A continuous methanol single cell protein production process and apparatus

Abstract

A continuous methanol single-cell protein production method and device, the production method is applied to the production device, including the following steps: the three-dimensional physical reaction space in the fermentation device is discretized into a three-dimensional body grid structure, and a plurality of flow field discrete sampling nodes are defined at the grid intersection; the real-time fluid viscosity is calculated, the partial pressure sampling dissolved oxygen value is weighted and compensated in the spatial dimension, and a global dissolved oxygen characterization tensor covering the whole space is reconstructed; the metabolic activity of microorganisms corresponding to each flow field discrete sampling node is evaluated, and a virtual metabolic field map is generated; the metabolic low-activity area in the virtual metabolic field map is identified, and a spatial differential compensation is driven to be performed by an execution unit. The present application realizes accurate reconstruction and prediction of the metabolic state of the whole space, greatly suppresses the environmental gradient through spatial differential compensation, significantly reduces the operation cost and load of the precision monitoring system, and improves the safety and product consistency of the bacterial growth.

Description

Technical Field

[0001] This invention relates to the field of bio-fermentation and automated control technology, and more specifically, to a continuous methanol single-cell protein production method and apparatus. Background Technology

[0002] The continuous fermentation of methanol single-cell protein (MSCP) is a highly complex biochemical process involving gas, liquid, and solid phases. In large-scale industrial fermentation plants, due to the significant cytotoxicity of the methanol substrate and the extremely high dissolved oxygen (DO) requirement of the microorganisms, traditional monitoring methods have the following significant drawbacks: Significant spatial environmental gradients: Due to uneven fluid mixing inside the fermenter, there are significant dissolved oxygen gradients and nutrient concentration differences at different heights and radial positions. Traditional single-point monitoring cannot represent the true metabolic state of the entire tank, easily leading to localized microbial hypoxia or methanol poisoning. The monitoring data is lagging and unreliable: Due to the physical membrane permeation resistance, the dissolved oxygen probe has a significant lag when reporting high oxygen consumption conditions; at the same time, expensive physicochemical analyses such as ICP monitoring will cause huge operating losses and instrument burden if they are kept running at a constant high frequency. Limited control methods: Existing aeration and feeding methods usually involve uniform regulation of the entire tank, which cannot accurately compensate for metabolic dead zones in specific spaces, thus restricting the yield and product quality consistency of continuous fermentation.

[0003] Therefore, the existing technology has problems and needs further improvement and development. Summary of the Invention

[0004] (I) Purpose of the invention: In order to solve the problems existing in the prior art, the purpose of the present invention is to provide a continuous methanol single-cell protein production method and apparatus.

[0005] (II) Technical Solution: To solve the above-mentioned technical problems, this technical solution provides a continuous methanol single-cell protein production method, which utilizes a single-cell protein production device with a fermentation unit, and includes the following steps: Step S1: Discretize the three-dimensional physical reaction space inside the fermentation device into a three-dimensional volumetric mesh structure, and define several flow field discrete sampling nodes at the mesh intersections; configure a multi-dimensional parameter set for each flow field discrete sampling node to store the empty flow field calibration vector at the coordinates of the flow field discrete sampling node; the empty flow field calibration vector is obtained from the cold fluid experiment before fermentation feeding, and includes the local oxygen supply gain coefficient and the local shear stress distribution coefficient; Step S2: Real-time acquisition of fermentation process operating parameters, including partial pressure sampled dissolved oxygen values ​​obtained by multiple dissolved oxygen detection devices, real-time oxygen consumption intensity obtained by exhaust gas detection device, stirring shaft power, and methanol residual concentration; using the real-time oxygen consumption intensity combined with the real-time fluid viscosity calculated based on stirring shaft power, spatial weighting compensation is performed on the partial pressure sampled dissolved oxygen values ​​to reconstruct a global dissolved oxygen characterization tensor covering the entire space; Step S3: Map the global dissolved oxygen characterization tensor to the empty flow field calibration vector stored at each of the flow field discrete sampling nodes, evaluate the microbial metabolic activity corresponding to each of the flow field discrete sampling nodes, and generate a virtual metabolic field map reflecting the spatial differences of the gas-liquid-solid three-phase environment by fitting a spatial interpolation algorithm. Step S4: Identify low-activity metabolic regions in the virtual metabolic field map in real time, locate the corresponding discrete sampling nodes of the flow field, and drive the execution unit to perform spatial difference compensation. Gas phase regulation: For the intake sector to which the metabolic low activity zone belongs, the gas volume ratio is adjusted according to the local oxygen supply gain coefficient of the corresponding discrete sampling node of the flow field; Liquid phase control: Non-uniform pulse feeding is performed in the metabolically low activity zone to eliminate the spatial environmental gradient caused by uneven mixing of the three phases during integrated continuous fermentation.

[0006] The continuous methanol single-cell protein production method further includes, in step S1, the multidimensional parameter set as follows: Local metabolic heat production coefficient: used to record the heat accumulation potential of each discrete sampling node of the flow field under unit bacterial density; Local fluid shear gradient: used to assess the risk of physical damage to the cell membrane at discrete sampling nodes of the flow field in real time during continuous fermentation, in conjunction with the stirring shaft power.

[0007] In the continuous methanol single-cell protein production method, the weighted compensation in step S2 specifically includes: The real-time oxygen consumption intensity is used as a feedforward gain to correct the hysteresis of the partial pressure sampled dissolved oxygen value in the gas-liquid mass transfer process; and the Reynolds number calculated based on the stirring shaft power is used to dynamically interpolate and verify the local oxygen supply gain coefficient at each discrete sampling node of the flow field.

[0008] The continuous methanol single-cell protein production method, wherein the liquid phase control in step S4 specifically includes: A single feeding action is defined as a discrete feeding instance, and a spatial response attribute is configured for each discrete feeding instance; Based on the differences in substrate demand in different coordinate regions of the virtual metabolic field map, differentiated pulse frequency offsets are assigned to independent feed ports located at different positions. By adjusting the trigger frequency of the discrete feeding instances on the time axis, dynamic smoothing of the local nutrient concentration around each discrete sampling node of the flow field can be achieved.

[0009] The continuous methanol single-cell protein production method, wherein the gas phase control in step S4 specifically includes: The central control unit calculates the spatial flux weighting factor for the corresponding gas supply sector based on the local oxygen supply gain coefficient of each discrete sampling node of the flow field. The central control unit identifies metabolically inactive regions through the virtual metabolic field map and, based on the spatial flux weighting factor, dynamically increases the air volume allocation ratio of the air intake sector to which the metabolically inactive region belongs, while keeping the total air intake constant. By adjusting the opening degree of the proportional solenoid valve of the corresponding sector, the critical flow velocity of the gas when passing through the micro-orifice of the distributor is changed, thereby achieving spatial gradient compensation for the local dissolved oxygen rate around the discrete sampling node of the flow field.

[0010] The continuous methanol single-cell protein production method further includes a physicochemical analysis scheduling based on monitoring window clipping, specifically: Define a key feature window for the fermentation process, which is determined based on the rate of change of parameters at each sampling node in the virtual metabolite map; When the virtual metabolic field map shows that the global environmental parameters are within the preset steady-state range, the sampling frequency of the physicochemical analysis task is reduced, thereby reducing the operating load of the ICP intelligent monitoring system. When the attribute fluctuation of any discrete sampling node of the flow field in the virtual metabolic field map exceeds the threshold, the high-frequency monitoring mode is activated.

