Centralized liquid cooling method and system based on phase change enhanced heat transfer and electronic equipment

Abstract

The invention discloses a centralized liquid cooling method and system based on phase change enhanced heat transfer and electronic equipment, and relates to the technical field of electronic equipment thermal management and liquid cooling, and the method comprises the steps: outputting the spatial and temporal distribution characteristics of heat flux density according to the thermal working condition information of a target area, and recognizing and judging the characteristics based on a coupling field data analysis method; according to the judgment characteristics, the regulation and control requirements of the heat exchange interface are determined, micro-scale adjustment is conducted on the heat exchange interface, and the physical property of the interface meets the phase change stabilization threshold value condition; carrying out phase change process stabilization control by taking the phase change stabilization threshold as a constraint condition and combining the real-time phase change signal, and outputting phase change state information; inputting the phase change state information into a centralized cycle control logic to form a cycle state feature; according to the circulation state characteristics, flow fluctuation in a circulation loop is actively controlled, and a phase change stabilization threshold value is dynamically updated; according to the invention, through multi-source coupling analysis and threshold value control, the problem of liquid cooling phase change instability is solved.

Description

Technical Field

[0001] This invention relates to the field of thermal management and liquid cooling technology for electronic devices, and more specifically, to a centralized liquid cooling method, system, and electronic device based on phase change enhanced heat transfer. Background Technology

[0002] In modern high-performance electronic devices, thermal management has become increasingly prominent as computing power and power density continue to improve. High-power electronic components such as chips, power modules, and data centers generate significant amounts of localized heat during operation. If this heat cannot be effectively dissipated, it can lead to excessive temperature rise, affecting performance, shortening lifespan, and even posing safety hazards. Traditional air cooling methods, limited by air's thermal conductivity, are insufficient to meet the heat dissipation requirements of high-power-density electronic systems. Liquid cooling technology, with its high specific heat capacity and high thermal conductivity, has become the mainstream solution. However, traditional liquid cooling systems still face a series of problems under high-load dynamic conditions, especially in applications requiring phase change enhanced heat transfer.

[0003] For example, the invention patent with announcement number CN120730713A discloses a single-phase and two-phase immersion liquid cooling method and system based on AI intelligent decision-making. The method includes injecting coupled data into a dynamic feature extraction engine, outputting a thermodynamic state evolution tensor containing features such as temperature change rate and load-heat flux density coupling coefficient. The thermodynamic state evolution tensor is input into a deep neural network model to calculate the temperature and pressure suitable for the current thermodynamic state evolution tensor, generating a set of closed-loop control instructions executable by the device. This closed-loop control instruction set is injected into the actuator group, which executes power reconfiguration and flow channel switching according to the instructions. In two-phase mode, the gaseous fluorinated liquid is liquefied and refluxed through a high-efficiency condenser, achieving adaptive switching of heat dissipation mode and thermal cycle reconstruction. The system includes a server, an AI algorithm controller, a coolant storage tank, a condenser, a circulating pump, electric valves, a pressure relief valve, and temperature sensors. This significantly improves heat dissipation efficiency and system reliability.

[0004] For example, the invention patent with announcement number CN120812913A discloses a thermosiphon-assisted natural circulation multi-stage liquid cooling method. This method involves setting up a double-layer heat exchange interface in the vicinity of the heat source, consisting of a lower high-thermal-conductivity layer and an upper low-thermal-conductivity layer, to create a local temperature rise abrupt change zone. When the temperature reaches the boiling threshold of the working liquid, a local thermosiphon flow jump is generated to drive liquid circulation. By adjusting the material and structural parameters of the double-layer heat exchange interface, the thermosiphon flow jump has a periodic re-triggering capability. Several distributed impedance control nodes are sequentially arranged along the thermosiphon flow path to establish a flow difference between adjacent nodes. Multiple stages of thermosiphon cavities are arranged in series along the flow path. By matching the start-up sequence of adjacent thermosiphon cavities, a progressive siphon drive is achieved. In the condensation outlet region of the last stage thermosiphon cavity, the condensate adheres and flows smoothly in the form of a liquid film. A liquid turning buffer zone is set after the condensation outlet to guide the condensate to a gentle return path, completing a closed-loop natural circulation.

[0005] The above-disclosed technical solutions have at least the following technical problems:

[0006] Traditional liquid cooling systems lack precise control over the phase change process, typically relying on constant flow rate or simple temperature feedback control, making it difficult to ensure that the phase change interface properties remain within a safe and stable range. Phase change instability can easily lead to localized liquid drying, decreased heat transfer efficiency, and even equipment damage. To address these issues, this invention proposes a solution. Summary of the Invention

[0007] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a centralized liquid cooling method, system, and electronic equipment based on phase change enhanced heat transfer, which solves the liquid cooling phase change instability problem through multi-source coupling analysis and threshold control.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] A centralized liquid cooling method based on phase change-enhanced heat transfer includes: outputting the spatiotemporal distribution characteristics of heat flux density based on the thermal condition information of the target area, and identifying and judging the characteristics based on coupled field data analysis; determining the heat exchange interface control requirements based on the judged characteristics, and performing micro-scale adjustment of the heat exchange interface to ensure that the interface properties meet the phase change stabilization threshold condition; using the phase change stabilization threshold as a constraint condition, combining real-time phase change signals to perform phase change process stabilization control, and outputting phase change state information; inputting the phase change state information into a centralized loop control logic to dynamically adjust the loop and form loop state characteristics; and actively controlling the flow fluctuations in the loop through a disturbance suppression method based on the loop state characteristics, and dynamically updating the phase change stabilization threshold based on the disturbance suppression results.

