PID (Proportion Integration Differentiation) control-based two-phase immersion liquid-cooled gas phase pressure regulation method and system

Abstract

The invention discloses a two-phase immersion liquid cooling gas-phase pressure adjusting method and system based on PID control, and relates to the technical field of liquid cooling gas pressure adjustment. The PID control-based two-phase immersion liquid cooling gas phase pressure regulation method comprises the following steps: S1, acquiring and preprocessing environment temperature data, pressure and volume data and execution feedback data in a liquid cooling gas pressure regulation process, and constructing a standardized gas-liquid control data set; s2, the gas phase space pressure intensity state is evaluated, and the inflation and deflation operation force of the elastic air bag is dynamically adjusted; s3, the air pressure adjusting trend is analyzed, and the air inlet and outlet rate of the two-way pump is adjusted in real time; and S4, comprehensively analyzing the current closed gas phase space state, and driving the volume adjustment instruction to adjust the amplitude. The problem that accurate matching control of evaporation and condensation processes is difficult to support due to the fact that a target pressure value lacks a joint matching mechanism for altitude change and load dynamics is solved.

Description

A Method and System for Two-Phase Immersion Liquid-Cooled Gas Phase Pressure Regulation Based on PID Control Technical Field

[0001] This invention relates to the field of liquid cooling gas pressure regulation technology, specifically to a two-phase immersion liquid cooling gas phase pressure regulation method and system based on PID control. Background Technology

[0002] With the widespread application of two-phase immersion liquid cooling in the thermal management of high heat flux devices, intelligent control modes with continuous response capabilities and adaptive regulation logic have gradually developed around the dynamic pressure regulation of the gas phase space. In a typical two-phase immersion liquid cooling structure, the working liquid coats the surface of the high-heat-generating unit. The heat drives the liquid to vaporize and form a stable gas-liquid coexistence interface, generating considerable gas phase volume changes, which in turn lead to periodic fluctuations in the pressure within the gas cavity.

[0003] For example, invention patent CN114545856B discloses a mechanical self-resetting air pressure detection and control device and method, relating to the field of air pressure transmission technology. Its key technical points are: it also includes a processor and a PLC controller; the side wall of the main air pressure pipe is connected to an adjusting cylinder, and an adjusting plug is movably connected inside the adjusting cylinder in a sealed manner; the adjusting cylinder is equipped with a driving component for driving the adjusting plug to move along the axis of the adjusting cylinder; an air pressure sensor is installed inside the main air pressure pipe and / or the adjusting cylinder; the output end of the air pressure sensor is connected to the input end of the processor, the output end of the processor is connected to the input end of the PLC controller, and the output end of the PLC controller is connected to the input end of the driving component; the processor is electrically connected to the air source device. This invention maintains a relatively stable air pressure inside the main air pressure pipe by compressing or diffusing gas, eliminating the need for frequent adjustments to the air source device, effectively ensuring the stability of the entire system, and reducing the frequency of air pressure detection.

[0004] For example, the invention patent with announcement number CN115390496B provides a multi-mode air pressure control device and control method based on the EtherCAT bus interface. The invention is connected through a master station control module, a slave station communication module, a controller module, an air source module and an air pressure output module. The air pressure output module is configured to adjust and deliver positive and / or negative output air pressure, thereby achieving the purpose of multi-mode air pressure control with rapid air pressure adjustment and precise air pressure adjustment switching, and thus improving the air pressure delivery efficiency and delivery accuracy.

[0005] However, existing liquid cooling pressure regulation schemes mostly employ setpoint control based on empirical rules or segmented logic-driven methods, lacking real-time feedback modeling and continuous dynamic control capabilities for the gas-liquid interaction mechanism. Especially when factors such as condenser lag, evaporator asymmetric heat transfer, and bidirectional pump inertia limitations exist, pressure regulation response lag, valve control oscillations, or localized overpressure within the cavity can easily occur, affecting the system's heat exchange efficiency and equipment safety boundaries. Furthermore, since the dynamic changes in gas phase pressure are simultaneously controlled by the asynchronous evolution of the coolant evaporation rate, the cavity free space compressibility, and the condenser-side exhaust capacity, a single proportional or fixed gain control strategy cannot achieve dual correction for instantaneous pressure deviations and cumulative errors. This results in problems such as regulation lag and large overshoot when facing rapid disturbances or conditions with multiple superimposed disturbances.

[0006] To address the above issues, there is an urgent need for a two-phase immersion liquid-cooled gas phase pressure regulation method and system based on PID control. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides a two-phase immersion liquid-cooled gas phase pressure regulation method and system based on PID control. This solves the problem that the target pressure value lacks a joint adaptation mechanism for altitude changes and load dynamics, making it difficult to support precise matching control of the evaporation and condensation processes.

[0008] Technical solution

[0009] To achieve the above objectives, the present invention provides the following technical solution: a two-phase immersion liquid-cooled gas phase pressure regulation method and system based on PID control, comprising: S1, collecting ambient temperature data, pressure and volume data, and execution feedback data during the liquid-cooled gas pressure regulation process, and preprocessing the collected ambient temperature data, pressure and volume data, and execution feedback data to construct a standardized gas-liquid control dataset; S2, evaluating the current gas phase space pressure state based on the standardized gas-liquid control dataset, and dynamically adjusting the inflation and deflation force of the elastic airbag based on the evaluation results; S3, analyzing the gas pressure regulation trend based on the standardized gas-liquid control dataset and combined with the environmental heat exchange characteristics, and adjusting the bidirectional pump inlet and outlet rates in real time based on the analysis results; S4, comprehensively analyzing the current closed gas phase space state using the gas phase space pressure state evaluation results and the gas pressure regulation trend analysis results as input, and driving the volume regulation command amplitude adjustment based on the analysis results.

[0010] Furthermore, the specific steps for collecting ambient temperature data, pressure and volume data, and execution feedback data during the liquid cooling pressure regulation process are as follows: Ambient temperature data is collected using high-precision temperature sensors deployed on the inner wall of the elastic airbag and along each air inlet and outlet path. This ambient temperature data includes: the altitude of the current deployment environment, the coolant temperature at the radiator inlet, the coolant temperature at the radiator outlet, the external ambient temperature, and the gas temperature inside the elastic airbag. Pressure and volume data are collected using pressure sensors. This pressure and volume data includes: the instantaneous pressure value of the gas phase region of the pressure sensor, the volume of the elastic airbag, and the average gas phase pressure value is calculated and recorded using the instantaneous pressure values ​​of each gas phase region. Execution feedback data is collected using a linkage control command recording device. This execution feedback data includes: the instantaneous power of the server, the bidirectional pump speed, and the real-time height of the coolant level.

[0011] Furthermore, the specific steps for preprocessing the collected ambient temperature data, pressure and volume data, and execution feedback data to construct a standardized gas-liquid control dataset are as follows: Multi-point temperature synchronization correction is performed on the ambient temperature data. For temperature values ​​collected from the inner wall of the elastic airbag, the air inlet and outlet paths, the radiator inlet and outlet, and the external environment, sampling delays between different sensor channels are eliminated through timestamp alignment. Simultaneously, atmospheric pressure influence correction is performed by combining altitude values ​​to ensure consistent temperature data representation under different altitude environments. Noise removal and mapping integration are performed on the pressure and volume data. The instantaneous pressure values ​​of the collected gas phase region need to be filtered using sliding window averaging and median filtering to remove abnormal fluctuations and short-term spikes, ensuring input stability. Temporal feature extraction is performed on the execution feedback data. The collected instantaneous server power, bidirectional pump speed, and coolant level values ​​are segmented and trend slopes are calculated. The standardized ambient temperature data, pressure and volume data, and execution feedback data are normalized, mapped to a unified dimension range, and stored to construct the standardized gas-liquid control dataset.

