Server heat dissipation device

Abstract

The invention relates to a server heat dissipation device, and the device comprises a heat exchange module which is disposed at a heating point of each processor, a heat exchange medium is accommodated in the heat exchange module, and the heat exchange medium absorbs the heat of the heating point to dissipate the heat of the processor; the state monitoring module comprises a processor state acquisition unit, and the processor state acquisition unit is connected with the mainboard and acquires running state signals of the processors in real time; the control module is used for judging the real-time heat dissipation requirements of the heat exchange modules corresponding to the processors according to the collected running state signals of the processors and outputting a flow adjusting instruction; and the flow adjusting module is used for receiving the flow adjusting instruction and adjusting the flowing flow of the heat exchange medium so as to adjust the heat dissipation speed of each processor. According to the heat dissipation device, the start-stop state of multiple processors can be captured in real time, the heat exchange medium flow of each heat exchange module is accurately adjusted, and heat dissipation according to needs is achieved.

Description

Technical Field

[0001] This application belongs to the field of data center operation and maintenance technology, and relates to a server heat dissipation device. Background Technology

[0002] With the rapid development of artificial intelligence, big data, and cloud computing technologies, the computing power demand of AI servers in data centers is growing exponentially. To meet the operational needs of core scenarios such as large model training and high-performance parallel computing, AI servers typically adopt multi-GPU cluster configurations, with a single server carrying up to 8, 16, or even more GPUs. Since GPUs generate extremely high heat flux density when operating at full load, and each GPU needs to maintain synchronized and efficient computation, their heat dissipation directly determines the server's operational stability, computing power output efficiency, and hardware lifespan. Therefore, efficient heat dissipation for high-speed GPUs is particularly important.

[0003] Existing cooling devices mostly use constant flow or simple flow distribution to control the cooling of each GPU. When a GPU suddenly starts or stops, the cooling demand of the corresponding heat sink will change drastically. When running, sufficient flow is required for cooling, and when stopped, only the basic circulation flow needs to be maintained. However, the existing control cannot respond to this change in time, which leads to the risk of overheating of running GPUs due to insufficient flow, or energy waste caused by flow redundancy of stopped GPUs.

[0004] Therefore, there is an urgent need for a server cooling device that can accurately match the dynamic heat dissipation requirements of the GPU to solve the problems existing in the current technology. Summary of the Invention

[0005] This application provides a server cooling device that captures the start-stop status of multiple processors in real time and precisely adjusts the flow rate of the heat exchange medium in each heat exchange module to achieve on-demand cooling. This solves the problem of overheating risk during operation and energy waste during shutdown caused by the inability of existing cooling devices to respond in time to the dynamic start-stop of processors.

[0006] This embodiment provides a server heat dissipation device. The server includes a motherboard and multiple processors. The heat dissipation device includes: a heat exchange module disposed at the heat points of each processor and containing a heat exchange medium that absorbs heat from the heat points to dissipate heat from the processors; a status monitoring module including a processor status acquisition unit connected to the motherboard to acquire the operating status signals of each processor in real time; a control module that, based on the acquired processor operating status signals, determines the real-time heat dissipation requirements of the heat exchange modules corresponding to each processor and outputs a flow rate adjustment command; and a flow rate adjustment module that receives the flow rate adjustment command and adjusts the flow rate of the heat exchange medium to adjust the heat dissipation speed of each processor.

[0007] The technical solution of this embodiment, through the above technical solution, the heat dissipation device can capture the start-stop state of multiple processors in real time, accurately adjust the flow rate of the heat transfer medium of each heat exchange module, and achieve on-demand heat dissipation, which not only avoids overheating during processor operation and energy waste during shutdown, but also ensures the efficient and stable operation of multiple processors, with strong adaptability and convenient operation and maintenance.

[0008] In this embodiment, the heat exchange module includes a cold plate, which is arranged at the heat generating part of each processor for dissipating heat from the processor; the state monitoring module includes a temperature sensor, which is arranged on the inlet branch and outlet branch of the cold plate for collecting the inlet temperature and outlet temperature of the heat transfer medium in real time; the control module is configured to: preset a first temperature threshold T1 and a second temperature threshold T2, and calculate the temperature difference ΔT between the inlet and outlet of each cold plate according to the collected inlet temperature and outlet temperature _i , according to the first temperature threshold T1, the second temperature threshold T2 and the actual temperature difference ΔT _i calculate the target flow rate Q of each cold plate _obj_i : when ΔT _i > T1, the target flow rate Q _obj_i = Q0 × [1 + (ΔT _i - T1) / T1]; when ΔT _i < T2, the target flow rate Q _obj_i = Q0 × [1 - (T2 - ΔT _i ) / T1]; when T2 ≤ ΔT_i ≤ T1, the target flow rate Q _obj_i = Q0; where Q0 is the preset basic flow rate of the cold plate.

[0009] The technical solution of this embodiment, by setting temperature sensors on the inlet and outlet branches of the cold plate, combined with the temperature difference threshold and the target flow rate algorithm, can accurately match the real-time heat dissipation requirements of each GPU. When the temperature difference is too large, increase the flow rate to strengthen heat dissipation, and when it is too small, optimize the flow rate to avoid waste, which not only ensures the stable operation of the GPU, but also realizes energy saving and improves the heat dissipation adaptability and efficiency.

