Closed-loop control digital microfluidic system and method thereof

Abstract

The invention discloses a closed-loop control digital microfluidic system and a method thereof, and relates to the technical field of microfluidics, and the method comprises the following steps: initializing a digital microfluidic chip electrode array, setting an electrode spacing, an electrode size and a voltage upper limit, collecting a no-load state signal, and generating initial reference data; liquid drops are formed at the preset position of the digital micro-fluidic chip, and according to the initial reference data, the initial position, the volume and the form information of the liquid drops are collected, and a liquid drop initial state data set is generated; inputting the initial state data set of the liquid drop into a pre-trained closed-loop prediction model, performing kinetic analysis and trajectory prediction calculation, generating an expected motion trajectory of the liquid drop, and comparing the expected motion trajectory with the real-time state data flow of the liquid drop to generate deviation feedback data of the liquid drop; and dynamically adjusting the initial driving voltage parameter according to the droplet deviation feedback data to generate a closed-loop driving voltage sequence. The liquid drop trajectory control precision is improved, and motion errors caused by environmental disturbance and electrode response differences are reduced.

Description

Technical Field

[0001] This invention relates to the field of microfluidics, and in particular to a closed-loop control digital microfluidic system and method thereof. Background Technology

[0002] Microfluidics is a cutting-edge technology for the precise manipulation of fluids at the microscale, widely used in fields such as bioanalysis, chemical reactions, drug screening, and laboratory automation. With the development of digital microfluidic chips, the electrode array-driven droplet manipulation method has become a key approach for achieving high-throughput, multifunctional droplet processing. Existing digital microfluidic methods typically rely on preset voltage sequences to drive droplets onto the chip surface for distribution, mixing, or transfer, achieving experimental operations through droplet generation, movement, and merging.

[0003] Existing digital microfluidic technologies still have room for improvement in droplet generation, trajectory prediction, and state feedback. In the early stages of droplet generation, the inhomogeneity of droplet volume, morphology, and initial position directly affects the stability of subsequent motion trajectories. Open-loop control methods struggle to calibrate deviations in a timely manner, leading to decreased droplet manipulation accuracy. In parallel multi-droplet operations or complex path planning, the lack of a systematic, closed-loop prediction and adjustment mechanism limits the scalability and automation level of droplet manipulation. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a closed-loop control digital microfluidic method to solve the problems of insufficient droplet motion accuracy and difficulty in real-time adjustment.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a closed-loop control digital microfluidic method, comprising,

[0008] Initialize the digital microfluidic chip electrode array, set the electrode spacing, electrode size and voltage limit, and collect the no-load state signal to generate initial reference data;

[0009] A droplet is formed at a predetermined location on the digital microfluidic chip, and the initial position, volume and morphology information of the droplet are collected based on the initial reference data to generate a droplet initial state dataset.

[0010] The initial state dataset of the droplet is input into the pre-trained closed-loop prediction model to perform dynamic analysis and trajectory prediction calculation, generate the expected motion trajectory of the droplet, and compare it with the real-time state data stream of the droplet to generate droplet deviation feedback data.

[0011] Based on the droplet deviation feedback data, the initial driving voltage parameters are dynamically adjusted to generate a closed-loop driving voltage sequence;

[0012] A closed-loop driving voltage sequence is applied to the electrode array, droplet state data is collected in real time, droplet operation commands are generated, and integrated to form a global droplet state matrix.

[0013] Based on the global droplet state matrix, droplet operation status is recorded, and a droplet operation log file is generated.

[0014] In a preferred embodiment of the closed-loop control digital microfluidic method of the present invention, the specific steps for generating initial reference data are as follows:

[0015] Read the digital microfluidic chip model and electrode array design parameters, calculate the electrode spacing, size and material properties, and generate electrode layout configuration data;

[0016] The electrode layout configuration data is loaded into the controller, the initial voltage upper limit is set and a low amplitude test voltage is applied, and at the same time, the no-load state sensor signal is acquired to generate no-load reference data.

[0017] Based on the no-load reference data, the local response and dielectric layer difference of each electrode are calculated to generate electrode calibration parameters, which are then integrated with the voltage distribution table to generate initial reference data.

[0018] As a preferred embodiment of the closed-loop control digital microfluidic method of the present invention, the specific steps for generating the droplet initial state dataset are as follows:

[0019] Based on the initial reference data, the target position and initial voltage configuration for droplet generation on the digital microfluidic chip are calculated to obtain droplet generation control commands.

[0020] According to the droplet generation control command, a voltage pulse is applied at a predetermined position on the digital microfluidic chip to form a droplet. At the same time, the position, volume and morphology sensor signals of the droplet are collected in real time during the droplet formation process to generate preliminary data on droplet formation.

[0021] The preliminary data on droplet formation are analyzed to extract the actual position, volume, and morphology of the droplets, and then integrated to generate the initial state data of the droplets.

[0022] As a preferred embodiment of the closed-loop control digital microfluidic method of the present invention, the pre-trained closed-loop prediction model is constructed based on the droplet initial state dataset, historical droplet motion data and electrode calibration parameters, through dynamic analysis and trajectory prediction algorithms, and is obtained by pre-training with historical droplet motion data.

[0023] As a preferred embodiment of the closed-loop control digital microfluidic method of the present invention, the specific steps for generating the expected trajectory of the droplet are as follows:

[0024] Input the initial state dataset of droplets into the pre-trained closed-loop prediction model to establish a droplet dynamics prediction environment;

[0025] In the droplet dynamics prediction environment, dynamic analysis is performed on the initial state dataset of droplets to generate preliminary driving voltage parameters;

[0026] Trajectory prediction calculations are performed on the initial driving voltage parameters to generate the expected trajectory of the droplet.

