A deep underground space multi-factor coupling environment simulation experiment device and method

Abstract

This invention discloses a multi-factor coupled environment simulation experimental device and method for deep underground space. The main experimental chamber unit and the buffer chamber unit can be flexibly spliced ​​together according to the experimental scenario to form a material performance experimental module or a human factors engineering experimental module, which is suitable for various experimental scenarios. The buffer chamber unit can isolate the external environment to ensure that the environment inside the main experimental chamber unit is not disturbed. Furthermore, the prediction and control unit constructs a digital twin of the main experimental chamber unit, dynamically restores the multi-field distribution state of the entire domain, predicts the evolution trend and local non-uniformity of the multi-field distribution state of the entire domain, and can reflect the environmental distribution state and uniformity within the main experimental chamber unit in real time. This allows for the correction of the output parameters of the temperature control unit and the gas control unit, ensuring the accuracy and uniformity of the multi-factor coupling of the environment within the main experimental chamber unit, and improving the stability and reliability of the experimental results.

Description

Technical Field

[0001] This invention relates to the field of building engineering technology, specifically to an environmental simulation experimental device for deep underground spaces. Background Technology

[0002] Deep underground space generally refers to usable space in deeper areas below the Earth's surface, such as deep areas starting from 50 meters or even 100 meters below the surface. It is becoming a new territory for national strategic reserves and urban development. Deep underground space generally has comprehensive environmental characteristics such as high temperature, high humidity, micro-pressure, and dynamic changes in gas composition, which directly affect the comfort of people living there and the long-term durability of building structural materials. Conducting research on the laws of human factors engineering and the evolution of material properties in deep environments is the core foundation for safe and livable design of deep underground space.

[0003] Chinese Patent CN 103439246 A discloses a modular multi-factor atmospheric environment simulation test device and method, which can simulate one or more coupled environments of environmental factors such as temperature, humidity, salt spray, light, rain, and polluted atmosphere. Its hardware system includes an environmental factor adjustment system, an air conditioning chamber system, an environmental simulation test chamber system, an overall control system of the device, and a hardware connection system that effectively connects the above four components. It can realize the switching and various combinations between different environmental factor adjustment systems, air conditioning chamber systems, and environmental simulation test chamber systems.

[0004] However, most existing environmental simulation test devices are single-function, fixed-size integrated chambers that can only conduct material performance experiments or human exposure experiments separately. They cannot simultaneously support human factors engineering index testing and research on the evolution of building material performance. As a result, the equipment utilization rate is low and the experimental cost is high. At the same time, the accuracy and uniformity of multi-factor coupling are poor, which cannot match the precise simulation requirements of deep space environment parameters and is prone to experimental data deviation.

[0005] Therefore, how to effectively improve the multi-factor coupling accuracy and uniformity of environmental simulation test devices, and make them applicable to various experimental scenarios, and improve the accuracy and reliability of the results of various experiments in deep underground spaces, has become an urgent problem to be solved in this field. Summary of the Invention

[0006] In view of the shortcomings of the existing technology, the purpose of this invention is to provide a simulation experimental device and method for multi-factor coupling environment in deep underground space that can effectively improve the accuracy and uniformity of multi-factor coupling, is applicable to various experimental scenarios, and is stable and reliable.

[0007] To achieve the above objectives, the present invention provides a multi-factor coupled environment simulation experimental device for deep underground space, comprising a main experimental chamber unit, a buffer chamber unit, a temperature control unit, a gas regulation unit, a prediction and regulation unit, and a centralized control unit. The main experimental cabin unit is configured as a modular structure, and several main experimental cabin units can be spliced ​​together to form a material performance experimental module. The buffer cabin unit can be spliced ​​together with the main experimental cabin unit to form a human factors engineering experimental module. The buffer cabin unit is used for personnel passage and to isolate the external environment from the main experimental cabin unit. The temperature control unit can monitor and adjust the temperature and humidity inside the main experimental chamber unit in real time, and the gas control unit can monitor and adjust the gas pressure and gas concentration inside the main experimental chamber unit in real time. The predictive control unit can construct a digital twin of the main experimental chamber unit, and dynamically restore the global multi-field distribution state of various environmental parameters in the main experimental chamber unit based on the monitoring data of the temperature control unit and the gas control unit. It can also construct a predictive model based on predictive control algorithms to predict the evolution trend and local non-uniformity of the global multi-field distribution state and generate correction signals. The centralized control unit can drive the temperature control unit and the gas regulation unit to operate based on preset deep target environmental parameters, and can also receive the correction signal to dynamically adjust the working state of the temperature control unit and the gas regulation unit.

[0008] Furthermore, both the main experimental cabin unit and the buffer cabin unit are equipped with standardized splicing interfaces for assembly.

[0009] Furthermore, the main experimental cabin unit is configured as a rectangular modular structure and is equipped with a sealed cabin door and a pressure-bearing observation window.

[0010] Furthermore, the buffer chamber unit is equipped with an independent pressure fine-tuning module and a temperature and humidity fine-tuning module, and is configured with a double-layered door that cooperates with the main experimental chamber unit. The pressure fine-tuning module and the temperature and humidity fine-tuning module can adjust the environmental parameters inside the buffer chamber unit to be consistent with those inside the main experimental chamber unit.

[0011] Furthermore, the temperature control unit integrates a cooling module, a heating module, a humidification module, and a dehumidification module, and is equipped with a multi-point temperature and humidity detection device.

[0012] Furthermore, the gas control unit includes an exhaust system, an air compressor unit, and a gas concentration monitoring module; the air compressor unit is used to supply compressed air to the exhaust system; the exhaust system is used to uniformly deliver gas into the main experimental chamber unit; and the gas concentration monitoring module is used to monitor the oxygen and carbon dioxide concentrations in the main experimental chamber unit in real time.

[0013] Furthermore, the prediction and control unit dynamically recreates the global distribution of the three-dimensional temperature field, airflow field, and concentration field within the main experimental chamber unit using a built-in fluid dynamics simulation model.

[0014] Furthermore, the predictive control unit is based on a predictive control algorithm and adopts a model predictive control framework. It uses a reduced-order multi-field model as the basis for prediction, and combines it with an LSTM / Transformer time series model to correct the nonlinear deviations of various environmental parameters. Based on historical environmental parameters and real-time global multi-field distribution characteristics, it predicts parameter drift and local inhomogeneity in advance.

[0015] Furthermore, it also includes a safety protection unit, which includes a pressure protection module, an electrical protection module, and a fire protection module.

[0016] To achieve the above objectives, the present invention provides a multi-factor coupled environment simulation experimental method for deep underground space, based on the aforementioned multi-factor coupled environment simulation experimental device for deep underground space, the experimental method comprising: The main experimental module and the buffer module are spliced ​​together according to the experimental scenario to form a material performance experimental module or a human factors engineering experimental module. The central control unit drives the temperature control unit and the gas regulation unit to operate based on preset deep target environmental parameters. The temperature control unit monitors and adjusts the temperature and humidity in the main experimental chamber unit in real time, and the gas regulation unit monitors and adjusts the gas pressure and gas concentration in the main experimental chamber unit in real time. The predictive control unit constructs a digital twin of the main experimental module, dynamically reconstructs the global multi-field distribution of various environmental parameters within the main experimental module, and predicts the evolution trend and local non-uniformity of the global multi-field distribution, generating correction signals. The central control unit receives correction signals and dynamically adjusts the operating status of the temperature control unit and the gas regulation unit.