[0010] In this embodiment, the flow rate adjustment module includes a flow rate regulating valve, which is arranged on the inlet branch of each cold plate and adjusts its opening degree according to the flow rate adjustment instruction to adjust the flow rate of the heat transfer medium in each cold plate; the state monitoring module includes a flow rate sensor arranged on the inlet branch of each cold plate, and the flow rate sensor is used to collect the actual flow rate of the heat transfer medium; the control module is configured to: collect the actual flow rate in the inlet branch through the flow rate sensor, and calculate the target opening degree θ of the flow rate regulating valve of each inlet branch according to the actual flow rate Q [[ID=�0]] _act_i of the inlet branch of each cold plate and the target flow rate Q _obj_i , and calculate the target opening degree θ of the flow rate regulating valve of each inlet branch _obj_iA pulse signal is sent to the corresponding flow control valve to control the valve to adjust the valve opening to the target value θ. _obj_i .

[0011] The technical solution of this embodiment can accurately respond to control commands, calibrate valve opening and actual flow in real time, ensure that the flow rate of heat exchange medium of each cold plate is accurately matched with the target value, improve the accuracy and stability of flow regulation, ensure that the heat dissipation requirements of each GPU are accurately met, and realize closed-loop control to reduce flow deviation and optimize heat dissipation effect and system reliability.

[0012] In this embodiment, the control module is configured to: preset the minimum sustaining flow rate Q of the cold plate. min The minimum sustaining flow rate Q min The target flow rate Q of the cold plate is determined when the processor is stopped and T2 ≤ ΔT_i ≤ T1, and the processor's running status is less than the base flow rate Q0. _obj_i =Q min .

[0013] With the technical solution of this embodiment, when the processor stops running and the temperature difference is within the normal range, the target flow rate automatically switches to a smaller minimum sustaining flow rate Q. min This avoids energy waste caused by redundant flow, maintains basic pipeline circulation, and ensures rapid response to heat dissipation needs when the processor restarts, thus balancing energy saving and timely system response.

[0014] In this embodiment, the heat exchange module further includes a main circulation pump. The output end of the main circulation pump is connected to each inlet branch through the main inlet channel, and its input end is connected to each outlet branch through the main outlet channel. The main circulation pump is used to drive the heat exchange medium in the cold plate to circulate and dissipate heat. The status monitoring module includes a pressure sensor, which is installed on the main outlet channel to collect the actual pressure P of the heat exchange medium in the main outlet channel. _act The control module is configured to: preset a maximum pressure threshold P max and minimum pressure threshold P min In P _act >P max or P _act <P min At that time, the target speed of the main circulation pump is calculated, and a speed adjustment command is output to the main circulation pump; the formula for calculating the target speed is: n _obj =n _act -ΔP K, ΔP=P _act -[(P min +P max [) / 2]; where n _act The actual speed of the main circulation pump, n _objtK is the target speed of the main circulation pump, and K is the preset pressure-speed adjustment coefficient.

[0015] The technical solution of this embodiment monitors the system pressure in real time through a pressure sensor, and combines a preset pressure threshold with a speed adjustment formula to control the target speed of the main circulation pump, quickly correct pressure deviations, ensure that the system pressure is stable within a safe range, avoid pipeline leaks, abnormal pump loads and other faults, ensure the reliable operation of the liquid cooling circuit when adjusting the flow of multiple branches, and improve the overall stability and safety of the system.

[0016] In this embodiment, the server cooling device further includes a host computer connected to the server; the control module is configured to: preset a safe temperature threshold T max and safety duration threshold t max ; In determining the collected liquid outlet temperature T of each cold plate _out_i greater than the safe temperature threshold T max At that time, the processor's overheating fault information is sent to the host computer; at the liquid outlet temperature T of each cold plate... _out_i greater than the safe temperature threshold T max And the duration exceeds the safe duration threshold t. max At that time, a shutdown signal is sent to the server.

[0017] The technical solution of this embodiment, by setting a safe temperature and duration threshold and combining it with a host computer linkage mechanism, can not only promptly report processor overheating faults for maintenance and repair, but also trigger server shutdown after the overheating timeout, effectively preventing processor damage due to long-term overheating, and balancing the timeliness of fault warning and the security of hardware protection.

[0018] In this embodiment, the server heat dissipation device further includes: a spray heat dissipation module, which includes spray heads arranged towards each secondary heat-generating point, dissipating heat by spraying cooling medium onto the secondary heat-generating points; a status monitoring module including a spray temperature sensor for collecting the real-time temperature of each secondary heat-generating point; a control module determining the real-time heat dissipation requirement of each secondary heat-generating point based on the collected real-time temperature, and outputting flow rate adjustment commands for each spray head; and a flow rate adjustment module including a spray solenoid valve, which opens or closes according to the flow rate adjustment commands for each spray head, thereby controlling the start and stop of spraying from each spray head to adjust the heat dissipation rate of each secondary heat-generating point; wherein, the secondary heat-generating points include other heat-generating points within the server besides the processor.

[0019] The technical solution of this embodiment is specifically designed for spray cooling of secondary heat-generating points in the server other than the processor. The spray is started and stopped precisely by real-time monitoring and linkage of temperature sensors with spray solenoid valves, which fills the heat dissipation gap of secondary heat-generating points, avoids the accumulation of heat and the resulting local overheating problem, and further ensures the overall reliability of the server.

