Intelligent control method and system for extraction process based on multi-dimensional dynamic algorithm

Abstract

This invention discloses an intelligent control method and system for the extraction process based on a multi-dimensional dynamic algorithm, belonging to the field of microfluidic intelligent control technology. The method includes: acquiring multi-dimensional real-time state data of the single-cell microenvironment within a microfluidic chip; performing flow-concentration field coupled simulation to generate simulation prediction results; generating microchannel reconstruction instructions to guide the fluid flow to avoid low-velocity stagnation regions and maintain laminar flow; and generating a trigger signal when the metabolite concentration in the multi-dimensional real-time state data reaches a preset peak window, controlling a microvalve to perform a precise micro-volume capture action. This invention solves the technical problems of existing technologies where stagnation regions easily appear in the microchannel flow field, and the fluid flow state is unstable, leading to inaccurate timing of metabolite sample capture and insufficient real-time detection accuracy. It achieves dynamic optimization of the microchannel flow field and precise capture of metabolite samples, improving the real-time performance and accuracy of single-cell metabolic detection.

Description

Technical Field

[0001] This invention relates to the field of microfluidic intelligent control technology, specifically to an intelligent control method and system for the extraction process based on a multi-dimensional dynamic algorithm. Background Technology

[0002] In microfluidic applications for single-cell metabolic detection, traditional techniques mostly employ fixed-structure microchannels, which cannot adjust the channel morphology according to dynamic changes in the cellular microenvironment. Low-velocity stagnation areas easily form within the channels, disrupting the laminar flow state and causing metabolite accumulation and uneven distribution. Furthermore, existing solutions lack the ability to co-simulate and predict flow and concentration fields, making it difficult to accurately capture peak metabolite concentration nodes. Sample capture timing deviations are significant, and data processing and instruction generation rely on general-purpose computing units, resulting in high overall latency. Ultimately, this leads to poor real-time performance in single-cell metabolic tracking and low sample capture accuracy, failing to meet the demands for high-precision, dynamic single-cell metabolic analysis.

[0003] Existing technologies suffer from technical problems such as the easy formation of stagnant areas in the microfluidic flow field and unstable fluid flow, leading to inaccurate timing of metabolite sample capture and insufficient real-time performance and accuracy of detection. Summary of the Invention

[0004] This application provides a method and system for intelligent control of the extraction process based on a multi-dimensional dynamic algorithm, which is used to address the technical problems in existing technologies, such as the easy formation of stagnant areas in the microchannel flow field, unstable liquid flow state, resulting in inaccurate timing of metabolite sample capture and insufficient real-time detection accuracy.

[0005] In view of the above problems, this application provides a method and system for intelligent control of the extraction process based on a multi-dimensional dynamic algorithm.

[0006] The first aspect of this application provides an intelligent control method for the extraction process based on a multi-dimensional dynamic algorithm, the method comprising:

[0007] Multi-dimensional real-time state data of the single-cell microenvironment within a microfluidic chip is acquired. This multi-dimensional real-time state data includes at least metabolite concentration gradient field data and cell physiological activity data. Based on this multi-dimensional real-time state data, a flow field-concentration field coupled simulation is performed using a virtual simulation model driven by an edge computing unit to generate simulation prediction results. The virtual simulation model integrates fluid dynamics equations and mass diffusion equations to predict the distribution of metabolites and low-velocity stagnation regions within the microchannel in real time. According to the simulation prediction results, microchannel reconstruction instructions are dynamically generated using structural optimization logic. Based on these instructions, the morphology of the flow divider barrier within the channel is adjusted in real time using a microelectromechanical adjustable structure to guide the fluid flow away from low-velocity stagnation regions and maintain laminar flow. When the metabolite concentration in the multi-dimensional real-time state data reaches a preset peak window, a trigger signal is generated to control a microvalve to perform a precise micro-capture action and immediately deliver the captured sample to a mass spectrometry analysis unit.

[0008] A second aspect of this application provides an intelligent control system for the extraction process based on a multi-dimensional dynamic algorithm, the system comprising:

[0009] The system comprises the following modules: a state data acquisition module for acquiring multi-dimensional real-time state data of the single-cell microenvironment within the microfluidic chip, including at least metabolite concentration gradient field data and cell physiological activity data; a simulation prediction result generation module for performing flow field-concentration field coupled simulation based on the multi-dimensional real-time state data using a virtual simulation model driven by an edge computing unit, generating simulation prediction results; and a reconstruction instruction generation module for dynamically generating microfluidic channel reconstruction instructions based on the simulation prediction results using structural optimization logic, and adjusting the morphology of the flow divider barrier within the channel in real-time using a microelectromechanical adjustable structure to guide the flow away from low-velocity stagnation areas and maintain laminar flow; and a trigger signal generation module for generating a trigger signal when the metabolite concentration in the multi-dimensional real-time state data reaches a preset peak window, controlling the microvalve to perform a precise micro-capture action, and immediately delivering the captured sample to the mass spectrometry analysis unit.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0011] This method acquires multi-dimensional real-time state data of the single-cell microenvironment within a microfluidic chip; performs flow-concentration field coupled simulation using a virtual simulation model driven by an edge computing unit, generating simulation prediction results; based on these prediction results, dynamically generates microchannel reconstruction instructions using structural optimization logic, and adjusts the morphology of the flow divider barriers within the channel in real time using a microelectromechanical adjustable structure based on these instructions, guiding the fluid flow to avoid low-velocity stagnation areas and maintaining laminar flow; when the metabolite concentration in the multi-dimensional real-time state data reaches a preset peak window, a trigger signal is generated to control a microvalve to perform a precise micro-capture action, and the captured sample is immediately delivered to the mass spectrometry analysis unit. This achieves the technical effect of dynamically optimizing the microchannel flow field and precisely capturing metabolite samples, improving the real-time performance and accuracy of single-cell metabolic detection. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A schematic diagram of the intelligent control method for the extraction process based on a multi-dimensional dynamic algorithm provided in this application embodiment;

[0014] Figure 2 A schematic diagram of the structure of an intelligent control system for the extraction process based on a multi-dimensional dynamic algorithm, provided in an embodiment of this application.