[0011] A continuous methanol single-cell protein production device, comprising a main fermentation unit, a sensing unit, an execution unit, and a central control unit; The main fermentation unit is a continuous three-phase reaction fermentation device with an internal three-dimensional reaction space, which is used to provide a physical site for gas-liquid-solid three-phase biochemical reactions. The sensing unit is used to acquire fermentation kinetics and metabolic parameters in real time, including multiple dissolved oxygen detection devices, exhaust gas detection devices and stirring shaft power detection devices that are spatially arrayed. The execution unit is used to perform differentiated resource allocation according to instructions, including a segmented gas distribution device and a multi-point feeding system; The central control unit is used to receive data from the sensing unit and calculate and generate a virtual metabolic field map, thereby driving the execution unit to perform partition compensation. The central control unit stores the empty flow field calibration vector.

[0012] The continuous methanol single-cell protein production device includes a stirring device on the central shaft of the fermentation device, which is used to stir the materials inside the fermentation device. The stirring device includes a stirring shaft set on the central axis of the fermenter, a stirring drive motor at one end of the stirring shaft, and three layers of stirring paddles installed in series along the axial direction. The three layers of stirring paddles divide the fermenter into a three-layer structure of bottom, middle and top layers in the vertical direction.

[0013] The continuous methanol single-cell protein production apparatus, wherein the segmented gas distribution device is used to adjust the oxygen supply intensity of different sectors according to the radial distribution differences of the virtual metabolic field map, including: At least four independent gas supply sectors arranged in a circular array, each equipped with an independent proportional solenoid valve and flow control unit.

[0014] The continuous methanol single-cell protein production apparatus, wherein the multi-point feeding system includes: At least three independent feeding branches are set at different height levels along the axis of the fermentation device; the feeding branches are used to match the layering of the three-dimensional volumetric mesh structure in the height dimension, and independently execute variable frequency pulse feeding according to the axial concentration difference of the virtual metabolic field map.

[0015] (III) Beneficial effects: The present invention provides a continuous methanol single-cell protein production method and apparatus, which realizes the accurate reconstruction and prediction of the metabolic state in the whole space, greatly smooths the environmental gradient through spatial difference compensation, significantly reduces the operating cost and load of the precision monitoring system, and improves the safety of cell growth and product consistency. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the steps in a continuous methanol single-cell protein production method according to the present invention; Figure 2 This is a schematic diagram of the structure of a fermenter in a continuous methanol single-cell protein production device according to the present invention. Figure 3 This is a schematic diagram of the segmented gas distribution device of a continuous methanol single-cell protein production apparatus according to the present invention. 100-Fermentation tank; 101-Full-length baffle; 102-Agitator shaft; 103-Agitator drive motor; 104-Dynamic torque sensor; 105-Bottom layer agitator; 106-Middle layer agitator; 107-Top layer agitator; 108-Dissolved oxygen detection device; 109-Outlet condenser; 110-Exhaust gas detection device; 200-Segmented gas distribution device; 201-Gas supply sector; 202-Inlet branch pipe. Detailed Implementation

[0017] The present invention will be further described in detail below with reference to preferred embodiments. More details are set forth in the following description in order to provide a full understanding of the present invention. However, the present invention can obviously be implemented in many other ways different from those described herein. Those skilled in the art can make similar extensions and derivations based on actual application situations without departing from the spirit of the present invention. Therefore, the scope of protection of the present invention should not be limited by the content of this specific embodiment.

[0018] The accompanying drawings are schematic diagrams of embodiments of the present invention. It should be noted that these drawings are for illustrative purposes only and are not drawn to scale, and should not be construed as limiting the actual scope of protection of the present invention.

[0019] A continuous methanol single-cell protein production apparatus includes a main fermentation unit, a sensing unit, an execution unit, and a central control unit. The sensing unit and the execution unit are mounted on the fermentation unit and are respectively connected to the central control unit.

[0020] The main fermentation unit is a continuous three-phase reaction fermentation device with an internal three-dimensional reaction space. The inner wall of the fermentation device is provided with a corrosion-resistant composite lining to provide a physical environment for gas-liquid-solid three-phase biochemical reactions.

[0021] The sensing unit is used to acquire fermentation kinetics and metabolic parameters in real time, including multiple dissolved oxygen detection devices, exhaust gas detection devices, and stirring shaft power detection devices arranged in a spatial array. The fermentation kinetic parameters focus on describing the mass transfer, mixing, and growth rate characteristics within the reactor, specifically characterized by physical quantities such as stirring shaft power and real-time oxygen consumption intensity. The metabolic parameters focus on describing the real-time transformation state of nutrients by the cells, specifically characterized by biochemical indicators such as partial pressure sampled dissolved oxygen values ​​and substrate residual concentration. Fermentation kinetics and metabolic parameters together constitute the data foundation for reconstructing a global virtual metabolic field map.

[0022] The execution unit is used to perform differentiated resource allocation according to instructions, including a segmented gas distribution device and a multi-point feeding system.

[0023] The central control unit is used to receive data from the sensing unit and calculate and generate a virtual metabolic field map, thereby driving the execution unit to perform partition compensation. The central control unit is a distributed control system (DCS) running a metabolic field rendering algorithm, and the central control unit stores the empty flow field calibration vector.

[0024] The fermentation apparatus is preferably a continuous three-phase reaction fermenter with a total volume of 600L, such as... Figure 2 As shown, its diameter-to-height ratio is between 1:2.5 and 1:3 to prolong the residence time of bubbles in the tank. The inner wall of the fermenter is lined with a 2mm thick polyvinylidene fluoride (PVDF) anti-corrosion lining to resist corrosion from high concentrations of methanol and continuous metabolic acidic byproducts. Four full-length baffles are uniformly welded along the circumference of the inner wall of the fermenter to eliminate fluid swirl. A stirring device is installed on the central shaft of the fermenter to stir the materials inside. The stirring device includes a stirring shaft located on the central shaft of the fermenter, a stirring drive motor at one end of the stirring shaft, and three layers of stirring blades installed in series along the axial direction. The three layers of stirring blades divide the fermenter into a three-layer structure in the vertical direction: bottom layer, middle layer, and top layer. The bottom stirring blade uses a six-bladed disc turbine, such as a Rushton turbine, to powerfully shear and break up the air intake at the bottom; the middle and top stirring blades use four-bladed wide-face thrust blades to enhance the axial fluid circulation.

[0025] Multiple dissolved oxygen detection devices, arranged in an array, can be positioned at 1 / 4, 1 / 2, and 3 / 4 of the fermenter's height from bottom to top, corresponding to the mixing zones of the bottom, middle, and top agitators, respectively. Three dissolved oxygen detection devices are inserted at 120° intervals along the circumference on each horizontal plane. These devices can be in-situ sterilizable optical dissolved oxygen probes. A total of nine dissolved oxygen probes are deployed throughout the fermenter. The insertion depth of the optical dissolved oxygen probes is divided into a shallow zone (approximately 0.1D) near the fermenter wall and a deep zone (approximately 0.3D) near the main axis. Physically, these probes directly map to discrete sampling nodes in the flow field of the continuous methanol single-cell protein production method, used to collect real-time dissolved oxygen values ​​within the corresponding coordinate regions. Where D refers to the inner diameter of the fermenter, 0.1D depth means that the tip of the optical dissolved oxygen probe extends from the wall of the fermenter to the center a distance of 10% of the inner diameter. For example, if the inner diameter of the tank is 600mm, the probe extends 60mm in. This mainly monitors the fluid state near the wall. 0.3D depth means that the tip of the optical dissolved oxygen probe extends a distance of 30% of the inner diameter. This is closer to the drainage area of ​​the agitator and can monitor the dissolved oxygen in the most active mass transfer area.

