AI dynamic optimization based ultralow temperature vacuum system and superconducting application control method

Abstract

The application discloses an AI dynamic optimization-based ultralow-temperature vacuum system and a superconducting application control method, relates to the field of ultralow-temperature vacuum, and comprises a vacuum cavity. The inner wall of the vacuum cavity is vertically and equidistantly provided with a plurality of second cold screens, and the inner wall of the vacuum cavity outside the plurality of second cold screens is provided with a plurality of first cold screens. Through a series of structures, heat conduction loss is effectively reduced, the problems of poor heat insulation performance and large temperature fluctuation of a traditional system are solved, an ultrahigh stability environment foundation is provided for superconducting experiments, energy consumption is reduced, and the cooling time is shortened. The application realizes multi-objective optimization of minimum energy consumption, fastest cooling speed and minimum temperature fluctuation, enhances the structural stability, sealing reliability and durability of the equipment under long-term high pressure and ultralow-temperature operation, reduces the maintenance demand, shortens the equipment downtime maintenance time, and reduces the operation cost. The application not only guarantees the temperature uniformity of the cold plate, but also improves the vacuum purity, thereby providing protection for superconducting experiments.

Description

Technical Field

[0001] This invention relates to the field of cryogenic vacuum, specifically to a cryogenic vacuum system and a control method for superconducting applications based on AI dynamic optimization. Background Technology

[0002] In the field of superconducting high technology, ultra-low temperature vacuum environment is a core prerequisite for carrying out precision experiments, special material preparation and high-end equipment operation.

[0003] Currently, traditional cryogenic vacuum systems suffer from insufficient temperature control precision, relying on manual experience or simple closed-loop control. This makes them ill-suited to handling complex situations such as internal heat load fluctuations and external environmental interference, resulting in poor cryogenic steady-state maintenance, large temperature fluctuations, and an inability to meet the stringent environmental stability requirements of superconducting experiments. Furthermore, these systems exhibit high energy consumption, with cooling power adjustments mismatched to actual load demands. The lack of scientific optimization strategies during cooling leads to energy waste, significant heat loss from the cold shield support structure, and limited insulation performance of the vacuum chamber. The absence of real-time monitoring of structural stress makes them prone to endplate deformation and seal failure under long-term high-pressure, cryogenic operation, impacting equipment lifespan. Moreover, existing systems often employ fixed, procedural control methods, lacking adaptive adjustment and predictive optimization capabilities. This prevents real-time optimization of control strategies based on experimental progress and equipment status changes, affecting the accuracy and reliability of superconducting experimental data. Summary of the Invention

[0004] The purpose of this invention is to provide an AI-based dynamically optimized cryogenic vacuum system and a control method for superconducting applications to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an AI-based dynamically optimized ultra-low temperature vacuum system, comprising a vacuum chamber, wherein a plurality of second cold screens are vertically and equidistantly arranged on the inner wall of the vacuum chamber, and a plurality of first cold screens are arranged on the inner wall of the vacuum chamber outside the plurality of second cold screens. A cryogenic pump group is connected to the bottom of the vacuum chamber via a flange, a pumping pump group is connected to the bottom of the vacuum chamber at the front end of the cryogenic pump group via a flange, and a compressor circuit is connected to the bottom of the vacuum chamber at the rear end of the cryogenic pump group via a flange. A smart valve group is connected between the pumping pump group and the vacuum chamber, and a variable power drive module is connected between the compressor circuit and the vacuum chamber. A distributed temperature sensor array is arranged on the inner wall of the second cold screens, an activated carbon adsorption plate is installed at the bottom of the vacuum chamber, and a cold plate is installed above the activated carbon adsorption plate.

[0006] Preferably, the first and second cold screens are connected to a low thermal conductivity material support frame on one side. This fixes the position of the cold screens while minimizing heat conduction through the support structure, ensuring the insulation effect of the cold screens. The first cold screen is connected to the first cold head of the cryogenic pump unit via a flexible heat conduction cable to achieve efficient transfer of cooling capacity and ensure that the first cold screen reaches the designed cooling temperature. The second cold screen and the cold plate are respectively connected to the second cold head of the cryogenic pump unit via flexible heat conduction cables, enabling the second cold screen and the cold plate to receive precise cooling capacity to meet the ultra-low temperature requirements of the core area. The first and second cold screens on one side of the low thermal conductivity material support frame are connected to a shape memory alloy. By utilizing the temperature deformation characteristics of the shape memory alloy, the spacing and fit of the cold screens can be dynamically adjusted to optimize the heat conduction path and achieve adaptive adjustment of thermal management.

