Intelligent microfluidic experiment system and method with digital twinning and AI auxiliary functions

Abstract

The invention provides an intelligent micro-fluidic experiment system and method with digital twinning and AI auxiliary functions. The intelligent micro-fluidic experiment system comprises a micro-fluidic chip, and the micro-fluidic chip is connected with a micro-sampling module and a collecting module which are arranged at the two ends of the micro-fluidic chip respectively; a control module is further arranged on one side of the micro-fluidic chip, and the control module is connected with a first constant-flux pump, a second constant-flux pump and a gas mass flow controller which are arranged on the micro-sampling module; a detection module is arranged between the control module and the micro-fluidic chip, and the control module is connected with the detection module. According to the intelligent microfluidic experiment system and method with the digital twinning and AI auxiliary functions, automation, intelligentization and high efficiency of a microfluidic experiment are achieved, particularly, due to application of the digital twinning and AI technology, the precision, depth and breadth of the experiment are remarkably improved, and the experiment efficiency is improved. And a powerful tool is provided for the application of the microfluidic technology in the fields of scientific research and teaching.

Description

Technical Field

[0001] The present invention relates to the field of micro-chemical engineering, and particularly to an intelligent microfluidic experimental system and method with digital twin and AI-assisted functions. Background Art

[0002] A microfluidic chip is a micro-scale chip with channels having an equivalent diameter of 10 to 1000 microns, which can achieve a rapid synthesis process in the field of organic synthesis and helps reduce the environmental impact of chemicals. It has a higher specific surface area, uniform flow distribution, and efficient heat and mass transfer capabilities, showing advantages unparalleled by traditional equipment in the chemical reaction process. However, existing traditional microfluidic experimental systems are usually composed of multiple independent devices, and different device parameters need to be set separately, resulting in a cumbersome operation process and low efficiency. In addition, traditional experimental systems lack data analysis and intelligent assistance functions, making it difficult to meet the growing scientific research needs and restricting the popularization and application of microfluidic experiments in chemical engineering professional teaching.

[0003] Digital twin technology optimizes system performance, predicts faults, and assists in decision-making by constructing a virtual model of a physical entity and real-time mapping its state and behavior; AI technology can analyze multi-modal data such as sensor data and image data in real time to achieve accurate prediction of parameters that are difficult to directly measure, and large language model technology can realize the design and intelligent evaluation of experimental schemes. However, existing microfluidic experimental systems have not fully integrated these advanced technologies, resulting in limited intelligence level and difficulty in meeting the dual needs of scientific research and teaching.

[0004] In this context, it is particularly urgent and necessary to introduce core technologies such as digital twin and AI technology into microfluidic experimental devices. This can not only meet the needs of automated experimental operations, real-time data collection and analysis in scientific research in the field of micro-chemical engineering, but also respond to the urgent need of the chemical industry for the cultivation of intelligent talents. Through the guidance of experimental scheme design and evaluation by digital twin technology and the large language model of artificial intelligence, students' understanding of chemical engineering technology in the industrial 4.0 era can be improved, their cognitive ability of intelligent chemical engineering can be enhanced, and the experimental scheme design ability and operation level of chemical engineering major students can be significantly improved. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent microfluidic experimental system and method with digital twin and AI-assisted functions, which realizes the automation, intelligence, and high efficiency of microfluidic experiments. In particular, the application of digital twin and AI technology significantly improves the accuracy, depth, and breadth of experiments, providing a powerful tool for the application of microfluidic technology in scientific research and teaching fields.

[0006] To achieve the above object, the present invention provides an intelligent microfluidic experimental system with digital twin and AI assistance functions, which includes a microfluidic chip. The microfluidic chip is respectively connected to a micro-sampling module and a collection module arranged at both ends thereof;

[0007] On one side of the microfluidic chip, a control module is further arranged. The control module is connected to a first peristaltic pump, a second peristaltic pump, and a gas mass flow controller arranged on the micro-sampling module;

[0008] A detection module is arranged between the control module and the microfluidic chip, and the control module is connected to the detection module.

[0009] Preferably, the detection module includes an electronic magnifier, a first temperature sensor, a second temperature sensor, a third temperature sensor, a fourth temperature sensor, a first pressure sensor, and a second pressure sensor. The electronic magnifier is located above the microfluidic chip. The first temperature sensor, the second temperature sensor, the third temperature sensor, and the fourth temperature sensor are respectively located at two inlets, one outlet, and the mixing area of the microchannel of the microfluidic chip.

