Two-phase cold plate heat dissipation control method and system based on vibration and flow coordinated regulation and control

Abstract

The invention discloses a two-phase cold plate heat dissipation control method and system based on vibration and flow coordinated regulation and control, and relates to the technical field of liquid cooling heat dissipation. The two-phase cold plate heat dissipation control method and system based on vibration and flow collaborative regulation and control comprises the steps that S1, cold plate collaborative heat dissipation data is collected in real time, and the cold plate collaborative heat dissipation data is preprocessed; s2, the heat dissipation degradation degree of the two-phase cold plate is evaluated, heat dissipation state classification judgment is executed, and a regulation and control instruction is issued; s3, executing the control instruction, evaluating the regulation and control response state of the cooling path, judging the regulation and control effect, and triggering a time-sharing coordinated regulation and control strategy; and S4, whether heat dissipation of the cooling path fails or not is judged, the thermal degradation trend of the two-phase cold plate is evaluated, and a liquid path switching strategy is executed. The problems that in an existing cold plate liquid cooling heat dissipation technology, oil films are generated by cooling liquid due to temperature changes, the oil films are prone to being attached to the inner walls of micro-pore channels, flowing is blocked, the heat exchange efficiency is reduced, and local overheating and unstable equipment performance are further caused are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of liquid cooling heat dissipation, in particular to a two-phase cold plate heat dissipation control method and system based on vibration and flow cooperative regulation. BACKGROUND

[0002] At present, in the thermal management scenarios of high-performance computing, data centers, servers and high heat flux chips, two-phase liquid cooling technology has become an important technical route to solve the chip heat dissipation bottleneck due to its high heat conduction capacity and compact integration characteristics. Among them, the two-phase cold plate as the core component of the hot end undertakes the key task of quickly transferring the heat of the chip to the cooling liquid, and its heat dissipation performance directly affects the stability and energy efficiency ratio of the whole cooling system. Therefore, around the key factors such as cold plate structure design, internal flow channel optimization, fin arrangement mode, etc., there are many technical solutions to improve the heat exchange efficiency and heat dissipation uniformity.

[0003] For example, the invention with publication number CN118119164A provides a liquid cooling heat dissipation plate, which comprises: a first substrate, a plurality of rows and columns of needle fins are formed on the first surface of the first substrate, the first surface of the first substrate also forms an inlet end and an outlet end which are longitudinally spaced apart and provide a flow path for the cooling liquid, a plurality of mounting areas are formed on the second surface of the first substrate, and the plurality of mounting areas are arranged along the transverse direction; at least one set of baffles, each set of baffles is inserted into the gap between the needle fins corresponding to the spacing area, the spacing area is the area between adjacent mounting areas, and each set of baffles comprises a plurality of baffles spaced apart along the longitudinal direction, and each baffle forms a notch facing the outlet end to block the cooling liquid from entering the gap between the needle fins; a second substrate connected to the first surface of the first substrate, the second substrate forms a groove, and the groove forms an inlet hole and an outlet hole. The liquid cooling heat dissipation plate of the invention can improve the heat dissipation efficiency and increase the reusability.

[0004] For example, the invention with publication number CN113365485A discloses a liquid cooling plate radiator, which comprises a radiator body, and a cooling liquid flow channel is formed in the radiator body for circulating flow of the cooling liquid; wherein the cooling liquid flow channel comprises a plurality of heat dissipation sub-flow channels, and a plurality of fin units are arranged on the heat dissipation sub-flow channels along the flow direction of the cooling liquid, the fin units have a plurality of fins extending side by side along the flow direction of the cooling liquid, and the fins of the fin units adjacent in front and back on the heat dissipation sub-flow channels are arranged in a staggered manner, thereby increasing the heat dissipation area and enhancing the disturbance of the cooling liquid when flowing in the heat dissipation sub-flow channel, and improving the heat dissipation efficiency; thereby solving the technical problems of limited heat dissipation area and low heat dissipation efficiency caused by the single straight line type of the cooling liquid flow channel, effectively increasing the contact area of the liquid cooling plate radiator and the cooling liquid, and improving the heat dissipation efficiency by increasing the disturbance of the cooling liquid, and the structure is simple.

[0005] However, although the above scheme has achieved certain results in structure optimization and heat exchange performance improvement, it still mainly focuses on the physical improvement of the cold plate body or the flow channel structure, and fails to conduct in-depth research on the dynamic collaborative control and heat dissipation state perception mechanism of the cooling path from the system level. In actual operation, with the nonlinear fluctuation of the chip heat load, the gradual deterioration of the coolant physical properties, and the frequent occurrence of micro-channel blockage caused by the dynamic change of the oil film, the traditional heat dissipation scheme based on static structure design has been difficult to cope with the complex heat dissipation situation in the multi-path parallel liquid cooling system.

[0006] Therefore, in view of the above problems, there is an urgent need for a two-phase cold plate heat dissipation control method and system based on vibration and flow collaborative regulation. SUMMARY

[0007] Technical problems to be solved In view of the deficiencies of the prior art, the present application provides a two-phase cold plate heat dissipation control method and system based on vibration and flow collaborative regulation, which solves the problem that in the existing cold plate liquid cooling heat dissipation technology, the oil film is generated due to the temperature change of the cooling liquid, these oil films are easy to adhere to the inner wall of the micro-pore, causing flow obstruction and heat exchange efficiency decline, and further causing local overheating and unstable equipment performance.

[0008] Technical scheme

[0009] To achieve the above purpose, the present application realizes the following technical scheme: a two-phase cold plate heat dissipation control method based on vibration and flow collaborative regulation, comprising: S1, real-time acquisition of cold plate collaborative heat dissipation data, and execution of time stamp alignment, sampling reconstruction, abnormal cleaning, noise suppression and normalization processing on the cold plate collaborative heat dissipation data; S2, based on the pre-processed cold plate collaborative heat dissipation data, establishing the spatial mapping relationship between the temperature monitoring area and the cooling path, evaluating the heat dissipation degradation degree of the two-phase cold plate, executing heat dissipation state classification determination, and issuing corresponding regulation instructions according to the classification result; S3, the execution unit receives and executes each control instruction, evaluates the regulation response state of each cooling path, determines the regulation effect, and triggers the dynamic iteration process of the time-sharing collaborative regulation strategy based on the regulation result; S4, continuously monitoring the oil film density evolution to determine whether the cooling path is heat dissipation failure, evaluating the thermal degradation trend of the two-phase cold plate, executing the liquid path switching strategy based on the evaluation result, and extracting the path operation characteristics to construct a strategy evolution sample set, so as to realize the continuous optimization of the cooling strategy.

[0010] Further, the specific steps of collecting the cold plate cooperative heat dissipation data in real time and performing timestamp alignment, sampling reconstruction, abnormal cleaning, noise suppression and normalization processing on the cold plate cooperative heat dissipation data are as follows: collecting the cold plate cooperative heat dissipation data in real time, the cold plate cooperative heat dissipation data including inlet temperature, outlet temperature, cooling liquid flow rate, specific heat capacity of cooling liquid, viscosity of cooling liquid, chip power consumption, chip temperature, shaker frequency, ambient temperature and oil film density; through constructing a time sequence synchronization method based on multi-channel timestamp alignment and sampling interval reconstruction, the cold plate cooperative heat dissipation data is processed uniformly across modules and time; through introducing an abnormal cleaning algorithm combining abnormal point identification and physical constraint screening, the pulse disturbance, sensor drift and physical inconsistency in the cold plate cooperative heat dissipation data are filtered and repaired; through adopting a smoothing denoising method combining sliding window filtering and dynamic threshold monitoring, the high-frequency jitter and measurement noise of the cold plate cooperative heat dissipation data are suppressed; through constructing a normalization processing method based on maximum value mapping and segmented stretching, the cold plate cooperative heat dissipation data is compressed across physical quantities; through introducing a joint algorithm of standard deviation standardization and centralization processing, the cold plate cooperative heat dissipation data is distributed standardized.

