Data processing system of stem cell culture equipment

By building a data processing system for stem cell culture equipment and using multi-source sensors and nonlinear state transition models for prediction and feedback correction, the problems of weak multi-parameter adjustment capabilities and control drift in existing equipment are solved, and efficient and flexible environmental adjustment is achieved.

CN120630681AInactive Publication Date: 2025-09-12SOUTH MEDICAL BIOLOGY (SHENZHEN) CO LTD
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202510760053.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-06-09
Publication Date
2025-09-12
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing stem cell culture equipment lacks the ability to capture the dynamic evolution of the complex coupling relationship between multiple parameters. The control strategies are mostly based on fixed rules or single variable adjustment, resulting in slow adjustment or excessive response, and lack of closed-loop feedback mechanism, which makes control drift prone.

Method used

A data processing system is adopted, including data acquisition, state modeling, trend prediction, optimization control and closed-loop feedback modules. Environmental state parameters and control operation parameters are obtained through multi-source sensors, state matrix and control matrix are constructed, and prediction is made using nonlinear state transition model. The model parameters are corrected through closed-loop feedback to achieve multi-parameter coordinated adjustment.

Benefits of technology

Real-time dynamic monitoring and high-precision adjustment of the stem cell culture environment are achieved, which enhances the adaptability and generalization ability of the system, effectively corrects control drift, and improves the flexibility and efficiency of control.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120630681A_ABST
    Figure CN120630681A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of data processing of stem cell culture equipment, and discloses a data processing system of stem cell culture equipment, which comprises the following modules: a data acquisition module generates a state and a control sequence by acquiring parameters in a stem cell culture process; the state modeling module constructs a multi-dimensional state matrix and a control matrix based on the collected data; the trend prediction module is used for constructing a state transition model and generating a prediction state vector at the next moment; the optimization control module solves a control strategy vector based on the prediction state vector; the control execution module is used for converting the control strategy vector into an actual control instruction; the closed-loop feedback module collects the actual state after adjustment and is used for correcting the state and the control matrix and dynamically updating the state transition model. According to the method, the state matrix and the control matrix based on the sliding time window are constructed, so that multi-dimensional time sequence information in the stem cell culture process is subjected to structured expression, and the effect of capturing the dynamic change trend of the environment in real time is achieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of stem cell culture equipment data processing, and in particular to a data processing system for stem cell culture equipment. Background Art

[0002] In recent years, the research and application of stem cells in the fields of regenerative medicine, disease model construction, and drug screening have continued to deepen, and higher requirements have been placed on the precise control of their culture process. Especially under large-scale and automated culture conditions, how to monitor changes in the culture environment in real time, effectively respond to system disturbances, and achieve high-precision regulation of complex physiological parameters has become one of the key links in the intelligent development of the stem cell industry.

[0003] Most existing stem cell culture equipment integrates basic sensors and controllers to monitor and regulate environmental factors such as temperature, pH, and dissolved oxygen concentration. Common solutions typically rely on preset thresholds to trigger control logic, such as activating a heater when the temperature falls below a certain threshold, or implement simple on / off control logic through methods such as programmable logic controllers (PLCs).

[0004] However, most existing stem cell culture devices lack continuous modeling of environmental and control parameter changes over time, making it difficult to capture the dynamic evolution of complex coupling relationships between multiple parameters. Control strategies are often based on fixed rules or single-variable adjustment models, resulting in slow or over-responsive regulation. Furthermore, they lack feedback mechanisms to correct execution errors, making control drift a common occurrence during long-term operation. Therefore, the present invention provides a data processing system for stem cell culture devices to address these shortcomings of the existing technology. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides a data processing system for stem cell culture equipment, which solves the problems of weak dynamic coordinated adjustment capability of multiple parameters, low state prediction accuracy and lack of closed-loop feedback in existing stem cell culture equipment.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A data processing system for a stem cell culture device includes the following modules:

[0007] A data acquisition module is used to obtain various environmental state parameters and control operation parameters during the stem cell culture process through multi-source sensors, and generate state data sequences and control data sequences corresponding to time;

[0008] A state modeling module is used to receive a state data sequence and a control data sequence, and construct a state matrix and a control matrix based on a sliding time window;

[0009] The trend prediction module is used to receive the state matrix and the control matrix, predict the environment and control operation state at the next moment by building a state transition model, and generate a predicted state vector;

[0010] an optimization control module, configured to receive a predicted state vector, construct a control objective function based on a deviation between the predicted state vector and a preset target state vector, and generate a control strategy vector by solving the control objective function;

[0011] The control execution module is used to convert the control strategy vector into actual control instructions to drive the various actuators of the stem cell culture equipment to complete the environmental parameter adjustment operation;

[0012] The closed-loop feedback module is used to collect the actual state of the actuator after adjustment, compare it with the predicted state, and then use it to correct the state matrix and the control matrix, and to update the parameters of the state transition model.