[0010] Preferably, a first liquid storage bottle, a second liquid storage bottle, and a gas cylinder are arranged on the micro-sampling module. The first liquid storage bottle, the second liquid storage bottle, and the gas cylinder are connected in parallel. The first liquid storage bottle is connected to the first peristaltic pump, the second liquid storage bottle is connected to the second peristaltic pump, and the gas cylinder is connected to the gas mass flow controller.

[0011] Preferably, a liquid collection bottle and an exhaust gas treatment bottle are arranged on the collection module. The liquid collection bottle and the exhaust gas treatment bottle are connected in sequence.

[0012] Preferably, the first pressure sensor is located between the microfluidic chip and the micro-sampling module; the second pressure sensor is located between the microfluidic chip and the liquid collection bottle.

[0013] Preferably, the other ends of the first peristaltic pump, the second peristaltic pump, and the gas mass flow controller are connected to the microfluidic chip.

[0014] Preferably, the internal channels of the microfluidic chip are of T-shaped structure, cross-shaped structure, and heart-shaped structure. The microfluidic chip is provided with standard threaded interfaces, and an LED illumination source is arranged on the back of the microfluidic chip.

[0015] Preferably, the control module is provided with an all-in-one computer, which is embedded with a data acquisition unit, a data processing unit, an enhancement unit, and a display unit. The enhancement unit is connected to the display unit, and the data processing unit is connected to the display unit;

[0016] The data acquisition unit is used to acquire experimental images and experimental data;

[0017] The data processing unit is used to process the experimental images and experimental data obtained by the data acquisition unit, and perform multimodal AI and theoretical analysis on the experimental images and experimental data;

[0018] The enhancement unit is used to respond to the user's scanning operation on the microfluidic chip identifier through a mobile device;

[0019] The display component displays the results processed and analyzed by the data processing unit, and displays the real-time experimental status information and the status of the digital twin system.

[0020] Preferably, a voice interaction component, an experimental scheme generation component, and an evaluation component are provided on the data processing unit. The voice interaction function is used to recognize the user's voice input, the experimental scheme generation function is used to generate an experimental scheme according to the user's voice or text input, and the evaluation component is used to perform intelligent evaluation on the user's experimental operations.

[0021] An intelligent microfluidic experiment method with digital twin and AI assistance functions includes the following steps:

[0022] Select the mode required for this experiment on the all-in-one computer, and input or adjust specific experimental parameters according to the selected experimental mode;

[0023] Micro-injection module preparation: Add liquid samples with different properties to the first liquid storage bottle and the second liquid storage bottle respectively according to the experimental requirements;

[0024] Start the experiment in the control module, and the micro-injection module starts to accurately deliver the liquid sample to the microfluidic chip at the set flow rate;

[0025] At the same time, the gas mass flow controller introduces gas into the microfluidic chip as needed;

[0026] During the experiment, the detection module continuously collects the experimental data inside the microfluidic chip;

[0027] The electronic magnifying glass takes an experimental image, and transmits it to the data processing unit together with the experimental data through the data acquisition unit;

[0028] After receiving the experimental data and experimental images, the data processing unit first performs preprocessing, and processes and analyzes the preprocessed data;

[0029] According to the current experimental type, perform real-time and accurate prediction on the key experimental parameters;

[0030] According to the analysis result of the data processing unit, the user issues an adjustment instruction to the system through the voice interaction component or text;

[0031] The system adjusts the experimental parameters in real time according to the user's instructions, and optimizes the experimental conditions to achieve the best experimental results;

[0032] Click the stop button in the control software to end the experiment.

[0033] Therefore, the present invention adopts the above intelligent microfluidic experimental system and method with digital twin and AI assistance functions, and the technical effects are as follows:

[0034] (1) The present invention realizes the high integration of the microfluidic experimental system, integrates the microfluidic chip, sample injection, collection, detection, communication, computer and data processing modules into one, and works together through standardized interfaces and communication protocols. Users can complete all operations through an all-in-one computer, getting rid of the disadvantages of the traditional system with scattered equipment and cumbersome operations, and greatly improving the experimental efficiency and convenience.

[0035] (2) The present invention introduces a multi-modal AI analysis model, which integrates and processes sensor data and image data, and can accurately predict key parameters that are difficult to directly measure by traditional methods, such as flow patterns, bubble surface area to volume ratio, mixing efficiency and mass transfer coefficient, etc. Combined with the digital twin system, users can intuitively compare the prediction results with the actual phenomena, significantly improving the depth of data analysis and the controllability of the experiment.