[0011] Further, based on the preprocessed cold plate cooperative heat dissipation data, the specific steps of establishing the spatial mapping relationship between the temperature monitoring area and the cooling path and evaluating the heat dissipation degradation degree of the two-phase cold plate are as follows: dividing the chip into multiple temperature monitoring areas, establishing the spatial correlation mapping relationship between the temperature monitoring area and the cooling path according to the two-phase cold plate structure and flow channel configuration; based on the preprocessed cold plate cooperative heat dissipation data, multiplying the cooling liquid flow rate by the difference between the outlet temperature and the inlet temperature, and then dividing by the chip power consumption to obtain the cooling liquid sensible heat transfer efficiency term; dividing the cooling liquid standard viscosity by the cooling liquid viscosity to obtain the flowability influence term; subtracting the oil film density to obtain the microchannel patency correction term; multiplying the cooling liquid sensible heat transfer efficiency term, the flowability influence term and the microchannel patency correction term in turn to obtain the comprehensive heat dissipation capacity term; subtracting the comprehensive heat dissipation capacity term to obtain the heat dissipation degradation degree evaluation value.

[0012] Further, the specific steps of executing the heat dissipation state classification determination and issuing corresponding control instructions according to the classification results are as follows: comparing the heat dissipation degradation degree evaluation value with the heat dissipation threshold value in real time, and classifying and controlling the operating state of the two-phase cold plate to respond: when the heat dissipation degradation degree evaluation value is less than or equal to the first-level heat dissipation threshold value, it is determined that the two-phase cold plate is in a normal heat dissipation state, and the existing heat dissipation configuration is maintained; when the heat dissipation degradation degree evaluation value is greater than the first-level heat dissipation threshold value and less than the second-level heat dissipation threshold value, it is determined that the two-phase cold plate is in a performance degradation state, and a gradual heat dissipation enhancement instruction is issued to the execution unit, and based on the incremental adjustment algorithm of the degradation gradient, the current period and the previous period heat dissipation degradation degree evaluation value change trend is used to adjust the cooling liquid flow and the vibrator frequency by using the adaptive proportional adjustment mechanism; when the heat dissipation degradation degree evaluation value is greater than or equal to the second-level heat dissipation threshold value, it is determined that the two-phase cold plate is in an overheating state, and a path-level local enhancement control is implemented: the real-time chip temperature of each temperature monitoring area is extracted, the temperature rise rate is calculated based on the continuous time difference, the abnormal chip area whose temperature rise rate exceeds the rate threshold value is identified, the corresponding cooling path is marked as a priority control path, and a forced parameter refresh instruction is issued to the execution unit, and the cooling liquid flow and the vibrator frequency of the priority control path are adjusted to the flow upper limit and the frequency upper limit, and the remaining cooling paths maintain the existing heat dissipation configuration.

[0013] Further, the specific steps of the execution unit receiving and executing various control instructions and evaluating the control response state of each cooling path are as follows: the execution unit receives various control instructions, sets the cooling liquid flow and the vibrator frequency to target values; the priority control path is adjusted first by using a higher step level, and the remaining cooling paths are executed according to a regular step level; the cold plate cooperative heat dissipation data after execution is collected within a fixed sampling period to evaluate the control response state of each cooling path: the cooling liquid flow is multiplied by the specific heat capacity of the cooling liquid, and the difference between the outlet temperature and the inlet temperature is multiplied to obtain the cooling liquid sensible heat transfer term; the cooling liquid sensible heat transfer term is divided by the chip power consumption, and the obtained ratio is subtracted by 1 to obtain the sensible heat transfer efficiency term; the cooling liquid viscosity is subtracted by the standard cooling liquid viscosity, and then divided by the standard cooling liquid viscosity to obtain the flowability deviation term; the cooling liquid sensible heat transfer efficiency term and the flowability deviation term are squared and added to obtain the control response deviation evaluation value.

[0014] Further, the specific steps of determining the control effect and triggering the dynamic iteration process of the time-sharing cooperative control strategy based on the control result are as follows: comparing the control response deviation evaluation value with the control deviation threshold value in real time, when the control response deviation evaluation value is less than or equal to the control deviation threshold value, it is determined that the control is effective, and the current cooling liquid flow and the vibrator frequency remain unchanged; when the control response deviation evaluation value is greater than the control deviation threshold value, it is determined that the control is ineffective, and the time-sharing cooperative control is executed: the vibrator frequency is preferentially increased, and the cooling liquid flow is increased after a delay of one sampling period; otherwise, only the vibrator frequency is increased.

[0015] Further, the specific steps of continuously monitoring the oil film density evolution to determine whether the cooling path fails to dissipate heat and evaluating the thermal degradation trend of the two-phase cold plate are as follows: after performing the time-sharing cooperative regulation, the oil film density change corresponding to each cooling path is continuously monitored, and if the oil film density still exceeds the density threshold in the last three sampling periods, the current cooling path is marked as a heat dissipation failure path; all cold plate cooperative heat dissipation data from the start of the regulation response to the period when the heat dissipation failure path is marked are extracted, and the corresponding sampling period number is counted; the current cooling liquid viscosity is reduced by the initial cooling liquid viscosity, and then divided by the product of the sampling period number and the standard viscosity of the cooling liquid, to obtain the cooling liquid viscosity degradation term; for each sampling period, the difference between the chip temperature and the ambient temperature is calculated, and the ratio of the chip power consumption is obtained, to obtain the unit power consumption thermal offset ratio; all unit power consumption thermal offset ratios are squared, and the average value is calculated based on the sampling period number, to obtain the chip thermal load abnormality term; the cooling liquid viscosity degradation term is squared, and then added to the chip thermal load abnormality term, to obtain the thermal degradation risk assessment value.

[0016] Further, the specific steps of executing a liquid path switching strategy based on the evaluation result are as follows: the thermal degradation risk assessment value is compared with the degradation intensity threshold in real time, when the thermal degradation risk assessment value is less than or equal to the degradation intensity threshold, the current regulation strategy configuration is maintained, and the current heat dissipation failure path is recorded as an acceptable risk level path; when the thermal degradation risk assessment value is greater than the degradation intensity threshold, the current cooling path is marked as an unrecoverable path, the liquid path switching strategy is executed, and the original cooling path heat dissipation task is interrupted, and the corresponding thermal load is migrated to the alternative two-phase cold plate area.

[0017] Further, the specific steps of extracting path running features to construct a strategy evolution sample set to realize continuous optimization of the cooling strategy are as follows: according to the marked acceptable risk paths and unrecoverable paths, the corresponding oil film density change, cooling liquid viscosity evolution and chip temperature response data are extracted to construct a regulation behavior feature set; the similarity between the regulation behavior features of the current cooling path and the historical path features in the regulation behavior feature set is calculated in real time, if the similarity exceeds the matching threshold, the optimization control instruction is generated according to the historical path associated cooling liquid flow rate set value and the vibrator frequency set value; otherwise, all cold plate cooperative heat dissipation data and regulation states of the current cooling path are constructed as strategy training samples and stored in the strategy training sample library for subsequent cooling strategy evolution and optimization.

