A dynamic process control method for ficius microcarpa probiotic fermentation process

Abstract

This invention belongs to the technical field of microbial detection and analysis instruments and systems. Specifically, it is a dynamic process control method for the fermentation process of *Prunus pedatus* probiotics. It employs an in-tank temperature change model to predict the in-tank temperature field distribution during a predicted period, obtaining a predicted in-tank temperature field. The predicted in-tank temperature field is then subjected to quantitative deviation analysis to determine whether adjustments to the fermenter temperature control parameters are necessary. If dynamic parameter adjustment is triggered, a dynamic adjustment phase is initiated. The predicted in-tank temperature field at the end of the lag period is taken as the current state, and the standard temperature field as the target state. A multivariable PID controller is constructed, using an incremental PID algorithm combined with a diagonal matrix decoupling method to iteratively calculate the adjustment values ​​of each control parameter and adjust the fermenter temperature control parameters accordingly. This invention achieves precise, stable, and efficient control of the fermentation temperature, improving the quality consistency and production efficiency of *Prunus pedatus* probiotic fermentation products.

Description

Technical Field

[0001] This invention belongs to the technical field of microbial detection and analysis instruments and other systems, specifically a dynamic process control method for the fermentation process of Prunus pubescens probiotics. Background Technology

[0002] Five-finger peach is a plant with important medicinal value. Its probiotic fermentation products are rich in polysaccharides, flavonoids, coumarins, and other active ingredients, showing broad application prospects in functional foods and pharmaceuticals. During the fermentation process of five-finger peach probiotics, temperature is one of the key environmental parameters affecting bacterial growth, metabolism, and product accumulation. Too low a fermentation temperature will inhibit bacterial activity and prolong the fermentation cycle; too high a temperature may lead to bacterial death or metabolic pathway deviation, reducing the yield of the target product. Therefore, achieving precise and stable temperature control within the fermentation tank is a core technological requirement for ensuring the consistency of five-finger peach probiotic fermentation product quality and production efficiency.

[0003] The existing temperature control methods for the fermentation process of Prunus pubescens probiotics mainly have the following technical problems:

[0004] Existing multivariable temperature control methods often employ simple decoupling strategies or directly tune the PID parameters for each loop independently, failing to effectively solve the coupling problem. This can easily lead to conflicting control actions, system oscillations, and excessively long settling times, affecting temperature stability and control robustness.

[0005] Most existing control methods employ fixed-cycle control strategies, or simply wait a preset fixed time after adjustment before making the next judgment. This approach either results in insufficient waiting time, causing the system to be unstable before making the next adjustment, leading to overshoot and oscillation; or excessive waiting time reduces control efficiency and fails to respond quickly to sudden changes in metabolic heat production. Summary of the Invention

[0006] To overcome the shortcomings of existing technologies and solve at least one of the technical problems mentioned in the background, this invention proposes a dynamic process control method for the fermentation process of *Prunus pedatus* probiotics. The method aims to achieve precise, stable, and efficient control of fermentation temperature by constructing a three-dimensional temperature acquisition network, establishing an adaptive dynamic model of fermenter temperature, introducing a predictive-driven advance judgment mechanism, and employing a decoupled-incremental multivariate PID control strategy. This will improve the quality consistency and production efficiency of *Prunus pedatus* probiotic fermentation products.

[0007] The technical solution adopted by this invention to solve the technical problem is: a dynamic process control method for the fermentation process of *Prunus pedatus* probiotics, comprising:

[0008] Construct a temperature acquisition network to collect the actual temperature field inside the tank;

[0009] A model of internal temperature change is constructed, the temperature control parameters of the fermenter are obtained and imported into the model for simulation, the simulated internal temperature field is obtained, and the simulation accuracy is evaluated by comparing it with the actual internal temperature field. If the simulation accuracy is low, the model of internal temperature change is iteratively trained and updated by combining the actual internal temperature field.

[0010] Specifically, basic information about the fermenter is collected, and based on physical and chemical laws, energy conservation equations and gas-solid two-phase flow equations are constructed by combining the basic information of the fermenter. A model of temperature change inside the tank is built by integrating these equations. The temperature control parameters of the fermenter are acquired in real time and imported into the constructed model of temperature change inside the tank. Through the simulation and calculation of the model of temperature change inside the tank, the simulated temperature data in the simulated temperature field inside the tank is obtained.

[0011] At the end of each evaluation period, for each measurement point, the simulated temperature data of the measurement point in all simulated tank temperature fields during the evaluation period are obtained, and integrated into a simulated temperature sequence according to the time sequence to obtain the simulated temperature sequence of each measurement point.

[0012] The temperature data of the measurement points in all actual tank temperature fields during the evaluation period are obtained, integrated into an actual temperature sequence according to the time sequence, and the actual temperature sequence is detected and replaced by linear interpolation to obtain the actual temperature sequence of each measurement point.

[0013] The root mean square error of the actual temperature field inside the tank and the simulated temperature field inside the tank at each measurement point is obtained and calculated. The root mean square error during the evaluation period is processed to obtain the spatial dimension temperature error. The Euclidean distance between the actual temperature sequence and the simulated temperature sequence at each measurement point is obtained and processed to obtain the temporal dimension temperature error.