[0012] Furthermore, the specific steps for evaluating the current gas phase space pressure state based on the standardized gas-liquid control dataset are as follows: multiply the instantaneous pressure values ​​of the gas phase region of each pressure sensor by the corresponding heat exchange capacity factor and sum them to obtain the weighted gas phase pressure total value; add the instantaneous power of the server to the altitude value of the current deployment environment and then subtract the heat exchange capacity factor to obtain the basic load assessment value; subtract the basic load assessment value from the weighted gas phase pressure total value to obtain the pressure deviation input value.

[0013] Furthermore, the specific steps for dynamically adjusting the inflation and deflation of the elastic airbag based on the evaluation results are as follows: Real-time comparison of the pressure deviation input value of the current gas phase region with the dynamic balance threshold: When the pressure deviation input value is less than or equal to the dynamic balance threshold, it is determined to be a stable range with good evaporation-condensation rate matching. The current working state of the bidirectional pump and the elastic airbag is maintained, without triggering additional gas inflation and deflation adjustments. Only the PID closed loop continuously corrects the pressure fluctuations slightly, while monitoring server power changes and condensation capacity trends to prevent threshold drift. When the pressure deviation input value is greater than the dynamic balance threshold, it is determined to be a pressure imbalance range. The execution path is selected based on the deviation sign and rate of change: When the deviation direction is negative, the gas collection tank is driven to fill the elastic airbag with inert gas, causing the elastic airbag to expand to increase the cavity pressure and suppress excessive evaporation. When the deviation direction is positive, the elastic airbag is controlled to discharge gas back to the gas collection tank, causing the elastic airbag to contract to release cavity space, reduce pressure, and promote phase change heat dissipation.

[0014] Furthermore, the specific steps for analyzing the air pressure regulation trend based on the standardized gas-liquid control dataset and combined with the environmental heat exchange characteristics are as follows: Calculate the difference between the coolant temperature at the radiator inlet and the coolant temperature at the outlet, add the instantaneous power of the server minus the bidirectional pump speed to obtain the net load-driven temperature difference; calculate the absolute value of the difference between the instantaneous pressure value and the average gas phase pressure value in the gas phase region, divide it by the average gas phase pressure value, multiply it by the real-time height of the coolant level, and take the logarithm to obtain the pressure path disturbance value; divide the net load-driven temperature difference by the external ambient temperature and multiply it by the pressure path disturbance value to obtain the gas phase regulation driving value.

[0015] Furthermore, the specific steps for adjusting the bidirectional pump's intake and exhaust rates in real time based on the analysis results are as follows: The calculation results of the current gas phase adjustment drive value are collected in real time to directly drive the bidirectional pump to achieve filling and exhaust control: When the gas phase adjustment drive value is positive, and the gas phase adjustment drive value generated by the change in gas phase state in two consecutive iterations shows a continuous increase, the bidirectional pump is automatically started to inject gas. Simultaneously, without causing gas turbulence, the upper limit of the injection rate is increased to rapidly improve the overall pressure response capability of the sealed chamber. At the same time, the pressure rise rate change during the injection process is recorded synchronously, and the bidirectional pump speed increase strategy is adjusted in real time; when the adjustment drive value is negative... The exhaust control process is triggered to achieve rapid depressurization by reducing the inlet pressure of the bidirectional pump. The liquid level recovery rate during the depressurization phase is correlated with the top vapor expansion margin to determine whether micro-liquid compensation injection needs to be triggered simultaneously. At the same time, the change trend of elastic bladder deformation during the critical exhaust process is used as feedback input to determine whether the exhaust has achieved the expected effect. If necessary, the switch to pressure stabilization mode is delayed. When the gas phase adjustment drive value is zero and the gas phase adjustment drive value generated by the change of gas phase state does not change between the two consecutive times, only the bidirectional pump is maintained to start and stop intermittently, and the high-frequency data recording stream is turned off and low-frequency monitoring is switched to. Furthermore, the specific steps for comprehensively analyzing the current closed gas phase space state using the gas phase space pressure state assessment results and gas pressure regulation trend analysis results as inputs are as follows: obtain the pressure deviation input value, subtract the volume of the elastic airbag from it, and then multiply it by the pressure deviation input value to obtain the elastic modulation pressure difference value; use the elastic modulation pressure difference value as the numerator, and take the absolute value of the gas temperature inside the elastic airbag plus the logarithm as the denominator to obtain the elastic correction ratio value; add one to the gas phase regulation drive value and then multiply it by the elastic correction ratio to obtain the elastic linkage output value.

[0016] Furthermore, the specific steps for adjusting the volume adjustment command amplitude based on the analysis results are as follows: Real-time comparison of the elastic linkage output value at the current measuring point with the elastic output evaluation threshold. The elastic output evaluation threshold is divided into a first output threshold and a second output threshold. When the elastic linkage output value is less than or equal to the second output threshold, the PID controller automatically starts steady-state observation, continuously sampling data only from the pressure sensor and the internal structure changes of the elastic airbag, and calculating the mean and standard deviation of the dynamic offset using a sliding window method. When the elastic linkage output value generated by the combined pressure deviation input value and the gas phase adjustment driving value for five consecutive times maintains a consistent trend direction, the PID controller pauses gas flow control and keeps the bidirectional pump in low-power standby mode. When the elastic linkage output value is greater than the second output threshold and less than or equal to the first output threshold: the PID controller automatically triggers gradual adjustment, and adjusts the unilateral micro-inflation and deflation operation of the elastic airbag according to the coordinated change trend of the real-time pressure deviation input value and the gas phase adjustment drive value, while reducing the bidirectional pump speed; when the elastic linkage output value is greater than the first output threshold: the PID controller immediately activates the high-priority linkage path and activates the elastic airbag emergency release and multi-channel sensor feedback mechanism: including high-flow bidirectional pump start-up, short-time high-frequency inflation and deflation switching and bypass channel redundancy coordinated activation, to complete the rapid correction of pressure deviation in the shortest time, while monitoring the execution channel temperature rise rate, elastic airbag deformation rate and response consistency in real time.

[0017] The second aspect of this invention provides a two-phase immersion liquid-cooled gas-phase pressure regulation system based on PID control, comprising: a multi-parameter dynamic data extraction module, used to collect ambient temperature data, pressure and volume data, and execution feedback data during the liquid-cooled gas pressure regulation process, and to preprocess the collected ambient temperature data, pressure and volume data, and execution feedback data to construct a standardized gas-liquid control dataset; a real-time pressure sensing module for the gas phase region, used to evaluate the current gas phase space pressure state based on the standardized gas-liquid control dataset, and to dynamically adjust the inflation and deflation force of the elastic airbag based on the evaluation results; a PID closed-loop control regulation module, used to analyze the gas pressure regulation trend based on the standardized gas-liquid control dataset and combined with the environmental heat exchange characteristics, and to adjust the bidirectional pump inlet and outlet rates in real time based on the analysis results; and an elastic pressure regulation linkage control module, used to comprehensively analyze the current closed gas phase space state with the gas phase space pressure state evaluation results and the gas pressure regulation trend analysis results as inputs, and to drive the volume regulation command amplitude adjustment based on the analysis results.

[0018] Beneficial effects The present invention has the following beneficial effects: (1) The two-phase immersion liquid cooling gas phase pressure regulation method and system based on PID control, by integrating the gas phase pressure deviation value, elastic air bag volume change and gas temperature feedback to construct a multi-parameter closed-loop control path, realizes high-frequency dynamic response to pressure fluctuations in the evaporation area, and significantly improves the system's ability to quickly adapt to sudden thermal disturbances.

