Vacuum heat insulation liquid nitrogen tank storage device and intelligent control system thereof

Abstract

The invention relates to the technical field of equipment intelligent control, in particular to a vacuum heat insulation liquid nitrogen container storage device and an intelligent control system thereof.The vacuum heat insulation liquid nitrogen container storage device comprises a physical device and an intelligent control device, and the physical device achieves heat insulation through a high vacuum interlayer between an outer container and an inner container and is provided with a liquid nitrogen cold source interlayer, a liquid nitrogen tower drum and other multiple cold source structures; the intelligent control device monitors environmental parameters in real time through the data acquisition module, constructs a virtual model through the digital twinborn simulation module, analyzes data and predicts faults through the intelligent diagnosis and prediction module, generates a control strategy through the self-adaptive optimization control module, and drives the valve body module to adjust liquid nitrogen supply through the distributed cooperative execution module. And after data fusion is carried out on system basic parameters and execution feedback, a control strategy is optimized through an online learning process, a closed-loop control system with an autonomous evolution capability is formed, and the control precision and the system adaptability are continuously improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control of equipment, and in particular to a vacuum-insulated liquid nitrogen tank storage device and an intelligent control system thereof. BACKGROUND

[0002] The vacuum-insulated liquid nitrogen tank technology is constructed based on the principle of multi-layer insulation, eliminates the convective heat transfer of gas molecules by forming a high-vacuum environment between the container walls, and reflects thermal radiation with the help of a low-emissivity material layer, thereby greatly reducing the transfer of heat from the outside to the internal liquid nitrogen; such a structure enables the liquid nitrogen to be maintained at an extremely low temperature for a long time, and is suitable for biological sample storage and superconducting device cooling, etc. occasions that require stable low temperature.

[0003] The existing vacuum-insulated liquid nitrogen tank technology has the following technical pain points, specifically, the high loss rate of liquid nitrogen evaporation is due to the low efficiency of the existing insulation structure, the heat invasion accelerates the vaporization of liquid nitrogen, and the inaccurate intelligent liquid level control is due to the lack of real-time monitoring and automatic adjustment mechanism, resulting in frequent temperature fluctuations during biological sample storage, for example, when storing precious cell lines, temperature instability may cause ice crystals to form and damage the cell membrane, and insufficient endurance time cannot maintain a constant low-temperature environment in the event of power interruption, increasing the risk of sample denaturation. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides a vacuum-insulated liquid nitrogen tank storage device and an intelligent control system thereof, which solves the technical problems of unstable temperature of biological sample storage and insufficient endurance time caused by high loss rate of liquid nitrogen evaporation and inaccurate intelligent liquid level control.

[0005] To solve the above technical problems, the specific content of the present application is as follows: In a first aspect, the present application provides a vacuum-insulated liquid nitrogen tank storage device, which comprises a physical device and an intelligent control device, wherein the intelligent control device is in communication connection with the physical device; The physical device comprises an outer shell, an inner shell, a tank opening, support columns, foot cups, universal wheels, lifting rings, a liquid nitrogen cold source interlayer, a low-temperature cold screen, a liquid nitrogen tower cylinder, a valve body module, an addition opening, a cryopreservation rack assembly, an aluminum pipe support plate, a cryopreservation rack support plate, a hollow fixing column, an aluminum pipe, an eccentric offset structure, and a fixing bracket; A high-vacuum interlayer is formed between the outer shell and the inner shell; The outer shell is provided with an eccentric offset structure at the top, and the center of the eccentric offset structure is provided with a tank opening; four support columns are welded around the outer side of the outer shell; The fixing bracket is welded to the bottom of the outer shell, and the foot cups and universal wheels are connected through threads at the four corners of the fixing bracket; Two lifting rings are symmetrically welded on both sides of the top of the outer shell; A liquid nitrogen cold source interlayer is welded around the outer wall of the inner container, and a gap of 20-100 mm is reserved between the liquid nitrogen cold source interlayer and the outer wall of the inner container; A plurality of welding support rods are welded between the liquid nitrogen cold source interlayer and the outer wall of the inner container; A liquid nitrogen tower cylinder is welded at the center of the inner container; A valve body module is installed on the outside of the outer container, and the valve body module is provided with an addition port connected to the gap between the liquid nitrogen cold source interlayer and the outer wall of the inner container and the inside of the liquid nitrogen tower cylinder through a pipeline; A cryopreservation rack assembly is installed inside the inner container, which includes aluminum pipe support plates and cryopreservation rack support plates arranged in parallel from top to bottom, and hollow fixing columns are welded vertically between the aluminum pipe support plates and the cryopreservation rack support plates, and the aluminum pipes pass through the aluminum pipe support plates and the cryopreservation rack support plates; A low-temperature cold screen is arranged in the gap between the liquid nitrogen cold source interlayer and the outer wall of the inner container, and the low-temperature cold screen is welded to the outer wall of the inner container; The intelligent control device includes a data acquisition module, a digital twin simulation module, an intelligent diagnosis and prediction module, a self-adaptive optimization control module, a distributed collaborative execution module, and a real-time feedback and online learning module; The data acquisition module acquires liquid nitrogen level, temperature, pressure, humidity, and external environment data in real time and transmits them to the digital twin simulation module; The digital twin simulation module receives the data transmitted by the data acquisition module, constructs a virtual model and performs dynamic simulation, and outputs the simulation results to the intelligent diagnosis and prediction module; The intelligent diagnosis and prediction module receives the simulation results output by the digital twin simulation module, performs data analysis and fault prediction, and outputs the prediction results to the self-adaptive optimization control module; The self-adaptive optimization control module receives the prediction results output by the intelligent diagnosis and prediction module, generates an optimized control strategy, and outputs control instructions to the distributed collaborative execution module; The distributed collaborative execution module receives the control instructions output by the self-adaptive optimization control module, converts the control instructions into physical actions, and drives the valve body module; The real-time feedback and online learning module obtains system basic parameters by monitoring the running state of the intelligent control device, receives execution feedback generated after the distributed collaborative execution module executes the control instructions, and optimizes the control strategy through an online learning process after data fusion of the system basic parameters and the execution feedback to generate learning results, and feeds back the learning results to the data acquisition module and the digital twin simulation module to realize closed-loop control optimization.

[0006] Further, the vacuum insulated liquid nitrogen tank storage device of the present application, the data acquisition module includes a plurality of sensors, the sensors are arranged in the inner container, the liquid nitrogen cold source interlayer and the liquid nitrogen tower barrel, optical fiber sensors and infrared thermal imagers are adopted, data is transmitted to the edge computing node built in the intelligent control device through the Internet of Things protocol, the edge computing node transmits the learning result to the data acquisition module and the digital twin simulation module after filtering and normalizing preprocessing of the data.

[0007] Further, the vacuum insulated liquid nitrogen tank storage device of the present application, the digital twin simulation module adopts computational fluid dynamics to simulate heat flow and liquid nitrogen evaporation process, receives the data transmitted by the data acquisition module, calibrates the model and outputs the simulation result to the intelligent diagnosis and prediction module.

[0008] Further, the vacuum insulated liquid nitrogen tank storage device of the present application, the intelligent diagnosis and prediction module adopts long short-term memory network to process time series data transmitted by the data acquisition module and the digital twin simulation module, integrates support vector machine for liquid nitrogen evaporation anomaly detection, receives the simulation result output by the digital twin simulation module, performs data analysis and fault prediction, and outputs the prediction result to the adaptive optimization control module.

[0009] Further, the vacuum insulated liquid nitrogen tank storage device of the present application, the adaptive optimization control module adopts non-dominated sorting genetic algorithm and model predictive control to dynamically adjust control parameters, receives the prediction result output by the intelligent diagnosis and prediction module, generates an optimized control strategy, and outputs a control instruction to the distributed collaborative execution module.

[0010] Further, the vacuum insulated liquid nitrogen tank storage device of the present application, the distributed collaborative execution module receives the control instruction output by the adaptive optimization control module, drives the inlet valve of the valve body module through the programmable logic controller connected by the adaptive optimization control module, adjusts the liquid nitrogen inflow rate, and feeds back the execution state to the real-time feedback and online learning module.

