Cell parallel culture environment regulation and control method and system for parallel biological reaction equipment

Abstract

The invention provides a cell parallel culture environment regulation and control method and system for parallel biological reaction equipment, and relates to the field of cell culture equipment and intelligent process control, and the system comprises a plurality of culture tanks arranged in parallel, independent gas supply branches corresponding to the culture tanks, a gas source unit, a gas path control unit, a parameter detection unit and a process control unit; each gas supply branch is connected with a gas source unit and is sequentially provided with a pressure reducing assembly, a filtering assembly, a proportional valve and a flow sensor so as to realize independent closed-loop control of the gas supply flow of each culture tank; a dissolved oxygen sensor, a pH sensor and a temperature sensor are arranged in each culture tank and are used for monitoring culture microenvironment parameters in real time. The process control unit establishes a mapping relation between culture process parameters and actuator control parameters based on a deep learning model, and cooperates with bottom layer gas flow closed-loop control to realize accurate, stable and non-interfering regulation and control of the multi-culture tank micro-physiological environment. The method is suitable for high-throughput cell parallel culture and process optimization scenes.

Description

Technical Field

[0001] This invention relates to the field of cell culture equipment and intelligent process control technology, and in particular to a method and system for regulating the parallel cell culture environment in parallel bioreactors. Background Technology

[0002] Cell culture processes place extremely high demands on the stability and controllability of the culture microenvironment. Key process parameters such as dissolved oxygen (DO), pH, temperature, and shear stress directly affect cell proliferation efficiency, differentiation status, and functional consistency. With the increasing demand for high-throughput cell culture and process screening, single culture tanks or centralized gas supply and control methods are no longer sufficient to meet the experimental and production needs of multiple culture units operating in parallel and with distinguishable parameters.

[0003] In the prior art, multi-culture tank systems usually adopt a shared gas source, a shared control strategy or a simple proportional control method, which has the following shortcomings: (1) the gas path coupling between each culture tank is serious, making it difficult to achieve truly independent control; (2) traditional PID or rule control methods are difficult to cope with the biological culture process with multiple variables, strong coupling and time-varying characteristics; (3) lack the ability to predict the evolution trend of process parameters, and the control lag is obvious, affecting the stability and repeatability of culture.

[0004] Therefore, there is an urgent need for a method to regulate the parallel cell culture environment in order to achieve independent, precise and adaptive adjustment of process parameters in multiple culture tanks, thereby improving the intelligence level and process consistency of cell culture systems. Summary of the Invention

[0005] To achieve the above-mentioned objectives and other advantages of the present invention, a first objective of the present invention is to provide a cell parallel culture environment control system for a parallel biological reaction device, comprising: Multiple culture jars configured in parallel; The gas source unit includes at least air, oxygen, carbon dioxide, and nitrogen gas sources; Multiple independent gas supply branches are provided, with the number corresponding to each culture tank. Each independent gas supply branch independently delivers the gas from the gas source unit to the corresponding individual culture tank. Each independent gas supply branch is equipped with a proportional valve and a flow sensor for closed-loop control of gas flow in a single tank. A parameter detection unit is installed in each of the culture tanks to monitor the culture microenvironment parameters in real time, which include at least dissolved oxygen concentration, pH value and temperature. The gas circuit control unit, which is communicatively connected to the proportional valve and the flow sensor on each of the independent gas supply branches, is configured to execute the underlying gas flow closed-loop control algorithm. The process control unit is communicatively connected to the parameter detection unit and the gas path control unit; the process control unit is configured to: The system operates an intelligent control model based on deep learning, which maps the real-time monitored microenvironment parameters to actuator control parameters and coordinates with the gas path control unit to achieve independent regulation of the microenvironment of each culture tank.

[0006] Furthermore, on each of the independent gas supply branches, a pressure reducing component and a high-efficiency air filter are sequentially arranged upstream of the proportional valve; and / or, a terminal filter is arranged between the proportional valve and the corresponding culture tank.

[0007] Furthermore, the actuator control parameters output by the process control unit act on the pneumatic control unit to adjust the opening degree of the proportional valve.

[0008] Furthermore, the system also includes a stirring device and a temperature control module corresponding to each culture tank; The actuator control parameters include the gas valve opening degree, stirring speed, and temperature control module drive voltage.