[0019] (2) The two-phase immersion liquid-cooled gas phase pressure regulation method and system based on PID control effectively balances the rhythm control and regulation amplitude of the charging and discharging action by introducing a flexible buffer structure and a bidirectional pump coordination mechanism, combined with the real-time adjustment logic of PID parameters, thereby reducing the risk of structural fatigue caused by high-frequency fluctuations.

[0020] (3) The two-phase immersion liquid-cooled gas phase pressure regulation method and system based on PID control, through multi-level output thresholds, triggers observation, slow adjustment and emergency control modes in stages, avoiding the problem of decreased thermal stability caused by adjustment lag or over-adjustment in conventional systems, and improving the intelligent coordination of the regulation path.

[0021] (4) The two-phase immersion liquid-cooled gas phase pressure regulation method and system based on PID control, by real-time calculation and component decoupling of the gas phase regulation driving value, takes into account the influence of factors such as temperature rise rate, condensation path and external disturbance, realizes accurate correction of the pressure evolution trend of the sealed gas cavity, and enhances the multi-condition adaptability of the regulation system.

[0022] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0023] Figure 1 is a flowchart of the two-phase immersion liquid-cooled gas phase pressure regulation method based on PID control according to the present invention; Figure 2 is a structural diagram of the two-phase immersion liquid-cooled gas phase pressure regulation system based on PID control according to the present invention; Figure 3 is a line graph of the elastic linkage output value involved in the present invention; Figure 4 is a schematic diagram of the PID control process involved in the present invention; Figure 5 is a schematic diagram of the system principle involved in the present invention; In the figures, 1, sealed box; 2, pressure sensor; 3, radiator; 4, server; 5, cooling coil; 6, PID controller; 7, bidirectional pump; 8, gas collection tank; 9, elastic airbag. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Please refer to Figures 1-5. This embodiment of the invention provides a technical solution: a two-phase immersion liquid-cooled gas-phase pressure regulation method and system based on PID control, comprising: S1, collecting ambient temperature data, pressure and volume data, and execution feedback data during the liquid-cooled gas pressure regulation process, and preprocessing the collected ambient temperature data, pressure and volume data, and execution feedback data to construct a standardized gas-liquid control dataset; S2, evaluating the current gas phase space pressure state based on the standardized gas-liquid control dataset, and dynamically adjusting the inflation and deflation force of the elastic airbag 9 based on the evaluation results; S3, analyzing the gas pressure regulation trend based on the standardized gas-liquid control dataset and combined with the environmental heat exchange characteristics, and adjusting the intake and exhaust rates of the bidirectional pump 7 in real time based on the analysis results; S4, comprehensively analyzing the current closed gas phase space state using the gas phase space pressure state evaluation results and the gas pressure regulation trend analysis results as input, and driving the volume regulation command amplitude adjustment based on the analysis results.

[0026] Specifically, the steps for collecting ambient temperature data, pressure and volume data, and execution feedback data during the liquid-cooled gas pressure regulation process are as follows: In the two-phase immersion liquid-cooled gas phase pressure regulation system based on PID control, the dynamic construction and control input support for the gas phase regulation drive logic are completed through multi-dimensional data acquisition and fusion processing of key environmental variables and structural response information. Specifically, the temperature data acquisition is accomplished by high-precision temperature sensors deployed on the inner wall of the elastic airbag 9 and along each air inlet and outlet path. The acquired data includes the current altitude of the deployment environment, the coolant temperature at the inlet of the radiator 3, the coolant temperature at the outlet of the radiator 3, the external ambient temperature, and the internal gas temperature of the elastic airbag 9, providing fundamental support for the analysis of temperature gradient trends and the evaluation of heat exchange efficiency. Pressure and volume data are acquired by pressure sensor 2, including the instantaneous pressure value of the gas phase region obtained by pressure sensor 2 and the current volume of the elastic airbag 9. At the same time, the instantaneous values ​​of each pressure sampling point are integrated and calculated to form an average gas phase pressure value, which is used to characterize the pressure state and fluctuation degree of the overall gas phase region. In addition, the execution feedback information is acquired in real time through the linkage control command recording device, including the current instantaneous power index of the server 4, the rotation speed parameters of the bidirectional pump 7, and the real-time height value of the coolant level. These feedback data are used together to describe the system's thermal load state, coolant supply circulation intensity, and dynamic changes in the liquid level.

[0027] This implementation plan achieves comprehensive perception and unified acquisition of multiple key operating parameters involved in the two-phase immersion liquid cooling regulation process. Through the fusion perception of various multi-source heterogeneous physical data, it provides fundamental support for the calculation of driving variables, construction of feedback paths, and evaluation of regulation response in the elastic regulation process.

[0028] Specifically, the collected ambient temperature data, pressure and volume data, and execution feedback data are preprocessed to construct a standardized gas-liquid control dataset. The specific steps are as follows: Multi-point temperature synchronization correction is performed on the ambient temperature data. The specific operations include: For the temperature data collected by temperature sensors deployed in the inner wall of the elastic airbag 9, the air inlet and outlet paths, the inlet and outlet of the radiator 3, and the external environment, a timestamp alignment mechanism is introduced to unify the sampling rhythm of each measuring point, thereby effectively eliminating the time offset problem caused by the different response rates of the sensing channels; Subsequently, atmospheric pressure influence correction is performed by combining the altitude values ​​collected from the external environment. Based on the disturbance characteristics of air pressure on gas heat transfer and heat capacity effect under different altitude conditions, corresponding correction factors are established to ensure the equivalent expression capability of temperature data in different deployment scenarios.

[0029] During the processing of pressure and volume data, the instantaneous pressure values ​​in the gas phase region collected by pressure sensor 2 are subjected to sliding window averaging and median filtering to filter out short-cycle spikes caused by airflow disturbances or sensor response hypersensitivity, thereby improving the input stability in subsequent drive control. In the execution feedback data section, based on the instantaneous power of server 4, the rotational speed of bidirectional pump 7, and the coolant level, combined with the task execution rhythm and load evolution trend, the data is divided into time-series segments. The slope and fluctuation amplitude of changes within each key interval are extracted to capture the evolutionary characteristics of the execution state. After the ambient temperature data, pressure and volume data, and execution feedback data are preprocessed separately, a unified normalization operation is performed, and all variables are mapped to a unified dimensional standard. Finally, a standardized gas-liquid control dataset is constructed, providing a data foundation for the linkage decision-making and flexible output response of PID controller 6.

[0030] In this implementation scheme, the collected multi-source heterogeneous data is unified, structured, and standardized to ensure that all input variables have consistent time accuracy, physical scale, and semantic expression during gas phase regulation and elastic linkage control, thereby improving the PID controller 6's perception accuracy and response robustness to changes in gas-liquid state. Multi-point temperature synchronization correction eliminates sensing delay differences, and altitude values ​​are combined to correct for air pressure disturbances, enhancing the universality of environmental temperature data. Noise reduction and mapping integration are performed on pressure and volume data to stabilize input sources and enhance the accuracy of system elastic modeling. Trend extraction and normalization are performed on execution feedback data to ensure that the control logic maintains drive consistency and response controllability even under dynamic evolution.

[0031] Specifically, based on the standardized gas-liquid control dataset, the specific steps for evaluating the current gas phase space pressure state are as follows: multiply the instantaneous pressure values ​​of the gas phase region of each pressure sensor by the corresponding heat exchange capacity factor and sum them up to obtain the weighted gas phase pressure total value; add the instantaneous power of server 4 to the altitude value of the current deployment environment and then subtract the heat exchange capacity factor to obtain the basic load assessment value; subtract the basic load assessment value from the weighted gas phase pressure total value to obtain the pressure deviation input value.