[0027] In a preferred embodiment of the closed-loop control digital microfluidic method of the present invention, the specific steps for generating droplet deviation feedback data are as follows:

[0028] Real-time acquisition of droplet current position, volume and morphology information, generating droplet real-time status data stream;

[0029] The real-time status data stream of the droplet is compared with the expected trajectory of the droplet to extract the deviation between the actual position, volume and shape of the droplet and the expected trajectory, and then integrated to generate droplet trajectory deviation data.

[0030] Based on the droplet trajectory deviation data, the amount and direction of the droplet deviation are calculated, and droplet deviation feedback data is generated.

[0031] In a preferred embodiment of the closed-loop control digital microfluidic method of the present invention, the specific steps for generating the closed-loop driving voltage sequence are as follows:

[0032] Analyze the droplet deviation feedback data for deviation amount and direction to generate feedback analysis data;

[0033] Based on the feedback analysis data, a voltage adjustment algorithm is used to calculate the correction value of the initial drive voltage parameters and generate a voltage adjustment scheme.

[0034] Based on the voltage adjustment scheme, the initial driving voltage parameters are dynamically modified to generate a closed-loop driving voltage sequence.

[0035] As a preferred embodiment of the closed-loop control digital microfluidic method of the present invention, the specific steps for forming the global droplet state matrix are as follows:

[0036] The closed-loop driving voltage sequence is loaded onto the electrode array, and the voltage amplitude, pulse timing and the corresponding application order of each electrode are analyzed to generate electrode driving task instructions.

[0037] According to the electrode driving task instructions, a closed-loop driving voltage is applied to the electrode array, and the droplet position, volume and morphology are collected in real time to generate a droplet state data stream.

[0038] The droplet state data stream is analyzed and integrated to generate droplet operation instructions, and the state of each droplet is summarized to form a global droplet state matrix.

[0039] In a preferred embodiment of the closed-loop control digital microfluidic method of the present invention, the specific steps for generating the droplet operation record file are as follows:

[0040] Based on the global droplet state matrix, the operation events of each droplet are numbered, classified, and time-stamped to generate droplet operation log entries;

[0041] The droplet operation log entries are summarized, formatted, and saved to generate a droplet operation record file.

[0042] Secondly, the present invention provides a closed-loop control digital microfluidic system, comprising,

[0043] The electrode initialization module is used to initialize the electrode array of the digital microfluidic chip, set the electrode spacing, electrode size and voltage limit, and collect no-load state signals to generate initial reference data.

[0044] The droplet generation module is used to form droplets at predetermined positions on the digital microfluidic chip, and to collect the initial position, volume and morphology information of the droplets based on the initial reference data to generate a droplet initial state dataset.

[0045] The trajectory prediction module is used to input the initial state dataset of the droplet into the pre-trained closed-loop prediction model, perform dynamic analysis and trajectory prediction calculation, generate the expected motion trajectory of the droplet, and compare it with the real-time state data stream of the droplet to generate droplet deviation feedback data.

[0046] The voltage optimization module is used to dynamically adjust the initial driving voltage parameters based on droplet deviation feedback data to generate a closed-loop driving voltage sequence.

[0047] The state control module is used to apply the closed-loop drive voltage sequence to the electrode array, collect droplet state data in real time, generate droplet operation commands, and integrate them to form a global droplet state matrix.

[0048] The record generation module is used to record droplet operation information based on the global droplet state matrix and generate droplet operation record files.

[0049] The beneficial effects of this invention are as follows: by analyzing the deviation amount and direction of the droplet deviation feedback data, and using a voltage adjustment algorithm to dynamically modify the initial driving voltage parameters, a closed-loop driving voltage sequence is generated, realizing real-time closed-loop control during the droplet motion process. It can immediately perform adaptive correction when the droplet deviates from the expected trajectory, ensuring that the droplet moves stably along the planned path. This achieves the effects of improving the accuracy of droplet trajectory control, reducing motion errors caused by environmental disturbances and differences in electrode response, and ensuring that the droplet operation process has high precision, dynamic adjustability, and high repeatability. Attached Figure Description

[0050] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the 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.

[0051] Figure 1 This is a flowchart of a closed-loop control digital microfluidic method.

[0052] Figure 2 This is a schematic diagram of a closed-loop control digital microfluidic system.

[0053] Figure 3 A flowchart for generating the initial state dataset of droplets.

[0054] Figure 4 A flowchart for generating the closed-loop drive voltage sequence. Detailed Implementation

[0055] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0056] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0057] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0058] Reference Figures 1-4As one embodiment of the present invention, this embodiment provides a closed-loop control digital microfluidic method, comprising the following steps:

[0059] S1. Initialize the digital microfluidic chip electrode array, set the electrode spacing, electrode size and voltage limit, and collect the no-load state signal to generate initial reference data.

[0060] S1.1 Read the digital microfluidic chip model and electrode array design parameters, calculate the electrode spacing, size and material properties, and generate electrode layout configuration data.

[0061] Specifically, the process involves reading the digital microfluidic chip model and electrode array design parameters, parsing and extracting parameters such as the number of electrodes, the number of array rows and columns, the cell spacing, the minimum allowable feature size, the dielectric layer thickness, and the material type. A cell-by-cell coordinate calculation method is used, with the array row and column number as an index, to sequentially calculate the row and column number of each electrode in the array, and to calculate the center coordinates of adjacent electrodes based on the cell spacing. A geometric quantization method based on adjacent cells is used to determine the width and length of the electrodes based on the minimum allowable feature size and the duty cycle (e.g., when the duty cycle is 0.6, the electrode width is the minimum allowable feature size × 0.6), and to calculate the electrode layer thickness and conductive layer thickness (e.g., the conductive layer thickness is 200 nanometers) in conjunction with the dielectric layer thickness and material type. The center coordinates, width, length, height, and material properties of each electrode are calculated and recorded by field, and these records are merged to generate electrode layout configuration data.