[0017] The present invention provides a multi-factor coupled environment simulation experimental device and method for deep underground space. The main experimental chamber unit and the buffer chamber unit can be flexibly spliced ​​together according to the experimental scenario to form a material performance experimental module or a human factors engineering experimental module, which is applicable to a variety of experimental scenarios. The buffer chamber unit can isolate the external environment to ensure that the environment inside the main experimental chamber unit is not disturbed. Furthermore, the prediction and control unit constructs a digital twin of the main experimental chamber unit, dynamically restores the multi-field distribution state of the entire domain, predicts the evolution trend and local non-uniformity of the multi-field distribution state of the entire domain, and can reflect the environmental distribution state and uniformity within the main experimental chamber unit in real time. This allows for the correction of the output parameters of the temperature control unit and the gas control unit, ensuring the accuracy and uniformity of the multi-factor coupling of the environment within the main experimental chamber unit, and improving the stability and reliability of the experimental results. Attached Figure Description

[0018] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0019] Figure 1System block diagram of the deep underground space multi-factor coupled environment simulation experimental device provided by the present invention; Figures 2 to 4 This is a schematic diagram of the structure of the main experimental cabin unit in this invention; Figure 5 This is a system block diagram of the buffer chamber unit in this invention; Figure 6 This is a system block diagram of the temperature control unit in this invention; Figure 7 This is a system block diagram of the gas control unit in this invention; Figure 8 This is a system block diagram of the predictive control unit in this invention; Figure 9 This is a system block diagram of the safety protection unit in this invention.

[0020] Figure label: 1. Main experimental chamber unit; 11. Sealed hatch; 12. Pressure-bearing observation window; 2. Buffer chamber unit; 21. Outer hatch; 22. Inner hatch; 23. Pressure fine-tuning module; 24. Temperature and humidity fine-tuning module; 3. Temperature control unit; 31. Refrigeration module; 32. Heating module; 33. Humidification module; 34. Dehumidification module; 35. Temperature and humidity detection device; 4. Gas control unit; 41. Air compressor unit; 42. Supply and exhaust system; 43. Compressed air filtration system; 44. Inlet and outlet solenoid valves; 45. Gas concentration monitoring module; 5. Predictive control unit; 6. Centralized control unit; 7. Safety protection unit; 71. Pressure protection module; 72. Electrical protection module; 73. Fire protection module. Detailed Implementation

[0021] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below with reference to specific illustrations.

[0022] See Figure 1 The image shows an example of a multi-factor coupled environment simulation experimental device for deep underground space provided by the present invention.

[0023] The deep underground space multi-factor coupled environment simulation experimental device in this example mainly includes the main experimental chamber unit 1, buffer chamber unit 2, temperature control unit 3, gas regulation unit 4, prediction and regulation unit 5, and centralized control unit 6.

[0024] The main experimental cabin unit 1 is configured as a modular structure. Several main experimental cabin units 1 can be spliced ​​together to form a material performance experimental module. The buffer cabin unit 2 can be spliced ​​with the main experimental cabin unit 1 to form a human factors engineering experimental module. The buffer cabin unit 2 is used for personnel passage, isolating the external environment from the main experimental cabin unit 1, so as to be suitable for various experimental scenarios and to ensure that the internal environment of the main experimental cabin unit 1 is not disturbed. Temperature control unit 3 can monitor and adjust the temperature and humidity in the main experimental chamber unit 1 in real time, and gas control unit 4 can monitor and adjust the gas pressure and gas concentration in the main experimental chamber unit 1 in real time. Predictive control unit 5 can construct a digital twin of the main experimental chamber unit 1, and dynamically restore the global multi-field distribution state of various environmental parameters in the main experimental chamber unit 1 based on the monitoring data of temperature control unit 3 and gas control unit 4. It can also construct a prediction model based on predictive control algorithm, predict the evolution trend and local non-uniformity of the global multi-field distribution state, and generate correction signals. The central control unit 6 can drive the temperature control unit 3 and the gas regulation unit 4 to operate based on preset deep target environmental parameters. It can also receive correction signals to dynamically adjust the working state of the temperature control unit 3 and the gas regulation unit 4, so as to correct the output parameters of the temperature control unit 3 and the gas regulation unit 4, ensure the accuracy and uniformity of multi-factor coupling of the environment in the main experimental cabin unit, and improve the stability and reliability of the experimental results.

[0025] Combination Figures 2 to 4 The main experimental cabin unit 1 is configured as a rectangular modular structure, preferably made of high-strength SPCC composite board modular structure, to provide experimental space. The main experimental cabin unit 1 is equipped with a sealed cabin door 11 and a pressure-bearing observation window 12.

[0026] Furthermore, the interior, floor, and lighting of the main experimental unit 1 are all configured to meet the requirements of cleanliness, antibacterial properties, and pressure resistance. Preferably, the interior enclosure is made of ≥50mm thick laboratory-specific antibacterial purification panels, the floor is made of ≥2.0mm thick homogeneous through-core PVC roll material with seamless splicing, and the inside and outside corners are equipped with R=50mm rounded clean aluminum profiles with embedded LED waterproof clean lights to ensure that the main experimental unit 1 provides a clean, antibacterial, and stable experimental space, thereby ensuring the accuracy of experimental results.

[0027] Experimental main cabin unit 1 can be used independently as a single section according to experimental needs. Several experimental main cabin units 1 can also be connected in series to expand the experimental space and form a material performance experimental module for experiments on the evolution of building material performance in deep underground spaces.

[0028] In conjunction with this, the buffer chamber unit 2 can be spliced ​​with the main experimental chamber unit 1 to form a human factors engineering experimental module for human exposure human factors engineering testing experiments in deep underground spaces. The buffer chamber unit 2 serves as a transition chamber for personnel to enter and exit the main experimental chamber unit 1 from the external environment. It can effectively isolate the external environment from the main experimental chamber unit 1, avoid interference to the internal environment of the main experimental chamber unit 1 caused by direct entry from the external environment, and ensure the stability of the experimental space.

[0029] Combination Figure 5Furthermore, the specific structure of the buffer chamber unit 2 is similar to that of the main experimental chamber unit 1. The buffer chamber unit 2 is also equipped with an outer hatch 21 and an inner hatch 22. The outer hatch 21 is coordinated with the external environment, and the inner hatch 22 is coordinated with the sealed hatch 11 of the main experimental chamber unit 1 to form a double-layer hatch, ensuring effective isolation and sealing between the buffer chamber unit 2 and the main experimental chamber unit 1.