[0026] To cover the entire space of the fermentation unit, the nine probes were arranged in a matrix of 3 layers and 3 angles: Vertical height: Installed at the height of the bottom, middle and top agitators to capture the dissolved oxygen gradient from bottom to top.

[0027] Horizontal radial: On the same height plane, some extend shallower by 0.1D depth, and some extend deeper by 0.3D depth.

[0028] Circumferential angle: The probes are installed at 120° offsets on the circumference to prevent them from interfering with each other in the flow field.

[0029] The exhaust gas from the top of the fermenter is connected in series with a tail gas detection device via a water outlet condenser. This tail gas detection device can be a mass spectrometer-type tail gas analyzer, used for millisecond-level output of real-time oxygen consumption intensity. The power output end of the stirring drive motor is connected to the stirring shaft via a coupling. A dynamic torque sensor is installed at the coupling to collect the actual input power of the stirring shaft in real time, i.e., the stirring shaft power.

[0030] The segmented gas distribution device is used to independently adjust the oxygen supply intensity of different sectors based on the radial distribution differences of the virtual metabolic field map, thereby eliminating the dissolved oxygen gradient on the horizontal plane and achieving gas phase regulation. The segmented gas distribution device includes at least four independent gas supply sectors arranged in a circumferential array, and each gas supply sector is equipped with an independent proportional solenoid valve and flow control unit.

[0031] Specifically, the segmented gas distribution device is located below the bottom turbine propeller. For example... Figure 3 As shown, the segmented gas distribution device is a ring structure composed of four independent microporous gas distribution pipes arranged in a 90° fan shape, with each fan-shaped area serving as a gas supply sector. Each fan-shaped microporous gas distribution pipe is connected to the main gas supply pipe outside the tank via an independent inlet branch pipe. A precision pressure reducing valve, a thermal mass flow controller (MFC), and a high-speed proportional solenoid valve are connected in series on each inlet branch pipe.

[0032] The multi-point feeding system enables liquid phase control and includes at least three independent feeding branches set at different height levels along the axial direction of the fermentation unit. These feeding branches are used to match the stratification of the three-dimensional volumetric mesh structure in the height dimension, and independently execute variable-frequency pulse feeding based on the axial concentration differences in the virtual metabolic field map to balance the substrate concentration at different height levels within the tank.

[0033] Specifically, the multi-point feeding system, targeting substrates such as methanol and ammonia, has independent feeding pipes welded at three height levels corresponding to the dissolved oxygen detection device in the fermenter, forming a three-layer feeding array. Each feeding pipe is controlled by an independent stepper motor-driven peristaltic pump, and a one-way check valve to prevent backflow is installed at the end of the feeding pipe. When the central control unit determines that a metabolically low activity zone exists in a certain level, it triggers the peristaltic pump at that level to perform high-frequency pulse injection.

[0034] The central control unit adopts a distributed control system (DCS) architecture based on the collaboration of an industrial computer and a programmable logic controller (PLC). The industrial computer has an embedded high-performance GPU module, and its memory is pre-loaded with the cold-state three-dimensional volumetric mesh coordinates of the fermenter and the corresponding no-load flow field calibration vector. The PLC communicates with all thermal mass flow controllers (MFC), peristaltic pumps, dissolved oxygen probes, and torque sensors via a fieldbus at millisecond levels, ensuring that the rendering output of the virtual metabolic field map and the action of the physical valves are highly synchronized on the time axis. The fieldbus can be Profinet or EtherCAT.

[0035] The main fermentation unit also includes a sampling bypass connected to the ICP intelligent monitoring system. The bypass is equipped with an automatic sampling valve for periodically extracting fermentation broth and sending it to the ICP intelligent monitoring system for online analysis of trace metal element concentrations.

[0036] A continuous methanol single-cell protein production method is applied to a continuous methanol single-cell protein production device with a fermentation unit.

[0037] like Figure 1 As shown, a continuous methanol single-cell protein production method includes the following steps: Step S1: Discretize the three-dimensional physical reaction space inside the fermentation device into a three-dimensional volumetric mesh structure, and define several flow field discrete sampling nodes at the mesh intersections; configure a multi-dimensional parameter set for each flow field discrete sampling node to store the empty flow field calibration vector at the coordinates of the flow field discrete sampling node; the empty flow field calibration vector is obtained from the cold fluid experiment before fermentation feeding, and includes the local oxygen supply gain coefficient and the local shear stress distribution coefficient; Step S2: Real-time acquisition of fermentation process operating parameters via sensor array. These operating parameters include partial pressure sampled dissolved oxygen values ​​acquired by multiple dissolved oxygen detection devices, real-time oxygen consumption intensity acquired by exhaust gas detection device, stirring shaft power, and residual methanol concentration. Using the real-time oxygen consumption intensity combined with the real-time fluid viscosity calculated based on stirring shaft power, spatial weighting compensation is applied to the partial pressure sampled dissolved oxygen values ​​to reconstruct a global dissolved oxygen characterization tensor covering the entire space. Step S3: The central control unit performs mapping operations on the global dissolved oxygen characterization tensor and the empty flow field calibration vector stored in each of the flow field discrete sampling nodes, evaluates the microbial metabolic activity corresponding to each of the flow field discrete sampling nodes, and generates a virtual metabolic field map reflecting the spatial differences of the gas-liquid-solid three-phase environment by fitting a spatial interpolation algorithm. Step S4: The central control unit identifies the low metabolic activity region in the virtual metabolic field map in real time, locates the corresponding discrete sampling node of the flow field, and automatically drives the execution unit to perform spatial difference compensation. Gas phase control: The linkage segmented gas distribution device independently adjusts the gas volume ratio for the intake sector to which the metabolic low activity zone belongs, based on the local oxygen supply gain coefficient of the corresponding discrete sampling node of the flow field. Liquid phase control: The multi-point feeding system is linked to perform non-uniform pulse feeding in the low metabolic activity zone; thereby eliminating the spatial environmental gradient caused by uneven mixing of the three phases during integrated continuous fermentation.

[0038] In step S1, the multidimensional parameter set further includes: a local metabolic heat production coefficient and a local fluid shear gradient. The local metabolic heat production coefficient is used to record the heat accumulation potential of each discrete sampling node of the flow field at a unit cell density. The local fluid shear gradient is used to assess the risk of physical damage to the cell membrane at the discrete sampling node of the flow field in real time during continuous fermentation, in conjunction with the stirring shaft power.

[0039] Step S1 specifically includes: The central control unit inputs the actual physical dimensions of the fermentation device, including tank diameter, tank height, baffle width, and the positions and blade shapes of the three-layer agitator, and performs geometric modeling to obtain the first model. The continuous cylindrical space in the first model is then divided into thousands of extremely small three-dimensional blocks to obtain a three-dimensional volumetric mesh structure. The intersections of the three-dimensional volumetric mesh structure are extracted as discrete sampling nodes for the flow field. In this invention, to reduce the computational load on the central control unit, the installation location of the dissolved oxygen detection device is used as the core anchor point, i.e., the discrete sampling node for the flow field, to construct the backbone mesh.

[0040] In the memory matrix of the central control unit, a unique spatial coordinate identifier is assigned to each discrete sampling node of the flow field. A multidimensional parameter set is allocated for each discrete sampling node of the flow field to store the calibration vector of the unloaded flow field obtained from the cold fluid experiment before fermentation feeding.