[0011] Further, the vacuum insulated liquid nitrogen tank storage device of the present application, the real-time feedback and online learning module receives the execution state feedback transmitted by the distributed collaborative execution module, monitors the system state, adopts an incremental learning algorithm trained by real-time feedback data and a federated learning algorithm fed back by the distributed collaborative execution module for online learning, and feeds back the learning result to the data acquisition module and the digital twin simulation module.

[0012] Further, the vacuum insulated liquid nitrogen tank storage device of the present application, a 20-100mm gap is left between the liquid nitrogen cold source interlayer and the inner container, liquid nitrogen is injected into the gap through the inlet to form upper and lower liquid nitrogen protective layers, and the liquid nitrogen level is adjusted by the distributed collaborative execution module.

[0013] Further, the vacuum insulated liquid nitrogen tank storage device adds liquid nitrogen as a very cold source inside the liquid nitrogen tower, and the liquid nitrogen cold source interlayer is injected with liquid nitrogen, which is added through an adding port controlled by a valve body module driven by a distributed collaborative execution module.

[0014] In a second aspect, the application provides an intelligent control system for a vacuum insulated liquid nitrogen tank storage device, comprising: The data acquisition module collects liquid nitrogen level, temperature, pressure, humidity and external environment data, and transmits them to the edge computing node built in the intelligent control device, which transmits the data to the digital twin simulation module after filtering and normalization preprocessing. The digital twin simulation module receives the data transmitted by the edge computing node, simulates the heat flow and liquid nitrogen evaporation process using computational fluid dynamics, constructs a virtual model for dynamic simulation, and outputs the simulation results to the intelligent diagnosis and prediction module. The intelligent diagnosis and prediction module receives the simulation results output by the digital twin simulation module, processes time series data using a long short-term memory network, and integrates a support vector machine for liquid nitrogen evaporation anomaly detection, and outputs the prediction results to the adaptive optimization control module. The adaptive optimization control module receives the prediction results output by the intelligent diagnosis and prediction module, generates an optimized control strategy using a non-dominated sorting genetic algorithm and model predictive control, and outputs control instructions to the distributed collaborative execution module. The distributed collaborative execution module receives the control instructions output by the adaptive optimization control module, adjusts the liquid nitrogen inflow rate through the adding port valve of the valve body module driven by the programmable logic controller, and transmits the execution state feedback to the real-time feedback and online learning module. The real-time feedback and online learning module obtains system basic parameters by monitoring the running state of the intelligent control device, receives execution feedback transmitted by the distributed collaborative execution module, and performs data fusion on the system basic parameters and execution feedback, and then performs online learning using an incremental learning algorithm and a federated learning algorithm to optimize the control strategy and generate learning results, and feeds back the learning results to the data acquisition module and the digital twin simulation module to form a closed-loop control system.

[0015] The application has the following advantages: The application effectively reduces the evaporation loss rate of liquid nitrogen and improves the temperature stability through the synergistic optimization of the multi-layer insulation structure in the physical device and the intelligent control device, specifically, the high-vacuum interlayer between the outer shell and the inner shell significantly inhibits heat conduction, the liquid nitrogen cold source interlayer and the liquid nitrogen tower cylinder form multiple cold source protection to reduce external heat intrusion; the intelligent control device monitors multi-dimensional parameters in real time through the data acquisition module, accurately predicts the liquid nitrogen evaporation trend through the digital twin simulation module, timely identifies abnormalities through the intelligent diagnosis and prediction module, dynamically adjusts the liquid level strategy through the self-adaptive optimization control module, accurately adjusts the liquid nitrogen supply through the distributed collaborative execution module, continuously optimizes the system performance through the real-time feedback and online learning module, thereby prolonging the endurance time, avoiding damage to biological samples due to temperature fluctuations, and reducing maintenance costs and improving equipment reliability. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. Obviously, for those skilled in the art, other drawings can also be obtained without creative labor based on the drawings.

[0017] Fig. 1 It is a whole structure schematic diagram of a vacuum heat insulation liquid nitrogen tank storage device and its intelligent control system.

[0018] Fig. 2 It is a cross-sectional structure diagram of a vacuum heat insulation liquid nitrogen tank storage device and its intelligent control system.

[0019] Fig. 3 It is a valve body module structure diagram of a vacuum heat insulation liquid nitrogen tank storage device and its intelligent control system.

[0020] Fig. 4 It is an inner shell structure schematic diagram of a vacuum heat insulation liquid nitrogen tank storage device and its intelligent control system.

[0021] Fig. 5 It is a cryopreservation rack assembly structure schematic diagram of a vacuum heat insulation liquid nitrogen tank storage device and its intelligent control system.

[0022] BRIEF DESCRIPTION OF DRAWINGS: 1-outer shell, 2-inner shell, 3-tank opening, 4-supporting column, 5-foot cup, 6-universal wheel, 7-lifting ring, 8-liquid nitrogen cold source interlayer, 9-low-temperature cold screen, 10-liquid nitrogen tower cylinder, 11-valve body module, 12-adding opening, 13-cryopreservation rack assembly, 14-aluminum tube supporting plate, 15-cryopreservation rack supporting plate, 16-hollow fixing column, 17-aluminum tube, 18-eccentric offset structure, 19-fixing support. DETAILED DESCRIPTION

[0023] In order to make the technical solutions of the present application clearer, the present application will be described below in conjunction with specific embodiments of the present application and corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application. The present application provided by each embodiment of the present application will be described in detail below in conjunction with the drawings. For the purpose of better understanding the present application, the present application will be further described in detail below.

[0024] In a first aspect, referring to Figs. 1 to 5 The present application provides a vacuum insulated liquid nitrogen tank storage device, comprising a physical device and an intelligent control device, the intelligent control device is in communication connection with the physical device, and the intelligent control device is used for controlling the physical device. The physical device comprises an outer barrel 1, an inner barrel 2, a tank opening 3, a support column 4, a foot cup 5, a universal wheel 6, a lifting ring 7, a liquid nitrogen cold source interlayer 8, a low-temperature cold screen 9, a liquid nitrogen tower cylinder 10, a valve body module 11, an adding opening 12, a cryopreservation rack assembly 13, an aluminum pipe support plate 14, a cryopreservation rack support plate 15, a hollow fixing column 16, an aluminum pipe 17, an eccentric offset structure 18 and a fixing support 19. A high-vacuum interlayer is formed between the outer barrel 1 and the inner barrel 2. The outer barrel 1 is provided with an eccentric offset structure 18 at the top, and the eccentric offset structure 18 is provided with a tank opening 3 in the center; four support columns 4 are welded on the outer side of the four peripheries of the outer barrel 1. The fixing support 19 is welded at the bottom of the outer barrel 1, and the foot cup 5 and the universal wheel 6 are connected through threads at the four corners of the fixing support 19. Two lifting rings 7 are symmetrically welded on both sides of the top of the outer barrel 1. The liquid nitrogen cold source interlayer 8 is welded around the outer wall of the inner barrel 2, and a gap of 20-100 mm is reserved between the liquid nitrogen cold source interlayer 8 and the outer wall of the inner barrel 2. A plurality of welding support rods are welded between the liquid nitrogen cold source interlayer 8 and the outer wall of the inner barrel 2. The liquid nitrogen tower cylinder 10 is welded at the center of the inner barrel 2. The valve body module 11 is installed outside the outer barrel 1, and the valve body module 11 is provided with an adding opening 12, which is connected to the gap between the liquid nitrogen cold source interlayer 8 and the outer wall of the inner barrel 2 and the inside of the liquid nitrogen tower cylinder 10 through a pipeline. The cryopreservation rack assembly 13 is installed inside the inner barrel 2, and the cryopreservation rack assembly 13 comprises an aluminum pipe support plate 14 and a cryopreservation rack support plate 15 arranged in parallel from top to bottom, a hollow fixing column 16 is vertically welded between the aluminum pipe support plate 14 and the cryopreservation rack support plate 15, and an aluminum pipe 17 penetrates through the aluminum pipe support plate 14 and the cryopreservation rack support plate 15. A low-temperature cold shield 9 is arranged in the gap between the liquid nitrogen cold source interlayer 8 and the outer wall of the inner container 2, and the low-temperature cold shield 9 is welded in close contact with the outer wall of the inner container 2; The intelligent control device comprises a data acquisition module, a digital twin simulation module, an intelligent diagnosis and prediction module, a self-adaptive optimization control module, a distributed collaborative execution module, and a real-time feedback and online learning module. The data acquisition module acquires liquid nitrogen liquid level, temperature, pressure, humidity and external environment data in real time and transmits them to the digital twin simulation module. The digital twin simulation module receives the data transmitted by the data acquisition module, constructs a virtual model and performs dynamic simulation, and outputs the simulation results to the intelligent diagnosis and prediction module. The intelligent diagnosis and prediction module receives the simulation results output by the digital twin simulation module, performs data analysis and fault prediction, and outputs the prediction results to the self-adaptive optimization control module. The self-adaptive optimization control module receives the prediction results output by the intelligent diagnosis and prediction module, generates an optimized control strategy, and outputs control instructions to the distributed collaborative execution module. The distributed collaborative execution module receives the control instructions output by the self-adaptive optimization control module, converts the control instructions into physical actions, and drives the valve body module. The real-time feedback and online learning module obtains system basic parameters by monitoring the running state of the intelligent control device, receives execution feedback generated by the distributed collaborative execution module after executing the control instructions, and optimizes the control strategy through an online learning process after data fusion of the system basic parameters and the execution feedback, and generates learning results, which are fed back to the data acquisition module and the digital twin simulation module to realize closed-loop control optimization.