[0030] The buffer chamber unit 2 is also equipped with an independent pressure fine-tuning module 23 and a temperature and humidity fine-tuning module 24. The pressure fine-tuning module 23 can adjust the gas pressure inside the buffer chamber unit 2, and the temperature and humidity fine-tuning module 24 can adjust the temperature and humidity inside the buffer chamber unit 2. Only after the environmental parameters inside the buffer chamber unit 2 are consistent with and stable with the main experimental chamber unit 1 can personnel enter the main experimental chamber unit 1 through the buffer chamber unit 2, thereby ensuring the environmental stability inside the main experimental chamber unit 1.

[0031] Here, the pressure fine-tuning module 23 and the temperature and humidity fine-tuning module 24 are conventional technical means in this field, and will not be described in detail here. As an example, the pressure fine-tuning module 23 can be composed of an independent miniature pressure reducing valve and a pressure sensor, and the temperature and humidity fine-tuning module 24 can be composed of an independent heater, a humidifier and a temperature and humidity sensor.

[0032] Both the main experimental chamber unit 1 and the buffer chamber unit 2 are equipped with standardized splicing interfaces. These standardized splicing interfaces include flange-type mechanical connection structures and pressure-bearing sealing rings, enabling several main experimental chamber units 1 to be spliced ​​together, as well as between the main experimental chamber unit 1 and the buffer chamber unit 2, to adapt to various experimental scenarios.

[0033] Furthermore, the main experimental cabin unit 1 is also equipped with a temperature control unit 3 and a gas regulation unit 4. The temperature control unit 3 and the gas regulation unit 4 can monitor and adjust the environmental parameters inside the main experimental cabin unit 1 to achieve environmental regulation of the main experimental cabin unit 1 and simulate the real environment of deep underground space.

[0034] Combination Figure 6 The temperature control unit 3 includes a cooling module 31, a heating module 32, a humidifying module 33, and a dehumidifying module 34. Each module is installed in the main experimental chamber unit 1 and can respectively cool, heat, humidify, and dehumidify the air in the main experimental chamber unit 1 to regulate the temperature and humidity in the main experimental chamber unit 1.

[0035] Furthermore, each module is connected to the central control unit 6, enabling the central control unit 6 to independently control the output parameters of each module based on preset deep target environmental parameters (such as target temperature and humidity) in order to precisely adjust the temperature and humidity inside the main experimental cabin unit 1.

[0036] Furthermore, the temperature control unit 3 also includes a temperature and humidity detection device 35, which is distributed at multiple points in the main experimental chamber unit 1. It can monitor the temperature and humidity values ​​of each area in the main experimental chamber unit 1 in real time, providing accurate monitoring data for the subsequent prediction and control unit 5, so that the prediction and control unit 5 can predict the global field distribution and uniformity of the temperature field in the main experimental chamber unit 1.

[0037] Combination Figure 7 Correspondingly, the gas control unit 4 includes an air compressor unit 41, an exhaust system 42, and a compressed air filtration system 43. The air compressor unit 41 serves as the source of pressure and gas, and its output end is connected to the compressed air filtration system 43. The compressed air filtration system 43 filters and compresses the air output from the air compressor unit 41 to remove odors, dust, and other harmful substances, and then inputs it into the exhaust system 42. The exhaust system 42 distributes the compressed air to various areas within the main experimental cabin unit 1.

[0038] The air supply and exhaust system 42 includes multiple air inlets located on the top and sides of the main experimental chamber unit 1 to form a distributed air intake structure. Each air inlet is equipped with an intake / exhaust solenoid valve 44 and a silencer. The intake branches of each air inlet merge into the main intake path, so that compressed air, after passing through the main intake path and the intake / exhaust solenoid valves 44, simultaneously enters the interior of the main experimental chamber unit 1 from multiple directions and positions through the distributed air inlets. This ensures uniform mixing of the airflow inside the chamber and avoids local parameter differences caused by intake from a single direction.

[0039] At the same time, the silencing device can reduce airflow noise in a timely manner and avoid noise interference with the experimental environment.

[0040] Furthermore, the exhaust system also includes an exhaust pipe, which connects to each intake branch and extends to the outside of the main experimental cabin unit 1, for discharging the gas inside the cabin.

[0041] Furthermore, the inlet and outlet solenoid valves 44 and the air compressor unit 41 are connected to the central control unit 6, enabling the central control unit 6 to control the opening degree of the inlet and outlet solenoid valves 44 and the output parameters of the air compressor unit 41 based on preset deep target environmental parameters (such as target gas concentration and gas flow field), so as to adjust the air intake flow rate and exhaust rate of each air inlet in the main experimental chamber unit 1, and achieve precise maintenance of oxygen concentration and carbon dioxide concentration in the chamber within the target range.

[0042] The gas control unit 4 also includes a gas concentration monitoring module 45, which is distributed at multiple points in the main experimental chamber unit 1. It can monitor the oxygen and carbon dioxide concentrations in each area of ​​the main experimental chamber unit 1 in real time, providing accurate monitoring data for the subsequent prediction and control unit 5. This facilitates the prediction and control unit 5 in predicting the global field distribution and uniformity of the airflow field and concentration field in the main experimental chamber unit 1.

[0043] The temperature control unit 3 and the gas regulation unit 4 thus cooperate with each other to provide a multi-factor coupled environment simulation for the main experimental chamber unit 1, which can efficiently restore the real environment of deep underground space and ensure the reliability of experimental results.

[0044] To improve the accuracy and uniformity of multi-factor coupling within the main experimental chamber unit 1, and to match the precise simulation requirements of deep underground space environmental parameters and reduce experimental data deviation, the predictive control unit 5 can predict the evolution trend, drift, and local non-uniformity of various environmental parameters within the main experimental chamber unit 1 in advance based on digital twin technology and predictive control algorithms. This allows for fine-tuning of the output parameters of the temperature control unit 3 and the gas control unit 4, ensuring the accuracy and uniformity of multi-factor coupling.

[0045] Combination Figure 8 The prediction and control unit 5 first constructs a digital twin of the main experimental chamber unit 1 to map the virtual body of the main experimental chamber unit 1 in real time, and receives the temperature and humidity values ​​of each area in the main experimental chamber unit 1 monitored and acquired by the temperature and humidity detection device 35, as well as the oxygen concentration and carbon dioxide concentration of each area in the main experimental chamber unit 1 monitored and acquired by the gas concentration monitoring module 45. Based on the monitoring data, the full-domain multi-field distribution state of each environmental parameter in the main experimental chamber unit 1 is dynamically restored.

[0046] As an example, the prediction and control unit 5 constructs a digital twin of the main experimental module 1 using the following steps: S11. Geometric and Physical Property Modeling. Based on the actual internal dimensions, enclosure structure layers, and current number of spliced ​​sections of the main experimental chamber unit 1, a 1:1 scale three-dimensional geometric model is established. The sealed chamber 11, pressure observation 12, and the building material placement area or personnel activity area within the main experimental chamber unit 1 are simultaneously recreated in the model. The actual installation positions and air outlet directions of the cooling module 31, heating module 32, humidification module 33, and dehumidification module 34 in the temperature control unit 3 are also simultaneously recreated, as well as the actual spatial coordinates of the air inlets and outlets of the air supply and exhaust system 42 located on the top and sides of the main experimental chamber unit 1. Furthermore, preset physical property parameters such as thermal conductivity, specific heat capacity, density, surface emissivity, and splice seam leakage rate are assigned to each enclosure surface. Simultaneously, preset characteristic parameters such as rated power, rated airflow, and response time constant are assigned to each temperature control and gas regulation device, thus forming a geometrically and physically integrated twin base.