[0041] A specific concentration of sodium carboxymethyl cellulose (CMC) aqueous solution was injected into the fermentation apparatus as a simulation solution to ensure its apparent viscosity matched that of a real high-density bacterial culture. The agitator was turned on to the target speed, and the segmented gas distribution device at the bottom was activated to introduce gas, simulating the fluid dynamics during actual fermentation. Cold-state fluid experiments refer to purely physical tests conducted without the introduction of living microorganisms or the occurrence of actual biochemical reactions.

[0042] The local oxygen supply gain coefficient was determined using a dynamic degassing method: First, pure nitrogen was introduced into the fermentation device to expel all oxygen from the simulated liquid until the dissolved oxygen detection device readings at all discrete sampling nodes of the flow field dropped to near zero. Then, the nitrogen was instantly switched to air, and the central control unit recorded the dissolved oxygen rise curves at each discrete sampling node of the flow field at high frequency. At the discrete sampling nodes near the bottom agitator, the bubbles were well broken up, resulting in a very rapid rise in dissolved oxygen; at the discrete sampling nodes near the top tank wall, the rise was very slow. The central control unit calculated the slope of each curve, i.e., the volumetric oxygen mass transfer coefficient. The gain / attenuation ratio of each discrete sampling node in the flow field relative to the average level of the entire fermentation device is obtained, yielding the local oxygen supply gain coefficient. The local shear stress distribution coefficient is calculated: the central control unit reads data from the dynamic torque sensor on the stirring shaft to obtain the total shaft power; combining this with fluid dynamics formulas, the closer the discrete sampling node is to the impeller tip, the greater the shear force, making it easier to break the cell wall; the farther the discrete sampling node is from the impeller tip, the smaller the shear force. The central control unit quantifies these mechanical gradients and stores them as the local shear stress distribution coefficient.

[0043] The weighted compensation in step S2 specifically involves: using the real-time oxygen consumption intensity as a feedforward gain to correct the hysteresis of the partial pressure sampled dissolved oxygen value in the gas-liquid mass transfer process; and using the Reynolds number Re calculated based on the stirring shaft power to dynamically interpolate and verify the local oxygen supply gain coefficient at each of the discrete sampling nodes of the flow field.

[0044] The specific implementation process of step S2 is as follows: During fermentation, the central control unit polls the sensing unit at a high frequency with a period of 100ms. At this time, the exhaust gas detection device detects a sharp drop in exhaust oxygen content and calculates a significant jump in the real-time oxygen consumption intensity (OUR), indicating that the bacteria have entered a high oxygen consumption state.

[0045] Simultaneously, nine dissolved oxygen monitoring devices distributed across three height levels within the fermentation unit transmit their monitored partial pressure dissolved oxygen values ​​to the central control unit. Due to physical resistance in gas-liquid mass transfer, there is a 5-10 second lag in the decrease of dissolved oxygen monitoring device readings. At this time, the central control unit does not directly use the probe readings for control; instead, it activates a weighted compensation logic: using the aforementioned rate of change in real-time oxygen consumption intensity as a feedforward gain, and superimposing the local fluid viscosity resistance calculated from the stirring shaft, it performs temporal and spatial advance corrections on the current readings of the nine dissolved oxygen monitoring devices, thereby reconstructing a three-dimensional global dissolved oxygen characterization tensor that includes the predicted trend.

[0046] In fermentation engineering, the real-time oxygen uptake rate (OUR) is not directly detected by dissolved oxygen monitoring devices, but is calculated based on the law of conservation of gas mass. The central control unit determines a significant jump in the real-time oxygen uptake rate (OUR) through the following specific process.

[0047] The central control unit incorporates a gas phase mass balance equation. Every specified interval, such as 100ms, the central control unit reads the intake air flow rate. Inlet oxygen concentration And the exhaust flow rate sent from the exhaust gas detection device to the central control unit and exhaust oxygen concentration When air is introduced, the oxygen content is typically 20.9%.

[0048] The formula for calculating real-time oxygen consumption intensity is: .

[0049] in, This represents the real-time volume of the fermentation liquid within the fermentation apparatus.

[0050] In order to filter out noise from the dissolved oxygen detection device and capture instantaneous metabolic changes, the central control unit does not only look at the absolute value of the real-time oxygen consumption intensity OUR, but also at the first derivative of the real-time oxygen consumption intensity, that is, its rate of change, as detailed below.

[0051] The central control unit sets a time sliding window, for example... =5 seconds, calculate the real-time slope of the change in real-time oxygen consumption intensity OUR: .when Exceeding the preset threshold of the central control unit When this occurs, the central control unit determines that the real-time oxygen consumption intensity (OUR) has significantly increased and triggers a subsequent advance correction procedure.

[0052] Because of a lag of 2-10 seconds, when the central control unit determines that the real-time oxygen consumption intensity OUR has increased significantly, the monitoring value of the dissolved oxygen detection device has not yet decreased. It is necessary to subtract the future trend of 5-10 seconds in advance, that is, to perform spatial dimension weighted compensation on the partial pressure sampled dissolved oxygen value, and reconstruct a global dissolved oxygen characterization tensor covering the entire space.

[0053] Specifically, it includes: The fermentation broth is a shear-thinning non-Newtonian fluid. The central control unit first calculates the current apparent fluid viscosity by using the real-time power P and rotational speed N of the stirring drive motor. The pre-stored no-load flow field calibration vector, and the central control unit assigns local mass transfer resistance penalty coefficients to the discrete sampling nodes of the flow field where the nine dissolved oxygen detection devices are located. Where i = 1, 2, ..., 9. In the dead zone far from the agitator, the fermentation broth has a higher viscosity and slower bubble mass transfer. Once high oxygen consumption occurs, oxygen is lost most rapidly here, hence the local mass transfer resistance penalty coefficient. The larger.

[0054] Dissolved oxygen values ​​were collected based on real-time partial pressure sampling from nine dissolved oxygen monitoring devices. The central control unit calculates nine compensated dissolved oxygen values ​​that include future trends based on the following formula. : ,in, This represents the current physical reading of the i-th dissolved oxygen detection device; The preset feedforward gain constant; This is the slope of the real-time oxygen consumption intensity (OUR) change that was just calculated; This is the mass transfer resistance penalty coefficient specific to the discrete sampling node of the flow field.

[0055] The central control unit calculated the nine compensated dissolved oxygen values. arrive Using a spatial interpolation algorithm, with these nine points as the framework and combined with the coordinate system (x, y, z) of the three-dimensional volumetric mesh, a smooth fitting calculation was performed on the mesh intersection points of the fermentation device to obtain... The three-dimensional data matrix, i.e., the global dissolved oxygen characterization tensor. .

[0056] Calculate the current apparent fluid viscosity Specifically, the central control unit reads the real-time variable frequency output power of the stirring motor. And deduct the mechanical losses of the transmission system and the no-load power of the motor itself. Calculate the actual effective stirring power transferred to the fermentation broth, i.e., the real-time power P: , The mechanical efficiency constant of the transmission system is given by: effective power P, current real-time rotational speed N, and real-time density of the fermentation broth. And the diameter D of the agitator in the fermentation device, calculate the current real-time power coefficient. : .