[0009] Furthermore, the intelligent control model is a multi-layer neural network model, which is obtained by training with historical cultivation data and is used to establish a nonlinear mapping relationship between the cultivation microenvironment parameters or their deviations and the actuator control parameters; The process control unit is further configured to: dynamically predict and output the optimal actuator control parameters based on the deviation between the real-time collected culture microenvironment parameters and the preset target values ​​through the intelligent control model.

[0010] Furthermore, the process control unit and the gas path control unit work together to form a feedforward-feedback composite control structure; The intelligent control model provides feedforward predictive control, and the gas path control unit provides PID regulation based on real-time flow feedback.

[0011] A second objective of this invention is to provide a method for regulating the parallel culture environment of cells in a parallel bioreactor, applied to the aforementioned system, comprising the following steps: Acquire real-time microenvironmental parameters for each culture tank, including at least dissolved oxygen concentration, pH value, and temperature; The microenvironment parameters are input into an intelligent control model trained based on deep learning, and the model outputs actuator control parameters for each culture tank. Based on the actuator control parameters and in coordination with the underlying gas flow closed-loop control, independent regulation of the microenvironment of each culture tank can be achieved.

[0012] Furthermore, the regulation process includes: When the fluctuation of the microenvironment parameters is within the linear or preset range, the underlying PID closed-loop control algorithm is used for adjustment. When the microenvironment parameters exhibit a nonlinear changing trend, the intelligent control model dynamically adjusts the control strategy.

[0013] Furthermore, the gas supply flow control quantity using the underlying PID closed-loop control algorithm satisfies the following relationship: in, The gas flow rate is controlled for the i-th culture tank. The deviation between the target training parameters and the real-time parameters, The proportional, integral, and derivative coefficients are adaptively adjusted by the intelligent control model based on historical training data.

[0014] Furthermore, the intelligent control model establishes a microenvironment parameter vector. With actuator control parameter vector Mapping relationship between them: in, Including dissolved oxygen, pH, and temperature parameters, This includes the gas valve opening, stirring speed, and temperature control module drive parameters. and These are the weights and bias parameters, respectively.

[0015] Furthermore, the intelligent control model is trained by minimizing a loss function and outputs the optimal actuator control strategy during online inference, forming a feedforward-feedback composite control structure in conjunction with the underlying PID control; wherein, the minimized loss function is: .

[0016] A third objective of the present invention is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0017] A fourth objective of the present invention is to provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.

[0018] Compared with the prior art, the beneficial effects of the present invention are: This invention provides a method and system for regulating the parallel cell culture environment in parallel bioreactors. It achieves physical and logical independence in gas supply and control for multiple culture tanks, introduces a deep learning-based intelligent process control model, improves regulation accuracy and response speed, and can adapt to the nonlinear, time-varying, and multivariate coupling characteristics of the cell culture process. It is suitable for high-throughput parallel cell culture and process optimization scenarios.

[0019] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0020] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the structure of a cell culture vessel. Figure 2 This is a schematic diagram of the process control unit; Figure 3 Diagram of deep learning model architecture; Figure 4 A flowchart of a method for regulating the parallel culture environment of cells used in parallel bioreactors; Figure 5 A schematic diagram of computer equipment; Figure 6 This is a schematic diagram of a computer-readable storage medium. Detailed Implementation

[0021] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0022] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0023] The drawing numbers in this application are only used to distinguish the steps in the scheme and are not used to limit the execution order of the steps. The specific execution order is as described in the specification.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0025] This invention provides an independent gas path and intelligent process control system and method for parallel cell culture. By constructing independent gas supply branches corresponding one-to-one with each culture vessel, and introducing a deep learning-based intelligent control model and a low-level closed-loop gas regulation strategy, independent, precise, and non-interfering control of the microphysiological environment parameters of multiple culture vessels is achieved. The specific scheme is as follows: Example 1 A cell parallel culture environment control system for parallel bioreactors, such as Figure 1 As shown, it includes: Multiple culture jars configured in parallel; The gas source unit includes at least air, oxygen, carbon dioxide, and nitrogen gas sources; Multiple independent gas supply branches are provided, with the number corresponding to each culture tank. Each gas supply branch is connected to a single culture tank and a gas source unit. Each independent gas supply branch independently delivers gas from the gas source unit to the corresponding single culture tank. Each independent gas supply branch is equipped with a proportional valve and a flow sensor for closed-loop control of gas flow in a single tank. A parameter detection unit is installed in each of the culture tanks and includes at least a dissolved oxygen sensor, a pH sensor and a temperature sensor. It is used to monitor the key process parameters of the culture microenvironment during the culture process in real time, namely the culture microenvironment parameters, which include at least dissolved oxygen concentration, pH value and temperature. The gas circuit control unit, which is communicatively connected to the proportional valve and the flow sensor on each of the independent gas supply branches, is configured to execute the underlying gas flow closed-loop control algorithm. The process control unit is communicatively connected to the parameter detection unit and the gas path control unit; the process control unit is configured to: The system runs an intelligent control model built on deep learning, processes the real-time acquired process parameters, maps them to corresponding actuator control parameters, including but not limited to gas valve opening, stirring speed and temperature control module drive voltage, and coordinates with the gas path control unit to achieve intelligent regulation of the microenvironment of each culture tank.