[0013] Preferably, the data acquisition module includes:

[0014] An environmental status acquisition unit, configured to periodically acquire parameters such as temperature, pH value, dissolved oxygen concentration, nutrient solution concentration, and gas flow rate in the stem cell culture environment using a multi-sensor system, including a temperature sensor, a pH sensor, a dissolved oxygen concentration sensor, a nutrient solution concentration sensor, and a gas flow rate sensor;

[0015] The control operation recording unit is used to record the parameter values ​​corresponding to the control instructions of various equipment during the cultivation process in real time, including heating power, flow valve opening, and pump speed adjustment;

[0016] The time synchronization unit is used to align the collected environmental state parameters and control operation parameters according to a unified time reference, and generate a state data sequence and a control data sequence that correspond to the time steps one by one.

[0017] Preferably, the state modeling module includes:

[0018] A state matrix construction unit is used to extract state information within a recent continuous time range based on the state data sequence, and to construct a state matrix by superposition according to time steps;

[0019] A control matrix construction unit is used to synchronously process the control data sequence within the corresponding time period to form a control matrix corresponding to the state matrix;

[0020] The window advancement unit is used to automatically advance the time window when new data is generated, and to update the state matrix and control matrix in real time.

[0021] Preferably, the state matrix construction unit sets a time window of fixed length, continuously selects state parameters of multiple time steps from the time-synchronized state data sequence to form a two-dimensional matrix, and forms a time evolution expression of the current system state.

[0022] Preferably, the trend prediction module includes:

[0023] A state model construction unit is used to establish a nonlinear state transfer model based on the constructed state matrix and control matrix, and train it through supervised learning;

[0024] The state prediction unit is used to drive the state transfer model according to the current state matrix and the control matrix, and output the predicted state vector of the next time step as the current data processing system's estimate of the future state;

[0025] The model updating unit is used to adaptively adjust the structure or parameters of the state transition model based on the prediction error information fed back by the closed-loop feedback module.

[0026] Preferably, the optimization control module includes:

[0027] A target state setting unit is used to receive a target state vector set externally or obtained through self-learning of the data processing system as a target for environmental control;

[0028] Deviation calculation unit, used to compare the predicted state vector with the target state vector to obtain the deviation results of each parameter;

[0029] The strategy generation unit is used to construct the control objective function according to the deviation result, and generate the control strategy vector through the optimization algorithm to guide the subsequent control execution module to complete the parameter adjustment.

[0030] Preferably, the strategy generation unit adopts a nonlinear optimization algorithm to solve the objective function, and the generated control strategy vector has the ability of multi-parameter collaborative regulation and supports dynamic coupling adjustment between different control variables.

[0031] Preferably, the control execution module includes:

[0032] An instruction parsing unit, configured to receive a control strategy vector and parse it into specific device control instructions;

[0033] The control interface unit is used to transmit the parsed instructions to the corresponding actuators, including heaters, gas flow valves, liquid injection pumps and pH adjustment components;

[0034] The execution monitoring unit is used to monitor the response of each actuator to the control instruction and return the response result as the actual control state to the closed-loop feedback module.

[0035] Preferably, the closed-loop feedback module includes:

[0036] The state acquisition and comparison unit is used to collect the actual environmental state of the stem cell culture device in real time through sensors after the actuator is adjusted, compare it one by one with the predicted state vector generated by the trend prediction module, and calculate the error of each environmental state;

[0037] The feedback correction unit calculates the difference between the predicted state and the actual state based on the error results obtained by the state acquisition and comparison unit, and uses the difference value to update the state matrix and control matrix in the state modeling module;

[0038] The model parameter updating unit is used to automatically update the state transition model parameters in the trend prediction module according to the correction information of the feedback correction unit to ensure the accuracy during the operation of the system.

[0039] A data processing method for a stem cell culture device is also provided, comprising the following steps:

[0040] Through multiple sensors and control devices, various environmental state parameters and control operation parameters during stem cell culture are collected in real time to generate state data sequences and control data sequences corresponding to time;

[0041] Based on the collected state data sequence and control data sequence, a state matrix and a control matrix are constructed using a sliding time window technology, and the current state expression of the system is gradually formed by extracting the state data and control data within a time period;

[0042] By receiving the constructed state matrix and control matrix, the system state is predicted based on the nonlinear state transition model, and the predicted state vector at the next moment is output as the estimation result of the future state of the data processing system;

[0043] According to the deviation between the predicted state vector and the preset target state vector, a control objective function is constructed. Through the optimization algorithm, the control objective function is solved and a control strategy vector is generated to adjust the equipment environmental parameters.

[0044] The generated control strategy vector is converted into specific control instructions to drive the actuator of the stem cell culture device to adjust the environmental parameters;

[0045] The actual state of the actuator after adjustment is collected by sensors and compared with the predicted state to calculate the error. Based on the error information, the state matrix and control matrix are corrected, and then the state transition model parameters in the trend prediction step are updated.

