Acid production rate regulation system based on real-time monitoring of fermentation activity of specific strains
By combining a specific strain-specific viability detection module with an improved convolutional neural network, the accuracy problem of monitoring the fermentation viability of specific strains in existing technologies has been solved, achieving stable regulation of acid production rate and improving the stability and production efficiency of the fermentation process.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- YANGZHOU YANGDA KANGYUAN DAIRY
- Filing Date
- 2026-03-14
- Publication Date
- 2026-06-09
AI Technical Summary
Existing acid production rate regulation systems based on real-time monitoring of fermentation activity of specific strains cannot accurately identify specific strains, are prone to cross-reaction with other bacteria and impurities in the fermentation system, and lack specificity in the prediction model, leading to fluctuations in acid production rate and fermentation failure.
By employing a specific strain-specific viability detection module, a data processing and analysis module, an acid production rate prediction module, a multi-dimensional collaborative regulation module, and a feedback correction module, combined with improved convolutional neural network and microfluidic chip technology, real-time viability monitoring and accurate prediction and stable regulation of acid production rate of specific strains can be achieved.
This method achieves stability and precision in acid production rate during the fermentation process of specific strains, reduces fermentation failure rate, and improves production efficiency and product quality.
Smart Images
Figure CN122168805A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of bio-fermentation engineering technology, specifically to an acid production rate regulation system based on real-time monitoring of the fermentation activity of specific strains. Background Technology
[0002] In the microbial fermentation acid production industry, the stability of the acid production rate directly determines the quality, yield, and production efficiency of the fermentation products. The core influencing factor of the acid production rate is the fermentation activity of a specific acid-producing strain—strain activity directly reflects its metabolic intensity, proliferation capacity, and acid production potential. Abnormal activity can lead to fluctuations in the acid production rate, prolonged fermentation cycle, increased product impurities, and even fermentation failure. Existing acid production rate control systems based on real-time monitoring of the fermentation activity of specific strains rely on general-purpose probes or detection methods that cannot accurately identify specific strains. These systems are prone to cross-reaction with other bacteria and impurities in the fermentation system. Predictive models often depend on auxiliary parameters such as pH and dissolved oxygen concentration of the fermentation system, failing to fully consider the core influence of strain activity. Furthermore, the model parameters are vague and lack specificity, making it impossible to predict the changing trend of the acid production rate in advance. They often employ single-parameter control modes, with a lack of coordination among the control units, which easily leads to fluctuations in the acid production rate. Summary of the Invention
[0003] To address the shortcomings of existing technologies, this invention provides an acid production rate control system based on real-time monitoring of the fermentation activity of specific strains, which solves the problems of the system's inability to accurately identify specific strains and its susceptibility to cross-reaction with other bacteria and impurities in the fermentation system.
[0004] To achieve the above objectives, the present invention provides the following technical solution: a system for regulating the acid production rate based on real-time monitoring of the fermentation activity of specific strains. This system is applicable to specific strains with a single or identical specific molecular markers and is used to regulate the acid production fermentation process of specific strains in a fermenter. It includes a specific strain-specific activity detection module formed by electrical connections to create a closed loop for signal transmission and execution, a data processing and analysis module, an acid production rate prediction module, a multi-dimensional collaborative regulation module, a feedback correction module, and an emergency module, wherein:
[0005] A specific strain-specific activity detection module is used to detect the fermentation activity of a specific strain in real time. The reaction system inside the fermenter where the specific strain performs acid-producing fermentation is the fermentation system.
[0006] The data processing and analysis module performs noise reduction, filtering, and standardization on the fermentation activity signal detected by the dedicated activity detection module. It then calculates the real-time activity index of a specific strain and constructs a three-dimensional correlation model that links the activity, acid production rate, and auxiliary parameters of the specific strain.
[0007] The acid production rate prediction module predicts the trend of acid production rate changes and outputs the prediction results based on the real-time vigor index of a specific strain, a three-dimensional correlation model, and auxiliary parameters of the fermentation system.