[0057] In the cold fluid experiment before fermentation feeding in step S1, the central control unit also used sodium carboxymethyl cellulose (CMC) simulation solutions of different concentrations to plot and store the power number-Reynolds number characteristic mapping curve specific to the fermentation device. In the transition or laminar flow regions of non-Newtonian fluids, the power number... With Reynolds number There exists a strict one-to-one correspondence. The central control unit calls upon this power metric-Reynolds number characteristic mapping curve and calculates it through a table lookup or polynomial fitting back-calculation algorithm. The calculated real-time power index Inversely calculate the current mixed Reynolds number. .

[0058] According to the standard definition formula of the stirring Reynolds number The central control unit calculates the apparent fluid viscosity of the fermentation broth at that instant by performing a transposition transformation. : Current apparent fluid viscosity This reflects the macroscopic viscosity of the current high-density bacterial solution. This value is then substituted into the calculation of the spatial mass transfer resistance penalty coefficient to accurately compensate for the dissolved oxygen value of the partial pressure sampling.

[0059] Since this invention relates to a gas-liquid-solid three-phase mixture system, the actual hydrodynamic density of the fermentation broth is not the aqueous constant. Gas is continuously introduced into the bottom of the fermentation device, causing the fluid to expand, and the concentration of solid-phase cells continuously increases with the growth of single-cell proteins. To accurately calculate the dimensionless power coefficient in the formula... The sensing unit also includes a dual-flange differential pressure level transmitter. The dual-flange differential pressure level transmitter includes two high-precision pressure probes respectively deployed at the bottom and top of the fermentation unit's liquid level, used to acquire the liquid column pressure difference in real time. The central control unit is based on the hydrostatic equation. Where h is the fixed axial height difference between the two probes, the gas density of the mixed phase including the bubble phase is calculated in real time. Its value is usually around 700 kg / m 3 Up to 950kg / m 3 The dynamic changes between them. This dynamic... Substitute the value into the power coefficient The calculation formula eliminates the viscosity estimation error caused by changes in ventilation volume or liquid level fluctuations, ensuring the absolute accuracy of Reynolds number back-calculation.

[0060] Since apparent viscosity and shear stress are inversely proportional by a power law, the central control unit is pre-programmed with the following inverse proportional mapping formula to calculate the local mass transfer resistance penalty coefficient of each discrete sampling node of the flow field in real time. : ,in, This is an empirical index for fluid rheology, specifically for high-density late-stage systems using methanol-based single-cell proteins. The value is usually between 0.4 and 0.6.

[0061] The specific implementation process of step S3 is as follows: The central control unit extracts two-dimensional tensor data from the discrete sampling nodes of the flow field: Extract the compensated dissolved oxygen values ​​containing future trends from the corresponding coordinates of the global dissolved oxygen characterization tensor reconstructed in step S2. ; Extract the multidimensional parameter set pre-stored in step S1, including the local oxygen supply gain coefficient of the node. and local shear stress distribution coefficient ; The central control unit performs mapping calculations using a preset metabolic evaluation cost function. Specifically, the central control unit calculates a normalized microbial metabolic activity index for each discrete sampling node of the flow field. The calculation model is as follows: ,in, This is the magnification factor. This represents the weighting constant for the inhibitory effect of shear force on cell metabolism. The physical logic of this formula is as follows: it predicts that regions with higher dissolved oxygen and better oxygen supply will have more vigorous metabolism; while in regions with greater shear force, cell metabolism will be inhibited due to physical damage.

[0062] The nine discrete sampling nodes of the flow field were calculated based on the above formula. After the value is calculated, the central control unit normalizes it, mapping it to the 0-100% range, and sets a threshold based on the physiological characteristics of the bacterial species. >75% is designated as the high metabolic activity zone, which is usually located in the area with moderate stirring paddle and sufficient dissolved oxygen. 40%≤ ≤75% is defined as the metabolically stable region; <40% is defined as a region of low metabolic activity, which is a region of low metabolic activity that requires physical compensation.

[0063] Due to the limited number of dissolved oxygen detection devices—only nine in this embodiment of the invention—while the three-dimensional volumetric mesh structure of the fermentation device contains tens of thousands of intersections, the central control unit uses these nine core flow field discrete sampling nodes. The value is the baseline data anchor point. A three-dimensional kriging space interpolation algorithm is invoked for all remaining nodes without dissolved oxygen detection devices. The values ​​are smoothly estimated, and the discrete nodes are generalized into a continuous three-dimensional scalar field. The three-dimensional kriging space interpolation algorithm not only considers the spatial distance, but also incorporates the fluid anisotropy variogram function of the radial and axial directions of the stirring shaft.

[0064] After the interpolation calculation is completed, the visualization module of the central control unit renders the three-dimensional scalar field as a virtual metabolic field map. On the visualization module, the virtual metabolic field map is presented in the form of a three-dimensional heatmap: for example, red represents high activity and blue represents low activity. The virtual metabolic field map is a matrix containing three-dimensional spatial coordinates (x, y, z) and corresponding microbial metabolic activity indices. This matrix directly serves as the underlying digital base, outputting to step S4 to accurately guide the coordinate addressing and operation of the linked segmented gas distribution device and the multi-point feeding system.

[0065] A three-dimensional scalar field is generated by fitting the fluid anisotropy variogram using a spatial interpolation algorithm. The specific process is as follows: Step 1: Calculation of anisotropic spatial distance based on flow field characteristics; In classical Euclidean geometry, spatial distance is isotropic. However, in a fermentation device, the mass transfer rate in the horizontal radial direction is much greater than that in the vertical axis. Therefore, the central control unit first introduces a fluid anisotropy range penalty factor into the spatial coordinate system to calculate arbitrary unknown grid nodes. Discrete sampling nodes of known flow field The anisotropic equivalent distance h between them: ,in, , , , which represent the radial, axial, and tangential distance differences between the two points in cylindrical coordinates; , , , respectively, are the spatial range constants in the corresponding directions. For the fermentation device, the axial range... Much smaller than radial range This mathematically reflects the physical reality that mixing between upper and lower layers is slow, while mixing within the same layer is fast. The aforementioned three-dimensional volumetric mesh structure is generated by spatial discretization based on a cylindrical coordinate system (r, θ, z) constructed with the stirring axis of the fermentation device as the Z-axis. Here, r corresponds to the radial depth dimension of the fermentation device, θ corresponds to the circumferential tangential dimension, and z corresponds to the axial height dimension. This cylindrical coordinate mesh perfectly matches the physical model of the fermentation device, which has circular boundary characteristics and rotating flow field properties.

[0066] Step 2: Construction of the theoretical variogram model; The central control unit uses a spherical model or an exponential model to construct an anisotropic variogram function. This is used to quantify the variation of metabolic activity differences between two points in space with the equivalent distance h. Here, we take the exponential model as an example: , among which, The nugget constant represents measurement error or random fluctuations at the microscopic scale; For the value of the off-center sill; To comprehensively assess the effective range, this index model indicates that nodes that are closer in distance have more similar metabolic states and smaller variogram values.

[0067] Step 3: Construction and solution of the unbiased optimal Kriging equations; For any unknown node in the blind zone of the 3D volume mesh Its estimated metabolic activity This is not a simple average of the surrounding known nodes, but rather the measured values ​​from the anchor points of nine known dissolved oxygen monitoring devices. Linear weighted combination: ; To ensure the unbiasedness of the estimation and minimize the estimation variance, the central control unit constructs the following Kriging matrix equations to solve for the optimal weight coefficients. : ,in, Let be the known values ​​of the mutation function between nodes. The variogram values ​​are the values ​​of the known nodes and the unknown nodes. It is a Lagrange multiplier.