[0046] The present invention provides a data processing system for a stem cell culture device. It has the following beneficial effects:

[0047] 1. By constructing a state matrix and control matrix based on a sliding time window, the present invention enables structured expression of multidimensional time series information during the stem cell culture process, achieving the effect of capturing the dynamic changing trends of the environment in real time. Traditional data processing systems mostly use static state points for control judgment, ignoring the historical trajectory of state evolution, resulting in low prediction accuracy and regulation lag, thereby effectively filling the gap in time series modeling.

[0048] 2. The present invention adopts a nonlinear state transition model combined with supervised learning to model the state change law, and introduces a model update mechanism in the prediction module, so that the system can continuously correct the internal structure parameters according to the actual operation process and has the ability of continuous learning. Compared with the previous control system based on fixed formulas or linear assumptions, it is difficult to adapt to the real problem of large state disturbances in complex biological environments. This significantly enhances the adaptability and generalization ability.

[0049] 3. During the generation of the optimized control strategy, this invention integrates error analysis between the predicted and target states and further introduces a nonlinear optimization algorithm to solve the control objective function, ultimately achieving coordinated regulation of multiple parameters. Unlike traditional control methods that primarily rely on independent adjustment of a single variable, this approach simultaneously considers the dynamic coupling relationships between multiple variables, resulting in more flexible control and more efficient regulation. It is particularly suitable for systems with highly correlated variables, such as biological culture.

[0050] 4. By constructing a complete closed-loop feedback mechanism, the present invention enables the system to compare the difference between the predicted value and the execution result in real time, and use the error information for state matrix correction and model parameter update, forming a high-frequency self-iteration driven by data. Compared with traditional control systems in the absence of feedback adaptation, it can effectively discover and correct control drift problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a system architecture diagram of the present invention;

[0052] Figure 2 The figure is a flow chart of the method steps of the present invention. DETAILED DESCRIPTION

[0053] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0054] Please see the attached Figure 1 , an embodiment of the present invention provides a data processing system for a stem cell culture device, comprising the following modules:

[0055] A data acquisition module is used to obtain various environmental state parameters and control operation parameters during the stem cell culture process through multi-source sensors, and generate state data sequences and control data sequences corresponding to time;

[0056] A state modeling module is used to receive a state data sequence and a control data sequence, and construct a state matrix and a control matrix based on a sliding time window;

[0057] The trend prediction module is used to receive the state matrix and the control matrix, predict the environment and control operation state at the next moment by building a state transition model, and generate a predicted state vector;

[0058] an optimization control module, configured to receive a predicted state vector, construct a control objective function based on a deviation between the predicted state vector and a preset target state vector, and generate a control strategy vector by solving the control objective function;

[0059] The control execution module is used to convert the control strategy vector into actual control instructions to drive the various actuators of the stem cell culture equipment to complete the environmental parameter adjustment operation;

[0060] The closed-loop feedback module is used to collect the actual state of the actuator after adjustment, compare it with the predicted state, and then use it to correct the state matrix and the control matrix, and to update the parameters of the state transition model.

[0061] In this embodiment, the data acquisition module is used to collect real-time environmental state parameters and control operation parameters within the stem cell culture device. This module uses a multi-sensor system to acquire multiple key environmental state parameters and control operation parameters during the stem cell culture process and converts them into time-dependent state data sequences and control data sequences. Specifically, the data acquisition module consists of three main units: an environmental state acquisition unit, a control operation recording unit, and a time synchronization unit.

[0062] In some embodiments, the environmental status acquisition unit periodically collects key parameters such as temperature, pH value, dissolved oxygen concentration, nutrient solution concentration, and gas flow rate in the stem cell culture environment through sensors. These parameters are crucial for the growth of stem cells and the stability of the culture environment. Specifically, the temperature can be monitored in real time by a temperature sensor (such as a thermocouple sensor or an RTD sensor), the pH value can be measured by an electrode-type pH sensor, the dissolved oxygen concentration is collected by a dissolved oxygen sensor, and the nutrient solution concentration and gas flow rate can be obtained by equipment such as a spectrometer or a mass flow meter.

[0063] Alternatively, a control operation recording unit records in real time the parameter values ​​corresponding to various device control commands during the stem cell culture process. These control commands include heating power, flow valve opening, pump speed adjustment, and more. Recording and collecting these control operation parameters is crucial for subsequent data processing and optimized control. The control operation recording unit ensures that the environmental and control data corresponding to each time step accurately reflect the current operating status.

[0064] Specifically, the time synchronization unit aligns environmental state parameters with control operation parameters according to a unified time base, ensuring temporal consistency between the generated state data sequence and the control data sequence. This unit generates precisely matched state and control data sequences for subsequent modules based on the collected data.

[0065] In one possible implementation, the state data sequence and the control data sequence are synchronously processed and represented by the following formula:

[0066] State data sequence representation:

[0067] S(t)={T(t),pH(t),O2(t),C nutrient (t),v gas (t)};

[0068] Among them, S(t) represents the environmental state data sequence at time t, T(t) is the temperature, pH(t) is the pH value, O2(t) is the dissolved oxygen concentration, C nutrient (t) is the concentration of nutrient solution, v gas (t) is the gas flow rate.

