A parameter cooperative control system based on enzymatic hydrolysis reaction of yeast extract

Abstract

This application discloses a parameter-coordinated control system based on the enzymatic hydrolysis reaction of yeast extract, relating to the field of microbial detection and analysis. The method includes: digitally simulating the enzymatic hydrolysis reaction using expected quality indicators of the yeast extract as processing constraints to generate a standard degree of hydrolysis curve and a standard product accumulation curve; setting an appropriate adjustment time step based on a preset reaction stage to obtain a feedback adjustment time cycle sequence, a standard degree of hydrolysis sequence, and a standard product accumulation sequence; and iteratively optimizing the enzymatic hydrolysis control parameters during the reaction process based on the feedback adjustment time cycle sequence, with the goal of approximating the standard degree of hydrolysis sequence and the standard product accumulation sequence, to obtain the optimal enzymatic hydrolysis control parameter sequence for iterative regulation of the enzymatic hydrolysis equipment. This solves the problem that existing enzymatic hydrolysis reaction control methods rely on experience or simple experimental designs, making it difficult to accurately control the reaction process, leading to unstable product quality and low yield.

Description

Technical Field

[0001] This application relates to the field of microbial detection and analysis technology, specifically to a parameter collaborative control system based on the enzymatic hydrolysis reaction of yeast extract. Background Technology

[0002] With the continuous development of biotechnology, yeast extracts are increasingly widely used in food, medicine, and bioengineering. As an important bioactive substance, the quality and yield of yeast extracts directly affect the quality and market competitiveness of downstream products. However, traditional enzymatic hydrolysis reaction control methods rely on experience or simple experimental designs, making it difficult to precisely control the reaction process, leading to problems such as unstable product quality and low yield. Summary of the Invention

[0003] This application provides a parameter-coordinated control system based on yeast extract enzymatic hydrolysis reaction, which solves the technical problem that existing enzymatic hydrolysis reaction control methods rely on experience or simple experimental design, making it difficult to accurately control the reaction process, resulting in unstable product quality and low yield.

[0004] The technical solution to the above-mentioned technical problems in this application is as follows:

[0005] In a first aspect, this application provides a parameter coordination control system based on the enzymatic hydrolysis reaction of yeast extract, comprising:

[0006] The enzymatic hydrolysis simulation module is used to digitally simulate the enzymatic hydrolysis reaction based on the expected quality indicators of yeast extract as processing constraints and the initial reaction conditions, generating standard hydrolysis degree curves and standard product cumulative curves.

[0007] The step size adjustment module is used to set an appropriate adjustment time step based on the preset reaction stage, and to divide the enzymatic hydrolysis reaction process, the standard degree of hydrolysis curve and the standard product accumulation curve respectively, and obtain the feedback adjustment time cycle sequence, the standard degree of hydrolysis sequence and the standard product accumulation sequence.

[0008] The equipment control module is used to adjust the time period sequence based on the feedback, with the standard degree of hydrolysis sequence and the standard product accumulation sequence as parameters for coordinated control, to iteratively optimize the enzymatic hydrolysis control parameters in the enzymatic hydrolysis reaction process, and to obtain the optimal enzymatic hydrolysis control parameter sequence for iterative control of the enzymatic hydrolysis reaction equipment.

[0009] This application provides one or more technical solutions, which have at least the following technical effects or advantages:

[0010] This application provides a parameter-coordinated control system for the enzymatic hydrolysis reaction of yeast extract. First, it accurately simulates the enzymatic hydrolysis reaction based on initial reaction conditions, generating standard hydrolysis degree curves and standard product accumulation curves to provide reference standards for subsequent enzymatic hydrolysis processes. Second, it divides the enzymatic hydrolysis process, standard hydrolysis degree curves, and standard product accumulation curves to obtain feedback adjustment time cycle sequences, standard hydrolysis degree sequences, and standard product accumulation amount sequences, ensuring precise control and optimization of the enzymatic hydrolysis process. Finally, based on the feedback adjustment time cycle sequences, iteratively optimizes the enzymatic hydrolysis control parameters, thereby solving the problems of unstable product quality and low yield in traditional enzymatic hydrolysis control methods, improving the quality and yield of yeast extract, and enhancing the quality and market competitiveness of downstream products.

[0011] Through the above technical solutions, the parameter coordination control system based on yeast extract enzymatic hydrolysis provided in this application embodiment can not only accurately simulate the enzymatic hydrolysis process, but also dynamically adjust according to real-time feedback data to ensure that the enzymatic hydrolysis reaction always takes place under optimal conditions. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a schematic diagram of a parameter collaborative control system based on the enzymatic hydrolysis reaction of yeast extract provided in an embodiment of this application.

[0014] The components represented by each number in the attached diagram are explained below:

[0015] Enzymatic hydrolysis simulation module 11, step size adjustment module 12, equipment control module 13. Detailed Implementation

[0016] This application provides a parameter-coordinated control system based on yeast extract enzymatic hydrolysis reaction, which addresses the technical problem that existing enzymatic hydrolysis reaction control methods rely on experience or simple experimental design, making it difficult to accurately control the reaction process, resulting in unstable product quality and low yield.

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0018] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0019] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use this application. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that this application can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid unnecessarily obscuring the description of this application. Therefore, this application is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0020] Examples, such as Figure 1 As shown in the embodiments of this application, a parameter synergistic control system based on yeast extract enzymatic hydrolysis reaction is provided, including:

[0021] Enzymatic hydrolysis simulation module 11 is used to perform digital simulation of enzymatic hydrolysis reaction based on the expected quality indicators of yeast extract as processing constraints and the initial reaction conditions, and generate standard hydrolysis degree curve and standard product cumulative curve.

[0022] In this embodiment, firstly, the expected quality indicators of the yeast extract are received, including the target degree of hydrolysis, amino acid nitrogen yield, nucleotide content, and glutamic acid content. Simultaneously, the initial conditions of the enzymatic hydrolysis reaction are obtained, covering variables and quantifications during the reaction process. Variables include temperature, pH, reaction time, and stirring speed; quantifications include yeast emulsion dry matter concentration, enzyme type, and enzyme dosage.

[0023] Based on the initial reaction conditions, mathematical models and algorithms are used to digitally simulate the enzymatic hydrolysis reaction. By simulating the interaction between the enzyme and the substrate, the changes in the degree of hydrolysis and the accumulation of products during the reaction are predicted. Finally, standard hydrolysis degree curves and standard product accumulation curves are generated as reference standards for subsequent enzymatic hydrolysis reaction control.