[0008] The multi-dimensional collaborative regulation module, based on a three-dimensional correlation model and prediction results, adopts a dynamic decoupling-feedforward collaborative regulation strategy to regulate the fermentation system and ensure that the acid production rate is stable within the preset target range.
[0009] The feedback correction module, based on the deviation between the actual acid production rate and the predicted acid production rate and the target acid production rate in the fermentation system, connects the output of the feedback correction module to the input of the acid production rate prediction module to correct the acid production rate prediction module and the control parameters.
[0010] The emergency module is used to monitor the rate and magnitude of viability decay of specific strains and trigger emergency strategies. The input of the emergency module is connected to the output of the data processing and analysis module.
[0011] Preferably, the specific strain-specific viability detection module includes a specific identification unit, an in-situ detection unit, a signal amplification unit, and a data output unit. The specific identification unit includes a specific probe that binds only to specific molecular markers of a specific strain, and the surface of the specific probe is modified with anti-degradation groups.
[0012] Preferably, the specific molecular markers of the specific strain are key acid-producing enzymes, specific metabolic intermediates, or cell surface antigens unique to the strain, and the dedicated probe is a fluorescent aptamer probe or an enzyme-linked probe that binds only to the specific molecular markers of the specific strain, does not cross-react with other bacteria or impurities in the fermentation system, and can withstand the interference of acid and alkali metabolites in the fermentation system. The in-situ detection unit uses a microfluidic chip combined with laser-induced fluorescence detection technology.
[0013] Preferably, the three-dimensional correlation model is used to determine the optimal vigor range of a specific strain, and the three-dimensional correlation model introduces a vigor index weighting factor, which can preferentially respond to small changes in the vigor index and realize early prediction and control of acid production rate.
[0014] Preferably, the formula for calculating the real-time viability index of the specific strain is:
[0015] ,
[0016] In the formula: These are the fitting coefficients. As a correction constant, , Regression analysis of gradient experimental data on key acid-producing enzymes and fluorescence intensity of a specific strain revealed that the gradient experiment covered the activity range of the key acid-producing enzymes of this strain. , This refers to the activity of key enzymes in acid production by the strain. The fluorescence intensity is denoted as .
[0017] Preferably, the multi-dimensional collaborative regulation module includes a substrate replenishment unit, a dissolved oxygen regulation unit, a temperature regulation unit, a pH regulation unit, a stirring regulation unit, and a metabolic regulator addition unit that work together to avoid fluctuations in the acid production rate caused by single parameter regulation. Specifically, the multi-dimensional collaborative regulation module calculates the action parameters of each regulation unit in real time through a three-dimensional correlation model, and dynamically allocates the action sequence and quantification range of each regulation unit according to the correlation between the real-time activity index and the acid production rate, so as to ensure that the acid production rate is stable within the preset target range.
[0018] Preferably, the acid production rate prediction module incorporates an improved convolutional neural network model for predicting acid production rate. The improved convolutional neural network adds a temporal attention weight allocation layer on the basis of a conventional convolutional neural network. The temporal attention weight allocation layer adopts a gated recurrent unit, which contains two hidden layers, each with 64-128 neurons and a time step of 10 minutes.
[0019] Preferably, the system further includes a model update module, which is used to optimize the parameters of the acid production rate prediction model based on the actual acid production data of the fermentation batch, and the optimization process combines the deviation between the vitality index and the actual acid production rate to achieve adaptive adjustment of the model.
[0020] Preferably, the emergency module triggers a corresponding emergency strategy based on a preset threshold for monitoring the degree of decline in the viability of a specific strain, wherein the preset threshold is set based on the inflection point of the decline in fermentation success rate.
[0021] This invention also discloses a method for regulating acid production rate based on real-time monitoring of the fermentation activity of specific strains, specifically including the following steps:
[0022] S1. Identify specific acid-producing strains, clarify their specific molecular markers and metabolic characteristics, and modify the exclusive probes to resist degradation.