[0026] To improve response speed and control accuracy, a robust system is constructed, providing an ideal engineering framework for adaptation and self-learning. The process control unit and the gas path control unit work together to form a feedforward-feedback composite control structure. The intelligent control model provides feedforward predictive control, and the gas path control unit provides PID regulation based on real-time flow feedback.

[0027] This architecture integrates the advantages of two classic control strategies. The intelligent control model, acting as an advanced feedforward controller, can predict and issue globally optimal control commands (such as target gas flow rates) in advance based on real-time monitored deviations in cultivation parameters, rapidly compensating for macroscopic dynamics and significantly shortening system response time. Simultaneously, the gas path control unit, as a low-level feedback (PID) controller, focuses on the precise execution of feedforward commands. Utilizing high-speed feedback from the flow sensor, it instantaneously fine-tunes the proportional valve opening to resist minor pipeline disturbances, ensuring ultra-high steady-state accuracy of flow control. The combination of these two approaches enables the system to respond quickly and accurately target the complex changes in the cultivation process.

[0028] This architecture clearly defines the functions of the control layers. The upper-layer deep learning model is responsible for complex decision-making, prediction, and multivariate optimization; the lower-layer PID pneumatic control is responsible for high-bandwidth, high-reliability execution. This division of labor greatly enhances the system's robustness. Even when the upper-layer model is updated online or encounters entirely new operating conditions, the reliable PID closed-loop at the bottom layer can still ensure basic stable operation and prevent runaway. At the same time, the precise execution at the bottom layer provides a reliable foundation for the predictions of the upper-layer model, allowing it to focus on generating better strategies.

[0029] This composite control structure is an ideal platform for implementing advanced control algorithms. The feedforward path provides a direct output interface for the online learning and real-time prediction of the deep learning model, enabling the model to continuously self-optimize based on the control effect. The feedback path provides a stable and measurable benchmark closed loop, which can accurately evaluate the effect of the feedforward control and generate high-quality feedback data for model training. This structure transforms the entire system from an automated device executing a preset strategy into an intelligent biological reaction platform capable of continuously learning and evolving from operational data.

[0030] To achieve both physical and logical independence, stable gas quality and pressure, and precise independent control capabilities, each independent gas supply branch is equipped with a pressure reducing assembly (including a pressure reducer and a pressure regulating valve) and a high-efficiency air filter (such as a HEPA filter) upstream of the proportional valve; and / or, a terminal filter (such as a needle filter) is installed between the proportional valve and the corresponding culture tank. Each gas supply branch is independent in both physical structure and control logic. After being output from compressed gas cylinders, each gas source undergoes pretreatment via a pressure reducer, HEPA filter, and pressure regulating valve before being distributed to multiple independent gas supply branches. Furthermore, the proportional valve, flow sensor, and needle filter installed sequentially before entering the culture tank enable independent control of the gas flow rate and gas composition for a single culture tank.

[0031] Each gas supply branch is completely isolated in physical structure and forms its own closed loop in control logic, ensuring that the gas supply and regulation between multiple culture tanks do not interfere with each other, fundamentally avoiding cross-coupling of gas components and flow rates, and providing a basis for conducting differentiated process experiments in parallel.

[0032] Through a three-stage pretreatment process involving a pressure reducer, HEPA filter, and pressure regulator, the gas source pressure is not only reduced and stabilized within the working range, but also particulate matter and microorganisms are effectively filtered out, providing a clean, high-quality gas source with constant pressure for cell culture and ensuring the basic stability of the culture environment.

[0033] On each branch, the coordinated operation of proportional valves, flow sensors, and terminal filters (such as needle filters) constructs a high-precision closed-loop gas flow control unit. This enables the system to independently and precisely adjust the inlet flow rate and gas composition according to the real-time needs of each culture tank, achieving customized and precise control of the microphysiological environment of individual culture tanks.