[0024] Specifically, step 11 in the system includes:

[0025] The expected quality indicators and initial reaction conditions of the yeast extract were obtained, wherein the expected quality indicators included the target degree of hydrolysis, amino acid nitrogen yield, 5'-nucleotide content and glutamic acid content, and the initial reaction conditions included the dry matter concentration of the yeast emulsion, the type of enzyme preparation and the amount of enzyme added;

[0026] Based on the historical processing logs of yeast extract, a set of sample quality indicators, a set of sample initial conditions, a set of sample standard hydrolysis degree curves, and a set of sample standard product cumulative curves were collected to train a machine learning model and construct an enzymatic hydrolysis reaction simulator. The products of the enzymatic hydrolysis reaction include amino acid nitrogen, 5'-nucleotides, and glutamic acid.

[0027] Using the enzymatic hydrolysis reaction simulator, the enzymatic hydrolysis reaction is digitally simulated according to the expected quality indicators and initial reaction conditions, and the standard degree of hydrolysis curve and standard product cumulative curve are output.

[0028] In this embodiment, firstly, the expected quality indicators of the yeast extract are obtained, including the target degree of hydrolysis, amino acid nitrogen yield, 5'-nucleotide content, and glutamic acid content. Simultaneously, the initial reaction conditions are obtained, including the dry matter concentration of the yeast emulsion, the type of enzyme preparation, and the amount of enzyme added.

[0029] Among them, the target degree of hydrolysis is an indicator that measures the extent to which proteins in yeast extracts are enzymatically hydrolyzed, which is related to the quality of the final product and the content of active ingredients; the amino acid nitrogen yield reflects the efficiency of protein conversion into amino acids during enzymatic hydrolysis and is a parameter for evaluating the nutritional value of yeast extracts; the content of 5'-nucleotides and glutamate, as functional components in yeast extracts, affect the bioactivity of the product.

[0030] The dry matter concentration of yeast emulsion is related to the concentration of substrate in the reaction system and the contact efficiency between enzyme and substrate; the type of enzyme preparation affects the specificity and efficiency of enzymatic hydrolysis, and different types of enzyme preparations have different substrate cleavage methods and product compositions; the amount of enzyme added determines the concentration of enzyme in the reaction system, which in turn affects the rate and extent of enzymatic hydrolysis.

[0031] Secondly, after obtaining the expected quality indicators and initial reaction conditions, based on the historical processing logs of yeast extract, a set of sample quality indicators, a set of sample initial conditions, a set of sample standard hydrolysis degree curves, and a set of sample standard product accumulation curves were collected. A machine learning model was then trained to construct an enzymatic hydrolysis reaction simulator. This simulator can simulate the enzymatic hydrolysis process under different initial conditions and predict changes in the degree of hydrolysis and product accumulation. The products of the enzymatic hydrolysis reaction include amino acid nitrogen, 5'-nucleotides, and glutamic acid.

[0032] Specifically, in the process of constructing the enzymatic hydrolysis reaction simulator, machine learning algorithms, such as those based on neural networks, are employed to deeply mine and analyze historical data, capturing the complex nonlinear relationships in the enzymatic hydrolysis reaction process. By continuously optimizing the model parameters, the prediction accuracy and generalization ability of the simulator are improved, ensuring that the enzymatic hydrolysis reaction simulator accurately predicts the standard degree of hydrolysis curve and the standard product accumulation curve under different reaction conditions.

[0033] For example, an enzymatic hydrolysis reaction simulator is constructed and trained based on a neural network, and the specific steps are as follows:

[0034] First, data preparation: collecting a set of sample quality indicators, a set of sample initial conditions, a set of sample standard hydrolysis degree curves, and a set of sample standard product cumulative curves, based on historical processing logs of yeast extract.

[0035] Secondly, in model construction, the number of nodes in the input layer is equal to the dimension of the input features. For example, if there are four features: sample quality index set, sample initial condition set, sample standard hydrolysis degree curve set, and sample standard product cumulative curve set, then the input layer contains four nodes. Set 1-3 hidden layers, and adjust the number of nodes in each layer through experiments, such as 64 or 32. The activation function is ReLU. The output layer generally does not use an activation function. If the output takes two nodes, directly output continuous values.

[0036] Next, the model is trained, and the predicted standard degree of hydrolysis curve and standard product accumulation curve are used as outputs. The training framework is constructed using the Adam optimizer and the mean squared error loss function, with a batch size of 32 and a total of 50 training epochs. An early stopping mechanism (patience=5) is introduced, which automatically terminates the training process when the validation set loss does not decrease for 5 consecutive epochs, resulting in a trained enzymatic reaction simulator. This effectively avoids model overfitting while ensuring that the model reaches a convergent state.

[0037] Finally, using the constructed enzymatic hydrolysis simulator, the enzymatic hydrolysis reaction was digitally simulated based on the input expected quality indicators and initial reaction conditions. During the simulation, the simulator calculated and output standard hydrolysis degree curves and standard product accumulation curves in real time, serving as the basis for subsequent enzymatic hydrolysis reaction control. By comparing the monitoring data from the actual reaction process with the standard curves, deviations in the reaction process were identified, and adjustment measures were taken to ensure that the enzymatic hydrolysis reaction always proceeded under optimal conditions, thereby improving the quality and yield of yeast extract.

[0038] The step size adjustment module 12 is used to set an appropriate adjustment time step based on a preset reaction stage, and to divide the enzymatic hydrolysis reaction process, the standard degree of hydrolysis curve and the standard product accumulation curve respectively, and to obtain the feedback adjustment time cycle sequence, the standard degree of hydrolysis sequence and the standard product accumulation sequence.

[0039] In this embodiment, an appropriate time step is preset based on the characteristics of different stages of the enzymatic hydrolysis reaction. In the early stage of the reaction, since the reaction system is not yet stable, a smaller time step is used for close monitoring to ensure that the initial changes in the reaction can be captured in time. As the reaction proceeds and the reaction system tends to stabilize, the time step is appropriately increased to reduce unnecessary computation and improve system efficiency.

[0040] Specifically, this module synchronously divides the enzymatic hydrolysis reaction process, the standard degree of hydrolysis curve, and the standard product accumulation curve. Based on a preset time step, the entire enzymatic hydrolysis reaction process is divided into multiple feedback adjustment time periods, forming a feedback adjustment time period sequence. Simultaneously, for each feedback adjustment time period, the corresponding degree of hydrolysis value and product accumulation value are extracted from the standard degree of hydrolysis curve and the standard product accumulation curve, respectively constructing the standard degree of hydrolysis sequence and the standard product accumulation sequence.