[0014] The spatial and temporal temperature errors are integrated to obtain the simulation error of the tank temperature change model. If the simulation error is greater than the error threshold, the simulation accuracy of the tank temperature change model is low. The actual temperature sequence of all measurement points during the evaluation period and the corresponding fermenter temperature control parameters are obtained. The parameters in the tank temperature change model are optimized and adjusted by combining the particle swarm optimization algorithm. Through iterative training, the simulation error of the tank temperature change model is reduced, and the updated tank temperature change model is obtained.

[0015] A tank temperature change model is used to predict the tank temperature field distribution during the prediction period, and the predicted tank temperature field is obtained. The predicted tank temperature field is then subjected to quantitative deviation analysis to determine whether the fermenter temperature control parameters need to be adjusted. If so, the parameters are dynamically adjusted.

[0016] Specifically, the prediction period is defined as a continuous future time interval, starting from the current data collection time and changing over time;

[0017] In the tank temperature change model, the fermentation tank temperature control parameters at the current acquisition time are input, the actual tank temperature field at the current acquisition time is obtained as the initial temperature field, the energy conservation equation and gas-solid two-phase flow equation in the tank temperature change model are solved, the finite difference method is used for numerical solution, the predicted temperature data of each spatial point at each acquisition time in the prediction period are obtained, and the predicted tank temperature field in the prediction period is generated by integrating them.

[0018] The predicted temperature field inside the tank and the standard temperature field within the prediction period are obtained and compared. In the spatial dimension, the spatial standard deviation between the predicted temperature field inside the tank and the standard temperature field is calculated. In the temporal dimension, the temporal fluctuation deviation within the prediction period is calculated.

[0019] If the spatial standard deviation is greater than the standard deviation threshold, or the time fluctuation deviation is greater than the fluctuation deviation threshold, it is determined that the temperature control parameters of the fermenter need to be adjusted.

[0020] If dynamic parameter adjustment is triggered, the dynamic adjustment phase is started. The predicted internal temperature field at the end of the hysteresis period is obtained as the current state, and the standard temperature field is obtained as the target state. A multivariable PID controller is constructed, and an incremental PID algorithm is adopted in combination with the diagonal matrix decoupling method. The adjustment values ​​of each control parameter are iteratively calculated and the temperature control parameters of the fermenter are adjusted.

[0021] Specifically, the method for adjusting the temperature control parameters of the fermenter is as follows:

[0022] The system obtains the adjustment values ​​of each control parameter at the current iteration step and sends them to the actuator to adjust the temperature control parameters of each fermenter. During the lag period starting from the iteration step, the system obtains the flag value. If the flag value is equal to the stable flag value, the system stops the lag period and predicts the temperature field inside the predicted tank during the prediction period. The system also calculates the spatial standard deviation and temporal fluctuation deviation during the prediction period to determine whether the temperature control parameters of the fermenter need to be adjusted.

[0023] If not needed, or if the control parameter adjustment values ​​calculated in three consecutive iterations are all less than the adjustment threshold, stop the iteration and end the dynamic adjustment phase; otherwise, set the flag value to 0 and proceed to the next iteration.

[0024] The flag value is obtained as follows:

[0025] Set a flag value and initialize it to 0. During the hysteresis period starting from the current iteration step, continuously monitor the temperature deviation between the actual temperature field inside the tank and the standard temperature field at adjacent acquisition times during the hysteresis period, and calculate the absolute difference between the temperature deviations at adjacent acquisition times. If the absolute difference is less than the absolute difference threshold, the flag value is incremented once. If it is greater than or equal to the absolute difference threshold, the flag value is set to 0.

[0026] Furthermore, the adjustment values ​​of each control parameter are obtained as follows:

[0027] Obtain the predicted tank temperature field at the end of the lag period starting from the current iteration step, as the current state before adjusting the fermenter temperature control parameters, and obtain the standard temperature field as the target state.

[0028] A multivariable PID controller is constructed. The current state and the target state are input into the multivariable PID controller. The diagonal matrix decoupling method is used to decompose the multivariable PID controller into independent single-loop control. An incremental PID algorithm is used to discretize the control parameter adjustment into iterative calculations under several hysteresis periods. The starting point of each hysteresis period is taken as an iteration step, and the adjustment values ​​of each control parameter at the current iteration step are calculated.

[0029] The beneficial effects of this invention are as follows:

[0030] 1. This invention constructs a comprehensive temperature acquisition network to accurately collect the actual temperature field inside the tank, and builds and iteratively updates the temperature change model inside the tank based on this network. This greatly improves the accuracy of understanding the temperature distribution inside the tank. Based on this model, the temperature field distribution inside the tank during the prediction period is predicted and the deviation is quantified. The temperature change trend can be predicted in advance. Once the simulation accuracy is found to be insufficient, the model is iterated in time to ensure the reliability of the prediction results. This provides a scientific basis for adjusting the temperature control parameters of the fermenter and effectively improves the accuracy and stability of temperature control.