[0034] In order to construct a hierarchical and modular intelligent control architecture, realize the advantages of feedforward and feedback control, and enhance the accuracy and reliability of control, the actuator control parameters output by the process control unit act on the pneumatic control unit to adjust the opening of the proportional valve.

[0035] The process control unit, as the upper-level intelligent decision-making center, focuses on multivariable, nonlinear optimization based on deep learning; the gas path control unit, as the lower-level execution core, is responsible for high-response, single-variable (flow rate) closed-loop precise regulation. This hierarchical design decouples complex algorithms from real-time control tasks, improving the overall stability, reliability, and maintainability of the system.

[0036] The upper-level intelligent model provides the optimal control strategy (actuator control parameters) in advance (feedforward) based on the changing trends of the cultivation parameters; the lower-level gas path control unit, based on real-time feedback from the flow sensor, performs rapid and precise fine-tuning of the proportional valve opening (feedback). The two work together to form a composite control, which significantly improves the system's response speed and control accuracy to dynamic changes in the cultivation process, and reduces overshoot and oscillation.

[0037] Actuator control parameters (such as the target gas flow rate) are ultimately driven by a specialized gas path control unit, rather than being directly controlled. This ensures that the control of the critical actuator (proportional valve) is accomplished by a module specializing in fast, stable closed-loop regulation, thereby achieving intelligent optimization while guaranteeing the accuracy and dynamic performance of the underlying gas flow regulation, making the entire control system both intelligent and reliable.

[0038] To achieve rapid, stable, and non-interfering control of the gas supply to each culture tank, this invention constructs an independent closed-loop control circuit for gas flow on each gas supply branch. For the first... The gas flow control model for a single gas supply branch is expressed as follows: in, For the first One gas supply branch at time Gas flow rate; The flow deviation of this gas supply branch is defined as the difference between the set flow rate and the real-time measured flow rate. , and These are the proportional, integral, and derivative control coefficients, respectively. The control coefficients... , and The process control unit can adaptively adjust based on historical operating data and cultivation status, providing a stable and predictable execution basis for upper-level intelligent control.

[0039] Closed-loop regulation of gas flow in a single culture tank is achieved through the coordinated action of a proportional valve and a flow sensor. Each gas supply branch is independent in physical structure and control logic, thereby avoiding coupling interference during the gas supply process.

[0040] In order to form a multivariate synergistic regulation capability for key microenvironment parameters, establish a globally optimal control strategy for complex biological processes, and significantly improve the consistency and reproducibility of the culture process, the system also includes a stirring device and a temperature control module corresponding to each culture tank. The actuator control parameters include the gas valve opening degree, stirring speed, and temperature control module drive voltage.

[0041] This invention is not limited to gas control, but incorporates the core physical factors affecting cell culture—temperature and the fluid dynamic environment (shear force)—into a unified intelligent control framework. By using the stirring speed, the driving voltage of the temperature control module, and the gas valve opening as actuator control parameters, the system can simultaneously optimize multiple key parameters such as dissolved oxygen, pH, temperature, and shear force generated by stirring, achieving comprehensive, integrated, and precise control of the cell culture microphysiological environment.

[0042] Deep learning models can learn and map the complex nonlinear, coupled relationships between multivariate inputs (such as dissolved oxygen, pH, and temperature) and multivariate outputs (gas, stirring, heating / cooling). This enables the system to dynamically calculate and output a set of optimal control commands that bring the overall culture environment closer to the set target based on the real-time process status, rather than adjusting individual parameters in isolation. This significantly improves the overall integrity and intelligence of complex biological process control.

[0043] By integrating stirring and temperature control in a closed-loop system, the system eliminates the operational fluctuations and lags caused by traditional manual or independent control methods. Precise speed control ensures stable mixing efficiency and shear force, while precise temperature control maintains consistency in enzyme activity and metabolic rate. These elements, combined with gas control, provide highly stable, uniform, and controllable culture conditions for cells and other microenvironment-sensitive organisms, greatly improving the reproducibility and reliability of experimental results across different batches and culture units, laying a solid foundation for high-throughput process development and large-scale production.