[0007] Preferably, the vacuum chamber includes an outer shell, which provides external structural support and protection for the chamber, resisting the influence of the external environment on the chamber. The outer shell has an inner wall shell, which forms a sealed space inside the chamber, ensuring the airtightness of the vacuum environment. A thermal insulation material is provided between the inner wall shell and the outer shell to block heat transfer between the outer shell and the inner wall shell, thereby enhancing the overall thermal insulation performance of the chamber. Distributed strain and temperature sensors are arranged inside the thermal insulation material to monitor the temperature change and structural stress state of the thermal insulation layer in real time, and to provide timely warnings of abnormal temperature conduction and structural deformation risks. The outer surface of the outer shell is provided with a support structure to provide stable installation support for the entire vacuum chamber, ensuring the structural stability of the equipment during operation.

[0008] Preferably, the activated carbon adsorption plate has a plurality of legs evenly distributed on its circumferential surface. These legs are installed inside the vacuum chamber to ensure the activated carbon adsorption plate is stably installed at the bottom of the chamber, while avoiding heat conduction and contamination caused by direct contact between the adsorption plate and the bottom of the chamber. The cold plate has a plurality of lifting brackets evenly distributed on its circumferential surface. The bottom ends of these lifting brackets are installed on the top of the activated carbon adsorption plate to support the cold plate above it, ensuring gas flow below the cold plate and achieving precise positioning and installation of the cold plate.

[0009] Preferably, the interior of the cold plate is provided with a spiral guide plate to guide liquid nitrogen to form a spiral flow path inside the cold plate, prolonging the residence time of liquid nitrogen and improving the temperature uniformity inside the cold plate. A liquid nitrogen inlet is provided at the middle position of the bottom end of the cold plate as a channel for liquid nitrogen to enter the cold plate, ensuring stable input of liquid nitrogen. A liquid nitrogen outlet is provided at one edge of the bottom end of the cold plate for discharging the liquid nitrogen after heat exchange, realizing the recycling of liquid nitrogen.

[0010] Preferably, both the liquid nitrogen inlet and the liquid nitrogen outlet are equipped with intelligent flow regulation units. The intelligent flow regulation unit contains a flow controller and a temperature sensor to monitor the liquid nitrogen flow and temperature data in real time. It can accurately adjust the flow according to AI instructions to ensure the temperature control accuracy of the cold plate. The liquid nitrogen inlet is connected to the second cold head of the cryogenic pump unit to obtain the cooling output of the cryogenic pump unit and further reduce the initial temperature of the liquid nitrogen. The liquid nitrogen outlet is connected to the compressor circuit to realize the recovery and circulation of liquid nitrogen after heat exchange and reduce operating costs.

[0011] Preferably, an AI optimization controller is provided on one side of the vacuum cavity. As the core control unit of the system, it receives data collected by various sensors, runs AI algorithms to perform analysis and decision-making, and sends control commands to various actuators to realize dynamic optimization control of the system.

[0012] A method for controlling superconducting applications in a cryogenic vacuum system includes the following steps: S1: System self-check and parameter initialization. The AI ​​optimization controller sends self-check commands to each subsystem; checks the communication status, motor insulation, and valve position feedback of the pump group, cryogenic pump group, and intelligent valve group; checks the operating status of the compressor circuit and variable power drive module to confirm that there are no alarm signals; reads the initial values ​​of the distributed temperature sensor array and distributed strain and temperature sensors to verify that the sensors are working properly; the controller loads the target process formula from the database, such as the target temperature of 4K. At the same time, it synchronizes the collected real-time physical parameters, such as cavity size and material properties, to the system's digital twin model to provide a high-fidelity virtual environment for subsequent AI prediction and optimization.

[0013] S2: Coordinated vacuuming and system precooling. The AI ​​controller sequentially opens the corresponding valves in the intelligent valve group, starts the pump group, and evacuates the vacuum chamber from atmospheric pressure to a medium vacuum range, such as 10. -1 -10 -2 The vacuum level data is fed back in real time by a vacuum gauge installed on the cavity. When the vacuum level reaches the starting condition of the cryogenic pump group, the controller starts the cryogenic pump group. At the same time, the compressor circuit is started, and the cooling power is controlled by the variable power drive module to start pre-cooling the first and second cold screens. The AI ​​model predicts the temperature field in real time based on the data of the distributed temperature sensor array, and dynamically adjusts the power of the refrigerator through the fuzzy PID algorithm to avoid thermal stress caused by sudden temperature drop and achieve stable and rapid pre-cooling.