[0018] The second aspect of the present application provides a two-phase cold plate heat dissipation control system based on vibration and flow collaborative regulation, comprising: a cold plate heat dissipation data acquisition and preprocessing module, a cold plate thermal state determination and regulation decision module, a control instruction execution and response evaluation module, and a cooling degradation identification and strategy optimization module, wherein: the cold plate heat dissipation data acquisition and preprocessing module is used to collect cold plate collaborative heat dissipation data in real time, and perform timestamp alignment, sampling reconstruction, abnormal cleaning, noise suppression and normalization processing on the cold plate collaborative heat dissipation data; the cold plate thermal state determination and regulation decision module is used to establish a spatial mapping relationship between the temperature monitoring area and the cooling path based on the preprocessed cold plate collaborative heat dissipation data, evaluate the heat dissipation degradation degree of the two-phase cold plate, perform heat dissipation state classification determination, and issue corresponding regulation instructions according to the classification results; the control instruction execution and response evaluation module is used for receiving and executing various control instructions by the execution unit, evaluating the regulation response state of each cooling path, determining the regulation effect, and triggering the dynamic iteration process of the time-sharing collaborative regulation strategy based on the regulation result; the cooling degradation identification and strategy optimization module is used to continuously monitor the oil film density evolution to determine whether the cooling path is heat dissipation failure, evaluate the thermal degradation trend of the two-phase cold plate, execute the liquid path switching strategy based on the evaluation result, and extract the path operation characteristics to construct a strategy evolution sample set, so as to realize the continuous optimization of the cooling strategy.

[0019] Advantages The present application has the following advantages: (1) The two-phase cold plate heat dissipation control method and system based on vibration and flow collaborative regulation, by establishing a spatial mapping relationship between the temperature monitoring area and the cooling path, integrating the cooling liquid sensible heat transfer efficiency, the flowability influence term and the microchannel patency correction term, establishing a heat dissipation degradation degree evaluation model, realizing the fine identification and classification control of the cooling path heat dissipation state, and improving the heat dissipation diagnosis accuracy and regulation basic response ability of the system.

[0020] (2) The two-phase cold plate heat dissipation control method and system based on vibration and flow collaborative regulation, by continuously monitoring the chip temperature evolution trend, identifying the local temperature rise abnormal area, dynamically determining the priority regulation path, and based on the time-sharing collaborative control strategy of vibration and flow parameters, quickly enhancing the local heat dissipation capacity, effectively dealing with the thermal unevenness and local thermal runaway problem of high heat density load area.

[0021] (3) The two-phase cold plate heat dissipation control method and system based on vibration and flow collaborative regulation, a regulation response deviation evaluation mechanism is constructed to determine whether the existing regulation scheme is effective, and when the response fails, a time-sharing regulation strategy of vibration first and then flow is triggered, realizing the ordered linkage and iterative optimization of vibration frequency and cooling flow, and improving the regulation stability and strategy convergence ability of the system under complex heat load changes.

[0022] (4) The two-phase cold plate heat dissipation control method and system based on vibration and flow collaborative regulation, by real-time monitoring of oil film density and cooling liquid viscosity evolution, a thermal degradation risk assessment index is established to determine whether the cooling path has a heat dissipation degradation trend, and when the degradation is serious, liquid path switching and task migration are performed, combined with the strategy sample library to realize continuous evolution of cooling path management and control strategy, and the fault tolerance and operation reliability of the system are enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 A flowchart of the two-phase cold plate heat dissipation control method based on vibration and flow collaborative regulation; Figure 2 A structure diagram of the two-phase cold plate heat dissipation control system based on vibration and flow collaborative regulation; Figure 3 A heat dissipation degradation evaluation diagram of the temperature monitoring area; Figure 4 A two-phase cold plate cooling structure composition schematic diagram.

[0024] In the figure, 1 is an inlet pipe opening, 2 is a vibrator, 3 is an outlet pipe opening, 4 is a two-phase cold plate, and 5 is a chip. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0026] Please refer to Figures 1-4 The embodiments of the present application provide a technical solution: a two-phase cold plate heat dissipation control method based on vibration and flow collaborative regulation, comprising: S1, real-time acquisition of cold plate collaborative heat dissipation data, and execution of timestamp alignment, sampling reconstruction, abnormal cleaning, noise suppression and normalization processing on the cold plate collaborative heat dissipation data; S2, based on the pre-processed cold plate collaborative heat dissipation data, a spatial mapping relationship between the temperature monitoring area and the cooling path is established, the heat dissipation degradation degree of the two-phase cold plate 4 is evaluated, the heat dissipation state classification is determined, and the corresponding regulation instruction is issued according to the classification result; S3, the execution unit receives and executes each control instruction, evaluates the regulation response state of each cooling path, determines the regulation effect, and triggers the dynamic iteration process of the time-sharing collaborative regulation strategy based on the regulation result; S4, continuously monitor the oil film density evolution to determine whether the cooling path is heat dissipation failure, evaluate the thermal degradation trend of the two-phase cold plate 4, execute the liquid path switching strategy based on the evaluation result, and extract the path operation characteristics to build a strategy evolution sample set to realize continuous optimization of the cooling strategy.

[0027] Specifically, the specific steps of real-time acquisition of cold plate cooperative heat dissipation data and performing timestamp alignment, sample reconstruction, abnormal cleaning, noise suppression and normalization processing on the cold plate cooperative heat dissipation data are as follows: real-time acquisition of cold plate cooperative heat dissipation data, the cold plate cooperative heat dissipation data including inlet temperature, outlet temperature, cooling liquid flow rate, specific heat capacity of cooling liquid, viscosity of cooling liquid, chip power consumption, chip temperature, shaker frequency, environmental temperature and oil film density. The acquisition method of each data is as follows: the inlet temperature and outlet temperature are acquired by high-precision thermocouple temperature sensors arranged at the inlet pipe port 1 and outlet pipe port 2 of the two-phase cold plate 4; the volume flow rate of the cooling liquid per unit time is collected by dynamic detection through the integration of a micro turbine flow sensor in the cooling liquid inlet path to obtain the cooling liquid flow rate; the specific heat capacity of the cooling liquid is calculated by calling the physical property fitting model in combination with the current sampling period environmental temperature, composition density parameters and the cooling liquid physical property database; the cooling liquid viscosity is obtained according to the relationship between the cooling liquid shear rate and stress by installing an online rotary micro viscosity sensor; the chip power consumption is obtained by period integration based on the power management unit output data combined with the current and voltage real-time sampling results; the chip temperature is obtained by arranging a micro thermistor array on the surface of the chip 5 to collect local temperature point data and calculating the chip temperature by a weighted algorithm; the shaker frequency is obtained by returning the frequency parameter from the vibration control unit and recording the current working frequency in real time; the environmental temperature is obtained by arranging a thermistor temperature probe at the external ventilation environment of the cold plate to collect the cooling environment parameters; the oil film density is obtained by measuring the oil film thickness and distribution through a capacitive thin film sensor arranged in the cooling path combined with the liquid density parameters. By constructing a time sequence synchronization method based on multi-channel timestamp alignment and sampling interval reconstruction, the cold plate cooperative heat dissipation data is time aligned and data frame unified processed, and a synchronization strategy including channel identification embedding, time interpolation mapping and nonlinear drift compensation is adopted, so that the cold plate cooperative heat dissipation data can be aligned with high precision under the condition of sampling error in each channel, and the consistency of the data basis for collaborative analysis is ensured. By introducing an abnormal cleaning algorithm combining abnormal point identification and physical constraint screening, the pulse disturbance, sensor drift and physically inconsistent values in the cold plate cooperative heat dissipation data are filtered and repaired, wherein the abnormal point identification is based on a data statistical distribution model and a time sequence jump detection mechanism, and the physical constraints include the heat exchange correlation constraint between the cooling liquid flow rate and the chip power consumption, the change function relationship between the cooling liquid viscosity and the temperature, etc., to ensure that the cold plate cooperative heat dissipation data has physical reasonableness and numerical continuity after cleaning. By adopting a smoothing denoising method combining sliding window filtering and dynamic threshold monitoring, the high-frequency jitter and measurement noise of the cold plate cooperative heat dissipation data are suppressed, the sliding window filtering dynamically adjusts the window length to adapt to the different change rates of the cooling liquid flow rate, chip temperature and shaker frequency, and the dynamic threshold mechanism adjusts the denoising strength in real time according to the sampling stability coefficient, thereby improving the resolvability and stability of the cold plate cooperative heat dissipation data.By constructing the normalization processing method based on the maximum value mapping and the segmented stretching, the cross-physical quantity numerical compression of the cold plate collaborative heat dissipation data is performed, and the analysis applicability under the unified dimension is ensured. By introducing the joint algorithm of the standard deviation standardization and the centralization processing, the distribution standardization of the cold plate collaborative heat dissipation data is performed. The long-term trend drift of the coolant flow, the chip power consumption and the vibrator frequency is processed by using the sliding mean value as the center of the zero mean centralization strategy. The scale scaling of the coolant viscosity, the oil film density and the temperature type variable is realized in combination with the standard deviation constraint. All the cold plate collaborative heat dissipation data can be input into the subsequent analysis model under the unified distribution structure, and the model stability and the generalization ability are improved.