[0015] Figure labeling: 10 for state data acquisition module, 20 for simulation prediction result generation module, 30 for reconstruction instruction generation module, and 40 for trigger signal generation module. Detailed Implementation

[0016] This application provides an intelligent control method and system for the extraction process based on a multi-dimensional dynamic algorithm, which addresses the technical problems of existing technologies, such as the easy formation of stagnant areas in the microchannel flow field, unstable liquid flow state, inaccurate timing of metabolite sample capture, and insufficient real-time detection accuracy.

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0018] Example 1, as Figure 1 As shown, this application provides an intelligent control method for the extraction process based on a multi-dimensional dynamic algorithm, the method comprising:

[0019] Step S100: Obtain multi-dimensional real-time state data of the single-cell microenvironment within the microfluidic chip, wherein the multi-dimensional real-time state data includes at least metabolite concentration gradient field data and cell physiological activity data.

[0020] Specifically, an array of optical sensing units is deployed at key nodes of the microfluidic channel to monitor the adsorption kinetics curves of target metabolites secreted by single cells at the solid-liquid interface in real time. By demodulating the changes in reflected light intensity, the local refractive index is inverted to obtain high spatiotemporal resolution metabolite concentration gradient field data. Simultaneously, the signals of electrochemical sensing units between the electrodes of the microfluidic channel are acquired, the Nyquist plot characteristics of the electrochemical impedance spectroscopy are analyzed, and cell adhesion impedance and membrane capacitance parameters are extracted to obtain cell physiological activity data characterizing the degree of cell metabolic activity. The above metabolite concentration gradient field data and cell physiological activity data are integrated to form multi-dimensional real-time state data of the single-cell microenvironment within the microfluidic chip, and cached in the random access memory of the edge computing unit.

[0021] Step S200: Based on the multi-dimensional real-time state data, a flow field-concentration field coupled simulation is performed through a virtual simulation model driven by an edge computing unit to generate simulation prediction results. The virtual simulation model integrates fluid dynamics equations and material diffusion equations to predict the distribution of metabolites in the microchannel and the low-velocity stagnation area in real time.

[0022] Specifically, multi-dimensional real-time state data is read through edge computing units, and metabolite concentration gradient field data is mapped to pre-divided finite element mesh nodes. Cell physiological activity data is used as a correction factor to calibrate the effective diffusion coefficient of the material diffusion equation. A virtual simulation model integrating fluid dynamics equations and material diffusion equations is used to perform coupled flow field-concentration field simulation. The Navier-Stokes equations are iteratively solved to obtain the velocity vector distribution in the microchannel. The metabolite diffusion flux vector is calculated by combining the convective mass transfer term. Metabolite accumulation areas and low-velocity stagnation areas are identified and marked. Finally, simulation prediction results containing accumulation coordinates, stagnation dead zone coordinates, and concentration prediction values ​​are generated.

[0023] Step S300: Based on the simulation prediction results, microchannel reconstruction instructions are dynamically generated using structural optimization logic. Based on the microchannel reconstruction instructions, the morphology of the flow divider barrier in the flow channel is adjusted in real time through the microelectromechanical adjustable structure to guide the liquid flow to avoid the low-velocity stagnation area and maintain the laminar flow state.

[0024] Specifically, the simulation prediction results are analyzed to extract the three-dimensional coordinate set of the low-velocity stagnation region and the metabolite enrichment region. A multi-objective optimization function is constructed with the goal of minimizing the flow loss rate and maximizing the average concentration in the enrichment region. The diversion barrier is discretized into deformable units and transformed into a mathematical programming problem of pseudo-density variables through structural optimization logic. After iterative optimization and convergence judgment, a binary microchannel reconstruction instruction with the optimal topology is generated. Then, according to the instruction, the microelectromechanical adjustable structure is driven to deflect and adjust the shape of the diversion barrier according to the displacement. Combined with the negative feedback mechanism, the angle is finely adjusted so that the liquid flow maintains a laminar flow state and avoids the low-velocity stagnation region and concentrates on flushing the metabolite enrichment region.

[0025] Step S400: When the concentration of metabolites in the multidimensional real-time status data reaches the preset peak window, a trigger signal is generated to control the microvalve to perform a micro-precision capture action and immediately deliver the captured sample to the mass spectrometry analysis unit.

[0026] Specifically, a nonlinear state-space model is constructed based on metabolite concentration gradient field data. The state gain matrix is ​​iteratively updated using a recursive calculation method. The estimated value of the true concentration of metabolites and the rate of change of concentration are calculated in real time. When the rate of change of concentration is detected to change from a positive value to a negative value and the absolute value exceeds the preset threshold, it is determined that the current moment has entered the preset peak window and a trigger signal is generated. The microvalve is controlled to perform a micro-liquid segment interception action, and the captured sample is delivered to the mass spectrometry analysis unit in real time through a micropump to complete the detection.

[0027] In one possible implementation, step S100 further includes:

[0028] Step S110: Deploy an array of optical sensing units at key nodes of the microfluidic channel to monitor the adsorption kinetics curve of the target metabolite secreted by a single cell at the solid-liquid interface in real time. Demodulate the change in reflected light intensity to invert the local refractive index and obtain metabolite concentration gradient field data with high spatiotemporal resolution.

[0029] Step S120: Synchronously acquire signals from the electrochemical sensing unit between the microfluidic electrodes, analyze the Nyquist plot features of the electrochemical impedance spectroscopy, and extract cell adhesion impedance and membrane capacitance parameters to obtain cell physiological activity data characterizing the degree of cell metabolic activity.

[0030] Step S130: Integrate the metabolite concentration gradient field data with the cell physiological activity data to form multi-dimensional real-time state data, and cache the multi-dimensional real-time state data in the random access memory of the edge computing unit.

[0031] Specifically, an array of optical sensing units is deployed at key monitoring nodes in the microfluidic channel to capture the adsorption process of target metabolites secreted by single cells on the solid-liquid interface of the microfluidic channel in real time through optical detection, and continuously collect and record the adsorption kinetic curves; the real-time changes in reflected light intensity are demodulated and numerically calculated, the refractive index distribution of the local area is inverted based on the changes in light intensity, and then the refractive index information is converted into concentration information, finally obtaining metabolite concentration gradient field data with high spatiotemporal resolution that can reflect the spatial distribution of metabolites.

