Continuous extraction self-adaptive optimization method for soybean protein based on membrane separation coupling enzymolysis

Abstract

The application discloses a continuous extraction self-adaptive optimization method for soybean protein based on membrane separation coupling enzymolysis, relates to the field of food engineering and biological separation technology, and the method constructs an integrated continuous system of enzymolysis-membrane separation, deploys a multi-parameter online sensing network to monitor key variables such as peptide concentration, turbidity and membrane flux in real time, and automatically adjusts pH, temperature, enzyme addition rate and membrane operation parameters based on a dynamic feedback control model; when the membrane flux attenuation or peptide generation rate is detected, the cross-flow speed and reaction conditions are automatically optimized to maintain efficient and stable operation, and the solvent is recycled in a closed loop through nanofiltration-reverse osmosis. The application aims to solve the problems of low extraction efficiency, unstable product quality and serious membrane pollution in traditional soybean protein extraction, significantly improve the soybean peptide yield and the proportion of target molecular weight segment, reduce energy consumption and water consumption, and enhance the system robustness and intelligent level.

Description

Technical Field

[0001] This invention relates to the fields of food engineering and bioseparation technology, and in particular to an adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis. Background Technology

[0002] With the continuous growth in demand for high-value utilization of plant protein, soybean protein extraction technology based on membrane separation and enzymatic hydrolysis has become a key technological path in the field of soybean deep processing due to its advantages such as high efficiency, environmental friendliness, and excellent functional properties of the resulting products. This technology releases active peptides in soybean protein through enzymatic hydrolysis and combines it with membrane separation to clarify, classify, and concentrate the product, thereby improving product purity and yield while preserving bioactivity. However, in current industrial practice, this type of coupled process still faces significant bottlenecks in terms of continuous operation capability, adaptive control mechanism of process parameters, and system-level extraction efficiency optimization, which seriously restricts its stability, economy, and scalability in large-scale production scenarios.

[0003] The continuous extraction process of soybean protein based on membrane separation coupled with enzymatic hydrolysis aims to achieve continuous operation throughout the entire process—from raw material feeding and enzymatic hydrolysis to product retention and solvent recovery—by integrating reaction and separation units. The core of this process lies in maintaining a dynamic balance between enzymatic hydrolysis kinetics and membrane mass transfer performance to ensure that target peptides are efficiently generated and separated immediately within the optimal reaction window, avoiding excessive hydrolysis or byproduct accumulation. However, existing technologies generally employ fixed parameter settings and batch-based operation logic, making it difficult to cope with dynamic changes in operating conditions such as fluctuations in raw material composition, enzyme activity decay, and membrane fouling evolution. This leads to decreased extraction efficiency, increased energy consumption, and product quality fluctuations during long-term system operation.

[0004] The existing process involves acid washing and defatting of soybean flour, Alcalase hydrolysis, and a three-stage membrane treatment process of microfiltration, ultrafiltration, and nanofiltration. While this method achieves functional enrichment of the product, the entire process is a discrete batch operation. There is a lack of real-time data interaction and parameter linkage between the enzymatic hydrolysis reactor and the membrane separation unit, making it impossible to dynamically adjust membrane operating conditions or enzymatic hydrolysis parameters according to the reaction progress. Furthermore, ceramic membranes are used for fractional concentration of soybean peptides, with a fixed concentration factor set as the operational endpoint. Although ceramic membranes have good fouling resistance, their process control relies entirely on preset thresholds, without the introduction of online sensors to monitor key state variables or feedback loops to drive adaptive parameter adjustment. This results in the system being unable to maintain optimal separation efficiency when the raw material protein content changes or when enzymatic hydrolysis is uneven, easily leading to problems such as a sudden drop in membrane flux, deviation of the target component retention rate, or unnecessary increases in energy consumption. Summary of the Invention