[0047] S12. Grid Discretization and Measurement Point Mapping. The prediction and control unit 5 performs grid discretization on the established three-dimensional geometric model and refines the grid at the near-wall surface, each air inlet, exhaust outlet, and hatch to obtain N computational grid nodes. Then, according to the actual installation coordinates of each measurement point of the temperature and humidity detection device 35 and each measurement point of the gas concentration monitoring module 45 in the main experimental cabin unit 1, each physical measurement point is mapped to the computational grid node with the closest spatial distance to it, and an index table with one-to-one correspondence between physical measurement points and twin nodes is established. Based on this, an observation matrix is ​​formed so that the measured data can be accurately projected to the corresponding spatial position in the twin.

[0048] S13. Construction of Offline Operating Condition Sample Library. Within the design and operation operating condition space of the main experimental unit 1, several representative operating conditions are generated using target temperature, target humidity, target gas pressure, target oxygen concentration, target carbon dioxide concentration, flow distribution ratio of each air inlet, and output power of each module of temperature control unit 3 as design variables. High-fidelity conjugate heat transfer and component transport numerical calculations are performed one by one. Temperature, velocity, humidity, and gas concentration values ​​of all grid nodes at each operating condition and time are collected to form a snapshot matrix, which serves as the physical prior knowledge base of the twin.

[0049] S14. Reduced-order basis function extraction. Prediction and control unit 5 performs eigenorthogonal decomposition on the snapshot matrix, extracts the first k spatial basis functions in descending order of energy proportion, and ensures that the cumulative energy proportion of the first k modes is not less than 99%, thereby compressing the original global multi-field model with N degrees of freedom into a k-dimensional time mode coefficient vector, where k is much smaller than N, thus obtaining a reduced-order multi-field model that can be solved in real time.

[0050] S15. Real-time data access and preprocessing. The prediction and control unit 5 receives the temperature and humidity values ​​of each area in the main experimental chamber unit 1 monitored by the temperature and humidity detection device 35 and the oxygen and carbon dioxide concentrations of each area monitored by the gas concentration monitoring module 45 in real time at a fixed sampling period; the raw data is sequentially time-scale aligned, dimensionally normalized, outlier removed and filtered and smoothed, and neighborhood interpolation and confidence reduction are performed on individual failed or abnormal measurement points to obtain usable observation vectors.

[0051] S16. Modal Coefficient Inversion and Optimal Estimation. On one hand, the spatial basis functions obtained in S14 are projected onto a finite number of measurement points using the observation matrix. The time modal coefficients on the measurement side are then obtained by inversely solving the sparse measurement point values ​​using the least squares method with L2 regularization. The regularization term is used to suppress ill-conditioned solutions and overfitting caused by the limited number of measurement points. On the other hand, the reduced-order multi-field model extrapolates forward based on the state at the previous moment to give the prior predicted values ​​of the time modal coefficients on the model side. Then, Kalman filtering is used to weight and fuse the inverted values ​​on the measurement side and the predicted values ​​on the model side. The weights of the two are automatically assigned according to the measurement noise covariance and the process noise covariance to obtain the optimal estimated values ​​of the time modal coefficients at that moment.

[0052] S17. Global Reconstruction and Residual Self-Verification. The optimally estimated time modal coefficients are linearly superimposed with the corresponding spatial basis functions to obtain the global three-dimensional distribution of temperature, humidity, airflow velocity, oxygen concentration, and carbon dioxide concentration within the main experimental unit 1. The reconstruction results are then substituted back to the locations of each physical measurement point and compared with the measured values ​​to obtain the residuals and root mean square errors. When the residuals exceed a preset threshold, online expansion of basis functions, supplementation of the operating condition sample library, and online identification and correction of uncertain parameters such as convective heat transfer coefficient, wall thermal resistance, and splice seam leakage rate are automatically triggered, enabling the twin to self-evolve with cabin aging, specimen replacement, and changes in the number of splice sections.

[0053] S18. Synchronous Execution and Twin Data Accumulation. S15 to S17 are executed in a rolling fashion at a frequency no greater than the control cycle of the central control unit 6, so that the virtual cabin and the real cabin are synchronized in time and space. At the same time, the geometric property parameters, real-time measurement point data, reconstructed full-field data, control commands of the central control unit 6, and correction signals generated by the predictive control unit 5 are uniformly stored in the twin database for subsequent predictive model training, control effect review, and experimental data traceability.

[0054] Compared to existing environmental simulation test devices that typically rely on readings from a single point or a few sensors and use their average value to characterize the entire cabin environment, failing to obtain the spatial distribution of environmental parameters within the cabin, or that introduce numerical simulations with offline verification during the design phase, each calculation can take several hours to tens of hours and cannot be synchronized with the operation process; this device integrates physical mechanism models, data-driven order reduction, and real-time observation assimilation. It provides physical priors with high-fidelity numerical calculations, compresses the global multi-field to a low-dimensional modal space using intrinsic orthogonal decomposition, making online solutions possible, and then uses sparse measurement point observations to perform optimal estimation of modal coefficients through regularized inversion and Kalman filtering. Residual-driven twin self-correction upgrades the twin from a static mirror to a self-evolving online state observer, resulting in low-dimensional, continuous, and differentiable modal state quantities output by the twin. This provides a computationally achievable state foundation for subsequent fluid dynamics simulation order reduction and predictive control, thereby fundamentally supporting the improvement of the accuracy and uniformity of multi-factor coupling within the main experimental cabin unit 1.

[0055] Furthermore, the prediction and control unit 5 dynamically restores the global distribution of the three-dimensional temperature field, airflow field, and concentration field within the main experimental chamber unit 1 through a built-in fluid dynamics simulation model, so as to achieve real-time synchronization between the virtual chamber and the real chamber of the main experimental chamber unit 1.

[0056] As an example, the fluid dynamics simulation model built into the prediction and control unit 5 dynamically recreates the global distribution of the three-dimensional temperature field, airflow field, and concentration field within the main experimental chamber unit 1 according to the following steps: S21. Configuration of the governing equations. Using the geometric model of the constructed digital twin as the computational domain, the three-dimensional unsteady Reynolds-averaged Navier-Stokes equations, energy equations, component transport equations, and turbulent transport equations are simultaneously established. The component transport equations are established for oxygen, carbon dioxide, and water vapor respectively, and the buoyancy term caused by temperature difference and the gas density change of the main experimental chamber unit 1 under micro-high pressure are taken into account. This allows the temperature field, airflow field, and concentration field to be solved in a coupled manner rather than independently: temperature difference changes air density and drives buoyancy airflow, airflow transport changes the spatial distribution of heat and components, and changes in components and moisture content, in turn, change the specific heat of air and the intensity of convective heat transfer, thus realistically reproducing the coupling relationship between multiple factors in the model.