[0028] In the embodiment, by constructing the methods of the multi-channel timestamp alignment, the sampling interval reconstruction, the abnormal point identification and the physical constraint screening, the sliding window filtering and the dynamic threshold monitoring, the maximum value mapping and the segmented stretching, the standard deviation standardization and the centralization processing, the inlet temperature, the outlet temperature, the coolant flow, the specific heat capacity of the coolant, the viscosity of the coolant, the chip power consumption, the chip temperature, the vibrator frequency, the environmental temperature and the oil film density in the cold plate collaborative heat dissipation data are comprehensively pretreated. The synchronization, the reliability, the stability and the standardization level of the data are significantly improved, and high-quality input is provided for the subsequent heat dissipation performance evaluation and the control strategy.

[0029] Specifically, based on the pre-processed cold plate cooperative heat dissipation data, the spatial mapping relationship between the temperature monitoring area and the cooling path is established, and the specific steps of evaluating the heat dissipation degradation degree of the two-phase cold plate 4 are as follows: the chip 5 is divided into multiple temperature monitoring areas, which correspond to the heat source distribution units in the physical structure of the chip 5 respectively, and combined with the multi-channel flow channel configuration and structural parameters inside the cold plate, according to the structure connection mode of the two-phase cold plate 4, the cooling liquid flow direction and the flow channel topological relationship, a one-to-one spatial correlation mapping relationship between the temperature monitoring area and the cooling path is established, and the accurate modeling of the heat conduction path is realized; based on the pre-processed cold plate cooperative heat dissipation data, including the inlet temperature, the outlet temperature, the cooling liquid flow, the chip power consumption, the cooling liquid sensible heat transfer efficiency item is calculated: the cooling liquid flow is multiplied by the temperature difference value between the outlet temperature and the inlet temperature, and then the result is divided by the chip power consumption value to obtain the unit power consumption sensible heat carrying capacity; the cooling liquid standard viscosity value is divided by the current cooling liquid viscosity to obtain the flowability influence item, which is used to reflect the deviation degree of the current cooling liquid flow characteristics relative to the ideal state; wherein, the cooling liquid standard viscosity value refers to the reference viscosity value measured under the standard working condition when the cooling liquid is in the optimal flow state, which is used as the benchmark for judging the current running viscosity; one is subtracted from the oil film density parameter value to obtain the microchannel patency correction item, which is used to quantify the influence of the micro resistance caused by the adhesion behavior of the oil film between the cooling liquid and the inner wall of the channel on the heat dissipation performance; the cooling liquid sensible heat transfer efficiency item, the flowability influence item and the microchannel patency correction item are sequentially multiplied to obtain the comprehensive heat dissipation capacity item, which reflects the overall heat exchange capacity after the coupling effect of the multi-dimensional heat dissipation influence factor; finally, one is subtracted from the comprehensive heat dissipation capacity item to obtain the heat dissipation degradation degree evaluation value, which is used as a quantitative index to characterize the heat performance degradation degree of the cooling path corresponding to the current temperature monitoring area, and provides accurate judgment basis for subsequent state classification and strategy regulation.

[0030] wherein the specific calculation formula of the heat dissipation degradation degree evaluation value is: ; In the formula, represents the heat dissipation degradation degree evaluation value, represents the cooling liquid flow, represents the outlet temperature, represents the inlet temperature, represents the chip power consumption, represents the cooling liquid viscosity, represents the cooling liquid standard viscosity, represents the oil film density.

[0031] In this embodiment, Table 1 is a data table of heat dissipation degradation assessment values, which shows the cold plate collaborative heat dissipation data and assessment results of the five temperature monitoring areas in the current control cycle, covering the values ​​of coolant flow rate, chip power consumption, coolant standard viscosity, coolant viscosity, oil film density, inlet temperature, outlet temperature and heat dissipation degradation assessment values. The specific details are as follows: In Region 1, the coolant flow rate is 3.080, the chip power consumption is 120, the standard viscosity of the coolant is 1.5, the coolant viscosity is 1.836, the oil film density is 0.211, the inlet temperature is 40°C, and the outlet temperature is 56.5°C. The calculated heat dissipation degradation assessment value is 0.73. In Region 2, the coolant flow rate is 6.387, the chip power consumption is 120, the standard viscosity of the coolant is 1.5, the coolant viscosity is 1.619, the oil film density is 0.294, the inlet temperature is 40°C, and the outlet temperature is 45.5°C. The calculated heat dissipation degradation assessment value is 0.82. In Region 3, the coolant flow rate is 7.312, the chip power consumption is 120, the standard viscosity of the coolant is 1.5, and the coolant viscosity is 1. In region 737, the oil film density is 0.170, the inlet temperature is 40°C, and the outlet temperature is 45.4°C, resulting in a calculated heat dissipation degradation assessment value of 0.76. In region 4, the coolant flow rate is 3.436 L, the chip power consumption is 120 kW, the standard viscosity of the coolant is 1.5, the coolant viscosity is 1.970, the oil film density is 0.293, the inlet temperature is 40°C, and the outlet temperature is 45.8°C, resulting in a calculated heat dissipation degradation assessment value of 0.91. In region 5, the coolant flow rate is 3.079 L, the chip power consumption is 120 kW, the standard viscosity of the coolant is 1.5, the coolant viscosity is 1.631, the oil film density is 0.193, the inlet temperature is 40°C, and the outlet temperature is 57.5°C, resulting in a calculated heat dissipation degradation assessment value of 0.67.

[0032] Table 1. Data Table of Heat Dissipation Deterioration Assessment Values

[0033] like Figure 3 The figure shows the changes in the heat dissipation degradation assessment values ​​of five temperature monitoring zones, reflecting the thermal performance of the two-phase cold plate 4 in different zones. The solid line graph represents the heat dissipation degradation assessment values ​​for each zone, reflecting the differences in heat dissipation status between different zones. Two threshold lines are set in the figure to assist in judgment: the first-level heat dissipation threshold is a thin dashed line, used to determine the starting point of performance degradation; the second-level heat dissipation threshold is a thick dashed line, used to represent the boundary between overheating and severe degradation. It can be seen from the figure that zones 1 and 5 are in a normal heat dissipation state, indicating good heat dissipation; zones 2 and 3 are in a state of performance degradation for the two-phase cold plate 4, requiring appropriate enhanced control; zone 4 is in an overheated state, and priority control measures should be implemented immediately to prevent local thermal runaway.

[0034] In the embodiment, by establishing a spatial mapping relationship between the temperature monitoring area and the cooling path, the cooling liquid flow rate, the inlet temperature, the outlet temperature, the chip power consumption, the cooling liquid viscosity, the cooling liquid standard viscosity, and the oil film density are introduced to form a multi-parameter calculation system, and the cooling plate collaborative heat dissipation data are introduced to realize the fine modeling of the cooling liquid sensible heat transfer efficiency term, the flowability influence term, and the micro-channel smoothness correction term; the comprehensive heat dissipation capacity term is constructed by multiplying a plurality of indexes, and the heat dissipation degradation degree evaluation value is obtained in reverse, which can accurately reflect the degradation trend of the heat exchange performance in the temperature monitoring area, significantly improve the accuracy of the heat dissipation performance analysis and the pertinence of the control strategy execution, and ensure stronger heat dissipation safety and adaptability under the condition of multi-factor interference.