[0010] In a preferred embodiment, the step of outputting the spatiotemporal distribution characteristics of heat flux density based on the thermal condition information of the target area, and identifying and judging the characteristics based on the coupled field data analysis method, specifically includes: acquiring the temperature field distribution, power time series, and fluid condition information of the target area, and simultaneously collecting phase change data to form multi-source data containing phase change precursor characteristics; preprocessing the multi-source data, and mapping the processed multi-source data to a common time scale and spatial coordinate system to form a coupled field data tensor; and performing inversion calculation on the coupled field data tensor through a physical information neural network to solve for the temporal distribution characteristics of heat flux density. The system generates a spatial distribution and outputs local energy input characteristics. Based on the heat flux density distribution and local energy input characteristics, combined with phase transition data, a coupled field feature set is constructed. Dynamic mode decomposition and Koopman modal analysis are performed on the coupled field feature set to identify the dominant dynamic modes and their growth rates, and modal features are extracted from them. The modal features are input into a Bayesian change point detection model to perform a probabilistic assessment of the phase transition state, obtaining a phase transition instability risk score with confidence intervals, and identifying the key modes that lead to instability. The phase transition instability risk score, key modes, and coupled field feature set are fused to generate a phase transition instability judgment feature set.

[0011] In a preferred embodiment, the step of inverting the coupled field data tensor using a physical information neural network to solve for the spatiotemporal distribution of heat flux density and output local energy input features is as follows: A multimodal feature embedding space is constructed based on self-supervised representation learning, mapping the temperature field and phase transition data into a unified latent space representation; the coupled field data tensor is weighted and encoded using a spatiotemporal attention mechanism to form a phase transition sensitive feature map; a temporal evolution feature sequence is generated by comparing and learning phase transition data from different time slices; a spatial adjacency graph is constructed based on the phase transition sensitive feature map and the temporal evolution feature sequence, and input into a cross-scale graph neural network to obtain the spatiotemporal distribution of heat flux density; energy activity clustering analysis is performed on the spatiotemporal distribution of heat flux density, and an energy active region label map is obtained based on local heat flux density peak values, gradient change amplitudes, and hotspot temporal characteristics; growth trajectory analysis is performed on the data within the energy active region, and a local energy input feature set is obtained by combining heat flux density changes, the phase transition sensitive feature map, and the temporal evolution sequence.

[0012] In a preferred embodiment, the energy activity clustering analysis of the spatiotemporal distribution of heat flux density, based on local heat flux density peak values, gradient change amplitudes, and hotspot temporal characteristics, yields an energy active area label map, specifically as follows: For each spatial unit in the spatiotemporal distribution of heat flux density, the maximum heat flux density value is extracted throughout the time series, which is the local heat flux peak value of that unit; the difference calculation is performed on the spatial neighborhood of the heat flux density distribution to obtain the amplitude of the heat flux density gradient change with spatial location, i.e., the gradient change amplitude; based on the data sequence of heat flux density changing over time, hotspot temporal characteristics are extracted; the local heat flux peak values, gradient change amplitudes, and hotspot temporal characteristics are normalized and combined into spatial unit feature vectors; the similarity of the feature vectors of spatial units is calculated based on the dynamic time warping method, and the feature vectors of spatial units are clustered using a hierarchical clustering algorithm, grouping spatial units with similar heat flux intensities and temporal evolution characteristics into the same cluster, obtaining the cluster label and intra-cluster statistical characteristics of each spatial unit; the cluster label of each spatial unit is mapped back to the original spatial coordinate system to form an energy active area label map.

[0013] In a preferred embodiment, the phase transition stabilization threshold is specifically defined as follows: A training dataset is constructed based on historical observation data and simulation results; a Bayesian probability model of instability probability is constructed based on the training data; the Bayesian probability model is solved using an online variational inference method, dynamically matching the posterior distribution of the input variables with the observation data, so that the critical heat flux density, nucleation density threshold, wettability threshold, and liquid replenishment capacity threshold converge to the posterior probability distribution in a statistical sense; the posterior probability distribution is input into the posterior prediction model to calculate the probability of phase transition instability caused by each physical parameter at different values, and an instability probability curve is generated; by comparing the instability probability rise intervals corresponding to different parameter values ​​with the critical change points, key inflection points of each physical property parameter are identified, and these key inflection points are used as candidate ranges for subsequent threshold selection; according to a preset risk level, the corresponding quantiles on the instability probability curves within the candidate range are selected as critical heat flux density, nucleation density threshold, wettability threshold, and liquid replenishment capacity threshold, thereby obtaining the final phase transition stabilization threshold.

[0014] In a preferred embodiment, the step of using a phase transition stabilization threshold as a constraint, combined with real-time phase transition signals, to perform phase transition process stabilization control and output phase transition state information specifically involves: performing denoising, time synchronization, and short-time window feature extraction on the real-time phase transition signal to form a phase transition feature vector at the current moment; comparing the phase transition feature vector with the phase transition stabilization threshold to output an instantaneous phase transition instability risk score for the current spatial unit; and when the instantaneous phase transition instability risk score reaches a warning level, using the instantaneous risk score as a priori, combined with the phase transition feature sequence from several past moments, and predicting the future evolution trend of the phase transition features based on a short-time prediction model to form... The system generates a short-term instability probability estimate for the future; based on the instantaneous phase transition instability risk score and the short-term instability probability estimate, and combined with the boundary conditions provided by the phase transition stabilization threshold, it generates control commands corresponding to the instability trend; it performs interface regulation on the phase transition region according to the control commands, and monitors the changes in phase transition characteristics in real time during the execution of the action to obtain instantaneous feedback of the action; based on the change in the deviation between the action feedback signal and the threshold, it determines whether the control action reduces the instantaneous phase transition instability risk; if the risk does not decrease, it performs a rollback adjustment; it integrates the phase transition characteristics, instantaneous phase transition instability risk score, short-term instability probability estimate, control action, and action effect into phase transition state information.

[0015] In a preferred embodiment, the step of inputting phase change state information into centralized loop control logic to dynamically adjust the loop and form loop state characteristics is as follows: Phase change state information is input into centralized loop control logic, and key variables are extracted; the extracted key variables are input into a loop-phase change coupling mapping model to establish a mapping relationship between local phase change demand and loop flow rate, pressure regulation direction, and vapor-liquid ratio target range, thus obtaining macro-control demand; based on the macro-control demand, the loop prediction model outputs the changing trends of loop flow rate, vapor-liquid ratio, and pressure within future control cycles, generating loop prediction targets; the loop prediction targets are input into a control strategy generation module, and an executable loop adjustment instruction set is generated by integrating flow coupling relationship, system stability boundary, and energy consumption conditions through constraint solving methods; the loop loop operation state is adjusted in real time according to the loop adjustment instructions, and system response data is collected during the adjustment process to form the current loop response sequence; time window analysis is performed on the loop response sequence to extract loop operation characteristics; the loop operation characteristics are correlated and integrated with phase change state information to generate loop state characteristics.