[0044] To achieve nonlinear, adaptive control of complex biological processes, and to realize a leap from passive response to active prediction in intelligent control, significantly improving control accuracy and process optimization efficiency, the intelligent control model is a multi-layer neural network model. It is trained using historical culture data to establish a nonlinear mapping relationship between the culture microenvironment parameters or their deviations and the actuator control parameters. The process control unit is further configured to dynamically predict and output the optimal actuator control parameters based on the deviations between the real-time collected culture microenvironment parameters and preset target values, using the intelligent control model.

[0045] Traditional PID or rule-based control struggles to effectively handle the complex dynamics of multi-parameter coupling, time-varying, and nonlinear processes in cell culture. This invention employs a multi-layer neural network model, capable of autonomously mining and establishing deep nonlinear mappings between culture environment parameters (or their deviations) and multi-actuator control commands by learning from vast amounts of historical data. This endows the system with powerful adaptive capabilities, enabling it to automatically adjust control strategies for different stages and dynamic changes in the culture process, effectively addressing complex conditions that traditional methods struggle to handle.

[0046] Unlike traditional feedback control (which adjusts only after a deviation occurs), the intelligent control model of this invention can dynamically predict and output a set of forward-looking optimal control parameters based on real-time collected parameter deviations and learned process dynamics. This predictive control can compensate for system disturbances in advance, significantly reduce control lag, and control the fluctuation range of key parameters (such as dissolved oxygen and pH) within a narrower range, thereby providing cells with an extremely stable microenvironment, which is crucial for maintaining cell pluripotency and promoting uniform growth.

[0047] The optimal actuator control parameters output by this model are data-driven global optimization solutions, rather than experience-based local adjustments. This ensures that, under any given operating condition, the system can calculate a control action combination that approximates the theoretical optimal combination, thereby achieving ultra-high precision control of microenvironment parameters. Furthermore, this data-driven model itself can continuously iterate and optimize with the accumulation of new data, constantly improving its control performance. It provides a powerful intelligent tool for quickly finding and locking in optimal cultivation process parameters, significantly shortening the process development cycle.

[0048] The principle of deep learning models is as follows Figure 3 As shown, the original samples include cell culture process parameters and actuator control parameters, randomly divided into training and test sets. In the initial training phase, the neural network randomly initializes its internal weights, and the process parameters and control parameters are activated by the neuron activation function. Weight Establish a mapping relationship between the bias and the process parameters. Input the process parameters into the neural network, using the loss function... With minimum as the optimization objective, the backpropagation algorithm is used to optimize its internal weights. Repeated iterations approximate the true data distribution. The model completes training set learning and verifies the predictive performance of the neural network structure using a test set. Using the tested neural network structure as the prediction model, online detection process parameters during culture are input to calculate the optimal actuator control parameters, thereby achieving precise regulation of the cellular microphysiological environment.

[0049] Specifically, suppose the system includes The first culture jar, for the first The key process parameters for each culture tank are defined as follows: Dissolved oxygen: pH: ,temperature: Shear force: The corresponding target settings are as follows: , , and Construct the process parameter deviation vector: Process control unit based on deep learning model Establish a mapping relationship between process parameter deviations and actuator control parameters: The actuator control parameter vector is as follows: These correspond to the gas valve opening adjustment command, the stirring speed adjustment command, and the temperature control module drive voltage, respectively.

[0050] The deep learning model adopts a multi-layer neural network structure, and the calculation expression of a single-layer neuron is as follows: Construct a loss function with the objective of minimizing future process parameter deviations: The model parameters are updated using the Backpropagation Through Time (BPTT) algorithm: After training, the model runs as an online prediction model, outputting the optimal control strategy for each actuator in real time, and working in conjunction with the underlying gas path closed-loop control.

[0051] In some embodiments, the culture chamber includes multiple culture tanks, a gas supply unit, a temperature control unit, and a drive unit, primarily used to provide the gas / liquid, temperature, and power required for cell culture. After the culture tanks are installed, the reaction flask caps are sealed, and the stirring drive assembly connects the power source and the culture tank's stirring paddle. Air, oxygen, carbon dioxide, and nitrogen are compressed in steel cylinders. The cylinder outlets, after passing through a pressure reducer, a HEPA filter, and a pressure regulating valve, are divided into eight branches, each connected to a different culture tank. Each culture tank's gas branch is connected in series with a proportional valve, a flow sensor, and a needle filter before entering the culture tank and then being discharged into the atmosphere. On each branch entering the culture tank, a proportional valve, in conjunction with a flow sensor, is used for closed-loop control of the gas flow rate for a single path; the flow control for each culture tank is independent. A pH sensor and an O2 concentration sensor are installed in each culture tank to measure the concentration of the corresponding gas in the supply gas, ensuring that the supply gas concentration meets the culture requirements. When the gas concentration exceeds the set range, the system issues an alarm.