[0041] Among them, the adaptation adjustment time step is set based on the preset reaction stage, including:

[0042] A predetermined reaction stage for obtaining yeast extract for enzymatic hydrolysis, wherein the predetermined reaction stage includes a cell wall disruption initiation period, an efficient hydrolysis period, a flavor optimization period, and a reaction termination period;

[0043] The importance of feedback regulation was evaluated for the cell wall breaking start-up period, efficient hydrolysis period, flavor optimization period and reaction termination period respectively, and multiple regulation importance coefficients were output. The mean value of the regulation importance coefficients was calculated. The regulation importance coefficients were obtained based on the product quality impact and reaction sensitivity of the reaction stage.

[0044] The ratio of the mean of the control importance coefficients to the control importance coefficients is set as the time compensation coefficient. The product of the time compensation coefficient and the initial adjustment time step is used as the adaptive adjustment time step. Multiple adaptive adjustment time steps are calculated based on the multiple control importance coefficients.

[0045] In this embodiment, the enzymatic hydrolysis reaction of yeast extract is first divided into stages, with the preset reaction stages including the cell wall breaking initiation stage, the efficient hydrolysis stage, the flavor optimization stage, and the reaction termination stage.

[0046] Furthermore, for each stage, the sensitivity of product quality to the influence of reaction parameters and the contribution of the reaction rate to the overall process at that stage are analyzed to quantify the control importance coefficient of each stage. The control importance coefficients of all stages are then arithmetically averaged to obtain the mean control importance coefficient, which serves as a benchmark reference value.

[0047] Among them, the impact of product quality during the cell-wall breaking start-up period is medium to high, laying the foundation for the reaction and affecting subsequent efficiency; the reaction sensitivity is medium, and parameter deviations can be partially compensated in subsequent stages; the control importance coefficient is 3.2.

[0048] The product quality is significantly affected during the high-efficiency hydrolysis period, directly determining the yield of the core product; the reaction sensitivity is high, with even small fluctuations in temperature / pH significantly affecting the reaction rate; the control importance coefficient is 4.8.

[0049] The flavor optimization period has a very high impact on product quality, determining the final flavor quality; the reaction is highly sensitive, requiring precise control to prevent bitterness; the control importance coefficient is 5.0.

[0050] The product quality is significantly affected during the reaction termination period, impacting product consistency and stability; the reaction is highly sensitive, requiring precise timing; the control importance coefficient is 4.5.

[0051] Secondly, by calculating the ratio of the control importance coefficient to the mean at each stage, the time compensation coefficient is determined. This coefficient reflects the adjustment range of the time step that needs to be adjusted at different stages.

[0052] For example, the control importance coefficient for the cell wall breaking start-up period is 3.2, for the efficient hydrolysis period it is 4.8, for the flavor optimization period it is 5.0, and for the reaction termination period it is 4.5. The average control importance coefficient is (3.2+4.8+5.0+4.5) / 4=4.375.

[0053] At this point, the time compensation coefficient for the cell wall breaking start-up period is 3.2 / 4.375=0.73; the time compensation coefficient for the efficient hydrolysis period is 4.8 / 4.375=1.10; the time compensation coefficient for the flavor optimization period is 5.0 / 4.375=1.14; and the time compensation coefficient for the reaction termination period is 4.5 / 4.375=1.03.

[0054] Finally, the initial set base adjustment time step is multiplied by the time compensation coefficient to obtain the dynamic adjustment time step adapted to each reaction stage, ensuring that an intensive monitoring strategy is adopted in the early stage of the reaction, an appropriate monitoring frequency is maintained in the stable period, and precise control is implemented before termination.

[0055] For example, if the initial adjustment time step t0 is set to 5 minutes, then the adaptation adjustment time step for the cell disruption start-up period is 5 × 0.73 = 3.65 minutes, the efficient hydrolysis period is 5 × 1.10 = 5.5 minutes, the flavor optimization period is 5 × 1.14 = 5.7 minutes, and the reaction termination period is 5 × 1.03 = 5.15 minutes.

[0056] The higher the regulatory importance coefficient, the shorter the reaction time and the higher the regulation frequency. By dynamically adjusting the time step, the system can achieve more precise control at different reaction stages, avoiding over-calculation while ensuring the monitoring density of key stages, thus ensuring a high degree of match between the time allocation and reaction characteristics of the entire enzymatic hydrolysis process.

[0057] Furthermore, the enzymatic hydrolysis reaction process, the standard degree of hydrolysis curve, and the standard product accumulation curve are divided to obtain the feedback adjustment time period sequence, the standard degree of hydrolysis sequence, and the standard product accumulation sequence, including:

[0058] Based on the preset sequence of reaction stages, the preset reaction time of the enzymatic hydrolysis process is divided according to the multiple adaptive adjustment time steps to generate a feedback adjustment time cycle sequence.

[0059] Based on the preset sequence of reaction stages, node data are extracted from the standard degree of hydrolysis curve and the standard product accumulation curve according to the multiple adaptive adjustment time steps to generate a standard degree of hydrolysis sequence and a standard product accumulation sequence.

[0060] In this embodiment, firstly, the total time of the enzymatic hydrolysis reaction is segmented according to the time sequence of the preset reaction stages, and the time step is adjusted as an interval to generate a feedback adjustment time cycle sequence.

[0061] For example, the cell disruption start-up period lasts 15 minutes and can be divided into 5 feedback cycles with a step size of 3.65 minutes; the efficient hydrolysis period lasts 60 minutes and can be divided into 11 feedback cycles with a step size of 5.5 minutes.

[0062] The sequence of the preset reaction stages is synchronized on the standard degree of hydrolysis curve and the standard product accumulation curve. The degree of hydrolysis value and the product accumulation value are extracted at the start time point of each feedback cycle to form a standard degree of hydrolysis sequence and a standard product accumulation sequence that completely correspond to the time points of the feedback cycle sequence.

[0063] By constructing a sequence with spatiotemporal synchronization, it is ensured that a corresponding standard reference value can be obtained for each feedback cycle, providing a comparison benchmark for subsequent parameter adjustment.

[0064] The equipment control module 13 is used to adjust the time period sequence based on the feedback, with the standard degree of hydrolysis sequence and the standard product accumulation sequence as the parameter collaborative control target, to iteratively optimize the enzymatic hydrolysis control parameters in the enzymatic hydrolysis reaction process, and to obtain the optimal enzymatic hydrolysis control parameter sequence to iteratively control the enzymatic hydrolysis reaction equipment.