[0014] S3: AI-optimized cooling and ultra-low temperature steady-state maintenance. When the system is pre-cooled to the base temperature, such as 80K, the AI ​​controller instructs the second cold head of the cryogenic pump unit to work at full capacity to cool the second cold screen and cold plate. Liquid nitrogen flows in from the liquid nitrogen inlet, flows out from the outlet after passing through the flow channel formed by the spiral guide plate. The intelligent flow regulation unit precisely controls the flow rate and temperature according to the AI ​​instructions to ensure that the temperature of the cold plate drops uniformly. The AI ​​controller runs reinforcement learning, such as dual deep Q network or model predictive control algorithm, to achieve multi-objective optimization such as minimum energy consumption, fastest cooling speed, and minimum temperature fluctuation. For example, by controlling the deformation of the shape memory alloy, the thermal conductivity between the cold screens is finely adjusted to achieve dynamic thermal management.

[0015] S4: Superconducting Experiment Operation and Real-time Optimization. After the system reaches and stabilizes at the target ultra-low temperature, the specific process parameters of the superconducting experiment are loaded. During the experiment, the extended state observer function of the AI ​​controller will estimate and compensate for unknown disturbances such as internal heat load fluctuations and external environmental changes in real time. The digital twin model will perform forward simulation to predict the changes in the system state in the future and make control adjustments in advance to achieve feedforward optimization.

[0016] S5: Safe shutdown, regeneration and data archiving. After the experiment, the AI ​​controller executes the safe shutdown procedure. First, it stops the experimental activities on the sample stage, and then controls the cooling system to slowly heat up to avoid damage to the equipment due to excessive heating. All data from the entire operation will be automatically archived. The AI ​​model will learn offline based on the data from this operation, update its model parameters, and achieve continuous self-evolution of performance.

[0017] Compared with the prior art, the beneficial effects of the present invention are: This AI-based dynamically optimized cryogenic vacuum system and superconducting application control method constructs a gradient insulation and dynamic thermal management system by combining a first and second cold screen structure with a low thermal conductivity material support frame and shape memory alloy, along with real-time sensing by a distributed temperature sensor array. This effectively reduces heat conduction loss and solves the problems of poor insulation performance and large temperature fluctuations in traditional systems, providing a highly stable environmental foundation for superconducting experiments.

[0018] This AI-based dynamic optimization-based ultra-low temperature vacuum system and superconducting application control method integrates fuzzy PID algorithm, reinforcement learning algorithm and digital twin model through AI-optimized controller to achieve dynamic adaptive adjustment of parameters such as cooling power, liquid nitrogen flow rate and vacuum degree. In the process of rapid cooling and ultra-low temperature maintenance, it accurately matches the load demand, reduces energy consumption and shortens the cooling time, and achieves multi-objective optimization of minimum energy consumption, fastest cooling speed and minimum temperature fluctuation.

[0019] This AI-based dynamically optimized cryogenic vacuum system and superconducting application control method utilizes a double-layer shell design for the vacuum chamber, integrated distributed strain and temperature sensors within the insulation material, and a comprehensive system self-checking and fault warning mechanism to monitor the structural stress and operating status of the equipment in real time. This proactively mitigates risks such as end plate deformation and seal failure. Furthermore, by combining the offline learning and self-evolution capabilities of the AI ​​model, the control strategy is continuously optimized, significantly enhancing the structural stability, sealing reliability, and durability of the equipment under long-term high-pressure and cryogenic operation. This reduces maintenance requirements, shortens equipment downtime for maintenance, and lowers operating costs.

[0020] This AI-based dynamically optimized cryogenic vacuum system and superconducting application control method extends the liquid nitrogen heat exchange path and residence time through the internal spiral guide plate of the cold plate and the intelligent flow regulation unit of the liquid nitrogen inlet and outlet. At the same time, it enhances the adsorption and purification effect of trace gases inside the vacuum chamber, ensuring both the temperature uniformity of the cold plate and the vacuum purity, thus providing a guarantee for superconducting experiments. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the overall structure of the present invention; Figure 2 This is a partial cross-sectional schematic diagram of the vacuum cavity structure of the present invention; Figure 3 This is a schematic diagram of the thermal insulation material structure of the present invention; Figure 4 This is a partial cross-sectional schematic diagram of the first and second cold screen structures of the present invention; Figure 5 This is a schematic diagram of the internal structure of the cold plate tray of the present invention; Figure 6 This is a bottom-view perspective view of the cold plate tray structure of the present invention.