[0031] 2. After determining that the temperature control parameters of the fermenter need to be adjusted, this invention enters the dynamic adjustment stage. A multivariable PID controller is constructed with the predicted internal temperature field at the end of the hysteresis period as the current state and the standard temperature field as the target state. The incremental PID algorithm combined with the diagonal matrix decoupling method is used to calculate the adjustment values ​​of each control parameter. This method fully considers the multivariable, strong coupling and hysteresis characteristics of the fermenter temperature control system, and can quickly and accurately adjust the control parameters so that the fermenter temperature can be quickly stabilized within the target range, thereby improving cement production efficiency and product quality.

[0032] More detailed explanation:

[0033] This invention simultaneously calculates the spatial dimension temperature error (arithmetic mean of root mean square error) and the temporal dimension temperature error (arithmetic mean of Euclidean distance), and integrates them into the simulation error. When the simulation error exceeds a threshold, a particle swarm optimization algorithm is used to iteratively train and update the model parameters. This ensures that the model's temperature prediction ability is consistent across different axial heights and radial positions within the fermenter, avoiding the masking of local "hot spots" or "cold spots" by averaging. It also ensures the model's ability to fit the dynamic temperature change trend during fermentation, preventing the model from only matching at a certain moment while the overall trend deviates. The model can simultaneously satisfy the constraints of "accurate spatial distribution" and "accurate temporal evolution," significantly improving the model's extrapolation prediction ability under varying operating conditions and reducing the risk of model failure due to changes in the metabolic state of the microorganisms.

[0034] This invention employs an updated model to predict the tank temperature field within a given timeframe, calculating the spatial standard deviation and temporal fluctuation deviation. If these deviations exceed a threshold, dynamic parameter adjustments are triggered. Before the actual temperature deviates from the standard field, the invention identifies future deviation trends through prediction, enabling pre-regulation and shortening the duration of temperature anomalies. This proactive control allows adjustments to be made when deviations are small, avoiding subsequent large-scale, high-frequency adjustments and extending the lifespan of actuators (such as valves and heaters). The metabolic heat production rates of *Ficus hirta* probiotics differ significantly across different growth stages (lag phase, logarithmic phase, and stationary phase). The predictive mechanism can respond in advance to sudden changes in heat production, preventing temperature overshoot or drop.

[0035] This invention employs a diagonal matrix decoupling method to decompose a multivariable PID controller into independent single-loop controls. There is inherent coupling between cooling water flow rate, heating power, and stirring speed (e.g., increasing stirring speed enhances the heat transfer coefficient and simultaneously increases heat generation). After decoupling, each single-loop PID responds only to its own temperature deviation, avoiding the oscillating cycle of "increased cooling, decreased temperature, increased heating, increased temperature, and increased cooling again." The decoupled single-loop PIDs can be tuned individually according to the single-variable system parameters, reducing the difficulty of on-site debugging and facilitating engineering application. Because coupling delay is eliminated, the response time of each control channel is shortened, reducing the overall system settling time.

[0036] The control parameter adjustment is discretized into iterative calculations over several hysteresis periods, with each hysteresis period starting as one iteration step, using an incremental PID algorithm. Incremental PID only outputs the change in the control quantity, not the absolute value. When the actuator reaches its limit, the integral term will not accumulate, preventing deep saturation. The total adjustment is output step-by-step, with the system response stabilizing within each hysteresis period before the next adjustment is made, avoiding over-adjustment due to hysteresis. The output change of incremental PID is step-like, avoiding the potential impact of a single large adjustment on the fermentation broth (especially for shear-sensitive probiotics).

[0037] This invention continuously monitors the absolute difference in actual temperature deviation between adjacent acquisition times during the hysteresis period. If the difference is less than a threshold multiple times consecutively, a flag is incremented, and the hysteresis period ends early after reaching a stable flag value. The system can adaptively adjust the control cycle according to the actual temperature change rate, iterating rapidly when changes are fast and converging early when changes are slow, avoiding resource waste caused by waiting at a fixed cycle. When the temperature has stabilized, the current adjustment sequence is stopped in a timely manner to prevent unnecessary fine-tuning from being repeatedly triggered due to residual deviations, reducing the number of actions of the actuators. Frequent start-stop or adjustment of actuators such as cooling water pumps, heaters, and stirring motors consumes additional energy. The flag value mechanism enables the system to exit dynamic adjustment as soon as it reaches stability, reducing energy consumption.

[0038] If the control parameter adjustment values ​​calculated in three consecutive iterations are all less than the adjustment threshold, the iteration stops and the dynamic adjustment phase ends. When the system approaches steady state, the control quantity adjustment value will be very small. Three consecutive judgments can filter out false judgments caused by single noise, ensuring that the adjustment only stops after the system has truly entered steady state. This avoids continuing meaningless iterative calculations in steady state, reduces the load on the controller (PLC or industrial computer), and facilitates coordinated operation with other control tasks (such as pH, dissolved oxygen). Attached Figure Description

[0039] The invention will now be further described with reference to the accompanying drawings.