[0005] The purpose of this invention is to provide an adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis, in order to solve the problems of low extraction efficiency, unstable product quality and severe membrane fouling in traditional soybean protein extraction.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: An adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis includes: Step 1, constructing a continuous enzymatic hydrolysis-membrane separation coupling system: Soybean protein raw material slurry is fed into the enzymatic hydrolysis reactor through a continuous feed pump. The microfiltration membrane module and ultrafiltration membrane module are directly connected at the outlet of the enzymatic hydrolysis reactor to form a continuous process integrating reaction and separation. The enzymatic hydrolysis reactor is equipped with an online pH adjustment unit, a temperature control unit and an enzyme solution metering and dosing unit. The microfiltration membrane module and ultrafiltration membrane module are respectively equipped with independent transmembrane pressure differential control valves and cross-flow circulation pumps. Step 2, Deploy a multi-parameter online sensor network: Install turbidity sensors, peptide concentration optical sensors, conductivity sensors and membrane flux monitoring units inside the enzymatic hydrolysis reactor and at the inlet and outlet of the membrane module to collect key state variable data in the reaction system in real time and transmit the data to the central control unit; Step 3, establish a dynamic feedback control model: The central control unit, based on real-time collected state variable data and combined with the preset target peptide molecular weight distribution range and membrane fouling index threshold, dynamically adjusts the pH value, temperature, enzyme addition rate of the enzymatic hydrolysis reactor, as well as the transmembrane pressure difference, crossflow velocity and concentration factor of the membrane module through an adaptive algorithm. Step 4, perform adaptive parameter optimization: When the peptide concentration growth rate is detected to be lower than the set threshold or the membrane flux decay rate exceeds the preset upper limit, the central control unit automatically triggers the parameter recalibration program, prioritizes adjusting the crossflow rate to alleviate membrane fouling, and simultaneously fine-tunes the enzymatic hydrolysis temperature to maintain the optimal enzyme activity window. Step 5, achieving product classification and solvent recovery: The permeate from the ultrafiltration membrane module enters the nanofiltration unit for desalination and removal of small molecule impurities. The retentate is output as a high-purity soybean peptide product. The nanofiltration permeate is treated by reverse osmosis and then reused for raw material slurry preparation, forming a closed-loop water circulation system.

[0007] In step 1, the enzymatic hydrolysis reactor adopts a fully mixed-flow reactor structure with a predetermined effective volume. It is equipped with a double-layer jacket for precise temperature control, and the temperature control accuracy meets the process requirements. The stirring speed is adjustable within a predetermined range to ensure that the enzyme and substrate are in full contact and that shear force does not destroy the enzyme activity.

[0008] In step 1, the microfiltration membrane module uses a ceramic membrane tube module with a predetermined pore size, and the operating pressure is maintained within a preset range. The ultrafiltration membrane module uses a polyethersulfone hollow fiber membrane with a predetermined molecular weight cutoff, and the operating pressure is maintained within a preset range. Both membrane modules are equipped with independent backwashing programs, and the backwashing cycle is dynamically set according to the membrane flux decay rate.

[0009] In step 2, the peptide concentration optical sensor is based on the principle of ultraviolet absorption spectroscopy. It detects the absorbance of aromatic amino acid residues at a specific wavelength and converts it into peptide concentration value through a calibration curve. The measurement range covers the range required by the process, and the response time is less than the preset time limit. The turbidity sensor adopts the 90-degree scattering light detection method, and the measurement range covers the expected operating conditions. The accuracy meets the process monitoring requirements.

[0010] The dynamic feedback regulation model in step 3 includes three core control equations: one of which is the enzymatic hydrolysis kinetic equation. ,in Let be the peptide concentration at time t. For effective enzyme concentration, and The first is the kinetic constant; the second is the membrane flux decay model. ,in For the initial flux, The pollution resistance coefficient, The first is the membrane surface concentration; the second is the comprehensive optimization objective function. ,in For peptide yield, This is the flux attenuation. Energy consumption per unit of product , , These are the weighting coefficients, and their sum is 1.

[0011] In step 4, the parameter recalibration procedure sets the membrane flux decay rate threshold to a preset percentage. When the actual decay rate exceeds this value, the crossflow velocity is automatically increased to a predetermined percentage range. At the same time, the enzymatic hydrolysis temperature is reduced by a specific amount within the preset temperature range to reduce the reaction rate and thus reduce the contribution of newly generated peptides to membrane fouling.

[0012] In step 5, the nanofiltration unit uses a composite membrane with a predetermined molecular weight cutoff, the operating pressure is maintained within a preset range, and the desalination rate reaches a preset standard; the recovery rate of the reverse osmosis unit is set to a predetermined value, the conductivity of the product water is lower than a preset limit, and the water quality meets the standard for raw material slurry preparation.