[0022] In the diagram: 1. Vacuum chamber; 101. Outer shell; 102. Inner wall shell; 103. Thermal insulation material; 104. Distributed strain and temperature sensors; 2. Support structure; 3. Cryogenic pump unit; 4. Pumping unit; 5. Intelligent valve unit; 6. Compressor circuit; 7. Variable power drive module; 8. AI optimization controller; 9. Distributed temperature sensor array; 10. First cold shield; 11. Second cold shield; 12. Low thermal conductivity material support frame; 13. Shape memory alloy; 14. Activated carbon adsorption plate; 15. Legs; 16. Lifting frame; 17. Spiral guide plate; 18. Cold plate tray; 19. Liquid nitrogen inlet; 20. Liquid nitrogen outlet. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] In the description of this invention, it should be noted that the terms "upper," "lower," "inner," "outer," "front end," "rear end," "both ends," "one end," and "the other end," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0025] like Figures 1 to 6 As shown, this embodiment utilizes an AI-dynamically optimized cryogenic vacuum system, including a vacuum chamber 1. The inner wall of the vacuum chamber 1 is vertically and equidistantly equipped with several second cold shields 11. Outside the second cold shields 11, the inner wall of the vacuum chamber 1 is equipped with several first cold shields 10. This double-layer cold shield structure forms a gradient thermal insulation barrier. The outer first cold shields 10 block ambient heat from being conducted to the core area of ​​the chamber, while the inner second cold shields 11 further enhance the cryogenic maintenance effect, synergistically improving the chamber's insulation performance. A cryogenic pump unit 3 is connected to the bottom of the vacuum chamber 1 via a flange, providing the core cooling power for the system and enabling the construction of a cryogenic environment. A pressure pump unit 4 is connected to the bottom of the vacuum chamber 1 at the front end of the cryogenic pump unit 3 via a flange, used to quickly extract gas from the chamber, creating the required vacuum preconditions for the start-up of the cryogenic pump unit 3. A pressure pump unit 4 is connected to the bottom of the vacuum chamber 1 at the rear end of the cryogenic pump unit 3 via a flange. The compressor circuit 6 realizes the circulation and recovery of refrigerant and pressure regulation, ensuring the continuous and stable operation of the refrigeration system. The pump group 4 and the vacuum chamber 1 are connected by an intelligent valve group 5, which precisely controls the opening and closing of the vacuum passage to match the vacuum adjustment requirements at different stages. The compressor circuit 6 and the vacuum chamber 1 are connected by a variable power drive module 7, which dynamically adjusts the compressor output power to adapt to real-time changes in the refrigeration load. The inner wall of the second cold screen 11 is equipped with a distributed temperature sensor array 9, which collects temperature data of the core area of ​​the chamber in all directions with high precision, providing a basic perception basis for AI-optimized control. An activated carbon adsorption plate 14 is installed at the bottom of the vacuum chamber 1 to adsorb trace gas molecules remaining in the chamber, further improving the vacuum purity. A cold plate 18 is installed above the activated carbon adsorption plate 14 to enhance local low temperature conduction and temperature uniformity control.

[0026] Specifically, the first cold screen 10 and the second cold screen 11 are connected to a low thermal conductivity material support frame 12 on one side. While fixing the position of the cold screen, the heat conduction through the support structure 2 is minimized, ensuring the heat insulation effect of the cold screen. The first cold screen 10 is connected to the first cold head of the cryogenic pump group 3 through a flexible heat conduction tape to achieve efficient transfer of cold energy and ensure that the first cold screen 10 reaches the design cooling temperature. The second cold screen 11 and the cold plate 18 are respectively connected to the second cold head of the cryogenic pump group 3 through flexible heat conduction tape, so that the second cold screen 11 and the cold plate 18 can obtain precise cold energy supply to meet the ultra-low temperature requirements of the core area. The first cold screen 10 and the second cold screen 11 are connected to a shape memory alloy 13 on one side of the low thermal conductivity material support frame 12. By utilizing the temperature deformation characteristics of the shape memory alloy 13, the spacing and bonding state of the cold screens are dynamically adjusted, the heat conduction path is optimized, and the adaptive adjustment of thermal management is achieved.