[0065] In this embodiment, based on the feedback adjustment time cycle sequence, and with the standard hydrolysis degree sequence and standard product accumulation sequence as optimization targets, the enzymatic hydrolysis reaction parameters are dynamically adjusted through an iterative algorithm. Actual hydrolysis degree and product accumulation data are collected within each feedback cycle, and deviation analysis is performed compared with the standard sequence. Optimization methods such as gradient descent or genetic algorithms are used to calculate the adjustment amounts for parameters such as temperature, pH, and stirring speed.

[0066] For example, when the actual degree of hydrolysis is lower than the standard value, the amount of enzyme added is increased or the reaction time is extended; if the product accumulation deviates from the expectation, the enzyme activity is optimized by adjusting the pH value. Through multiple iterations, an optimal parameter sequence covering the entire reaction cycle is generated and transmitted to the enzymatic hydrolysis equipment in real time, achieving precise closed-loop control from the initial conditions to the termination stage. This process ensures the product yield and quality stability at each stage by dynamically matching parameter adjustments with the reaction progress.

[0067] Specifically, step 13 in the system includes:

[0068] In the feedback adjustment time period sequence, the first feedback adjustment time period is selected as the first feedback adjustment time period, and the adjacent period of the first feedback adjustment time period is set as the second feedback adjustment time period. The second standard degree of hydrolysis and the second standard product accumulation corresponding to the second feedback adjustment time period are obtained.

[0069] The enzymatic hydrolysis reaction equipment is controlled to perform the enzymatic hydrolysis reaction within the first feedback adjustment time period according to the preset enzymatic hydrolysis control parameters, and the first monitoring data sequence set within the first feedback adjustment time period is monitored and acquired. The enzymatic hydrolysis control parameters include heating temperature, pH value and stirring speed, and the monitoring data includes reaction solution pH, reaction solution temperature, real-time conductivity, reaction solution viscosity and soluble solids accumulation.

[0070] The first predicted degree of hydrolysis and the first predicted cumulative amount of product are predicted based on the first monitoring data sequence set.

[0071] With the goal of approximating the second standard degree of hydrolysis and the second standard product accumulation, the enzymatic hydrolysis control parameters in the enzymatic hydrolysis process are iteratively optimized based on the first predicted degree of hydrolysis and the first predicted product accumulation, and the second optimal enzymatic hydrolysis control parameters are output to optimize the control of the enzymatic hydrolysis equipment within the second feedback adjustment time period.

[0072] The time cycle sequence is adjusted according to the feedback, and the enzymatic control parameters in the enzymatic hydrolysis process are iteratively optimized, as well as the enzymatic hydrolysis equipment is iteratively controlled, until the enzymatic hydrolysis of the yeast extract is completed.

[0073] In this embodiment of the application, firstly, the first feedback adjustment time period in the feedback adjustment time period sequence is selected as the first feedback adjustment time period, and the adjacent period is selected as the second feedback adjustment time period. The second standard degree of hydrolysis and the cumulative amount of the second standard product corresponding to the second feedback adjustment time period are obtained as optimization targets.

[0074] Next, the enzymatic hydrolysis equipment is started according to the preset enzymatic hydrolysis control parameters, such as heating temperature, pH value, and stirring speed. The reaction is carried out within the first feedback adjustment time period. At the same time, data such as pH, temperature, conductivity, viscosity and cumulative amount of soluble solids in the reaction solution are monitored and recorded in real time to form the first monitoring data sequence set.

[0075] Subsequently, using a pre-trained prediction model, the first predicted degree of hydrolysis and the first predicted cumulative amount of product are calculated based on the first monitoring data sequence set.

[0076] Furthermore, by comparing the deviations between the predicted values ​​and the second standard degree of hydrolysis and the second standard product accumulation corresponding to the second feedback adjustment time period, optimization methods such as gradient descent or genetic algorithm are used to dynamically adjust the enzymatic hydrolysis control parameters and generate the second optimal enzymatic hydrolysis control parameters.

[0077] Finally, the second optimal enzymatic hydrolysis control parameters are applied to the enzymatic hydrolysis reaction equipment within the second feedback adjustment time period, achieving iterative optimization of parameters and precise control of the equipment. This process is repeated cyclically until all feedback adjustment time periods are covered, completing the full-process enzymatic hydrolysis reaction control of the yeast extract.

[0078] Further, based on the first monitoring data sequence set, the first predicted degree of hydrolysis and the first predicted cumulative amount of product are predicted, including:

[0079] Based on the historical processing logs of yeast extract, multiple sample monitoring datasets were collected, and the historical degree of hydrolysis corresponding to different sample monitoring datasets was obtained as the sample degree of hydrolysis, and multiple sample degrees of hydrolysis were obtained.

[0080] The multiple sample monitoring datasets and multiple sample hydrolysis degrees are used as training data, and Q sample training sets are obtained by random selection with replacement, where Q is an integer greater than or equal to 10.

[0081] The Q sample training sets are used to train BP neural networks until convergence, resulting in Q hydrolysis degree prediction models. A hydrolysis degree predictor is then constructed by integrating these models based on the mean fusion strategy.

[0082] Using the hydrolysis degree predictor, a first predicted hydrolysis degree is obtained based on the first monitoring data sequence set;

[0083] A product accumulation predictor is constructed based on a BP neural network. The product accumulation is predicted based on the first predicted degree of hydrolysis, the first viscosity of the reaction liquid, and the first soluble solids accumulation, and the first predicted product accumulation is output.

[0084] In this embodiment, firstly, multiple sets of sample monitoring datasets are extracted from the historical processing logs of yeast extract. Each set of data includes characteristic parameters such as pH, temperature, conductivity, viscosity, and cumulative soluble solids in the reaction solution, and the corresponding historical degree of hydrolysis is recorded as a label value. Simultaneously, the historical degree of hydrolysis corresponding to different sample monitoring datasets is obtained as the sample degree of hydrolysis, resulting in multiple sample degrees of hydrolysis.

[0085] Secondly, multiple sample monitoring datasets and multiple sample hydrolysis degrees are used as training data. Q sample training sets are generated by random sampling with replacement, where Q≥10, to ensure that each training set contains different proportions of sample data to enhance the model's generalization ability.