[0013] The central control unit has a built-in historical operating condition database that stores a sufficient amount of batch operation data, including raw material protein content, batch differences in enzyme activity, ambient temperature and humidity, and corresponding optimal operating parameter sets. When a new batch starts, it automatically matches similar operating conditions and initializes control parameters, shortening the time required for system stabilization.

[0014] In the method, the residence time of the enzymatic hydrolysis reaction is controlled within a predetermined time period by adjusting the ratio of the feed flow rate to the reactor volume, ensuring that the target peptide is retained and separated by the membrane module before significant secondary hydrolysis occurs, and that the molecular weight distribution of the peptide is concentrated in a preset range, with a proportion exceeding a preset ratio.

[0015] The entire system is equipped with an abnormal early warning mechanism. When any sensor data deviates from the normal range by more than a preset percentage for a predetermined period of time, or when the membrane pressure difference suddenly increases beyond the preset pressure value, the system will automatically reduce its load and issue a maintenance prompt to prevent equipment damage and product quality fluctuations.

[0016] Compared with the prior art, the beneficial technical effects of the present invention are as follows: This invention breaks through the limitations of traditional batch operations. Through the integrated continuous process design of enzymatic hydrolysis and membrane separation, the soybean protein extraction process is transformed from intermittent to truly continuous production. The daily processing capacity of a single line is significantly improved, the equipment utilization rate is greatly increased, and the energy consumption and labor cost per unit product are significantly reduced. At the same time, the closed-loop water recycling system significantly reduces process water consumption, which meets the requirements of green manufacturing.

[0017] This invention introduces a multi-parameter online sensing and dynamic feedback control model. The system can sense dynamic disturbances such as raw material fluctuations, enzyme activity changes, and membrane fouling evolution in real time, and autonomously adjust key operating parameters to ensure that high extraction efficiency and stable product quality are maintained under different operating conditions. Experiments show that under conditions of fluctuating raw material protein content, the standard deviation of peptide yield is significantly lower than that of the fixed parameter system.

[0018] This invention effectively suppresses excessive hydrolysis side reactions by precisely controlling the enzymatic hydrolysis reaction window and the instant separation mechanism, resulting in a higher proportion of the target molecular weight range in the obtained soybean peptide product, which is significantly higher than that of traditional methods. At the same time, adaptive cross-flow regulation significantly reduces the membrane flux decay rate, extends the service life of the membrane module, and reduces the frequency of downtime for cleaning.

[0019] This invention integrates historical data-driven initial parameter matching and real-time optimization algorithms to form a closed-loop control architecture of "perception-decision-execution". This not only improves the system's automation level but also provides a data foundation for process knowledge accumulation and continuous optimization. The system can be seamlessly integrated with the factory's MES system, supporting remote monitoring and predictive maintenance, and enhancing the reliability and scalability of industrial applications. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall technical architecture of the adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis proposed in this invention. Figure 2 This is a schematic diagram of the core principle framework of the dynamic feedback control model in this invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Example 1 To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0023] The operation steps in this embodiment are as follows: Figure 1 As shown, Figure 1 This is a schematic diagram of the overall technical architecture of the adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis proposed in this invention. The specific operation is as follows: Step 1: Construct a continuous enzymatic hydrolysis-membrane separation coupling system, which specifically includes the following operations: The pretreated soybean protein raw material slurry is fed into the enzymatic hydrolysis reactor at a constant flow rate through a high-precision volumetric continuous feed pump. The flow rate control accuracy of the feed pump is ±0.5% to ensure feed stability.