[0027] Furthermore, the vacuum chamber 1 includes an outer shell 101, which provides external structural support and protection for the chamber, resisting the influence of the external environment on the chamber. An inner wall shell 102 is provided inside the outer shell 101, forming a sealed space inside the chamber to ensure the airtightness of the vacuum environment. A thermal insulation material 103 is provided between the inner wall shell 102 and the outer shell 101 to block heat transfer between the outer shell 101 and the inner wall shell 102, thereby enhancing the overall thermal insulation performance of the chamber. Distributed strain and temperature sensors 104 are arranged inside the thermal insulation material 103 to monitor the temperature change and structural stress state of the thermal insulation layer in real time, and to provide timely warnings of abnormal temperature conduction and structural deformation risks. A support structure 2 is provided on the outer surface of the outer shell 101 to provide stable installation support for the entire vacuum chamber 1, ensuring the structural stability of the equipment during operation.

[0028] Furthermore, several feet 15 are evenly distributed on the circumferential surface of the activated carbon adsorption plate 14. The feet 15 are installed inside the vacuum chamber 1 to ensure the stable installation of the activated carbon adsorption plate 14 at the bottom of the chamber, while avoiding heat conduction and contamination caused by direct contact between the adsorption plate and the bottom of the chamber. Several lifting brackets 16 are evenly distributed on the circumferential surface of the cold plate 18. The bottom end of the lifting bracket 16 is installed on the top of the activated carbon adsorption plate 14. The lifting bracket 16 supports the cold plate 18 above the activated carbon adsorption plate 14, ensuring gas flow below the cold plate 18 and achieving precise positioning and installation of the cold plate 18.

[0029] Furthermore, a spiral guide plate 17 is provided inside the cold plate 18 to guide liquid nitrogen to form a spiral flow path inside the cold plate 18, prolonging the residence time of liquid nitrogen and improving the temperature uniformity inside the cold plate 18. A liquid nitrogen inlet 19 is provided at the middle position of the bottom end of the cold plate 18 as a channel for liquid nitrogen to enter the cold plate 18, ensuring stable input of liquid nitrogen. A liquid nitrogen outlet 20 is provided at one edge of the bottom end of the cold plate 18 to discharge the liquid nitrogen after heat exchange, realizing the recycling of liquid nitrogen.

[0030] Furthermore, both the liquid nitrogen inlet 19 and the liquid nitrogen outlet 20 are equipped with intelligent flow regulation units. The intelligent flow regulation units contain flow controllers and temperature sensors, which monitor the liquid nitrogen flow and temperature data in real time and adjust the flow precisely according to AI instructions to ensure the temperature control accuracy of the cold plate 18. The liquid nitrogen inlet 19 is connected to the second cold head of the cryogenic pump group 3 to obtain the cooling output of the cryogenic pump group 3 and further reduce the initial temperature of the liquid nitrogen. The liquid nitrogen outlet 20 is connected to the compressor circuit 6 to realize the recovery and circulation of liquid nitrogen after heat exchange and reduce operating costs.

[0031] Furthermore, an AI optimization controller is installed on one side of the vacuum chamber 1. As the core control unit of the system, it receives data collected by various sensors, runs AI algorithms to perform analysis and decision-making, and sends control commands to various actuators to realize dynamic optimization control of the system.

[0032] A method for controlling superconducting applications in a cryogenic vacuum system includes the following steps: S1: System self-check and parameter initialization. The AI ​​optimization controller 8 sends self-check commands to each subsystem; checks the communication status, motor insulation, and valve position feedback of the pump group 4, cryogenic pump group 3, and intelligent valve group 5; checks the operating status of the compressor circuit 6 and variable power drive module 7 to confirm that there are no alarm signals; reads the initial values ​​of the distributed temperature sensor array 9 and the distributed strain and temperature sensor 104 to verify that the sensors are working properly; the controller loads the target process formula from the database, such as the target temperature of 4K. At the same time, it synchronizes the collected real-time physical parameters, such as cavity size and material properties, to the system's digital twin model to provide a high-fidelity virtual environment for subsequent AI prediction and optimization.

[0033] S2: Coordinated vacuuming and system precooling. The AI ​​controller sequentially opens the corresponding valves in the intelligent valve group 5, starts the pump group 4, and pumps the vacuum chamber 1 from atmospheric pressure to a medium vacuum range, such as 10. -1 -10 -2The vacuum level data is fed back in real time by a vacuum gauge installed on the cavity. When the vacuum level reaches the starting condition of the cryogenic pump group 3, the controller starts the cryogenic pump group 3. At the same time, the compressor circuit 6 is started, and the cooling power is controlled by the variable power drive module 7 to start pre-cooling the first cold screen 10 and the second cold screen 11. The AI ​​model predicts the temperature field in real time based on the data of the distributed temperature sensor array 9, and dynamically adjusts the power of the refrigerator through the fuzzy PID algorithm to avoid the thermal stress caused by the sudden drop in temperature and achieve stable and rapid pre-cooling.