[0035] Specifically, the specific steps of performing heat dissipation state classification determination and issuing corresponding control instructions according to the classification results are as follows: based on the heat dissipation degradation degree evaluation value corresponding to each temperature monitoring area at the current moment, the numerical relationship between the heat dissipation degradation degree evaluation value and the first-level heat dissipation threshold value and the second-level heat dissipation threshold value is compared in real time, the operating state of the two-phase cold plate 4 is accurately classified, and the corresponding control response mechanism is triggered: when the heat dissipation degradation degree evaluation value is less than or equal to the first-level heat dissipation threshold value, it is determined that the two-phase cold plate 4 is in a normal heat dissipation state, and all cooling paths maintain the original configuration of the current cooling liquid flow rate set value and the vibrator frequency set value; when the heat dissipation degradation degree evaluation value is greater than the first-level heat dissipation threshold value and less than the second-level heat dissipation threshold value, it is determined that the two-phase cold plate 4 is in a heat dissipation performance decline state, and a gradual heat dissipation enhancement instruction is immediately issued to the execution unit, which is calculated based on the incremental adjustment algorithm of the degradation gradient, dynamically fuses the heat dissipation degradation degree evaluation value change amplitude corresponding to the current sampling period and the previous sampling period, and adjusts the cooling liquid flow rate set value and the vibrator frequency set value of the corresponding cooling path respectively to realize the heat transfer capacity improvement; when the heat dissipation degradation degree evaluation value is greater than or equal to the second-level heat dissipation threshold value, it is determined that the two-phase cold plate 4 is in an overheating state, and the path-level local enhancement control mechanism is immediately activated: the chip temperature 5 corresponding to all temperature monitoring areas is extracted in real time, the temperature rise rate value of the chip temperature rise is obtained based on the continuous time difference calculation model, the chip 5 area whose temperature rise rate exceeds the set rate threshold value is identified, and then the corresponding cooling path is marked as a priority control path, and a forced parameter refresh instruction is immediately issued to the execution unit to adjust the cooling liquid flow rate set value and the vibrator frequency set value of the priority control path to the flow rate upper limit and the frequency upper limit, and the remaining cooling paths that are not determined to be abnormal maintain the current heat dissipation parameter configuration unchanged, so as to realize the priority response control of the hot spot area and the stable maintenance of the overall heat balance.

[0036] In the embodiment, by constructing a classification control response mechanism based on the heat dissipation degradation degree evaluation value, fine identification and precise control of the operating state of the two-phase cold plate 4 are realized. Based on the real-time comparison result between the heat dissipation degradation degree evaluation value and the first-level heat dissipation threshold value and the second-level heat dissipation threshold value, the normal heat dissipation state, the heat dissipation performance decline state and the overheating state are strictly divided into three operating categories, and the cooling path is ensured to obtain differential regulation and response under different operating conditions. By issuing progressive heat dissipation enhancement instructions and forced parameter refresh instructions to the execution unit, adaptive fine-tuning control of the cooling liquid flow rate set value and the vibrator frequency set value in the performance decline state is realized, and the upper limit refresh control of the cooling liquid flow rate set value and the vibrator frequency set value of the preferential regulation path in the overheating state is realized, which ensures that the abnormal heat area can be preferentially intervened and responded, effectively improves the dynamic regulation and control capability and safety fault tolerance performance of the heat management, and further enhances the stability, adaptability and energy efficiency optimization level of the two-phase cold plate 4 in the high-load continuous operation scene.

[0037] Specifically, the execution unit receives and executes each control instruction, and the specific steps of evaluating the regulation and response state of each cooling path are as follows: the execution unit receives each control instruction issued based on the heat dissipation state classification determination result, and configures the cooling liquid flow rate set value and the vibrator frequency set value as target set values according to the instruction requirements; for the cooling path marked as the preferential regulation path, a higher step level is used for preferential parameter adjustment, and the remaining cooling paths use a conventional step level to execute the parameter adjustment process; within a fixed sampling period, the execution unit synchronously collects cold plate cooperative heat dissipation data, including cooling liquid flow rate, cooling liquid specific heat capacity, outlet liquid temperature, inlet liquid temperature, chip power consumption, cooling liquid viscosity and cooling liquid standard viscosity, and performs regulation and response effect evaluation on the collected cold plate cooperative heat dissipation data; first, the cooling liquid flow rate is multiplied by the cooling liquid specific heat capacity, and the difference between the outlet liquid temperature and the inlet liquid temperature is multiplied, to calculate the cooling liquid sensible heat transfer term; then, the cooling liquid sensible heat transfer term is divided by the chip power consumption, and the sensible heat transfer efficiency term is obtained by subtracting the ratio; the cooling liquid viscosity is subtracted from the cooling liquid standard viscosity, and then divided by the cooling liquid standard viscosity, to obtain the flowability deviation term reflecting the change trend of the cooling liquid dynamic performance; the square of the sensible heat transfer efficiency term and the flowability deviation term is executed respectively, and the square results of the two terms are added, to finally obtain the regulation and response deviation evaluation value for determining the dynamic response capability and regulation accuracy of the cooling path.

[0038] wherein, the specific calculation formula of the regulation and response deviation evaluation value is: ; In the formula, represents the regulation and response deviation evaluation value, represents the cooling liquid flow rate, represents the cooling liquid specific heat capacity, represents the temperature of the liquid outlet, represents the temperature of the liquid inlet, represents the power consumption of the chip, represents the viscosity of the cooling liquid, represents the standard viscosity of the cooling liquid.

[0039] In the embodiment, the execution unit receives and executes the control instructions, and dynamically analyzes the cooling liquid cooperatively cooling data, so that the response state of the cooling path is accurately quantitatively evaluated. The combined evaluation method of the sensible heat transfer efficiency term and the flow deviation term can comprehensively reflect the change trend of the cooling liquid heat transfer capacity and flow performance. The quantitative feedback of the cooling path adjustment effect is realized by regulating the response deviation evaluation value, so that the present application not only has a real-time response mechanism, but also has a dynamic tuning capability, thereby improving the stability and adaptability of the cooling system, and significantly enhancing the heat dissipation reliability and control accuracy of the two-phase cooling plate 4.

[0040] Specifically, the specific steps of determining the regulation effect and triggering the dynamic iteration process of the time-sharing cooperative regulation strategy based on the regulation result are as follows: comparing the regulation response deviation evaluation value and the regulation deviation threshold value in real time, when the regulation response deviation evaluation value is less than or equal to the regulation deviation threshold value, it is determined that the control strategy in the current period is effective, and the target cooling liquid flow and the target vibrator frequency are maintained unchanged; when the regulation response deviation evaluation value is greater than the regulation deviation threshold value, it is determined that the current control strategy does not achieve the regulation target, and enters the time-sharing cooperative regulation stage: the vibrator frequency parameter is preferentially improved, the oil film density influence is improved by enhancing the micro-channel disturbance, and the cooling liquid flow parameter is simultaneously improved after a complete sampling period is delayed, so as to enhance the heat convection heat transfer effect; when the delay condition is not met, only the vibrator frequency parameter is improved, so as to control the step response rhythm of the adjustment rate and realize the dynamic feedback regulation of the cooling path.

[0041] In the embodiment, by comparing the regulation response deviation evaluation value and the regulation deviation threshold value in real time, the dynamic adjustment process of the cooling liquid flow and the vibrator frequency has a clear determination mechanism, a rhythm control mechanism and a target feedback mechanism. By adjusting the vibrator frequency and the cooling liquid flow in sequence, the system can effectively identify the control lag problem in the regulation failure scene and improve the response sensitivity to the cooling liquid cooperatively cooling data. The above mechanism improves the identification accuracy of the two-phase cooling plate heat dissipation control system to the invalid state of regulation, ensures the progressive, stability and convergence of the time-sharing cooperative regulation strategy, so as to realize the stable control of the chip temperature and the continuous enhancement of the local heat dissipation performance.