[0032] Simultaneously with optical sensing monitoring, the AC excitation response signal output by the electrochemical sensing unit between the working electrode and the counter electrode inside the microchannel is acquired. The impedance of this signal is calculated using an electrochemical workstation to obtain an electrochemical impedance spectrum and plot the corresponding Nyquist plot. Based on the spectral characteristics of the Nyquist plot, such as the low-frequency real intercept and the high-frequency semicircle diameter, the interfacial cell adhesion impedance formed after cell adhesion in the microchannel and the membrane capacitance parameters characterizing cell membrane properties are extracted. Based on the above electrical parameters, the physiological state and secretory activity of the cell are quantitatively characterized, and finally, cell physiological activity data that can directly reflect the metabolic activity of a single cell are obtained.

[0033] The acquired metabolite concentration gradient field data and the acquired cell physiological activity data are time-aligned and fused to form multi-dimensional real-time state data containing metabolite spatial distribution and cell physiological state. The data is encapsulated according to the addressing format of the edge computing unit and written and cached in real time in the random access memory (RAM) of the edge computing unit, providing a high-speed real-time data source for subsequent flow field-concentration field coupled simulation.

[0034] In one possible implementation, step S200 further includes:

[0035] Step S210: The edge computing unit calls the data interface to read the multi-dimensional real-time state data from the random access memory, and maps the metabolite concentration gradient field data in the multi-dimensional real-time state data onto the pre-divided finite element mesh nodes.

[0036] Step S220: Using the virtual simulation model to call the flow field-concentration field coupled simulation, read the cell physiological activity data in the multi-dimensional real-time state data as a correction factor, substitute it into the material diffusion equation integrated by the virtual simulation model, and dynamically adjust the diffusion coefficient tensor.

[0037] Step S230: Execute the fluid dynamics equations integrated in the virtual simulation model, calculate the three-dimensional velocity vector distribution of each grid cell in the microchannel, and solve the material diffusion equation by combining the adjusted diffusion coefficient tensor to obtain the spatial distribution cloud map of metabolite concentration.

[0038] Step S240: Identify the regions in the distribution cloud map where the concentration is higher than a first preset threshold as metabolite accumulation regions, and mark the grid cells in the three-dimensional velocity vector distribution where the velocity modulus is lower than a second preset threshold as low velocity stagnation regions, and encapsulate the geometric coordinate information of the metabolite accumulation regions and the low velocity stagnation regions into the simulation prediction results.

[0039] Specifically, the edge computing unit retrieves cached multi-dimensional real-time state data from the random access memory through its built-in data reading interface. The metabolite concentration gradient field data is then mapped one by one onto the finite element nodes of the microchannel region that have been pre-meshed according to the spatial correspondence, providing accurate initial concentration field input conditions for the flow field-concentration field coupled simulation.

[0040] A virtual simulation model integrating fluid dynamics equations and mass diffusion equations is driven by an edge computing unit. The flow field-concentration field coupled simulation calculation is initiated. Cell physiological activity data is read from multi-dimensional real-time state data and normalized into a numerical correction factor. This correction factor is substituted into the mass diffusion equation to dynamically correct and update the diffusion coefficient tensor, which characterizes the diffusion ability of metabolites in the equation, in real time. This ensures that the simulation parameters are consistent with the actual metabolic state of a single cell and improves the accuracy of the coupled simulation.

[0041] A virtual simulation model driven by edge computing units is used to numerically iterate and solve the integrated fluid dynamics equations, namely the Navier-Stokes equations, based on pre-divided finite element meshes and microchannel boundary conditions. The three-dimensional velocity vector, pressure value, and velocity component of each element are calculated mesh by mesh to form a complete flow field distribution. Then, the diffusion coefficient tensor corrected by cell physiological activity data is substituted into the mass diffusion equation, and the convective mass transfer term driven by the velocity vector is introduced. The molecular diffusion flux and convective diffusion flux are coupled and superimposed to obtain the diffusion flux vector. After full-field iterative solution by the finite element method, the real-time distribution of metabolite concentration in the microchannel space is output in the form of a cloud map, that is, the spatial distribution cloud map of metabolite concentration is obtained.

[0042] The obtained metabolite concentration spatial distribution cloud map is subjected to a full-field threshold traversal judgment. Grid regions with concentration values ​​higher than a first preset threshold are identified and marked as metabolite accumulation areas. At the same time, the modulus of the three-dimensional velocity vector distribution is calculated, and grid cells with velocity modulus lower than a second preset threshold are identified as low-velocity stagnation areas. Subsequently, the three-dimensional geometric coordinate information corresponding to the above-mentioned metabolite accumulation areas and low-velocity stagnation areas is extracted. The coordinate set, concentration prediction value and velocity feature data are uniformly packaged and processed to form the simulation prediction results for subsequent flow channel reconstruction.

[0043] In one possible implementation, step S220 further includes:

[0044] Step S221: Based on the flow field-concentration field coupled simulation, extract the cell membrane permeability parameters reflected by the cell physiological activity data in the multi-dimensional real-time state data, and use the cell membrane permeability parameters to calibrate the effective diffusion coefficient in the material diffusion equation integrated by the virtual simulation model.

[0045] Step S222: Using the fluid dynamics equations and the boundary conditions of the microchannel wall, iteratively solve the Navier-Stokes equations to obtain the velocity vector distribution of the microchannel cross section and longitudinal direction, and record the pressure value and velocity component of each grid node.

[0046] Step S223: Using the mass diffusion equation and the convective mass transfer term brought about by the velocity vector distribution, calculate the diffusion flux vector of metabolites in the flow channel. The diffusion flux vector is formed by the superposition of molecular diffusion flux and convective diffusion flux.

[0047] Step S224: Integrate the velocity vector distribution and the diffusion flux vector, and generate the simulation prediction result from the virtual simulation model, which includes the set of metabolite aggregation coordinates, the set of flow dead zone coordinates, and the corresponding concentration prediction values.

[0048] Specifically, in the process of conducting flow field-concentration field coupled simulation, cell membrane permeability parameters that can reflect the physiological state of cells are extracted from multi-dimensional real-time state data. These parameters are used as calibration factors and substituted into the material diffusion equation integrated in the virtual simulation model. The effective diffusion coefficient in the equation is corrected and numerically calibrated in real time, so that the simulation calculation is more in line with the real physical process of single-cell metabolite secretion and diffusion.