[0025] The present application provides a kind of vacuum heat insulation liquid nitrogen tank storage device, including physical device and intelligent control device, intelligent control device and physical device establish communication connection, realize intelligent monitoring and control.The physical device constitutes the core structure of storage tank, involves heat insulation design, support system, cold source module and sample storage component, each component works cooperatively to maintain low temperature environment and prolong endurance time.

[0026] The outer container 1 and the inner container 2 form a high-vacuum interlayer, which eliminates the heat convection of gas molecules through vacuum pumping, thereby significantly reducing the heat transfer from the external environment to the inner container 2, achieving high-efficiency heat insulation. The top of the outer container 1 adopts an eccentric offset structure 18, which facilitates the access of biological samples by operators and reduces heat exchange when the tank opening 3 is opened. The diameter of the tank opening 3 is suitable for different sample access requirements.

[0027] The outer shell 1 is provided with four support columns 4 around the four corners, which are connected with the tank body by welding to provide structural stability and disperse the load. The bottom is provided with a fixed support 19, which is installed with a foot cup 5 and a universal wheel 6 at the four corners. The foot cup 5 is used to fix the position of the tank body, and the universal wheel 6 is convenient for moving the tank body, thereby enhancing the portability of the equipment. The top of the outer shell 1 is provided with two lifting rings 7 symmetrically on both sides, which are used for lifting and transporting operations to realize safe hoisting of the tank body.

[0028] The liquid nitrogen cold source sandwich 8 is arranged around the inner wall of the tank body and is made of stainless steel material and connected with the tank body by welding. The thickness of the liquid nitrogen cold source sandwich 8 is optimized to resist heat and withstand internal pressure. The welded support rod is arranged between the liquid nitrogen cold source sandwich 8 and the tank body, which enhances the mechanical strength of the liquid nitrogen cold source sandwich 8 and prevents deformation or failure.

[0029] The liquid nitrogen tower cylinder 10 is arranged in the center of the inner shell of the tank body and is connected with the tank body by welding. The diameter and height of the liquid nitrogen tower cylinder 10 are designed to contain liquid nitrogen as the core cold source. The liquid nitrogen evaporated in the liquid nitrogen tower cylinder 10 provides a continuous cooling effect. The valve body module 11 is arranged outside the tank body, which is provided with an independent inlet 12 for injecting liquid nitrogen. The valve body module 11 adjusts the flow of liquid nitrogen through an intelligent control device.

[0030] The cryopreservation rack assembly 13 is arranged inside the inner shell 2, which is sequentially arranged from top to bottom as an aluminum pipe support plate 14 and a cryopreservation rack support plate 15. The aluminum pipe support plate 14 and the cryopreservation rack support plate 15 are fixed by a hollow fixing column 16, which provides vertical support and allows air circulation. The aluminum pipe 17 passes through the aluminum pipe support plate 14 and the cryopreservation rack support plate 15, which is used to place sample containers. The arrangement of the aluminum pipe 17 optimizes the use of space and promotes uniform temperature distribution.

[0031] The intelligent control device is communicatively connected with the physical device. The intelligent control device monitors the liquid level, temperature, pressure, humidity and external environment data of the liquid nitrogen in real time through a data acquisition module, constructs a virtual model for dynamic simulation through a digital twin simulation module, analyzes data and predicts faults through an intelligent diagnosis and prediction module, generates a control strategy through a self-adaptive optimization control module, drives the valve body module 11 to perform actions through a distributed collaborative execution module, adjusts system parameters through a real-time feedback and online learning module, and forms a closed-loop control to maintain stable low temperature and reduce liquid nitrogen evaporation loss. The logical association of each component, from structural insulation to intelligent regulation, cooperatively realizes super-long endurance and safe storage of samples.

[0032] The data acquisition module collects liquid nitrogen level, temperature, pressure, humidity and external environment data in real time. The data acquisition module includes a plurality of sensors arranged in the inner container, the liquid nitrogen cold source interlayer and the liquid nitrogen tower cylinder. The sensors are of optical fiber sensor and infrared thermal imager type. The collected data is transmitted to the edge computing node built in the intelligent control device through the Internet of Things protocol. The edge computing node performs filtering and normalization preprocessing on the data to improve data accuracy and reliability.

[0033] The digital twin simulation module receives the preprocessed data transmitted by the data acquisition module, constructs a virtual model and performs dynamic simulation. The digital twin simulation module simulates the heat flow and liquid nitrogen evaporation process by using computational fluid dynamics method. The virtual model is calibrated by real-time data to accurately reflect the thermodynamic behavior of the physical system, and the simulation results are output.

[0034] The intelligent diagnosis and prediction module receives the simulation results output by the digital twin simulation module, performs data analysis and fault prediction. The intelligent diagnosis and prediction module processes time series data and detects liquid nitrogen evaporation anomalies by using machine learning methods to identify potential fault patterns and output prediction results.

[0035] The adaptive optimization control module receives the prediction results output by the intelligent diagnosis and prediction module, generates an optimized control strategy, and dynamically adjusts control parameters by using an optimization algorithm to cope with system changes and external disturbances, and outputs control instructions.

[0036] The distributed collaborative execution module receives the control instructions output by the adaptive optimization control module, converts the control instructions into physical actions, and drives the inlet valve of the valve body module through the programmable logic controller to adjust the liquid nitrogen inflow rate, thereby achieving precise control of liquid nitrogen supply.

[0037] The real-time feedback and online learning module obtains system basic parameters by monitoring the running state of the intelligent control device, receives execution feedback generated by the distributed collaborative execution module after executing the control instructions, and optimizes the control strategy through the online learning process after data fusion of the system basic parameters and the execution feedback to generate learning results. The learning results are fed back to the data acquisition module and the digital twin simulation module to realize closed-loop control optimization.

[0038] The real-time feedback and online learning module obtains system basic parameters by continuously monitoring the running state of the intelligent control device. The system basic parameters include the working state, resource load condition and performance indicators of each module, providing basic data support for subsequent optimization. At the same time, the real-time feedback and online learning module receives execution feedback generated by the distributed collaborative execution module after executing the control instructions. The execution feedback includes the action state of the valve body module, the liquid nitrogen inflow rate adjustment record and other real-time operation information.

[0039] The system basis parameters and the execution feedback are integrated through multi-source data fusion technology, a data weighting and feature extraction method is used, a unified data set is formed, and comprehensive input is provided for the online learning process. The online learning process dynamically updates the model parameters based on the incremental learning algorithm, and integrates the federal learning framework to aggregate the local learning results of the distributed nodes, thereby realizing continuous optimization of the control strategy.

[0040] The optimized control strategy generates learning results, including updated model parameters and adjustment rules, which are fed back to the data acquisition module and the digital twin simulation module. The data acquisition module uses the learning results to calibrate the sensor data acquisition accuracy, and the digital twin simulation module adjusts the virtual model parameters according to the learning results, thereby improving the simulation accuracy.

[0041] Through the above steps, the real-time feedback and online learning module form a closed-loop control optimization, enabling the intelligent control device to adapt to external environmental changes and internal state fluctuations, gradually improving system control accuracy and stability. The entire process relies on data-driven and machine learning technology to realize the autonomous evolution capability of the intelligent control device.