[0020] In this embodiment, the control module is configured to: preset the spray start-up temperature threshold T _spray Spray stop temperature threshold T _stop According to the temperature T of the secondary heating point _ j and spray start-up temperature threshold T _spray Spray stop temperature threshold T _stop Output spray adjustment command: When T _j >T _spray When T is activated, an opening command is output to the solenoid valve of the corresponding secondary heating point; when T _j <T _stop When T is active, a shut-off command is output to the solenoid valve of the corresponding secondary heating point; when T is active... _stop <T _j <T _spray When it is determined that no spray adjustment command needs to be output, it is clear that no such command is required.

[0021] The technical solution of this embodiment, by preset spray start and stop temperature thresholds and combining them with the real-time temperature of the secondary heat points to accurately control the spray solenoid valve switch, enables the secondary heat points to be sprayed for heat dissipation as needed. This avoids local overheating caused by insufficient heat dissipation and prevents energy waste caused by excessive spraying, thus ensuring the stable operation of the entire server.

[0022] In this embodiment, the spray cooling module further includes spray branch pipes, a spray main pipe, a spray power unit, and a spray liquid storage unit. The input end of the spray main pipe is connected to the spray liquid storage unit via the spray power unit. The output end of the spray main pipe extends to the secondary heating point area and branches off to form spray branch pipes corresponding to each secondary heating point. The end of each spray branch pipe is connected to the spray head of each secondary heating point. A spray solenoid valve is installed on each spray branch pipe to control the start and stop of spraying on the corresponding spray branch pipe. The status monitoring module includes a spray flow sensor and a spray pressure sensor installed on the spray main pipe. The spray flow sensor and the spray pressure sensor are used to collect the flow rate and pressure of the cooling medium in the spray main pipe, respectively. The control module is configured to adjust the output power of the spray power unit based on the flow rate and pressure data of the cooling medium collected by the spray flow sensor and the spray pressure sensor, thereby adjusting the flow rate and pressure of the cooling medium.

[0023] The technical solution of this embodiment uses flow and pressure sensors to monitor data in real time, and adjusts the output power of the spray power unit in conjunction with the branch pipe solenoid valve to precisely control the start and stop. This ensures stable flow and pressure of the cooling medium, ensuring uniform and efficient spraying of secondary heat-generating points, while also dynamically adapting to heat dissipation needs, avoiding abnormal system pressure, and improving the stability and reliability of spray heat dissipation.

[0024] In this embodiment, the spray power unit includes a spray pump, and the flow rate of the cooling medium is adjusted by regulating the rotation speed of the spray pump. The control module is configured to: preset the target flow rate of each spray branch pipe, and calculate the target flow rate Q of the main spray pipe based on the number of opening spray solenoid valves. _spray Compare the actual flow rate Q of the main sprinkler pipe. _spray_act and target traffic Q _spray Size: when Q _spray_act <Q _spray When Q is reached, the rotational speed of the spray pump is increased; when Q is reached... _spray_act >Q _spray When this is the case, reduce the speed of the spray pump.

[0025] The technical solution of this embodiment dynamically matches the actual flow rate with the target flow rate by adjusting the speed of the spray pump, ensuring that the flow rate of the spray system is stable and accurate. This not only meets the heat dissipation needs of each secondary heat-generating point, but also avoids flow redundancy or insufficiency, thus ensuring the heat dissipation effect and energy efficiency of the system.

[0026] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application, it can be implemented according to the contents of the specification. In order to make the above and other objects, features and advantages of this application more apparent, specific embodiments of this application are given below. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a schematic diagram of the heat dissipation device of this application.

[0029] Figure 2 This is a flowchart illustrating the overall process of heat dissipation in the heat dissipation device of this application.

[0030] Figure 1Components: 1. Processor; 2. Cold plate; 3. Temperature sensor; 4. Inlet branch; 5. Outlet branch; 6. Flow regulating valve; 7. Flow sensor; 8. Main circulation pump; 9. Main inlet; 10. Main outlet; 11. Pressure sensor; 12. Condensation device; 13. Total flow sensor; 14. Secondary heating point; 15. Spray pump; 16. Spray temperature sensor; 17. Spray solenoid valve; 18. Spray branch pipe; 19. Main spray pipe; 20. Spray pressure sensor; 21. Spray flow sensor; 22. Spray storage unit; 23. Collection tank; 24. Pressure relief valve. Detailed Implementation

[0031] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this application.

[0032] The prefixes such as "first" and "second" used in this application embodiment are merely for distinguishing different descriptive objects and do not limit the position, order, priority, quantity, or content of the described objects. The use of ordinal numbers and other prefixes used to distinguish descriptive objects in this application embodiment does not constitute a limitation on the described objects. The description of the described objects is given in the claims or the context of the embodiments, and should not constitute unnecessary restrictions due to the use of such prefixes. Furthermore, in the description of this embodiment, unless otherwise stated, "multiple" means two or more.

[0033] The technical solutions of the embodiments of this application will be described below with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B; the term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone.