[0016] In a preferred embodiment, the active control of flow fluctuations in the loop based on the loop state characteristics and the dynamic updating of the phase transition stabilization threshold based on the disturbance suppression results are as follows: The amplitude, dominant frequency component, phase drift, and spatial distribution of the current flow fluctuations in the loop are extracted based on the loop state characteristics and input as disturbance characterization vectors into the disturbance identification model to identify disturbance patterns; based on the identified disturbance patterns, the disturbance characteristics are mapped to the corresponding flow control quantities, and the disturbance suppression controller determines the active adjustment direction and intensity to be applied; the timing response of the control action is predicted based on the loop prediction model to generate a disturbance suppression command that can be executed in real time in the next control cycle, and this command is applied to the loop for active suppression; during the disturbance suppression process, the following steps are taken: The system collects pressure drop changes, flow recovery rates, fluctuation attenuation coefficients, and local vapor-liquid ratio stabilization states in the circulating loop to form a disturbance response sequence. This sequence is then input into a disturbance suppression effectiveness evaluation model to quantify the effectiveness of disturbance suppression. Based on the actual effect of disturbance suppression, residual fluctuation characteristics affecting local phase transition behavior are extracted. These residual fluctuation characteristics, along with the phase transition stabilization threshold, are input into a Bayesian posterior update mechanism. This mechanism automatically corrects the posterior distributions of the critical heat flux density, nucleation density threshold, wettability threshold, and liquid replenishment capacity threshold under the new circulating environment, resulting in an updated posterior distribution. Based on the updated posterior distribution, the quantiles of each physical threshold are recalculated according to a preset risk level. These quantiles are then integrated to generate an updated set of phase transition stabilization thresholds for use as stabilization constraints in the next control cycle.

[0017] The system based on a centralized liquid cooling method for phase change enhanced heat transfer includes a feature determination module, a phase change stabilization threshold module, a phase change state module, a circulation loop module, and a disturbance suppression module, which are interconnected. The feature determination module outputs the spatiotemporal distribution characteristics of heat flux density based on the thermal condition information of the target area and identifies the features using coupled field data analysis. The phase change stabilization threshold module determines the heat exchange interface control requirements based on the features and performs micro-scale adjustments to the heat exchange interface to ensure that the interface properties meet the phase change stabilization threshold condition. The phase change state module uses the phase change stabilization threshold as a constraint and combines real-time phase change signals to perform phase change process stabilization control and outputs phase change state information. The circulation loop module inputs the phase change state information into the centralized circulation control logic, dynamically adjusts the circulation loop, and forms circulation state characteristics. The disturbance suppression module actively controls the flow fluctuations in the circulation loop based on the circulation state characteristics using a disturbance suppression method and dynamically updates the phase change stabilization threshold based on the disturbance suppression results.

[0018] An electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform any one of the centralized liquid cooling methods based on phase change enhanced heat transfer.

[0019] The technical effects and advantages of the centralized liquid cooling method, system, and electronic equipment based on phase change enhanced heat transfer in this invention are as follows:

[0020] 1. This invention proposes the concept of a "phase change stabilization threshold" for phase change process control. Through Bayesian probabilistic modeling and online variational inference, combined with historical observation data and real-time simulation results, statistical learning and dynamic optimization are performed on physical properties such as critical heat flux density, nucleation density, wettability, and liquid replenishment capacity to achieve constrained control of the phase change process. Real-time phase change signals are compared and analyzed with the threshold. Through short-time prediction models and control feedback mechanisms, microscale adjustment and dynamic control of the heat transfer interface are achieved, effectively suppressing phase change instability and improving the system's heat transfer efficiency and operational safety.

[0021] 2. This invention establishes a centralized loop control logic, coupling phase change state information with dynamic adjustment of the loop. Through a loop-phase change mapping model and a loop prediction model, it achieves optimized control of macroscopic flow rate, pressure, and vapor-liquid ratio. Combined with a disturbance suppression module, this invention can actively control flow fluctuations in the loop. Through disturbance identification, control command generation, and effect evaluation, it achieves rapid stabilization of the flow state. Simultaneously, the disturbance response feedback is used to dynamically update the phase change stabilization threshold, forming a closed-loop intelligent adjustment mechanism. This collaborative control strategy enables the liquid cooling system to maintain efficient and stable operation under complex thermal conditions, significantly reducing the risk of local overheating and phase change instability. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the centralized liquid cooling method based on phase change enhanced heat transfer according to the present invention.

[0023] Figure 2 This is a schematic diagram of the system structure of the centralized liquid cooling method based on phase change enhanced heat transfer according to the present invention. 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0025] Example 1, Figure 1 This invention presents a centralized liquid cooling method based on phase change-enhanced heat transfer, comprising:

[0026] S1, based on the thermal condition information of the target area, outputs the spatiotemporal distribution characteristics of heat flux density, and identifies and judges the characteristics based on the coupled field data analysis method;

[0027] In this embodiment, based on the thermal condition information of the target area, the spatiotemporal distribution characteristics of the heat flux density are output, and the characteristics are identified and judged based on the coupled field data analysis method, as follows:

[0028] The temperature field distribution, power time series and fluid condition information of the target area are acquired, and phase change data are collected at the same time to form multi-source data containing phase change precursor characteristics. The phase change data includes acoustic signals, optical reflection signals and micro-vibration signals.

[0029] Multi-source data is processed by time alignment, spatial interpolation and noise suppression, and the processed multi-source data is uniformly mapped to a common time scale and spatial coordinate system to form a coupled field data tensor;

[0030] The tensor of coupled field data is inverted and calculated using a physical information neural network to solve the spatiotemporal distribution of heat flux density and obtain local energy input characteristics to characterize the local energy injection status of the phase transition process. These local energy input characteristics describe the energy driving intensity required for local bubble nucleus formation, vapor-liquid interface migration, and local film evolution.