[0062] It should also be noted that the unit-by-unit coordinate calculation method refers to calculating the center coordinates of each electrode one by one, using the row and column numbers of the electrode in the array as indexes; the adjacent unit geometric quantization method refers to determining the geometric dimensions (such as width and length) of adjacent electrodes into specific values ​​through quantization calculation based on the electrode spacing and the minimum allowable feature size.

[0063] S1.2 Load the electrode layout configuration data into the controller, set the initial voltage upper limit and apply a low amplitude test voltage, and at the same time collect the no-load state sensor signal to generate no-load reference data.

[0064] Specifically, after loading the electrode layout configuration data into the controller, the electrode coordinates, area parameters, and connection methods in the electrode layout configuration data are read sequentially according to the electrode number. An initial voltage upper limit is set in the controller, and a low-amplitude test voltage is applied to each electrode in sequence according to the electrode number. For example, a low-amplitude test voltage of 0.5V is applied. During the application of the low-amplitude test voltage, the controller's sampling command is called simultaneously to obtain the voltage quantization value, current quantization value, and impedance quantization value of the no-load state sensor signal. After the low-amplitude test voltage is applied to all electrodes, the voltage quantization value, current quantization value, and impedance quantization value of the corresponding no-load state sensor signal are arranged according to the electrode number to generate no-load reference data.

[0065] S1.3. Based on the no-load reference data, calculate the local response and dielectric layer difference of each electrode, generate electrode calibration parameters, and integrate them with the voltage distribution table to generate initial reference data.

[0066] Specifically, based on the no-load reference data, the voltage quantization value, current quantization value, and impedance quantization value are read sequentially according to the electrode number. Utilizing the no-load response characteristics of each electrode under the same low-amplitude test voltage, the local response of each electrode is obtained through the differential ratio method. For example, the difference in impedance quantization values ​​between adjacent electrodes is used as the local response quantization result. The local response of each electrode is compared with the dielectric layer thickness quantization value and material parameter quantization value recorded in the electrode layout configuration data. The dielectric layer difference quantization value is calculated according to the dielectric layer difference evaluation rules. The local response quantization value and dielectric layer difference quantization value of each electrode are combined to form the electrode calibration parameters. After generating the electrode calibration parameters, the target voltage quantization value is read from the voltage allocation table according to the electrode number. The local response quantization value and dielectric layer difference quantization value in the electrode calibration parameters are superimposed on the target voltage quantization value to generate the adjusted voltage quantization value. All adjusted voltage quantization values ​​are arranged according to the electrode number to generate the initial reference data.

[0067] It should also be noted that the differential ratio method calculates the signal difference between the no-load responses of adjacent electrodes and then performs a ratio operation with the reference signal to quantify the response differences between adjacent electrodes, so as to obtain the local calibration parameters of each electrode, which facilitates subsequent voltage adjustment and precise control of droplet movement.

[0068] The dielectric layer difference assessment rule is to compare the response signals of each electrode under no-load conditions, and combine the material dielectric constant and layer thickness information to quantitatively evaluate the dielectric layer characteristics covered by different electrodes, thereby determining the impact of dielectric layer differences on the local response of the electrodes, which is used to generate electrode calibration parameters to ensure the accuracy of voltage application and the controllability of droplet movement.

[0069] The voltage allocation table is a voltage mapping table pre-defined according to the chip design and the target droplet operation process. It is used to assign a basic driving voltage and application time parameters to each electrode and can be obtained through experimental calibration or simulation calculation.

[0070] S2. A droplet is formed at a predetermined position on the digital microfluidic chip, and the initial position, volume and morphology information of the droplet are collected based on the initial reference data to generate a droplet initial state dataset.

[0071] S2.1. Based on the initial reference data, calculate the target position and initial voltage configuration for droplet generation on the digital microfluidic chip to obtain the droplet generation control command.

[0072] Specifically, based on the electrode layout configuration data, electrode calibration parameters, and no-load response information in the initial reference data, the target coordinate mapping method is used to calculate the droplet generation position of the digital microfluidic chip unit by unit coordinate. Combined with the droplet volume and generation sequence, the initial voltage configuration corresponding to each droplet is calculated, and droplet generation control instructions are integrated to generate droplet generation control instructions. For example, the applied voltage amplitude and duration are assigned to each target position.

[0073] It should also be noted that the target coordinate mapping method refers to mapping the target spatial coordinates of each droplet to the corresponding electrode array electrode according to the predetermined droplet generation position on the digital microfluidic chip. By considering the electrode geometry configuration, electrode spacing and droplet volume, the corresponding voltage amplitude and duration are assigned to each electrode array electrode to form a droplet generation control command.

[0074] Droplet volume and generation sequence refer to the specific volume setting for each droplet on a digital microfluidic chip to ensure that droplets are generated according to the path and functional requirements, and to determine the order in which the droplets are generated during the generation process. The droplet volume is used to calculate the amplitude and duration of the required applied voltage, and the generation sequence is used to arrange the time sequence of electrode driving, thereby ensuring that the droplets are formed sequentially according to the target position and do not interfere with each other.

[0075] S2.2. According to the droplet generation control command, a voltage pulse is applied to a predetermined position on the digital microfluidic chip to form a droplet. At the same time, the position, volume and morphology sensor signals of the droplet are collected in real time to generate preliminary data on droplet formation.

[0076] Specifically, according to the droplet generation control command, a voltage pulse with a set amplitude and duration is applied to the electrode array electrodes at a predetermined position on the digital microfluidic chip, such as a 5-volt voltage pulse lasting 20 milliseconds, causing the droplet to detach from the liquid storage area and generate a droplet at the target position. At the same time, position sensors, volume sensors, and morphology sensors are used to collect the position information, volume change, and morphological feature signals of the droplet during the generation process in real time, and integrate them according to the time series to generate preliminary droplet formation data. The preliminary droplet formation data includes droplet image frames acquired by high-speed imaging and the parsed position information / volume / morphological features.