[0034] S3: AI-optimized cooling and ultra-low temperature steady-state maintenance. When the system is pre-cooled to the base temperature, such as 80K, the AI ​​controller instructs the second cold head of the cryogenic pump group 3 to work at full capacity to cool the second cold screen 11 and the cold plate 18. Liquid nitrogen flows in from the liquid nitrogen inlet 19, flows out from the outlet after passing through the flow channel formed by the spiral guide plate 17. The intelligent flow regulation unit precisely controls the flow rate and temperature according to the AI ​​instructions to ensure that the temperature of the cold plate 18 drops uniformly. The AI ​​controller runs reinforcement learning, such as dual deep Q network or model predictive control algorithm, to achieve multi-objective optimization such as minimum energy consumption, fastest cooling speed, and minimum temperature fluctuation. For example, by controlling the deformation of the shape memory alloy 13, the thermal conductivity between the cold screens is finely adjusted to achieve dynamic thermal management.

[0035] S4: Superconducting Experiment Operation and Real-time Optimization. After the system reaches and stabilizes at the target ultra-low temperature, the specific process parameters of the superconducting experiment are loaded. During the experiment, the extended state observer function of the AI ​​controller will estimate and compensate for unknown disturbances such as internal heat load fluctuations and external environmental changes in real time. The digital twin model will perform forward simulation to predict the changes in the system state in the future and make control adjustments in advance to achieve feedforward optimization.

[0036] S5: Safe shutdown, regeneration and data archiving. After the experiment, the AI ​​controller executes the safe shutdown procedure. First, it stops the experimental activities on the sample stage, and then controls the cooling system to slowly heat up to avoid damage to the equipment due to excessive heating. All data from the entire operation will be automatically archived. The AI ​​model will learn offline based on the data from this operation, update its model parameters, and achieve continuous self-evolution of performance.

[0037] The usage method of this embodiment is as follows: First, complete the system installation and preliminary preparations, ensuring that the thermal insulation material 103 between the outer shell 101 and the inner wall shell 102 of the vacuum chamber 1 is properly laid, and that the distributed strain and temperature sensor 104 is properly wired. Figure 3 As shown; check the connection between the first cold shield 10, the second cold shield 11, the low thermal conductivity material support frame 12, and the shape memory alloy 13 to ensure the flexible heat transfer cable is reliably connected to the cold head of the cryogenic pump unit 3. Figure 4As shown; the activated carbon adsorption plate 14 is fixed to the bottom of the vacuum chamber 1 by the bracket 15, and the cold plate 18 is installed above the activated carbon adsorption plate 14 by the lifting bracket 16, ensuring that the spiral guide plate 17 is not deformed. The liquid nitrogen inlet 19 and liquid nitrogen outlet 20 are connected and sealed to the corresponding pipelines as shown. Figure 5 , Figure 6 As shown; finally, connect the communication lines between the AI ​​optimization controller 8 and each subsystem, connect the power supply, liquid nitrogen storage device and compressor circuit 6, and start the system. After starting the system, the AI ​​optimization controller 8 automatically executes step S1, system self-check and parameter initialization: send self-check commands to pump group 4, cryogenic pump group 3, intelligent valve group 5, etc., to verify communication, motor insulation and valve feedback status, and at the same time read the initial values ​​of distributed temperature sensor array 9 and distributed strain and temperature sensor 104. After confirming that there are no abnormalities, load the target process formula such as 4K ultra-low temperature target, and synchronize parameters such as cavity size and material properties to the digital twin model. After the self-check is passed, proceed to step S2, coordinated vacuuming and system precooling: the AI ​​controller opens the corresponding valves of intelligent valve group 5 in sequence, and starts pump group 4 to pump the vacuum chamber 1 from atmospheric pressure to 10kJ / kg. -1 -10 -2The vacuum level data is fed back to the controller in real time within the medium vacuum range of Pa. When the vacuum level meets the start-up conditions of the cryogenic pump group 3, the cryogenic pump group 3 and compressor circuit 6 are started. The cooling power is adjusted by the variable power drive module 7 to pre-cool the first cold screen 10 and the second cold screen 11. The AI ​​model predicts the temperature field based on the data of the distributed temperature sensor array 9 and dynamically adjusts the cooling power through the fuzzy PID algorithm to avoid the generation of thermal stress. After the system is pre-cooled to the base temperature of 80K, it enters the S3 step of AI optimization cooling and ultra-low temperature steady state maintenance: the AI ​​controller instructs the second cold head of the cryogenic pump group 3 to work at full capacity. Liquid nitrogen flows into the cold plate 18 through the liquid nitrogen inlet 19, exchanges heat along the flow channel of the spiral guide plate 17, and is discharged from the liquid nitrogen outlet 20. The intelligent flow regulation unit accurately controls the liquid nitrogen flow and temperature according to the AI ​​instructions. At the same time, the AI ​​controller runs the reinforcement learning algorithm to fine-tune the heat conduction state of the cold screen by controlling the deformation of the shape memory alloy 13 to achieve the lowest energy consumption and the fastest cooling. The system undergoes multi-objective optimization to minimize speed and temperature fluctuations until it stabilizes at the target ultra-low temperature. Then, step S4, superconducting experimental operation and real-time optimization, loads specific process parameters for the superconducting experiment. During the experiment, the AI ​​controller compensates for internal heat load fluctuations and external environmental changes in real time through an expanded state observer. A digital twin model performs forward-looking simulations to adjust control strategies in advance, ensuring a stable experimental environment. After the experiment, step S5, safe shutdown, regeneration, and data archiving, is executed: the AI ​​controller first stops the sample stage experiment and then controls the cooling system to slowly raise the temperature to avoid equipment damage. Data such as temperature, vacuum, and energy consumption during operation are automatically archived. The AI ​​model learns offline based on this data, updating parameters to achieve continuous performance evolution. Throughout the entire process, the system status can be monitored in real time through the AI ​​optimization controller. If an alarm signal occurs, the system will automatically locate the fault and provide a solution, ensuring operational safety and reliability.