[0034] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0035] With the rapid development of artificial intelligence, big data, and cloud computing technologies, the computing power demand of AI servers in data centers is growing exponentially. To meet the operational needs of core scenarios such as large model training and high-performance parallel computing, AI servers typically adopt multi-GPU cluster configurations, with a single server carrying up to 8, 16, or even more GPUs. Since GPUs generate extremely high heat flux density when operating at full load, and each GPU needs to maintain synchronized and efficient computation, their heat dissipation directly determines the server's operational stability, computing power output efficiency, and hardware lifespan. Meanwhile, although secondary heat-generating components such as memory modules and expansion cards generate less heat per unit than GPUs, the accumulated heat from the dense arrangement of multiple components can still cause localized increases in server internal temperature. If heat dissipation is not timely, problems such as memory module frequency reduction and data transmission errors can easily occur, affecting the overall reliability of the machine. Against this backdrop, traditional air cooling methods, due to their low cooling efficiency, high noise, and high energy consumption, are no longer sufficient. Liquid cooling technology, with its higher heat dissipation power density, has become the mainstream cooling solution for AI servers. However, in actual operation, the GPU of an AI server is often in a dynamic start-stop state. This dynamic operating condition exposes many technical shortcomings of existing liquid cooling temperature control devices:

[0036] Existing devices mostly employ constant flow or simple equal-flow control methods. When the GPU suddenly starts or stops, the heat dissipation demand of the corresponding cold plate changes drastically. During operation, sufficient flow is required for heat dissipation, while during shutdown, only a basic circulation flow needs to be maintained. However, existing control systems cannot respond to this change in a timely manner, leading to the risk of overheating in running GPUs due to insufficient flow, or energy waste due to flow redundancy in stopped GPUs. Secondly, the switching between GPU start-up and shutdown states is abrupt. Existing temperature control systems have delays in monitoring the GPU's operating status, and the flow regulation mechanism's response is not timely, easily resulting in excessive temperature fluctuations, affecting the stability of GPU operation. On the other hand, when multiple cold plates are connected in parallel, the pipeline pressure stability is poor. Sudden changes in flow in some branches can cause pressure fluctuations in the entire liquid cooling circuit. Without effective pressure compensation and flow coordination control, this may cause pipeline leaks, abnormal pump loads, and other malfunctions, reducing the reliability of the liquid cooling system. Meanwhile, the lack of heat dissipation for secondary heat sources means that existing solutions do not cover secondary heat sources such as memory modules, and the accumulated heat cannot be effectively dissipated, which can easily lead to local overheating problems. Finally, there is insufficient adaptability. Existing control methods are mostly designed for servers with a fixed number of GPUs. When the number of GPUs in the AI ​​server is adjusted, the control logic cannot be flexibly adapted and the parameters need to be readjusted, which increases the operation and maintenance costs.

[0037] Therefore, please see Figure 1To address the aforementioned technical problems, this embodiment proposes a server heat dissipation device. The server includes multiple processors 1, a motherboard, expansion cards, and memory modules, among other components. To accurately match the dynamic heat dissipation requirements of the multiple processors 1, the heat dissipation device includes: A heat exchange module is provided at the heat-generating points of each processor 1. It contains a heat exchange medium, which absorbs the heat from the heat-generating points to dissipate heat from the processor 1. The status monitoring module includes a processor 1 status acquisition unit, which is connected to the motherboard and collects the running status signals of each processor 1 in real time, and promptly captures the start and stop status of each processor 1. The control module determines the real-time heat dissipation requirements of the heat exchange modules corresponding to each processor 1 based on the collected operating status signals of processor 1, and outputs flow adjustment commands. The flow regulation module receives the flow regulation command and adjusts the flow rate of the heat exchange medium to regulate the heat dissipation rate of each processor 1.

[0038] For example, taking an AI server equipped with 8 GPUs (Graphics Processing Units) as an application scenario, the GPU status acquisition unit collects the operating status signals of the 8 GPUs in real time. The obtained signal indicates that two of the GPUs have stopped running. At this time, the control module outputs an instruction to reduce the flow rate of the heat exchange medium of the two GPUs. After receiving the adjustment instruction, the flow rate adjustment module reduces the flow rate accordingly, thereby reducing unnecessary heat dissipation work and avoiding energy waste.

[0039] Through the above technical solution, the heat dissipation device can capture the start-up and shutdown status of the multi-processor 1 in real time, accurately adjust the heat exchange medium flow rate of each heat exchange module, and realize heat dissipation on demand. This not only avoids overheating of the processor 1 during operation and energy waste during shutdown, but also ensures the efficient and stable operation of the multi-processor 1. It is highly adaptable and easy to maintain.

[0040] In some embodiments, the control module uses an industrial-grade PLC controller as the core control unit, with a built-in embedded control program, possessing data processing, logic judgment, and instruction output functions. This controller interacts with the GPU status acquisition unit, various sensors, regulating valves, and circulating pumps via a communication interface to achieve centralized control of the entire system. The controller has a built-in parameter storage unit, pre-setting parameters such as flow baseline values, temperature thresholds, and pressure thresholds corresponding to different numbers of running GPUs. Parameter configuration and modification via a host computer are supported, improving system adaptability.

[0041] In some embodiments, the heat exchange module includes a cold plate 2, which is disposed at the heat - generating part of each GPU for cooling the GPU. Specifically, the cold plate 2 can be a micro - channel phase - change cold plate. Each GPU is correspondingly provided with an independent micro - channel phase - change cold plate. The cold plate uses a metal with a high thermal conductivity coefficient as the substrate, and a high - density micro - channel array is integrated on the surface to enhance the heat - exchange area and heat - transfer efficiency. The cold plate uses a low - boiling - point fluorinated liquid as the heat - exchange medium, and uses the phase - change process of the fluorinated liquid absorbing heat and vaporizing near the heat - generating part of the GPU to quickly take away a large amount of heat.