[0077] It should also be noted that the electrode array at the predetermined position refers to the specific electrodes arranged on the digital microfluidic chip according to the droplet operation plan, which are used to apply voltage to generate or move droplets. Each electrode has a defined spatial coordinate and electrical connection, which is used to precisely control the formation position and initial movement direction of the droplet on the chip.

[0078] A voltage pulse with set amplitude and duration refers to an electrical pulse signal with a defined voltage magnitude (e.g., in volts) and duration (e.g., in milliseconds) applied to a specific electrode of a digital microfluidic chip. It is used to drive droplets to form or move at a predetermined position. The amplitude controls the magnitude of the force on the droplet, and the duration determines the time the droplet is subjected to force, thereby achieving precise droplet generation and motion control.

[0079] S2.3 Analyze the preliminary data of droplet formation, extract the actual position, volume and morphology information of the droplets, and integrate them to generate the initial state data of the droplets.

[0080] Specifically, image processing methods are used to analyze the coordinates of the positional information of the initial droplet formation data to extract the actual position of each droplet on the chip; numerical calculations are performed on the volume changes of the initial droplet formation data to obtain the actual volume of the droplets; contour recognition and morphological feature extraction are performed on the morphological feature signals of the initial droplet formation data to obtain the actual shape of the droplets; and the actual position, volume, and morphological information of the droplets are integrated to generate the initial state data of the droplets.

[0081] It should also be noted that image processing methods refer to the methods and means of obtaining the spatial location, size, shape or other feature information of a target object by preprocessing, segmenting, extracting features and analyzing acquired digital images.

[0082] S3. Input the initial state dataset of the droplet into the pre-trained closed-loop prediction model, perform dynamic analysis and trajectory prediction calculation, generate the expected motion trajectory of the droplet, and compare it with the real-time state data stream of the droplet to generate droplet deviation feedback data.

[0083] S3.1 The pre-trained closed-loop prediction model is constructed based on the droplet initial state dataset, historical droplet motion data, and electrode calibration parameters, through dynamic analysis and trajectory prediction algorithms, and is obtained through pre-training using historical droplet motion data.

[0084] Specifically, the initial state dataset of droplets, historical droplet motion data, and electrode calibration parameters are standardized to unify the data format and dimensions. Based on droplet dynamics analysis methods, trajectory analysis and motion feature extraction are performed on the historical droplet motion data to form a training sample set. A trajectory prediction algorithm is used to learn from the training sample set, constructing a droplet motion prediction relationship, and the prediction parameters are iteratively adjusted to minimize the deviation between the actual motion and the expected trajectory. The predicted parameters obtained from the training, the initial state dataset of droplets, and the electrode calibration parameters are combined to form a pre-trained closed-loop prediction model for subsequent droplet dynamics analysis and trajectory prediction.

[0085] It should also be noted that historical droplet motion data refers to the set of motion information of droplets collected and recorded in real time during previous operations on digital microfluidic chips, including the specific position coordinates, volume changes and morphological characteristics of droplets at different time points. It is used to reflect the motion law and response characteristics of droplets under different electrode excitations and can provide a reference for dynamic analysis and trajectory prediction.

[0086] Droplet dynamics analysis refers to the method of modeling and calculating the motion law of droplets on microfluidic chips driven by electrodes based on droplet initial state datasets, historical droplet motion data and electrode calibration parameters. By considering the changes in droplet position, volume, shape and the effect of electrode electric field, the motion trend and dynamic response characteristics of droplets are obtained.

[0087] Trajectory prediction algorithms are methods that, based on droplet dynamics analysis, use historical droplet motion patterns and current initial states to numerically predict the motion path of a droplet over a future time period, generating the droplet's expected trajectory and providing a reference for closed-loop control.

[0088] S3.2 Input the initial state dataset of the droplets into the pre-trained closed-loop prediction model to establish a droplet dynamics prediction environment.

[0089] Specifically, the initial state dataset of droplets is loaded one by one into the pre-trained closed-loop prediction model according to the position, volume and morphology information of each droplet. By reading historical droplet motion data and electrode calibration parameters, the motion response of droplets under electrode driving action is calculated according to the droplet dynamics analysis method. The initial state and historical motion law of each droplet are mapped to the droplet dynamics prediction environment to establish the droplet dynamics prediction environment. For example, corresponding electrode force parameters and time steps are assigned to each droplet to form an initial calculation environment that can be used for trajectory prediction.

[0090] It should also be noted that the droplet dynamics analysis method refers to the calculation method that combines fluid dynamics and electrodynamics based on the initial state dataset of the droplet, the physical properties of the droplet, and the electrode calibration parameters. It models and simulates the motion law of the droplet under the electric field driven on the digital microfluidic chip. Specifically, it includes: calculating the position, velocity, and shape changes of the droplet under the driving force generated by the voltage applied by each electrode, surface tension, and viscous resistance; iteratively updating the droplet state step by step over time; and predicting the dynamic response of the droplet on the microfluidic channel or open surface, thereby providing an accurate dynamic basis for subsequent trajectory prediction and closed-loop control.

[0091] S3.3 In the droplet dynamics prediction environment, perform dynamic analysis on the initial state dataset of the droplets to generate preliminary driving voltage parameters.

[0092] Specifically, in the droplet dynamics prediction environment, droplet dynamics analysis methods are used to calculate the initial position, volume, and morphology information of each droplet in the initial state dataset. Combined with electrode calibration parameters and historical droplet motion data, the forces acting on the droplet under the applied voltage of each electrode are simulated step by step over time, including electrodynamic force, surface tension, and viscous drag. The droplet position and morphology are updated, and the dynamic response of the droplet along the predetermined path is calculated sequentially. Finally, the initial driving voltage parameters that can drive the droplet to move along the expected trajectory are determined, such as the voltage amplitude and application sequence of each electrode.