[0049] Based on the fluid dynamics equations in the virtual simulation model, and combined with constraints such as the no-slip boundary condition of the microchannel wall and the pressure boundary conditions at the inlet and outlet, the Navier-Stokes equations are solved using the finite element numerical iterative algorithm. The complete velocity vector distribution in the cross-section and longitudinal channel of the microchannel is calculated sequentially. The pressure value and velocity components in the X, Y, and Z directions corresponding to each finite element grid node are recorded simultaneously, providing accurate flow field basic data for subsequent concentration field coupling calculations.

[0050] The material diffusion equation in the virtual simulation model is invoked, and the obtained velocity vector distribution is used as input to construct the convective mass transfer term generated by fluid flow. First, the molecular diffusion flux of metabolites is calculated based on Fick's law and the calibrated effective diffusion coefficient. Then, the convective diffusion flux is calculated based on the velocity vector and the metabolite concentration field. Finally, the molecular diffusion flux and the convective diffusion flux are superimposed according to the vector operation rules to accurately solve for the diffusion flux vector of metabolites in the microchannel.

[0051] The obtained velocity vector distribution and diffuse flux vector are fused and matched in spatial dimension. Based on the velocity magnitude, the flow dead zone region is determined, and based on the diffuse flux and concentration field gradient, the metabolite accumulation region is identified. The virtual simulation model traverses all finite element mesh nodes, extracts the metabolite accumulation coordinate set and the flow dead zone coordinate set, and associates the corresponding concentration prediction values ​​of each node. The unified structured output is the simulation prediction result.

[0052] In one possible implementation, step S300 further includes:

[0053] Step S310: Analyze the simulation prediction results, extract the three-dimensional coordinate set representing the low-flow-rate stagnation region and the three-dimensional coordinate set representing the metabolite enrichment region, and calculate the fluid volume fraction occupied by the low-flow-rate stagnation region.

[0054] Step S320: Construct a multi-objective optimization function with weighted coefficients, with the first optimization objective being to minimize the flow loss rate through the low-flow-rate stagnation region and the second optimization objective being to maximize the average concentration of metabolites in the metabolite enrichment region.

[0055] Step S330: Through the structural optimization logic, the flow divider barrier in the microchannel is discretized into multiple deformable units, and the multi-objective optimization function is transformed into a mathematical programming problem about the pseudo-density variables of each deformable unit.

[0056] Step S340: Based on the mathematical programming problem, perform an iterative optimization calculation process, adjust the pseudo-density value of each deformable element until the Pareto front solution set of the multi-objective optimization function converges, select the solution with the best comprehensive performance, and generate a microchannel reconstruction instruction for reconstructing the physical morphology of the flow channel.

[0057] Specifically, spatial analysis and data extraction are performed on the simulation prediction results. The three-dimensional coordinate sets corresponding to the low-velocity retention area and the metabolite enrichment area are selected and extracted respectively. Then, based on the unit volume information of the finite element mesh, the sum of the volumes of all mesh units belonging to the low-velocity retention area is calculated and compared with the total fluid volume in the microchannel to obtain the fluid volume fraction occupied by the low-velocity retention area.

[0058] The first optimization objective is to reduce the flow loss rate of fluid in the low-velocity stagnation region, and the second optimization objective is to increase the average concentration of metabolites in the metabolite enrichment region. By setting weighting coefficients to allocate the optimization weights of the two objectives, the flow loss control and metabolite enrichment effect are balanced, and a multi-objective optimization function that can be used for numerical solution is constructed.

[0059] By employing topology optimization logic, the flow divider barriers inside the microchannel are spatially discretized, dividing the overall structure into several deformable units of uniform size. A pseudo-density variable in the range of 0 to 1 is introduced for each deformable unit to characterize the state of the unit's physical presence or hollowness. Thus, the multi-objective optimization function, which includes flow loss rate and average metabolite concentration, is transformed into a continuous mathematical programming problem with pseudo-density of each unit as the design variable and satisfying volume constraints.

[0060] Iterative optimization calculations are performed on the established mathematical programming problem. The pseudo-density of deformable elements at each grid node is used as the design variable. The adjoint variable method is used to solve the sensitivity derivative corresponding to the multi-objective optimization function. Combining the sign and magnitude of the derivative, the pseudo-density of each node on the microchannel wall and the diversion barrier is iteratively updated to increase the material density in the high-sensitivity region and decrease the material density in the low-sensitivity region. After each iteration, it is checked whether the value of the multi-objective optimization function is within the preset convergence interval. If the solution set is determined to be convergent, the current pseudo-density distribution matrix is ​​set as the optimal topology. Then, the matrix is ​​binarized by thresholding to distinguish physical entities and void regions to obtain the binary driving signal matrix. Finally, the matrix is ​​encoded into a time-series pulse sequence and outputs a microchannel reconstruction command that can be used to reconstruct the physical morphology of the flow channel.

[0061] In one possible implementation, step S340 further includes:

[0062] Step S341: Calculate the sensitivity derivative of each design variable in the multi-objective optimization function, where the design variable is the pseudo density of deformable elements corresponding to each grid node.

[0063] Step S342: Based on the positive and negative directions and magnitude of the sensitivity derivative, update the pseudo-density distribution of the corresponding nodes on the flow channel wall and the diversion barrier, including increasing the material density in the high-sensitivity region and decreasing the material density in the low-sensitivity region.

[0064] Step S343: Determine whether the multi-objective optimization function value of the current iteration step falls within the preset convergence interval. If so, determine the current pseudo-density distribution matrix as the optimal topology.

[0065] Step S344: The optimal topology is binarized to generate a binary driving signal matrix representing physical entities and gaps, and the binary driving signal matrix is ​​encoded into a time-series pulse sequence as a microchannel reconstruction instruction.

[0066] Specifically, the pseudo-density of deformable elements at each grid node is used as the design variable. The adjoint variable method is used to solve the sensitivity derivative. First, the adjoint equation is established by combining the finite element calculation results of the flow field and concentration field. Then, the original multi-objective optimization function and constraint conditions are solved numerically. The sensitivity derivative value corresponding to each pseudo-density variable is calculated in turn to clarify the magnitude and direction of the influence of the parameter changes of each element on the optimization objective.

