of 0.6-0.8, and a mutation probability of 0.01-0.

05. The formula evaluation and optimization module further evaluates the generated intelligent formula and optimizes and adjusts the formula based on the evaluation results. The evaluation indicators include cost indicators and raw material availability indicators, with the cost indicator having a weight of 0.4 and the raw material availability indicator having a weight of 0.

6. The formula with a comprehensive evaluation score of ≥80 is the optimal formula.

2. The AI-predictive-guided intelligent formulation system for γ-PGA fermentation medium according to claim 1, characterized in that, The AI ​​prediction model construction module uses a deep neural network algorithm, which includes an input layer, 3-5 hidden layers, and an output layer. The number of neurons in the input layer is equal to the key feature dimension, i.e., the number of principal components extracted by PCA. Each hidden layer has 32-128 neurons, and the output layer has 1 neuron, corresponding to the predicted value of γ-PGA fermentation yield. The activation function is the ReLU function, the loss function is the mean squared error function, the optimizer is the Adam optimizer, and the learning rate is 0.001-0.

01. After training with 5-fold cross-validation, the model prediction accuracy is ≥92%. This algorithm uses a multi-layer neuron structure to perform complex nonlinear mapping on the input feature data to accurately predict the yield of γ-PGA fermentation products.

3. The AI-predictive-guided intelligent formulation system for γ-PGA fermentation medium according to claim 2, characterized in that, The intelligent formula generation module uses a genetic algorithm for optimization. The parameters of the genetic algorithm are set as follows: population size 50-100, number of iterations 50-100 generations, crossover probability 0.6-0.8, mutation probability 0.01-0.05, and fitness function is the reciprocal of the absolute value of the difference between the predicted γ-PGA yield and the target yield. By simulating the natural selection and genetic mechanism in the process of biological evolution, the optimal fermentation medium formula that satisfies the target yield is searched in the solution space. The formula generation time is ≤2 hours.

4. The AI-predictive-guided intelligent formulation system for γ-PGA fermentation medium according to claim 3, characterized in that, The cost indicator is calculated using the following formula: ; Where C represents the total cost of the fermentation medium. This represents the unit price of the i-th raw material. This indicates the amount of the i-th ingredient used in the formula. This indicates the total number of types of ingredients in the formula.

5. The AI-predictive-guided intelligent formulation system for γ-PGA fermentation medium according to claim 4, characterized in that, The raw material availability index is calculated using the following formula: ； Where A represents the raw material availability index value, This represents the current inventory level of the i-th raw material. This represents the average demand for the i-th raw material over a certain period of time. This indicates the total number of types of ingredients in the formula.

6. The AI-predictive-guided intelligent formulation system for γ-PGA fermentation medium according to claim 5, characterized in that, The data acquisition module is also connected to an external database to obtain publicly available research data related to γ-PGA fermentation, thereby enriching the data sources and improving the accuracy and generalization ability of the model.

7. The AI-predictive-guided intelligent formulation system for γ-PGA fermentation medium according to claim 6, characterized in that, The system also includes a visualization module, which is used to visualize the generated intelligent formula, evaluation results, and fermentation process-related data, making it convenient for users to understand and analyze them intuitively.

8. A method for intelligent formulation of γ-PGA fermentation medium, characterized in that, Includes the following steps: S1 collects various data related to γ-PGA fermentation, including historical fermentation medium formulation data, fermentation product yield data under different formulations, and fermentation process environmental parameter data. This module is connected to the NCBI fermentation database and the CNKI academic database to obtain publicly available γ-PGA fermentation-related research data, with a data sample size of ≥1000 groups. Historical fermentation medium formulation data includes the types and contents of carbon sources, nitrogen sources, and inorganic salts. Fermentation process environmental parameter data includes temperature 25-37℃, pH 5.5-7.5, dissolved oxygen 3-8mg / L, and stirring speed 100-500r / min. S2, clean the collected data, delete outliers that deviate from the mean by 3 times the standard deviation, normalize the data, and use the min-max standardization method to unify the data to the [0,1] range, thereby eliminating noise and outliers in the data; S3. Key features related to the γ-PGA fermentation medium formulation were extracted from the preprocessed data, and principal component analysis was used to extract principal components with a cumulative contribution rate of ≥90% as model input features. S4. Based on the extracted feature data, an AI prediction model for predicting the yield of γ-PGA fermentation products was constructed using a deep neural network algorithm. The model was trained and validated using a 5-fold cross-validation method, and the model prediction accuracy was ≥92%. S5. Based on the target yield of 5-30 g / L of γ-PGA, combined with the AI ​​prediction model, a genetic algorithm is used to generate an intelligent formula for fermentation medium that meets the target yield. The population size is 50-100, the number of iterations is 50-100 generations, the crossover probability is 0.6-0.8, and the mutation probability is 0.01-0.

05. S6, Evaluate the generated smart formula. Evaluation indicators include cost indicators, according to the formula. Calculations and raw material availability indicators are based on the formula. Calculation, where The optimal formula is the one with a comprehensive evaluation score of ≥80 points, which is the average demand over 30 days. If the score is lower than 80 points, the types and contents of raw materials are adjusted and the formula is regenerated until the requirements are met, thus obtaining the final intelligent formula for the fermentation medium.

9. The AI-predictive-guided intelligent formulation method for γ-PGA fermentation medium according to claim 8, characterized in that, When building the AI ​​prediction model, a 5-fold cross-validation method is used to train and validate the model. The dataset is randomly divided into 5 equal parts, and 4 parts are used as the training set and 1 part is used as the validation set. This is repeated 5 times, and the average accuracy of the 5 validations is taken as the final performance index of the model to improve the stability and accuracy of the model. The model prediction accuracy is ≥92%.

10. The AI-predictive-guided intelligent formulation method for γ-PGA fermentation medium according to claim 8, characterized in that, After generating the intelligent formula, the formula information is stored in the database for subsequent querying and analysis. At the same time, the formula data in the database is updated and optimized based on the actual fermentation production situation.

Images