[0093] It should also be noted that the droplet dynamics prediction environment refers to a simulation computing environment built on the basis of the droplet initial state dataset, historical droplet motion data and electrode calibration parameters, through dynamic analysis methods and trajectory prediction algorithms. It is used to simulate the motion behavior and morphological changes of droplets under different electrode voltages. It can predict and verify the expected motion trajectory of droplets under virtual conditions, and provide a basis for deviation analysis and driving voltage adjustment for closed-loop control.

[0094] S3.4 Perform trajectory prediction calculations on the initial driving voltage parameters to generate the expected trajectory of the droplet.

[0095] Specifically, in the droplet dynamics prediction environment, a trajectory prediction algorithm is used to apply the initial driving voltage parameters to each droplet in the initial state dataset in sequence. The position changes and morphological evolution of the droplets under the action of each electrode voltage are calculated according to the time step. Combined with the droplet dynamics analysis results, the future motion path of the droplets is interpolated and the trajectory is reconstructed to generate the expected motion trajectory of the droplets, including the spatial coordinates, volume and morphological state of each droplet at each time point. For example, the position change curve and volume change curve of the droplets are recorded at millisecond time intervals.

[0096] It should also be noted that trajectory prediction algorithms refer to methods for calculating and extrapolating the future motion path of droplets on microfluidic chips using droplet initial state datasets, historical droplet motion data, and electrode calibration parameters. These methods typically include time series analysis, solving dynamic models, and numerical integration to generate trajectory data of the droplet's expected position, volume, and morphology changing over time, providing reference and deviation feedback for closed-loop control.

[0097] S3.5. Real-time acquisition of the droplet's current position, volume, and morphology information to generate a real-time droplet status data stream.

[0098] Specifically, a high-speed image acquisition method is used to continuously capture images of the current position of the droplet on the digital microfluidic chip. Each frame of the image is processed to extract the spatial coordinates, volume, and morphological features of the droplet. The information of each frame of the droplet is integrated in chronological order to generate continuous time series data containing the droplet's position, volume, and morphology, thus generating a real-time state data stream of the droplet.

[0099] It should also be noted that the high-speed image acquisition method refers to continuously acquiring images of droplet motion on a microfluidic chip at a high frame rate (e.g., thousands of frames per second or higher) using a high-speed camera or high-speed imaging machine, and storing each frame in digital form to ensure that the minute position, volume and morphological changes of the droplet in a short period of time can be captured, so as to achieve continuous and accurate recording of the entire process of droplet motion.

[0100] S3.6. Compare the real-time status data stream of the droplet with the expected trajectory of the droplet, extract the deviation between the actual position, volume and shape of the droplet and the expected trajectory, and integrate them to generate droplet trajectory deviation data.

[0101] Specifically, using the Euclidean distance calculation method, the position of each droplet in the real-time state data stream is compared with the position of the droplet at the corresponding time point of the expected motion trajectory, and the position deviation vector is calculated. At the same time, the difference between the real-time volume of the droplet and the expected volume of the droplet is calculated to extract the volume deviation. The difference between the real-time morphological features of the droplet (such as aspect ratio and roundness) and the expected morphological features of the droplet is calculated to extract the morphological deviation. The position deviation, volume deviation and morphological deviation are integrated according to the time series to generate droplet trajectory deviation data. Each droplet contains records of position deviation, volume deviation and morphological deviation at each time point.

[0102] S3.7. Based on the droplet trajectory deviation data, calculate the amount and direction of the droplet deviation, and generate droplet deviation feedback data.

[0103] Specifically, based on the droplet trajectory deviation data, the Euclidean distance calculation method is used to calculate the square root of the sum of the squares of the coordinate differences between the actual position and the expected position of each droplet to obtain the droplet deviation amount. The droplet deviation direction is determined by vector direction calculation. Combining the droplet deviation amount and deviation direction information, droplet deviation feedback data is generated, such as recording the deviation amount and corresponding direction of each droplet in tabular form.

[0104] It should be noted that introducing a pre-trained closed-loop prediction model based on initial droplet state data, historical droplet motion data, and electrode calibration parameters into the digital microfluidic control process upgrades droplet motion control from the traditional method that relies on fixed rules, empirical parameters, or static path planning to an intelligent closed-loop control mode with active prediction, real-time comparison, and error quantification capabilities. In existing technologies, droplet motion trajectories typically rely on fixed driving voltage sequences or simple empirical correction methods, which are insufficient to cope with trajectory deviations caused by droplet volume fluctuations, differences in local electrode responses, and dielectric layer inhomogeneities.

[0105] S4. Based on the droplet deviation feedback data, the initial driving voltage parameters are dynamically adjusted to generate a closed-loop driving voltage sequence.

[0106] S4.1 Analyze the droplet deviation feedback data for deviation amount and deviation direction, and generate feedback analysis data.

[0107] Specifically, based on the droplet deviation feedback data, the deviation amount and deviation direction information of each droplet are extracted, the deviation amount is numerically normalized, and the deviation direction is angularly standardized. The processed deviation amount and deviation direction are integrated according to the droplet sequence to form feedback analysis data, which includes a complete dataset containing the deviation amount and deviation direction of each droplet, and arranged according to the droplet arrangement order.

[0108] S4.2 Based on the feedback analysis data, a voltage adjustment algorithm is used to calculate the correction value of the initial drive voltage parameters and generate a voltage adjustment scheme.