[0042] In some embodiments, the server cooling device further includes an auxiliary module, and the auxiliary module includes a host computer monitoring terminal and an alarm unit. The host computer monitoring terminal is connected to the main controller through Ethernet, and can display information such as the status of each GPU, cold - plate parameters, valve opening degree, pump body speed, etc. in real time, and supports historical data query and control parameter setting. The alarm unit includes an audible and visual alarm and a fault indicator light. When situations such as GPU over - temperature, pipeline pressure abnormality, valve failure, etc. occur, the control module triggers the alarm unit to act, and at the same time displays the fault information on the host computer, facilitating the maintenance personnel to process it in time.

[0043] In some embodiments, the status monitoring module includes a temperature sensor 3, which is disposed on the liquid inlet branch 4 and the liquid outlet branch 5 of the cold plate 2 for collecting the liquid inlet temperature and the liquid outlet temperature of the heat - exchange medium in real time.

[0044] By setting the temperature sensor 3, the server cooling device can also implement the function of adjusting the flow rate of the heat - exchange medium according to the real - time heat - dissipation situation of each GPU. To achieve the above function, the control module is configured as follows: preset a first temperature threshold T1 and a second temperature threshold T2, calculate the temperature difference ΔT between the liquid inlet and outlet of each cold plate 2 according to the collected liquid inlet temperature and liquid outlet temperature _i and calculate the target flow rate Q of each cold plate 2 according to the first temperature threshold T1, the second temperature threshold T2 and the actual temperature difference ΔT _i : _obj_i : When ΔT _i > T1, the target flow rate Q _obj_i = Q0×[1+(ΔT _i - T1) / T1]; When ΔT _i < T2, the target flow rate Q _obj_i = Q0×[1-(T2 - ΔT _i ) / T1]; When T2≤ΔT_i≤T1, the target flow rate Q _obj_i = Q0; Wherein, Q0 is the preset basic flow rate of the cold plate 2.

[0045] For example, the first temperature threshold T1 is 10℃, and the second temperature threshold T2 is 5℃. When ΔT _i When the temperature is >10℃, the correction factor for the target flow rate is ΔT. _i / 10, the maximum value of this correction factor does not exceed 1.5. When ΔT _i When the temperature is below 5℃, the correction factor for the target flow rate is 0.5 + ΔT. _i / 10, the minimum value of this correction factor is not less than 0.7.

[0046] The above technical solution, by setting a temperature sensor 3 in the liquid inlet / outlet branch 5 of the cold plate 2, combined with the temperature difference threshold and target flow algorithm, can accurately match the real-time heat dissipation needs of each GPU. When the temperature difference is too large, the flow rate is increased to enhance heat dissipation, and when the temperature difference is too small, the flow rate is optimized to avoid waste. This ensures the stable operation of the GPU, achieves energy saving, and improves heat dissipation adaptability and efficiency.

[0047] There exists a working condition where the inlet and outlet temperature difference ΔT of the cold plate 2 corresponding to a certain GPU is... _i When the temperature is within the normal range of 5℃ to 10℃, and the GPU status acquisition unit confirms that the GPU has stopped running, the GPU generates almost no new heat, leaving only a small amount of residual heat. At this point, if the preset base flow rate (Q) is still applied... min Adjusting the flow rate of the cold plate 2 corresponding to the GPU would result in redundant heat exchange medium flow. This means that a large amount of heat is not needed to be removed through this flow, and the excess heat exchange medium circulation would consume additional energy, causing ineffective flow of the heat exchange medium and ultimately leading to unnecessary energy waste, thus violating the energy-saving design intent of the cooling system. Therefore, to avoid this situation, in some embodiments, the control module is configured to preset the minimum sustaining flow rate Q of the cold plate 2. min And minimum sustaining flow Q min <Base flow Q0. When the collected GPU running status is stopped and T2≤ΔT_i≤T1, determine the target flow Q of the cold plate 2. _obj_i =Q min .

[0048] In order to regulate the flow rate of the heat exchange medium in each cold plate 2, in some embodiments, the flow regulation module includes a flow regulation valve 6, which is disposed on the liquid inlet branch 4 of each cold plate 2 and its opening is adjusted according to the flow regulation command to regulate the flow rate of the heat exchange medium in each cold plate 2.

[0049] Specifically, the flow regulating valve 6 can be an electric proportional regulating valve, which is driven by a stepper motor and receives pulse signals sent by the control module to adjust the valve opening.

[0050] After calculating the target flow rate of the heat exchange medium for each cold plate 2, the valve opening needs to be adjusted according to the actual flow rate of the heat exchange medium in each cold plate 2 to complete the flow regulation. Therefore, in some embodiments, the status monitoring module includes a flow sensor 7 installed on the liquid inlet branch 4 of each cold plate 2. The flow sensor 7 is used to collect the actual flow rate of the heat exchange medium. The control module is configured to: collect the actual flow rate in the liquid inlet branch 4 through the flow sensor 7, and adjust the valve opening according to the actual flow rate Q of the liquid inlet branch 4 of each cold plate 2. _act_i With target traffic Q _obj_i The target opening θ of the flow control valve 6 in each inlet branch 4 is calculated using a PID (proportional-integral-derivative) control algorithm. _obj_i A pulse signal is sent to the corresponding flow control valve 6 to control the flow control valve 6 to adjust the valve opening to the target value θ. _obj_i .