[0031] Based on the heat flux density distribution and local energy input characteristics, combined with the spectral characteristics of acoustic signals, the scattering characteristics of optical reflection signals, and the amplitude changes of micro-vibration signals, a coupled field feature set is constructed.

[0032] Dynamic mode decomposition and Koopman mode analysis are performed on the coupled field feature set to identify the dominant dynamic modes and their growth rates, and modal features that can characterize phase transition instability trends are extracted from them. These modal features are used to characterize potential phase transition instability trends, including local bubble period synchronization, interface perturbation amplification, and critical evolution paths for thin film formation.

[0033] Modal features are input into a Bayesian change point detection model to perform a probabilistic assessment of the phase transition state, obtain a phase transition instability risk score with confidence intervals, and identify the key modes that lead to instability.

[0034] The phase transition instability risk score, key modes and coupled field feature set are fused to generate a phase transition instability judgment feature set.

[0035] In this embodiment, the tensor of the coupled field data is inverted using a physical information neural network to solve for the spatiotemporal distribution of heat flux density, and local energy input characteristics are obtained to characterize the local energy injection status of the phase transition process, as follows:

[0036] A multimodal feature embedding space is constructed based on self-supervised representation learning, which maps temperature field, acoustic signal, optical reflection signal and micro-vibration signal into a unified latent space representation;

[0037] By using a spatiotemporal attention mechanism to weighted encode the coupled field data tensor, the response capability to sudden changes in heat flux density, enhanced local bubble activity regions, and interface disturbance accumulation regions is enhanced, thereby forming a spatiotemporal feature map that highlights the phase transition sensitive region, namely the phase transition sensitive feature map.

[0038] By comparing and learning phase transition data from different time slices, the physical information neural network can automatically capture the temporal synchronicity of local heat flow driving trends, interface disturbance growth trends, and bubble behavior from temporal changes, and generate a time evolution feature sequence. The time evolution features include: the peak heat flux density of each spatial unit at each time point and the time of peak occurrence, as well as the rate of change of heat flux density at adjacent time points.

[0039] Based on the phase transition sensitive feature map and the time evolution feature sequence, a spatial adjacency relationship graph is constructed and input into a cross-scale graph neural network, which infers the heat flux density from multi-scale spatial correlation and temporal changes, and obtains the spatiotemporal distribution results of the heat flux density, including local heat flux peaks, hot spot migration paths and heat flux gradient distribution.

[0040] Energy activity clustering analysis was performed on the spatiotemporal distribution of heat flux density. Based on the local heat flux density peak, gradient change amplitude, and hot spot time sequence characteristics, an energy active region label map was obtained. The energy active region label map includes the bubble nucleus activation region, the interface heating region, and the thin film transition tendency region.

[0041] Growth trajectory analysis is performed on the data within the energy-active region. By combining heat flux density changes, phase transition sensitive feature maps, and time evolution sequences, the growth trajectory of the driving force for bubble nucleation, the cumulative trend of interfacial movement potential energy, the energy evolution path of thin film formation, and the trajectory of the latent heat release rate of phase transition are extracted, thereby obtaining a local energy input feature set characterizing the local energy driving mechanism of phase transition.

[0042] In this embodiment, a spatial adjacency graph is constructed based on the phase transition sensitive feature map and the temporal evolution feature sequence, and then input into a cross-scale graph neural network. This network infers the heat flux density from multi-scale spatial correlations and temporal variations, resulting in the spatiotemporal distribution of the heat flux density, as detailed below:

[0043] Inputting a phase transition sensitive feature map from spatiotemporal attention encoding and the time-varying heat flux peak, gradient, and phase transition signal sequence of each spatial unit, the node features of each spatial unit are formed.

[0044] Each spatial unit is regarded as a graph node, and the node features are phase transition sensitive features and time evolution features. Edges are established based on spatial proximity and feature similarity to form an adjacency matrix.

[0045] The adjacency matrix is ​​input into a cross-scale graph neural network, which outputs the heat flux density value of the spatial unit corresponding to each node, including the prediction results at different time points.

[0046] Map all node predictions back to the original spatial coordinate system to form a complete spatiotemporal distribution of heat flux density.

[0047] In this embodiment, energy activity clustering analysis is performed on the spatiotemporal distribution of heat flux density. Based on the local heat flux density peak value, gradient change amplitude, and hotspot temporal characteristics, an energy active area label map is obtained, as follows:

[0048] For each spatial unit in the spatiotemporal distribution of heat flux density, the maximum heat flux density value is extracted from the entire time series, which is the local heat flux peak value of that unit;

[0049] By performing differential calculations on the spatial neighborhood of the heat flux density distribution, the magnitude of the change in heat flux density gradient with spatial location is obtained, i.e., the gradient change magnitude.

[0050] Based on the data sequence of heat flux density changing over time, hotspot temporal features are extracted, namely the trajectory of heat flux peak change and peak occurrence time of each spatial unit over time, which are used to reflect the emergence, dissipation and migration trends of hotspots.

[0051] The local heat flux peak, gradient change amplitude, and hot spot time sequence characteristics are normalized and combined into a feature vector for each spatial unit, i.e., the spatial unit feature vector.

[0052] The similarity of feature vectors of spatial units is calculated based on the dynamic time warping (DTW) method, and the feature vectors of spatial units are clustered by hierarchical clustering algorithm. Spatial units with similar heat flux intensity and time evolution characteristics are divided into the same cluster, and the cluster label and intra-cluster statistical characteristics of each spatial unit are obtained.

[0053] The clustering labels of each spatial unit are mapped back to the original spatial coordinate system to form an energy-active region label map.

[0054] S2. Based on the judgment characteristics, determine the heat exchange interface control requirements and perform micro-scale adjustment on the heat exchange interface so that the interface properties meet the phase change stabilization threshold conditions.

[0055] In this embodiment, based on the judgment characteristics, the heat exchange interface control requirements are determined, and the heat exchange interface is microscale adjusted so that the interface properties meet the phase change stabilization threshold condition, as follows:

[0056] Based on the judgment characteristics, the local phase transition behavior is analyzed by Bayesian inference statistical method, and the local heat flux density critical value, nucleation density threshold, wettability threshold and liquid supply capacity threshold are output to form the phase transition stabilization threshold.