[0109] Specifically, a voltage adjustment algorithm is used, taking the deviation amount and direction of each droplet as input, to calculate the correction value of the initial driving voltage parameters, expressed as:

[0110] ;

[0111] in, Represents droplets Preliminary drive voltage parameter correction values ​​(in V). This represents the vertical voltage adjustment factor (unit: V / μm). Represents droplets The deviation (in μm). Represents droplets The direction of deviation (in rad). This represents the horizontal voltage adjustment factor (unit: V / μm). Indices representing droplet indices, Indicates the horizontal direction;

[0112] The initial driving voltage parameter correction values ​​of all droplets are then integrated according to the electrode arrangement order to generate a voltage adjustment scheme that includes the correction voltage of each droplet.

[0113] It should also be noted that the electrode arrangement order refers to the order formed by sequentially numbering the electrodes from the starting position to the ending position according to the fixed spatial arrangement of each electrode in the digital microfluidic chip. For example, it can be sequentially arranged from left to right according to the row direction of the digital microfluidic chip, or sequentially arranged from top to bottom according to the column direction. The electrode arrangement order is used to clarify the position order of each electrode on the digital microfluidic chip, so that subsequent processing steps can process and output the electrode driving voltage item by item based on a consistent spatial arrangement reference.

[0114] The vertical voltage adjustment coefficient is used to convert the vertical deviation of the droplet into a voltage correction value. It can be determined through experimental calibration or numerical simulation. For example, a known voltage pulse is applied, the vertical displacement of the droplet is collected, and the voltage is calculated according to the linear relationship between the voltage and the deviation. The horizontal voltage adjustment coefficient is used to convert the horizontal deviation of the droplet into a voltage correction value. It can be determined through experimental calibration or numerical simulation by fitting the relationship between the known voltage input and the horizontal displacement of the droplet.

[0115] S4.3. Based on the voltage adjustment scheme, dynamically modify the initial drive voltage parameters to generate a closed-loop drive voltage sequence.

[0116] Specifically, after obtaining the voltage adjustment scheme, the preliminary driving voltage parameter correction values ​​of the corresponding electrodes in the voltage adjustment scheme are read sequentially according to the electrode arrangement order. The original driving voltage in the preliminary driving voltage parameters is added to the corresponding preliminary driving voltage parameter correction value to obtain the updated electrode driving voltage value. The same numerical superposition is performed on each electrode in the arrangement order so that all the original driving voltages are replaced with the corresponding updated electrode driving voltage values. After all the electrodes have been updated, all the updated electrode driving voltage values ​​arranged in the electrode arrangement order are integrated to generate a closed-loop driving voltage sequence.

[0117] It should be noted that the data-driven voltage adaptive control method based on real-time deviation feedback makes the driving voltage no longer a preset fixed sequence, but a dynamic, continuous and precise correction based on the amount and direction of droplet deviation. In the existing technology, digital microfluidics usually relies on static driving voltage meters or adjustment methods based on simple rules. When the droplet experiences positional shift, volume change, or environmental interference, due to the lack of real-time feedback closed loop, the control cannot correct the specific deviation point by point, which makes the droplet prone to deviating from the trajectory or even operation failure.

[0118] S5. Apply the closed-loop driving voltage sequence to the electrode array, collect droplet state data in real time, generate droplet operation commands, and integrate them to form a global droplet state matrix.

[0119] S5.1 Load the closed-loop driving voltage sequence into the electrode array, analyze the voltage amplitude, pulse timing and the corresponding application order of each electrode, and generate electrode driving task instructions.

[0120] Specifically, the closed-loop driving voltage sequence is read item by item according to the electrode arrangement order. The voltage amplitude, pulse start time, pulse duration and corresponding electrode position of each item in the closed-loop driving voltage sequence are analyzed in sequence. The analyzed voltage amplitude, pulse start time and pulse duration are matched one by one with the corresponding electrodes in the electrode array according to the electrode arrangement order. Based on the analyzed corresponding electrode position, pulse start time and pulse duration, the electrode driving task instruction is generated, which includes the specific application order, application time and application amplitude.

[0121] S5.2. According to the electrode driving task instruction, apply a closed-loop driving voltage to the electrode array and collect the droplet position, volume and morphology in real time to generate a droplet state data stream.

[0122] Specifically, according to the electrode position, application time, and application amplitude recorded in the electrode driving task instruction, a closed-loop driving voltage is applied to each corresponding electrode in the electrode array. During the application of the closed-loop driving voltage, image data of the droplet's location is continuously acquired at a preset sampling frequency using high-speed imaging. Each frame of the acquired image is input into an image processing method, and target contour extraction, boundary localization, and region area estimation are performed on each frame. The position, volume, and morphological parameters of the droplet in each frame are calculated sequentially. The droplet position, volume, and morphological parameters obtained from each frame are integrated according to the image acquisition time sequence to form a data sequence containing continuous droplet state information, which is output as a droplet state data stream.

[0123] It should also be noted that the specific steps for setting the sampling frequency include: before applying the closed-loop driving voltage, determining the sampling frequency range based on the droplet velocity and the processing time of the image processing method. For example, when the droplet velocity is relatively fast, the sampling frequency range is set to 100 frames per second to 300 frames per second; selecting a target frequency for acquiring image data within the sampling frequency range and recording the target frequency as the preset sampling frequency; writing the preset sampling frequency into the image acquisition command to constrain the image acquisition interval of the high-speed imaging method during the droplet state acquisition process, so that the high-speed imaging method performs continuous image acquisition operations according to the preset sampling frequency. The sampling frequency should meet the requirement of acquiring at least N frames per droplet motion cycle (e.g., N≥10), and the specific value can be determined according to the droplet velocity and image processing time.

[0124] The image processing method is a process that uses image preprocessing, target contour recognition, and feature quantization to extract droplet position, volume, and morphology from droplet image frames acquired by high-speed imaging. The image preprocessing generates an image containing only the droplet region through grayscale conversion, binarization, and background removal. Target contour recognition extracts droplet contour pixels through edge detection. Feature quantization calculates numerical information about droplet position, volume, and morphology by performing area statistics, centroid coordinate calculation, and shape feature parameter calculation on the contour pixels.