[0086] Next, a BP neural network model is trained using Q sample training sets. The network weights are iteratively adjusted to continuously converge the prediction error, ultimately resulting in Q independently trained sub-models for predicting the degree of hydrolysis. The BP neural network is an error backpropagation network that minimizes the error between the network output and the actual target through gradient descent. A mean fusion strategy is used to weighted average the outputs of the Q sub-models, constructing an integrated degree of hydrolysis predictor. This predictor effectively reduces the risk of overfitting by integrating the prediction results of multiple base models.

[0087] For example, a hydrolysis degree predictor is constructed and trained based on a BP neural network, and the specific steps are as follows:

[0088] First, data preparation involved collecting multiple sample monitoring datasets and multiple sample hydrolysis degrees, based on historical processing logs of yeast extract.

[0089] Secondly, in model building, the number of nodes in the input layer is equal to the dimension of the input features. For example, if there are Q training samples with a total of 6 features, then the input layer contains 6 nodes. Set 1-3 hidden layers, and adjust the number of nodes in each layer through experiments, such as 64, 32, etc. The activation function is ReLU. The output layer generally does not use an activation function. For example, if the output takes 2 nodes, directly output continuous values.

[0090] Next, for model training, the predicted degree of hydrolysis is used as the output. The Adam optimizer and mean squared error loss function are used to build the training framework. The batch size is set to 32 and the total number of training rounds is 50. An early stopping mechanism is introduced (patience=5, where Patience is a hyperparameter used to control the triggering condition of the early stopping mechanism). When the validation set loss does not decrease for 5 consecutive rounds, the training process is automatically terminated, and the trained degree of hydrolysis predictor is obtained. This effectively avoids model overfitting and ensures that the model reaches a convergent state.

[0091] In the actual prediction phase, the first monitoring data sequence set collected within the first feedback adjustment time period is input into the hydrolysis degree predictor, and the first predicted hydrolysis degree is output.

[0092] Meanwhile, a product accumulation predictor is constructed based on the BP neural network architecture. The product accumulation predictor uses the first predicted degree of hydrolysis, the first reaction liquid viscosity, and the first soluble solids accumulation as input features, and outputs the first predicted product accumulation through nonlinear mapping.

[0093] For example, the product accumulation predictor is constructed and trained based on a BP neural network, and the specific steps are as follows:

[0094] First, data preparation: the first predicted degree of hydrolysis, the first reaction solution viscosity, and the first cumulative amount of soluble solids were collected based on the historical processing logs of yeast extract.

[0095] Secondly, in model building, the number of nodes in the input layer is equal to the dimension of the input features. For example, if there are three features, namely the first predicted degree of hydrolysis, the first reaction liquid viscosity, and the first soluble solids accumulation, then the input layer contains three nodes. Set 1-3 hidden layers, and adjust the number of nodes in each layer through experiments, such as 64 or 32. The activation function is ReLU. The output layer generally does not use an activation function. For example, if the output takes two nodes, directly output continuous values.

[0096] Next, during model training, the cumulative predicted output is used as the output. The Adam optimizer and mean squared error loss function are used to construct the training framework. The batch size is set to 32 and the total number of training rounds is 50. An early stopping mechanism is introduced (patience=5, where Patience is a hyperparameter used to control the triggering condition of the early stopping mechanism). When the validation set loss does not decrease for 5 consecutive rounds, the training process is automatically terminated, and the trained cumulative predictor of the output is obtained. This effectively avoids model overfitting while ensuring that the model reaches a convergent state.

[0097] In the actual prediction stage, the first predicted degree of hydrolysis, the first reaction liquid viscosity, and the first soluble solids accumulation are input into the product accumulation predictor, and the first predicted product accumulation predictor is output.

[0098] The method of using the hydrolysis degree predictor to predict the first predicted hydrolysis degree based on the first monitoring data sequence set includes:

[0099] The monitoring data of the latest monitoring node is extracted from the first monitoring data sequence set to obtain the pH of the first reaction solution, the temperature of the first reaction solution, the first real-time conductivity, the viscosity of the first reaction solution, and the cumulative amount of the first soluble solids.

[0100] The data volatility of several monitoring data sequences of several monitoring indicators in the first monitoring data sequence set is calculated and weighted and fused to obtain the first comprehensive volatility coefficient.

[0101] The ratio of the first comprehensive volatility coefficient to the historical maximum comprehensive volatility coefficient within the historical time range is multiplied by Q and rounded to obtain the number of adaptive models selected, K, where K is greater than or equal to 2 and less than or equal to Q.

[0102] K hydrolysis degree prediction models are randomly selected from the Q hydrolysis degree prediction models. The prediction is made based on the pH of the first reaction solution, the temperature of the first reaction solution, the first real-time conductivity, the viscosity of the first reaction solution, and the first cumulative amount of soluble solids. The first predicted hydrolysis degree is then output.

[0103] In this embodiment of the application, firstly, the data of the latest monitoring node are extracted from the first monitoring data sequence set, including the pH value of the first reaction liquid, the temperature of the first reaction liquid, the first real-time conductivity, the viscosity of the first reaction liquid, and the cumulative amount of the first soluble solids, as input features of the prediction model.

[0104] Simultaneously, data volatility analysis was conducted on key monitoring indicators such as pH, temperature, conductivity, and viscosity of the reaction liquid in the first monitoring data sequence set. The degree of volatility was quantified by calculating the standard deviation or coefficient of variation of each indicator, and the volatility indicators were weighted and fused to obtain the first comprehensive volatility coefficient that comprehensively reflects the data volatility.

[0105] For example, assuming the weights of pH, temperature, conductivity, and viscosity of the reaction solution are 0.3, 0.3, 0.2, and 0.2, respectively, and the standard deviation of pH is 0.15, the standard deviation of temperature is 0.25, the standard deviation of conductivity is 0.1, and the standard deviation of viscosity is 0.2, then the first comprehensive fluctuation coefficient is 0.3×0.15+0.3×0.25+0.2×0.1+0.2×0.2=0.19.

[0106] Secondly, the ratio of the first comprehensive volatility coefficient to the historical maximum comprehensive volatility coefficient within the preset historical time range is calculated, and the ratio is multiplied by the total number of models Q and rounded to determine the number of suitable models to be selected, K. The value of K is between 2 and Q, which ensures that the predictive advantages of multiple models can be utilized while avoiding the increase in computational complexity caused by too many models.

[0107] For example, when the first comprehensive volatility coefficient is 0.19, the historical maximum comprehensive volatility coefficient is 0.5, and the total number of models Q is 10, then K = (0.19 / 0.5) × 10, rounded down, is 4. That is, 4 models are randomly selected from the Q hydrolysis degree prediction models to participate in the prediction.