[0042] Specifically, the specific steps of continuously monitoring the oil film density evolution to determine whether the cooling path fails to dissipate heat and evaluating the thermal degradation trend of the two-phase cold plate 4 are as follows: after completing the time-sharing cooperative regulation of the cooling liquid flow and the vibrator frequency, the oil film density in the cold plate cooperative heat dissipation data is continuously collected based on a fixed sampling period mechanism, and the change sequence of the oil film density corresponding to each cooling path is dynamically tracked and analyzed; if the monitored oil film density in the last three sampling periods exceeds the oil film density threshold value, the cooling path is determined as a heat dissipation failure path, and is marked as a heat dissipation failure path state; then, all the cold plate cooperative heat dissipation data from the time when the first regulation response instruction is triggered to the time when the cooling path is marked as a heat dissipation failure path state is extracted, and the total number of sampling periods in the time period is obtained. Based on the sampling period data, the following calculation process is performed: the cooling liquid viscosity monitored at the current time is subtracted from the cooling liquid viscosity at the initial time, and the result is divided by the product of the number of sampling periods and the standard viscosity of the cooling liquid, to obtain the cooling liquid viscosity degradation term, which is used to represent the influence of the cooling liquid property degradation trend on the heat exchange capacity; for each sampling period, the difference between the chip temperature and the ambient temperature is calculated, and the ratio of the chip power consumption is obtained, to obtain the unit power consumption thermal offset ratio, which is used as a thermal load characteristic term of the heat dissipation abnormality; further, the unit power consumption thermal offset ratio in all sampling periods is squared, and the mean value is calculated according to the number of sampling periods, to form the chip thermal load abnormality term, which accurately depicts the degree of deviation of the local thermal environment of the chip 5 from the stable working condition; finally, the cooling liquid viscosity degradation term is squared and added to the chip thermal load abnormality term, to form the thermal degradation risk assessment value, which is used to quantitatively evaluate the overall thermal performance degradation level of the two-phase cold plate 4 on the heat dissipation regulation failure path, and provides a decision basis for the liquid path switching strategy.

[0043] wherein the specific calculation formula of the thermal degradation risk assessment value is: ; In the formula, denotes the thermal degradation risk assessment value, denotes the current cooling liquid viscosity, denotes the initial cooling liquid viscosity, denotes the standard viscosity of the cooling liquid, denotes the chip temperature in the i th sampling period, denotes the ambient temperature in the i th sampling period, denotes the chip power consumption in the i th sampling period, denotes the number of sampling periods.

[0044] In the embodiment, by continuously monitoring the oil film density in the cold plate cooperative heat dissipation data, and combining the cooling liquid viscosity, the cooling liquid standard viscosity, the cooling liquid initial viscosity, the chip temperature, the environmental temperature, the chip power consumption and the sampling period number, a heat dissipation failure path recognition mechanism and a heat degradation risk evaluation model based on the dynamic evolution of the oil film density and the multi-parameter cross calculation are constructed, the accurate determination of whether the cooling path is heat dissipation failure and the quantitative analysis of the four heat degradation trends of the two-phase cold plate are realized, the early identification ability of the heat dissipation abnormal state and the decision effectiveness of the liquid path control strategy are improved, and the high reliability and response accuracy of the thermal management system under complex operating conditions are ensured.

[0045] Specifically, based on the comparison result between the heat degradation risk evaluation value and the degradation intensity threshold value, the specific steps of the liquid path switching strategy are as follows: comparing the heat degradation risk evaluation value and the degradation intensity threshold value in real time, when the heat degradation risk evaluation value is less than or equal to the degradation intensity threshold value, it is considered that the heat dissipation performance of the current cooling path has a degradation trend but is still within the controllable range, the cooling liquid flow and the vibrator frequency are maintained unchanged, the current cooling path is marked as an acceptable risk level path and the degradation characteristic information is recorded; when the heat degradation risk evaluation value is greater than the degradation intensity threshold value, it is considered that the heat dissipation capacity of the current cooling path is not recoverable, and the current cooling path is immediately marked as an unrecoverable path, and the liquid path switching strategy is executed, the heat dissipation task of the current cooling path is interrupted, the cooling liquid flow and the vibrator 2 driving are stopped, and the corresponding heat load is migrated to other two-phase cold plate 4 regions which are not marked as unrecoverable paths, so as to realize continuous heat dissipation and ensure the continuity and stability of thermal management.

[0046] In the embodiment, by comparing the heat degradation risk evaluation value and the degradation intensity threshold value in real time, the quantitative determination of the heat degradation risk of the cooling path is realized, when the heat degradation risk evaluation value is within the acceptable range, the cooling liquid flow and the vibrator frequency setting value are maintained unchanged, and the current cooling path is recorded as an acceptable risk level path; when the heat degradation risk evaluation value exceeds the degradation intensity threshold value, the current cooling path is marked as an unrecoverable path, and the liquid path switching strategy is immediately executed to terminate the cooling liquid flow and the vibrator 2 driving of the cooling path, and the corresponding heat load is migrated to other two-phase cold plate 4 regions which are not marked as unrecoverable paths, thereby improving the robustness and fault tolerance of the heat dissipation configuration response, effectively avoiding the device stability risk caused by local overheating, and further enhancing the adaptive control ability of the heat dissipation system in the degradation evolution process.

[0047] Specifically, the specific steps of the extraction path running feature construction strategy evolution sample set to realize the continuous optimization of the cooling strategy are as follows: according to the cooling path marked as an acceptable risk path and an unrecoverable path through the comparison result of the thermal degradation risk assessment value and the degradation intensity threshold value, the oil film density change data, the cooling liquid viscosity evolution data and the chip temperature response data in the cooling liquid cooperative heat dissipation data associated with the cooling path are extracted respectively, and a regulation behavior feature set with stable structure and clear feature parameters is constructed based on the three types of data; the multidimensional similarity between the regulation behavior feature of the current cooling path and the historical path feature in the regulation behavior feature set is calculated in real time, if the similarity exceeds the matching threshold, the cooling liquid flow rate set value and the vibrator frequency set value associated with the historical path are extracted, the optimization control instruction of the target cooling path is generated according to the optimization parameters, and the cooling configuration update is completed; if the similarity index value of the regulation behavior feature of the current cooling path and all historical paths does not exceed the matching threshold, all cooling liquid cooperative heat dissipation data of the current cooling path in the current regulation period and the corresponding regulation response state parameters are jointly constructed into a complete strategy training sample, and are stored in the strategy training sample library in time sequence as a data source for the cooling strategy evolution and control rule update.

[0048] In the embodiment, by constructing the regulation behavior feature set and classifying the cooling path state according to the thermal degradation risk assessment value and the degradation intensity threshold value, the setting of the cooling liquid flow rate and the vibrator frequency is more in line with the historical effective regulation experience. When the cooling path feature meets the similarity requirement, the historical setting parameters can be accurately called to generate the optimization control instruction; when it does not meet the requirement, it is automatically stored as a strategy evolution sample, continuously enriches the training data, and improves the adaptability and intelligent level of the cooling strategy.