[0042] Specifically, the vacuum insulated liquid nitrogen tank storage device described in the present application, the data acquisition module includes a plurality of sensors, the sensors are arranged in the inner container 2, the liquid nitrogen cold source interlayer 8 and the liquid nitrogen tower cylinder 10, the optical fiber sensor and the infrared thermal imager are used, the data is transmitted to the edge computing node built-in in the intelligent control device through the Internet of Things protocol, the edge computing node filters and normalizes the data for pretreatment, and then transmits the data to the digital twin simulation module.

[0043] In the data acquisition module of the present application, the sensors include optical fiber sensors and infrared thermal imagers, the optical fiber sensors are installed inside the inner container 2 for monitoring temperature changes, and the infrared thermal imagers are arranged on the surface of the liquid nitrogen cold source interlayer 8 for detecting heat distribution, the data is transmitted to the edge computing node built-in in the intelligent control device through the Internet of Things protocol, the edge computing node executes a filtering algorithm to remove noise interference, and performs normalization processing to unify the data scale, the pretreated data is transmitted to the digital twin simulation module to provide basic input for model construction, thereby realizing seamless connection between data acquisition and subsequent simulation.

[0044] Specifically, the vacuum insulated liquid nitrogen tank storage device described in the present application, the digital twin simulation module uses computational fluid dynamics to simulate heat flow and liquid nitrogen evaporation process, receives the data transmitted by the data acquisition module, calibrates the model and outputs the simulation results to the intelligent diagnosis and prediction module.

[0045] The application adopts a computational fluid dynamics method to establish a heat transfer and fluid dynamics model in the digital twin simulation module, simulates the liquid nitrogen evaporation process and thermal flow behavior, receives preprocessed data transmitted by the data acquisition module, calibrates virtual model parameters to improve simulation accuracy, calibrates the model state based on real-time data dynamic adjustment, outputs simulation results to the intelligent diagnosis and prediction module, and is used to support fault analysis and prediction.

[0046] Specifically, the vacuum insulated liquid nitrogen tank storage device of the application, the intelligent diagnosis and prediction module adopts a long short-term memory network to process time series data transmitted by the data acquisition module and the digital twin simulation module, integrates a support vector machine for liquid nitrogen evaporation anomaly detection, receives simulation results output by the digital twin simulation module, performs data analysis and fault prediction, and outputs prediction results to the adaptive optimization control module.

[0047] In the intelligent diagnosis and prediction module, the application adopts a long short-term memory network to process time series data to capture long-term dependencies, integrates a support vector machine algorithm for liquid nitrogen evaporation anomaly detection, identifies deviations and potential fault patterns in the liquid nitrogen evaporation process, receives simulation results output by the digital twin simulation module, performs data analysis and fault prediction, and outputs prediction results to the adaptive optimization control module to provide a basis for control strategy generation.

[0048] Specifically, the vacuum insulated liquid nitrogen tank storage device of the application, the adaptive optimization control module adopts a non-dominated sorting genetic algorithm and model predictive control to dynamically adjust control parameters, receives prediction results output by the intelligent diagnosis and prediction module, generates an optimized control strategy, and outputs control instructions to the distributed collaborative execution module.

[0049] In the adaptive optimization control module, the application adopts a non-dominated sorting genetic algorithm for multi-objective optimization, dynamically adjusts control parameters in combination with model predictive control, generates an optimized control strategy to minimize liquid nitrogen evaporation loss and maintain temperature stability, receives prediction results output by the intelligent diagnosis and prediction module, optimizes control instructions and outputs them to the distributed collaborative execution module, and realizes the adaptive ability of the system to respond to external changes and disturbances.

[0050] Specifically, the vacuum insulated liquid nitrogen tank storage device of the application, the distributed collaborative execution module receives control instructions output by the adaptive optimization control module, drives the valve of the joining port 12 of the valve body module 11 through the programmable logic controller connected by the adaptive optimization control module, adjusts the liquid nitrogen inflow rate, and transmits the execution state feedback to the real-time feedback and online learning module.

[0051] The application receives the control instruction output by the adaptive optimization control module through the programmable logic controller in the distributed collaborative execution module, drives the valve of the joining port 12 of the valve body module 11, adjusts the inflow rate of liquid nitrogen, and converts the control instruction into physical action to execute the liquid nitrogen supply management.

[0052] Specifically, the vacuum insulated liquid nitrogen tank storage device of the application receives the execution state feedback transmitted by the distributed collaborative execution module, monitors the system state, adopts the incremental learning algorithm trained by real-time feedback data and the federated learning algorithm fed back by the distributed collaborative execution module for online learning, and feeds back the learning result to the data acquisition module and the digital twin simulation module.

[0053] In the real-time feedback and online learning module, the model parameters are updated in real time based on new data using the incremental learning algorithm, the learning results are aggregated from distributed nodes using the federated learning algorithm, online learning is performed to optimize system performance, execution state feedback is received from the distributed collaborative execution module to monitor system state, and learning results are fed back to the data acquisition module and the digital twin simulation module to form a closed-loop control system for continuous improvement of accuracy and efficiency.

[0054] Specifically, the vacuum insulated liquid nitrogen tank storage device of the application leaves a 20-100mm gap between the liquid nitrogen cold source interlayer 8 and the inner container 2, injects liquid nitrogen into the gap through the joining port 12 to form upper and lower liquid nitrogen protective layers, and adjusts the liquid nitrogen level through the distributed collaborative execution module.

[0055] In the design of the liquid nitrogen cold source interlayer 8, a gap is reserved between the inner container 2, liquid nitrogen is injected into the gap through the joining port 12 to form upper and lower liquid nitrogen protective layers, which are used to enhance the heat insulation effect and temperature stability, the liquid nitrogen level is adjusted by the distributed collaborative execution module, and the distributed collaborative execution module drives the valve body module 11 to control the valve of the joining port 12 to maintain the balance of liquid nitrogen supply and system endurance.

[0056] Specifically, the vacuum insulated liquid nitrogen tank storage device of the application, liquid nitrogen is added inside the liquid nitrogen tower cylinder 10 as a very cold source, liquid nitrogen is injected into the liquid nitrogen cold source interlayer 8, liquid nitrogen is added through the joining port 12, the joining port 12 is controlled by the valve body module 11, and the valve body module 11 is driven by the distributed collaborative execution module.

[0057] In the configuration of the liquid nitrogen tower cylinder 10, liquid nitrogen is added inside as the core cold source, liquid nitrogen is injected into the liquid nitrogen cold source interlayer 8 at the same time, the joining port 12 is controlled by the valve body module 11, the valve body module 11 is driven by the distributed collaborative execution module, the collaborative operation of liquid nitrogen supply management is realized, and the liquid nitrogen tower cylinder 10 is connected to the tank body by welding for providing centralized cooling and supporting structural integrity.

[0058] In a second aspect, the present application provides an intelligent control system for a vacuum-insulated liquid nitrogen tank storage device, comprising: The data acquisition module collects liquid nitrogen level, temperature, pressure, humidity and external environment data, and transmits them to the edge computing node built in the intelligent control device. The edge computing node transmits the data to the digital twin simulation module after filtering and normalization preprocessing. The digital twin simulation module receives the data transmitted by the edge computing node, simulates the heat flow and liquid nitrogen evaporation process by computational fluid dynamics, constructs a virtual model for dynamic simulation, and outputs the simulation results to the intelligent diagnosis and prediction module. The intelligent diagnosis and prediction module receives the simulation results output by the digital twin simulation module, processes time series data using long short-term memory network, and integrates support vector machine for liquid nitrogen evaporation anomaly detection, and outputs the prediction results to the adaptive optimization control module. The adaptive optimization control module receives the prediction results output by the intelligent diagnosis and prediction module, generates an optimized control strategy using non-dominated sorting genetic algorithm and model predictive control, and outputs control instructions to the distributed collaborative execution module. The distributed collaborative execution module receives the control instructions output by the adaptive optimization control module, drives the valve adjustment of the inlet valve 12 of the valve module 11 through the programmable logic controller to regulate the liquid nitrogen inflow rate, and transmits the execution state feedback to the real-time feedback and online learning module. The real-time feedback and online learning module receives the execution state feedback transmitted by the distributed collaborative execution module, performs online learning using incremental learning algorithm and federated learning algorithm, and feeds back the learning results to the data acquisition module and the digital twin simulation module to form a closed-loop control system.