[0108] Finally, K models are randomly selected from the Q pre-trained hydrolysis degree prediction models. Input features such as the pH value of the first reaction solution, the temperature of the first reaction solution, the first real-time conductivity, the viscosity of the first reaction solution, and the first cumulative amount of soluble solids are input into these K models respectively. Each model outputs a predicted hydrolysis degree value. By averaging or weighted averaging these K predicted values, the final first predicted hydrolysis degree is obtained, which provides a basis for the iterative optimization of subsequent enzymatic hydrolysis reaction parameters.

[0109] Furthermore, with the goal of approximating the second standard degree of hydrolysis and the second standard product accumulation, the enzymatic hydrolysis control parameters in the enzymatic hydrolysis process are iteratively optimized based on the first predicted degree of hydrolysis and the first predicted product accumulation, outputting the second optimal enzymatic hydrolysis control parameters, including:

[0110] The first enzymatic hydrolysis control parameter is randomly selected within the adjustment space of the enzymatic hydrolysis control parameter of the enzymatic hydrolysis reaction equipment;

[0111] A prediction model for enzymatic hydrolysis reaction was constructed based on a BP neural network.

[0112] Using the enzymatic hydrolysis reaction prediction model, based on the first enzymatic hydrolysis control parameters, the first predicted degree of hydrolysis and the first predicted product accumulation, the enzymatic hydrolysis reaction prediction within the second feedback adjustment time period is performed, and the second virtual degree of hydrolysis and the second virtual product accumulation are output.

[0113] Based on the second standard degree of hydrolysis and the cumulative amount of the second standard product, the deviations of the second virtual degree of hydrolysis and the cumulative amount of the second virtual product are calculated respectively, and the differences are weighted to obtain the first overall reaction deviation.

[0114] Next, a second enzymatic hydrolysis control parameter is randomly selected within the adjustment space of the enzymatic hydrolysis control parameter, and the second overall reaction deviation is analyzed and obtained.

[0115] Continue iterative selection of enzymatic hydrolysis control parameters and iterative analysis of reaction deviation until the second optimal convergence number is reached. Output the enzymatic hydrolysis control parameter corresponding to the minimum reaction deviation as the second optimal enzymatic hydrolysis control parameter.

[0116] In this embodiment of the application, firstly, a first set of enzymatic hydrolysis control parameters is randomly selected as initial parameters within the adjustment space of the enzymatic hydrolysis control parameters of the enzymatic hydrolysis reaction equipment. This parameter adjustment space covers the preset range of key parameters such as heating temperature, pH value, and stirring speed.

[0117] Secondly, an enzymatic hydrolysis reaction prediction model is constructed based on the BP neural network architecture. The input layer of the model contains enzymatic hydrolysis control parameters, including temperature, pH, rotation speed, first predicted degree of hydrolysis and first predicted product accumulation, etc. The hidden layer adopts a two-layer structure and is equipped with the ReLU activation function. The output layer directly maps the second virtual degree of hydrolysis and the second virtual product accumulation.

[0118] Next, using the trained prediction model, the first set of enzymatic hydrolysis control parameters and prediction features are input into the model to simulate the enzymatic hydrolysis reaction process within the second feedback adjustment time period, and the corresponding second virtual degree of hydrolysis and second virtual product accumulation are output.

[0119] Furthermore, based on the preset second standard degree of hydrolysis and the second standard product accumulation, the absolute deviation between the virtual value and the standard value is calculated respectively, and the first overall reaction deviation is synthesized according to the weighted ratio of degree of hydrolysis and product accumulation, with the degree of hydrolysis weighting 0.6 and the product accumulation weighting 0.4.

[0120] Then, within the parameter adjustment space, a second set of enzymatic hydrolysis control parameters is randomly selected, and the above prediction and deviation calculation process is repeated to obtain the second overall reaction deviation. By iterating multiple times to select different parameter combinations and analyzing the corresponding reaction deviations, when the number of iterations reaches the preset second optimization convergence number, such as 20 iterations, the parameter combination with the smallest cumulative reaction deviation is selected as the second optimal enzymatic hydrolysis control parameter, which is used to guide the regulation of the enzymatic hydrolysis reaction equipment in the next cycle.

[0121] Specifically, the methods for setting the second optimization convergence number include:

[0122] Based on the first standard degree of hydrolysis and the first standard product accumulation corresponding to the first feedback adjustment time period, the deviation of the first predicted degree of hydrolysis and the first predicted product accumulation is calculated and weighted to obtain the first control deviation amplitude.

[0123] The product of the first control deviation amplitude and the constant L is used as the deviation compensation coefficient, and the sum of 1 and the deviation compensation coefficient is used as the first optimization number compensation coefficient, wherein L is greater than 20 and less than 50.

[0124] The product of the first optimization number compensation coefficient and the preset initial optimization convergence number is set as the second optimization convergence number.

[0125] In this embodiment, firstly, using the first standard degree of hydrolysis and the first standard product accumulation actually reached within the first feedback adjustment period as benchmark values, the absolute deviation between the first predicted degree of hydrolysis and the first standard degree of hydrolysis, as well as the absolute deviation between the first predicted product accumulation and the first standard product accumulation, are calculated. By setting a weighting ratio of 0.6 for the degree of hydrolysis and 0.4 for the product accumulation, the two absolute deviations are combined into a first control deviation amplitude, reflecting the degree of deviation between the current predicted value and the standard value.

[0126] Secondly, the first control deviation amplitude is multiplied by a preset constant L to obtain the deviation compensation coefficient, where L ranges from 20 to 50. This coefficient is used to dynamically adjust the number of optimization iterations. By summing 1 with the deviation compensation coefficient, a first optimization iteration compensation coefficient is constructed. This coefficient increases linearly with the increase of the actual control deviation, ensuring that when the predicted result deviates significantly from the standard value, the system automatically increases the number of parameter optimization iterations.

[0127] Finally, the compensation coefficient for the first optimization attempt is multiplied by the preset initial optimization convergence attempt to obtain the dynamically adjusted second optimization convergence attempt. The dynamic setting mechanism enables the system to automatically adjust the optimization intensity according to the actual response deviation, ensuring convergence accuracy while avoiding unnecessary consumption of computational resources.