[0032] The formula for calculating the pressure deviation input value is: In the formula, N represents the number of pressure sensors; This represents the instantaneous pressure value of the gas phase region of the i-th pressure sensor, used to construct a representative pressure field. It is the original data source for sensing the evaporation dynamics in a closed environment, originating from an industrial-grade pressure sensor array deployed at the top of the gas phase region of the cooling tank. This represents the instantaneous power of server 4, used to quantify the intensity of coolant evaporation driven by the current heat source. It is a fundamental variable for dynamically evaluating steam load and originates from the power acquisition interface of the server power management unit or main control chip. The altitude value of the current deployment environment is used to correct the boiling point threshold of the coolant. It is an important external parameter for determining the evaporation initiation condition and reference pressure offset, and it comes from the embedded pressure sensor 2. This represents the dynamic weighting factor of the i-th pressure sensor, with a value ranging from 0.2 to 1. It is used to measure the dynamic weight of the contribution of each pressure channel in the gas phase region to the overall deviation, and is derived from the fluctuation response intensity, historical stability deviation, and structural layout importance level within the current window. In the specific calculation, firstly, the gas phase pressure value change sequence of each sensor channel is extracted, and its change slope, fluctuation frequency, and abrupt change amplitude are calculated to quantify the current disturbance level. Then, combined with the fluctuation benchmark of the channel in the steady state stage, its reference amplitude, stable frequency band, and rhythm distribution are extracted, and multi-dimensional feature alignment is performed, including transient-steady state amplitude matching rate, period difference degree, and interference sensitivity overlap ratio. Combined with the spatial coupling and heat exchange coverage of the channel in the structural layout, dynamic weight aggregation is performed to finally determine the dynamic weighting factor corresponding to the channel. When the channel fluctuates significantly, changes are unstable, and deviates from the historical benchmark by a large margin within the current window, the dynamic weighting factor value is increased to the range of 0.9 to 1 to enhance its contribution to the pressure deviation output. Conversely, if the channel state is stable and the trend is highly predictable, the dynamic weighting factor is compressed to the range of 0.2 to 0.4 to avoid unnecessary interference with the overall control decision. The heat exchange capacity factor, ranging from 0.1 to 0.9, dynamically characterizes the instantaneous heat exchange efficiency of the current liquid cooling system. It is derived from the temperature gradient at the gas-liquid interface, the degree of heat exchange area exposure, and the flow status of the condensation path. In the specific calculation, firstly, parameters such as the condensation temperature distribution in the core liquid cooling region, the evaporator outlet temperature difference, and the coolant flow rate are jointly analyzed to identify the heat exchange hindrance characteristics and flow field inhomogeneity indicators at the current stage. Then, combined with changes in liquid level height, bubble obstruction ratio, and the trend of heat flux direction, the heat exchange smoothness under the current state is assessed, and the thermal resistance coupling factor is calculated. Finally, the heat exchange capacity factor is determined based on the severity of the heat exchange obstacle. When there are problems such as condensation path blockage, sudden increase in heat load, or evaporator backflow lag, the heat exchange capacity factor value increases to between 0.8 and 0.9 to reflect the current state of heat exchange obstruction. If the heat exchange channel is smooth and the temperature difference is stable, the heat exchange capacity factor value decreases to between 0.1 and 0.3, making the response to abnormal gas pressure more sensitive and improving the dynamic adjustment accuracy and cooling rhythm adaptability.

[0033] In this implementation scheme, the pressure state of the gas phase region in a two-phase immersion liquid-cooled environment is dynamically evaluated to construct a regulating benchmark that accurately reflects the current gas phase control requirements. By integrating instantaneous readings from multiple pressure sensors, server load power, deployment ambient temperature and altitude, and evaporator heat exchange capacity, and introducing a dynamic weighting mechanism and multi-source disturbance correction terms, a differential input quantity for the target gas phase pressure control section is formed, which is used as the error driving term for the subsequent PID controller. This formula, on the one hand, can integrate the representativeness and response importance of the regions where different pressure sensors are located, reflecting the comprehensive pressure level of the overall gas phase region through weighted superposition; on the other hand, by introducing server operating power, environmental deployment altitude, and current heat exchange capacity, it eliminates the offset effect caused by external disturbances on the pressure control target, improving the robustness and adaptability of the deviation input. In addition, the formula structure supports deployment in a sealed enclosure 1 in a distributed heterogeneous scenario, has good scalability and practicality, and can effectively drive subsequent fine control to achieve dynamic gas pressure stability control and energy efficiency optimization scheduling.

[0034] Specifically, the steps for dynamically adjusting the inflation and deflation of the elastic airbag 9 based on the evaluation results are as follows: Real-time comparison of the pressure deviation input value of the current gas phase region with the dynamic balance threshold: When the pressure deviation input value is less than or equal to the dynamic balance threshold, it is determined that the current state is in a stable range where the evaporation rate and condensation rate are well matched. At this time, the current operating state of the bidirectional pump 7 and the elastic airbag 9 is maintained, without triggering any form of gas injection or discharge. Closed-loop regulation is only performed based on the slight fluctuations monitored by the PID controller 6 and pressure sensor 2 to ensure that the overall gas-liquid phase is in a steady-state circulation mode. Simultaneously, auxiliary monitoring is performed by combining the real-time power changes of the server 4 and the coolant temperature trend to identify potential energy consumption fluctuations and heat... Threshold drift risk caused by load changes; when the pressure deviation input value is greater than the dynamic equilibrium threshold, it indicates that the gas phase region is in an unbalanced state. Path decision is made based on the positive or negative direction and rate of change of the pressure deviation: when the deviation is negative, that is, the pressure is lower than the stable expected level, inert gas is injected from the gas collecting tank 8 into the elastic gas bag 9. The expansion of the elastic gas bag 9 increases the pressure inside the sealed box 1 to counteract the negative pressure trend caused by the increase in the evaporation rate; when the deviation is positive, that is, the pressure is higher than the stable threshold, the elastic gas bag 9 is controlled to release gas into the gas collecting tank 8, and the contraction releases the available space inside the sealed box 1 to quickly reduce the pressure and accelerate heat transfer through expansion phase change, ultimately restoring the dynamic balance of the evaporation and condensation process.

[0035] In this implementation scheme, the stability judgment and adaptive response path selection of the gas-liquid phase change regulation state are achieved by evaluating the relationship between the pressure deviation input value and the dynamic equilibrium threshold in the gas phase region in real time. When the pressure is stable, unnecessary gas regulation operations are avoided, reducing energy consumption and device fatigue. When a pressure imbalance trend occurs, the filling and discharging processes between the gas collecting tank 8 and the elastic air bladder 9 can be triggered according to the direction of the deviation. The variable volume characteristics of the air bladder structure are used to quickly and flexibly adjust the pressure in the cavity, ensuring that a reasonable gas-liquid phase balance is maintained when facing dynamic load disturbances such as condensation hysteresis or enhanced evaporation, effectively improving the continuity and accuracy of the overall temperature control response.

[0036] Specifically, based on a standardized gas-liquid control dataset and combined with environmental heat exchange characteristics, the analysis of the gas pressure regulation trend involves the following steps: Calculate the difference between the coolant temperature at the inlet and outlet of radiator 3, add the instantaneous power of server 4 minus the bidirectional pump speed, to obtain the net load-driven temperature difference; calculate the absolute value of the difference between the instantaneous pressure value and the average gas phase pressure value in the gas phase region, divide it by the average gas phase pressure value, multiply it by the real-time height of the coolant level, and take the logarithm to obtain the pressure path disturbance value; divide the net load-driven temperature difference value by the external ambient temperature and multiply it by the pressure path disturbance value to obtain the gas phase regulation driving value.