[0068] Step 4: Global mesh traversal and 3D scalar field generation; The central control unit matrix calculation module calculates the matrix inversion operation for the current unknown node. optimal weight set and calculate ; Subsequently, the central control unit traverses all the grid intersections formed after the discretization of the three-dimensional physical reaction space of the fermentation device, for example, 50*50*100 voxel nodes, and iteratively executes the first to third steps described above for each node. Finally, it outputs a list containing all spatial coordinates and their corresponding... The estimated values ​​are a continuous data matrix, which is a three-dimensional scalar field reflecting the spatial differences in the gas, liquid, and solid three-phase environments.

[0069] Solving for the weight coefficient set With Lagrange multiplier The specific process is as follows: For any unknown node The matrix solving module of the central control unit automatically generates the corresponding linear algebraic matrix equations based on the aforementioned constructed Kriging equation system. Where: K is the known spatial covariance matrix: a 10*10 matrix whose internal elements are the variogram values ​​between each pair of the 9 known nodes. The structure is as follows: the last row and last column are set to a constant 1, and the bottom right corner of the diagonal is 0, to reflect the unbiased constraint condition; D is the distance vector: a 10*1 column vector containing the distances from 9 known probe nodes to unknown nodes. variogram value The last element is a constant 1; W is the unknown solution vector: a 10*1 column vector containing the 9 optimal weight coefficients to be determined. - And the last Lagrange multiplier .

[0070] Since the elements in matrix K and vector D can be obtained in real time through coordinate distance and theoretical variability function, they are all known quantities. The central control unit uses Gauss-Jordan elimination or LU decomposition to perform inversion operation on matrix K, and executes... In this calculation step, the central control unit simultaneously calculates the current unknown node. Dedicated optimal weight set With Lagrange multiplier Get Then, the central control unit can substitute the values ​​into the weighted formula to calculate the estimated metabolic activity of that node. .

[0071] The gas-phase control in step S4 specifically includes: The central control unit calculates the spatial flux weighting factor for the corresponding gas supply sector based on the local oxygen supply gain coefficient of each discrete sampling node of the flow field. The central control unit identifies metabolically inactive regions through the virtual metabolic field map, and dynamically increases the air volume allocation ratio of the air intake sector to which the metabolically inactive region belongs, while keeping the total air intake constant, based on the spatial flux weighting factor, and correspondingly reduces the air volume allocation ratio of other sectors in metabolic steady state or high-activity regions. By adjusting the opening degree of the proportional solenoid valve of the corresponding sector, the critical flow velocity of the gas when passing through the micro-orifice of the distributor is changed, thereby achieving spatial gradient compensation for the local dissolved oxygen rate around the discrete sampling node of the flow field.

[0072] The effective range of the differential distribution of the bottom gas phase is limited to the area between the bottom and middle layers. For the metabolically low-activity zone at the top layer, the central control unit no longer relies on the distribution of the bottom gas phase sector, but instead switches the weight to the liquid phase compensation of the multi-point feeding system.

[0073] The liquid phase control in step S4 specifically includes: The central control unit defines a single feeding action as a discrete feeding instance and configures a spatial response attribute for each discrete feeding instance. Based on the differences in substrate demand in different coordinate regions of the virtual metabolic field map, differentiated pulse frequency offsets are assigned to independent feed ports located at different positions. By adjusting the trigger frequency of the discrete feeding instances on the time axis, dynamic smoothing of the local nutrient concentration around each discrete sampling node of the flow field can be achieved.

[0074] The central control unit retrieves the continuous data matrix generated in step S3 at millisecond intervals. When a certain coordinate is identified... Metabolic activity index When the concentration is less than 40%, the region is identified as a metabolically inactive area, and the following dual-pathway compensation mechanism is activated: The central control unit first calls the local oxygen supply gain coefficient of the flow field discrete sampling node corresponding to the coordinate node. ,right In regions with lower natural mass transfer resistance, the spatial flux weighting factor is calculated. ,in, and It exhibits an inverse correlation mapping and is used to correct the compensation strength.

[0075] In order not to interfere with the total pressure and defoaming balance of the fermentation device, the central control unit adjusts the proportional solenoid valves of the corresponding sectors in the segmented distribution device according to the spatial flux weighting factor instruction.

[0076] By increasing the valve opening in this specific sector, the critical flow velocity of gas passing through the micropores in that area is increased. The increased flow velocity enhances the shearing effect at the gas-liquid interface, generating higher-density microbubbles, thereby significantly increasing the dissolved oxygen rate at this local node and eliminating the dissolved oxygen gradient in a spatial dimension.

[0077] The central control unit defines a single peristaltic pump push action as a discrete feeding instance, and each instance contains specific spatial response attributes: feeding amount and coordinate orientation.

[0078] For the identified low-activity area coordinates, the central control unit locates the independent feeding port at that height level and calculates the pulse frequency offset. , Depending on the substrate demand gap in the region, by Bias determines.

[0079] Normal zone: Maintain basic feeding frequency For example, it can be triggered once every 10 seconds.

[0080] Metabolic inactivity region: Assign a positive offset Increase the trigger frequency to = + For example, it can be triggered once every 2 seconds.

[0081] Because each discrete feeding instance has a small volume, such as 5 mL, the high-frequency pulse injection allows the methanol substrate to be rapidly diluted by the local fluid the moment it enters the tank, avoiding the risk of toxicity caused by excessively high local methanol concentration due to continuous feeding, and achieving dynamic smoothing of local nutrient concentration.

[0082] Through the coordinated action of gas-phase control by adjusting the distribution ratio and microbubble density, and liquid-phase control by adjusting the pulse frequency and discrete distribution, the metabolically inactive zone in the fermentation unit that was originally in a state of suffocation or starvation is precisely repaired.

[0083] By continuously cycling through steps S2 to S4, the biological metabolic activity at each sampling node throughout the entire tank is increased. The process tends to be consistent, reducing its fluctuation range to within ±5% of the average value, thereby eliminating the spatial environmental gradient caused by uneven mixing of the three phases during integrated continuous fermentation.

[0084] This section describes the experimental procedures for cold fluid before fermentation.

[0085] Step 1: Simulate fluid preparation and mesh initialization; With the fermenter unloaded, a prepared simulated solution was injected into the fermentation unit. To accurately reproduce the non-Newtonian fluid characteristics of methanol-containing single-cell protein during the later stages of continuous fermentation, a 0.1%-1.5% (w / w) sodium carboxymethyl cellulose (CMC) aqueous solution was used as the simulated solution, ensuring its apparent viscosity matched that of the target high-density bacterial solution. Simultaneously, the distributed control system (DCS) of the central control unit initialized the three-dimensional volumetric mesh and activated the probes of nine dissolved oxygen detection devices distributed at different coordinate levels within the tank, establishing them as discrete sampling nodes for the actual flow field. To accurately reproduce the non-Newtonian fluid characteristics of methanol-containing single-cell protein during the later stages of continuous fermentation when the cell dry weight (DCW) reaches above 30 g / L, a 0.5% (w / w) CMC aqueous solution was used as an example. At a 0.5% (w / w) CMC aqueous solution, the rheological index and consistency coefficient of the simulated solution accurately simulated the shear-thinning behavior of the target high-density bacterial solution, ensuring that the unloaded flow field calibration vector obtained under cold-state experiments has real industrial guiding significance.