[0036] (3) The present invention integrates voice interaction and experimental scheme generation functions based on large language models. Users can operate the system through voice commands, and use the fine-tuned large language model to intelligently generate experimental schemes according to their own needs. The model can also evaluate the user's experimental operations and give scores. This innovation significantly reduces the threshold of experimental design, improves the experimental efficiency, and provides personalized guidance for users. Description of the Drawings

[0037] Figure 1 It is a schematic diagram of the intelligent microfluidic experimental system of the present invention;

[0038] Figure 2 It is a flowchart of the intelligent microfluidic experimental system of the present invention;

[0039] Figure 3 It is a schematic diagram of the control module of the intelligent microfluidic experimental system of the present invention;

[0040] 1. Microfluidic chip; 2. Micro-sampling module; 201. First flat pump; 202. Second flat pump; 203. Gas mass flow controller; 204. First liquid storage bottle; 205. Second liquid storage bottle; 206. Gas cylinder; 3. Collection module; 301. Liquid collection bottle; 302. Tail gas treatment bottle; 4. Control module; 401. Data acquisition unit; 402. Data processing unit; 4021. Voice interaction component; 4022. Experimental scheme generation component; 4023. Evaluation component; 403. Enhancement unit; 404. Display unit; 5. Detection module; 501. Electronic magnifier; 502. First temperature sensor; 503. Second temperature sensor; 504. Third temperature sensor; 505. Fourth temperature sensor; 506. First pressure sensor; 507. Second pressure sensor. Detailed implementation mode

[0041] The technical solution of the present invention will be further described below through the accompanying drawings and embodiments.

[0042] Unless otherwise defined, the technical terms or scientific terms used in the present invention shall have the ordinary meanings understood by those with ordinary skills in the field to which the present invention belongs.

[0043] Embodiment 1

[0044] An intelligent microfluidic experiment system with digital twin and AI assistance functions, as Figure 1 , 2 shown. The solid arrows indicate the conveying direction of materials in the system, and the dashed arrows indicate the transmission direction of data between modules. The experimental system includes a microfluidic chip 1, and the microfluidic chip 1 is respectively connected to a micro-sampling module 2 and a collection module 3 provided at both ends thereof. A control module 4 is also provided on one side of the microfluidic chip 1, and a detection module 5 is provided between the control module 4 and the microfluidic chip 1, and the control module 4 is connected to the detection module 5.

[0045] The detection module 5 includes an electronic magnifier 501, a first temperature sensor 502, a second temperature sensor 503, a third temperature sensor 504, a fourth temperature sensor 505, a first pressure sensor 506, and a second pressure sensor 507; the electronic magnifier 501 is located above the microfluidic chip 1; the first temperature sensor 502, the second temperature sensor 503, the third temperature sensor 504, and the fourth temperature sensor 505 are respectively located at two inlets, one outlet of the microfluidic chip, and the mixing area of the microchannel, and the first pressure sensor 506 is located between the microfluidic chip 1 and the micro-sampling module 2; the second pressure sensor 507 is located between the microfluidic chip 1 and the liquid collection bottle 301.

[0046] The micro-injection module 2 is provided with a first liquid storage bottle 204, a second liquid storage bottle 205 and a gas cylinder 206. The first liquid storage bottle 204, the second liquid storage bottle 205 and the gas cylinder 206 are connected in parallel. The first liquid storage bottle 204 is connected to the first peristaltic pump 201, the second liquid storage bottle 205 is connected to the second peristaltic pump 202, and the gas cylinder 206 is connected to the gas mass flow controller 203. The control module 4 is connected to the first peristaltic pump 201, the second peristaltic pump 202 and the gas mass flow controller 203 provided on the micro-injection module 2. The other ends of the first peristaltic pump 201, the second peristaltic pump 202 and the gas mass flow controller 203 are connected to the microfluidic control chip.

[0047] The collection module 3 is provided with a liquid collection bottle 301 and an exhaust gas treatment bottle 302, and the liquid collection bottle 301 and the exhaust gas treatment bottle 302 are connected in sequence.

[0048] The control module 4 is provided with an all-in-one computer, and the all-in-one computer is embedded with a data acquisition unit and a data processing unit 402.

[0049] The first pressure sensor 506 and the second pressure sensor 507 are respectively located at both ends of the microfluidic chip 1. The first temperature sensor 502, the second temperature sensor 503, the third temperature sensor 504 and the fourth temperature sensor 505 are respectively located at the two inlets, one outlet and the mixing area of the microchannel of the microfluidic chip. The first pressure sensor 506 is located between the microfluidic chip 1 and the micro-injection module 2. The second pressure sensor 507 is located between the microfluidic chip 1 and the liquid collection bottle 301.

[0050] The internal channels of the microfluidic chip 1 are of T-shaped structure, cross-shaped structure and heart-shaped structure. The microfluidic chip 1 is provided with a standard threaded interface, and an LED illumination source is arranged on the back of the microfluidic chip 1 to assist the detection module 5 in collecting experimental images.