[0037] The formula for calculating the gas-phase regulation driving value is: In the formula, It represents the coolant temperature at the radiator inlet, used to assess the baseline state before the condenser absorbs heat, and is the original temperature variable for judging the intensity of heat load input, which comes from the temperature sensor installed at the radiator inlet 3. This indicates the coolant temperature at the radiator outlet, which reflects the thermal state of the condenser after heat exchange and is an important basis for evaluating the strength of the condensation effect. It is derived from the temperature sensors installed on the three sides of the radiator. This represents the instantaneous power of server 4, which reflects the level of heat release caused by the current computing load. It is an important input for determining the evaporation trend and is derived from the energy consumption monitoring port. The speed of the bidirectional pump 7 is used to reflect the current level of inflation and deflation of the elastic airbag. It is a real-time control feedback variable for outputting the action and is derived from the built-in rotary encoder of the bidirectional pump 7. This represents the external ambient temperature, used to measure the degree of heat transfer limitation of the condenser. It is an external disturbance factor affecting the performance of the condenser, originating from an air temperature probe deployed on the outside. It represents the instantaneous pressure value in the gas phase region, used to characterize the pressure state generated by evaporation in real time. It is a direct adjustment variable for judging pressure deviation and control response amplitude, and is derived from pressure sensor 2 installed at the top of the gas phase space. This represents the average gas phase pressure value, used to standardize the current pressure deviation and improve control adaptability under cross-load scenarios. It is obtained by averaging the instantaneous pressure values ​​of each gas phase region. It represents the real-time height of the coolant level and is used to determine whether the current coolant capacity is within the normal operating range. It is an important physical indicator for early warning of uneven evaporation or abnormal replenishment, and is derived from capacitive level sensors installed on the side walls and bottom of the coolant tank.

[0038] In this implementation scheme, the comprehensive influence of dynamic adjustment on the gas phase side affecting the actuator's driving behavior during the two-phase immersion liquid cooling environment is characterized, serving as a key intermediate variable for regulating the stability of the gas-liquid balance. Its calculation comprehensively considers the state changes before and after condensation and heat absorption, the server's current instantaneous power, the real-time responsiveness of the cooling system's actuators, and the external ambient temperature background. Simultaneously, the logarithmic term in the formula amplifies the adjustment sensitivity significantly deviating from the steady-state pressure range, enhancing the adaptive response capability to gas phase dynamic inhomogeneities caused by multi-source disturbances. The gas phase adjustment driving value ultimately acts on the actuator's step command scheduling, opening control, and elastic airbag 9 inflation / deflation rhythm adjustment, thereby achieving closed-loop control of gas-liquid interface stability, vapor release intensity, and condensation rate regulation.

[0039] Specifically, the steps for adjusting the intake and exhaust rates of the bidirectional pump 7 in real time based on the analysis results are as follows: The calculation results of the current gas phase regulation drive value are collected in real time to directly drive the bidirectional pump 7 to achieve filling and exhaust control. When the gas phase regulation drive value is positive, and the gas phase regulation drive value generated by the natural change in gas phase state increases continuously between two consecutive values, the bidirectional pump 7 is automatically started to inject gas. Simultaneously, without causing gas turbulence, the upper limit of the injection rate is increased from the initial rate to the upper limit of the injection threshold in high-response mode to accelerate the pressure establishment process in the gas phase region inside the sealed chamber 1. During the injection process, the continuous response value of the pressure sensor 2 is recorded at 0.1-second intervals to form an injection pressure rise rate curve. The rotational speed increase strategy of the bidirectional pump 7 is adjusted in real time, i.e., the speed is increased smoothly in a linear increment during the initial stage, and the speed increase is stopped after the gas injection rate and pressure rise rate reach a threshold coupling relationship, maintaining the current stable rotational speed. When the regulation drive value is negative, the exhaust control process is immediately switched to, and directional pressure release is achieved by adjusting the inlet valve pressure of the bidirectional pump 7 to quickly release the gas inside the sealed chamber 1. The gas phase pressure drops back to the target lower limit, and the rate of change of coolant level and the degree of compression of the top vapor space are monitored in real time during the exhaust process. If the liquid level recovery rate is lower than the target threshold and the vapor compression margin is insufficient, the micro-liquid compensation injection mechanism is triggered. At the same time, the volume change rate of the elastic airbag 9 is used as a closed-loop feedback indicator to determine whether the exhaust has achieved the steady-state target. If the feedback deformation tends to be constant within a preset time, the exhaust is determined to be completed and the pressure stabilization mode is switched after a delay of 500ms: the bidirectional pump 7 maintains a constant speed and suspends the response to the new round of adjustment command, only maintaining the balance within the pressure fluctuation range; when the gas phase adjustment drive value is zero and the gas phase adjustment drive value generated by the natural gas phase state change remains unchanged, it indicates that the gas-liquid state is in a static and stable stage. The bidirectional pump 7 enters an intermittent start and stop state, triggering a low-speed operation detection only once every 10 seconds. At the same time, the high-frequency data recording channel is closed and the low-frequency monitoring mode is switched to. In the low-frequency monitoring mode, only the periodic polling of the status of key components and the identification of abnormal values ​​are retained to support the subsequent automatic exit or re-entry of the dynamic adjustment process.

[0040] In this implementation scheme, dynamic regulation and stable maintenance of the gas phase pressure within the sealed chamber 1 are achieved. Based on the real-time trend of the gas phase regulation drive value, the bidirectional pump 7 is driven to perform gas injection, discharge, or intermittent maintenance operations, thereby achieving precise response control to pressure fluctuations within the chamber caused by the evaporation and condensation process. By setting a speed ramp-up strategy, the smoothness of pressure rise during gas injection is ensured, preventing turbulence. During the discharge process, the deformation of the elastic airbag 9 is introduced as feedback, which helps to determine whether the discharge is sufficient, thereby improving the closed-loop adaptive capability of the regulation. At the same time, a pressure stabilization mode and a low-frequency monitoring mechanism are introduced to avoid energy waste and hardware fatigue caused by frequent operation when the fluctuation of the gas phase regulation drive value tends to stabilize, thereby improving long-term operational stability and response robustness while ensuring thermal control efficiency.

[0041] Specifically, using the gas phase space pressure state assessment results and gas pressure regulation trend analysis results as inputs, the comprehensive analysis of the current closed gas phase space state is carried out in the following steps: obtain the pressure deviation input value, subtract the volume of the elastic airbag from one and then multiply it with the pressure deviation input value to obtain the elastic modulation pressure difference value; take the elastic modulation pressure difference value as the numerator, and take the absolute value of the gas temperature inside the elastic airbag and add one to it, then take the logarithm as the denominator to obtain the elastic correction ratio value; add one to the gas phase regulation drive value and then multiply it with the elastic correction ratio to obtain the elastic linkage output value.

[0042] The formula for calculating the elastic linkage output value is: In the formula, This represents the pressure deviation input value, which reflects the instantaneous deviation formed by the combined effect of multiple input parameters under a specific air pressure environment. It is a driving variable for regulating the intensity of elastic response. The volume of the elastic airbag is used to characterize the total volume state used for gas phase pressure buffering and regulation. It is the core variable for analyzing the degree of elastic response and is derived from the cavity pressure conversion of the elastic airbag 9. The temperature of the gas inside the elastic airbag 9 is used to quantify the sensitivity of the current elastic structure to thermal disturbances. It is an important parameter for adjusting the response speed of volume change and is derived from a high-precision temperature sensor installed on the inner wall of the elastic airbag 9. This represents the gas phase regulation driving value, used to quantify the comprehensive control intensity required for gas phase regulation in the current environment, and is the rate adjustment factor for subsequent linkage response output.