[0069] Control data sequence representation:

[0070] C(t)={P heat (t),v flow (t),R pump (t)};

[0071] Where C(t) is the control data sequence at time t, P heat (t) is the heating power, v flow (t) is the flow valve opening, R pump (t) is the pump speed adjustment.

[0072] During the data synchronization process, the time synchronization unit aligns these diversified environment and control parameters with the time axis, generating a one-to-one correspondence between time steps, thereby forming a complete state data sequence and control data sequence.

[0073] In some embodiments, the function of the data acquisition module is not limited to real-time data acquisition, but can also buffer and store data through a built-in cache mechanism to ensure that the integrity and timeliness of the data are not affected when the device is operated frequently or the sensor response is delayed.

[0074] In this embodiment, the state modeling module, after collecting continuous state data sequences and control data sequences, constructs a state matrix and a control matrix based on a sliding time window mechanism, thereby achieving structured modeling of the system's current operating state. As a key data conversion link following the data acquisition module, the state modeling module provides a high-dimensional time series representation for subsequent prediction and decision-making processes, ensuring the traceability of state evolution and the completeness of the modeling logic.

[0075] Generally speaking, the state modeling module includes a state matrix construction unit, a control matrix construction unit and a window advancement unit. These units work together to organize and update multidimensional time series data.

[0076] Specifically, the state matrix construction unit first extracts data from the time-synchronized state data sequence within a continuous time period and samples the state parameters within a fixed-length time window. By stacking the data for each time step row by row, a two-dimensional matrix is ​​formed that reflects the state's temporal trend. This process can be formalized as follows:

[0077]

[0078] Among them, X t Represents the state matrix at the current time t, with a window length of n. S(t) represents the environmental state data sequence at time t, including state parameters such as temperature, pH, and dissolved oxygen concentration; t-n+1 to t represents the state data sequence from time t-n+1 to time t within the time window.

[0079] As an option, the control matrix construction unit and the state matrix construction unit operate synchronously, extracting the control vector from the control data sequence in the corresponding time period and forming the control matrix in chronological order, as follows:

[0080]

[0081] Among them, U t represents the control matrix in the current time window, C(t) is the control operation parameter vector acting on the system at time t, including heating power, valve opening, and pump speed, etc., and t-n+1 to t represents the control data sequence from time t-n+1 to time t in the time window.

[0082] In one possible implementation, the window advancement unit is used to advance the time window by one time step when new data is sampled. Each time a new set of state and control data is received, the system performs the following update operations:

[0083] Delete the earliest row in the state matrix and control matrix;

[0084] Add the latest row of data to the bottom of the matrix;

[0085] The updated matrix serves as the input state at the current time.

[0086] This sliding window strategy ensures that the state modeling module has real-time update capabilities and can continuously reflect the latest trends in the equipment operating status.

[0087] In some embodiments, to enhance the expressiveness of state modeling, differential terms may be introduced when constructing the state matrix to construct a state matrix based on incremental changes, for example:

[0088] ΔX t =X t -X t-1 ;

[0089] Where ΔX t Represents the state change matrix, which is used to characterize the change amplitude of the state variable in the time dimension and helps to reveal the potential dynamic structure. t Represents the state matrix at the current time t, X t-1 Represents the state matrix at the previous time t-1.

[0090] In addition, during the matrix construction process, normalization operations can be introduced to normalize physical quantities of different dimensions to a unified range, thereby improving the consistency and stability of subsequent modeling. Common normalization methods are:

[0091]

[0092] Among them, S i represents the original state parameters, are the minimum and maximum values ​​in the historical data collection, is the normalized state quantity.

[0093] In some implementation schemes, to improve system response speed, the window advancement unit can also integrate an asynchronous update mechanism, allowing windows to be pushed in advance when some states are not refreshed, and filling missing values ​​through weighted interpolation to enhance the system's fault tolerance.

[0094] In this embodiment, the trend prediction module receives the state matrix and control matrix output by the state modeling module. By constructing and driving a state transition model, it estimates the system state at the next moment. This module, located between the modeling and optimization control stages of the data processing system, is a key technical foundation for implementing feedforward control logic.

[0095] Generally speaking, the trend prediction module includes a state model construction unit, a state prediction unit, and a model update unit. These units work together to complete the learning, application, and adaptive adjustment of the state transition mechanism.

[0096] Specifically, the state model building unit is used to build the state matrix X t With the control matrix U t , constructing a nonlinear state transition model. This model is used to capture the dynamic evolution of environmental parameters as control operations change during stem cell culture. In one possible implementation, machine learning structures such as feedforward neural networks (FNNs), long short-term memory networks (LSTMs), or convolutional temporal networks (TCNs) can be used to model nonlinear state transitions.