[0023] S2. Obtain historical fermentation data for this specific strain, and based on ≥3 batches of fermentation experimental data, construct a three-dimensional correlation model of "vibration-acid production rate-auxiliary parameters", introduce a vibration index weighting factor, and determine the optimal vibration range and emergency threshold.
[0024] S3. Connect the seed culture of the specific strain, start the system, and the specific strain-specific viability detection module detects the strain viability in real time through the modified special probe;
[0025] S4, the data processing and analysis module performs noise reduction, filtering, standardization and correlation analysis on vitality data and fermentation auxiliary parameters, and calculates the real-time vitality index by fitting formula to determine the vitality status.
[0026] S5. The acid production rate prediction module inputs real-time data and uses an improved convolutional neural network model to predict the future acid production rate trend and prediction value. If the prediction value deviates from the target range, it sends a control command to the multi-dimensional collaborative control module.
[0027] S6. The multi-dimensional collaborative regulation module executes quantitative collaborative regulation if the activity decay slope is lower than the emergency threshold according to the regulation command; otherwise, it jumps to S7. The quantitative collaborative regulation operation is as follows: adopting a dynamic decoupling-feedforward collaborative regulation strategy to quantitatively regulate the parameters of each unit and maintain the acid production rate stability.
[0028] S7. The feedback correction module compares the deviation between the actual acid production rate and the predicted acid production rate and the target acid production rate in real time and corrects them in a timely manner; the emergency module monitors the degree of strain activity decay in real time and performs emergency regulation in a timely manner.
[0029] S8. Fermentation ends, the system is shut down, and all data is recorded. The model update module optimizes the parameters of the acid production rate prediction model based on the fermentation data and the deviation between the vitality index and the actual acid production rate.
[0030] Beneficial effects
[0031] This invention provides a system for regulating acid production rate based on real-time monitoring of the fermentation activity of specific bacterial strains. Compared with existing technologies, it has the following advantages:
[0032] (1) The acid production rate regulation system based on real-time monitoring of fermentation activity of specific strains, by setting up a specific activity detection module for specific strains in the system, using a specific probe combined with specific molecular markers of specific strains, the probe is modified to resist degradation, so it can avoid being degraded by the metabolites of miscellaneous bacteria and does not cross-react with miscellaneous bacteria and impurities; combined with microfluidic chip online ultraviolet sterilization and laser-induced fluorescence detection technology, aseptic in-situ detection is realized, ensuring the accuracy and real-time nature of the detection data, and solving the pain points of poor detection specificity and signal distortion in existing detection systems.
[0033] (2) The acid production rate regulation system based on real-time monitoring of fermentation activity of specific strains, by setting an acid production rate prediction module in the system, adopts an improved convolutional neural network with a time-series attention weight allocation layer to accurately predict the trend of acid production rate change and achieve early regulation. It adopts a multi-dimensional collaborative regulation strategy to avoid fluctuations caused by single parameter regulation. The emergency threshold adopts a quantitative linear decay slope to accurately capture the activity decay trend and improve fermentation stability.