[0040] Figure 1 This is a flowchart illustrating the steps of a dynamic process control method for the fermentation process of *Prunus pedatus* probiotics according to an embodiment of the present invention. Detailed Implementation

[0041] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0042] Example 1

[0043] Please see Figure 1 As shown in the embodiment of the present invention, a dynamic process control method for the fermentation process of *Prunus pedatus* probiotics includes the following steps:

[0044] S1: Construct a temperature acquisition network to collect the actual temperature field inside the tank;

[0045] Several temperature sensors are evenly arranged along the axial and radial directions on the inner wall of the fermenter to form a temperature acquisition network. The temperature sensors are high-precision thermocouples, and their placement is determined according to the structure of the fermenter to ensure that the temperature acquisition network can cover areas at different heights and radial positions inside the decomposition tank, comprehensively capturing the temperature information inside the tank. Each temperature sensor in the temperature acquisition network is marked as a measurement point.

[0046] The period during which the temperature inside the decomposition tank is continuously collected through the temperature acquisition network is marked as a temperature analysis period. The sampling frequency is set and initialized at the beginning of the temperature analysis period. According to the set sampling frequency, the temperature data inside the decomposition tank is collected in real time through the temperature acquisition network to obtain the temperature data at each measurement point. The temperature data at each measurement point is integrated to obtain the discrete temperature data inside the decomposition tank.

[0047] Each moment when the temperature acquisition network collects temperature data inside the decomposition tank according to the sampling frequency is marked as the acquisition moment;

[0048] For each measurement point Assigning three-dimensional spatial coordinates ,in , Let be the planar coordinates on the circumference of the fermenter wall. Let i represent the axial height coordinate of the fermenter, and i denote the measurement point. The number;

[0049] Data processing is performed on the collected discrete temperature data to generate a continuous actual internal temperature field. Specifically, based on the discrete temperature data and the three-dimensional spatial coordinates of the corresponding measurement points in the fermentation tank, cubic spline interpolation is used to calculate the temperature value at each three-dimensional spatial coordinate in the internal space of the tank. The calculated temperature value and the discrete temperature data are then integrated according to the position coordinates to generate the actual internal temperature field.

[0050] It should be noted that the purpose of this step is to obtain comprehensive and accurate temperature information inside the decomposition tank, ensuring the integrity and continuity of temperature data in space and time, providing a reliable basis for adjusting temperature control parameters, avoiding control deviations caused by missing or inaccurate data, and improving the accuracy and reliability of temperature monitoring.

[0051] S2: Construct a dynamic model of the internal temperature of the tank, obtain the temperature control parameters of the fermenter and import them into the internal temperature change model for simulation, obtain the simulated internal temperature field, compare it with the actual internal temperature field of the tank to evaluate the simulation accuracy, and if the simulation accuracy is low, combine the actual internal temperature field of the tank to iteratively train and update the internal temperature change model.

[0052] Collect basic information about the fermenter, including equipment structural parameters, material characteristic parameters, and operating condition parameters.

[0053] The equipment structural parameters include the nominal volume of the fermenter, height-to-diameter ratio, type and number of stirring paddles, number and size of baffles, and heat exchange area and structure of the jacket (or coil); the material characteristic parameters include the specific heat capacity, density, viscosity, and thermal conductivity of the fermentation medium, as well as the cell growth kinetics, product generation kinetics, and metabolic heat production coefficient of the Five-Finger Peach probiotic fermentation process; and the operating condition parameters include the stirring speed range, cooling water temperature and flow rate range, and heating medium parameters.

[0054] Based on the principles of bioreaction engineering and heat transfer, and combined with the basic information of the fermenter, a heat balance equation and a liquid-phase mixing heat transfer equation for the fermenter are constructed, and an integrated dynamic temperature model of the fermenter is obtained. This model uses the fermenter temperature control parameters as input to establish the relationship between the control parameters and the temperature field distribution inside the tank. In the model, heat accumulation = microbial metabolic heat production + stirring heat production - jacket / coil heat transfer - exhaust sensible heat - tank wall heat dissipation. The equations are numerically solved using the finite difference method or a reduced-order computational fluid dynamics (CFD) model to obtain and output the simulated temperature field inside the tank.

[0055] The fermenter temperature control parameters include: jacket cooling water flow rate (or cooling water valve opening), heating power (or steam valve opening), and stirring speed.

[0056] Based on the same sampling frequency as the temperature acquisition network, the fermenter temperature control parameters are acquired in real time during the temperature analysis period. The fermenter temperature control parameters are then imported into the constructed tank temperature change model. Through model simulation and calculation, simulated temperature data under the current fermenter temperature control parameters are obtained. All simulated temperature data are then integrated to obtain the simulated tank temperature field.

[0057] The simulated tank temperature field obtained by the tank temperature change model is compared and analyzed with the actual tank temperature field collected to evaluate the simulation accuracy of the tank temperature change model.

[0058] Specifically, based on any temperature analysis period, an evaluation period is set within the temperature analysis period. The interval between adjacent evaluation moments is an evaluation period. The duration of evaluation periods within the same temperature analysis period is the same. The starting point of the temperature analysis period is the first evaluation moment within the temperature analysis period.