[0057] By comparing the judgment characteristics with the phase change stabilization threshold, the interface region that needs to be regulated is determined, and the heat exchange interface regulation target is output, including the specific range that the nucleation density, wettability and liquid supply capacity need to reach.

[0058] Based on the heat exchange interface control target, select the control method, such as microstructure adjustment, surface chemical modification / coating, microchannel flow or jet strategy, to form an implementable control scheme for local property control;

[0059] The adjustment scheme is applied to the interface region to adjust the local nucleation density, wettability and liquid replenishment capacity to make them close to or reach the stabilization threshold, and the adjusted interface physical properties are output, including the nucleation density distribution, wettability and liquid replenishment capacity.

[0060] The adjusted interface properties are used as input for subsequent phase change process stabilization control and centralized liquid cooling cycle dynamic adjustment.

[0061] In this embodiment, based on the judgment characteristics, the local phase transition behavior is analyzed using Bayesian inference statistical methods, and the local heat flux density critical value, nucleation density threshold, wettability threshold, and liquid replenishment capacity threshold are output to form the phase transition stabilization threshold, as detailed below:

[0062] Based on historical observation data and simulation results, a training dataset is constructed, which includes local heat flux density, nucleation density, wettability, liquid replenishment capacity, and corresponding phase transition instability labels.

[0063] Based on the training data, a Bayesian probability model of instability probability is constructed, with local heat flux density, nucleation density, wettability and liquid replenishment capacity as input variables and the probability of phase change instability events as output variables.

[0064] By solving the Bayesian probability model using an online variational inference method, the posterior distribution of the input variables is dynamically matched with the observed data, so that the critical heat flux density, nucleation density threshold, wettability threshold, and liquid replenishment capacity threshold converge to the posterior probability distribution in a statistical sense.

[0065] The posterior probability distribution is input into the posterior prediction model to calculate the probability of phase transition instability caused by each physical parameter at different values, and to generate an instability probability curve describing the change of phase transition stability with parameter variation.

[0066] By comparing the range of rising instability probability and critical change points corresponding to different parameter values, the key inflection points of each physical property parameter transitioning from a stable state to an unstable state are identified, and these key inflection points are used as candidate ranges for subsequent threshold selection.

[0067] Based on the preset risk level (such as the optimal point determined when the instability probability reaches 10% or 20%), the corresponding quantiles on the instability probability curve are selected from the candidate range as the critical heat flux density, nucleation density threshold, wettability threshold and liquid replenishment capacity threshold, thereby obtaining the final phase change stabilization threshold.

[0068] In this embodiment, the judgment features are compared with the phase transition stabilization threshold to determine the interface region that needs to be regulated, as follows:

[0069] The judgment features are compared with the phase transition stabilization thresholds obtained through the Bayesian inference statistical model. The degree of deviation is measured by the differences in the critical value of heat flux density, nucleation density threshold, wettability threshold and liquid replenishment capacity threshold, forming an instant phase transition instability risk score that can quantify the tendency of local phase transition instability.

[0070] The interface spatial units are screened based on the instantaneous phase transition instability risk score. Regions whose instantaneous phase transition instability risk scores exceed the phase transition stabilization threshold are marked as interface regions that need to be regulated, thereby obtaining information on the regulated region including its location distribution, area range, and main deviation indicators.

[0071] S3 uses the phase transition stabilization threshold as a constraint condition, combines the real-time phase transition signal to perform phase transition process stabilization control, and outputs phase transition state information;

[0072] In this embodiment, the phase transition stabilization threshold is used as a constraint condition, and the phase transition process is stabilized and controlled in conjunction with the real-time phase transition signal. The phase transition state information is then output, as follows:

[0073] Denoising, time synchronization, and short-window feature extraction are performed on the real-time phase transition signal to form the phase transition feature vector at the current moment;

[0074] The phase transition eigenvector is compared with the phase transition stabilization threshold to output the instantaneous phase transition instability risk score of the current spatial unit;

[0075] When the instantaneous phase transition instability risk score reaches the warning level, the instantaneous risk score is used as a priori. Combined with the phase transition characteristic sequence of several past moments, the future evolution trend of the phase transition characteristics is predicted based on the short-term prediction model to form a future short-term instability probability estimate.

[0076] Based on the instantaneous phase transition instability risk score and short-term instability probability estimate, and combined with the boundary conditions provided by the phase transition stabilization threshold, control instructions corresponding to the instability trend are generated, including adjustment type, adjustment area, target amplitude and execution duration.

[0077] According to the control command, the interface regulation or working fluid adjustment action is performed in the phase change region, and the changes in phase change characteristics are monitored in real time during the execution of the action to obtain the immediate feedback of the action.

[0078] Based on the change in the deviation between the action feedback signal and the threshold, it is determined whether the control action reduces the risk of instantaneous phase transition instability; if the risk has not decreased, a rollback adjustment is performed.

[0079] Phase transition characteristics, instantaneous phase transition instability risk score, short-term instability probability estimation, control actions, and action effects are integrated into phase transition state information.

[0080] S4 inputs the phase change state information to the centralized loop control logic, dynamically adjusts the loop loop, and forms loop state characteristics;

[0081] In this embodiment, the phase transition state information is input to the centralized loop control logic to dynamically adjust the loop and form loop state characteristics, as follows:

[0082] The phase transition state information is input into the centralized loop control logic, and the local risk level, phase transition region distribution, interface response effect and phase transition evolution trend are analyzed to extract key variables that may affect the macroscopic behavior of the loop.

[0083] The extracted key variables are input into the cycle-phase change coupling mapping model to establish the mapping relationship between local phase change demand and cycle flow rate, pressure regulation direction and vapor-liquid ratio target range, thereby obtaining the system-level macro-control demand.

[0084] Based on the needs of macro-control, the cyclic prediction model is used to calculate the changing trends of cyclic flow rate, vapor-liquid ratio and pressure in the future control cycle, and generate cyclic prediction targets, namely the target flow rate, vapor-liquid ratio offset direction and pressure correction amount required for the next control cycle.