[0051] The control module 4 is provided with an all-in-one computer, and the all-in-one computer is embedded with a data acquisition unit 401, a data processing unit 402, an enhancement unit 403 and a display unit 404. The enhancement unit 403 is connected to the display unit 404, and the data processing unit 402 is connected to the display unit 404;

[0052] The data acquisition unit 401 is used to acquire experimental images and experimental data;

[0053] The data processing unit 402 is used to process the experimental images and experimental data obtained by the data acquisition unit 401, and analyze the experimental images and experimental data;

[0054] The enhancement unit 403 is used to respond to the scanning operation of the user on the microfluidic chip 1 identifier through a mobile device;

[0055] The display component displays the results processed and analyzed by the data processing unit 402, showing the real-time status information of the experiment.

[0056] A voice interaction component 4021, an experiment scheme generation component 4022, and an evaluation component 4023 are provided on the data processing unit 402. The voice interaction function is used to recognize the user's voice input, the experiment scheme generation function is used to generate an experiment scheme according to the user's voice or text input, and the evaluation component 4023 is used to intelligently evaluate the user's experimental operations.

[0057] The microfluidic chip 1 is manufactured using high-precision 3D printing technology and is made of a translucent resin material with good biocompatibility and optical transparency. The outer dimensions of the microfluidic chip 1 are: 150 mm in length, 100 mm in width, and 5 mm in thickness; microchannels for fluid mixing and mass transfer are machined inside the chip, and the cross-sectional dimensions of the microchannels are: 1000 μm in width and 280 μm in height; the microchannels adopt a specific geometric structure design in the mixing area, including but not limited to at least one of a T-shaped, cross-shaped, or heart-shaped structure, to enhance the fluid mixing efficiency; in order to facilitate connection with the external pipeline system and ensure good sealing performance, 2 inlets and 1 outlet, a total of 3 fluid inlets and outlets, are provided on the microfluidic chip 1. These inlets and outlets all use 1 / 4-28UNF standard threaded interfaces and can be reliably connected to standard pipe fittings.

[0058] The configuration of the micro-syringe injection module 2 is as follows: The first liquid storage bottle 204 and the second liquid storage bottle 205 are two independent storage bottles. The volumes of the first liquid storage bottle 204 and the second liquid storage bottle 205 are both 500 mL, which are respectively used to store two liquid samples with different properties or colors; Two high-precision first peristaltic pumps 201 and second peristaltic pumps 202 are configured, corresponding to the above-mentioned first liquid storage bottle 204 and second liquid storage bottle 205 respectively, for independently controlling the pumping of the two liquids. The flow rate adjustment ranges of the first peristaltic pump 201 and the second peristaltic pump 202 are from 0.1 mL / min to 100 mL / min, the maximum working pressure is 25 MPa, the flow rate repeatability accuracy is ±0.5% F.S, the flow rate accuracy is ±1.0% F.S, and they are equipped with RS232 communication interfaces and support the Modbus communication protocol to facilitate data communication and remote control with the all-in-one computer; A gas cylinder 206 with a volume of 0.4 L is configured to store the gas required during the experiment, such as air or carbon dioxide, etc.; A high-precision gas mass flow controller 203 is adopted, with a measurement range of 2 sccm to 500 sccm, an accuracy of ±1.0% F.S, a repeatability accuracy of ±0.2% F.S, and a linearity of ±0.5% F.S. The maximum pressure resistance value of this gas mass flow controller 203 is 10 MPa, and the response time is less than 1 second, which can quickly and stably adjust the gas flow rate. At the same time, this controller is equipped with an RS232 communication interface and supports the Modbus communication protocol, which is convenient for data exchange and transmission of control instructions with the all-in-one computer 8.

[0059] The configuration of the collection module 3 is as follows: A solvent storage bottle with a volume of 500 mL is used as the liquid collection bottle 301 to collect the reacted liquid or waste liquid discharged from the outlet of the microfluidic chip 1; A tail gas treatment bottle 302 with a volume of 500 mL is adopted, and the inside of the tail gas treatment bottle 302 is filled with 400 mL of tail gas absorption liquid. The specific composition of the tail gas absorption liquid is selected according to the types of tail gases that may be generated during the experiment to ensure that the harmful gases that may be emitted at the outlet of the microfluidic chip 1 can be effectively absorbed and neutralized.

[0060] Each component device in the micro-syringe injection module 2, the microfluidic chip 1, and the collection module 3 is equipped with a 1 / 4-28UNF standard thread interface. These standard thread interfaces are connected through corrosion-resistant and high-temperature-resistant polytetrafluoroethylene pipelines. The specific connection method is: Connect each device in sequence according to the process flow with polytetrafluoroethylene pipelines to form a closed fluid path.