[0059] At the end of each evaluation period, the actual and simulated internal temperature fields of the tank are obtained at each acquisition time within the evaluation period, based on any measurement point. The temperature data of the measurement points in all actual tank temperature fields during the evaluation period are obtained and integrated into an actual temperature sequence according to the time series. The system acquires simulated temperature data from measurement points within all simulated tanks during the evaluation period, and integrates these data into a simulated temperature sequence based on the time series. ;

[0060] Outlier detection and processing are performed on the actual temperature series. The principle is to calculate the mean of the temperature data at the measurement point within the actual temperature series. and standard deviation If the temperature data at a certain acquisition time t satisfy:

[0061] ;

[0062] Then determine the temperature data at time t. For outliers, use temperature data from adjacent sampling times within the actual temperature sequence to perform linear interpolation to replace the outliers;

[0063] The temperature data of each measurement point at each acquisition time during the evaluation period is traversed, outliers are filtered, processed and replaced, the actual temperature sequence of each measurement point is updated, noise interference is eliminated, and the corresponding temperature data is replaced in the actual tank temperature field at the corresponding acquisition time to update the actual tank temperature field.

[0064] The simulated tank temperature field and the actual tank temperature field were compared in both spatial and temporal dimensions during the evaluation period.

[0065] In the spatial dimension, based on the updated actual tank temperature field, at the same acquisition time, for each measurement point in the decomposed tank, the root mean square error between the temperature data of the measurement point in the actual tank temperature field and the simulated tank temperature field and the simulated temperature data is calculated, and the root mean square error calculated at all acquisition times is arithmetically averaged to obtain the spatial dimension temperature error.

[0066] In the time dimension, based on the updated actual temperature sequence, for the same measurement point, the Euclidean distance between the actual temperature sequence and the simulated temperature sequence of the measurement point is calculated, and the Euclidean distance calculated for each measurement point is arithmetically averaged to obtain the time dimension temperature error.

[0067] The error analysis of the simulated tank temperature field and the actual tank temperature field in the spatial and temporal dimensions during the evaluation period is integrated. The obtained spatial and temporal temperature errors are then averaged to obtain the simulation error of the tank temperature change model.

[0068] The simulation error of the obtained tank temperature change model is compared with the preset error threshold.

[0069] If the simulation error of the tank temperature change model is less than or equal to the error threshold, the simulation accuracy of the tank temperature change model is considered to be high.

[0070] If the simulation error of the internal temperature change model is greater than the error threshold, it is determined that the simulation accuracy of the internal temperature change model is low. Combining the actual temperature sequence of all measurement points during the evaluation period and the corresponding fermenter temperature control parameters, the particle swarm optimization algorithm is used to optimize and adjust the parameters in the internal temperature change model. Through iterative training, the simulation error of the internal temperature change model is reduced, and an updated internal temperature change model is obtained.

[0071] It should be noted that the purpose of this step is to establish the relationship between the temperature control parameters of the fermenter and the temperature field distribution. Through simulation and optimization of the temperature change model, it can more accurately reflect the temperature change law in the decomposition tank, improve the adaptability of the temperature change model to actual working conditions, provide effective theoretical model support for temperature prediction and parameter adjustment, and reduce the uncertainty in actual production.

[0072] S3: Using the tank temperature change model, the tank temperature field distribution during the prediction period is predicted to obtain the predicted tank temperature field. The predicted tank temperature field is then subjected to quantitative deviation analysis to determine whether the fermenter temperature control parameters need to be adjusted. If so, dynamic parameter adjustment is triggered.

[0073] The obtained tank temperature change model is used to predict the tank temperature field distribution during the prediction period;

[0074] Specifically, the prediction period is defined as a continuous future time interval. ,in This is the current data collection time. To predict the duration, the starting point of the prediction period is synchronized with the data collection time of the temperature analysis period to ensure that the temperature data in the time series are aligned and the prediction period changes over time.

[0075] Based on the obtained tank temperature change model, the fermenter temperature control parameters at the current acquisition time are input. The fermenter temperature control parameters include cooling water flow rate, heating power, and stirring speed. The actual tank temperature field at the current acquisition time is obtained as the initial temperature field. The initial temperature field represents the initial conditions at the start of the prediction period.

[0076] After inputting the fermenter temperature control parameters and initial temperature field at the current acquisition time into the in-tank temperature change model, the microbial metabolic heat production rate at the current time (which can be estimated from the online respiratory quotient or offline specific growth rate) needs to be input when solving the model. The fourth-order Runge-Kutta method is used for time progression to obtain the predicted temperature data of each spatial point at each acquisition time within the prediction period, and the data are integrated to generate the predicted in-tank temperature field. ;

[0077] Obtain a standard temperature field. The standard temperature field should be set according to the optimal growth temperature range of Prunus cerevisiae probiotics (e.g., 30-32℃) and the requirement to avoid temperature fluctuations affecting the accumulation of metabolites. It represents the ideal temperature distribution inside the fermenter.