[0034] (3) The acid production rate regulation system based on real-time monitoring of fermentation activity of specific strains can adapt to the fluctuations of different fermentation batches by setting a model update module in the system, which can be applied to multiple batches of fermentation of the same strain, thereby reducing production costs. Attached Figure Description
[0035] Figure 1 This is a block diagram illustrating the main principle of the present invention;
[0036] Figure 2 This is a schematic diagram of the specific strain-specific viability detection module of the present invention;
[0037] Figure 3 This is a schematic diagram of the principle of the multi-dimensional collaborative control module of the present invention;
[0038] Figure 4 This is the main flowchart within the present invention. Detailed Implementation
[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0040] Reference 1- Figure 4 The present invention provides the following three technical solutions:
[0041] The first implementation method is an acid production rate regulation system based on real-time monitoring of the fermentation activity of specific strains. This system is applicable to specific strains with a single or identical specific molecular markers and is used to regulate the acid production fermentation process of these strains in a fermenter. It includes a strain-specific activity detection module with an electrical connection forming a closed loop for signal transmission and execution, a data processing and analysis module, an acid production rate prediction module, a multi-dimensional collaborative regulation module, a feedback correction module, and an emergency module, wherein:
[0042] A specific strain-specific activity detection module is used to detect the fermentation activity of specific strains in real time. The reaction system in the fermenter where the specific strain performs acid-producing fermentation is called the fermentation system. The specific strain-specific activity detection module includes a specific identification unit, an in-situ detection unit, a signal amplification unit, and a data output unit. The specific identification unit contains a specific probe that binds only to the specific molecular markers of the specific strain. The surface of the specific probe is modified with anti-degradation groups. This anti-degradation modification treatment can prevent degradation by the metabolic products of other bacteria in the fermentation system, thus achieving specific detection.
[0043] Specific molecular markers for specific strains are key acid-producing enzymes, specific metabolic intermediates, or cell surface antigens unique to the strain. Dedicated probes are fluorescent aptamer probes or enzyme-linked probes that bind only to the specific molecular markers of that particular strain, do not cross-react with other bacteria or impurities in the fermentation system, and have anti-degradation groups that are molecular weight specific to the strain. The polyethylene glycol groups, and the modification density It can withstand the interference of acid and alkali metabolites in the fermentation system. The in-situ detection unit uses a microfluidic chip combined with laser-induced fluorescence detection technology to realize aseptic continuous sampling and in-situ detection of the fermentation system. The microfluidic chip integrates an online ultraviolet sterilization unit, which sterilizes for 30 seconds per cycle to ensure the accuracy and real-time nature of the detection data.
[0044] The formula for calculating the real-time viability index of a specific strain is:
[0045] ,
[0046] In the formula: These are the fitting coefficients. As a correction constant, , Regression analysis of gradient experimental data on key acid-producing enzymes and fluorescence intensity of a specific strain revealed that the gradient experiment covered the activity range of the key acid-producing enzymes of this strain. , This refers to the activity of key enzymes in acid production by the strain. Fluorescence intensity;
[0047] The data processing and analysis module performs noise reduction, filtering, and standardization on the fermentation activity signal detected by the dedicated activity detection module. It then calculates the real-time activity index of a specific strain and constructs a three-dimensional correlation model linking the activity, acid production rate, and auxiliary parameters of the specific strain based on the fermentation system's pH, dissolved oxygen concentration, fermentation temperature, substrate concentration, and stirring rate. This three-dimensional correlation model is used to determine the optimal activity range for a specific strain, and by introducing a activity index weighting factor, it can preferentially respond to small changes in the activity index, enabling early prediction and control of the acid production rate.
[0048] The acid production rate prediction module predicts the trend of acid production rate changes and outputs the prediction results based on the real-time vigor index of specific strains, a three-dimensional correlation model, and auxiliary parameters of the fermentation system. The multi-dimensional synergistic regulation module, based on the three-dimensional correlation model and the prediction results, adopts a dynamic decoupling-feedforward synergistic regulation strategy to regulate the fermentation system and ensure that the acid production rate is stable within the preset target range. The multi-dimensional synergistic regulation module includes substrate supplementation unit, dissolved oxygen regulation unit, temperature regulation unit, pH regulation unit, stirring regulation unit, and metabolic regulator addition unit that work together to avoid fluctuations in the acid production rate caused by single parameter regulation. Specifically, the multi-dimensional synergistic regulation module calculates the action parameters of each regulation unit in real time through the three-dimensional correlation model, and dynamically allocates the action sequence and quantification range of each regulation unit according to the correlation between the real-time vigor index and the acid production rate to ensure that the acid production rate is stable within the preset target range.