[0052] Process control unit such as Figure 2 As shown, a deep learning-based control model is adopted, and key process parameters (dissolved oxygen, pH, temperature and shear force) affecting cell culture are selected as control targets. The internal weights are optimized by processing and learning input data using a multi-level neural network structure, and the mapping relationship between culture process parameters and actuator control parameters (rotation speed, valve opening and Peltier voltage) is determined to predict the optimal control strategy.

[0053] This invention achieves physical isolation at the gas supply level through independent gas supply branches, enables rapid and stable single-variable control through bottom-level flow closed-loop, and realizes prediction and optimization of multi-variable coupled process parameters through upper-level intelligent control. The control loops of each culture tank operate in parallel without interfering with each other, thus achieving independent and stable control of multiple culture tanks.

[0054] Example 2 A method for regulating the parallel cell culture environment in a parallel bioreactor is provided, applied to the system described above. For a detailed description of the system, please refer to the corresponding description in the system embodiments described above; it will not be repeated here. Figure 4 As shown, the method includes the following steps: S100. Acquire real-time microenvironmental parameters of each culture tank, wherein the microenvironmental parameters include at least dissolved oxygen concentration, pH value and temperature; Specifically, by means of parameter detection units installed in each of the culture tanks, key process parameters of the culture microenvironment, i.e., culture microenvironment parameters, are monitored in real time during the culture process. S200. The microenvironment parameters are input into the intelligent control model trained by deep learning, and the actuator control parameters for each culture tank are output. The parameters are then applied to the gas supply, stirring and temperature control modules of each culture tank, which can form a closed-loop intelligent control process in which multiple culture tanks operate in parallel and are independent of each other.

[0055] S300: Based on the actuator control parameters and in coordination with the underlying gas flow closed-loop control, independent regulation of the microenvironment of each culture tank is achieved.

[0056] In the system, the process control unit and the gas path control unit work together to form a feedforward-feedback composite control structure. The intelligent control model provides feedforward predictive control, while the gas path control unit provides PID regulation based on real-time flow feedback.

[0057] To achieve optimal allocation of control resources and maximize system efficiency, enhance the system's adaptability at different cultivation stages, and construct a safety assurance mechanism that balances robustness and intelligence, the regulation process includes: When the fluctuation of the microenvironment parameters is within the linear range or a preset range (small disturbance range), the underlying PID closed-loop control algorithm is used for adjustment. When the microenvironment parameters exhibit a nonlinear changing trend, the intelligent control model dynamically adjusts the control strategy to achieve precise, stable, and independent regulation of the microenvironment in multiple culture tanks.

[0058] The aforementioned control process intelligently allocates control tasks based on the nature (linear / nonlinear) of microenvironment parameter fluctuations. For normal, small-amplitude linear disturbances, the computationally efficient and responsive underlying PID closed-loop control directly handles them. This fully leverages the reliability and economy of the PID algorithm in dealing with small disturbances, avoiding frequent calls to complex intelligent models and reducing the system's computational load. When a nonlinear, large-amplitude trend in parameters is detected, the system automatically switches to the intelligent control model, utilizing its powerful ability to handle complex, nonlinear dynamics, ensuring that the system always operates in the most efficient and economical manner.

[0059] Cell culture typically involves different growth phases (such as the inoculation phase, logarithmic growth phase, and stationary phase), each with significantly different metabolic activities and response patterns to environmental disturbances. The adaptive switching strategy of this invention perfectly matches this characteristic. During metabolically stable phases, PID control maintains basic stability; during critical phases with drastic metabolic changes and prone to nonlinear fluctuations (such as the rapid proliferation phase), an intelligent model takes over, providing a higher level of customized control. This ensures that the system can adapt to the inherent dynamic changes of the culture process throughout, providing optimal control for each phase.

[0060] The aforementioned control process essentially constructs a flexible and reliable control safety net. PID control, as a stable and mature baseline control, ensures the system's basic stable operation under any circumstances (the foundation of robustness). The intelligent model, as a performance enhancement layer, significantly improves control quality when conditions permit (the manifestation of intelligence). The smooth or conditional switching between the two prevents the risks that may arise from the failure of a single control mode. For example, when the intelligent model is updated online or encounters extreme and unseen situations, the system can rely on the underlying PID control to maintain stability, thereby greatly improving the safety and success rate of the entire breeding process.