[0097] The state transition model can be expressed as:

[0098]

[0099] in, Represents the predicted next time step state vector, f θ (·) is the state transition function represented by parameter θ, which can be trained by supervised learning with the goal of minimizing the prediction error.

[0100] The state prediction unit receives the trained model and the state matrix X in the given current time window t With the control matrix U t Under the condition of , output the next moment predicted state vector This vector serves as a quantitative estimate of the system's future state change trend and is used to guide the generation of subsequent control strategies.

[0101] As an option, to improve the adaptability of the model at different training stages, the model update unit adaptively adjusts the model parameter θ based on the error information between the actual state and the predicted state provided by the closed-loop feedback mechanism. This update process can be completed using a gradient descent optimization algorithm. The common error function form is as follows:

[0102]

[0103] in, Denotes the prediction loss function, S actual(t+1) is the real state vector collected at the next time step.

[0104] In some embodiments, in order to enhance the stability of model updates, momentum terms or regularization terms may be introduced for improvement, such as:

[0105]

[0106] Among them, θ represents the model parameters, η is the learning rate, λ is the regularization coefficient, Represents the L2 regularization term, which aims to prevent the model from overfitting. Represents the loss function The gradient with respect to the model parameters θ.

[0107] In one possible implementation, the trend prediction module can use a recursive structure to iteratively advance the prediction results of consecutive time steps, thereby completing the multi-step prediction process. It can be input into the state transition model together with the known control input C(t+1) to calculate And so on, the predicted trajectory generation is achieved.

[0108] In order to improve the generalization ability of the model, the state model construction unit can adopt a multi-model structure to perform weighted fusion of different types of prediction sub-models. The fusion function can adopt a weighted average or gating mechanism to achieve a stronger characterization of uncertainty and nonlinearity.

[0109] In this embodiment, the optimization control module receives the predicted state vector generated by the trend prediction module and performs a variance analysis based on the predicted target state vector. This module then constructs a control objective function and solves the resulting control strategy vector, which drives the control execution module. As a key component of closed-loop regulation, this module directly determines the stem cell culture system's response efficiency and control accuracy to deviations from the state.

[0110] Generally speaking, the optimization control module includes a target state setting unit, a deviation calculation unit, and a strategy generation unit. Each unit plans the adjustment strategy for the future state of the system based on the collaborative strategy.

[0111] Specifically, the target state setting unit is used to obtain the environmental target parameters required for the system to operate. The target state vector can be preset by the user or dynamically updated by the system based on the learning behavior of the historical training process. The target state vector can be expressed as:

[0112]

[0113] Among them, T * , pH * 、 They represent the target values ​​of temperature, pH value, dissolved oxygen concentration, nutrient solution concentration and gas flow rate under the ideal state of the system respectively; S target is the target state vector.

[0114] As an option, the deviation calculation unit is used to convert the predicted state vector With the target state vector S target Compare and calculate the deviation of each environmental parameter. The deviation can be formally expressed as:

[0115]

[0116] Among them, E(t+1) represents the state deviation vector in the next time step, which has the same dimension as the state vector and is used to quantify the difference between the current predicted state and the expected state.

[0117] In a possible implementation, the strategy generation unit constructs a control objective function based on the deviation vector E(t+1), and solves the objective function through an optimization algorithm to generate a control strategy vector C for control execution. * (t+1), the following optimization problem:

[0118]

[0119] in, is the control objective function, d is the state dimension, w i is the weighting coefficient of each environmental parameter, represents the i-th state value predicted under the control input C(t+1), λ is the control strategy regularization coefficient, which is used to limit the sharp fluctuation of the control command amplitude, S target,i is the i-th target state value, is the regularization term.

[0120] In some embodiments, in order to enhance the dynamic adaptability of the control strategy, the strategy generation unit can solve the above-mentioned objective function based on a nonlinear optimization algorithm, such as sequential quadratic programming (SQP), particle swarm optimization (PSO) or reinforcement learning strategy optimization (RL-based optimization), so that the generated control strategy vector has the ability to coordinate multi-parameter control and supports the coupling adjustment mechanism between control variables.

[0121] In some technical implementations, in order to avoid unstable jitter of the control strategy between multiple objectives, a strategy smoothing term can be introduced to constrain the control changes between consecutive time steps. For example:

[0122]

[0123] in, is the cost function term for controlling smoothness, C(t+1) is the control strategy vector at the current time t+1, C(t) is the control strategy vector at the previous time t, γ is the smoothing adjustment factor, ‖C(t+1)-C(t)‖ 2 It represents the difference norm between control strategies and is used to constrain the strategy update rate to prevent radical adjustments from causing disturbances to the system.

[0124] In this embodiment, the control execution module receives the control strategy vectors generated by the optimization control module and accurately converts them into corresponding control instructions, thereby driving the various actuators in the stem cell culture device to perform the corresponding environmental adjustment operations. This module plays a central role in strategy implementation and execution response in the data processing system, and is a crucial bridge for translating optimization calculation results into physical operations.