[0057] S22. Real-time online binding of boundary conditions. Unlike conventional simulations that use fixed boundary conditions, the boundary conditions of this model are dynamically updated by measured data and real-time control commands. The real-time output power and outlet air temperature and humidity of the refrigeration module 31, heating module 32, humidification module 33, and dehumidification module 34 in the temperature control unit 3 are used as internal heat sources and heat flow boundaries. The real-time supply air temperature and humidity of each air inlet of the supply and exhaust system 42, as well as the air flow of each branch calculated from the opening of each inlet and exhaust solenoid valve 44, are used as the mass flow inlet boundary of each air inlet, thus realistically reflecting the distributed multi-point air intake structure in the model. The exhaust port is set as the pressure outlet boundary, and the outlet pressure is taken from the measured chamber pressure of the gas control unit 4. The enclosure surface is set as a third-type boundary that takes into account convection and radiation, and the wall temperature is given in real time by the wall thermal resistance network of the twin combined with the external ambient temperature. The heat, moisture, oxygen consumption, and carbon dioxide production of the specimen or test subject are set as volume source terms according to their actual arrangement.

[0058] S23. Hierarchical Solution of High-Fidelity Offline and Order Reduction Online. In the offline stage, the above equations are solved in high-fidelity transient mode on a fine mesh, covering the design working space and saving a snapshot of the calculation. In the online stage, the control equations are projected onto a low-dimensional subspace spanned by the first k-order spatial basis functions obtained by eigenorthogonal decomposition using Galerkin projection. This transforms the partial differential equations, which originally contained hundreds of thousands to millions of degrees of freedom, into ordinary differential equations containing only k unknowns. The solution time per time step is reduced from the order of hours to the order of milliseconds, thus making the progress of the fluid dynamics simulation faster than the control cycle of the centralized control unit 6.

[0059] S24. Real-time assimilation correction driven by sparse observations. In each sampling period, the measured values ​​of the temperature and humidity detection device 35 and the gas concentration monitoring module 45 are substituted as observations. Kalman filtering or ensemble Kalman filtering is used to jointly estimate and correct the reduced-order state variables, as well as uncertain parameters such as convective heat transfer coefficient, splice seam leakage, and actual temperature and humidity of the supply air. This is to compensate for the reduced-order truncation error, turbulence model error, and boundary condition estimation error, and to prevent the simulation results from cumulatively deviating from the actual cabin state during long-term operation.

[0060] S25. Global Field Reconstruction and Uniformity Feature Extraction. The corrected time modal coefficients are linearly superimposed with the spatial basis functions of temperature, velocity, and concentration to reconstruct the three-dimensional temperature field, three-dimensional airflow field (including velocity vector distribution and turbulent kinetic energy distribution), and three-dimensional concentration field (including oxygen concentration, carbon dioxide concentration, and relative humidity distribution) within the main experimental unit 1. Based on this, the feature quantities characterizing uniformity are further calculated, including the difference between the global maximum and minimum values ​​of each field, spatial standard deviation, maximum value of spatial gradient modulus and its coordinates, volume ratio and spatial location of low-speed recirculation zone and airflow dead zone, air age and ventilation efficiency distribution, etc., thereby quantifying whether the distribution is uniform, where there is non-uniformity, and the degree of non-uniformity into numerical indicators that can be directly used for control.

[0061] S26. Mirror Synchronization and Result Output. The above reconstruction results are presented in real time on the human-machine interface of the central control unit 6 in the form of three-dimensional cloud map, arbitrary cross-sectional slice and isosurface, making the virtual cabin a real-time visualized mirror of the real cabin; at the same time, the reconstructed time modal coefficient vector and uniformity feature quantity are output to the prediction model as input for subsequent evolution trend prediction and local non-uniformity prediction.

[0062] Compared to existing technologies where fluid dynamics simulation is typically used only for offline verification during the device design phase, with boundary conditions set by humans and fixed values, the calculation process is disconnected from actual operation, and the simulation results cannot be used for operational control. Furthermore, temperature, airflow, and concentration are monitored and controlled as independent single fields, ignoring the strong coupling relationship between the three, which easily leads to mutual interference between control loops. This device integrates a fluid dynamics simulation model into the predictive control unit 5, dynamically binding its boundary conditions with measured data and real-time control commands. This transforms the simulation from a design tool into an on-loop state observer during operation. Simultaneously, it employs a layered architecture combining offline high-fidelity simulation with online reduced-order projection, reducing online computation by three to five orders of magnitude and shortening the overall field reconstruction cycle from the control cycle. Furthermore, it continuously assimilates and corrects the reduced-order model using sparse measured data, preserving the speed advantage of reduced-order modeling while suppressing the accuracy loss introduced by it. It outputs a quantized uniformity index based on the simultaneous solution of three coupled fields, rather than individual fields. Isolated solutions enable real-time observation of local hotspots, low-temperature corners, airflow dead zones, carbon dioxide accumulation zones, and humidity stratification within the main experimental unit 1 during operation, rather than post-hoc speculation. This transforms previously unobservable spatial non-uniformity into observable and quantifiable features. The three-field coupled solution avoids mutual cancellation or superposition of the regulatory effects of temperature control unit 3 and gas regulation unit 4, resulting in overshoot. The output spatial distribution characteristics, rather than single-point scalars, provide richer input to the prediction model, enabling subsequent regulation to be implemented according to spatial location differences. This directly improves the accuracy and uniformity of multi-factor coupling within the main experimental unit 1.

[0063] Furthermore, the predictive control unit 5 can also construct a predictive model based on predictive control algorithms to predict the evolution trend and local non-uniformity of the multi-field distribution state across the entire domain and generate correction signals.

[0064] Specifically, the predictive control unit 5 is based on predictive control algorithms and adopts a model predictive control framework. It uses a reduced-order multi-field model as the basis for prediction, and combines it with an LSTM / Transformer time series model to correct the nonlinear deviations of various environmental parameters. Based on historical environmental parameters and real-time global multi-field distribution characteristics, it predicts parameter drift and local inhomogeneity in advance.

[0065] As an example, the prediction and control unit 5 constructs a prediction model according to the following steps, and predicts parameter drift and local inhomogeneities accordingly: S31. Feature Set Construction. Two types of features are extracted from the twin database: The first type is the time series of historical environmental parameters, including the historical readings of each measuring point of the temperature and humidity detection device 35 and the gas concentration monitoring module 45, the historical output power of each module of the temperature control unit 3, the historical opening degree of each inlet and outlet solenoid valve 44, the historical output pressure and pressurization rate of the air compressor unit 41, the gas pressure inside the chamber, as well as the amount of test specimens or the number of test subjects, the opening and closing events of the chamber doors, and disturbances such as the temperature and humidity of the external environment; the second type is the global multi-field distribution state features, including the time modal coefficient vector obtained in the previous steps, the spatial standard deviation and range of each field, the maximum value of the gradient modulus and its coordinates, the proportion of the airflow dead zone volume, and uniformity indicators such as air age statistics. After unifying and normalizing the two types of features, sliding window slicing is performed according to the input time window and the prediction time domain to form the training and inference sample sets.