[0067] The sensitivity derivative values ​​of each grid node are read. First, the direction of pseudo-density adjustment is determined according to the sign of the derivative. Then, the influence strength is determined according to the magnitude of the derivative. The pseudo-density values ​​of the corresponding nodes of the microchannel wall and the diversion barrier are updated one by one. The material pseudo-density is increased in the area with large sensitivity amplitude and decreased in the area with small sensitivity amplitude, thus completing this round of parameter iteration update.

[0068] Extract the calculation result of the multi-objective optimization function in the current iteration step, and compare it with the pre-set convergence interval threshold. If the calculated value of the function is within the range of the interval, the iterative optimization process is determined to be converged, and the currently stored pseudo-density distribution matrix is ​​directly determined as the optimal topology of the microchannel diversion barrier.

[0069] Binarization is performed by setting a boundary threshold for the pseudo-density matrix corresponding to the optimal topology. Cells larger than the threshold are marked as physical entities, and cells smaller than the threshold are marked as gaps, thereby generating a binary driving signal matrix. Then, the matrix data is serialized according to the predetermined encoding rules and compiled into a continuous time-series pulse sequence, which finally forms a microchannel reconfiguration instruction that can directly drive hardware execution.

[0070] In one possible implementation, step S300 further includes:

[0071] Step S350: Analyze the binary drive signal in the microchannel reconstruction instruction and convert it into the execution displacement of each baffle in the microelectromechanical adjustable structure.

[0072] Step S360: Drive the matrix-distributed microelectromechanical adjustable structure to deflect according to the executed displacement amount, so as to physically reconstruct the morphology of the flow channel diversion barrier.

[0073] Step S370: Monitor the velocity distribution of the reconstructed flow channel cross section, and fine-tune the deflection angle of the microelectromechanical adjustable structure through a negative feedback mechanism to maintain the liquid flow in a laminar state and bypass the low velocity stagnation area, thereby concentrating on flushing the metabolite enrichment area.

[0074] Specifically, the preset signal mapping lookup table and calibration conversion coefficients are retrieved, and the binary drive signal in the microchannel reconstruction command is decoded bit by bit. The numerical conversion is completed by combining the mechanical stroke range of the baffle, and the discrete binary data is accurately converted into the actual linear displacement or deflection displacement of each baffle in the microelectromechanical adjustable structure.

[0075] The displacement of each baffle is converted into a driving voltage / pulse signal. Each group of microelectromechanical adjustable structures is controlled sequentially according to the matrix array addressing method, so that the baffle is precisely deflected to the designated position. Through the coordinated change of the overall attitude, the physical shape of the diversion barrier and the flow guiding structure in the microchannel are reconstructed.

[0076] The flow velocity data of the entire cross-section of the reconstructed flow channel is collected in real time by a flow velocity sensor. The measured flow velocity is compared with the standard laminar flow velocity threshold and the regional flow velocity target value. An angle correction amount is generated by relying on a closed-loop negative feedback control algorithm. The deflection angle of the baffles of each microelectromechanical adjustable structure is dynamically adjusted to stabilize the laminar flow state of the liquid flow, guide the fluid to avoid the low flow velocity stagnation area, and at the same time allow the liquid flow to concentrate on the metabolite enrichment area to complete the directional flushing.

[0077] In one possible implementation, step S400 further includes:

[0078] Step S410: Construct a nonlinear state-space model of metabolite concentration changes over time, and use the metabolite concentration gradient field data in the multi-dimensional real-time state data as the observation input.

[0079] Step S420: Using a recursive calculation method, the state gain matrix is ​​iteratively updated using the state estimate from the previous moment and the observation vector at the current moment, and the true concentration estimate of the metabolite and its rate of change over time are calculated in real time.

[0080] Step S430: When the rate of change is detected to change from positive to negative and the absolute value exceeds the preset threshold, the current moment is determined to be the preset peak window, a trigger signal is generated, the micro valve is controlled to intercept the micro liquid segment, and the micro liquid segment is delivered to the mass spectrometry analysis unit for mass spectrometry detection in real time through the micro pump.

[0081] Specifically, a nonlinear state-space model of metabolite concentration evolution over time is constructed using a nonlinear regression neural network. The model is divided into a state deduction branch and an observation mapping branch: first, a fully connected neural network is used to fit the state equation to learn the nonlinear temporal changes in concentration caused by metabolite diffusion, convection, and scouring within the microchannel; then, a shallow network is built to construct the observation equation and establish the mapping relationship between the internal state and the measured data; the model is trained under supervision using historical concentration time-series samples and concentration gradient field samples collected on-site, and the network weights and biases are optimized through backpropagation. After convergence, the network structure and parameters are fixed, and finally, the real-time collected and preprocessed metabolite concentration gradient field data is continuously fed into the observation terminal as dynamic input to the model.

[0082] Extended Kalman filtering is used to perform recursive operations. The estimated state value of metabolite concentration at the previous time step is retrieved and combined with the current concentration gradient field observation vector. The Kalman gain matrix is ​​updated round by round according to the filtering iteration formula to complete state prediction and error correction in sequence. The real estimated value of metabolite concentration is output in real time, and the rate of change of concentration over time is calculated synchronously through differential operation.

[0083] The system continuously monitors the numerical value and sign of the rate of change in metabolite concentration. Once the rate of change is detected to turn from positive to negative and its absolute value is greater than the preset threshold, the current period is immediately identified as the concentration peak window, and a hardware trigger signal is output to drive the microvalve to complete the interception of a micro-liquid segment. The micropump then quickly delivers the liquid segment to the mass spectrometry analysis unit to carry out the corresponding mass spectrometry detection operation.

[0084] In one possible implementation, step S200 further includes:

[0085] The edge computing unit achieves hardware acceleration through a field-programmable gate array, controlling the delay time from collecting the multi-dimensional real-time status data to generating the flow channel reconstruction instruction within a preset range.

[0086] Specifically, the edge computing unit relies on field-programmable gate arrays to achieve hardware acceleration, and performs parallel processing on the entire process from multi-dimensional real-time status data acquisition, operation and analysis to the generation of flow channel reconstruction instructions. It strictly controls the overall transmission and computing latency within a preset range, fully meeting the real-time operation requirements of single-cell metabolic dynamic tracking scenarios.