[0078] Based on the predicted tank temperature field at each collection time within the prediction period, and compared with the standard temperature field, deviation analysis is performed in the spatial and temporal dimensions.

[0079] In the spatial dimension, for the predicted temperature field inside the tank and the standard temperature field, the predicted temperature data of each measurement point in the predicted temperature field inside the tank at each acquisition time t are obtained, using the formula:

[0080] ;

[0081] The spatial standard deviation within the prediction period was calculated. Where n is the total number of measurement points, and i represents the number of measurement points. The number, For measurement points Predicted temperature data at acquisition time t For measurement points Standard temperature data in a standard temperature field;

[0082] In terms of time dimension, calculate the time fluctuation deviation within the prediction period. The formula is:

[0083] ;

[0084] It should be noted that the spatial standard deviation reflects the degree of deviation of the predicted temperature field inside the tank from the standard temperature field in terms of spatial distribution, while the temporal fluctuation deviation measures the stability of the predicted temperature data over time.

[0085] If the spatial standard deviation within the prediction period is greater than the standard deviation threshold, or the temporal fluctuation deviation within the prediction period is greater than the fluctuation deviation threshold, it is determined that the spatial distribution between the predicted temperature field inside the tank and the standard temperature field is too large or the temperature fluctuation exceeds the limit within the prediction period. It is determined that the temperature control parameters of the fermenter need to be adjusted, triggering dynamic parameter adjustment.

[0086] Conversely, if the fermenter temperature control parameters do not need to be adjusted, the current fermenter temperature control parameters will be maintained, and prediction and monitoring will continue.

[0087] It should be noted that the purpose of this step is to predict the temperature change trend inside the tank in advance, detect the deviation of the temperature field from the standard state in a timely manner, and provide a clear decision basis for adjusting the control parameters by quantifying the deviation index, so as to make the temperature control forward-looking and avoid the impact of abnormal temperature on the quality and efficiency of the fermentation process of Five-Finger Peach Probiotics.

[0088] S4: If dynamic parameter adjustment is triggered, the dynamic adjustment phase is started. The predicted tank temperature field at the end of the hysteresis period is obtained as the current state, and the standard temperature field is obtained as the target state. A multivariable PID controller is constructed. The incremental PID algorithm is used in combination with the diagonal matrix decoupling method to iteratively calculate the adjustment value of each control parameter and adjust the fermenter temperature control parameters.

[0089] The specific steps for adjusting the temperature control parameters of the fermenter are as follows;

[0090] If dynamic parameter adjustment is triggered, the dynamic adjustment phase is started. During the dynamic adjustment phase, the predicted tank temperature field at the end of the lag period starting from the current acquisition time is obtained as the current state before adjusting the fermenter temperature control parameters, and the standard temperature field is obtained as the target state.

[0091] The hysteresis period refers to the time elapsed from the start of the actuator's action (such as adjusting the cooling water valve or changing the heating power) until the temperature sensor inside the fermenter detects a stable temperature change. This period is mainly determined by the response time of the temperature sensor, the mixing time of the liquid inside the fermenter, and the heat transfer hysteresis of the jacket / coil.

[0092] A multivariable PID controller is constructed, with the current state and target state input into it. Due to coupling between cooling water flow rate, heating power, and stirring speed (e.g., increasing stirring speed enhances the heat transfer coefficient and increases heat generation; cooling water flow rate and heating power directly compete), a diagonal matrix decoupling method is used to decompose the multivariable PID controller into independent single-loop control. An incremental PID algorithm is employed, discretizing the control parameter adjustment into iterative calculations over several hysteresis periods. The starting point of each hysteresis period is considered as the iteration step. The calculation formula is as follows:

[0093] ;

[0094] Where k represents the total number of iterations up to the current time, and m represents the number of iterations. j represents the index of the fermenter temperature control parameter. When j=1, it represents the cooling water flow rate adjustment; when j=2, it represents the heating power adjustment; and when j=3, it represents the stirring speed adjustment. Represents the proportionality coefficient. Represents the integral coefficient. Denotes the differential coefficient. This represents the adjustment amount of the fermenter temperature control parameter at index j at iteration step k. This represents the change in temperature deviation between the iteration number k and the previous iteration number;

[0095] ;

[0096] in, This represents the temperature deviation between the target state and the temperature at iteration step k.

[0097] ;

[0098] in, This represents the time corresponding to iteration step k. Indicates the duration of the previous lag period;

[0099] A detailed explanation of the coupling relationship: In this control scheme, the coupling manifests as follows: an increase in cooling water flow will lower the temperature, which may lead the heating system to increase its output to maintain the temperature; an increase in stirring speed will both increase heat generation and improve heat dissipation efficiency. Through a decoupling matrix, this mutual influence is compensated for in the feedforward manner, allowing each single-loop PID to respond independently to temperature deviations.