[0066] S32. Dual-channel prediction model construction. The prediction model adopts a model prediction control framework, which consists of a mechanism channel and a data channel connected in parallel. The mechanistic channel, based on the aforementioned reduced-order multi-field model, extrapolates the time modal coefficients and global multi-field distribution of each step in the prediction time domain within milliseconds, given a future control sequence, thus describing the dominant dynamics of the system. The data channel employs a time-series model composed of a cascaded or parallel LSTM network and a Transformer network. Its input is the historical characteristic time window described in S31, and its output is not the environmental parameters themselves, but rather the residual sequence between the predicted and actual values ​​from the mechanistic channel. This residual sequence is used to correct nonlinear biases that the mechanistic model fails to characterize, such as the hysteresis effect of humid air phase change and moisture absorption / desorption by enclosure materials, the randomness of heat and moisture production during subject metabolism, the dead zone and nonlinear response characteristics of actuators, sensor slow drift, and the time-varying characteristics of seam leakage with pressure and temperature. The LSTM unit excels at capturing short-term inertia and hysteresis, while the Transformer's self-attention mechanism excels at capturing long-term dependencies and intervariate coupling relationships between the temperature, humidity, pressure, and concentration channels. Adding the outputs of the two channels yields the mixed prediction value.

[0067] S33. Training and Online Adaptation. The data channel is trained offline using historical operational data. The samples are divided into training, validation, and test sets according to time sequence. The loss function is based on the mean square error with a weighted term for the uniformity index deviation, so that the network not only focuses on whether the parameter values ​​are accurate, but also on whether the spatial distribution is uniform. After the model is deployed, incremental learning and online fine-tuning are performed using a sliding window, so that the prediction model can continuously adapt to changes in the number of splicing sections of the main experimental module 1, specimen replacement, and equipment aging.

[0068] S34. Parameter Drift Prediction. For each step of the mixed prediction value in the prediction time domain, the deviation sequence between it and the deep target environmental parameters set by the centralized control unit 6 is calculated one by one, and the slope of the deviation is calculated. When the absolute value of the deviation is expected to exceed the allowable control bandwidth (e.g., temperature ±0.3℃, relative humidity ±2%RH, gas pressure ±0.5kPa, oxygen concentration ±0.2%, carbon dioxide concentration ±100ppm), or the deviation slope remains the same sign for several consecutive steps and exceeds the set threshold, it is determined that parameter drift will occur, and the type of drifting parameter, drift direction, expected drift amplitude, and expected time of exceeding the limit are output synchronously, so that the identification is completed several control cycles before the deviation actually occurs.

[0069] S35. Local Non-uniformity Prediction. For the global multi-field distribution reconstructed at each step within the prediction time domain, its uniformity index is calculated. When the predicted temperature range, concentration range, or spatial standard deviation exceeds a set threshold, local non-uniformity is determined to occur. Furthermore, since low-order modes mainly characterize the overall global level and high-order modes mainly characterize the spatial non-uniform structure, by analyzing the energy evolution of the coefficients of each mode in the prediction time domain, the spatial morphology of non-uniformity can be inverted, thereby locating the spatial coordinate range, peak amplitude, and development trend of the non-uniform region. Based on the pre-established spatial adjacency relationship, the region is mapped to the nearest air inlets and temperature control modules, and a quadruple consisting of non-uniformity location, non-uniformity amplitude, development trend, and associated actuators is output.

[0070] S36. Confidence Assessment and Conservatism Adaptation. Confidence intervals for prediction results are given using ensemble prediction or Monte Carlo random deactivation. When the confidence level is lower than a set value, the constraint boundaries of subsequent controls are automatically tightened to avoid making aggressive adjustments when the model uncertainty is high.

[0071] Compared to existing environmental simulation test devices that generally employ setpoint feedback control, where the controller only acts after a deviation has occurred and been captured by the sensor, this approach suffers from significant adjustment lag and overshoot oscillations for controlled objects with large inertia and long time lag, such as large-space cabins. Furthermore, data-driven prediction relies entirely on black-box networks to directly predict environmental parameters, lacking physical constraints, requiring massive amounts of training data, and exhibiting poor sample extrapolation capabilities. This device employs a hybrid prediction architecture with parallel mechanistic and data channels. A reduced-order multi-field model undertakes the dominant dynamics, while a temporal network only handles residual compensation. This reduces the target dimension and amplitude that the network needs to fit, thereby lowering the training data requirements and overfitting risk, while maintaining the good extrapolation reliability of the physical model. Moreover, the prediction object is expanded from a single-point scalar to a global field, making the previously unpredictable local inhomogeneities predictable and locatable. The prediction results not only indicate whether limits have been exceeded but also provide the drift direction, the expected time of exceeding limits, and the coordinates of the inhomogeneous region, providing a direct basis for subsequent directional correction and improving the engineering feasibility and long-term reliability of this device.

[0072] Furthermore, the predictive control unit 5 anticipates parameter drift and local unevenness in advance and generates correction signals. The centralized control unit 6 can fine-tune the output parameters of the temperature control unit 3 and the gas control unit 4 (including the flow distribution of each air inlet, temperature and humidity output power, pressurization rate, valve opening, etc.) based on the correction signals, forming a closed-loop control of "virtual prediction - real control - data feedback - model iteration". Compared with the "setpoint feedback control" in the existing technology, this device solves the problem of poor uniformity of parameters in large space chambers and system response lag from the root through predictive and precise control, and can effectively improve the accuracy and uniformity of multi-factor coupling in the main experimental chamber unit 1.

[0073] As an example, the predictive control unit 5 generates a correction signal based on the prediction result according to the following steps, and the centralized control unit 6 performs fine-tuning control based on the correction signal: S41. Constructing a rolling optimization problem. The predictive control unit 5 establishes an objective function within the prediction and control time domains. This objective function consists of a weighted sum of four terms: the square of the tracking deviation between the predicted values ​​of each environmental parameter and the deep target environmental parameter; the square of the global multi-field uniformity index; the square of the control increment; and the system energy consumption term. Its constraints include the physical and rate boundaries of each actuator, namely, the upper and lower limits of the output power of the cooling module 31 and heating module 32; the upper and lower limits of the humidification and dehumidification capacities of the humidification module 33 and dehumidification module 34; the opening range and maximum rate of change of each inlet and outlet solenoid valve 44; the upper limit of the output pressure of the air compressor unit 41 and the limit of its pressure increase / decrease rate; and the safety boundaries of the cabin gas pressure, oxygen concentration, and carbon dioxide concentration. Since the predictive model has been reduced in order, this optimization problem has a low dimensionality and can be solved within one control cycle using a quadratic programming algorithm.

[0074] S42. Solution and First Step Extraction. Solving the above optimization problem yields the optimal control increment sequence in the control time domain. Following the rolling time domain principle, only the optimal control increment of the first step is taken as the actual correction amount for this cycle, while the remaining steps are only used to ensure the foresight of the optimization. The next cycle is solved again using the latest observation data, thereby realizing rolling optimization and feedback correction.