[0085] The cyclic prediction target is input into the control strategy generation module. Through the constraint solving method, the flow coupling relationship, system stability boundary and energy consumption conditions are integrated to generate an executable cyclic adjustment instruction set, including the flow adjustment of the main loop and branches, the adjustment range of the vapor-liquid ratio and the pressure setpoint offset.

[0086] The operating status of the circulating loop is adjusted in real time according to the cyclic adjustment command, and system response data such as circulating pressure, main loop and branch loop flow, loop temperature difference and steam content change are collected during the adjustment process to form the current cyclic response sequence;

[0087] Time window analysis was performed on the cyclic response sequence to extract characteristic quantities that characterize the cyclic operation state, such as pressure fluctuation intensity, flow distribution consistency, vapor-liquid ratio stability, and phase change load balance.

[0088] By linking and integrating the cyclic operation characteristics with the phase transition state information, a cyclic state characteristic that can reflect the operational stability of the cyclic loop under the current control is generated.

[0089] S5. Based on the characteristics of the cyclic state, the flow fluctuations in the cyclic loop are actively controlled by the disturbance suppression method, and the phase change stabilization threshold is dynamically updated based on the disturbance suppression results.

[0090] In this embodiment, based on the characteristics of the cyclic state, the flow fluctuations in the cyclic loop are actively controlled using a disturbance suppression method, and the phase transition stabilization threshold is dynamically updated based on the disturbance suppression results, as detailed below:

[0091] Based on the characteristics of the cyclic state, the amplitude, dominant frequency component, phase drift and spatial distribution of the current flow fluctuation are extracted and used as disturbance characterization vectors to input into the disturbance identification model to identify the dominant disturbance mode that may cause phase transition instability.

[0092] Based on the identified disturbance patterns, the disturbance characteristics are mapped to the corresponding flow control variables (such as instantaneous flow deviation, local pressure drop offset, abnormal changes in vapor holdup, etc.). The disturbance suppression controller determines the direction and intensity of the active adjustment to be applied, including control actions such as flow adjustment, bypass liquid replenishment, local throttling, or pressure compensation.

[0093] Based on the LSTM cyclic prediction model, the timing response of the control action is predicted, and a disturbance suppression command that can be executed in real time in the next control cycle is generated and applied to the cyclic loop, thereby actively suppressing flow fluctuations, pressure pulsations and vapor-liquid interface disturbances.

[0094] During the disturbance suppression process, the pressure drop change, flow recovery rate, fluctuation attenuation coefficient, and local vapor-liquid ratio stabilization status of the circulation loop are collected to form a disturbance response sequence.

[0095] The perturbation response sequence is input into the perturbation suppression effect evaluation model. The effectiveness of perturbation suppression is quantified by comparing indicators such as fluctuation amplitude, dominant frequency drift and phase transition response delay before and after perturbation suppression.

[0096] Based on the actual effect of disturbance suppression, residual fluctuation characteristics of residual fluctuations on local phase change behavior are extracted. These residual fluctuation characteristics include local nucleation delay, vapor holdup recovery deviation, and heat flux density stabilization shift, to characterize the degree of influence of disturbances on the phase change process.

[0097] The residual fluctuation characteristics and the previous phase transition stabilization threshold are input into the Bayesian posterior update mechanism, so that the threshold model can automatically correct the posterior distribution of the critical heat flux density, nucleation density threshold, wettability threshold and liquid replenishment capacity threshold in the new cyclic environment, so that these thresholds are updated synchronously with the strength of the disturbance and the phase transition response.

[0098] Based on the updated posterior distribution, the quantiles of each physical threshold are recalculated according to the preset risk level, so that the quantiles can reflect the true stable boundary under the current cyclic perturbation state, and no longer depend on the original static thresholds.

[0099] The quantiles are integrated to generate an updated set of phase transition stabilization thresholds, including critical heat flux density, nucleation density, wettability threshold, and liquid replenishment capacity threshold. The risk level or applicable priority of each threshold is marked for use as stabilization constraints in the next control cycle.

[0100] Example 2, Figure 2 The present invention provides a centralized liquid cooling method based on phase change enhanced heat transfer, comprising a feature judgment module, a phase change stabilization threshold module, a phase change state module, a circulation loop module, and a disturbance suppression module, with connections between the modules;

[0101] The feature determination module is used to output the spatiotemporal distribution characteristics of heat flux density based on the thermal condition information of the target area, and to identify and determine features based on the coupled field data analysis method.

[0102] The phase change stabilization threshold module is used to determine the heat exchange interface control requirements based on the judgment characteristics, and to perform micro-scale adjustment of the heat exchange interface so that the interface properties meet the phase change stabilization threshold conditions.

[0103] The phase transition state module is used to stabilize the phase transition process by using the phase transition stabilization threshold as a constraint and combining it with the real-time phase transition signal, and outputs the phase transition state information.

[0104] The loop module is used to input phase change state information into the centralized loop control logic, dynamically adjust the loop, and form loop state characteristics.

[0105] The disturbance suppression module is used to actively control the flow fluctuations in the loop based on the characteristics of the loop state and through the disturbance suppression method, and to dynamically update the phase change stabilization threshold based on the disturbance suppression results.

[0106] In detail, the modules in the centralized liquid cooling method device based on phase change enhanced heat transfer described in the embodiments of the present invention employ the same technical means as the centralized liquid cooling method based on phase change enhanced heat transfer shown in the accompanying drawings, and can produce the same technical effects, which will not be repeated here.

[0107] This embodiment includes electronic devices that may include processors, memory, communication buses, and communication interfaces. It may also include computer programs stored in the memory and capable of running on the processor, such as a model generation program for a centralized liquid cooling method based on phase change enhanced heat transfer.

[0108] The processor is the control unit of the electronic device. It connects to various components of the electronic device through various interfaces and lines. It performs various functions of the electronic device and processes data by running or executing programs or modules stored in the memory and calling data stored in the memory.

[0109] The memory includes at least one type of readable storage medium. In some embodiments, the memory may be an internal storage unit of an electronic device, such as a portable hard drive. The memory can be used to store not only application software installed on the electronic device but also various types of data.