[0061] The configuration of the detection module 5 is as follows: The electronic magnifier 501 adopts a high-resolution imaging device, and its core component is a 1 / 1.9-inch color CMOS image sensor. This sensor has the characteristics of high sensitivity and low noise. The pixel size of the sensor is 3.75μm × 3.75μm, and it supports the autofocus function, capable of clearly capturing the flow state and microscopic details of the fluid inside the microfluidic chip 1; The magnifier bracket is made of high-strength aluminum alloy material, with good stability and adjustment flexibility. It can support the electronic magnifier 501 and allow it to move freely up to 5 cm and be precisely positioned in all directions; Four high-precision digital temperature sensors are set on the back of the microfluidic chip 1, including the first temperature sensor 502, the second temperature sensor 503, the third temperature sensor 504, and the fourth temperature sensor 505, which are respectively located at the two inlets, one outlet, and the mixing area of the microchannel of the microfluidic chip 1, used to monitor the temperature changes at key positions in real time. The temperature data is output as 12-bit digital quantity, with a corresponding temperature resolution of 0.0625°C, a measurement range of -55°C to +125°C, and a measurement accuracy of ±0.5°C; A high-precision first pressure sensor 506 and a second pressure sensor 507 are respectively set at the inlet and outlet pipelines of the microfluidic chip 1, used to monitor the pressure changes inside the microfluidic chip 1 in real time. The measurement ranges of the first pressure sensor 506 and the second pressure sensor 507 are -0.1MPa to 2MPa, and they output 24-bit digital quantity through the RS485 interface. The measurement accuracy is ±0.1%F.S, the long-term stability is 0.1%F.S / year, and the temperature drift coefficient is ±0.01%F.S / °C.

[0062] The data transmission and control process is as follows: The detection module 5 and the micro-injection module 2 establish a data connection with the data acquisition unit 401 in the all-in-one computer through the RS232 or RS485 serial communication interface. The temperature, pressure data collected by the detection module 5 and the image data captured by the electronic magnifier 501, together with the flow rate data in the micro-injection module 2, are all transmitted to the data acquisition unit 401 in the form of digital signals in real time; There is a specially developed control software running on the all-in-one computer. This software provides a user-friendly experimental interface, namely the display unit 404. The experimental interface of the display unit 404 displays the received temperature, pressure, flow rate, and image data in real time, providing an intuitive experimental status monitoring for users. Users can set and adjust the experimental parameters through this experimental interface; The all-in-one computer generates corresponding control instructions according to the user's settings and sends these instructions to the first peristaltic pump 201, the second peristaltic pump 202, and the gas mass flow controller 203 in the micro-injection module 2 through the data acquisition unit 401 in real time. Thus, remote precise control of the fluid flow rate in the microfluidic chip 1 is achieved, such as starting and stopping, adjusting the flow rate size, etc.

[0063] The all-in-one computer integrates multiple experimental modes and advanced digital twin technologies, as follows: Four typical microfluidic experiments and their detailed experimental manuals are pre-configured in the all-in-one computer, namely the gas-liquid two-phase flow pattern transition experiment in microchannels, the gas-liquid two-phase mass transfer process experiment, the liquid-liquid mixing experiment, and the bubble rupture experiment, providing users with rich experimental options and reducing the difficulty of experimental operations. For the above four preset experiments and combined with the experimental conditions specified in the experimental manuals, the present invention constructs a digital twin system of the intelligent microfluidic experimental system based on the computational fluid dynamics simulation technology of the finite difference method. This digital twin system can real-time simulate the fluid flow state inside the microfluidic chip 1 and predict the experimental results. The digital twin system is also provided with a display unit 404 with an augmented reality display function. Users can scan a specific identifier pre-set on the all-in-one computer or the microfluidic chip 1 through a mobile device (such as a smartphone or a tablet computer). After the scanning operation is triggered, the digital twin system can respond to this operation and superimpose and display the real-time simulation information of the fluid state inside the microfluidic chip 1 on the screen of the mobile device in the form of augmented reality. This intuitive display method greatly enhances users' understanding and control of the experimental process.