[0125] Generally speaking, the control execution module includes an instruction parsing unit, a control interface unit, and an execution monitoring unit. Each unit works closely together to ensure the accurate implementation of the control strategy and complete feedback of the execution status.

[0126] Specifically, the instruction parsing unit receives the control strategy vector C * (t+1), and maps each component into a low-level control instruction that can be recognized by the device. The control strategy vector can generally be expressed as:

[0127]

[0128] in, Indicates the heater output power, is the opening percentage of the gas flow valve, is the speed setting value of the liquid injection pump, ΔpH * It is the compensation control amount of the pH adjustment unit.

[0129] In one possible implementation, the instruction parsing process can use a linear transformation relationship based on mapping rules to restore the normalized strategy vector components to the actual device input. For example:

[0130]

[0131] Among them, P real is the actual heating power value corresponding to the control instruction, P min With P max are the minimum and maximum power supported by the device, is the normalized strategy component (the value range is [0,1]).

[0132] After receiving the parsed control command, the control interface unit sends the command to the target actuator via a communication bus or driver protocol (such as Modbus, CAN, RS485, etc.). Actuators may include but are not limited to: constant temperature heaters, dissolved oxygen regulating gas valves, nutrient solution injection pumps, acid and alkali solution injection modules, etc.

[0133] As an option, to ensure the stability and responsiveness of the instruction execution process, the control interface unit can have a built-in instruction verification mechanism to automatically correct or resend the instruction format, instruction range and transmission anomalies, thereby enhancing the robustness of the overall system.

[0134] In some embodiments, the control execution module may quantify the closed-loop relationship between the control signal and the execution result in the following form:

[0135] R error (t+1)=u exec (t+1)-u target (t+1);

[0136] Among them, R error (t+1) represents the control execution error, u exec (t+1) is the actual actuator output value, u target (t+1) is the theoretical target control input value. This error measure is used to evaluate the response performance of the actuator and provide a reference for system stability analysis.

[0137] In some implementations, the control execution module also supports diagnostic judgment logic for execution anomalies, such as equipment jamming, electrical signal interruption, instruction lag, and other issues, and can upload relevant fault codes to the upper-level management system for easy maintenance and traceability.

[0138] The execution monitoring unit is used to monitor the operational feedback status of each actuator in real time, including device response delay, execution result status code, current set value and actual output value. This feedback information will be further processed and utilized by the closed-loop feedback module.

[0139] In this embodiment, the closed-loop feedback module is used to implement real-time corrections and model updates to the state modeling module and trend prediction module by comparing actual execution results with predicted states after the control strategy is implemented. This module is logically closely connected to the control execution module and is a key component in achieving the system's adaptive adjustment and continuous learning capabilities.

[0140] Generally speaking, the closed-loop feedback module includes a state acquisition and comparison unit, a feedback correction unit, and a model parameter update unit. The units form a continuous closed-loop structure of the feedback path, which helps to eliminate the cumulative offset caused by disturbances, modeling errors or execution deviations.

[0141] Specifically, after the actuator completes the environmental adjustment operation, the state acquisition and comparison unit obtains the actual state value of the system again through the real-time sensor connected to the data acquisition module to form the real state vector S at the current moment. actual (t+1). Then, the unit compares the actual state with the state vector predicted by the trend prediction module at the previous moment. Perform item-by-item comparison and calculate the error information at the environmental parameter level. The error expression can be formalized as:

[0142]

[0143] Among them, ε(t+1) is the prediction error vector at time t+1, which has the same dimension as the state vector and reflects the degree of deviation of the current model from the future state prediction.

[0144] Alternatively, the feedback correction unit uses the error information to adjust the state matrix and control matrix in the state modeling module. This process typically uses a weighted update strategy or a sliding window compensation mechanism to enhance the accuracy of the matrix while preserving the continuity of the time series. The correction operation can be expressed as:

[0145]

[0146] in, is the updated state matrix, X t Current state matrix, η is the update step coefficient, ΔX t is the error propagation matrix, Represents the synchronous compensation of the error in the time window.

[0147] In one possible implementation, the model parameter update unit dynamically corrects the state transition model parameter θ in the trend prediction module based on the statistical trend of the prediction error and the historical error. Structural learning can be completed using online gradient descent or Bayesian regression, thereby improving the robustness and prediction accuracy of the system under long-term operation.

[0148] The data processing method for a stem cell culture device described below and the data processing system for a stem cell culture device described above may correspond to each other.

[0149] Please see the attached Figure 2 The present invention also provides a data processing method for a stem cell culture device, comprising the following steps:

[0150] S1. Using multiple sensors and control devices, various environmental state parameters and control operation parameters are collected in real time during the stem cell culture process to generate state data sequences and control data sequences corresponding to time;

[0151] S2. Based on the collected state data sequence and control data sequence, the sliding time window technology is used to construct the state matrix and control matrix. By extracting the state data and control data within the time period, the current state expression of the system is gradually formed;

[0152] S3, by receiving the constructed state matrix and control matrix, predicting the state of the system based on the nonlinear state transition model, and outputting the predicted state vector at the next moment as the estimation result of the future state of the data processing system;

[0153] S4. Based on the deviation between the predicted state vector and the preset target state vector, a control objective function is constructed. Through the optimization algorithm, the control objective function is solved and a control strategy vector is generated to adjust the equipment environment parameters.