[0024] The enzymatic hydrolysis reactor adopts a fully mixed-flow reactor structure with an effective volume of 2000L. It features a double-jacketed structure, with heat transfer oil circulated within the jacket as the heat exchange medium. The temperature difference between the inlet and outlet of the heat transfer oil is controlled by a temperature control unit, achieving a temperature control accuracy of ±0.2℃ for the reaction system. The reactor is equipped with a variable-frequency speed-regulating impeller, allowing for stepless adjustment of the stirring speed from 50rpm to 300rpm. The impeller blades are three-bladed swept-back, and the shear rate gradient is controlled within... to This design ensures thorough mixing and contact between the enzyme and substrate while preventing inactivation due to conformational disruption caused by localized high shear forces. A sanitary quick-connect clamp directly connects the microfiltration membrane module inlet to the enzymatic digester outlet, creating a direct-connection reaction-separation integrated process without intermediate storage tanks, eliminating dead zones and residues caused by batch switching. The microfiltration membrane module uses zirconia ceramic membrane tubes with a pore size of 0.2 μm, each membrane tube being 1500 mm long, with an inner diameter of 8 mm and an outer diameter of 12 mm. The total filtration area of ​​the membrane module is 15... The operating pressure is maintained within the range of 0.15 MPa to 0.35 MPa. The microfiltration membrane module outlet is divided into two paths: one path allows the concentrate to be refluxed back to the inlet of the enzymatic hydrolysis reactor to form a cross-flow circulation, and the other path allows the permeate to enter the ultrafiltration membrane module. The ultrafiltration membrane module uses a polyethersulfone hollow fiber membrane with a molecular weight cutoff of 5 kDa, a membrane fiber inner diameter of 0.8 mm, an outer diameter of 1.2 mm, and a total effective membrane area of ​​30 μm. The operating pressure is maintained within the range of 0.2MPa to 0.5MPa. Each microfiltration and ultrafiltration membrane module is equipped with an independent transmembrane differential pressure control valve, which is a proportional pneumatic control valve with a response time of 0.5s and a control accuracy of ±0.01MPa. Simultaneously, each membrane module is equipped with an independent cross-flow circulation pump, a magnetically driven centrifugal pump with a flow rate range of 0 to 50 m³ / h and a head of 0 to 40 m. The cross-flow velocity is precisely adjusted within the range of 1.0 m / s to 3.5 m / s via a frequency converter. Furthermore, both membrane modules integrate a backwashing program. The backwashing medium is clean compressed air or permeate water, the backwashing pressure is 0.4MPa, and the backwashing duration is 30s. The backwashing cycle is not fixed but dynamically calculated based on the membrane flux decay rate. When the membrane flux decay rate exceeds 0.8% / min, the system automatically triggers a backwash command.

[0025] Step 2 involves deploying a multi-parameter online sensor network, specifically as follows: Three turbidity sensors and two peptide concentration optical sensors are uniformly arranged axially inside the enzymatic hydrolysis reactor. One set of conductivity sensors and membrane flux monitoring units are installed at the feed inlet, concentrate outlet, and permeate outlet of the microfiltration membrane module. The same type of sensor array is also configured at corresponding positions on the ultrafiltration membrane module. The turbidity sensor uses a 90-degree scattered light detection method, with an 860nm infrared LED as the light source and a silicon photodiode as the detector. The measurement range covers 0 NTU to 1000 NTU, with a measurement accuracy of ±2%FS and a response time of less than 5 seconds. The peptide concentration optical sensor is based on the principle of ultraviolet absorption spectroscopy, using a built-in deuterium lamp as the light source. A monochromator is used to select a wavelength of 280nm to detect the absorbance of aromatic amino acid residues (mainly tryptophan and tyrosine). The optical path length is 10mm, the measurement range is 0 mg / L to 200 mg / L, the resolution is 0.1 mg / L, and the response time is less than 3 seconds. This sensor has been calibrated using standard peptide solutions before leaving the factory. The calibration equation is as follows: Where C is the peptide concentration (mg / L). The absorbance at 280 nm is given, and a and b are calibration coefficients stored in the sensor's local memory. The conductivity sensor adopts a four-electrode structure, with a measurement range of 0 μS / cm to 20000 μS / cm, an accuracy of ±1%FS, and a temperature compensation range of 0℃ to 80℃. The membrane flux monitoring unit consists of a high-precision electromagnetic flowmeter and membrane area parameters. The flowmeter accuracy is ±0.3%, the sampling frequency is 1Hz, and it calculates the volumetric flux per unit membrane area J in real time. All sensors transmit the collected state variable data to the central control unit in 1-second cycles via the Industrial Ethernet protocol. The data packet format includes timestamp, sensor ID, raw value, engineering unit value, and status flag bit to ensure data integrity and timing consistency.