[0043] In this implementation example, the pressure deviation input value for Example 1 is set to 0.18, the elastic airbag volume is 0.12, the internal gas temperature of the elastic airbag is 32°C, and the gas phase adjustment drive value is 0.20; the pressure deviation input value for Example 2 is set to 0.22, the elastic airbag volume is 0.15, the internal gas temperature of the elastic airbag is 35°C, and the gas phase adjustment drive value is 0.25; the pressure deviation input value for Example 3 is set to 0.26, the elastic airbag volume is 0.14, the internal gas temperature of the elastic airbag is 33°C, and the gas phase adjustment drive value is 0.22; the pressure deviation input value for Example 4 is set to 0.30, and the elastic airbag volume... The elastic linkage output value is 0.10, the internal gas temperature of the elastic airbag is 36°C, and the gas phase adjustment drive value is 0.28. For Example 5, the pressure deviation input value is set to 0.24, the elastic airbag volume is 0.13, the internal gas temperature of the elastic airbag is 34°C, and the gas phase adjustment drive value is 0.23. For Example 6, the pressure deviation input value is set to 0.20, the elastic airbag volume is 0.11, the internal gas temperature of the elastic airbag is 31°C, and the gas phase adjustment drive value is 0.21. For Example 7, the pressure deviation input value is set to 0.28, the elastic airbag volume is 0.16, the internal gas temperature of the elastic airbag is 37°C, and the gas phase adjustment drive value is 0.27. The elastic linkage output values ​​for each example are calculated, as shown in Table 1.

[0044] Table 1. Elastic Linkage Output Value Data Table

[0045] Figure 3 shows a line graph of the elastic linkage output value provided in the examples of this application. As can be seen from Table 1 and Figure 3, Example 4 has the highest elastic linkage output value, indicating that under conditions of large pressure deviation input value, low elastic airbag volume, and moderate temperature, the response amplitude to gas phase regulation is the most significant. This shows that the combination has higher gas injection linkage efficiency and system response sensitivity under the current conditions, making it suitable as a key control scenario to improve dynamic regulation capabilities. It should be prioritized for deployment in high-frequency strain situations to enhance control timeliness and feedback accuracy. In contrast, Example 1 has the lowest elastic linkage output value. Its pressure deviation input value is small, and under conditions of low temperature and high airbag volume, the output response is relatively weak. This reflects that the linkage capability of airbag regulation to system disturbances is limited under this combination. It is suitable as a maintenance condition in the steady-state operation phase, reducing its priority in participating in high-frequency dynamic adjustment, thereby reducing ineffective scheduling and redundant resource waste. The line graph of the elastic linkage output value clearly reveals the influence trend of different input parameter combinations on linkage performance, which helps to optimize the pneumatic regulation path from the perspective of multi-parameter coupling and improve the overall response coordination and execution efficiency.

[0046] Specifically, the steps for adjusting the volume adjustment command amplitude based on the analysis results are as follows: The elastic linkage output value of the current measuring point is compared with the elastic output evaluation threshold in real time. The elastic output evaluation threshold is divided into two levels: a first output threshold and a second output threshold, corresponding to different levels of control strategies and response priorities. When the elastic linkage output value is less than or equal to the second output threshold, it is determined to be a stable operating section. The PID controller 6 automatically enters the steady-state observation process, continuously sampling only the minute deformations of the internal structure of the pressure sensor 2 and the elastic airbag 9. The mean and standard deviation of the offset are dynamically calculated using a sliding window statistical method to determine whether the change is continuous. If the elastic linkage output value generated by the pressure deviation input value and the gas phase adjustment driving value is consistent in the direction of change for five consecutive times, it is determined that it is in a state of slow pressure evolution. The PID controller 6 then pauses the gas flow control logic and adjusts the bidirectional pump 7 to a low-power standby state to reduce energy consumption and mechanical shock frequency.

[0047] When the elastic linkage output value is greater than the second output threshold and less than or equal to the first output threshold, it enters the response warning state. The PID controller 6 automatically triggers the slow adjustment mechanism. Based on the combined change trend of the real-time collected pressure deviation input value and the gas phase adjustment drive value, it drives the elastic airbag 9 to perform a small-amplitude unidirectional inflation and deflation adjustment, and cooperates to reduce the speed of the bidirectional pump 7 to achieve a small-amplitude, high-precision adjustment process, and suppress the further amplification of the abnormal pressure trend.

[0048] When the elastic linkage output value exceeds the first output threshold, it is determined to be a section of significant pressure imbalance. The PID controller 6 immediately switches to the high-priority linkage path, starts the emergency control process, and quickly activates the deformation response mechanism of the elastic airbag 9. In conjunction with the high-flow pump start-up operation of the bidirectional pump 7, it performs short-term high-frequency rapid switching of inflation and deflation. At the same time, the linkage bypass channel provides collaborative compensation to ensure that the pressure deviation correction task is completed in the shortest possible time. Simultaneously, it monitors the deformation rate of the elastic airbag 9, the temperature rise rate of the execution channel, and the consistency of the response, thereby improving the control stability and extreme intervention efficiency under multi-source feedback linkage.

[0049] Figure 4 shows a schematic diagram of the PID control process involved in this invention, used to adjust the internal pressure of the gas phase region of the sealed housing to maintain it near the target value and ensure stable operation. The process begins in the PID control initialization phase, where the PID controller 6 sets a target pressure. As the ideal pressure value required for the gas phase zone of the sealed enclosure, the actual pressure P of the current gas phase zone is then collected in real time by pressure sensor 2, and the pressure deviation is subsequently calculated. This signal is then used as input to a PID controller for processing, resulting in a continuous control signal output. This signal represents the intensity and direction of the action required by the system to adjust its current state; the judgment logic module determines based on... Compared with the pressure threshold, when When the pressure exceeds the pressure threshold, the system enters the adjustment execution branch, and then... The sign determines the adjustment direction, when At that time, inert gas is injected into the airbag to increase the airbag volume and increase the overall gas phase pressure. When gas is released from the airbag, it reduces the airbag volume and lowers the overall gas phase pressure; when When the pressure is less than or equal to the pressure threshold, no gas injection or discharge operation is triggered; the current pressure state is simply maintained.

[0050] Figure 5 shows the system schematic of this invention, which mainly embodies the closed-loop control mechanism of internal air pressure regulation, heat exchange, and airbag linkage within the sealed cavity. The core consists of a sealed housing 1, which contains a pressure sensor 2 for real-time acquisition of pressure changes in the gas phase region. A cooling coil 5 handles the liquid phase heat exchange process, and a server 4 serves as the main heat source input, releasing some residual heat through a radiator 3. An elastic airbag 9 is positioned at the top of the gas phase region. Driven by the adjustment command of the PID controller 6, it achieves dynamic pressure response regulation through the gas filling and discharging path between the bidirectional pump 7 and the gas collection tank 8. When the internal air pressure deviates, the PID controller 6 calculates the gas phase regulation drive signal based on the feedback value from the pressure sensor 2 and drives the bidirectional pump 7 to inject or release gas, thereby controlling the deformation of the elastic airbag 9 to increase or decrease the gas phase pressure. Simultaneously, the figure emphasizes the linkage control path centered on the PID controller 6 through a dashed box, coupling the pressure detection, elastic regulation, and airflow transport modules into a closed loop to ensure high responsiveness and high stability of the gas-liquid interface fluctuation control.

[0051] In this implementation scheme, based on the real-time monitoring results of the elastic linkage output value, the control rhythm and response priority of the bidirectional pump 7 and the elastic airbag 9 are dynamically adjusted to realize a graded gas phase regulation control strategy, thereby improving the adaptive intervention capability and resource scheduling efficiency for pressure deviation. By comparing the elastic linkage output value with the elastic output evaluation threshold, low power consumption operation can be maintained in the stable range, slow adjustment control can be performed in the deviation warning range, and an emergency linkage compensation mechanism can be activated in the significant imbalance range. This enables multi-level rapid correction and precise control of the internal pressure of the sealed box 1, ensuring the continuity, consistency and efficiency of pressure response in the gas-liquid coupling environment.