[0154] S5, converting the generated control strategy vector into specific control instructions to drive the actuator of the stem cell culture device to adjust the environmental parameters;

[0155] S6. The actual state of the actuator after adjustment is collected through the sensor, and compared with the predicted state, and the error is calculated. According to the error information, the state matrix and the control matrix are corrected, and then the state transition model parameters in the trend prediction step are updated.

[0156] In step S1, various environmental state parameters and control operation parameters during the stem cell culture process are collected in real time through multiple sensors and control devices, generating state data sequences and control data sequences corresponding to time. Specifically, sensors include but are not limited to temperature sensors, pH sensors, dissolved oxygen sensors, and nutrient solution concentration sensors. The real-time sampling frequency can be dynamically adjusted based on the characteristics of the culture process. Control devices include heating devices, gas control valves, liquid injection pumps, etc., which synchronously record corresponding operating parameters to form a multi-dimensional, time-series data source.

[0157] In step S2, a sliding time window technique is used to construct the state and control matrices based on the collected state and control data sequences. This technique uses a fixed-length time window to slice the continuously collected state and control data into matrices representing multiple time periods. The rows and columns of the matrices correspond to different parameters and time sampling points, gradually forming a structured representation of the dynamic evolution of the system's current multi-parameters. This representation exhibits stronger temporal correlation than traditional single-point sampling, facilitating the capture of changing environmental state trends and control response patterns.

[0158] In step S3, the system state is predicted based on the nonlinear state transition model by receiving the constructed state matrix and control matrix. Specifically, a trained nonlinear mapping function is used to transform the current and historical states, outputting the predicted state vector for the next moment. This prediction process not only considers the independent changes of a single variable but also reflects the dynamic coupling between multiple environmental parameters through the model, significantly improving the accuracy of future state estimation and providing a scientific basis for subsequent control strategies.

[0159] In step S4, a control objective function is constructed based on the deviation between the predicted state vector and the preset target state vector. This objective function comprehensively considers the deviation magnitudes of multiple environmental parameters and their weighted impact on the cultivation effect, combined with the control cost term, to form a multi-objective optimization problem. By introducing an optimization algorithm such as gradient descent or genetic algorithm, this function is solved to obtain the optimal control strategy vector, achieving reasonable coordinated regulation of environmental parameters.

[0160] In step S5, the generated control strategy vector is converted into specific control instructions to drive the actuators of the stem cell culture device to adjust environmental parameters. These control instructions include information such as actuator activation, adjustment amplitude, and duration. This ensures that the actuators accurately respond to the calculated results, maintaining the culture environment parameters within a predetermined range to meet the physiological needs of stem cell culture.

[0161] In step S6, sensors collect the actual state of the actuator after adjustment, compare it with the predicted state, and calculate the error. This error information serves as a feedback signal, acting on the dynamic correction process of the state matrix and control matrix to correct modeling deviations. The error-driven mechanism is further used to update the state transition model parameters in the trend prediction, enabling the model to learn online and adapt itself, thereby continuously optimizing prediction accuracy and control effectiveness, ensuring the stability and reliability of the training process.

[0162] The method of this embodiment can be used to execute the above system embodiment. Its principles and technical effects are similar and will not be described in detail here.

[0163] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A data processing system for a stem cell culture device, characterized in that: Includes the following modules: A data acquisition module is used to obtain various environmental state parameters and control operation parameters during the stem cell culture process through multi-source sensors, and generate state data sequences and control data sequences corresponding to time; A state modeling module is used to receive a state data sequence and a control data sequence, and construct a state matrix and a control matrix based on a sliding time window; The trend prediction module is used to receive the state matrix and the control matrix, predict the environment and control operation state at the next moment by building a state transition model, and generate a predicted state vector; an optimization control module, configured to receive a predicted state vector, construct a control objective function based on a deviation between the predicted state vector and a preset target state vector, and generate a control strategy vector by solving the control objective function; The control execution module is used to convert the control strategy vector into actual control instructions to drive the various actuators of the stem cell culture equipment to complete the environmental parameter adjustment operation; The closed-loop feedback module is used to collect the actual state of the actuator after adjustment, compare it with the predicted state, and then use it to correct the state matrix and the control matrix, and to update the parameters of the state transition model.