[0087] Example 2 is based on the same inventive concept as the intelligent control method for the extraction process based on a multi-dimensional dynamic algorithm in the previous examples, such as... Figure 2 As shown, this application provides an intelligent control system for the extraction process based on a multi-dimensional dynamic algorithm. The system and method embodiments in this application are based on the same inventive concept. The system includes:

[0088] The state data acquisition module 10 is used to acquire multi-dimensional real-time state data of the single-cell microenvironment within the microfluidic chip. The multi-dimensional real-time state data includes at least metabolite concentration gradient field data and cell physiological activity data.

[0089] The simulation prediction result generation module 20 is used to perform flow field-concentration field coupled simulation based on the multi-dimensional real-time state data and through the virtual simulation model driven by the edge computing unit to generate simulation prediction results. The virtual simulation model integrates fluid dynamics equations and material diffusion equations to predict the distribution of metabolites in the microchannel and the low-velocity stagnation area in real time.

[0090] The reconfiguration instruction generation module 30 is used to dynamically generate microchannel reconfiguration instructions based on the simulation prediction results using structural optimization logic, and based on the microchannel reconfiguration instructions, adjust the morphology of the flow divider barrier in the flow channel in real time through the microelectromechanical adjustable structure to guide the liquid flow to avoid the low-velocity stagnation area and maintain the laminar flow state.

[0091] The trigger signal generation module 40 is used to generate a trigger signal when the concentration of metabolites in the multi-dimensional real-time status data reaches a preset peak window, control the microvalve to perform a micro-precision capture action, and immediately deliver the captured sample to the mass spectrometry analysis unit.

[0092] Furthermore, the system is also used to implement the following functions:

[0093] An array of optical sensing units is deployed at key nodes of the microfluidic channel to monitor the adsorption kinetics curves of target metabolites secreted by single cells at the solid-liquid interface in real time. By demodulating the changes in reflected light intensity, the local refractive index is inverted to obtain high spatiotemporal resolution metabolite concentration gradient field data. Simultaneously, signals from electrochemical sensing units between the microfluidic electrodes are acquired, and the Nyquist plot features of the electrochemical impedance spectroscopy are analyzed to extract cell adhesion impedance and membrane capacitance parameters, thereby obtaining cell physiological activity data characterizing the degree of cell metabolic activity. The metabolite concentration gradient field data and the cell physiological activity data are integrated to form multi-dimensional real-time status data, which is then cached in the random access memory of the edge computing unit.

[0094] Furthermore, the system is also used to implement the following functions:

[0095] The edge computing unit calls the data interface to read the multi-dimensional real-time state data from the random access memory, and maps the metabolite concentration gradient field data in the multi-dimensional real-time state data to the pre-divided finite element mesh nodes; the virtual simulation model calls the flow field-concentration field coupled simulation, reads the cell physiological activity data in the multi-dimensional real-time state data as a correction factor, substitutes it into the material diffusion equation integrated by the virtual simulation model, and dynamically adjusts the diffusion coefficient tensor; the fluid dynamics equation integrated by the virtual simulation model is executed to calculate the three-dimensional velocity vector distribution of each mesh unit in the microchannel, and solves the material diffusion equation in combination with the adjusted diffusion coefficient tensor to obtain the distribution cloud map of metabolite concentration in space; the region with concentration higher than a first preset threshold in the distribution cloud map is identified as a metabolite accumulation region, and the mesh unit with velocity modulus lower than a second preset threshold in the three-dimensional velocity vector distribution is marked as a low velocity stagnation region, and the geometric coordinate information of the metabolite accumulation region and the low velocity stagnation region is encapsulated into the simulation prediction result.

[0096] Furthermore, the system is also used to implement the following functions:

[0097] Based on the flow field-concentration field coupled simulation, the cell membrane permeability parameter reflected by the cell physiological activity data in the multi-dimensional real-time state data is extracted. The effective diffusion coefficient in the material diffusion equation integrated by the virtual simulation model is calibrated using the cell membrane permeability parameter. The Navier-Stokes equation is iteratively solved using the fluid dynamics equation and the boundary conditions of the microchannel wall to obtain the velocity vector distribution of the microchannel cross section and longitudinal direction, and the pressure value and velocity component of each grid node are recorded. The diffusion flux vector of metabolites in the channel is calculated using the material diffusion equation and the convective mass transfer term brought about by the velocity vector distribution. The diffusion flux vector is composed of the superposition of molecular diffusion flux and convective diffusion flux. The velocity vector distribution and the diffusion flux vector are fused, and the simulation prediction result containing the set of metabolite aggregation coordinates, the set of flow dead zone coordinates, and the corresponding concentration prediction value is generated by the virtual simulation model.

[0098] Furthermore, the system is also used to implement the following functions:

[0099] The simulation prediction results are analyzed to extract the three-dimensional coordinate sets representing the low-velocity stagnation region and the metabolite enrichment region, and the fluid volume fraction occupied by the low-velocity stagnation region is calculated. A multi-objective optimization function with weighted coefficients is constructed, with the first optimization objective being minimizing the flow loss rate through the low-velocity stagnation region and the second optimization objective being maximizing the average metabolite concentration in the metabolite enrichment region. Through the structural optimization logic, the flow divider barrier within the microchannel is discretized into multiple deformable units, and the multi-objective optimization function is transformed into a mathematical programming problem concerning the pseudo-density variables of each deformable unit. Based on the mathematical programming problem, an iterative optimization calculation process is executed, adjusting the pseudo-density values ​​of each deformable unit until the Pareto front solution set of the multi-objective optimization function converges. The solution with the best overall performance is selected to generate a microchannel reconstruction command for reconstructing the physical morphology of the flow channel.

[0100] Furthermore, the system is also used to implement the following functions:

[0101] Calculate the sensitivity derivative of each design variable in the multi-objective optimization function, where the design variable is the pseudo-density of deformable elements corresponding to each grid node; based on the positive and negative directions and magnitudes of the sensitivity derivatives, update the pseudo-density distribution of nodes corresponding to the flow channel wall and flow divider barrier, including increasing the material density in high-sensitivity regions and decreasing the material density in low-sensitivity regions; determine whether the multi-objective optimization function value of the current iteration step falls within a preset convergence interval; if so, determine the current pseudo-density distribution matrix as the optimal topology; binarize the optimal topology to generate a binary driving signal matrix representing physical entities and voids, and encode the binary driving signal matrix into a time-series pulse sequence as a microchannel reconstruction command.