[0128] For example, if the absolute deviation between the first predicted degree of hydrolysis and the first standard degree of hydrolysis is 0.05, and the absolute deviation between the first predicted cumulative product amount and the first standard cumulative product amount is 0.03, then the first control deviation amplitude is 0.6 × 0.05 + 0.4 × 0.03 = 0.042. Next, multiplying the first control deviation amplitude by a preset constant L (e.g., L = 30) yields a deviation compensation coefficient of 0.042 × 30 = 1.26. Adding 1 to the deviation compensation coefficient yields a first optimization number compensation coefficient of 1 + 1.26 = 2.26. Finally, multiplying the first optimization number compensation coefficient by a preset initial optimization convergence number (e.g., initially set to 10 times) yields a second optimization convergence number of 2.26 × 10 = 22.6, which, after rounding, becomes 23 times.

[0129] In summary, compared with existing technologies, this application achieves accurate prediction of the degree of hydrolysis and product accumulation during the enzymatic hydrolysis of yeast extract and intelligent optimization of control parameters by constructing an integrated prediction model based on BP neural network and combining it with a dynamic parameter optimization mechanism.

[0130] This application provides a parameter-coordinated control system for the enzymatic hydrolysis reaction of yeast extract. First, it accurately simulates the enzymatic hydrolysis reaction based on initial reaction conditions, generating standard hydrolysis degree curves and standard product accumulation curves to provide reference standards for subsequent enzymatic hydrolysis processes. Second, it divides the enzymatic hydrolysis process, standard hydrolysis degree curves, and standard product accumulation curves, obtaining feedback adjustment time cycle sequences, standard hydrolysis degree sequences, and standard product accumulation amount sequences to ensure precise control and optimization of the enzymatic hydrolysis process. Finally, based on the feedback adjustment time cycle sequences, iteratively optimizes the enzymatic hydrolysis control parameters, thereby solving the problems of unstable product quality and low yield in traditional enzymatic hydrolysis control methods, improving the quality and yield of yeast extract, and enhancing the quality and market competitiveness of downstream products. Through the above technical solution, the parameter-coordinated control system for the enzymatic hydrolysis reaction of yeast extract provided in this application not only achieves accurate simulation of the enzymatic hydrolysis process but also dynamically adjusts based on real-time feedback data, ensuring that the enzymatic hydrolysis reaction always proceeds under optimal conditions.

[0131] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0132] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0133] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A parameter-coordinated control system based on the enzymatic hydrolysis reaction of yeast extract, characterized in that, include: The enzymatic hydrolysis simulation module is used to digitally simulate the enzymatic hydrolysis reaction based on the expected quality indicators of yeast extract as processing constraints and the initial reaction conditions, generating standard hydrolysis degree curves and standard product cumulative curves. The step size adjustment module is used to set an appropriate adjustment time step based on the preset reaction stage, and to divide the enzymatic hydrolysis reaction process, the standard degree of hydrolysis curve and the standard product accumulation curve respectively, and obtain the feedback adjustment time cycle sequence, the standard degree of hydrolysis sequence and the standard product accumulation sequence. The equipment control module is used to adjust the time period sequence based on the feedback, with the standard degree of hydrolysis sequence and the standard product accumulation sequence as the parameter collaborative control target, to iteratively optimize the enzymatic hydrolysis control parameters in the enzymatic hydrolysis reaction process, and to obtain the optimal enzymatic hydrolysis control parameter sequence to iteratively control the enzymatic hydrolysis reaction equipment. Using the expected quality indicators of yeast extract as processing constraints, a digital simulation of the enzymatic hydrolysis reaction was performed based on the initial reaction conditions to generate standard hydrolysis degree curves and standard product accumulation curves, including: The expected quality indicators and initial reaction conditions of the yeast extract were obtained, wherein the expected quality indicators included the target degree of hydrolysis, amino acid nitrogen yield, 5'-nucleotide content and glutamic acid content, and the initial reaction conditions included the dry matter concentration of the yeast emulsion, the type of enzyme preparation and the amount of enzyme added; Based on the historical processing logs of yeast extract, a set of sample quality indicators, a set of sample initial conditions, a set of sample standard hydrolysis degree curves, and a set of sample standard product cumulative curves were collected to train a machine learning model and construct an enzymatic hydrolysis reaction simulator. The products of the enzymatic hydrolysis reaction include amino acid nitrogen, 5'-nucleotides, and glutamic acid. Using the enzymatic hydrolysis reaction simulator, the enzymatic hydrolysis reaction is digitally simulated according to the expected quality indicators and initial reaction conditions, and the standard degree of hydrolysis curve and standard product accumulation curve are output. The time step for adaptation is set based on the preset reaction stage, including: A predetermined reaction stage for obtaining yeast extract for enzymatic hydrolysis, wherein the predetermined reaction stage includes a cell wall disruption initiation period, an efficient hydrolysis period, a flavor optimization period, and a reaction termination period; The importance of feedback regulation was evaluated for the cell wall breaking start-up period, efficient hydrolysis period, flavor optimization period and reaction termination period respectively, and multiple regulation importance coefficients were output. The mean value of the regulation importance coefficients was calculated. The regulation importance coefficients were obtained based on the product quality impact and reaction sensitivity of the reaction stage. The ratio of the mean of the control importance coefficients to the control importance coefficients is set as the time compensation coefficient. The product of the time compensation coefficient and the initial adjustment time step is used as the adaptive adjustment time step. Multiple adaptive adjustment time steps are calculated based on the multiple control importance coefficients.

2. The parameter synergistic control system based on yeast extract enzymatic hydrolysis reaction according to claim 1, characterized in that, The enzymatic hydrolysis reaction process, the standard degree of hydrolysis curve, and the standard product accumulation curve are divided into segments to obtain the feedback adjustment time period sequence, the standard degree of hydrolysis sequence, and the standard product accumulation sequence, including: Based on the preset sequence of reaction stages, the preset reaction time of the enzymatic hydrolysis process is divided according to the multiple adaptive adjustment time steps to generate a feedback adjustment time cycle sequence. Based on the preset sequence of reaction stages, node data are extracted from the standard degree of hydrolysis curve and the standard product accumulation curve according to the multiple adaptive adjustment time steps to generate a standard degree of hydrolysis sequence and a standard product accumulation sequence.