[0038] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An AI dynamic optimization based ultra-low temperature vacuum system comprising a vacuum chamber (1), characterized in that: The inner wall of the vacuum chamber (1) is vertically and equidistantly provided with several second cold screens (11), and the inner wall of the vacuum chamber (1) outside the several second cold screens (11) is provided with several first cold screens (10). The bottom of the vacuum chamber (1) is connected to a cryogenic pump group (3) through a flange. The bottom of the vacuum chamber (1) at the front end of the cryogenic pump group (3) is connected to a pressure pump group (4) through a flange. The bottom of the vacuum chamber (1) at the rear end of the cryogenic pump group (3) is connected to a compressor circuit (6) through a flange. The pressure pump group (4) and the vacuum chamber (1) are connected together by an intelligent valve group (5). The compressor circuit (6) and the vacuum chamber (1) are connected together by a variable power drive module (7). The inner wall of the second cold screen (11) is provided with a distributed temperature sensor array (9). The bottom of the vacuum chamber (1) is equipped with an activated carbon adsorption plate (14), and a cold plate tray (18) is installed above the activated carbon adsorption plate (14).

2. The AI dynamic optimization based ultralow temperature vacuum system according to claim 1, wherein: The first cold screen (10) and the second cold screen (11) are connected to a low thermal conductivity material support frame (12) on one side. The first cold screen (10) is connected to the first cold head of the cryogenic pump group (3) through a flexible heat conduction cable. The second cold screen (11) and the cold plate (18) are connected to the second cold head of the cryogenic pump group (3) through flexible heat conduction cables. The first cold screen (10) and the second cold screen (11) on one side of the low thermal conductivity material support frame (12) are connected to a shape memory alloy (13).

3. The AI dynamic optimization based ultralow temperature vacuum system according to claim 1, wherein: The vacuum cavity (1) contains an outer shell (101), and an inner wall shell (102) is provided inside the outer shell (101). A thermal insulation material (103) is provided between the inner wall shell (102) and the outer shell (101). Distributed strain and temperature sensors (104) are arranged inside the thermal insulation material (103). A support structure (2) is provided on the outer surface of the outer shell (101).