[0064] The AI multimodal analysis function is integrated in the data processing unit 402 of the present invention, and its specific implementation is as follows: The data processing unit 402 receives various experimental data transmitted from the data acquisition unit 401, including numerical data collected by the first temperature sensor 502, the second temperature sensor 503, the third temperature sensor 504, the fourth temperature sensor 505, the first pressure sensor 506, and the second pressure sensor 507, flow rate data provided by the micro-injection module 2, and experimental process image data captured by the electronic magnifier 501. The time interval for data acquisition is set to 500 ms to ensure the real-time nature of the data; for the possible problem of missing numerical data during the acquisition process, linear interpolation is used for data filling to ensure the integrity of the data sequence. All numerical data are normalized to scale them to a unified numerical range to eliminate the influence brought by different dimensions and numerical range differences, and improve the efficiency and accuracy of subsequent model processing; for the possible problem of missing image data, an image generation technology based on a generative adversarial network is used for intelligent filling. This method learns the internal rules and features from the existing image data and generates images that match the missing image data to ensure the coherence of the image sequence; the preprocessed multi-source heterogeneous data will be input into a pre-trained multimodal deep learning model based on the Transformer architecture. Different modalities of data such as numerical and image data are fused and processed, and according to the current type of experiment being carried out (e.g., microchannel gas-liquid two-phase flow pattern transition, gas-liquid two-phase mass transfer process, liquid-liquid mixing, or bubble rupture experiment), key experimental parameters are predicted in real time and accurately. These key parameters include, but are not limited to, physical quantities such as flow pattern, bubble surface area to volume ratio, liquid mixing efficiency, and gas-liquid mass transfer coefficient that are difficult to directly measure by sensors.

[0065] The data processing unit 402 of the present invention further integrates a voice interaction component 4021 and an experimental scheme generation component 4022. The specific implementation is as follows: The voice interaction component 4021 is implemented by fine-tuning based on existing open-source large language models. For example, models such as ChatGLM and DeepSeek can be used. For the professional terms and common expressions in the field of chemical experiments, the selected large language model is specifically fine-tuned using the existing chemical experiment scheme database to enable it to better understand voice commands related to chemical experiments; The voice interaction component 4021 can accurately recognize the voice commands input by the user through integrated voice recognition and convert the voice signal into text information. The converted text information will be displayed in real time on the display screen of the all-in-one computer for the user to confirm to ensure the accurate transmission of the command; The experimental scheme generation component 4022 also relies on the above-mentioned fine-tuned open-source large language model. When the user puts forward an experimental requirement through the voice interaction component 4021, the large language model first converts the text information obtained by voice recognition into a high-dimensional semantic vector. Then, using the cosine similarity calculation method between vectors, it retrieves in the pre-constructed experimental scheme design library to find 1-2 candidate experimental schemes with the highest similarity to the user's demand vector; Subsequently, the large language model will conduct an in-depth comparative analysis of the user's demand and the retrieved candidate experimental schemes to identify the similarities and differences between the two. Based on its reasoning ability, the model will combine the specific details of the user's demand and the advantages and disadvantages of the candidate schemes to conduct intelligent scheme adjustment and optimization, and finally generate a customized experimental scheme for the current user's demand and present it to the user for reference. The scheme can include detailed information such as experimental steps, required reagents, instrument equipment, parameter settings, etc.

[0066] The data processing unit 402 of the present invention is also provided with an evaluation component 4023 to comprehensively and objectively evaluate the user's experimental operations. The specific implementation is as follows: For various experiments preset in the all-in-one computer 8, the evaluation function combines the advantages of the AI multi-modal analysis model and the fine-tuned large language model to achieve intelligent evaluation of the experimental operation process and results. For the results of experimental operations, such as visually observable indicators like the color uniformity after liquid-liquid mixing and the flow pattern under specific conditions, the AI multi-modal analysis model in the data processing unit 402 is mainly used for evaluation. This model can process and analyze the image data collected by the electronic magnifier 501, the numerical data collected by the first temperature sensor 502, the second temperature sensor 503, the third temperature sensor 504, the fourth temperature sensor 505, the first pressure sensor 506, the second pressure sensor 507, and the flow rate data provided by the micro-injection module 2, and compare it with the standard results of the preset experiment, so as to quantify the difference between the current experimental operation result and the expected result and give a completion score. For the evaluation of the standardization and safety of experimental operation steps, a comprehensive judgment is made by combining the preset experimental rule library and the fine-tuned large language model. The large language model can understand the text description of the experimental operation steps and compare it with the preset experimental rules to determine whether the operation steps meet the experimental specifications and safety requirements. Finally, the data processing unit 402 will integrate the evaluation results of the AI multi-modal analysis model and the LLM to generate a comprehensive experimental operation evaluation report. This report can comprehensively reflect the user's level in aspects such as experimental design ability, operation skills, and safety awareness, and provide valuable feedback and improvement suggestions for the user.