[0049] The feedback correction module, based on the deviation between the actual acid production rate and the predicted and target acid production rates in the fermentation system, connects its output to the input of the acid production rate prediction module to correct the prediction module and its control parameters. The emergency module monitors the rate and magnitude of viability decay of specific strains and triggers emergency strategies. Its input is connected to the output of the data processing and analysis module. The emergency module triggers corresponding emergency strategies based on a preset threshold for the degree of viability decay of specific strains. This preset threshold is set based on the inflection point of the fermentation success rate decline and is defined as the linear decay slope of the viability index within 10 minutes. Early warning of declining vitality can prevent fermentation failure.
[0050] The main difference between the second implementation method and the first implementation method is that:
[0051] The acid production rate prediction module incorporates an improved convolutional neural network (CNN) model. This improved CNN, building upon conventional CNNs, adds a temporal attention weight allocation layer. This layer employs a gated recurrent unit (GRU) to assign different weights to the strain vigor temporal data over the past 10 minutes, reducing interference from irrelevant data. The GRU contains two hidden layers, each with 64-128 neurons and a time step of 10 minutes. This improved CNN with the temporal attention weight allocation layer accurately predicts the trend of acid production rate changes, enabling early intervention and addressing the shortcomings of existing predictions, such as lag and insufficient accuracy. A multi-dimensional collaborative control strategy is employed, with six control units working collaboratively to dynamically allocate the timing and quantization amplitude of action, avoiding fluctuations caused by single-parameter control. The emergency threshold uses a quantized linear decay slope to accurately capture the vigor decay trend, improving fermentation stability and reducing the fermentation failure rate.
[0052] The main difference between the third and second implementation methods is that:
[0053] The system also includes a model update module, which optimizes the parameters of the acid production rate prediction model based on the actual acid production data of the fermentation batches. The optimization process incorporates the deviation between the activity index and the actual acid production rate to achieve adaptive adjustment of the model. Through adaptive optimization, it is applicable to multiple batches of fermentation of the same strain and can be widely used in acid production fermentation of specific strains in the fields of food, medicine, and biochemicals to improve production efficiency and product quality and reduce production costs.
[0054] This invention also discloses a method for regulating acid production rate based on real-time monitoring of the fermentation activity of specific strains, specifically including the following steps:
[0055] S1. Identify specific acid-producing strains, clarify their specific molecular markers and metabolic characteristics, and modify the exclusive probes to resist degradation.
[0056] S2. Obtain historical fermentation data for this specific strain, and based on ≥3 batches of fermentation experimental data, construct a three-dimensional correlation model of "vibration-acid production rate-auxiliary parameters", introduce a vibration index weighting factor, and determine the optimal vibration range and emergency threshold.
[0057] S3. Connect the seed culture of the specific strain, start the system, and the specific strain-specific viability detection module detects the strain viability in real time through the modified special probe;
[0058] S4, the data processing and analysis module performs noise reduction, filtering, standardization and correlation analysis on vitality data and fermentation auxiliary parameters, and calculates the real-time vitality index by fitting formula to determine the vitality status.
[0059] S5. The acid production rate prediction module inputs real-time data and uses an improved convolutional neural network model to predict the future acid production rate trend and prediction value. If the prediction value deviates from the target range, it sends a control command to the multi-dimensional collaborative control module.
[0060] S6. The multi-dimensional collaborative regulation module executes quantitative collaborative regulation if the activity decay slope is lower than the emergency threshold according to the regulation command; otherwise, it jumps to S7. The quantitative collaborative regulation operation is as follows: adopting a dynamic decoupling-feedforward collaborative regulation strategy to quantitatively regulate the parameters of each unit and maintain the acid production rate stability.
[0061] S7. The feedback correction module compares the deviation between the actual acid production rate and the predicted acid production rate and the target acid production rate in real time and corrects them in a timely manner; the emergency module monitors the degree of strain activity decay in real time and performs emergency regulation in a timely manner.