[0125] S5.3 Analyze and integrate the droplet state data stream to generate droplet operation instructions, and summarize the state of each droplet to form a global droplet state matrix.

[0126] Specifically, a time series analysis method is used to perform time-series parsing on the droplet state data stream. The droplet position, volume, and morphology information of each frame are categorized according to timestamp and droplet index, generating a time series for each droplet. A moving average method and outlier removal method are used to smooth the trajectory and remove outliers from each droplet time series, resulting in a cleaned droplet state sequence. Based on the cleaned droplet state sequence, a threshold comparison method and a state machine determination method are used to determine the operation events to be executed, generating corresponding droplet operation instructions. These instructions include the operation type, target electrode, application time, and operation parameters. For example, when the droplet position change exceeds 80% of the center distance between adjacent electrodes and the volume change is less than 5%, it is determined as a movement event; when the volume change exceeds 10%, it is determined as a generation or merging event; and when the deviation exceeds a set threshold and the droplet is near the expected path, it is determined as a correction event. Simultaneously, the latest state of each droplet at the current time point is recorded as a droplet state entry. All droplet operation instructions are merged into an operation instruction set according to the execution sequence, and all droplet state entries are summarized according to droplet index and timestamp to form a global droplet state matrix.

[0127] It should also be noted that time series analysis is a method that analyzes data arranged in chronological order to identify the trends, periodicity, and regularity of data changes over time, thereby generating a continuous state sequence for each droplet.

[0128] The moving average method calculates the average value of consecutive data points in a time series by setting a window size, which smooths short-term fluctuations in the droplet state sequence and reduces noise interference.

[0129] Outlier removal methods identify and exclude droplet state data that deviate from the normal range based on statistical characteristics or set thresholds to ensure the accuracy of sequence analysis.

[0130] The threshold comparison method is to set upper and lower thresholds for various state parameters of the droplet and judge the parameters such as the current position, volume or shape of the droplet. If the parameters exceed or fall below the upper or lower thresholds, the droplet is determined to be in a specific state or a corresponding operation is triggered.

[0131] The state machine decision method defines the droplet's operational state as a finite set of states, and performs state transitions based on the droplet's current state and input conditions (such as position, volume, and shape) to determine the operation that the droplet should perform.

[0132] S6. Based on the global droplet state matrix, record the droplet operation status and generate a droplet operation log file.

[0133] S6.1. Based on the global droplet state matrix, number, classify, and time-mark the operation events of each droplet to generate droplet operation log entries.

[0134] Specifically, the state information of each droplet is read sequentially in the global droplet state matrix. Each operation event is uniquely numbered according to the operation event type such as the droplet's current position, volume, and shape, and classified according to the event category. At the same time, a time stamp is generated for each event according to the preset sampling frequency. The number, category, and time stamp are integrated to generate a droplet operation log entry. For example, the number is 001~n, the category example is generation, movement, and correction, and the time stamp example is a millisecond-level timestamp.

[0135] S6.2 Summarize, format, and save the droplet operation log entries to generate a droplet operation record file.

[0136] Specifically, the droplet operation log entries are read sequentially according to the droplet number. The number, category, and timestamp of each log entry are uniformly formatted, for example, the number is formatted as a fixed-length character, the category uses a uniform string identifier, and the timestamp is represented by a millisecond-level timestamp. All log entries are summarized according to the droplet number and time order to generate a droplet operation record set, and the droplet operation record set is saved to generate a droplet operation record file, such as a CSV or JSON format file, for subsequent querying and analysis.

[0137] This embodiment also provides a closed-loop control digital microfluidic system, including: an electrode initialization module for initializing the electrode array of the digital microfluidic chip, setting the electrode spacing, electrode size, and voltage upper limit, and acquiring no-load state signals to generate initial reference data; a droplet generation module for forming droplets at predetermined positions on the digital microfluidic chip, and acquiring the initial position, volume, and morphology information of the droplets based on the initial reference data to generate a droplet initial state dataset; a trajectory prediction module for inputting the droplet initial state dataset into a pre-trained closed-loop prediction model, performing dynamic analysis and trajectory prediction calculations, generating the expected motion trajectory of the droplets, and comparing it with the real-time state data stream of the droplets to generate droplet deviation feedback data; a voltage optimization module for dynamically adjusting the initial driving voltage parameters based on the droplet deviation feedback data to generate a closed-loop driving voltage sequence; a state control module for applying the closed-loop driving voltage sequence to the electrode array, acquiring droplet state data in real time, generating droplet operation commands, and integrating them to form a global droplet state matrix; and a record generation module for recording the droplet operation status based on the global droplet state matrix to generate a droplet operation record file.

[0138] In summary, this invention achieves real-time closed-loop control during droplet motion by analyzing the deviation amount and direction of droplet deviation feedback data and dynamically modifying the initial driving voltage parameters using a voltage adjustment algorithm to generate a closed-loop driving voltage sequence. This enables immediate adaptive correction when the droplet deviates from the expected trajectory, ensuring stable droplet movement along the planned path, improving droplet trajectory control accuracy, reducing motion errors caused by environmental disturbances and electrode response differences, and ensuring high precision, dynamic adjustability, and high repeatability in droplet operation.

[0139] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A closed-loop control digital microfluidic method, characterized in that: include, Initialize the digital microfluidic chip electrode array, set the electrode spacing, electrode size and voltage limit, and collect the no-load state signal to generate initial reference data; A droplet is formed at a predetermined location on the digital microfluidic chip, and the initial position, volume and morphology information of the droplet are collected based on the initial reference data to generate a droplet initial state dataset. The initial state dataset of the droplet is input into the pre-trained closed-loop prediction model to perform dynamic analysis and trajectory prediction calculation, generate the expected motion trajectory of the droplet, and compare it with the real-time state data stream of the droplet to generate droplet deviation feedback data. Based on the droplet deviation feedback data, the initial driving voltage parameters are dynamically adjusted to generate a closed-loop driving voltage sequence; A closed-loop driving voltage sequence is applied to the electrode array, droplet state data is collected in real time, droplet operation commands are generated, and integrated to form a global droplet state matrix. Based on the global droplet state matrix, droplet operation status is recorded, and a droplet operation log file is generated.