[0049] As Figure 2As shown, the second aspect of the present application provides a two-phase cold plate heat dissipation control system based on vibration and flow cooperative regulation, comprising: a cold plate heat dissipation data acquisition and preprocessing module, a cold plate thermal state determination and regulation decision module, a control instruction execution and response evaluation module, and a cooling degradation identification and strategy optimization module, wherein: the cold plate heat dissipation data acquisition and preprocessing module is used to collect cold plate cooperative heat dissipation data in real time, and perform timestamp alignment, sampling reconstruction, abnormal cleaning, noise suppression and normalization processing on the cold plate cooperative heat dissipation data; the cold plate thermal state determination and regulation decision module is used to establish a spatial mapping relationship between the temperature monitoring area and the cooling path based on the preprocessed cold plate cooperative heat dissipation data, evaluate the heat dissipation degradation degree of the two-phase cold plate 4, perform heat dissipation state classification determination, and issue corresponding regulation instructions according to the classification results; the control instruction execution and response evaluation module is used for receiving and executing various control instructions by the execution unit, evaluating the regulation response state of each cooling path, determining the regulation effect, and triggering the dynamic iteration process of the time-sharing cooperative regulation strategy based on the regulation result; the cooling degradation identification and strategy optimization module is used to continuously monitor the oil film density evolution to determine whether the cooling path is heat dissipation failure, evaluate the thermal degradation trend of the two-phase cold plate 4, execute the liquid path switching strategy based on the evaluation result, and extract the path operation characteristics to construct a strategy evolution sample set, so as to realize the continuous optimization of the cooling strategy.

[0050] As shown, Figure 4 As shown, the region division and key component layout of the two-phase cold plate 4 in the present application are clearly shown, which provides structural support basis for the heat dissipation control principle, specifically including: the liquid inlet pipe port 1, through which the cooling liquid enters the two-phase cold plate 4 as the starting position of the heat transfer medium flow; the vibrator 2 is arranged on the liquid inlet passage, which is used to adjust the flow state of the cooling liquid, enhance the local disturbance and flow uniformity, and realize the convection enhancement and heat exchange efficiency improvement; the liquid outlet pipe port 3, through which the cooling liquid is discharged after completing the sensible heat and latent heat absorption, realizing the complete cooling cycle; the two-phase cold plate 4 is the core heat exchange component, which is internally arranged with multiple groups of parallel microchannel structures for absorbing chip heat load; the chip 5 is the passive heat dissipation target, and the heat generated by operation is quickly exported through the two-phase cold plate 4. This figure shows the spatial relationship between each structure through wireframe three-dimensional perspective, highlights the nested layout of the vibrator 2 and the flow passage system, and the direct contact relationship between the chip 5 and the two-phase cold plate 4, which provides a clear basis for subsequent cooling path division, regulation region setting and thermal state mapping.

[0051] In the embodiment, a two-phase cold plate heat dissipation control system based on vibration and flow cooperative regulation is constructed to form a closed-loop thermal management mechanism from data acquisition to strategy optimization. The system relies on multiple source time series data in the cold plate cooperative heat dissipation data, such as inlet temperature, outlet temperature, cooling liquid flow, specific heat capacity of cooling liquid, viscosity of cooling liquid, chip power consumption, chip temperature, vibrator frequency, environmental temperature and oil film density, to complete the whole process operation of data acquisition and preprocessing, thermal state determination and regulation decision, control instruction execution and response evaluation, cooling degradation identification and strategy optimization in stages. Through the spatial mapping between the temperature monitoring area and the cooling path, the degree of heat dissipation degradation is accurately identified; through the comparison of hierarchical threshold values, heat dissipation control instructions are dynamically issued to realize the adaptive adjustment of cooling liquid flow and vibrator frequency; the control effect is judged based on the regulation response deviation evaluation value, and whether the heat dissipation is invalid is judged by monitoring the evolution of oil film density in multiple cycles; finally, the strategy evolution sample set is constructed combining the path running characteristics and the regulation results, to support the self-learning and optimization upgrade of the cooling strategy, so as to improve the real-time, accuracy and intelligent level of the cooling control.

[0052] It should be noted that the relative terms, such as first and second, and the like, are used herein only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between such entities or operations. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0053] The preferred embodiments of the application disclosed above are only used to help explain the application. The preferred embodiments do not describe all the details of the application, nor limit the application to the specific embodiments described. It is obvious that many modifications and variations can be made according to the content of the present specification. The present specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the application, so that those skilled in the art can well understand and utilize the application. The application is limited only by the claims and their full scope and equivalents.

Claims

1. A two-phase cold plate heat dissipation control method based on vibration and flow cooperative regulation, characterized in that, The method comprises the following steps: S1, real-time acquisition of cold plate cooperative heat dissipation data, and performing timestamp alignment, sampling reconstruction, abnormality cleaning, noise suppression and normalization processing on the cold plate cooperative heat dissipation data; S2, based on the preprocessed cold plate cooperative heat dissipation data, the spatial mapping relationship between the temperature monitoring area and the cooling path is established, the heat dissipation degradation degree of the two-phase cold plate (4) is evaluated, the heat dissipation state classification determination is performed, and the corresponding control instruction is issued according to the classification result; S3, the execution unit receives and executes each control instruction, evaluates the control response state of each cooling path, determines the control effect, and triggers the dynamic iteration process of the time-sharing cooperative control strategy based on the control result; S4, continuously monitor the oil film density evolution to determine whether the cooling path is heat dissipation failure, evaluate the thermal degradation trend of the two-phase cold plate (4), execute the liquid path switching strategy based on the evaluation result, and extract the path operation characteristics to construct a strategy evolution sample set to realize continuous optimization of the cooling strategy.

2. The two-phase cold plate heat dissipation control method based on vibration and flow collaborative regulation according to claim 1, characterized in that: The specific steps of the real-time acquisition of cold plate cooperative heat dissipation data, and the timestamp alignment, sampling reconstruction, abnormality cleaning, noise suppression and normalization processing on the cold plate cooperative heat dissipation data are as follows: Real-time acquisition of cold plate cooperative heat dissipation data, including inlet temperature, outlet temperature, cooling liquid flow, specific heat capacity of cooling liquid, viscosity of cooling liquid, chip power consumption, chip temperature, vibrator frequency, environmental temperature and oil film density; Through constructing a time sequence synchronization method based on multi-channel timestamp alignment and sampling interval reconstruction, cross-module time alignment and data frame unified processing are performed on the cold plate cooperative heat dissipation data; By introducing an abnormal point identification and physical constraint screening combined abnormality cleaning algorithm, pulse disturbance, sensor drift and physical inconsistency in the cold plate cooperative heat dissipation data are filtered and repaired; by adopting a sliding window filtering and dynamic threshold monitoring combined smoothing denoising method, high-frequency jitter and measurement noise of the cold plate cooperative heat dissipation data are suppressed; by constructing a normalization processing method based on maximum value mapping and segmented stretching, the cold plate cooperative heat dissipation data are compressed across physical quantity values; by introducing a standard deviation standardization and centralization processing combined algorithm, the cold plate cooperative heat dissipation data are distributed standardized.

3. The two-phase cold plate heat dissipation control method based on vibration and flow collaborative regulation according to claim 1, characterized in that: The specific steps of establishing the spatial mapping relationship between the temperature monitoring area and the cooling path based on the preprocessed cold plate cooperative heat dissipation data, and evaluating the heat dissipation degradation degree of the two-phase cold plate are as follows: Divide the chip (5) into multiple temperature monitoring areas, and establish the spatial correlation mapping relationship between the temperature monitoring area and the cooling path according to the structure and flow channel configuration of the two-phase cold plate (4); Based on the preprocessed cold plate cooperative heat dissipation data, multiply the cooling liquid flow by the difference between the outlet temperature and the inlet temperature, and then divide by the chip power consumption to obtain the cooling liquid sensible heat transfer efficiency term; divide the cooling liquid standard viscosity by the cooling liquid viscosity to obtain the flowability influence term; subtract the oil film density to obtain the microchannel patency correction term; Multiply the cooling liquid sensible heat transfer efficiency term, the flowability influence term and the microchannel patency correction term in turn to obtain the comprehensive heat dissipation capacity term; subtract the comprehensive heat dissipation capacity term to obtain the heat dissipation degradation degree evaluation value.