[0062] S8. Fermentation complete. Shut down the system, record all data. The model update module optimizes the parameters of the acid production rate prediction model based on the fermentation data and the deviation between the activity index and the actual acid production rate.
[0063] Furthermore, all content not described in detail in this specification is existing technology known to those skilled in the art, and the model parameters of each electrical appliance are not specifically limited; conventional equipment can be used.
[0064] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0065] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A system for regulating acid production rate based on real-time monitoring of fermentation activity of specific bacterial strains, characterized in that, The system is applicable to specific strains with single or identical specific molecular markers and is used to regulate the acid-producing fermentation process of specific strains in fermenters. It includes a strain-specific viability detection module formed by electrical connections to create a closed loop for signal transmission and execution, a data processing and analysis module, an acid production rate prediction module, a multi-dimensional collaborative regulation module, a feedback correction module, and an emergency module, wherein: A specific strain-specific activity detection module is used to detect the fermentation activity of a specific strain in real time. The reaction system inside the fermenter where the specific strain performs acid-producing fermentation is the fermentation system. The data processing and analysis module performs noise reduction, filtering, and standardization on the fermentation activity signal detected by the dedicated activity detection module. It then calculates the real-time activity index of a specific strain and constructs a three-dimensional correlation model that links the activity, acid production rate, and auxiliary parameters of the specific strain. The acid production rate prediction module predicts the trend of acid production rate changes and outputs the prediction results based on the real-time vigor index of a specific strain, a three-dimensional correlation model, and auxiliary parameters of the fermentation system. The multi-dimensional collaborative regulation module, based on a three-dimensional correlation model and prediction results, adopts a dynamic decoupling-feedforward collaborative regulation strategy to regulate the fermentation system and ensure that the acid production rate is stable within the preset target range. The feedback correction module, based on the deviation between the actual acid production rate and the predicted acid production rate and the target acid production rate in the fermentation system, connects the output of the feedback correction module to the input of the acid production rate prediction module to correct the acid production rate prediction module and the control parameters. The emergency module is used to monitor the rate and magnitude of viability decay of specific strains and trigger emergency strategies. The input of the emergency module is connected to the output of the data processing and analysis module.
2. The acid production rate regulation system based on real-time monitoring of fermentation activity of specific strains according to claim 1, characterized in that: The specific strain-specific viability detection module includes a specific identification unit, an in-situ detection unit, a signal amplification unit, and a data output unit. The specific identification unit contains a specific probe that binds only to specific molecular markers of a specific strain, and the surface of the specific probe is modified with anti-degradation groups.
3. The acid production rate regulation system based on real-time monitoring of fermentation activity of specific strains according to claim 2, characterized in that: The specific molecular markers of the specific strain are key acid-producing enzymes, specific metabolic intermediates, or cell surface antigens unique to the strain. The dedicated probe is a fluorescent aptamer probe or an enzyme-linked probe that binds only to the specific molecular markers of the specific strain and does not cross-react with other bacteria or impurities in the fermentation system. It can tolerate the interference of acid and alkali metabolites in the fermentation system. The in-situ detection unit uses a microfluidic chip combined with laser-induced fluorescence detection technology.
4. The acid production rate regulation system based on real-time monitoring of fermentation activity of specific strains according to claim 1, characterized in that: The three-dimensional correlation model is used to determine the optimal vigor range of a specific strain. Furthermore, the three-dimensional correlation model introduces a vigor index weighting factor, which can preferentially respond to small changes in the vigor index and achieve early prediction and control of acid production rate.
5. The acid production rate regulation system based on real-time monitoring of fermentation activity of specific strains according to claim 1, characterized in that: The formula for calculating the real-time viability index of the specific strain is as follows: , In the formula: These are the fitting coefficients. As a correction constant, , Regression analysis of gradient experimental data on key acid-producing enzymes and fluorescence intensity of a specific strain revealed that the gradient experiment covered the activity range of the key acid-producing enzymes of this strain. , This refers to the activity of key enzymes in acid production by the strain. The fluorescence intensity is denoted as .