[0067] An intelligent microfluidic experiment method with digital twin and AI assistance functions includes the following steps:

[0068] On the control software interface of the all-in-one computer, select the mode required for this experiment from the four preset typical microfluidic experiments (microchannel gas-liquid two-phase flow pattern transition experiment, gas-liquid two-phase mass transfer process experiment, liquid-liquid mixing experiment, bubble rupture experiment).

[0069] According to the selected experimental mode, input or adjust specific experimental parameters, such as fluid properties (density, viscosity, etc.), flow rate, temperature, pressure, etc.

[0070] Set the time interval for data acquisition (such as 500 ms) to ensure the real-time and accuracy of the data.

[0071] Micro-injection module preparation: According to the experimental requirements, add liquid samples with different properties to the first liquid storage bottle and the second liquid storage bottle respectively, and ensure that the volume of both storage bottles is 500 mL.

[0072] Configure a high-precision peristaltic pump with a flow rate adjustment range of 0.1 mL / min to 100 mL / min, a flow rate repeatability error of ±0.1% F.S, and a flow rate accuracy of ±1% F.S.

[0073] Set the range of the gas mass flow controller to sccm to 500 sccm, with an accuracy of ±1.0% F.S, a repeatability accuracy of ±0.2% F.S, a response time of less than 1 second, and a pressure resistance value of 10 MPa.

[0074] Start the experiment in the control software, and the micro-syringe module begins to accurately deliver liquid samples to the microfluidic chip according to the set flow rate.

[0075] Meanwhile, the gas mass flow controller introduces gas into the microfluidic chip as needed.

[0076] During the experiment, the detection module real-time collects data such as the temperature, pressure, and fluid flow state inside the microfluidic chip.

[0077] The electronic magnifying glass takes images of the experimental process and transmits them to the data processing unit through the data acquisition unit together with data such as temperature and pressure.

[0078] After receiving the data, the data processing unit first performs preprocessing, including filling missing numerical data using linear interpolation method, normalization processing to eliminate dimension differences, etc.

[0079] For the problem of missing image data, intelligent filling is performed using image generation technology based on generative adversarial networks.

[0080] The preprocessed multi-source heterogeneous data is input into a pre-trained multi-modal deep learning model based on the Transformer architecture for processing and analysis.

[0081] According to the current experimental type, real-time and accurate predictions are made for key experimental parameters (such as flow pattern, bubble surface area to volume ratio, liquid mixing efficiency, and gas-liquid mass transfer coefficient, etc.).

[0082] According to the analysis results of the data processing unit, the user can issue adjustment instructions to the system through the voice interaction component, such as changing the liquid flow rate, adjusting the temperature or pressure, etc.

[0083] The system adjusts the experimental parameters in real time according to the user's instructions and optimizes the experimental conditions to achieve the best experimental results.

[0084] Click the stop button in the control software to end the experiment. The micro-syringe module and the gas mass flow controller stop working.

[0085] Save all the data collected in this experiment (including temperature, pressure, flow rate, images, etc.) in the specified folder of the all-in-one computer.

[0086] As needed, the data can be exported to CSV, Excel or other format files for subsequent analysis and report writing.

[0087] Turn off the power switches of all devices and disconnect the power cords and data cables.

[0088] Clean the dirt and residual liquid on the surfaces of devices such as the microfluidic chip, micro-injection module, and collection module.

[0089] Regularly maintain the equipment, check whether the performance and accuracy of the equipment meet the requirements, and promptly replace worn or aging parts.

[0090] Therefore, the present invention adopts the above intelligent microfluidic experimental system and method with digital twin and AI assistance functions, realizing real-time monitoring and precise control of the internal flow state, temperature and pressure of the microfluidic chip, and supporting data transmission and remote control through RS232 / RS485 interfaces. The system integrates preset modes of various typical microfluidic experiments, and innovatively constructs a digital twin system based on finite difference method computational fluid dynamics simulation. Combining with augmented reality technology, it significantly improves the user's perception and interaction experience of the internal state of the chip. In addition, the data processing module integrates AI multi-modal analysis, voice interaction, intelligent experimental scheme generation and experimental operation evaluation functions, greatly improving the intelligent level of the experimental system and promoting the wide application of microfluidic technology in fields such as chemical engineering. The intelligent microfluidic experimental system proposed by the present invention, with its automation, intelligence and high efficiency, provides a powerful tool for microfluidic experimental research and has broad application prospects.

[0091] Finally, 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 them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. Intelligent microfluidic experimental system with digital twin and AI-assisted functions, characterized by: It includes a microfluidic chip, and the microfluidic chip is respectively connected to a micro-injection module and a collection module arranged at two ends thereof; A control module is also provided on one side of the microfluidic chip, and the control module is connected to a first advection pump, a second advection pump, and a gas mass flow controller provided on the micro-injection module; A detection module is arranged between the control module and the microfluidic chip, and the control module is connected to the detection module.