[0059] The real-time feedback and online learning module continuously monitors the running state of the intelligent control device to obtain system basic parameters, including the working state, resource load condition and performance index of each module, to provide basic data support for subsequent optimization. The module also receives the execution feedback generated after the distributed collaborative execution module executes the control instructions, including the action state of the valve module, the liquid nitrogen inflow rate adjustment record and other real-time operation information. The system basic parameters and the execution feedback are integrated through multi-source data fusion technology, and a unified data set is formed through data weighting and feature extraction method, providing comprehensive input for online learning process.

[0060] The online learning process dynamically updates model parameters based on an incremental learning algorithm, and integrates a federated learning framework to aggregate local learning results of distributed nodes, thereby realizing continuous optimization of the control strategy. The optimized control strategy generates learning results, including updated model parameters and adjustment rules, which are fed back to the data acquisition module and the digital twin simulation module. The data acquisition module calibrates the sensor data acquisition accuracy using the learning results, and the digital twin simulation module corrects the virtual model parameters according to the learning results, thereby improving the simulation accuracy. Through the above steps, a closed-loop control optimization is formed between the real-time feedback and the online learning module, enabling the intelligent control device to adapt to external environmental changes and internal state fluctuations, and gradually improving the system control precision and stability. The entire process relies on data-driven and machine learning techniques to realize the autonomous evolution capability of the intelligent control device.

[0061] The real-time feedback and online learning module uses a feature-level fusion method in the data fusion stage to associate and analyze the temperature gradient distribution in the system basic parameters and the valve opening change in the execution feedback, and extracts the key feature vector as the input of the online learning model. The incremental learning algorithm processes real-time data streams through a sliding window mechanism, dynamically adjusts the neural network weights, and avoids model overfitting problems. The federated learning framework integrates the local training results of multiple distributed nodes through an encrypted parameter aggregation method, while ensuring data privacy and improving the model generalization capability. The learning result feedback mechanism adopts an asynchronous update strategy, the data acquisition module optimizes the sampling frequency and filtering parameters according to the learning results, and the digital twin simulation module reduces the model prediction error through parameter correction, forming a collaborative optimization cycle.

[0062] During the operation of the module, the acquisition of system basic parameters covers the full life cycle state of the intelligent control device, including processor load, memory usage, and communication delay indicators, and abnormal fluctuation patterns are identified through time series analysis. The analysis of execution feedback data focuses on valve module response delay and liquid nitrogen flow deviation, and establishes a mapping relationship with control instructions. The data fusion process introduces an entropy weight method to determine the weight of multi-source data, enhancing the contribution of key information. The online learning model uses a long short-term memory network structure to capture long-term dependencies in system dynamic characteristics, and the federated learning round interval is adaptively adjusted according to network conditions, balancing learning efficiency and communication overhead. The feedback path of the learning result is designed as a two-way channel, the data acquisition module prioritizes high-priority calibration instructions, and the digital twin simulation module uses a gradual parameter update strategy to minimize system disturbances.

[0063] The real-time feedback and online learning module optimizes the closed-loop control through an iterative learning mechanism to improve performance, and each learning cycle includes four stages of data collection, fusion, training, and feedback. The module uses a multi-threshold alarm mechanism to monitor the operating state and trigger the adaptive learning process in a timely manner. The feedback data is pre-processed to remove noise interference and retain key control features. The data fusion algorithm combines principal component analysis dimension reduction and Kalman filter prediction to improve input data quality. The online learning process introduces reinforcement learning elements, guides control strategy optimization through a reward function, and selects federal learning nodes based on a reputation evaluation model to achieve collaboration reliability. After the learning results are fed back, the system verifies the effectiveness of the strategy through A / B testing, gradually promotes the optimization scheme, and forms a stable and reliable self-evolving system.

[0064] The physical device and the intelligent control device are cooperated to solve the problems of high evaporation loss rate of liquid nitrogen and inaccurate intelligent liquid level control. The high vacuum interlayer is formed between the outer shell and the inner shell in the physical device, which effectively eliminates the heat convection of gas and reduces the heat invasion; the liquid nitrogen cold source interlayer and the liquid nitrogen tower cylinder provide multiple cold source protection to reduce the evaporation rate of liquid nitrogen. In the intelligent control device, the data acquisition module monitors the liquid level, temperature, pressure, humidity and external environment data of liquid nitrogen in real time, obtains high-precision information through optical fiber sensors and infrared thermal imagers, and transmits data to edge computing nodes for filtering and normalization preprocessing. The digital twin simulation module simulates the heat flow and liquid nitrogen evaporation process by using computational fluid dynamics, constructs a virtual model and dynamically calibrates it, and outputs the simulation results. The intelligent diagnosis and prediction module processes time series data, uses long short-term memory network and support vector machine for anomaly detection and fault prediction, and generates prediction results. The adaptive optimization control module uses non-dominated sorting genetic algorithm and model predictive control to dynamically adjust control parameters and generate optimized control strategies. The distributed collaborative execution module drives the inlet valve of the valve body module through the programmable logic controller, adjusts the liquid nitrogen inflow rate, and accurately controls the liquid level. The real-time feedback and online learning module monitors the system state, uses incremental learning algorithm and federal learning algorithm for online learning, feeds back the learning results to the data acquisition module and the digital twin simulation module, forms a closed-loop control system, and continuously optimizes the performance, thereby maintaining temperature stability and prolonging the endurance time.

[0065] The embodiment of the present application relates to the collaborative work flow of the physical structure and intelligent control system of the vacuum insulated liquid nitrogen tank storage device. A high vacuum interlayer is formed between the outer tank 1 and the inner tank 2 in the physical device, and the heat transfer from the outside environment to the inner tank 2 is significantly reduced by eliminating the convection heat transfer of gas molecules through vacuumizing treatment, thereby realizing high-efficiency heat insulation. An eccentric offset structure 18 is arranged at the top of the outer tank 1, and a tank opening 3 is arranged at the center of the eccentric offset structure 18, so as to facilitate the access of biological samples by the operator and reduce heat exchange. Four support columns 4 are welded on the outer side of the four walls of the outer tank 1, so as to provide structural stability and load dispersion. A fixed support 19 is welded at the bottom of the outer tank 1, four corners of the fixed support 19 are connected with a foot cup 5 and a universal wheel 6 through threads, the foot cup 5 is used for fixing the position of the tank body, and the universal wheel 6 facilitates the movement of the device. Two lifting rings 7 are symmetrically welded on both sides of the top of the outer tank 1, and are used for lifting and transporting operations.

[0066] A liquid nitrogen cold source interlayer 8 is welded around the outer wall of the inner tank 2, a gap is reserved between the liquid nitrogen cold source interlayer 8 and the outer wall of the inner tank 2, and the width of the gap is suitable for the heat resistance optimization requirement. A plurality of welding support rods are welded between the liquid nitrogen cold source interlayer 8 and the outer wall of the inner tank 2, so as to enhance the mechanical strength and prevent deformation. A liquid nitrogen tower cylinder 10 is welded at the center of the inner tank 2, the liquid nitrogen tower cylinder 10 contains liquid nitrogen as a core cold source, and provides a continuous cooling effect. A valve body module 11 is installed outside the outer tank 1, the valve body module 11 is provided with an adding port 12, the adding port 12 is connected with the gap between the liquid nitrogen cold source interlayer 8 and the outer wall of the inner tank 2 and the inside of the liquid nitrogen tower cylinder 10 through a pipeline, and liquid nitrogen injection is realized. A cryopreservation rack assembly 13 is installed inside the inner tank 2, the cryopreservation rack assembly 13 includes an aluminum pipe support plate 14 and a cryopreservation rack support plate 15 which are arranged in parallel from top to bottom, a hollow fixing column 16 is vertically welded between the aluminum pipe support plate 14 and the cryopreservation rack support plate 15, and an aluminum pipe 17 penetrates through the aluminum pipe support plate 14 and the cryopreservation rack support plate 15, which is used for placing sample containers and promoting uniform temperature distribution. A low-temperature cold screen 9 is arranged in the gap between the liquid nitrogen cold source interlayer 8 and the outer wall of the inner tank 2, and the low-temperature cold screen 9 is welded in close contact with the outer wall of the inner tank 2, so as to enhance the heat insulation effect.