[0052] The second aspect of this invention provides a two-phase immersion liquid-cooled gas-phase pressure regulation system based on PID control, comprising: a multi-parameter dynamic data extraction module, used to collect multiple key parameters in real time throughout the liquid-cooled gas pressure regulation process, including ambient temperature data, instantaneous pressure and volume data of the gas phase space, and execution feedback data including the operating power of server 4, the start-stop frequency of bidirectional pump 7, and the duration of gas charging and discharging. After collection, the above data will undergo unified preprocessing steps of timestamp alignment, noise filtering, and unit normalization to construct a standardized gas-liquid control dataset.

[0053] The real-time pressure sensing module in the gas phase region relies on a standardized gas-liquid control dataset to continuously monitor the pressure state of the gas phase region inside the sealed box 1. When pressure fluctuations or deviations are detected, the module dynamically adjusts the inflation and deflation amplitude and response time of the elastic airbag 9 to achieve sensitive control of pressure changes.

[0054] The PID closed-loop control module further predicts the pressure regulation trend based on the standardized gas-liquid control dataset and the heat exchange characteristics of the current environment. It also adjusts the intake and exhaust rates of the bidirectional pump 7 in real time according to the pressure deviation, so that the output gas flow rate matches the desired regulation target, thereby enhancing the robustness and real-time performance of the gas pressure control.

[0055] The elastic pressure regulation linkage control module integrates pressure state assessment results and regulation trend analysis results to comprehensively assess the response inertia, gas elastic buffering capacity, and volume regulation requirements of the current closed gas phase space. It then dynamically generates and adjusts the execution amplitude of volume control commands to ensure a highly coordinated linkage effect between pressure regulation and volume response, thereby maintaining dynamic stability and response continuity.

[0056] In this implementation scheme, the multi-parameter dynamic data extraction module constructs a unified, standardized data foundation usable by the control module, forming a complete data stream. After time synchronization, unit normalization, and noise reduction preprocessing, it outputs a standardized gas-liquid control dataset, ensuring the accuracy and consistency of subsequent control logic. The real-time pressure sensing module in the gas phase region focuses on monitoring and evaluating the real-time pressure state of the gas phase space within the sealed enclosure 1. Based on the numerical changes collected by pressure sensor 2 in the standardized gas-liquid control dataset, it dynamically identifies the stability and deviation trend of the current pressure and determines whether to trigger the inflation / deflation operation of the elastic airbag 9. This allows for early response and adjustment to potential pressure imbalances, improving the sensitivity and initiative of pressure control. The PID closed-loop control adjustment module is used to construct the automatic feedback control core of the system. By analyzing the deviation between the current pressure value and the target pressure, and combining the contextual information about environmental heat exchange characteristics in the standardized data, the PID algorithm outputs a control signal to precisely adjust the inflation / deflation rate of the bidirectional pump 7. To eliminate pressure deviation and stabilize the dynamics of the gas-liquid interface, while simultaneously considering hysteresis and disturbance suppression, the elastic pressure regulation linkage control module undertakes the task of coordinating the linkage between gas pressure regulation and structural volume response. Using the results of gas phase pressure state assessment and PID regulation trend analysis as input, it comprehensively analyzes the volume regulation requirements of the closed gas phase region, the deformation capability of the elastic airbag, and the inertia of gas-liquid regulation, thereby dynamically controlling the volume regulation amplitude and rhythm of the elastic airbag to ensure that the gas pressure regulation strategy achieves rapid response and effective implementation at the spatial structure level.

[0057] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0058] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A two-phase immersion liquid-cooled gas phase pressure regulation method based on PID control, characterized in that: include: S1. Collect ambient temperature data, pressure and volume data, and execution feedback data during the liquid cooling gas pressure regulation process. Preprocess the collected ambient temperature data, pressure and volume data, and execution feedback data to construct a standardized gas-liquid control dataset. S2. Based on the standardized gas-liquid control dataset, evaluate the current gas phase space pressure state and dynamically adjust the inflation and deflation force of the elastic airbag (9) based on the evaluation results. S3. Based on the standardized gas-liquid control dataset, analyze the gas pressure regulation trend in combination with the environmental heat exchange characteristics and adjust the intake and exhaust rate of the bidirectional pump (7) in real time based on the analysis results. S4. Using the gas phase space pressure state evaluation results and the gas pressure regulation trend analysis results as input, comprehensively analyze the current closed gas phase space state and drive the volume regulation command amplitude adjustment based on the analysis results.

2. The two-phase immersion liquid-cooled gas phase pressure regulation method based on PID control according to claim 1, characterized in that: The specific steps for collecting ambient temperature data, pressure and volume data, and execution feedback data during the liquid cooling pressure regulation process are as follows: Ambient temperature data is collected by high-precision temperature sensors installed on the inner wall of the elastic airbag (9) and on each air inlet and outlet path. The ambient temperature data includes: the altitude of the current deployment environment, the coolant temperature at the inlet of the radiator (3), the coolant temperature at the outlet of the radiator (3), the external ambient temperature, and the internal gas temperature of the elastic airbag (9); Pressure and volume data are collected by pressure sensors. The pressure and volume data includes: the instantaneous pressure value of the gas phase region of the pressure sensor (2), the volume of the elastic airbag (9), and the average gas phase pressure value is calculated and recorded by the instantaneous pressure value of each gas phase region; Execution feedback data is collected by the linkage control command recording device. The execution feedback data includes: the instantaneous power of the server (4), the rotation speed of the bidirectional pump (7), and the real-time height of the coolant level.

3. The two-phase immersion liquid-cooled gas phase pressure regulation method based on PID control according to claim 1, characterized in that: The specific steps for preprocessing the collected ambient temperature data, pressure and volume data, and execution feedback data to construct a standardized gas-liquid control dataset are as follows: Multi-point temperature synchronization correction is performed on the ambient temperature data. For the temperature values ​​collected from the inner wall of the elastic airbag (9), the air inlet and outlet paths, the inlet and outlet of the radiator (3), and the external environment, the sampling delay between different sensing channels is eliminated by the timestamp alignment method. At the same time, atmospheric pressure influence correction is performed in combination with the altitude value to ensure that the temperature data has consistent expression ability under different altitude environments. The pressure and volume data are denoised and mapped and integrated. The instantaneous pressure values ​​of the collected gas phase region need to be removed by sliding window averaging and median filtering to remove abnormal fluctuations and short-term peak values ​​to ensure input stability. For the execution feedback data, time-series features are extracted, and the instantaneous power of the server (4), the rotation speed of the bidirectional pump (7) and the coolant level are segmented and the trend slope is calculated. The environmental temperature data, pressure and volume data and execution feedback data after standardization are normalized and mapped to a unified dimension range, and stored to construct a standardized gas-liquid control dataset.

4. The two-phase immersion liquid-cooled gas phase pressure regulation method based on PID control according to claim 1, characterized in that: The specific steps for evaluating the current gas phase space pressure state based on the standardized gas-liquid control dataset are as follows: multiply the instantaneous pressure values ​​of the gas phase region of each pressure sensor by the corresponding heat exchange capacity factor and sum them up to obtain the weighted gas phase pressure total value; add the instantaneous power of the server (4) to the altitude value of the current deployment environment and then subtract the heat exchange capacity factor to obtain the basic load evaluation value; subtract the basic load evaluation value from the weighted gas phase pressure total value to obtain the pressure deviation input value.