[0100] At each iteration step, the calculated control parameter adjustment value is sent to the actuator to adjust the temperature control parameters of each fermenter. During the hysteresis period starting from the time corresponding to the iteration step, the temperature deviation between the actual temperature field inside the tank and the standard temperature field at adjacent acquisition times is continuously monitored, and the absolute difference between the temperature deviations at adjacent acquisition times is calculated. A flag value is set and initialized to 0. If the absolute difference between the temperature deviations at adjacent acquisition times is less than the absolute difference threshold, the flag value is incremented. If the absolute difference between the temperature deviations at adjacent acquisition times is greater than or equal to the absolute difference threshold, the flag value is set to 0. The absolute difference threshold is set as follows: within two consecutive sampling periods, the rate of change of the tank temperature is lower than the minimum value allowed by the sensor accuracy and mixed noise.

[0101] If the flag value equals the stable flag value, stop the hysteresis period and predict the predicted tank temperature field at each collection time within the prediction period. Based on the predicted tank temperature field at each collection time within the prediction period, compare it with the standard temperature field, perform deviation analysis in the spatial and temporal dimensions, calculate the spatial standard deviation and temporal fluctuation deviation within the prediction period, and determine whether the fermenter temperature control parameters need to be adjusted based on the prediction results.

[0102] If not needed, or if the control parameter adjustment values ​​calculated in three consecutive iterations are all less than the adjustment threshold, stop the iteration and end the dynamic adjustment phase; otherwise, set the flag value to 0 and proceed to the next iteration.

[0103] It should be noted that the purpose of this step is to achieve dynamic and precise adjustment of the temperature control parameters of the fermenter, effectively overcome the system lag and parameter coupling problems, and through continuous predictive analysis and closed-loop control, enable the temperature inside the tank to quickly and stably approach the standard temperature field, improve the accuracy and stability of temperature control during the fermentation of Prunus cerasifera probiotics, ensure the smooth progress of the production process, and improve product quality and production efficiency.

[0104] The technical solution of this invention is as follows: A temperature acquisition network is constructed to collect the actual internal temperature field of the tank. An internal temperature change model is constructed, and the fermentation tank temperature control parameters are imported into the internal temperature change model for simulation. The simulated internal temperature field is obtained, and the simulation accuracy is evaluated by comparing it with the actual internal temperature field. If the simulation accuracy is low, the internal temperature change model is iteratively trained and updated using the actual internal temperature field. The internal temperature change model is used to predict the internal temperature field distribution during the prediction period, resulting in a predicted internal temperature field. The predicted internal temperature field is then subjected to quantitative deviation analysis to determine whether the fermentation tank temperature control parameters need to be adjusted. If so, dynamic parameter adjustment is triggered. If dynamic parameter adjustment is triggered, the dynamic adjustment phase is initiated. The predicted internal temperature field at the end of the lag period is taken as the current state, and the standard temperature field is taken as the target state. A multivariable PID controller is constructed, and an incremental PID algorithm combined with a diagonal matrix decoupling method is used to calculate the adjustment values ​​of each control parameter and adjust the fermentation tank temperature control parameters.

[0105] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A dynamic process control method for the fermentation process of *Prunus armeniaca* probiotics, characterized in that: include: Construct a temperature acquisition network to collect the actual temperature field of the fermenter; A fermenter temperature change model is constructed, the fermenter temperature control parameters are obtained and imported into the fermenter temperature change model for simulation, the simulated internal temperature field is obtained, and the simulation accuracy is evaluated by comparing it with the actual internal temperature field. If the simulation accuracy is low, the internal temperature change model is iteratively trained and updated by combining the actual internal temperature field. A tank temperature change model is used to predict the tank temperature field distribution during the prediction period, and the predicted tank temperature field is obtained. The predicted tank temperature field is then subjected to quantitative deviation analysis to determine whether the fermenter temperature control parameters need to be adjusted. If so, the parameters are dynamically adjusted. If dynamic parameter adjustment is triggered, the dynamic adjustment phase is initiated. The predicted internal temperature field at the end of the hysteresis period is obtained as the current state, and the standard temperature field is obtained as the target state. A multivariable PID controller is constructed, and an incremental PID algorithm is adopted in combination with the diagonal matrix decoupling method. The adjustment values ​​of each control parameter are iteratively calculated and the temperature control parameters of the fermenter are adjusted.

2. The dynamic process control method for the fermentation process of *Prunus armeniaca* probiotics according to claim 1, characterized in that: The method for evaluating simulation accuracy is as follows: The root mean square error of the actual temperature field inside the tank and the simulated temperature field inside the tank at each measurement point is obtained and calculated. The root mean square error during the evaluation period is processed to obtain the spatial dimension temperature error. The Euclidean distance between the actual temperature sequence and the simulated temperature sequence at each measurement point is obtained and processed to obtain the temporal dimension temperature error. By integrating the spatial and temporal temperature errors, the simulation error of the tank internal temperature change model is obtained. If the simulation error is greater than the error threshold, the simulation accuracy of the tank internal temperature change model is considered low.