[0110] The communication bus is configured to enable communication between the memory and at least one processor.

[0111] The communication interface is used for communication between the aforementioned electronic device and other devices, including a network interface and a user interface.

[0112] The figure only shows an electronic device with components. Those skilled in the art will understand that the structure shown in the figure does not constitute a limitation on the electronic device and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0113] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0114] The distributed energy dispatch optimization model generation program stored in the memory of the electronic device is a combination of multiple instructions. When run in the processor, it can implement the steps in the centralized liquid cooling method based on phase change enhanced heat transfer described above.

[0115] Specifically, the processor's specific implementation system of the above instructions can be found in the description of the relevant steps in the corresponding embodiments of the accompanying drawings, which will not be repeated here.

[0116] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0117] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0118] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0119] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0120] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0121] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A centralized liquid cooling method based on phase change-enhanced heat transfer, characterized in that, include: Based on the thermal condition information of the target area, the spatiotemporal distribution characteristics of heat flux density are output, and the characteristics are identified and judged based on the coupled field data analysis method. Based on the identified characteristics, the heat exchange interface control requirements are determined, and the heat exchange interface is microscale adjusted so that the interface properties meet the phase change stabilization threshold conditions. Using the phase transition stabilization threshold as a constraint, and combining it with the real-time phase transition signal, the phase transition process is stabilized and controlled, and the phase transition state information is output. Phase change state information is input into centralized loop control logic to dynamically adjust the loop and form loop state characteristics; Based on the characteristics of the cyclic state, the flow fluctuations in the cyclic loop are actively controlled by the perturbation suppression method, and the phase transition stabilization threshold is dynamically updated based on the perturbation suppression results.

2. The centralized liquid cooling method based on phase change enhanced heat transfer according to claim 1, characterized in that, The process involves outputting the spatiotemporal distribution characteristics of heat flux density based on the thermal condition information of the target area, and identifying and judging the characteristics based on the coupled field data analysis method, as detailed below: Acquire temperature field distribution, power time series and fluid condition information of the target area, and simultaneously collect phase change data to form multi-source data containing phase change precursor characteristics; Multi-source data is preprocessed and then mapped to a common time scale and spatial coordinate system to form a coupled field data tensor. The spatiotemporal distribution of heat flux density is solved by inverting the coupled field data tensor through a physical information neural network, and the local energy input characteristics are output. Based on the heat flux density distribution and local energy input characteristics, combined with phase transition data, a coupled field feature set is constructed; Dynamic mode decomposition and Koopman modal analysis are performed on the coupled field feature set to identify the dominant dynamic modes and their growth rates, and modal features are extracted from them. Modal features are input into a Bayesian change point detection model to perform a probabilistic assessment of the phase transition state, obtain a phase transition instability risk score with confidence intervals, and identify the key modes that lead to instability. The phase transition instability risk score, key modes and coupled field feature set are fused to generate a phase transition instability judgment feature set.

3. The centralized liquid cooling method based on phase change enhanced heat transfer according to claim 2, characterized in that, The process involves inverting the coupled field data tensor using a physical information neural network to solve for the spatiotemporal distribution of heat flux density and outputting local energy input features, as detailed below: A multimodal feature embedding space is constructed based on self-supervised representation learning, which maps temperature field and phase transition data into a unified latent space representation; A phase transition sensitive feature map is formed by weighting the coupled field data tensor through a spatiotemporal attention mechanism. By comparing and learning phase transition data from different time slices, a time evolution feature sequence is generated. Based on the phase transition sensitive feature map and the time evolution feature sequence, a spatial adjacency graph is constructed and input into a cross-scale graph neural network to obtain the spatiotemporal distribution results of heat flux density; Energy activity clustering analysis was performed on the spatiotemporal distribution of heat flux density, and an energy active area label map was obtained based on local heat flux density peaks, gradient variation amplitudes, and hotspot temporal characteristics. Growth trajectory analysis is performed on the data within the energy-active region. By combining the changes in heat flux density, the phase transition sensitivity feature map, and the time evolution sequence, a local energy input feature set is obtained.

4. The centralized liquid cooling method based on phase change enhanced heat transfer according to claim 3, characterized in that, The energy activity clustering analysis of the spatiotemporal distribution of heat flux density is performed. Based on the local heat flux density peak value, gradient change amplitude, and hotspot temporal characteristics, an energy active area label map is obtained, as follows: For each spatial unit in the spatiotemporal distribution of heat flux density, the maximum heat flux density value is extracted from the entire time series, which is the local heat flux peak value of that unit; By performing differential calculations on the spatial neighborhood of the heat flux density distribution, the magnitude of the change in heat flux density gradient with spatial location is obtained, i.e., the gradient change magnitude. Based on the data sequence of heat flux density changing over time, hotspot temporal features are extracted; The local heat flux peak, gradient change amplitude, and hot spot time sequence characteristics are normalized and combined into a spatial unit feature vector; The similarity of feature vectors of spatial units is calculated based on the dynamic time warping method, and the feature vectors of spatial units are clustered by the hierarchical clustering algorithm. Spatial units with similar heat flux intensity and time evolution characteristics are divided into the same cluster, and the cluster label and intra-cluster statistical characteristics of each spatial unit are obtained. The clustering labels of each spatial unit are mapped back to the original spatial coordinate system to form an energy-active region label map.

5. The centralized liquid cooling method based on phase change enhanced heat transfer according to claim 1, characterized in that, The phase transition stabilization threshold is as follows: A training dataset is constructed based on historical observation data and simulation results; Based on the training data, a Bayesian probability model for instability probability is constructed; By solving the Bayesian probability model using an online variational inference method, the posterior distribution of the input variables is dynamically matched with the observed data, so that the critical heat flux density, nucleation density threshold, wettability threshold, and liquid replenishment capacity threshold converge to the posterior probability distribution in a statistical sense. The posterior probability distribution is input into the posterior prediction model to calculate the probability of phase transition instability caused by each physical parameter at different values, and an instability probability curve is generated. By comparing the range of instability probability increase and the critical change point corresponding to different parameter values, the key turning point of each physical property parameter is identified, and the key turning point is used as the candidate range for subsequent threshold selection. Based on the preset risk level, the corresponding quantiles on the instability probability curve are selected from the candidate range as the critical heat flux density, nucleation density threshold, wettability threshold and liquid replenishment capacity threshold, thereby obtaining the final phase change stabilization threshold.