6. The acid production rate regulation system based on real-time monitoring of fermentation activity of specific strains according to claim 1, characterized in that: The multi-dimensional collaborative regulation module includes a substrate replenishment unit, a dissolved oxygen regulation unit, a temperature regulation unit, a pH regulation unit, a stirring regulation unit, and a metabolic regulator addition unit that work together to avoid fluctuations in the acid production rate caused by single parameter regulation. Specifically, the multi-dimensional collaborative regulation module calculates the action parameters of each regulation unit in real time through a three-dimensional correlation model, and dynamically allocates the action sequence and quantification range of each regulation unit according to the correlation between the real-time activity index and the acid production rate, so as to ensure that the acid production rate is stable within the preset target range.
7. The acid production rate regulation system based on real-time monitoring of fermentation activity of specific strains according to claim 1, characterized in that: The acid production rate prediction module incorporates an improved convolutional neural network model for predicting acid production rate. The improved convolutional neural network adds a temporal attention weight allocation layer on the basis of a conventional convolutional neural network. The temporal attention weight allocation layer adopts a gated recurrent unit, which contains two hidden layers with 64-128 neurons in each layer and a time step of 10 minutes.
8. The acid production rate regulation system based on real-time monitoring of fermentation activity of specific strains according to claim 6, characterized in that: The system also includes a model update module, which is used to optimize the parameters of the acid production rate prediction model based on the actual acid production data of the fermentation batch. The optimization process incorporates the deviation between the activity index and the actual acid production rate to achieve adaptive adjustment of the model.
9. The acid production rate regulation system based on real-time monitoring of fermentation activity of specific strains according to claim 1, characterized in that: The emergency module triggers a corresponding emergency strategy based on a preset threshold for monitoring the degree of decline in the viability of a specific strain. The preset threshold is set based on the inflection point of the decline in fermentation success rate.
10. A method for regulating acid production rate based on real-time monitoring of fermentation activity of specific bacterial strains, characterized in that: The acid production rate regulation system based on real-time monitoring of fermentation activity of specific strains according to any one of claims 1-9 specifically includes the following steps: S1. Identify specific acid-producing strains, clarify their specific molecular markers and metabolic characteristics, and modify the exclusive probes to resist degradation. S2. Obtain historical fermentation data for this specific strain, and based on ≥3 batches of fermentation experimental data, construct a three-dimensional correlation model of "vibration-acid production rate-auxiliary parameters", introduce a vibration index weighting factor, and determine the optimal vibration range and emergency threshold. S3. Connect the seed culture of the specific strain, start the system, and the specific strain-specific viability detection module detects the strain viability in real time through the modified special probe; S4, the data processing and analysis module performs noise reduction, filtering, standardization and correlation analysis on vitality data and fermentation auxiliary parameters, and calculates the real-time vitality index by fitting formula to determine the vitality status. S5. The acid production rate prediction module inputs real-time data and uses an improved convolutional neural network model to predict the future acid production rate trend and prediction value. If the prediction value deviates from the target range, it sends a control command to the multi-dimensional collaborative control module. S6. The multi-dimensional collaborative regulation module executes quantitative collaborative regulation if the activity decay slope is lower than the emergency threshold according to the regulation command; otherwise, it jumps to S7. The quantitative collaborative regulation operation is as follows: adopting a dynamic decoupling-feedforward collaborative regulation strategy to quantitatively regulate the parameters of each unit and maintain the acid production rate stability. S7. The feedback correction module compares the deviation between the actual acid production rate and the predicted acid production rate and the target acid production rate in real time and corrects them in a timely manner; the emergency module monitors the degree of strain activity decay in real time and performs emergency regulation in a timely manner. S8. Fermentation ends, the system is shut down, and all data is recorded. The model update module optimizes the parameters of the acid production rate prediction model based on the fermentation data and the deviation between the vitality index and the actual acid production rate.