[0051] In some embodiments, the heat exchange module further includes a position feedback sensor, which is disposed within the flow regulating valve 6 and is used to collect the opening degree of the flow regulating valve 6.

[0052] Through the above scheme, the device can accurately respond to control commands, calibrate the valve opening and actual flow in real time, ensure that the flow of heat exchange medium in each cold plate 2 is accurately matched with the target value, improve the accuracy and stability of flow regulation, ensure that the heat dissipation requirements of each GPU are accurately met, and at the same time realize closed-loop control, reduce flow deviation, and optimize heat dissipation effect and system reliability.

[0053] After performing the flow regulation operation, to ensure that the heat dissipation requirements of each GPU are accurately matched and to guarantee the heat dissipation effect and system stability, the flow regulation accuracy of each cold plate branch 2 needs to be detected and verified in real time. If the detection result shows that the deviation between the actual flow and the target flow exceeds the preset accuracy threshold, a secondary adjustment mechanism needs to be initiated immediately. Therefore, in some embodiments, the control module is configured to: preset the flow deviation rate α, and adjust the valve opening of the flow regulation valve 6 to the target value θ. _obj_i Then, the actual flow rate Q of the liquid inlet branch 4 of each cold plate 2 was collected again. _act_i If |Q _act_ iQ _obj_i |≤α Q _obj_i If the flow rate regulation of the cold plate 2 is within the acceptable range, no further adjustment is required; if |Q _act_ iQ _obj_i |>α Q _obj_i If the flow rate regulation of the cold plate 2 is not up to standard, it needs to be readjusted based on the actual flow rate Q of the liquid inlet branch 4 of each cold plate 2. _act_i With target traffic Q _obj_iThe target opening θ of the flow control valve 6 in each inlet branch 4 is calculated using a PID control algorithm. _obj_i A pulse signal is sent to the corresponding flow control valve 6 to control the flow control valve 6 to adjust the valve opening to the target value θ. _obj_i Until the actual flow rate Q of the inlet branch 4 _act_i Until the target is met.

[0054] In some embodiments, the heat exchange module further includes a main circulation pump 8. The output end of the main circulation pump 8 is connected to each liquid inlet branch 4 through the main liquid inlet 9, and its input end is connected to each liquid outlet branch 5 through the main liquid outlet 9. The main circulation pump 8 is used to drive the heat exchange medium in the cold plate 2 to circulate and dissipate heat from the GPU.

[0055] In some embodiments, the heat exchange module further includes a condensation device 12. The condensation device 12 is disposed on the liquid inlet main line 9 or the liquid outlet main line, and is used to cool down the high-temperature heat exchange medium after heat exchange. The cooled heat exchange medium continues to circulate into the cold plate 2 to dissipate heat from the GPU.

[0056] Since simultaneous adjustment of flow rates in multiple branches may cause fluctuations in the main circuit pressure, the control module needs to adjust the speed of the main circulation pump 8 in conjunction with the pressure adjustment to ensure system pressure stability. Therefore, in some embodiments, the status monitoring module includes a pressure sensor 11, which is installed on the main inlet 9 to collect the actual pressure P of the heat exchange medium in the main inlet 9. _act The control module is configured to: preset the maximum pressure threshold P max and minimum pressure threshold P min In P _act >P max or P _act <P min At that time, calculate the target speed n of the main circulation pump 8. _obj It also outputs a speed adjustment command to the main circulation pump 8. The main circulation pump 8 receives the speed adjustment command from the control module and adjusts the speed to the target speed n. _obj .

[0057] Specifically, the formula for calculating the target rotational speed is: n _obj =n _act -ΔP K, ΔP=P _act -[(P min +P max [) / 2]; where n _act The actual speed of the main circulation pump 8; n _obj The target speed of the main circulation pump 8; K is the preset pressure-speed adjustment coefficient, which is an empirical value, to ensure the target speed n _obj Within the maximum speed range of the main circulation pump 8.

[0058] Furthermore, the control module is also configured to: preset the total flow deviation rate β; and adjust the flow rate Q according to the target flow rate Q of each inlet branch 4. _obj_i Calculate the total target flow rate Q of the main inlet channel 9. _obj_total After the speed adjustment of the main circulation pump 8 is completed, the total actual flow rate Q of the main inlet channel 9 is collected. _total ; Calculate the total target flow Q _obj_total With total actual flow Q _total Deviation between: if |Q _total -Q _obj_total |≤β Q _obj_total If the total actual flow meets the requirements, no further pressure compensation is needed; if |Q _total -Q _obj_total |>β Q _obj_total Then, continue to calculate the target speed of the main circulation pump 8 and output a speed adjustment command to the main circulation pump 8.

[0059] In the above technical solution, the system pressure is monitored in real time by the pressure sensor 11 of the main inlet 9. Combined with the preset pressure threshold and the exclusive speed adjustment formula, the target speed of the main circulation pump 8 is controlled in a coordinated manner. The pressure deviation caused by the flow adjustment of multiple branches is quickly corrected to ensure that the system pressure is stable within a safe range, avoid pipeline leakage, abnormal pump load and other faults, ensure the reliable operation of the liquid cooling circuit, and improve the overall stability and safety of the system.