[0061] To achieve rapid, stable, and non-interfering control of the gas supply to each culture tank, this invention constructs an independent closed-loop control circuit for gas flow on each gas supply branch. For the first... The gas flow control model for a single gas supply branch is expressed as follows: in, For the first One gas supply branch at time Gas flow rate; The flow deviation of this gas supply branch is defined as the difference between the set flow rate and the real-time measured flow rate. , and These are the proportional, integral, and derivative control coefficients, respectively. The control coefficients... , and The process control unit can adaptively adjust based on historical operating data and cultivation status, providing a stable and predictable execution basis for upper-level intelligent control.

[0062] Preferably, the intelligent control model is a multi-layer neural network model, which is obtained through training on historical culture data and is used to establish a nonlinear mapping relationship between the culture microenvironment parameters or their deviations and the actuator control parameters. Specifically, step S200 involves: dynamically predicting and outputting the optimal actuator control parameters based on the deviations between the real-time collected culture microenvironment parameters and preset target values ​​using the intelligent control model.

[0063] Specifically, the intelligent control model establishes a microenvironment parameter vector. With actuator control parameter vector Mapping relationship between them: in, Including dissolved oxygen, pH, and temperature parameters, This includes the gas valve opening, stirring speed, and temperature control module drive parameters. and These are the weights and bias parameters, respectively.

[0064] To establish a data-driven, goal-oriented model optimization paradigm, achieve efficient collaboration between offline training and online inference, and form a performance-quantifiable and evolvable intelligent control closed loop, the intelligent control model is trained by minimizing a loss function and outputs the optimal actuator control strategy during online inference, forming a feedforward-feedback composite control structure in conjunction with the underlying PID control; wherein, the minimized loss function is: .

[0065] This invention provides an independent gas path control system for parallel cell culture and a process control method based on intelligent algorithms. Through a multi-source independent gas supply structure and a multi-level intelligent control algorithm, it achieves independent, precise and adaptive regulation of microenvironment parameters in multiple culture tanks.

[0066] Example 3 A computer device 400, such as Figure 5 As shown, the device includes a memory 410, a processor 420, and a computer program 430 stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a method for regulating the parallel culture environment of cells in a parallel bioreactor. For a detailed description of the method, please refer to the corresponding description in the above method embodiments; it will not be repeated here.

[0067] Example 4 A computer-readable storage medium, such as Figure 6 As shown, a computer program is stored thereon, which, when executed by a processor, implements the steps of a method for regulating the parallel culture environment of cells in a parallel bioreactor. For a detailed description of the method, please refer to the corresponding description in the above method embodiments, and will not be repeated here.

[0068] The number of devices and processing scale described herein are for the purpose of simplifying the description of the invention. Applications, modifications, and variations of the invention will be readily apparent to those skilled in the art.

[0069] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

[0070] The apparatus, computer device, and non-volatile computer storage medium and method provided in the embodiments of this specification are corresponding. Therefore, the apparatus, computer device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, computer device, and non-volatile computer storage medium will not be repeated here.

[0071] Those skilled in the art will also know that, besides implementing the controller in the form of purely computer-readable program code, the same functions can be achieved by logically programming the method steps, making the controller take the form of logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers (PLCs), and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the devices included within it for implementing various functions can also be considered structures within that hardware component. Alternatively, the devices for implementing various functions can be considered as both software units implementing the method and structures within a hardware component.

[0072] The systems, apparatuses, or units described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. For ease of description, the above apparatuses are described separately as various units based on their functions. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0073] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0074] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0075] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0076] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0077] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0078] This specification may be described in the general context of computer-executable instructions, such as program units, that are executed by a computer. Generally, program units include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification may also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program units may reside in local and remote computer storage media, including storage devices.

[0079] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0080] The above description is merely an embodiment of this specification and is not intended to limit the scope of one or more embodiments of this specification. Various modifications and variations can be made to one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of one or more embodiments of this specification.