[0075] S43. Sub-component encoding of correction signals. The above-mentioned optimal control increments are decomposed according to the actuator mapping and encapsulated into structured multi-component correction signals, specifically including: temperature correction component, i.e., the output power increment of the cooling module 31 and the heating module 32; humidity correction component, i.e., the output increment of the humidification module 33 and the dehumidification module 34; airflow and concentration correction component, i.e., the opening increment of the inlet and outlet solenoid valves 44 corresponding to each inlet of the air supply and exhaust system 42, which is given in vector form for each inlet, constituting the differentiated flow distribution between each inlet; pressure correction component, i.e., the output pressure increment and pressurization rate increment of the air compressor unit 41, and the opening increment of the exhaust branch. Each component also carries a timestamp, validity period, priority, and confidence label obtained from S36.

[0076] S44. Targeted Correction for Localized Inhomogeneities. For the inhomogeneous region located in S35, the predictive control unit 5 calls a pre-calibrated influence coefficient matrix. This matrix is ​​obtained offline by applying unit step disturbances to the opening of each air inlet and the output power of each temperature control module on a digital twin, and recording the response amplitude of each spatial partition within the cabin. It is used to quantitatively describe the influence intensity of a change in a certain actuator on a certain spatial partition. By left-multiplying the inhomogeneity to be eliminated by the generalized inverse of this matrix, the differentiated increments to be allocated to each relevant actuator can be solved, thereby achieving targeted correction where the inhomogeneity is, rather than uniformly raising and lowering the output of the entire cabin.

[0077] S45. Safety Verification and Strategy Overlay of Central Control Unit 6. After receiving the correction signal, Central Control Unit 6 first performs a safety verification: verifying whether the corrected target value exceeds the safety boundary, whether the rate of change exceeds the limit, and whether it conflicts with the interlocking conditions of Safety Protection Unit 7. If the limit is exceeded, saturation limiting is applied according to the boundary and an alarm is issued. After the verification is passed, the correction signal is overlaid with the reference control quantity generated by it based on the preset deep target environment parameters to form a composite control strategy that combines reference feedforward, predictive feedforward, and bottom-level feedback. The reference control quantity maintains the steady-state operating point, the correction signal provides advance compensation based on prediction, and the local proportional-integral control loop of each actuator retains the immediate feedback backup for sudden disturbances. The scope of action of the three does not conflict with each other and the priority is clear.

[0078] S46. Command Issuance and Closed-Loop Iteration. The centralized control unit 6 issues the superimposed control quantities in separate channels: temperature and humidity components are sent to the power regulators or frequency converters of the refrigeration module 31, heating module 32, humidification module 33, and dehumidification module 34 in the temperature control unit 3; the opening component is sent port by port to the inlet and outlet solenoid valves 44 at each air inlet of the air supply and exhaust system 42; and the pressure and rate components are sent to the frequency converter and pressure setting loop of the air compressor unit 41. After execution, the new data collected by the temperature and humidity detection device 35 and the gas concentration monitoring module 45 are sent back to the prediction and control unit 5 to update the state of the digital twin, correct the fluid dynamics simulation model, and fine-tune the prediction model parameters, and then enter the next round of rolling, thus forming a closed loop of virtual prediction, real control, data feedback, and model iteration.

[0079] S47. Graded Response Mechanism. The centralized control unit 6 implements a graded response based on the severity of the predicted deviation and the proximity of the expected over-limit time: when the deviation is minor and the expected over-limit time is far away, it only adjusts the opening degree of the inlet and outlet solenoid valves 44 and other actuators with low inertia and fast response to complete the fine adjustment with minimal energy cost; when the deviation is large, it further adjusts the output power of each module of the temperature control unit 3 and the output of the air compressor unit 41; when the predicted result touches the safety boundary, it is upgraded to the safety mode, and the centralized control unit 6 links the safety protection unit 7 to act and alarm.

[0080] Compared with the "setpoint feedback control" in existing technologies, this device solves the problems of poor uniformity of parameters and system response lag in large space cabins from the root through predictive precision control, and can effectively improve the accuracy and uniformity of multi-factor coupling in the main experimental cabin unit 1.

[0081] Combination Figure 9 In some embodiments, the experimental apparatus further includes a safety protection unit 7, which includes a pressure protection module 71, an electrical protection module 72, and a fire protection module 73.

[0082] The pressure protection module 71 constructs a four-level progressive protection system: The first level is the software-based pressure limiting and normal regulation implemented by the centralized control unit 6 based on the dual pressure detection systems (electronic pressure transmitters and independent mechanical pressure gauges are set independently); the second level is a dual-path automatic overpressure relief device, which consists of two independent pressure taps, pressure transmitters, control circuits, and fail-safe relief valves (automatically open in case of power failure or gas failure). Path A and Path B serve as backups for each other. When the chamber pressure reaches the activation value, the device automatically opens to release pressure, and in case of a single-path failure, the other path automatically engages, achieving redundant operation; the third level is an automatic mechanical pressure relief safety valve, which operates based on the force balance principle of the gas pressure inside the chamber and the spring preload, without relying on any power source, gas source, or control system. When the chamber pressure exceeds the limit, the valve disc automatically opens to release pressure, forming a purely mechanical protection barrier; the fourth level is a dual-path manual emergency pressure relief device, with independent mechanical lever valves installed inside and outside the chamber. Either device can be operated independently, achieving emergency pressure relief completely independent of electricity and automatic systems. The four-level protection system progressively reduces reliance on external energy and progressively expands the fault coverage, ensuring that subsequent protection systems can still operate reliably even if any link fails.

[0083] Furthermore, the electrical protection module 72 includes an all-around equipotential bonding system, an automatic short-circuit protection system, and an automatic leakage protection system. The equipotential bonding system uses the metal cabin as a reference body, and connects all exposed conductive parts, such as the cabin shell, metal casings of various equipment, pipes, and metal door frames, to the main equipotential grounding terminal block in a radial manner via a dedicated grounding conductor. This short-circuits all simultaneously accessible conductive parts to the same potential, limiting the human contact voltage to below a safe value. The automatic short-circuit protection system is equipped with circuit breakers at three levels: the main incoming line, branch circuits, and terminal circuits. Each level is selectively set to ensure that only the faulty branch circuit is disconnected in the event of a fault, without causing a complete loss of power to the entire cabin. The automatic leakage protection system is equipped with high-sensitivity instantaneous residual current devices (rated operating current not exceeding 30mA) in the cabin sockets and equipment circuits, and time-delayed protectors in the power distribution mains, forming a tiered coordination. Simultaneously, the cabin prioritizes the use of extra-low voltage power supply to reduce the risk of electric shock and ignition from the source.

[0084] Meanwhile, the fire-fighting module 73 uses B1-grade fire-retardant materials for the decorative components inside the main experimental cabin unit 1, combined with a high-strength cabin structure, to suppress the spread of fire.

[0085] The alarm and trip signals of the pressure protection module 71 and the electrical protection module 72 are connected to the centralized control unit 6. In the event of an anomaly, the centralized control unit 6 automatically switches to a safe mode, maintaining emergency lighting and emergency communication, and guiding personnel to evacuate in an orderly manner. The fire detection signal of the fire protection module 73 is connected to the power distribution circuit via a hard interlock, automatically cutting off non-fire protection power in the event of a fire. Each protection module maintains physical independence in terms of pressure taps, pipelines, power supply, and signal channels to avoid common-cause failures, providing complete intrinsic safety assurance for simulation experiments of multi-factor coupled environments in deep underground spaces.