4. The two-phase cold plate heat dissipation control method based on vibration and flow collaborative regulation according to claim 1, characterized in that: The specific steps of the execution heat dissipation state classification determination and issuing corresponding regulation instructions according to the classification results are as follows: Real-time comparison of heat dissipation degradation evaluation value and heat dissipation threshold value, classification regulation response of two-phase cold plate (4) operating state: When the heat dissipation degradation evaluation value is less than or equal to the first-level heat dissipation threshold value, it is determined that the two-phase cold plate (4) is in a normal heat dissipation state, and the existing heat dissipation configuration is maintained; When the heat dissipation degradation evaluation value is greater than the first-level heat dissipation threshold value and less than the second-level heat dissipation threshold value, it is determined that the two-phase cold plate (4) is in a performance degradation state, and a gradual heat dissipation enhancement instruction is issued to the execution unit, based on the incremental adjustment algorithm of the degradation gradient, according to the change trend of the heat dissipation degradation evaluation value of the current period and the previous period, the self-adaptive proportional adjustment mechanism is used to adjust the cooling liquid flow and the vibrator frequency; When the heat dissipation degradation evaluation value is greater than or equal to the second-level heat dissipation threshold value, it is determined that the two-phase cold plate (4) is in an overheating state, and the path-level local enhancement regulation is implemented: the real-time chip temperature of each temperature monitoring area is extracted, the temperature rise rate is calculated based on the continuous time difference, the abnormal chip (5) area whose temperature rise rate exceeds the rate threshold value is identified, the corresponding cooling path is marked as the priority regulation path, and the execution unit is issued with the forced parameter refresh instruction, the cooling liquid flow and the vibrator frequency of the priority regulation path are adjusted to the flow upper limit and the frequency upper limit, and the remaining cooling paths maintain the existing heat dissipation configuration.

5. The method of claim 1, wherein the method is characterized by: The specific steps of the execution unit receiving and executing various control instructions and evaluating the regulation response state of each cooling path are as follows: The execution unit receives various control instructions, sets the cooling liquid flow and the vibrator frequency to the target value; for the priority regulation path, a higher step level is used for adjustment, and the remaining cooling paths are executed according to the normal step level; In a fixed sampling period, the cold plate cooperative heat dissipation data after execution is collected, and the regulation response state of each cooling path is evaluated: the cooling liquid flow and the specific heat capacity of the cooling liquid are multiplied, and the difference between the outlet temperature and the inlet temperature is multiplied to obtain the cooling liquid sensible heat transfer term; the cooling liquid sensible heat transfer term is divided by the chip power consumption, and the obtained ratio is subtracted by 1 to obtain the sensible heat transfer efficiency term; the cooling liquid viscosity is subtracted from the cooling liquid standard viscosity, and then divided by the cooling liquid standard viscosity to obtain the flow deviation term; the cooling liquid sensible heat transfer efficiency term and the flow deviation term are squared and added to obtain the regulation response deviation evaluation value.

6. The method of claim 1, wherein the method is characterized by: The specific steps of the regulation effect determination and the dynamic iteration process of the time-sharing cooperative regulation strategy triggered based on the regulation result are as follows: Real-time comparison of the regulation response deviation evaluation value and the regulation deviation threshold value, when the regulation response deviation evaluation value is less than or equal to the regulation deviation threshold value, it is determined that the regulation is effective, and the current cooling liquid flow and the vibrator frequency remain unchanged; When the regulation response deviation evaluation value is greater than the regulation deviation threshold value, it is determined that the regulation is invalid, and the time-sharing cooperative regulation is executed: the vibrator frequency is preferentially improved, and the cooling liquid flow is improved after a delay of one sampling period; otherwise, only the vibrator frequency is improved.

7. The method of claim 1, wherein the method is characterized by: The specific steps of the continuous monitoring of the oil film density evolution to determine whether the cooling path fails to dissipate heat and evaluate the thermal degradation trend of the two-phase cold plate are as follows: After performing the time-sharing collaborative regulation, the oil film density changes corresponding to each cooling path are continuously monitored. If the oil film density still exceeds the density threshold in three consecutive sampling periods, the current cooling path is marked as a cooling failure path; All cold plate collaborative cooling data from the start of the regulation response to the period when the cooling failure path is marked are extracted, and the corresponding sampling period number is counted; The current coolant viscosity is reduced by the initial coolant viscosity, and then divided by the product of the sampling period number and the standard viscosity of the coolant to obtain the coolant viscosity degradation term; for each sampling period, the difference between the chip temperature and the ambient temperature is calculated, and the ratio of the chip power consumption is obtained to obtain the unit power consumption thermal offset ratio; All unit power consumption thermal offset ratios are squared, and the average value is calculated based on the sampling period number to obtain the chip thermal load abnormality term; The thermal degradation risk assessment value is obtained by squaring the coolant viscosity degradation term and adding the chip thermal load abnormality term.

8. The method of claim 1, wherein the method is a method of two-phase cold plate heat dissipation control based on vibration and flow co-regulation. The specific steps of executing the liquid path switching strategy based on the evaluation result are as follows: Real-time comparison of the thermal degradation risk assessment value and the degradation intensity threshold value is performed. When the thermal degradation risk assessment value is less than or equal to the degradation intensity threshold value, the current regulation strategy configuration is maintained, and the current cooling failure path is recorded as an acceptable risk level path; When the thermal degradation risk assessment value is greater than the degradation intensity threshold value, the current cooling path is marked as an unrecoverable path, and the liquid path switching strategy is executed to interrupt the original cooling path cooling task and migrate the corresponding thermal load to the alternative two-phase cold plate (4) region.

9. The two-phase cold plate heat dissipation control method based on vibration and flow co-regulation according to claim 8, characterized in that: The specific steps of the strategy evolution sample set construction strategy to realize the continuous optimization of the cooling strategy are as follows: According to the marked acceptable risk paths and unrecoverable paths, the oil film density change, coolant viscosity evolution and chip temperature response data are extracted to construct a regulation behavior feature set; The similarity between the regulation behavior features of the current cooling path and the historical path features in the regulation behavior feature set is calculated in real time. If the similarity exceeds the matching threshold, the optimization control instruction is generated according to the historical path associated coolant flow rate set value and vibrator frequency set value; otherwise, the cooling plate collaborative cooling data and regulation state of the current cooling path are constructed as a strategy training sample and stored in a strategy training sample library for subsequent cooling strategy evolution and optimization.

10. A two-phase cold plate heat dissipation control system based on vibration and flow cooperative regulation, characterized in that: It includes: A cold plate cooling data acquisition and preprocessing module, a cold plate thermal state determination and regulation decision module, a control instruction execution and response evaluation module, and a cooling degradation identification and strategy optimization module, wherein: The cold plate cooling data acquisition and preprocessing module is used to acquire cold plate collaborative cooling data in real time, and perform timestamp alignment, sampling reconstruction, abnormality cleaning, noise suppression and normalization processing on the cold plate collaborative cooling data; The cold plate thermal state determination and regulation decision module is used to establish a spatial mapping relationship between the temperature monitoring area and the cooling path based on the preprocessed cold plate collaborative cooling data, evaluate the cooling degradation degree of the two-phase cold plate (4), perform cooling state classification determination, and issue corresponding regulation instructions according to the classification result; The control instruction execution and response evaluation module is configured to execute the control instructions received by the execution unit, evaluate the response state of the regulation of each cooling path, determine the regulation effect, and trigger a dynamic iteration process of the time-sharing cooperative regulation strategy based on the regulation result. The cooling degradation identification and strategy optimization module is configured to continuously monitor the oil film density evolution to determine whether the heat dissipation of the cooling path fails, evaluate the thermal degradation trend of the two-phase cold plate (4), execute a liquid path switching strategy based on the evaluation result, and extract path operation characteristics to construct a strategy evolution sample set, so as to continuously optimize the cooling strategy.

Images