3. The dynamic process control method for the fermentation process of *Prunus pedunculata* probiotics according to claim 2, characterized in that: The actual temperature sequence and the simulated temperature sequence are obtained as follows: For each measurement point, the simulated temperature data of the measurement point in all simulated tank temperature fields during the evaluation period are obtained, and integrated into a simulated temperature sequence according to the time sequence to obtain the simulated temperature sequence of each measurement point; Temperature data of measurement points in all actual tank temperature fields during the evaluation period are obtained, integrated into actual temperature sequences according to time sequence, and outlier detection and replacement are performed on the actual temperature sequences using linear interpolation to obtain the actual temperature sequence of each measurement point.

4. The dynamic process control method for the fermentation process of *Prunus armeniaca* probiotics according to claim 3, characterized in that: The simulated temperature data is acquired in the following way: Basic information about the fermenter is collected, and energy conservation equations and gas-solid two-phase flow equations are constructed based on this information. A model of temperature change inside the fermenter is then built by integrating these equations. Real-time temperature control parameters of the fermenter are imported into the constructed model of temperature change inside the fermenter. Simulated temperature data in the simulated temperature field inside the fermenter is obtained through simulation and calculation using the model of temperature change inside the fermenter.

5. The dynamic process control method for the fermentation process of *Prunus pedunculata* probiotics according to claim 2, characterized in that: The method for iteratively training and updating the tank temperature change model is as follows: The actual temperature sequence of all measurement points during the evaluation period and the corresponding fermenter temperature control parameters are obtained. The parameters in the tank temperature change model are optimized and adjusted by combining the particle swarm optimization algorithm. Through iterative training, the simulation error of the tank temperature change model is reduced, and the updated tank temperature change model is obtained.

6. The dynamic process control method for the fermentation process of *Prunus armeniaca* probiotics according to claim 1, characterized in that: The method for determining whether the fermenter temperature control parameters need to be adjusted is as follows: The predicted temperature field inside the tank and the standard temperature field within the prediction period are obtained and compared. In the spatial dimension, the spatial standard deviation between the predicted temperature field inside the tank and the standard temperature field is calculated. In the temporal dimension, the temporal fluctuation deviation within the prediction period is calculated. If the spatial standard deviation is greater than the standard deviation threshold, or the time fluctuation deviation is greater than the fluctuation deviation threshold, it is determined that the temperature control parameters of the fermenter need to be adjusted.

7. The dynamic process control method for the fermentation process of *Prunus armeniaca* probiotics according to claim 6, characterized in that: The method for obtaining the predicted temperature field inside the tank is as follows: The prediction period is defined as a continuous future time interval, starting from the current data collection time and changing over time. In the tank temperature change model, the fermenter temperature control parameters at the current acquisition time are input, the actual tank temperature field at the current acquisition time is obtained as the initial temperature field, the energy conservation equation and the gas-solid two-phase flow equation in the tank temperature change model are solved, and the finite difference method is used for numerical solution to obtain the predicted temperature data of each spatial point at each acquisition time in the prediction period, and the data are integrated to generate the predicted tank temperature field in the prediction period.

8. The dynamic process control method for the fermentation process of *Prunus armeniaca* probiotics according to claim 1, characterized in that: The method for adjusting the temperature control parameters of the fermenter is as follows: An incremental PID algorithm is used to iterate the temperature control parameters of the fermenter, obtain the adjustment values ​​of each control parameter at the current iteration step, and send them to the actuator to adjust the temperature control parameters of each fermenter. During the lag period starting from the iteration step, a flag value is obtained. If the flag value is equal to the stable flag value, the lag period is stopped and the predicted temperature field inside the tank during the prediction period is predicted. The spatial standard deviation and time fluctuation deviation during the prediction period are calculated to determine whether the temperature control parameters of the fermenter need to be adjusted. If not needed, or if the control parameter adjustment values ​​calculated in three consecutive iterations are all less than the adjustment threshold, stop the iteration and end the dynamic adjustment phase; otherwise, set the flag value to 0 and proceed to the next iteration.

9. The dynamic process control method for the fermentation process of *Prunus armeniaca* probiotics according to claim 8, characterized in that: The method for obtaining the adjustment values ​​of each control parameter is as follows: Obtain the predicted tank temperature field at the end of the lag period starting from the current iteration step, as the current state before adjusting the fermenter temperature control parameters, and obtain the standard temperature field as the target state. A multivariable PID controller is constructed. The current state and the target state are input into the multivariable PID controller. The diagonal matrix decoupling method is used to decompose the multivariable PID controller into independent single-loop control. An incremental PID algorithm is used to discretize the control parameter adjustment into iterative calculations under several hysteresis periods. The starting point of each hysteresis period is taken as an iteration step, and the adjustment values ​​of each control parameter at the current iteration step are calculated.

10. The dynamic process control method for the fermentation process of *Prunus armeniaca* probiotics according to claim 8, characterized in that: The flag value is obtained in the following way: Set a flag value and initialize it to 0. During the hysteresis period starting from the current iteration step, continuously monitor the temperature deviation between the actual temperature field inside the tank and the standard temperature field at adjacent acquisition times within the hysteresis period, and calculate the absolute difference between the temperature deviations at adjacent acquisition times. If the absolute difference is less than the absolute difference threshold, the flag value is incremented once. If it is greater than or equal to the absolute difference threshold, the flag value is set to 0.

Images