[0060] In some embodiments, the server cooling device also has overheat warning and overheat protection functions. To achieve this function, the control module is configured to: preset a safe temperature threshold T. max and safety duration threshold t max ; In determining the collected liquid outlet temperature T of each cold plate 2 _out_i greater than the safe temperature threshold T max At that time, the overheating fault information of the GPU is sent to the host computer; at the liquid outlet temperature T of each cold plate 2 _out_i greater than the safe temperature threshold T max And the duration exceeds the safe duration threshold t. max At that time, a shutdown signal is sent to the server.

[0061] In the above technical solution, by setting safe temperature and duration thresholds and combining them with the host computer linkage mechanism, it can not only promptly report processor 1 overheating faults for maintenance and repair, but also trigger server shutdown after the overheating timeout, effectively preventing processor 1 from being damaged due to long-term overheating, thus balancing the timeliness of fault warning and the security of hardware protection.

[0062] In addition, although individual components such as memory modules and expansion cards in a server generate less heat than GPUs, the cumulative heat generated by the dense arrangement of multiple components can still cause the local ambient temperature inside the server to rise. If heat dissipation is not timely, problems such as memory module frequency reduction and data transmission errors may occur, affecting the overall reliability of the machine.

[0063] Therefore, to dissipate heat from the secondary heat sources 13 in the server, which include other heat sources within the server besides the GPU, the server cooling device further includes a spray cooling module in some embodiments.

[0064] The spray cooling module includes spray heads arranged towards each secondary heat source 13. These spray heads dissipate heat from the secondary heat sources 13 by spraying cooling medium onto them. The status monitoring module also includes a spray temperature sensor 15, which collects the real-time temperature of each secondary heat source 13. The control module is configured to determine the real-time cooling requirements of each secondary heat source 13 based on its collected real-time temperature and output flow rate adjustment commands for each spray head. The flow rate adjustment module also includes a spray solenoid valve 16, which opens or closes according to the flow rate adjustment commands from each spray head, thereby controlling the start and stop of spraying and adjusting the cooling rate of each secondary heat source 13.

[0065] Specifically, the spray temperature sensor 15 is a surface-mount temperature sensor, which is attached to the surface of secondary heat-generating points 13 such as memory modules and southbridge chips. One surface-mount temperature sensor is assigned to each group of secondary heat-generating points 13, and its temperature sampling frequency is consistent with the frequency of GPU operation status sampling. This is used to monitor the temperature changes of the secondary heat-generating points 13 in real time, providing data support for spray cooling control. In addition, the cooling medium for the spray cooling module is also a low-boiling-point fluorinated liquid working fluid.

[0066] In the above technical solution, the temperature of the secondary heat point 13 is captured in real time by the spray temperature sensor 15, the control module accurately judges the heat dissipation requirements, and controls the start and stop of the spray head by linkage with the spray solenoid valve 16, so as to realize the heat dissipation of the secondary heat point 13 on demand, fill the heat dissipation gap of non-core heat-generating components, avoid local heat accumulation affecting the overall reliability of the server, and reduce energy waste by starting and stopping on demand.

[0067] In some embodiments, to achieve the purpose of activating the spray system for timely heat dissipation when each secondary heating point 13 generates heat and stopping the spray system to save energy when the temperature is normal, the control module is configured to: preset the spray system activation temperature threshold T. _spray Spray stop temperature threshold T _stop According to the temperature T of secondary heating point 13 _ j and spray start-up temperature threshold T _spraySpray stop temperature threshold T _stop Output spray adjustment commands: When T _j >T _spray When T is activated, an opening command is output to the spray solenoid valve 16 corresponding to the secondary heating point 13; when T _j <T _stop When T is active, a shut-off command is output to the spray solenoid valve 16 corresponding to the secondary heating point 13; when T is active... _stop <T _j <T _spray When it is determined that no spray adjustment command needs to be output, it is clear that no such command is required.

[0068] In the above technical solution, by preset spray start and stop temperature thresholds and combining them with the real-time temperature of the secondary heating point 13 to accurately control the solenoid valve switch, the secondary heating point 13 can achieve on-demand heat dissipation by "spraying when the temperature exceeds the standard and stopping when the temperature is normal". This ensures timely dissipation of accumulated heat from the secondary heating point 13 and avoids local overheating, while also preventing excessive spraying that would waste energy, thus balancing heat dissipation effectiveness and energy saving.

[0069] In some embodiments, in order to enable the spray cooling module to dynamically match the heat dissipation of the secondary heat source 13, the spray cooling module also has the function of real-time monitoring of the flow rate and pressure of the cooling medium, and real-time adjustment of the flow rate and pressure of the cooling medium: The spray cooling module also includes spray branch pipes 17, a spray main pipe 18, a spray power unit, and a spray liquid storage unit 21. The input end of the spray main pipe 18 is connected to the spray liquid storage unit 21 via the spray power unit. The output end of the spray main pipe 18 extends to the secondary heating point 13 area and branches off to form spray branch pipes 17 corresponding to each secondary heating point 13. The end of each spray branch pipe 17 is connected to the spray head of each secondary heating point 13. A spray solenoid valve 16 is installed on each spray branch pipe 17 to control the start and stop of spraying on the corresponding spray branch pipe 17. The status monitoring module includes a spray flow sensor 20 and a spray pressure sensor 19 installed on the spray main pipe 18. The spray flow sensor 20 and the spray pressure sensor 19 are used to collect the flow rate and pressure of the cooling medium in the spray main pipe 18, respectively. The control module is configured to adjust the output power of the spray power unit based on the flow rate and pressure data of the cooling medium collected by the spray flow sensor 20 and the spray pressure sensor 19, thereby adjusting the flow rate and pressure of the cooling medium.