[0026] Step 3: Establish a dynamic feedback control model, such as... Figure 2 As shown, the central control unit dynamically adjusts key operating parameters based on real-time acquired state variable data, combined with the preset target peptide molecular weight distribution range (1kDa to 5kDa accounting for ≥85%) and membrane fouling index threshold (defined as J / J0≤0.7), through an adaptive algorithm. The core of this dynamic feedback regulation model lies in two mathematical expressions: one is the enzymatic hydrolysis kinetic equation, used to predict the peptide generation rate. in, This represents the peptide concentration (mg / L) at time t. This represents the current effective enzyme concentration (U / mL). and The first is a pH- and temperature-dependent kinetic constant, whose value is identified and updated in real time using online peptide concentration data; the second is a membrane flux decay model, used to assess the degree of membrane fouling. in, Initial membrane flux ( ), Pollution resistance coefficient ( ), The membrane surface concentration (mg / L) is calculated from the concentration factor and feed concentration; the central control unit incorporates a multivariate predictive controller (MPC) to comprehensively optimize the objective function. To optimize the objective, among which, For real-time peptide yield based on material balance, This represents the current flux decay. Energy consumption per unit product (kWh / kg peptide), weighting coefficient , , The sum is 1; the controller performs rolling optimization every 10 seconds, outputting the pH setpoint (range 7.5 to 9.0), temperature setpoint (range 50℃ to 60℃), and enzyme addition rate (range 0.1U / ) of the enzymatic hydrolysis reactor for the next control cycle. Up to 0.5U / The system includes transmembrane pressure differential settings for microfiltration and ultrafiltration membrane modules (0.15–0.35 MPa and 0.2–0.5 MPa, respectively), cross-flow velocity settings (1.0–3.5 m / s), and upper limits for concentration factor (3 to 8 times). pH adjustment is achieved by injecting 1 mol / L NaOH or HCl solution via a metering pump, with an adjustment accuracy of ±0.05. The enzyme addition unit uses a peristaltic pump with a flow rate range of 0.1 mL / min to 5 mL / min and a pulsation error of <1%.

[0027] Step 4 involves adaptive parameter optimization, the specific mechanism of which is as follows: The central control unit continuously monitors the peptide concentration increase rate. With membrane flux decay rate ;when For 30 consecutive seconds, the concentration of the drug was below 0.5 mg / L. or More than 0.5L / When this occurs, the system automatically triggers a parameter recalibration procedure. This procedure prioritizes increasing the crossflow velocity to 120% to 150% of its current value, for example, from 2.0 m / s to 2.4 m / s to 3.0 m / s, to enhance membrane shear force, remove deposited contaminants, and alleviate concentration polarization. Simultaneously, it fine-tunes the enzymatic hydrolysis temperature, decreasing it by 1 to 3°C within a preset temperature range (e.g., 55°C ± 2°C), for example, from 55°C to 53°C, to reduce the enzyme catalytic reaction rate and decrease the contribution of newly generated small peptides to membrane fouling. This dual regulation strategy is achieved through decoupled control logic, where crossflow velocity regulation dominates membrane fouling suppression, and temperature regulation dominates reaction rate matching; the two work synergistically. The system maintains a steady state; the magnitude and direction of parameter adjustments are determined by similar cases in the historical operating condition database. The database stores more than 500 batches of operating data, covering different raw material protein contents (35% to 50%), batch differences in enzyme activity (nominal value ±15%), environmental temperature and humidity (15℃ to 35℃, 40%RH to 80%RH), and corresponding optimal operating parameter sets. When a new batch is started, the system automatically extracts the protein content, moisture, and ash content data from the current raw material test report, performs Euclidean distance matching with the database, and selects the 10 nearest neighbor historical parameter sets as the initial control set, which greatly shortens the time required for the system to go from startup to stable operation (usually <15min).