3. The parameter synergistic control system based on yeast extract enzymatic hydrolysis reaction according to claim 1, characterized in that, Based on the feedback adjustment time period sequence, and with the goal of approximating the standard degree of hydrolysis sequence and the standard product accumulation sequence as the parameters for coordinated control, the enzymatic hydrolysis control parameters in the enzymatic hydrolysis reaction process are iteratively optimized to obtain the optimal enzymatic hydrolysis control parameter sequence for iterative regulation of the enzymatic hydrolysis reaction equipment, including: In the feedback adjustment time period sequence, the first feedback adjustment time period is selected as the first feedback adjustment time period, and the adjacent period of the first feedback adjustment time period is set as the second feedback adjustment time period. The second standard degree of hydrolysis and the second standard product accumulation corresponding to the second feedback adjustment time period are obtained. The enzymatic hydrolysis reaction equipment is controlled to perform the enzymatic hydrolysis reaction within the first feedback adjustment time period according to the preset enzymatic hydrolysis control parameters, and the first monitoring data sequence set within the first feedback adjustment time period is monitored and acquired. The enzymatic hydrolysis control parameters include heating temperature, pH value and stirring speed, and the monitoring data includes reaction solution pH, reaction solution temperature, real-time conductivity, reaction solution viscosity and soluble solids accumulation. The first predicted degree of hydrolysis and the first predicted cumulative amount of product are predicted based on the first monitoring data sequence set. With the goal of approximating the second standard degree of hydrolysis and the second standard product accumulation, the enzymatic hydrolysis control parameters in the enzymatic hydrolysis process are iteratively optimized based on the first predicted degree of hydrolysis and the first predicted product accumulation, and the second optimal enzymatic hydrolysis control parameters are output to optimize the control of the enzymatic hydrolysis equipment within the second feedback adjustment time period. The time cycle sequence is adjusted according to the feedback, and the enzymatic control parameters in the enzymatic hydrolysis process are iteratively optimized, as well as the enzymatic hydrolysis equipment is iteratively controlled, until the enzymatic hydrolysis of the yeast extract is completed.

4. The parameter synergistic control system based on yeast extract enzymatic hydrolysis reaction according to claim 3, characterized in that, Based on the first monitoring data sequence set, the first predicted degree of hydrolysis and the first predicted cumulative amount of product are predicted, including: Based on the historical processing logs of yeast extract, multiple sample monitoring datasets were collected, and the historical degree of hydrolysis corresponding to different sample monitoring datasets was obtained as the sample degree of hydrolysis, and multiple sample degrees of hydrolysis were obtained. The multiple sample monitoring datasets and multiple sample hydrolysis degrees are used as training data, and Q sample training sets are obtained by random selection with replacement, where Q is an integer greater than or equal to 10. The Q sample training sets are used to train BP neural networks until convergence, resulting in Q hydrolysis degree prediction models. A hydrolysis degree predictor is then constructed by integrating these models based on the mean fusion strategy. Using the hydrolysis degree predictor, a first predicted hydrolysis degree is obtained based on the first monitoring data sequence set; A product accumulation predictor is constructed based on a BP neural network. The product accumulation is predicted based on the first predicted degree of hydrolysis, the first viscosity of the reaction liquid, and the first soluble solids accumulation, and the first predicted product accumulation is output.

5. The parameter synergistic control system based on yeast extract enzymatic hydrolysis reaction according to claim 4, characterized in that, Using the hydrolysis degree predictor, a first predicted hydrolysis degree is predicted based on the first monitoring data sequence set, including: The monitoring data of the latest monitoring node is extracted from the first monitoring data sequence set to obtain the pH of the first reaction solution, the temperature of the first reaction solution, the first real-time conductivity, the viscosity of the first reaction solution, and the cumulative amount of the first soluble solids. The data volatility of several monitoring data sequences of several monitoring indicators in the first monitoring data sequence set is calculated and weighted and fused to obtain the first comprehensive volatility coefficient. The ratio of the first comprehensive volatility coefficient to the historical maximum comprehensive volatility coefficient within the historical time range is multiplied by Q and rounded to obtain the number of adaptive models selected, K, where K is greater than or equal to 2 and less than or equal to Q. K hydrolysis degree prediction models are randomly selected from the Q hydrolysis degree prediction models. The prediction is made based on the pH of the first reaction solution, the temperature of the first reaction solution, the first real-time conductivity, the viscosity of the first reaction solution, and the first cumulative amount of soluble solids. The first predicted hydrolysis degree is then output.

6. The parameter synergistic control system based on yeast extract enzymatic hydrolysis reaction according to claim 3, characterized in that, With the goal of approximating the second standard degree of hydrolysis and the second standard product accumulation, the enzymatic hydrolysis control parameters in the enzymatic hydrolysis process are iteratively optimized based on the first predicted degree of hydrolysis and the first predicted product accumulation, outputting the second optimal enzymatic hydrolysis control parameters, including: The first enzymatic hydrolysis control parameter is randomly selected within the adjustment space of the enzymatic hydrolysis control parameter of the enzymatic hydrolysis reaction equipment; A prediction model for enzymatic hydrolysis reaction was constructed based on a BP neural network. Using the enzymatic hydrolysis reaction prediction model, based on the first enzymatic hydrolysis control parameters, the first predicted degree of hydrolysis and the first predicted product accumulation, the enzymatic hydrolysis reaction prediction within the second feedback adjustment time period is performed, and the second virtual degree of hydrolysis and the second virtual product accumulation are output. Based on the second standard degree of hydrolysis and the cumulative amount of the second standard product, the deviations of the second virtual degree of hydrolysis and the cumulative amount of the second virtual product are calculated respectively, and the differences are weighted to obtain the first overall reaction deviation. Next, a second enzymatic hydrolysis control parameter is randomly selected within the adjustment space of the enzymatic hydrolysis control parameter, and the second overall reaction deviation is analyzed and obtained. Continue iterative selection of enzymatic hydrolysis control parameters and iterative analysis of reaction deviation until the second optimal convergence number is reached. Output the enzymatic hydrolysis control parameter corresponding to the minimum reaction deviation as the second optimal enzymatic hydrolysis control parameter.

7. The parameter synergistic control system based on yeast extract enzymatic hydrolysis reaction according to claim 6, characterized in that, The methods for setting the second optimization convergence number include: Based on the first standard degree of hydrolysis and the first standard product accumulation corresponding to the first feedback adjustment time period, the deviation of the first predicted degree of hydrolysis and the first predicted product accumulation is calculated and weighted to obtain the first control deviation amplitude. The product of the first control deviation amplitude and the constant L is used as the deviation compensation coefficient, and the sum of 1 and the deviation compensation coefficient is used as the first optimization number compensation coefficient, wherein L is greater than 20 and less than 50. The product of the first optimization number compensation coefficient and the preset initial optimization convergence number is set as the second optimization convergence number.

Images