[0086] Step 2: Extraction of local shear stress distribution coefficient; The stirring device of the fermentation unit is started, and a gradient speed curve is set. At each speed increment, the distributed control system (DCS) of the central control unit records the stirring shaft power fed back by the dynamic torque sensor at the coupling. Based on the input power and the real-time rheological curve of the simulated liquid, the DCS calculates the average energy dissipation rate of the entire tank using built-in empirical fluid dynamics formulas. Subsequently, combining the spatial coordinates of the nine discrete flow field sampling nodes, such as the shallow wall region at 0.1D and the deep impeller region at 0.3D, the velocity gradient at each node is calculated and normalized to generate the local shear stress distribution coefficient for the corresponding coordinates. This coefficient reflects the potential mechanical damage intensity to the bacterial cell membrane at different locations.

[0087] Step 3: Calibrate the local oxygen supply gain coefficient based on the dynamic degassing method; Under the set stirring speed and aeration rate, pure nitrogen gas is first introduced into the tank through a segmented gas distribution device to drive out dissolved oxygen in the simulated liquid until the partial pressure sampling dissolved oxygen values ​​obtained by the probes of the nine dissolved oxygen detection devices all drop below 5%. Then, the air intake is instantly switched to standard air or oxygen-enriched air, simultaneously triggering high-frequency data acquisition by the central control unit. Each of the nine dissolved oxygen detection devices records its dissolved oxygen recovery curve at its spatial coordinates, i.e., the exponential increase curve of dissolved oxygen (DO) concentration over time. The central control unit performs logarithmic slope fitting on each recovery curve to calculate the local volumetric oxygen mass transfer coefficient at each node. Locally from each node Compared with the average of the whole tank By performing ratio calculations, the local oxygen supply gain coefficient of each node can be obtained. This coefficient accurately characterizes the inherent physical differences in the oxygen acquisition capacity of different areas of the entire tank under the same air intake conditions.

[0088] Step 4: Encapsulation and preloading of calibration vectors; The central control unit packages the local shear stress distribution coefficient obtained in step two with the local oxygen supply gain coefficient obtained in step three to generate a tensor data structure that corresponds one-to-one with the three-dimensional spatial coordinates, namely the unloaded flow field calibration vector.

[0089] Ultimately, the empty flow field calibration vector is written into the multidimensional parameter set in the memory of the central control unit, serving as the static physical reference for rendering the virtual metabolic field map and executing the zonal automated closed-loop control after the fermentation officially starts.

[0090] A continuous methanol single-cell protein production method also includes physicochemical analysis scheduling based on monitoring window trimming: Define a key feature window for the fermentation process, which is determined based on the rate of change of parameters at each sampling node in the virtual metabolite map; When the virtual metabolic field map shows that the global environmental parameters are within the preset steady-state range, the central control unit automatically adjusts the sampling frequency of the physicochemical analysis task to reduce the operating load of the ICP intelligent monitoring system for continuous methanol single-cell protein production; the ICP intelligent monitoring system is used to monitor the real-time concentration of trace metal elements in the fermentation broth. When the attribute fluctuation of any discrete sampling node in the flow field of the virtual metabolic field map exceeds a threshold, the high-frequency monitoring mode is automatically activated. The attributes of the discrete sampling node refer to the multi-dimensional real-time state data carried by the physical coordinate node, including: physical attributes: apparent viscosity and shear stress at the point; biochemical attributes: partial pressure sampled dissolved oxygen value and local metabolic heat production rate at the point; and state-derived attributes: metabolic activity index of the point in the virtual metabolic field map.

[0091] Specifically, The central control unit defines the current fermentation window in real time by calculating the first derivative of the attributes of each node in the virtual metabolic field map, i.e. the rate of change.

[0092] Steady-state window: when the activity of biological metabolism... When the rate of change of key indicators such as dissolved oxygen value and real-time oxygen consumption intensity (OUR) is less than 2% for 10 consecutive minutes, the central control unit enters the steady-state window. At this time, the fermentation shows dynamic equilibrium and the material conversion is extremely stable.

[0093] Disturbance window: When the attribute fluctuation of any node exceeds the preset threshold, such as a sudden drop in dissolved oxygen or a sudden change in viscosity, the central control unit instantly determines to enter the disturbance window.

[0094] When the central control unit identifies the steady-state window, it sends a command to the ICP intelligent monitoring system to extend the interval of physicochemical sampling from every 15 minutes to every 2 hours. At this time, since the environmental parameters of the entire domain are within the preset steady-state range, the consumption rate of trace elements is constant, and high-frequency monitoring is not required. This significantly reduces the operating load of the peristaltic pump and plasma torch of the ICP intelligent monitoring system, and reduces the consumption of argon and diluent.

[0095] The central control unit maintains millisecond-level scanning of the virtual metabolic field map. When the property fluctuation of any discrete sampling node in the flow field exceeds a threshold, such as a local dissolved oxygen change rate exceeding 10% / min:

[0096] If the central control unit determines that there is a potential risk of metabolic imbalance or substrate accumulation, it will immediately and automatically interrupt the tailored physicochemical analysis task and forcibly wake up the ICP intelligent monitoring system to enter high-frequency monitoring mode, such as every 5 minutes. After obtaining the high-frequency physicochemical analysis results, the central control unit will combine these trace element data to further modify the gas phase regulation and liquid phase regulation compensation instructions in step S4 to achieve rapid suppression of abnormal fluctuations.

[0097] A continuous methanol single-cell protein production method and apparatus, through three-dimensional volumetric mesh discretization and virtual metabolic field mapping rendering, completely solves the problem that single-point monitoring in traditional fermentation processes cannot represent the entire tank state: By using a matrix distribution of nine probes and a spatial interpolation algorithm, it can identify metabolically low-activity zones at any coordinate within the fermentation unit in real time, avoiding growth stagnation caused by localized microbial hypoxia or substrate scarcity; it can reduce the fluctuation range of microbial metabolic activity at each sampling node of the fermentation unit to within ±5% of the average value, ensuring high consistency of product quality in continuous production. In fermentation engineering, the physical lag of dissolved oxygen detection devices is a major cause of control failure: This invention uses real-time oxygen consumption intensity as a feedforward gain and superimposes a fluid viscosity resistance factor calculated based on stirring power to perform advance correction of the monitored values; the reconstructed global dissolved oxygen characterization tensor not only covers the entire space but also includes predictions of metabolic trends, enabling the execution unit to perform pre-compensation before physical parameters exceed limits. This invention employs differentiated resource allocation for low-activity zones: a linked segmented gas distribution device, while maintaining a constant total gas intake, adjusts the proportional solenoid valves of local sectors to change the critical flow rate of micropores, enhancing local oxygen supply. It also optimizes mass transfer efficiency through the generation of high-density microbubbles, avoiding waste from ineffective ventilation. The multi-point feeding system, using frequency modulation without amplitude modulation, ensures substrate supply while guaranteeing instantaneous dilution of methanol upon entry into the tank, completely mitigating the risk of methanol poisoning that may result from continuous feeding. The newly added monitoring window trimming logic reflects a deep integration of fermentation process control and equipment maintenance: when the overall environment is in a steady-state window, the sampling frequency of the ICP intelligent monitoring system is automatically reduced, effectively lowering the operating load on expensive analytical instruments and the consumption of reagents such as argon; when abnormal fluctuations occur in the physical or biochemical properties of discrete sampling nodes in the flow field, high-frequency monitoring is immediately activated, enabling on-demand sampling and significantly extending the service life of precision detection equipment while ensuring the safety of continuous fermentation. The safety boundary of the cell growth environment has been strengthened: the multidimensional parameter set introduces the local fluid shear gradient, and combined with the stirring shaft power, the risk of cell membrane damage is assessed in real time to prevent cell inactivation due to excessive stirring; by recording the local metabolic heat production coefficient, potential heat accumulation risk points are warned in advance, ensuring the thermodynamic balance of the continuous fermentation process.