[0086] This constitutes the deep underground space multi-factor coupled environment simulation experimental device provided by the present invention.

[0087] This invention also provides a method for simulating the multi-factor coupled environment of deep underground space, based on a simulation device for the multi-factor coupled environment of deep underground space constructed by the above scheme. The method includes: The main experimental module 1 and the buffer module 2 are spliced ​​together according to the experimental scenario to form a material performance experimental module or a human factors engineering experimental module. The central control unit 6 drives the temperature control unit 3 and the gas regulation unit 4 to operate based on preset deep target environmental parameters. The temperature control unit 3 monitors and adjusts the temperature and humidity in the main experimental chamber unit 1 in real time, and the gas regulation unit 4 monitors and adjusts the gas pressure and gas concentration in the main experimental chamber unit 1 in real time. The prediction and control unit 5 constructs a digital twin of the main experimental module 1, dynamically restores the global multi-field distribution state of various environmental parameters in the main experimental module 1, and predicts the evolution trend and local non-uniformity of the global multi-field distribution state, generating correction signals. The central control unit 6 receives correction signals and dynamically adjusts the working status of the temperature control unit 3 and the gas regulation unit 4.

[0088] The present invention provides a multi-factor coupled environment simulation experimental device and method for deep underground space. The main experimental chamber unit and the buffer chamber unit can be flexibly spliced ​​together according to the experimental scenario to form a material performance experimental module or a human factors engineering experimental module, which is applicable to a variety of experimental scenarios. The buffer chamber unit can isolate the external environment to ensure that the environment inside the main experimental chamber unit is not disturbed. Furthermore, the prediction and control unit constructs a digital twin of the main experimental chamber unit, dynamically restores the multi-field distribution state of the entire domain, predicts the evolution trend and local non-uniformity of the multi-field distribution state of the entire domain, and can reflect the environmental distribution state and uniformity within the main experimental chamber unit in real time. This allows for the correction of the output parameters of the temperature control unit and the gas control unit, ensuring the accuracy and uniformity of the multi-factor coupling of the environment within the main experimental chamber unit, and improving the stability and reliability of the experimental results.

[0089] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A multi-factor coupled environment simulation experimental device for deep underground space, characterized in that, It includes the main experimental cabin unit, buffer cabin unit, temperature control unit, gas regulation unit, predictive regulation unit, and centralized control unit. The main experimental cabin unit is configured as a modular structure, and several main experimental cabin units can be spliced ​​together to form a material performance experimental module. The buffer cabin unit can be spliced ​​together with the main experimental cabin unit to form a human factors engineering experimental module. The buffer cabin unit is used for personnel passage and to isolate the external environment from the main experimental cabin unit. The temperature control unit can monitor and adjust the temperature and humidity inside the main experimental chamber unit in real time, and the gas control unit can monitor and adjust the gas pressure and gas concentration inside the main experimental chamber unit in real time. The predictive control unit can construct a digital twin of the main experimental chamber unit, and dynamically restore the global multi-field distribution state of various environmental parameters in the main experimental chamber unit based on the monitoring data of the temperature control unit and the gas control unit. It can also construct a predictive model based on predictive control algorithms to predict the evolution trend and local non-uniformity of the global multi-field distribution state and generate correction signals. The centralized control unit can drive the temperature control unit and the gas regulation unit to operate based on preset deep target environmental parameters, and can also receive the correction signal to dynamically adjust the working state of the temperature control unit and the gas regulation unit.

2. The deep underground space multi-factor coupled environment simulation experimental device according to claim 1, characterized in that, Both the main experimental cabin unit and the buffer cabin unit are equipped with standardized splicing interfaces for assembly.

3. The deep underground space multi-factor coupled environment simulation experimental device according to claim 1, characterized in that, The main experimental cabin unit is configured as a rectangular modular structure and is equipped with a sealed cabin door and a pressure-bearing observation window.

4. The deep underground space multi-factor coupled environment simulation experimental device according to claim 1, characterized in that, The buffer chamber unit is equipped with an independent pressure fine-tuning module and a temperature and humidity fine-tuning module, and is also equipped with a double-layer door that works in conjunction with the main experimental chamber unit. The pressure fine-tuning module and the temperature and humidity fine-tuning module can adjust the environmental parameters inside the buffer chamber unit to be consistent with those inside the main experimental chamber unit.

5. The deep underground space multi-factor coupled environment simulation experimental device according to claim 1, characterized in that, The temperature control unit integrates a cooling module, a heating module, a humidification module, and a dehumidification module, and is equipped with a multi-point temperature and humidity detection device.

6. The deep underground space multi-factor coupled environment simulation experimental device according to claim 1, characterized in that, The gas control unit includes an exhaust system, an air compressor unit, and a gas concentration monitoring module; the air compressor unit is used to supply compressed air to the exhaust system; the exhaust system is used to uniformly deliver gas into the main experimental chamber unit; and the gas concentration monitoring module is used to monitor the oxygen and carbon dioxide concentrations in the main experimental chamber unit in real time.

7. The deep underground space multi-factor coupled environment simulation experimental device according to claim 1, characterized in that, The prediction and control unit dynamically recreates the global distribution of the three-dimensional temperature field, airflow field, and concentration field within the main experimental chamber unit using a built-in fluid dynamics simulation model.

8. The deep underground space multi-factor coupled environment simulation experimental device according to claim 1, characterized in that, The predictive control unit is based on a predictive control algorithm and adopts a model predictive control framework. It uses a reduced-order multi-field model as the basis for prediction, and combines it with an LSTM / Transformer time series model to correct the nonlinear deviations of various environmental parameters. Based on historical environmental parameters and real-time global multi-field distribution characteristics, it predicts parameter drift and local inhomogeneity in advance.

9. The deep underground space multi-factor coupled environment simulation experimental device according to claim 1, characterized in that, It also includes a safety protection unit, which includes a pressure protection module, an electrical protection module, and a fire protection module.

10. A multi-factor coupled environment simulation experimental method for deep underground space, characterized in that, Based on the deep underground space multi-factor coupled environment simulation experimental device according to any one of claims 1-9, the experimental method includes: The main experimental module and the buffer module are spliced ​​together according to the experimental scenario to form a material performance experimental module or a human factors engineering experimental module. The central control unit drives the temperature control unit and the gas regulation unit to operate based on preset deep target environmental parameters. The temperature control unit monitors and adjusts the temperature and humidity in the main experimental chamber unit in real time, and the gas regulation unit monitors and adjusts the gas pressure and gas concentration in the main experimental chamber unit in real time. The predictive control unit constructs a digital twin of the main experimental module, dynamically reconstructs the global multi-field distribution of various environmental parameters within the main experimental module, and predicts the evolution trend and local non-uniformity of the global multi-field distribution, generating correction signals. The central control unit receives correction signals and dynamically adjusts the operating status of the temperature control unit and the gas regulation unit.

Images