[0102] Furthermore, the system is also used to implement the following functions:

[0103] The binary drive signal in the microchannel reconstruction command is analyzed and converted into the execution displacement of each baffle in the microelectromechanical adjustable structure. The matrix-distributed microelectromechanical adjustable structure is driven to deflect according to the execution displacement to physically reconstruct the flow barrier morphology within the flow channel. The flow velocity distribution of the reconstructed flow channel cross section is monitored, and the deflection angle of the microelectromechanical adjustable structure is finely adjusted through a negative feedback mechanism to maintain the liquid flow in a laminar state and bypass the low-velocity stagnation area, thus concentrating on flushing the metabolite-rich area.

[0104] Furthermore, the system is also used to implement the following functions:

[0105] A nonlinear state-space model of metabolite concentration changing over time is constructed, and the metabolite concentration gradient field data in the multi-dimensional real-time state data is used as the observation input. Through recursive calculation, the state gain matrix is ​​iteratively updated using the state estimate value of the previous moment and the observation vector of the current moment, and the true concentration estimate value of the metabolite and its rate of change over time are calculated in real time. When the rate of change is detected to change from a positive value to a negative value and the absolute value exceeds a preset threshold, the current moment is determined to be a preset peak window, a trigger signal is generated, a microvalve is controlled to intercept a trace liquid segment, and the trace liquid segment is delivered to the mass spectrometry analysis unit for mass spectrometry detection in real time through a micropump.

[0106] Furthermore, the system is also used to implement the following functions:

[0107] The edge computing unit achieves hardware acceleration through a field-programmable gate array, controlling the delay time from collecting the multi-dimensional real-time status data to generating the flow channel reconstruction instruction within a preset range.

[0108] It should be noted that the order of the embodiments described above is for descriptive purposes only and does not represent the superiority or inferiority of the embodiments. Specific embodiments of this specification have been described above. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0109] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0110] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method for intelligent control of the extraction process based on a multi-dimensional dynamic algorithm, characterized in that, The method includes: Acquire multi-dimensional real-time state data of the single-cell microenvironment within a microfluidic chip, wherein the multi-dimensional real-time state data includes at least metabolite concentration gradient field data and cell physiological activity data. Based on the multi-dimensional real-time status data, a flow field-concentration field coupled simulation is performed by a virtual simulation model driven by an edge computing unit to generate simulation prediction results. The virtual simulation model integrates fluid dynamics equations and mass diffusion equations to predict the distribution of metabolites in the microchannel and the low-velocity stagnation area in real time. Based on the simulation prediction results, microchannel reconstruction instructions are dynamically generated using structural optimization logic. Based on the microchannel reconstruction instructions, the morphology of the flow divider barrier in the flow channel is adjusted in real time through a microelectromechanical adjustable structure to guide the liquid flow to avoid low-velocity stagnation areas and maintain laminar flow. When the concentration of metabolites in the multidimensional real-time status data reaches a preset peak window, a trigger signal is generated to control the microvalve to perform a micro-precision capture action and immediately deliver the captured sample to the mass spectrometry analysis unit.

2. The intelligent control method for the extraction process based on a multi-dimensional dynamic algorithm as described in claim 1, characterized in that, Acquire multi-dimensional real-time state data of the single-cell microenvironment within a microfluidic chip, including: An array of optical sensing units is deployed at key nodes of the microchannel to monitor the adsorption kinetics curve of target metabolites secreted by single cells at the solid-liquid interface in real time. The local refractive index is inverted by demodulating the change in reflected light intensity to obtain metabolite concentration gradient field data with high spatiotemporal resolution. The signals of the electrochemical sensing unit between the microchannel electrodes were acquired simultaneously, the Nyquist plot characteristics of the electrochemical impedance spectroscopy were analyzed, and the cell adhesion impedance and membrane capacitance parameters were extracted to obtain cell physiological activity data characterizing the degree of cell metabolic activity. The metabolite concentration gradient field data is integrated with the cell physiological activity data to form multi-dimensional real-time state data, and the multi-dimensional real-time state data is cached in the random access memory of the edge computing unit.

3. The intelligent control method for the extraction process based on a multi-dimensional dynamic algorithm as described in claim 1, characterized in that, Based on the aforementioned multi-dimensional real-time state data, a flow field-concentration field coupled simulation is performed using a virtual simulation model driven by an edge computing unit to generate simulation prediction results, including: The edge computing unit calls the data interface to read the multi-dimensional real-time status data from the random access memory, and maps the metabolite concentration gradient field data in the multi-dimensional real-time status data to the pre-divided finite element mesh nodes. The virtual simulation model is used to call the flow field-concentration field coupled simulation, and the cell physiological activity data in the multi-dimensional real-time state data is read as a correction factor and substituted into the material diffusion equation integrated by the virtual simulation model to dynamically adjust the diffusion coefficient tensor. The fluid dynamics equations integrated in the virtual simulation model are executed to calculate the three-dimensional velocity vector distribution of each grid cell in the microchannel, and the material diffusion equation is solved by combining the adjusted diffusion coefficient tensor to obtain the spatial distribution cloud map of metabolite concentration. The regions in the distribution cloud map with concentrations higher than a first preset threshold are identified as metabolite accumulation areas, and the grid cells in the three-dimensional velocity vector distribution with velocity modulus lower than a second preset threshold are marked as low-velocity stagnation areas. The geometric coordinate information of the metabolite accumulation areas and the low-velocity stagnation areas is encapsulated into the simulation prediction results.

4. The intelligent control method for the extraction process based on a multi-dimensional dynamic algorithm as described in claim 3, characterized in that, The method of using the virtual simulation model to call the flow field-concentration field coupled simulation also includes: Based on the flow field-concentration field coupled simulation, the cell membrane permeability parameter reflected by the cell physiological activity data in the multi-dimensional real-time state data is extracted, and the effective diffusion coefficient in the material diffusion equation integrated by the virtual simulation model is calibrated using the cell membrane permeability parameter. By combining the fluid dynamics equations with the boundary conditions of the microchannel wall, the Navier-Stokes equations are solved iteratively to obtain the velocity vector distribution of the microchannel cross section and longitudinal direction, and the pressure value and velocity component of each grid node are recorded. Using the aforementioned material diffusion equation and the convective mass transfer term resulting from the velocity vector distribution, the diffusion flux vector of metabolites within the flow channel is calculated. This diffusion flux vector is formed by the superposition of molecular diffusion flux and convective diffusion flux. By integrating the velocity vector distribution and the diffusion flux vector, the virtual simulation model generates simulation prediction results that include a set of metabolite aggregation coordinates, a set of flow dead zone coordinates, and corresponding concentration prediction values.