[0067] An intelligent control device is communicatively connected with the physical device, so as to realize intelligent monitoring and control. A data acquisition module acquires liquid nitrogen level, temperature, pressure, humidity and external environment data in real time, the data acquisition module includes a plurality of sensors arranged in the inner tank 2, the liquid nitrogen cold source interlayer 8 and the liquid nitrogen tower cylinder 10, the sensors adopt optical fiber sensors and infrared thermal imagers, and the acquired data is transmitted to an edge computing node built in the intelligent control device through an Internet of Things protocol. The edge computing node performs filtering and normalization preprocessing on the data, so as to improve the data accuracy, and the preprocessed data is transmitted to a digital twin simulation module.

[0068] The digital twin simulation module receives the data transmitted by the data acquisition module, simulates the heat flow and liquid nitrogen evaporation process using computational fluid dynamics, constructs a virtual model and performs dynamic simulation. The digital twin simulation module calibrates the virtual model in real time through real-time data, accurately reflects the thermodynamic behavior of the physical system, and outputs the simulation results to the intelligent diagnosis and prediction module. The intelligent diagnosis and prediction module receives the simulation results output by the digital twin simulation module, processes time series data using a long short-term memory network, integrates a support vector machine for liquid nitrogen evaporation anomaly detection, identifies potential fault patterns, performs data analysis and fault prediction, and outputs the prediction results to the adaptive optimization control module.

[0069] The adaptive optimization control module receives the prediction results output by the intelligent diagnosis and prediction module, dynamically adjusts the control parameters using a non-dominated sorting genetic algorithm and model predictive control, and generates an optimized control strategy. The optimized control strategy aims to minimize liquid nitrogen evaporation loss and maintain temperature stability, and the adaptive optimization control module outputs control instructions to the distributed collaborative execution module. The distributed collaborative execution module receives the control instructions output by the adaptive optimization control module, drives the valve of the valve module 11 through the programmable logic controller, adjusts the liquid nitrogen inflow rate, and realizes accurate control of liquid nitrogen supply. The distributed collaborative execution module transmits execution state feedback to the real-time feedback and online learning module.

[0070] The real-time feedback and online learning module obtains system basic parameters by monitoring the running state of the intelligent control device. The system basic parameters include the working state and performance indicators of each module. The real-time feedback and online learning module simultaneously receives the execution feedback transmitted by the distributed collaborative execution module. The execution feedback includes the action state of the valve module 11 and the liquid nitrogen inflow rate adjustment record. The system basic parameters and execution feedback are integrated through multi-source data fusion technology, and a unified data set is formed using data weighting and feature extraction methods. The online learning process dynamically updates model parameters based on an incremental learning algorithm, and integrates a federated learning framework to aggregate local learning results of distributed nodes, realizing continuous optimization of the control strategy. The optimized control strategy generates learning results, including updated model parameters and adjustment rules. The learning results are fed back to the data acquisition module and the digital twin simulation module. The data acquisition module uses the learning results to calibrate the sensor data acquisition accuracy, and the digital twin simulation module adjusts the virtual model parameters according to the learning results to improve the simulation accuracy, forming a closed-loop control system.

[0071] The whole implementation process is optimized through the cooperation of the multiple cold source structure of the physical device and the intelligent control device, effectively reduces the evaporation loss rate of liquid nitrogen, improves the temperature stability, and prolongs the endurance time. The high vacuum interlayer and the liquid nitrogen cold source interlayer 8 in the physical device reduce the external heat invasion, the intelligent control device realizes the autonomous evolution ability through the data driving and the machine learning technology, adapts to the external environment change and the internal state fluctuation, and solves the technical problems of temperature instability and insufficient endurance in biological sample storage.

[0072] Embodiment 1 of the present application: The implementation of the vacuum insulated liquid nitrogen tank storage device involves the cooperative configuration of physical devices and intelligent control devices. A high-vacuum interlayer is formed between the outer shell 1 and the inner shell 2 in the physical device, and the heat transfer is significantly reduced by eliminating the convection heat transfer of gas molecules through vacuumization processing. The outer shell 1 is provided with an eccentric offset structure 18 at the top, and a tank opening 3 is formed in the center of the eccentric offset structure 18, which facilitates the access of biological samples and reduces heat exchange. Four support columns 4 are welded around the outer side of the outer shell 1 to provide structural support, and a fixed support 19 is welded at the bottom to enhance portability through threaded connection with the foot cup 5 and the universal wheel 6. The outer wall of the inner shell 2 is welded with a liquid nitrogen cold source interlayer 8, and a gap is reserved between the liquid nitrogen cold source interlayer 8 and the outer wall of the inner shell 2. A low-temperature cold screen 9 is arranged in the gap and welded to the outer wall of the inner shell 2 to enhance the heat insulation effect. A liquid nitrogen tower cylinder 10 is welded at the center of the inner shell 2 to contain liquid nitrogen as the core cold source. A valve body module 11 is installed outside the outer shell 1, which is provided with an addition port 12 connected to the gap between the liquid nitrogen cold source interlayer 8 and the outer wall of the inner shell 2 and the inside of the liquid nitrogen tower cylinder 10 through a pipeline to realize the injection of liquid nitrogen. A cryopreservation rack assembly 13 is installed inside the inner shell 2, which includes an aluminum tube support plate 14 and a cryopreservation rack support plate 15, between which a hollow fixed column 16 is vertically welded, and an aluminum tube 17 penetrates through the aluminum tube support plate 14 and the cryopreservation rack support plate 15 for placing sample containers. The intelligent control device collects liquid nitrogen level, temperature, pressure, humidity and external environment data in real time through a data acquisition module, which is set in the inner shell 2, the liquid nitrogen cold source interlayer 8 and the liquid nitrogen tower cylinder 10 using optical fiber sensors and infrared thermographs. The data is filtered and normalized by the edge computing node, and then transmitted to the digital twin simulation module. The digital twin simulation module simulates the heat flow and liquid nitrogen evaporation process using computational fluid dynamics, constructs a virtual model and outputs the simulation results to the intelligent diagnosis and prediction module. The intelligent diagnosis and prediction module analyzes the data and predicts faults using long short-term memory networks and support vector machines, and outputs the prediction results to the adaptive optimization control module. The adaptive optimization control module generates an optimized control strategy using non-dominated sorting genetic algorithm and model predictive control, and outputs control instructions to the distributed collaborative execution module. The distributed collaborative execution module drives the valve adjustment of the addition port 12 of the valve body module 11 through a programmable logic controller to regulate the inflow rate of liquid nitrogen, and feeds back the execution state to the real-time feedback and online learning module. The real-time feedback and online learning module monitors the system state, optimizes the control strategy based on the execution feedback and system basic parameters using incremental learning algorithm and federated learning algorithm, and generates learning results feedback to the data acquisition module and the digital twin simulation module, forming a closed-loop control system to improve temperature stability and endurance time.

[0073] Embodiment 2 of the present application: In the biological sample storage scenario, the vacuum insulated liquid nitrogen tank storage device realizes self-adaptive regulation and control through the intelligent control device. The gap between the liquid nitrogen cold source interlayer 8 and the inner container 2 in the physical device is injected with liquid nitrogen to form upper and lower protective layers. Liquid nitrogen is added to the liquid nitrogen tower 10 as an extremely cold source, and the liquid nitrogen supply is adjusted through the valve body module 11 controlling the addition port 12. The data acquisition module monitors environmental parameters in real time, and sensor data is transmitted to the edge computing node for processing through the Internet of Things protocol. The digital twin simulation module receives preprocessed data, calibrates the virtual model to accurately simulate liquid nitrogen evaporation and heat flow, and outputs simulation results for fault prediction. The intelligent diagnosis and prediction module processes time series data, identifies liquid nitrogen evaporation anomalies, and outputs prediction results. The self-adaptive optimization control module dynamically adjusts control parameters to generate optimization strategies to minimize evaporation loss. The distributed collaborative execution module drives the valve body module 11 to perform liquid nitrogen regulation actions and returns execution feedback. The real-time feedback and online learning module fuses multi-source data and updates model parameters through online learning. The learning results are used to calibrate the data acquisition and simulation modules to achieve system autonomous evolution. Through the coordinated optimization of the multiple cold source structure of the physical device and the intelligent control device, the whole process effectively reduces heat intrusion, maintains a low-temperature environment, and solves the problems of temperature fluctuations and insufficient endurance.