4. The AI dynamic optimization based ultralow temperature vacuum system according to claim 1, wherein: The activated carbon adsorption plate (14) has several legs (15) evenly spaced on its circumferential surface. The legs (15) are installed inside the vacuum chamber (1). The cold plate (18) has several lifting frames (16) evenly spaced on its circumferential surface. The bottom of the lifting frames (16) is installed on the top of the activated carbon adsorption plate (14). 5.The AI dynamic optimization based ultralow temperature vacuum system according to claim 1, wherein: The interior of the cold plate (18) is provided with a spiral guide plate (17), a liquid nitrogen inlet (19) is provided at the middle position of the bottom end of the cold plate (18), and a liquid nitrogen outlet (20) is provided at one edge of the bottom end of the cold plate (18). 6.The AI dynamic optimization based ultralow temperature vacuum system of claim 5, wherein: Both the liquid nitrogen inlet (19) and the liquid nitrogen outlet (20) are equipped with intelligent flow regulation units. The intelligent flow regulation units contain flow controllers and temperature sensors. The liquid nitrogen inlet (19) is connected to the second cold head of the cryogenic pump group (3), and the liquid nitrogen outlet (20) is connected to the compressor circuit (6). 7.The AI dynamic optimization based ultralow temperature vacuum system according to claim 1, wherein: An AI optimization controller (8) is provided on one side of the vacuum cavity (1).

8. A superconducting application control method for an ultra-low temperature vacuum system based on AI dynamic optimization, characterized in that: Includes the following steps: S1: System self-check and parameter initialization, AI optimization controller (8) sends self-check instructions to each subsystem; checks the communication status, motor insulation and valve position feedback of the pump group (4), cryogenic pump group (3) and intelligent valve group (5); checks the operating status of compressor circuit (6) and variable power drive module (7) to confirm that there is no alarm signal; reads the initial values ​​of distributed temperature sensor array (9) and distributed strain and temperature sensor (104) to verify that the sensors are working normally; the controller loads the target process formula from the database, the target temperature is 4K, and at the same time, synchronizes the collected real-time physical parameters, cavity size and material properties to the digital twin model of the system to provide a high-fidelity virtual environment for subsequent AI prediction optimization; S2: Coordinated vacuuming and system precooling, the AI ​​controller sequentially opens the corresponding valves in the intelligent valve group (5), starts the pump group (4), and pumps the vacuum chamber (1) from atmospheric pressure to a medium vacuum range, such as 10. -1 -10 -2 Pa, the vacuum level data is fed back in real time by the vacuum gauge installed on the cavity. When the vacuum level reaches the starting condition of the cryogenic pump group (3), the controller starts the cryogenic pump group (3), and at the same time, starts the compressor circuit (6) and controls the cooling power through the variable power drive module (7) to start pre-cooling the first cold screen (10) and the second cold screen (11). The AI ​​model predicts the temperature field in real time based on the data of the distributed temperature sensor array (9) and dynamically adjusts the power of the refrigerator through the fuzzy PID algorithm to avoid the thermal stress caused by the sudden drop in temperature and achieve stable and fast pre-cooling. S3: AI optimizes cooling and maintains ultra-low temperature steady state. When the system is pre-cooled to the base temperature, such as 80K, the AI ​​controller instructs the second cold head of the cryogenic pump group (3) to work at full capacity to cool the second cold screen (11) and the cold plate (18). Liquid nitrogen flows in from the liquid nitrogen inlet (19), flows out from the outlet (20) after passing through the flow channel formed by the spiral guide plate (17). The intelligent flow regulation unit accurately controls the flow and temperature according to the AI ​​instruction to ensure that the temperature of the cold plate (18) drops uniformly. The AI ​​controller runs reinforcement learning, dual deep Q network or model predictive control algorithm to achieve multi-objective optimization such as the lowest energy consumption, the fastest cooling speed, and the smallest temperature fluctuation. By controlling the deformation of the shape memory alloy (13), the thermal conductivity between the cold screens is finely adjusted to achieve dynamic thermal management. S4: Superconducting Experiment Operation and Real-time Optimization. After the system reaches and stabilizes at the target ultra-low temperature, the specific process parameters of the superconducting experiment are loaded. During the experiment, the extended state observer function of the AI ​​controller will estimate and compensate for unknown disturbances such as internal heat load fluctuations and external environmental changes in real time. The digital twin model will perform forward simulation to predict the changes in the system state in the future and make control adjustments in advance to achieve feedforward optimization. S5: Safe shutdown, regeneration and data archiving. After the experiment, the AI ​​controller executes the safe shutdown procedure. First, it stops the experimental activities on the sample stage, and then controls the cooling system to slowly heat up to avoid damage to the equipment due to excessive heating. All data from the entire operation will be automatically archived. The AI ​​model will learn offline based on the data from this operation, update its model parameters, and achieve continuous self-evolution of performance.

Images