6. The centralized liquid cooling method based on phase change enhanced heat transfer according to claim 1, characterized in that, The phase transition process is stabilized and controlled by using a phase transition stabilization threshold as a constraint and combining it with real-time phase transition signals, and the phase transition state information is output as follows: Denoising, time synchronization, and short-window feature extraction are performed on the real-time phase transition signal to form the phase transition feature vector at the current moment; The phase transition eigenvector is compared with the phase transition stabilization threshold to output the instantaneous phase transition instability risk score of the current spatial unit; When the instantaneous phase transition instability risk score reaches the warning level, the instantaneous risk score is used as a priori. Combined with the phase transition characteristic sequence of several past moments, the future evolution trend of the phase transition characteristics is predicted based on the short-term prediction model to form a future short-term instability probability estimate. Based on the instantaneous phase transition instability risk score and short-term instability probability estimate, and combined with the boundary conditions provided by the phase transition stabilization threshold, control commands corresponding to the instability trend are generated. The interface is adjusted according to the control command, and the changes in phase change characteristics are monitored in real time during the execution of the action to obtain immediate feedback of the action. Based on the change in the deviation between the action feedback signal and the threshold, it is determined whether the control action reduces the risk of instantaneous phase transition instability. If the risk has not decreased, a rollback adjustment is performed. Phase transition characteristics, instantaneous phase transition instability risk score, short-term instability probability estimation, control actions, and action effects are integrated into phase transition state information.

7. The centralized liquid cooling method based on phase change enhanced heat transfer according to claim 1, characterized in that, The phase transition state information is input into the centralized loop control logic to dynamically adjust the loop and form loop state characteristics, as detailed below: The phase transition state information is input into the centralized loop control logic, and key variables are extracted. The extracted key variables are input into the cycle-phase change coupling mapping model to establish the mapping relationship between local phase change demand and circulation flow rate, pressure regulation direction and vapor-liquid ratio target range, so as to obtain the macro-control demand. Based on the needs of macro-control, the cyclic prediction model outputs the changing trends of cyclic flow rate, vapor-liquid ratio and pressure in the future control cycle, and generates cyclic prediction targets. The cyclic prediction target is input into the control strategy generation module, and an executable cyclic adjustment instruction set is generated by integrating the flow coupling relationship, system stability boundary and energy consumption conditions through constraint solving methods. The operating status of the loop is adjusted in real time according to the cyclic adjustment command, and system response data is collected during the adjustment process to form the current cyclic response sequence; Perform time window analysis on the cyclic response sequence to extract cyclic operation characteristics; By associating and integrating the cyclic operation characteristics with the phase transition state information, cyclic state characteristics are generated.

8. The centralized liquid cooling method based on phase change enhanced heat transfer according to claim 1, characterized in that, Based on the characteristics of the cyclic state, the flow fluctuations in the cyclic loop are actively controlled using a disturbance suppression method, and the phase transition stabilization threshold is dynamically updated based on the disturbance suppression results, as detailed below: Based on the characteristics of the cyclic state, the amplitude, dominant frequency component, phase drift and spatial distribution of the current flow fluctuation are extracted and used as disturbance characterization vectors to input into the disturbance identification model to identify the disturbance pattern. Based on the identified disturbance patterns, the disturbance characteristics are mapped to the corresponding flow control quantities, and the disturbance suppression controller determines the direction and intensity of the active adjustment to be applied. Based on the cyclic prediction model, the timing response of the control action is predicted, and a disturbance suppression command that can be executed in real time in the next control cycle is generated and applied to the cyclic loop for active suppression. During the disturbance suppression process, the pressure drop change, flow recovery rate, fluctuation attenuation coefficient, and local vapor-liquid ratio stabilization state of the circulation loop are collected to form a disturbance response sequence. Input the perturbation response sequence into the perturbation suppression effectiveness evaluation model to quantify the effectiveness of perturbation suppression; Based on the actual effect of perturbation suppression, the residual wave characteristics generated by residual waves on local phase transition behavior are extracted. The residual fluctuation characteristics and the phase transition stabilization threshold are input into the Bayesian posterior update mechanism, so that the threshold model can automatically correct the posterior distribution of the critical heat flux density, nucleation density threshold, wettability threshold and liquid replenishment capacity threshold under the new cyclic environment, thereby obtaining the updated posterior distribution. Based on the updated posterior distribution, the quantiles of each physical threshold are recalculated according to the preset risk level. The quantiles are integrated to generate an updated set of phase transition stabilization thresholds, which are used for stabilization constraints in the next control cycle.

9. A system using a centralized liquid cooling method based on phase change enhanced heat transfer as described in any one of claims 1-8, characterized in that, It includes a feature judgment module, a phase transition stabilization threshold module, a phase transition state module, a loop module, and a disturbance suppression module, and the modules are interconnected. The feature determination module is used to output the spatiotemporal distribution characteristics of heat flux density based on the thermal condition information of the target area, and to identify and determine features based on the coupled field data analysis method. The phase change stabilization threshold module is used to determine the heat exchange interface control requirements based on the judgment characteristics, and to perform micro-scale adjustment of the heat exchange interface so that the interface properties meet the phase change stabilization threshold conditions. The phase transition state module is used to stabilize the phase transition process by using the phase transition stabilization threshold as a constraint and combining it with the real-time phase transition signal, and outputs the phase transition state information. The loop module is used to input phase change state information into the centralized loop control logic, dynamically adjust the loop, and form loop state characteristics. The disturbance suppression module is used to actively control the flow fluctuations in the loop based on the characteristics of the loop state and through the disturbance suppression method, and to dynamically update the phase change stabilization threshold based on the disturbance suppression results.

10. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor to enable the at least one processor to perform the centralized liquid cooling method based on phase change enhanced heat transfer as described in any one of claims 1 to 8.

Images