5. The intelligent control method for the extraction process based on a multi-dimensional dynamic algorithm as described in claim 1, characterized in that, Based on the simulation prediction results, microchannel reconfiguration instructions are dynamically generated using structural optimization logic, including: The simulation prediction results are analyzed to extract the three-dimensional coordinate set representing the low-flow-rate retention region and the three-dimensional coordinate set representing the metabolite enrichment region, and the fluid volume fraction occupied by the low-flow-rate retention region is calculated. A multi-objective optimization function with weighted coefficients is constructed, with minimizing the flow loss rate through the low-flow-rate retention area as the first optimization objective and maximizing the average metabolite concentration in the metabolite enrichment area as the second optimization objective. Through the structural optimization logic, the flow divider barrier in the microchannel is discretized into multiple deformable units, and the multi-objective optimization function is transformed into a mathematical programming problem about the pseudo-density variables of each deformable unit. Based on the mathematical programming problem, an iterative optimization calculation process is performed to adjust the pseudo-density values ​​of each deformable element until the Pareto front solution set of the multi-objective optimization function converges. The solution with the best overall performance is selected to generate a microchannel reconstruction instruction for reconstructing the physical morphology of the flow channel.

6. The intelligent control method for the extraction process based on a multi-dimensional dynamic algorithm as described in claim 5, characterized in that, The iterative optimization calculation process includes: Calculate the sensitivity derivative of each design variable in the multi-objective optimization function, where the design variable is the pseudo density of the deformable element corresponding to each grid node; Based on the positive and negative directions and magnitude of the sensitivity derivative, update the pseudo-density distribution of the corresponding nodes on the flow channel wall and the diversion barrier, including increasing the material density in the high-sensitivity region and decreasing the material density in the low-sensitivity region. Determine whether the multi-objective optimization function value of the current iteration step falls within the preset convergence interval. If so, determine the current pseudo-density distribution matrix as the optimal topology. The optimal topology is binarized to generate a binary driving signal matrix representing physical entities and gaps. The binary driving signal matrix is ​​then encoded into a time-series pulse sequence as a microchannel reconfiguration instruction.

7. The intelligent control method for the extraction process based on a multi-dimensional dynamic algorithm as described in claim 1, characterized in that, Based on the microchannel reconfiguration command, the morphology of the flow divider barrier within the channel is adjusted in real time through a microelectromechanical adjustable structure to guide the liquid flow away from low-velocity stagnation regions and maintain a laminar flow state, including: The binary drive signal in the microchannel reconfiguration instruction is analyzed and converted into the displacement of each baffle in the microelectromechanical adjustable structure. The microelectromechanical adjustable structure, which is driven in a matrix distribution, deflects according to the displacement to physically reconstruct the morphology of the flow channel's diversion barrier. The flow velocity distribution of the reconstructed flow channel cross section is monitored, and the deflection angle of the microelectromechanical adjustable structure is finely adjusted through a negative feedback mechanism to maintain the liquid flow in a laminar state and bypass the low-velocity stagnation area, thereby concentrating on flushing the metabolite-rich area.

8. The intelligent control method for the extraction process based on a multi-dimensional dynamic algorithm as described in claim 1, characterized in that, When the concentration of metabolites in the multi-dimensional real-time status data reaches a preset peak window, a trigger signal is generated to control the microvalve to perform a precise micro-capture action and immediately deliver the captured sample to the mass spectrometry analysis unit, including: A nonlinear state-space model of metabolite concentration changing over time is constructed, and the metabolite concentration gradient field data in the multi-dimensional real-time state data is used as the observation input; By using a recursive calculation method, the state gain matrix is ​​iteratively updated using the state estimate from the previous moment and the observation vector from the current moment, and the true concentration estimate of metabolites and their rate of change over time are calculated in real time. When the rate of change is detected to change from positive to negative and the absolute value exceeds a preset threshold, the current moment is determined to be a preset peak window, a trigger signal is generated, the microvalve is controlled to intercept a trace liquid segment, and the trace liquid segment is delivered to the mass spectrometry analysis unit for mass spectrometry detection in real time through a micropump.

9. The intelligent control method for the extraction process based on a multi-dimensional dynamic algorithm as described in claim 1, characterized in that, The edge computing unit achieves hardware acceleration through a field-programmable gate array, controlling the delay time from collecting the multi-dimensional real-time status data to generating the flow channel reconstruction instruction within a preset range.

10. An intelligent control system for the extraction process based on a multi-dimensional dynamic algorithm, characterized in that, The system is used to implement the intelligent control method for the extraction process based on a multi-dimensional dynamic algorithm as described in any one of claims 1-9, and the system comprises: The state data acquisition module is used to acquire multi-dimensional real-time state data of the single-cell microenvironment within the microfluidic chip. The multi-dimensional real-time state data includes at least metabolite concentration gradient field data and cell physiological activity data. The simulation prediction result generation module is used to perform flow field-concentration field coupled simulation based on the multi-dimensional real-time state data and through the virtual simulation model driven by the edge computing unit to generate simulation prediction results. The virtual simulation model integrates fluid dynamics equations and material diffusion equations to predict the distribution of metabolites in the microchannel and the low-velocity stagnation area in real time. The reconfiguration instruction generation module is used to dynamically generate microchannel reconfiguration instructions based on the simulation prediction results using structural optimization logic, and based on the microchannel reconfiguration instructions, adjust the morphology of the flow divider barrier in the flow channel in real time through the microelectromechanical adjustable structure to guide the liquid flow to avoid the low flow velocity stagnation area and maintain the laminar flow state. The trigger signal generation module is used to generate a trigger signal when the concentration of metabolites in the multi-dimensional real-time status data reaches a preset peak window, control the microvalve to perform a micro-precision capture action, and immediately deliver the captured sample to the mass spectrometry analysis unit.

Images