[0098] The above description illustrates preferred embodiments of the present invention and helps those skilled in the art to more fully understand the technical solution of the present invention. However, these embodiments are merely illustrative and should not be construed as limiting the specific implementation of the present invention to these embodiments. For those skilled in the art, several simple deductions and modifications can be made without departing from the inventive concept, and all such modifications should be considered within the protection scope of the present invention.

Claims

1. A continuous methanol single-cell protein production method, using a single-cell protein production device with a fermentation unit, characterized in that, Includes the following steps: Step S1: Discretize the three-dimensional physical reaction space inside the fermentation device into a three-dimensional volumetric mesh structure, and define several flow field discrete sampling nodes at the mesh intersections; configure a multi-dimensional parameter set for each flow field discrete sampling node to store the empty flow field calibration vector at the coordinates of the flow field discrete sampling node; the empty flow field calibration vector is obtained from the cold fluid experiment before fermentation feeding, and includes the local oxygen supply gain coefficient and the local shear stress distribution coefficient; Step S2: Real-time acquisition of fermentation process operating parameters, including partial pressure sampled dissolved oxygen values ​​obtained by multiple dissolved oxygen detection devices, real-time oxygen consumption intensity obtained by exhaust gas detection device, stirring shaft power, and methanol residual concentration; using the real-time oxygen consumption intensity combined with the real-time fluid viscosity calculated based on stirring shaft power, spatial weighting compensation is performed on the partial pressure sampled dissolved oxygen values ​​to reconstruct a global dissolved oxygen characterization tensor covering the entire space; Step S3: Map the global dissolved oxygen characterization tensor to the empty flow field calibration vector stored at each of the flow field discrete sampling nodes, evaluate the microbial metabolic activity corresponding to each of the flow field discrete sampling nodes, and generate a virtual metabolic field map reflecting the spatial differences of the gas-liquid-solid three-phase environment by fitting a spatial interpolation algorithm. Step S4: Identify low-activity metabolic regions in the virtual metabolic field map in real time, locate the corresponding discrete sampling nodes of the flow field, and drive the execution unit to perform spatial difference compensation. Gas phase regulation: For the intake sector to which the metabolic low activity zone belongs, the gas volume ratio is adjusted according to the local oxygen supply gain coefficient of the corresponding discrete sampling node of the flow field; Liquid phase control: Non-uniform pulse feeding is performed in the metabolically low activity zone to eliminate the spatial environmental gradient caused by uneven mixing of the three phases during integrated continuous fermentation.

2. The continuous methanol single-cell protein production method according to claim 1, characterized in that, The multidimensional parameter set in step S1 also includes: Local metabolic heat production coefficient: used to record the heat accumulation potential of each discrete sampling node of the flow field under unit bacterial density; Local fluid shear gradient: used to assess the risk of physical damage to the cell membrane at discrete sampling nodes of the flow field in real time during continuous fermentation, in conjunction with the stirring shaft power.

3. The continuous methanol single-cell protein production method according to claim 1, characterized in that, The weighted compensation in step S2 is specifically as follows: The real-time oxygen consumption intensity is used as a feedforward gain to correct the hysteresis of the partial pressure sampled dissolved oxygen value in the gas-liquid mass transfer process; and the Reynolds number calculated based on the stirring shaft power is used to dynamically interpolate and verify the local oxygen supply gain coefficient at each discrete sampling node of the flow field.

4. The continuous methanol single-cell protein production method according to claim 1, characterized in that, The liquid phase control in step S4 specifically includes: A single feeding action is defined as a discrete feeding instance, and a spatial response attribute is configured for each discrete feeding instance; Based on the differences in substrate demand in different coordinate regions of the virtual metabolic field map, differentiated pulse frequency offsets are assigned to independent feed ports located at different positions. By adjusting the trigger frequency of the discrete feeding instances on the time axis, dynamic smoothing of the local nutrient concentration around each discrete sampling node of the flow field can be achieved.

5. The continuous methanol single-cell protein production method according to claim 1, characterized in that, The gas-phase control in step S4 specifically includes: The central control unit calculates the spatial flux weighting factor for the corresponding gas supply sector based on the local oxygen supply gain coefficient of each discrete sampling node of the flow field. The central control unit identifies metabolically inactive regions through the virtual metabolic field map and, based on the spatial flux weighting factor, dynamically increases the air volume allocation ratio of the air intake sector to which the metabolically inactive region belongs, while keeping the total air intake constant. By adjusting the opening degree of the proportional solenoid valve of the corresponding sector, the critical flow velocity of the gas when passing through the micro-orifice of the distributor is changed, thereby achieving spatial gradient compensation for the local dissolved oxygen rate around the discrete sampling node of the flow field.

6. The continuous methanol single-cell protein production method according to claim 1, characterized in that, It also includes physicochemical analysis scheduling based on monitoring window clipping, specifically: Define a key feature window for the fermentation process, which is determined based on the rate of change of parameters at each sampling node in the virtual metabolite map; When the virtual metabolic field map shows that the global environmental parameters are within the preset steady-state range, the sampling frequency of the physicochemical analysis task is reduced, thereby reducing the operating load of the ICP intelligent monitoring system. When the attribute fluctuation of any discrete sampling node of the flow field in the virtual metabolic field map exceeds the threshold, the high-frequency monitoring mode is activated.

7. A continuous methanol single-cell protein production device, characterized in that, It includes the main fermentation unit, sensing unit, execution unit, and central control unit; The main fermentation unit is a continuous three-phase reaction fermentation device with an internal three-dimensional reaction space, which is used to provide a physical site for gas-liquid-solid three-phase biochemical reactions. The sensing unit is used to acquire fermentation kinetics and metabolic parameters in real time, including multiple dissolved oxygen detection devices, exhaust gas detection devices and stirring shaft power detection devices that are spatially arrayed. The execution unit is used to perform differentiated resource allocation according to instructions, including a segmented gas distribution device and a multi-point feeding system; The central control unit is used to receive data from the sensing unit and calculate and generate a virtual metabolic field map, thereby driving the execution unit to perform partition compensation. The central control unit stores the empty flow field calibration vector.

8. The continuous methanol single-cell protein production apparatus according to claim 7, characterized in that, The fermentation device is equipped with a stirring device on its central shaft, which is used to stir the materials inside the fermentation device. The stirring device includes a stirring shaft set on the central axis of the fermenter, a stirring drive motor at one end of the stirring shaft, and three layers of stirring paddles installed in series along the axial direction. The three layers of stirring paddles divide the fermenter into a three-layer structure of bottom, middle and top layers in the vertical direction.

9. The continuous methanol single-cell protein production apparatus according to claim 7, characterized in that, The segmented gas distribution device is used to adjust the oxygen supply intensity of different sectors according to the radial distribution differences of the virtual metabolic field map, including: At least four independent gas supply sectors arranged in a circular array, each equipped with an independent proportional solenoid valve and flow control unit.

10. A continuous methanol single-cell protein production apparatus according to claim 7, characterized in that, The multi-point feeding system includes: At least three independent feeding branches are set at different height levels along the axis of the fermentation device; the feeding branches are used to match the layering of the three-dimensional volumetric mesh structure in the height dimension, and independently execute variable frequency pulse feeding according to the axial concentration difference of the virtual metabolic field map.

Images