[0074] The technical terms of the present application are explained as follows: Computational fluid dynamics is a simulation technology that solves fluid motion and control equations through numerical methods, used to simulate flow behavior and heat transfer processes in complex systems. In the present application, the digital twin simulation module uses computational fluid dynamics to simulate the liquid nitrogen evaporation process and heat flow characteristics. By constructing a virtual model to analyze the heat exchange mechanism between the liquid nitrogen cold source interlayer 8 and the inner container 2, the model parameters are calibrated to accurately predict temperature distribution and liquid nitrogen evaporation rate, providing a high-precision simulation data basis for intelligent diagnosis.

[0075] Long short-term memory network is a recurrent neural network structure that can handle long-term dependencies, suitable for modeling and predicting time series data. In the present application, the intelligent diagnosis and prediction module uses long short-term memory network to process temperature, pressure and time series data collected by sensors, identifies abnormal patterns in the liquid nitrogen evaporation process, and predicts potential failures by analyzing historical data trends, providing dynamic decision-making basis for adaptive control.

[0076] Support vector machine is a classification algorithm based on statistical learning theory, which realizes data classification and anomaly detection by finding the optimal hyperplane. In the present application, the intelligent diagnosis and prediction module integrates support vector machine for liquid nitrogen evaporation anomaly detection, analyzes the feature vectors of liquid nitrogen level and temperature data, identifies abnormal points deviating from normal state, realizes early fault warning, and forms a complementary detection mechanism with long short-term memory network.

[0077] Non-dominated Sorting Genetic Algorithm (NSGA) is a multi-objective optimization algorithm that balances the trade-offs between conflicting objectives through a set of Pareto-optimal solutions. In this invention, the adaptive optimization control module employs NSGA to dynamically adjust control parameters while simultaneously optimizing both the minimization of liquid nitrogen evaporation loss and the maximization of temperature stability, generating a set of non-dominated solutions and providing an optimization direction for model predictive control.

[0078] Model Predictive Control (MPC) is a feedforward-feedback control strategy based on a dynamic model, achieving system control through rolling optimization and feedback correction. In this invention, the adaptive optimization control module integrates the MPC framework, utilizing the predictive model provided by the digital twin simulation module to rollingly optimize the liquid nitrogen supply strategy, and precisely controlling the liquid nitrogen inflow rate by adjusting the valve opening of the inlet valve 12 of the valve module 11 in real time.

[0079] Programmable Logic Controller (PLC) is a digital computer designed specifically for industrial environments, used for automated control of manufacturing processes by monitoring input signals and controlling output devices through programmed logic. In this invention, the distributed collaborative execution module drives the inlet valve of the valve module 11 through the programmable logic controller, receives the control instructions output by the adaptive optimization control module, accurately adjusts the liquid nitrogen inflow rate, realizes real-time regulation of liquid nitrogen supply, and feeds back the execution status to the real-time feedback and online learning module, achieving accurate execution of physical actions.

[0080] Incremental Learning Algorithm is a machine learning technique that allows models to gradually adapt to new data without the need to retrain the entire data set, achieving continuous learning through dynamic parameter updates. In this invention, the real-time feedback and online learning module uses the incremental learning algorithm to process real-time feedback data, optimizes the control strategy based on newly collected system state and execution feedback information, gradually adjusts model parameters to improve system adaptability and control accuracy, and avoids historical data redundancy and model degradation.

[0081] Federated Learning Algorithm is a distributed machine learning framework that protects data privacy by training models on local devices and only aggregating model updates to achieve collaborative learning. In this invention, the real-time feedback and online learning module integrates the federated learning algorithm, aggregates the local learning results fed back by the distributed collaborative execution module, integrates optimization experience from multiple nodes, enhances the model generalization ability, and at the same time realizes that sensitive operation data does not leave the local node, maintaining system security.

[0082] Internet of Things (IoT) Protocol is a set of communication standards used to connect IoT devices and enable data exchange, supporting low power consumption and long-range transmission. In this invention, the data acquisition module transmits sensor data from the inner container 2, the liquid nitrogen cold source interlayer 8, and the liquid nitrogen tower cylinder 10 to the edge computing node built-in the intelligent control device through the IoT protocol, achieving reliable and efficient communication of monitoring data, and providing stable data flow for subsequent processing.

[0083] Edge computing node is a computing device deployed near data sources for local data processing and analysis, reducing latency and network load. In the invention, the edge computing node receives sensor data transmitted by the data acquisition module, performs real-time computing tasks, and preliminarily processes parameters such as liquid nitrogen level, temperature, and pressure to provide preprocessed input for the digital twin simulation module, improving system response speed.

[0084] Data filtering is a signal processing technique that removes noise and outliers through algorithms to improve data quality and reliability. In the invention, the edge computing node performs filtering on the collected sensor data, using digital filters to eliminate environmental interference and measurement errors, so that the data transmitted to the digital twin simulation module accurately reflects the true state of the system, reducing simulation errors.

[0085] Normalization preprocessing is a data standardization method that converts feature values of different scales to a unified range, eliminating dimensional effects. In the invention, the edge computing node performs normalization preprocessing on the filtered data, scaling liquid nitrogen level, temperature, and other parameters to a standard interval, making the data suitable for machine learning model processing and improving the simulation accuracy and convergence speed of the digital twin simulation module.

[0086] Digital twin simulation is a technology that simulates and predicts real-world behavior by building a virtual model of the physical system, using computational fluid dynamics to dynamically simulate heat flow and liquid nitrogen evaporation. In the invention, the digital twin simulation module receives liquid nitrogen level, temperature, and pressure data transmitted by the data acquisition module, builds a high-precision virtual model to calibrate the thermodynamic state of the inner container 2 and the liquid nitrogen cold source interlayer 8, and outputs simulation results to the intelligent diagnosis and prediction module, realizing real-time prediction and optimization of system behavior.

[0087] Time series data processing is a method of analyzing data points arranged in chronological order to identify trends, periodicity, and abnormal patterns. In the invention, the intelligent diagnosis and prediction module uses a long short-term memory network to process time series data transmitted by the data acquisition module and the digital twin simulation module, analyzes the variation law of liquid nitrogen level and temperature, and predicts the evaporation trend of liquid nitrogen to provide decision basis for the adaptive optimization control module.

[0088] Anomaly detection is a technique for identifying data that deviates from normal patterns by statistical or machine learning methods to discover potential faults. In the invention, the intelligent diagnosis and prediction module integrates a support vector machine for liquid nitrogen evaporation anomaly detection, analyzes sensor data of the liquid nitrogen cold source interlayer 8 and the liquid nitrogen tower cylinder 10, identifies evaporation rate anomalies and temperature deviations, and timely warns of fault risks.

[0089] Multi-source data fusion is a technique that integrates data from different sources to form a unified dataset through weighting and feature extraction. In this invention, the real-time feedback and online learning module fuses the system base parameters with the execution feedback of the distributed collaborative execution module, determines the data weight using the entropy weight method, enhances the collaborative analysis of the liquid nitrogen inflow rate and valve state information, and provides comprehensive input for online learning.

[0090] Closed-loop control optimization is a technique that dynamically adjusts system behavior through a feedback mechanism to achieve continuous performance improvement. In this invention, the system monitors the valve state of the inlet valve 12 of the valve module 11 through the real-time feedback and online learning module, feeds the learning results back to the data acquisition module and digital twin simulation module, forms a closed-loop control optimization, accurately adjusts the liquid nitrogen supply, and maintains temperature stability.

[0091] Autonomous evolution capability is a technique that enables a system to adapt to environmental changes through machine learning without human intervention to continuously optimize performance. In this invention, the real-time feedback and online learning module uses incremental learning algorithms and federated learning algorithms for online learning, dynamically updates model parameters, enables the intelligent control device to adapt to external disturbances of the inner container 2 and the liquid nitrogen cold source interlayer 8, and achieves system autonomous evolution and long-term reliability improvement.