[0028] Step 5 achieves product fractionation and solvent recovery, with the following specific process: The retentate (containing the target peptide) from the ultrafiltration membrane module is transported to the finished product storage tank via a sterile diaphragm pump, output as a high-purity soybean peptide product; its permeate (containing salts, small molecule sugars, and free amino acids) enters the nanofiltration unit, which uses a polyamide composite membrane with a molecular weight cutoff of 150 Da and an effective membrane area of ​​10... ², the operating pressure is maintained within the range of 1.0 MPa to 2.0 MPa, the desalination rate is ≥95%, and the small molecule impurity removal rate is ≥90%; the retentate from the nanofiltration unit can be selectively refluxed to the ultrafiltration feed end to improve the yield, or collected as a by-product; the nanofiltration permeate (mainly pure water) enters the reverse osmosis unit, which uses a seawater desalination-grade reverse osmosis membrane, with a recovery rate set at 75%, an operating pressure of 1.5 MPa, and a product water conductivity ≤50 μS / cm, fully meeting the standards for raw material slurry preparation water (conductivity <100 μS / cm); reverse osmosis concentrate... After evaporation and crystallization, the water is transported off-site for disposal. The entire water recycling path forms a closed-loop water circulation system, reducing process water consumption to less than 30% of that of traditional processes. In addition, the system is equipped with an abnormal early warning mechanism. When any sensor data deviates from the normal range by more than 15% for more than 60 seconds, or when the pressure difference of the microfiltration / ultrafiltration membrane suddenly increases by more than 0.1 MPa, the central control unit immediately executes a load reduction command: the feed flow rate is reduced to 50%, the enzyme addition rate is reduced to zero, the cross-flow pump maintains its rated speed, and at the same time, a maintenance prompt window pops up on the human-machine interface, records the fault code and timestamp, and pushes the alarm information to the factory's MES system through the OPC UA protocol to prevent equipment damage and product quality fluctuations.

[0029] To further illustrate the practical application effects of this invention, the following specific application scenario is constructed: A soybean deep-processing enterprise uses the method of this invention to process a production line with a daily processing capacity of 10 tons of dry-based soybean protein powder. The raw material slurry has a solid content of 12% and a pH of 8.2, and is fed into a 2000L fully mixed-flow enzymatic hydrolysis reactor via a continuous feed pump at a flow rate of 8.33 m³ / h. The initial reactor settings are 55℃, pH 8.5, and the Alcalase 2.4L FG enzyme dosing rate is 0.3 U / L. The online peptide concentration sensor displays the peptide concentration in real time as it increases from 0 mg / L to 150 mg / L, with an increase rate of 2.1 mg / L. The flux is within the normal range. The initial flux of the microfiltration membrane module is 80 L / m³. After running for 30 minutes, the flux dropped to 68 L / min. The decay rate is 0.4 L / m³. No recalibration was triggered. At 45 minutes into the run, a batch change in raw materials caused the protein content to rise from 45% to 48%, while the peptide concentration increase rate plummeted to 0.3 mg / L. Meanwhile, the microfiltration flux decay rate increased to 0.6 L / L. The central control unit immediately triggered parameter recalibration: the crossflow velocity increased from 2.2 m / s to 2.8 m / s (127%), and the enzymatic hydrolysis temperature decreased from 55℃ to 52℃. After 10 minutes, the peptide concentration increase rate recovered to 1.8 mg / L. The flux decay rate decreased to 0.35 L / The system returned to steady state. The final product contained 89.2% 1kDa–5kDa peptides, with a peptide yield of 92.5%. The membrane module operated continuously for 120 hours, 40% longer than the fixed-parameter system. Nanofiltration desalination reached 96.3%, and the reverse osmosis permeate conductivity was 42 μS / cm, all of which was recycled for slurry preparation, saving approximately 60 tons of water per day.

[0030] In another example, a sudden drop in ambient temperature to 10°C increased the heat load on the jacket of the enzymatic hydrolysis reactor, causing temperature fluctuations of ±1.5°C. The central control unit retrieved a low-temperature case from the historical database (ambient temperature 12°C, protein content 42%) and automatically increased the initial stirring speed from 150 rpm to 200 rpm to enhance heat transfer. It also increased the preset enzyme addition rate by 10% to compensate for the decrease in enzyme activity caused by the low temperature. The system reached the set steady state within 8 minutes, and the peptide concentration curve deviated from the standard operating condition by less than 5%, demonstrating the strong robustness of this invention.