Claims

1. A cell parallel culture environment control system for parallel bioreactors, characterized in that, include: Multiple culture jars configured in parallel; The gas source unit includes at least air, oxygen, carbon dioxide, and nitrogen gas sources; Multiple independent gas supply branches are provided, with the number corresponding to each culture tank. Each independent gas supply branch independently delivers the gas from the gas source unit to the corresponding individual culture tank. Each independent gas supply branch is equipped with a proportional valve and a flow sensor for closed-loop control of gas flow in a single tank. A parameter detection unit is installed in each of the culture tanks to monitor the culture microenvironment parameters in real time, which include at least dissolved oxygen concentration, pH value and temperature. The gas circuit control unit, which is communicatively connected to the proportional valve and the flow sensor on each of the independent gas supply branches, is configured to execute the underlying gas flow closed-loop control algorithm. The process control unit is communicatively connected to the parameter detection unit and the gas path control unit; the process control unit is configured to: The system operates an intelligent control model based on deep learning, which maps the real-time monitored microenvironment parameters to actuator control parameters and coordinates with the gas path control unit to achieve independent regulation of the microenvironment of each culture tank.

2. The cell parallel culture environment control system for a parallel bioreactor as described in claim 1, characterized in that, On each of the independent gas supply branches, a pressure reducing component and a high-efficiency air filter are sequentially installed upstream of the proportional valve; and / or, a terminal filter is installed between the proportional valve and the corresponding culture tank.

3. The cell parallel culture environment control system for a parallel bioreactor as described in claim 1, characterized in that, The actuator control parameters output by the process control unit act on the pneumatic control unit to adjust the opening of the proportional valve.

4. The cell parallel culture environment control system for a parallel bioreactor as described in claim 1, characterized in that, The system also includes a stirring device and a temperature control module corresponding to each culture tank; The actuator control parameters include the gas valve opening degree, stirring speed, and temperature control module drive voltage.

5. The cell parallel culture environment control system for a parallel bioreactor as described in claim 1, characterized in that, The intelligent control model is a multi-layer neural network model, which is obtained by training through historical cultivation data and is used to establish a nonlinear mapping relationship between the cultivation microenvironment parameters or their deviations and the actuator control parameters. The process control unit is further configured to: dynamically predict and output the optimal actuator control parameters based on the deviation between the real-time collected culture microenvironment parameters and the preset target values ​​through the intelligent control model.

6. A cell parallel culture environment control system for a parallel bioreactor as described in claim 1 or 3, characterized in that, The process control unit and the gas path control unit work together to form a feedforward-feedback composite control structure; The intelligent control model provides feedforward predictive control, and the gas path control unit provides PID regulation based on real-time flow feedback.

7. A method for regulating the parallel cell culture environment in a parallel bioreactor, applied to the system as described in any one of claims 1 to 6, characterized in that, Includes the following steps: Acquire real-time microenvironmental parameters for each culture tank, including at least dissolved oxygen concentration, pH value, and temperature; The microenvironment parameters are input into an intelligent control model trained based on deep learning, and the model outputs actuator control parameters for each culture tank. Based on the actuator control parameters and in coordination with the underlying gas flow closed-loop control, independent regulation of the microenvironment of each culture tank can be achieved.

8. The method for regulating the parallel cell culture environment in a parallel bioreactor as described in claim 7, characterized in that, The regulation process includes: When the fluctuation of the microenvironment parameters is within the linear or preset range, the underlying PID closed-loop control algorithm is used for adjustment. When the microenvironment parameters exhibit a nonlinear changing trend, the intelligent control model dynamically adjusts the control strategy.

9. The method for regulating the parallel cell culture environment in a parallel bioreactor as described in claim 8, characterized in that, The gas supply flow control quantity using the underlying PID closed-loop control algorithm satisfies the following relationship: in, The gas flow rate is controlled for the i-th culture tank. The deviation between the target training parameters and the real-time parameters, The proportional, integral, and derivative coefficients are adaptively adjusted by the intelligent control model based on historical training data.

10. A method for regulating the parallel culture environment of cells in a parallel bioreactor as described in claim 8, characterized in that, The intelligent control model establishes a microenvironment parameter vector. With actuator control parameter vector Mapping relationship between them: in, Including dissolved oxygen, pH, and temperature parameters, This includes the gas valve opening, stirring speed, and temperature control module drive parameters. and These are the weights and bias parameters, respectively.

11. The method for regulating the parallel cell culture environment in a parallel bioreactor as described in claim 10, characterized in that, The intelligent control model is trained by minimizing a loss function and outputs the optimal actuator control strategy during online inference, forming a feedforward-feedback composite control structure in conjunction with the underlying PID control; wherein, the minimized loss function is: 。 12. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 7 to 11.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 7 to 11.

Images