Claims

1. A vacuum-insulated liquid nitrogen tank storage device, characterized in that, It includes a physical device and an intelligent control device, wherein the intelligent control device establishes a communication connection with the physical device and is used to control the physical device; The physical device includes an outer liner (1), an inner liner (2), a can opening (3), a support column (4), a foot cup (5), casters (6), a lifting ring (7), a liquid nitrogen cold source interlayer (8), a low-temperature cold shield (9), a liquid nitrogen tower (10), a valve module (11), an inlet (12), a cryogenic rack assembly (13), an aluminum tube support plate (14), a cryogenic rack support plate (15), a hollow fixed column (16), an aluminum tube (17), an eccentric off-center structure (18), and a fixed bracket (19). A high-vacuum interlayer is formed between the outer liner (1) and the inner liner (2); The outer liner (1) has an eccentric opening structure (18) at the top, and the opening (3) is opened in the center of the eccentric opening structure (18). The outer liner (1) has four supporting columns (4) welded to its outer sides; The bottom of the outer liner (1) is welded with a fixed bracket (19), and the four corners of the fixed bracket (19) are connected to the foot cup (5) and the caster wheel (6) by threads. Two lifting rings (7) are symmetrically welded to the top two sides of the outer liner (1); A liquid nitrogen cold source interlayer (8) is welded around the outer wall of the inner liner (2), and a gap of 20-100mm is reserved between the liquid nitrogen cold source interlayer (8) and the outer wall of the inner liner (2); Multiple welded support rods are welded between the liquid nitrogen cold source interlayer (8) and the outer wall of the inner liner (2); The inner liner (2) is welded to the center of the liquid nitrogen tower cylinder (10); The outer liner (1) is equipped with a valve body module (11). The valve body module (11) is provided with an inlet (12). The inlet (12) is connected to the gap between the liquid nitrogen cold source jacket (8) and the outer wall of the inner liner (2) and the inside of the liquid nitrogen tower (10) through a pipe. The inner liner (2) is equipped with a cryopreservation rack assembly (13). The cryopreservation rack assembly (13) includes an aluminum tube support plate (14) and a cryopreservation rack support plate (15) arranged in parallel from top to bottom. A hollow fixing column (16) is vertically welded between the aluminum tube support plate (14) and the cryopreservation rack support plate (15). An aluminum tube (17) passes through the aluminum tube support plate (14) and the cryopreservation rack support plate (15). A low-temperature cold shield (9) is installed in the gap between the liquid nitrogen cold source interlayer (8) and the outer wall of the inner liner (2), and the low-temperature cold shield (9) is welded to the outer wall of the inner liner (2). The intelligent control device includes a data acquisition module, a digital twin simulation module, an intelligent diagnosis and prediction module, an adaptive optimization control module, a distributed collaborative execution module, and a real-time feedback and online learning module. The data acquisition module collects real-time data on liquid nitrogen level, temperature, pressure, humidity, and external environment, and transmits it to the digital twin simulation module. The digital twin simulation module receives data transmitted from the data acquisition module, constructs a virtual model and performs dynamic simulation, and outputs the simulation results to the intelligent diagnosis and prediction module. The intelligent diagnosis and prediction module receives the simulation results output by the digital twin simulation module, performs data analysis and fault prediction, and outputs the prediction results to the adaptive optimization control module. The adaptive optimization control module receives the prediction results output by the intelligent diagnosis and prediction module, generates an optimized control strategy, and outputs control commands to the distributed collaborative execution module. The distributed collaborative execution module receives control commands from the adaptive optimization control module, converts the control commands into physical actions, and drives the valve body module. The real-time feedback and online learning module obtains basic system parameters by monitoring the operating status of the intelligent control device. At the same time, it receives execution feedback generated by the distributed collaborative execution module after executing control commands. After fusing the basic system parameters and execution feedback, it optimizes the control strategy and generates learning results through an online learning process, and feeds the learning results back to the data acquisition module and the digital twin simulation module.

2. The vacuum-insulated liquid nitrogen tank storage device according to claim 1, characterized in that, The data acquisition module includes multiple sensors, which are installed in the inner liner (2), the liquid nitrogen cold source interlayer (8), and the liquid nitrogen tower (10). Fiber optic sensors and infrared thermal imagers are used. The data is transmitted to the edge computing node built into the intelligent control device through the Internet of Things protocol. The edge computing node filters and normalizes the data before transmitting it to the digital twin simulation model.

3. The vacuum-insulated liquid nitrogen tank storage device according to claim 2, characterized in that, The digital twin simulation module uses computational fluid dynamics to simulate thermal flow and liquid nitrogen evaporation processes, receives data transmitted from the data acquisition module, calibrates the model, and outputs simulation results to the intelligent diagnosis and prediction module.

4. The vacuum-insulated liquid nitrogen tank storage device according to claim 3, characterized in that, The intelligent diagnosis and prediction module uses a long short-term memory network to process the time series data transmitted by the data acquisition module and the digital twin simulation module, integrates a support vector machine to detect liquid nitrogen evaporation anomalies, receives the simulation results output by the digital twin simulation module, performs data analysis and fault prediction, and outputs the prediction results to the adaptive optimization control module.

5. The vacuum-insulated liquid nitrogen tank storage device according to claim 4, characterized in that, The adaptive optimization control module uses a non-dominated sorting genetic algorithm and model predictive control to dynamically adjust control parameters. It receives prediction results from the intelligent diagnosis and prediction module, generates an optimized control strategy, and outputs control commands to the distributed collaborative execution module.

6. The vacuum-insulated liquid nitrogen tank storage device according to claim 5, characterized in that, The distributed collaborative execution module receives the control instructions output by the adaptive optimization control module, drives the valve at the inlet (12) of the valve body module (11) through the programmable logic controller connected to the adaptive optimization control module, adjusts the liquid nitrogen inflow rate, and transmits the execution status feedback to the real-time feedback and online learning module.

7. The vacuum-insulated liquid nitrogen tank storage device according to claim 6, characterized in that, The real-time feedback and online learning module receives execution status feedback transmitted from the distributed collaborative execution module, monitors the system status, and performs online learning using an incremental learning algorithm trained on real-time feedback data and a federated learning algorithm fed back from the distributed collaborative execution module. The learning results are then fed back to the data acquisition module and the digital twin simulation module.

8. The vacuum-insulated liquid nitrogen tank storage device according to claim 7, characterized in that, A 20-100mm gap is left between the liquid nitrogen cold source jacket (8) and the inner liner (2). Liquid nitrogen is injected into the gap through the inlet (12) to form upper and lower liquid nitrogen protective layers. The liquid nitrogen level is adjusted by the distributed collaborative execution module.

9. The vacuum-insulated liquid nitrogen tank storage device according to claim 8, characterized in that, Liquid nitrogen is added inside the liquid nitrogen tower (10) as an extreme cold source. Liquid nitrogen is injected into the liquid nitrogen cold source jacket (8). Liquid nitrogen is added through the inlet (12). The inlet (12) is controlled by the valve body module (11). The valve body module (11) is driven by the distributed collaborative execution module.

10. An intelligent control system for a vacuum-insulated liquid nitrogen tank storage device, applied to the vacuum-insulated liquid nitrogen tank storage device as described in any one of claims 1 to 9, characterized in that, include: The data acquisition module collects liquid nitrogen level, temperature, pressure, humidity and external environmental data, and transmits them to the edge computing node built into the intelligent control device. The edge computing node performs filtering and normalization preprocessing on the data before transmitting it to the digital twin simulation module. The digital twin simulation module receives data transmitted from edge computing nodes, uses computational fluid dynamics to simulate thermal flow and liquid nitrogen evaporation processes, constructs a virtual model for dynamic simulation, and outputs the simulation results to the intelligent diagnosis and prediction module. The intelligent diagnosis and prediction module receives the simulation results output by the digital twin simulation module, uses a long short-term memory network to process time series data, integrates a support vector machine to detect liquid nitrogen evaporation anomalies, and outputs the prediction results to the adaptive optimization control module. The adaptive optimization control module receives the prediction results output by the intelligent diagnosis and prediction module, uses a non-dominated sorting genetic algorithm and model predictive control to generate an optimized control strategy, and outputs control commands to the distributed cooperative execution module. The distributed collaborative execution module receives the control instructions output by the adaptive optimization control module, drives the valve body module (11) inlet (12) valve to adjust the liquid nitrogen inflow rate through the programmable logic controller, and transmits the execution status feedback to the real-time feedback and online learning module. The real-time feedback and online learning module obtains basic system parameters by monitoring the operating status of the intelligent control device. At the same time, it receives execution feedback transmitted by the distributed collaborative execution module. After fusing the basic system parameters and execution feedback, it uses incremental learning and federated learning algorithms for online learning to optimize the control strategy and generate learning results. The learning results are then fed back to the data acquisition module and the digital twin simulation module.

Images