[0031] Example 2 In another embodiment, a plug flow reactor is used instead of a fully mixed flow reactor for the enzymatic hydrolysis reactor. Static mixing elements are installed inside, and the standard deviation of the residence time distribution is controlled within 10% of the average residence time. The feed flow rate to reactor volume ratio is precisely set to ensure an average residence time of 90 minutes, guaranteeing that the target peptides are retained and separated by the membrane module before significant secondary hydrolysis occurs. The microfiltration membrane module is replaced with a silicon carbide ceramic membrane with a pore size of 0.1 μm, extending the pH tolerance range to 1–14, making it suitable for more demanding cleaning conditions. A 205 nm wavelength pass-through is added to the peptide concentration optical sensor. The system uses a method to detect peptide bond absorption and integrates it with 280nm data to improve measurement accuracy in low-concentration regions. A fuzzy logic controller is introduced into the dynamic feedback control model to classify the membrane fouling index into three levels: "mild," "moderate," and "severe," each corresponding to a different crossflow velocity enhancement strategy. The nanofiltration unit employs vibrating membrane technology, using high-frequency mechanical vibration (50Hz, 0.5mm amplitude) to further suppress membrane fouling, increasing the desalination rate to 98%. The central control unit connects to a cloud-based AI platform, utilizing federated learning technology to aggregate multi-plant operating data and continuously optimize kinetic constants. , With pollution coefficient The predictive model enables cross-regional sharing of process knowledge.

[0032] Example 3 In another embodiment, the ultrafiltration membrane module adopts a two-stage series structure, with the first stage having a molecular weight cutoff of 10 kDa and the second stage having 3 kDa, achieving more refined peptide fractionation. An intermediate buffer tank and pH adjustment unit are set between the two stages to ensure that the pH of the second-stage feed is precisely controlled within the range of the enzyme's optimal pH ± 0.1. An online sensing network is equipped with a near-infrared spectrometer, installed on the side wall of the enzymatic hydrolysis reactor, which predicts the substrate conversion rate and peptide molecular weight distribution in real time through a PLS regression model. The central control unit has a built-in digital twin module that drives a virtual reactor model based on real-time data, predicting membrane fouling trends and pre-adjusting parameters 10 minutes in advance. The permeate from the reverse osmosis unit is used as dilution water for the membrane module online cleaning (CIP) system to reduce fresh water consumption. The anomaly warning mechanism is upgraded to multivariate statistical process control (MSPC), which uses a principal component analysis (PCA) model to detect anomalies in the covariance structure of sensor data, reducing the false alarm rate to below 1%. The entire system is certified to the IEC 62443 safety standard and supports remote OTA firmware upgrades to ensure long-term operational reliability.

[0033] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0034] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis, characterized in that, Includes the following steps: A continuous enzymatic hydrolysis-membrane separation coupling system was constructed. Soybean protein raw material slurry was fed into the enzymatic hydrolysis reactor through a continuous feed pump. The outlet of the enzymatic hydrolysis reactor was directly connected to the microfiltration membrane module and the ultrafiltration membrane module, forming a continuous process integrating reaction and separation. The enzymatic hydrolysis reactor was equipped with an online pH adjustment unit, a temperature control unit and an enzyme solution metering and dosing unit. The microfiltration membrane module and the ultrafiltration membrane module were respectively equipped with independent transmembrane pressure differential control valves and cross-flow circulation pumps. Deploy a multi-parameter online sensor network, and install turbidity sensors, peptide concentration optical sensors, conductivity sensors and membrane flux monitoring units inside the enzymatic hydrolysis reactor and at the inlet and outlet of the membrane module to collect key state variable data in the reaction system in real time and transmit them to the central control unit. A dynamic feedback control model is established. The central control unit, based on real-time collected state variable data and combined with the preset target peptide molecular weight distribution range and membrane fouling index threshold, dynamically adjusts the pH value, temperature, enzyme addition rate of the enzymatic hydrolysis reactor, as well as the transmembrane pressure difference, cross-flow velocity and concentration factor of the membrane module through an adaptive algorithm. Adaptive parameter optimization is performed. When the peptide concentration growth rate is detected to be lower than the set threshold or the membrane flux decay rate exceeds the preset upper limit, the central control unit automatically triggers the parameter recalibration program, prioritizes adjusting the crossflow rate to alleviate membrane fouling, and simultaneously fine-tunes the enzymatic hydrolysis temperature to maintain the optimal enzyme activity window. The system achieves product classification and solvent recovery. The permeate from the ultrafiltration membrane module enters the nanofiltration unit for desalination and removal of small molecule impurities. The retentate is output as a high-purity soybean peptide product. The nanofiltration permeate is treated by reverse osmosis and then reused for the preparation of raw material slurry, forming a closed-loop water circulation system.