2. The closed-loop control digital microfluidic method as described in claim 1, characterized in that: The specific steps for generating the initial reference data are as follows: Read the digital microfluidic chip model and electrode array design parameters, calculate the electrode spacing, size and material properties, and generate electrode layout configuration data; The electrode layout configuration data is loaded into the controller, the initial voltage upper limit is set and a low amplitude test voltage is applied, and at the same time, the no-load state sensor signal is acquired to generate no-load reference data. Based on the no-load reference data, the local response and dielectric layer difference of each electrode are calculated to generate electrode calibration parameters, which are then integrated with the voltage distribution table to generate initial reference data.

3. The closed-loop control digital microfluidic method as described in claim 1, characterized in that: The specific steps for generating the initial state dataset of the droplets are as follows. Based on the initial reference data, the target position and initial voltage configuration for droplet generation on the digital microfluidic chip are calculated to obtain droplet generation control commands. According to the droplet generation control command, a voltage pulse is applied at a predetermined position on the digital microfluidic chip to form a droplet. At the same time, the position, volume and morphology sensor signals of the droplet are collected in real time during the droplet formation process to generate preliminary data on droplet formation. The preliminary data on droplet formation are analyzed to extract the actual position, volume, and morphology of the droplets, and then integrated to generate the initial state data of the droplets.

4. The closed-loop control digital microfluidic method as described in claim 1, characterized in that: The pre-trained closed-loop prediction model is constructed based on the droplet initial state dataset, historical droplet motion data, and electrode calibration parameters, through dynamic analysis and trajectory prediction algorithms, and is obtained through pre-training using historical droplet motion data.

5. The closed-loop control digital microfluidic method as described in claim 1, characterized in that: The specific steps for generating the expected trajectory of the droplets are as follows. Input the initial state dataset of droplets into the pre-trained closed-loop prediction model to establish a droplet dynamics prediction environment; In the droplet dynamics prediction environment, dynamic analysis is performed on the initial state dataset of droplets to generate preliminary driving voltage parameters; Trajectory prediction calculations are performed on the initial driving voltage parameters to generate the expected trajectory of the droplet.

6. The closed-loop control digital microfluidic method as described in claim 1, characterized in that: The specific steps for generating droplet deviation feedback data are as follows. Real-time acquisition of droplet current position, volume and morphology information, generating droplet real-time status data stream; The real-time status data stream of the droplet is compared with the expected trajectory of the droplet to extract the deviation between the actual position, volume and shape of the droplet and the expected trajectory, and then integrated to generate droplet trajectory deviation data. Based on the droplet trajectory deviation data, the amount and direction of the droplet deviation are calculated, and droplet deviation feedback data is generated.

7. The closed-loop control digital microfluidic method as described in claim 1, characterized in that: The specific steps for generating the closed-loop driving voltage sequence are as follows. Analyze the droplet deviation feedback data for deviation amount and direction to generate feedback analysis data; Based on the feedback analysis data, a voltage adjustment algorithm is used to calculate the correction value of the initial drive voltage parameters and generate a voltage adjustment scheme. Based on the voltage adjustment scheme, the initial driving voltage parameters are dynamically modified to generate a closed-loop driving voltage sequence.

8. The closed-loop control digital microfluidic method as described in claim 1, characterized in that: The specific steps for forming the global droplet state matrix are as follows: The closed-loop driving voltage sequence is loaded onto the electrode array, and the voltage amplitude, pulse timing and the corresponding application order of each electrode are analyzed to generate electrode driving task instructions. According to the electrode driving task instructions, a closed-loop driving voltage is applied to the electrode array, and the droplet position, volume and morphology are collected in real time to generate a droplet state data stream. The droplet state data stream is analyzed and integrated to generate droplet operation instructions, and the state of each droplet is summarized to form a global droplet state matrix.

9. The closed-loop control digital microfluidic method as described in claim 1, characterized in that: The specific steps for generating the droplet operation log file are as follows. Based on the global droplet state matrix, the operation events of each droplet are numbered, classified, and time-stamped to generate droplet operation log entries; The droplet operation log entries are summarized, formatted, and saved to generate a droplet operation record file.

10. A closed-loop control digital microfluidic system, based on the closed-loop control digital microfluidic method according to any one of claims 1 to 9, characterized in that: include, The electrode initialization module is used to initialize the electrode array of the digital microfluidic chip, set the electrode spacing, electrode size and voltage limit, and collect no-load state signals to generate initial reference data. The droplet generation module is used to form droplets at predetermined positions on the digital microfluidic chip, and to collect the initial position, volume and morphology information of the droplets based on the initial reference data to generate a droplet initial state dataset. The trajectory prediction module is used to input the initial state dataset of the droplet into the pre-trained closed-loop prediction model, perform dynamic analysis and trajectory prediction calculation, generate the expected motion trajectory of the droplet, and compare it with the real-time state data stream of the droplet to generate droplet deviation feedback data. The voltage optimization module is used to dynamically adjust the initial driving voltage parameters based on droplet deviation feedback data to generate a closed-loop driving voltage sequence. The state control module is used to apply the closed-loop drive voltage sequence to the electrode array, collect droplet state data in real time, generate droplet operation commands, and integrate them to form a global droplet state matrix. The record generation module is used to record droplet operation information based on the global droplet state matrix and generate droplet operation record files.

Images