2. The intelligent microfluidic experimental system with digital twin and AI-assisted functions according to claim 1, characterized in that: The detection module includes an electronic magnifying glass, a first temperature sensor, a second temperature sensor, a third temperature sensor, a fourth temperature sensor, a first pressure sensor, and a second pressure sensor; the electronic magnifying glass is located in front of the microfluidic chip; the first temperature sensor, the second temperature sensor, the third temperature sensor, and the fourth temperature sensor are respectively located at two inlets, one outlet, and a mixing area of ​​a microchannel of the microfluidic chip.

3. The intelligent microfluidic experimental system with digital twin and AI-assisted functions according to claim 1, characterized in that: The micro-injection module is provided with a first liquid storage bottle, a second liquid storage bottle and a gas cylinder, the first liquid storage bottle, the second liquid storage bottle and the gas cylinder are connected in parallel, the first liquid storage bottle is connected to a first horizontal flow pump, the second liquid storage bottle is connected to a second horizontal flow pump, and the gas cylinder is connected to a gas mass flow controller.

4. The intelligent microfluidic experimental system with digital twin and AI-assisted functions according to claim 1, characterized in that: The collection module is provided with a liquid collection bottle and an exhaust gas treatment bottle, and the liquid collection bottle and the exhaust gas treatment bottle are connected in sequence.

5. The intelligent microfluidic experimental system with digital twin and AI-assisted functions according to claim 2, characterized in that: The first pressure sensor is located between the microfluidic chip and the micro-injection module; the second pressure sensor is located between the microfluidic chip and the liquid collection bottle.

6. The intelligent microfluidic experimental system with digital twin and AI-assisted functions according to claim 1, characterized in that: The other ends of the first horizontal flow pump, the second horizontal flow pump and the gas mass flow controller are connected to the microfluidic control chip.

7. The intelligent microfluidic experimental system with digital twin and AI-assisted functions according to claim 1, characterized in that: The internal channel of the microfluidic chip is a T-shaped structure, a cross-shaped structure and a heart-shaped structure. A standard threaded interface is arranged on the microfluidic chip. An LED lighting source is arranged on the back of the microfluidic chip.

8. The intelligent microfluidic experimental system with digital twin and AI-assisted functions according to claim 1, characterized in that: The control module is provided with an all-in-one computer, wherein the all-in-one computer is embedded with a data acquisition unit, a data processing unit, an enhancement unit, and a display unit, wherein the enhancement unit is connected to the display unit, and the data processing unit is connected to the display unit; The data acquisition unit is used to acquire experimental images and experimental data; The data processing unit is used to process the experimental images and experimental data obtained by the data acquisition unit, and perform multimodal AI and theoretical analysis on the experimental images and experimental data; The enhancement unit is used to respond to a user's scanning operation of the microfluidic chip identifier through a mobile device; The display component displays the results of the analysis processed by the data processing unit, and displays the real-time status information of the experiment and the status of the digital twin system.

9. The intelligent microfluidic experimental system with digital twin and AI-assisted functions according to claim 8, characterized in that: The data processing unit is provided with a voice interaction component, an experimental plan generation component, and an evaluation component. The voice interaction function is used to recognize the user's voice input, the experimental plan generation function is used to generate an experimental plan according to the user's voice or text input, and the evaluation component is used to perform intelligent evaluation on the experimental operations.

10. Intelligent microfluidic experimental method with digital twin and AI-assisted functions, characterized in that: The following steps are involved: Select the mode required for this experiment on the computer all-in-one machine, and input or adjust specific experimental parameters according to the selected experimental mode; Preparation of micro-injection module: according to the experimental requirements, add liquid samples of different properties into the first liquid storage bottle and the second liquid storage bottle respectively; The experiment is started in the control module, and the microinjection module begins to accurately deliver liquid samples to the microfluidic chip at the set flow rate; At the same time, a gas mass flow controller introduces gas into the microfluidic chip as needed; During the experiment, the detection module collects experimental data inside the microfluidic chip in real time; The electronic magnifying glass takes the experimental image and transmits it to the data processing unit through the data acquisition unit together with the experimental data; After receiving the experimental data and experimental images, the data processing unit first performs preprocessing, and then processes and analyzes the preprocessed data; According to the current experiment type, the key experimental parameters are predicted in real time and accurately; Based on the analysis results of the data processing unit, the user issues adjustment instructions to the system through the voice interaction component or text; The system adjusts the experimental parameters in real time according to the user's instructions and optimizes the experimental conditions to achieve the best experimental results; Click the stop button in the control software to end the experiment.

Images