2. The adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis according to claim 1, characterized in that, The enzymatic hydrolysis reactor adopts a fully mixed-flow reactor structure with an internal double-jacket for precise temperature control. The stirring speed is adjustable from 50 rpm to 300 rpm, and the stirring blades are three-bladed swept-back type. The shear rate gradient is controlled within... to between.

3. The adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis according to claim 1, characterized in that, The microfiltration membrane module uses a ceramic membrane tube module with a pore size of 0.2 μm, and the operating pressure is maintained in the range of 0.15 MPa to 0.35 MPa; the ultrafiltration membrane module uses a polyethersulfone hollow fiber membrane with a molecular weight cutoff of 5 kDa, and the operating pressure is maintained in the range of 0.2 MPa to 0.5 MPa; both membrane modules are equipped with independent backwashing programs, and the backwashing cycle is dynamically set according to the membrane flux decay rate.

4. The adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis according to claim 1, characterized in that, The peptide concentration optical sensor is based on the principle of ultraviolet absorption spectroscopy. It detects the absorbance of aromatic amino acid residues at a wavelength of 280 nm and converts it into peptide concentration value through a pre-stored calibration curve. The turbidity sensor adopts the 90-degree scattered light detection method, with an 860 nm infrared LED as the light source and a silicon photodiode as the detector.

5. The adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis according to claim 1, characterized in that, The dynamic feedback control model includes an enzymatic hydrolysis kinetic equation, a membrane flux decay model, and a comprehensive optimization objective function. The enzymatic hydrolysis kinetic equation is used to predict the change of peptide concentration over time, the membrane flux decay model is used to assess the degree of membrane fouling, and the comprehensive optimization objective function uses peptide yield, membrane flux retention rate, and energy consumption per unit product as optimization variables. The multivariate predictive controller performs rolling optimization every 10 seconds.

6. The adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis according to claim 1, characterized in that, The parameter recalibration procedure was performed when the peptide concentration increase rate was below 0.5 for 30 consecutive seconds. Or the membrane flux decay rate exceeds When triggered, the crossflow rate automatically increases to 120% to 150% of the current value, while the enzymatic hydrolysis temperature decreases by 1°C to 3°C within the preset temperature range.

7. The adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis according to claim 1, characterized in that, The central control unit has a built-in historical operating condition database that stores no less than 500 batches of operating data. When a new batch starts, it performs a similarity match between the current raw material protein content, moisture and ash content data and the database, and selects the nearest neighbor set of historical parameters as the initial control parameters.

8. The adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis according to claim 1, characterized in that, The residence time of the enzymatic hydrolysis reaction is controlled by adjusting the feed flow rate and reactor volume ratio so that the target peptides are retained and separated by the membrane module before significant secondary hydrolysis occurs. The molecular weight distribution of the obtained peptides is concentrated in the range of 1kDa to 5kDa, and the proportion of peptides in this range is not less than 85%.

9. The adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis according to claim 1, characterized in that, The nanofiltration unit uses a composite membrane with a molecular weight cutoff of 150 Da, and the operating pressure is maintained within the range of 1.0 MPa to 2.0 MPa, with a desalination rate of not less than 95%; the reverse osmosis unit has a recovery rate set at 75%, and the permeate conductivity is not higher than 50%. It meets the water standards for preparing raw material slurry.

10. The adaptive optimization method for continuous extraction of soybean protein based on membrane separation coupled with enzymatic hydrolysis according to claim 1, characterized in that, The system is equipped with an abnormal early warning mechanism. When any sensor data deviates from the normal range by more than 15% for more than 60 seconds, or when the membrane pressure difference suddenly increases by more than 0.1MPa, the central control unit automatically executes a load reduction command, reducing the feed flow rate to 50%, the enzyme addition rate to zero, and maintaining the rated speed of the cross-flow pump while issuing a maintenance prompt.

Images