2. The data processing system for a stem cell culture device according to claim 1, characterized in that: The data acquisition module includes: An environmental status acquisition unit, configured to periodically acquire parameters such as temperature, pH value, dissolved oxygen concentration, nutrient solution concentration, and gas flow rate in the stem cell culture environment using a multi-sensor system, including a temperature sensor, a pH sensor, a dissolved oxygen concentration sensor, a nutrient solution concentration sensor, and a gas flow rate sensor; The control operation recording unit is used to record the parameter values ​​corresponding to the control instructions of various equipment during the cultivation process in real time, including heating power, flow valve opening, and pump speed adjustment; The time synchronization unit is used to align the collected environmental state parameters and control operation parameters according to a unified time reference, and generate a state data sequence and a control data sequence that correspond to the time steps one by one.

3. The data processing system for a stem cell culture device according to claim 1, characterized in that: The state modeling module includes: A state matrix construction unit is used to extract state information within a recent continuous time range based on the state data sequence, and to construct a state matrix by superposition according to time steps; A control matrix construction unit is used to synchronously process the control data sequence within the corresponding time period to form a control matrix corresponding to the state matrix; The window advancement unit is used to automatically advance the time window when new data is generated, and to update the state matrix and control matrix in real time.

4. The data processing system for a stem cell culture device according to claim 3, characterized in that: The state matrix construction unit sets a time window of fixed length and continuously selects state parameters of multiple time steps from the time-synchronized state data sequence to form a two-dimensional matrix, thereby forming a time evolution expression of the current system state.

5. The data processing system for a stem cell culture device according to claim 1, characterized in that: The trend prediction module includes: A state model construction unit is used to establish a nonlinear state transfer model based on the constructed state matrix and control matrix, and train it through supervised learning; The state prediction unit is used to drive the state transfer model according to the current state matrix and the control matrix, and output the predicted state vector of the next time step as the current data processing system's estimate of the future state; The model updating unit is used to adaptively adjust the structure or parameters of the state transition model based on the prediction error information fed back by the closed-loop feedback module.

6. The data processing system for a stem cell culture device according to claim 1, characterized in that: The optimization control module includes: A target state setting unit is used to receive a target state vector set externally or obtained through self-learning of the data processing system as a target for environmental control; Deviation calculation unit, used to compare the predicted state vector with the target state vector to obtain the deviation results of each parameter; The strategy generation unit is used to construct the control objective function according to the deviation result, and generate the control strategy vector through the optimization algorithm to guide the subsequent control execution module to complete the parameter adjustment.

7. The data processing system for a stem cell culture device according to claim 6, characterized in that: The strategy generation unit adopts a nonlinear optimization algorithm to solve the objective function, and the generated control strategy vector has the ability of multi-parameter collaborative regulation and supports dynamic coupling adjustment between different control variables.

8. The data processing system for a stem cell culture device according to claim 1, characterized in that: The control execution module includes: An instruction parsing unit, configured to receive a control strategy vector and parse it into specific device control instructions; The control interface unit is used to transmit the parsed instructions to the corresponding actuators, including heaters, gas flow valves, liquid injection pumps and pH adjustment components; The execution monitoring unit is used to monitor the response of each actuator to the control instruction and return the response result as the actual control state to the closed-loop feedback module.

9. The data processing system for a stem cell culture device according to claim 1, characterized in that: The closed-loop feedback module includes: The state acquisition and comparison unit is used to collect the actual environmental state of the stem cell culture device in real time through sensors after the actuator is adjusted, compare it one by one with the predicted state vector generated by the trend prediction module, and calculate the error of each environmental state; The feedback correction unit calculates the difference between the predicted state and the actual state based on the error results obtained by the state acquisition and comparison unit, and uses the difference value to update the state matrix and control matrix in the state modeling module; The model parameter updating unit is used to automatically update the state transition model parameters in the trend prediction module according to the correction information of the feedback correction unit to ensure the accuracy during the operation of the system.

10. A data processing method for a stem cell culture device, applied to a data processing system for a stem cell culture device according to any one of claims 1 to 9, characterized in that: The following steps are involved: Through multiple sensors and control devices, various environmental state parameters and control operation parameters during stem cell culture are collected in real time to generate state data sequences and control data sequences corresponding to time; Based on the collected state data sequence and control data sequence, a state matrix and a control matrix are constructed using a sliding time window technology, and the current state expression of the system is gradually formed by extracting the state data and control data within a time period; By receiving the constructed state matrix and control matrix, the system state is predicted based on the nonlinear state transition model, and the predicted state vector at the next moment is output as the estimation result of the future state of the data processing system; According to the deviation between the predicted state vector and the preset target state vector, a control objective function is constructed. Through the optimization algorithm, the control objective function is solved and a control strategy vector is generated to adjust the equipment environmental parameters. The generated control strategy vector is converted into specific control instructions to drive the actuator of the stem cell culture device to adjust the environmental parameters; The actual state of the actuator after adjustment is collected by sensors and compared with the predicted state to calculate the error. Based on the error information, the state matrix and control matrix are corrected, and then the state transition model parameters in the trend prediction step are updated.

Citation Information

Cited By

  • Cell culture dynamic regulation and control system based on multi-modal time sequence data

    CN122105016A