5. The two-phase immersion liquid-cooled gas phase pressure regulation method based on PID control according to claim 1, characterized in that: The specific steps for dynamically adjusting the inflation and deflation of the elastic airbag (9) based on the evaluation results are as follows: Real-time comparison of the pressure deviation input value of the current gas phase region with the dynamic balance threshold: When the pressure deviation input value is less than or equal to the dynamic balance threshold, it is determined to be a stable range with good evaporation-condensation rate matching, maintaining the current working state of the bidirectional pump (7) and the elastic airbag (9), without triggering additional gas inflation and deflation regulation, only continuously relying on the PID closed loop to make slight corrections to pressure fluctuations, while monitoring the power change and condensation capacity trend of the server (4) to prevent threshold critical drift; When the pressure deviation input value is greater than the dynamic balance threshold, it is determined to be a pressure imbalance range, and the execution path is selected according to the deviation sign and change rate: When the deviation direction is negative, the gas collection tank (8) is driven to fill the elastic airbag (9) with inert gas, so that the elastic airbag (9) expands to increase the cavity pressure and suppress excessive evaporation; When the deviation direction is positive, the elastic airbag (9) is controlled to discharge gas back to the gas collection tank (8), so that the elastic airbag (9) contracts to release the cavity space, reduce the pressure and promote phase change heat dissipation.

6. The two-phase immersion liquid-cooled gas phase pressure regulation method based on PID control according to claim 1, characterized in that: The specific steps for analyzing the gas pressure regulation trend based on the standardized gas-liquid control dataset and combined with the environmental heat exchange characteristics are as follows: Calculate the difference between the coolant temperature at the inlet and outlet of the radiator (3), add the instantaneous power of the server (4) minus the bidirectional pump speed to obtain the net load-driven temperature difference; calculate the absolute value of the difference between the instantaneous pressure value and the average gas pressure value in the gas phase region and divide it by the average gas pressure value, then multiply it by the real-time height value of the coolant surface and take the logarithm to obtain the pressure path disturbance value; divide the net load-driven temperature difference by the external ambient temperature and multiply it by the pressure path disturbance value to obtain the gas phase regulation driving value.

7. The two-phase immersion liquid-cooled gas phase pressure regulation method based on PID control according to claim 1, characterized in that: The specific steps for adjusting the intake and exhaust rates of the bidirectional pump (7) in real time based on the analysis results are as follows: Real-time acquisition of the calculation results of the current gas phase adjustment drive value to directly drive the bidirectional pump (7) to achieve filling and exhaust control: When the gas phase adjustment drive value is positive and the gas phase adjustment drive value generated by the change of gas phase state in the previous two times is continuously increasing, the bidirectional pump (7) is automatically started to inject gas. At the same time, without causing gas turbulence, the upper limit of the gas injection rate is increased to quickly improve the overall pressure response capability of the sealed box (1). At the same time, the pressure rise rate change during the gas injection process is recorded synchronously, and the speed increase strategy of the bidirectional pump (7) is adjusted in real time. When the adjustment drive value is negative, the exhaust control process is triggered to achieve rapid pressure release by reducing the inlet pressure of the bidirectional pump (7), and the liquid level recovery rate during the pressure release stage is compared with the top vapor expansion. The margin is used for correlation analysis to determine whether micro-liquid compensation injection needs to be triggered synchronously. At the same time, the change trend of the elastic airbag (9) deformation during the critical exhaust process is used as feedback input. When the elastic airbag (9) deformation at the current moment is less than the elastic airbag (9) deformation at the previous moment, and the liquid level recovery rate is less than or equal to the top vapor expansion margin, the pressure holding mode is switched. When the elastic airbag (9) deformation at the current moment is greater than or equal to the elastic airbag (9) deformation at the previous moment, or the liquid level recovery rate is greater than the top vapor expansion margin, the pressure holding mode is delayed. When the gas phase adjustment drive value is zero, and the gas phase adjustment drive value generated by the change of gas phase state in the previous two times has not changed, only the bidirectional pump (7) is maintained to start and stop intermittently, and the high frequency data recording stream is turned off and low frequency monitoring is switched.

8. The two-phase immersion liquid-cooled gas phase pressure regulation method based on PID control according to claim 1, characterized in that: The specific steps for comprehensively analyzing the current closed gas phase space state using the gas phase space pressure state assessment results and gas pressure regulation trend analysis results as inputs are as follows: Obtain the pressure deviation input value, subtract the volume of the elastic airbag from it, and then multiply it by the pressure deviation input value to obtain the elastic modulation pressure difference value; use the elastic modulation pressure difference value as the numerator, and take the absolute value of the gas temperature inside the elastic airbag plus the logarithm as the denominator to obtain the elastic correction ratio value; add one to the gas phase regulation drive value and then multiply it by the elastic correction ratio to obtain the elastic linkage output value.

9. The two-phase immersion liquid-cooled gas phase pressure regulation method based on PID control according to claim 1, characterized in that: The specific steps for adjusting the volume adjustment command amplitude based on the analysis results are as follows: compare the elastic linkage output value of the current measuring point with the elastic output evaluation threshold in real time. The elastic output evaluation threshold is divided into a first output threshold and a second output threshold. When the elastic linkage output value is less than or equal to the second output threshold, the PID controller (6) automatically starts steady-state observation, only continuously samples the internal structural changes of the pressure sensor (2) and the elastic airbag (9), and calculates the mean and standard deviation of the dynamic offset in a sliding window manner. When the elastic linkage output value generated by the pressure deviation input value and the gas phase adjustment driving value is consistent in the direction of change for five consecutive times, the PID controller (6) suspends the gas flow control and keeps the bidirectional pump (7) in a low-power standby state. When the elastic linkage output value is greater than the second output threshold and less than or equal to the first output threshold: the PID controller automatically triggers gradual adjustment, and adjusts the single-sided micro inflation and deflation operation of the elastic airbag (9) according to the coordinated change trend of the real-time pressure deviation input value and the gas phase adjustment drive value, while reducing the speed of the bidirectional pump (7); When the elastic linkage output value is greater than the first output threshold: the PID controller immediately activates the high-priority linkage path and activates the emergency release of the elastic airbag (9) and the multi-channel sensing feedback mechanism: including the start-up of the high-flow bidirectional pump (7), the short-time high-frequency inflation and deflation switching and the redundant activation of the bypass channel, to complete the rapid correction of the pressure offset in the shortest time, while monitoring the temperature rise rate of the execution channel, the deformation rate of the elastic airbag (9) and the response consistency in real time.

10. A two-phase immersion liquid-cooled gas phase pressure regulation system based on PID control, using the two-phase immersion liquid-cooled gas phase pressure regulation method based on PID control as described in any one of claims 1-9, characterized in that: include: The multi-parameter dynamic data extraction module is used to collect ambient temperature data, pressure and volume data, and execution feedback data during the liquid cooling gas pressure regulation process. It also preprocesses the collected ambient temperature data, pressure and volume data, and execution feedback data to construct a standardized gas-liquid control dataset. The real-time pressure sensing module in the gas phase region is used to evaluate the current pressure state of the gas phase space based on a standardized gas-liquid control dataset, and to dynamically adjust the inflation and deflation force of the elastic airbag (9) based on the evaluation results. The PID closed-loop control adjustment module is used to analyze the gas pressure regulation trend based on the standardized gas-liquid control dataset and combined with the environmental heat exchange characteristics, and adjust the intake and exhaust rates of the bidirectional pump (7) in real time based on the analysis results; the elastic pressure regulation linkage control module is used to comprehensively analyze the current closed gas phase space state with the gas phase space pressure state assessment results and gas pressure regulation trend analysis results as inputs, and drive the volume adjustment command amplitude adjustment based on the analysis results.

Images