[0070] Specifically, the spray power unit includes a spray pump 14, which can be a small variable frequency spray pump 14. By adjusting the speed of the spray pump 14, the flow rate and pressure of the cooling medium can be adjusted.

[0071] Furthermore, to achieve flow and pressure regulation of the cooling medium in the spray cooling module, the control module is configured to: preset the target flow rate of each spray branch pipe 17, and calculate the target flow rate Q of the spray main pipe 18 based on the number of opening spray solenoid valves 16. _spray Compare the actual flow rate Q of the sprinkler main pipe 18. _spray_act and target traffic Q _spray Size: when Q _spray_act <Q _spray When the speed of the spray pump 14 is increased, both the flow rate and pressure of the cooling medium will increase accordingly; when Q _spray_act >Q _spray When the speed of the spray pump 14 is reduced, the flow rate and pressure of the cooling medium will also decrease.

[0072] In some embodiments, the spray cooling module further includes a liquid collection tank 22, which is disposed below the secondary heating point 13 and is used to collect the cooling medium sprayed out after cooling the secondary heating point 13.

[0073] In the above technical solution, the flow sensor 7 and pressure sensor 11 monitor data in real time, and calculate the target flow rate of the main pipe by combining the number of openings of the spray solenoid valve 16. The speed of the variable frequency spray pump 14 is adjusted in linkage to dynamically match the actual flow rate with the target flow rate. This ensures that the flow rate and pressure of the cooling medium are stable, allowing each secondary heat-generating point 13 to obtain uniform and efficient spray heat dissipation. It can also flexibly adapt to different heat dissipation needs, avoid abnormal system pressure or flow redundancy / insufficiency, and take into account the spray heat dissipation effect, system stability and energy saving.

[0074] In some embodiments, the server heat dissipation device also has the functions of pressure abnormality warning and pressure abnormality protection. To achieve this function, the flow regulation module also includes a pressure relief valve 23, which is set on the liquid inlet main line 9 or liquid outlet main line of the heat exchange module and is used to relieve pressure on the heat exchange medium in the heat exchange module.

[0075] The control module is configured to: preset high-voltage alarm threshold P alarm_H Low-voltage alarm threshold P alarm_L and the high-pressure relief threshold P relief_H In P _act >P alarm_H or P _act <P alarm_L At the same time, it outputs a stop command to the main circulation pump 8 and a close command to the full flow regulating valve 6, and sends low-pressure or high-pressure alarm fault information to the host computer, causing the host computer to display the alarm information to the staff; in P _act >P relief_H When the pressure relief valve 23 is open, an opening control command is sent to the pressure relief valve 23 to release the pressure of the heat exchange medium in the heat exchange module.

[0076] In the above technical solution, by presetting multiple pressure thresholds and configuring pressure relief valve 23, the valve can be quickly shut off and the fault can be reported when the pressure exceeds the standard, which is convenient for staff to handle in a timely manner. At the same time, the pressure relief valve 23 can be automatically opened to relieve pressure when the pressure exceeds the high pressure relief threshold. This dual protection avoids risks such as pipeline leakage and pump damage, and greatly improves the system's pressure safety protection capability and operational reliability.

[0077] In some embodiments, a total flow sensor 13 is also provided on the main liquid inlet 9 of the heat exchange module to collect the total flow rate of the heat exchange medium on the main liquid inlet 9.

[0078] In some embodiments, see Figure 2 The control module is configured as follows: The reference speed n0 of the main circulation pump 8 and the reference speed n of the spray pump 14 are preset. _spray .

[0079] Before cooling the server, the device is first initialized: the controller reads various preset parameters from the built-in storage unit; it sends initial opening commands to each flow regulating valve 6, and confirms the valve operation is normal through the position feedback sensor; the main circulation pump 8 is started until its speed reaches the reference speed n0, and the output signals of the flow sensor 7 and pressure sensor 11 are monitored to confirm that the circulation loop of the cooling module is operating normally; the spray pump 14 is started until its speed reaches the reference speed n. _spray The system sends start / stop commands to the spray solenoid valves 16 of each spray branch pipe 17, monitors the feedback signals of the spray flow sensor 20, spray pressure sensor 19 and spray solenoid valve 16 to confirm that the spray cooling module is operating normally; and reads the initial status of all GPUs and the initial temperature of all secondary heat points 13 through the status monitoring module. If some GPUs have been started or the temperature of secondary heat points 13 is abnormal, the relevant information is recorded.

[0080] The following data will be continuously collected according to the preset collection cycle: GPU Status Data: The GPU status acquisition unit collects the running status of each GPU, denoted as S_i, i=1,2,...,n, where n is the total number of GPUs. The running status of the GPU is recorded through S_i=1 and S_i=0, where S_i=1 indicates that the i-th GPU is running and S_i=0 indicates that the i-th GPU is stopped. Temperature data: The liquid inlet temperature T of the i-th cold plate 2 is collected by temperature sensor 3 through the liquid inlet branch 4 and liquid outlet branch 5 of each cold plate 2. _in_i and outlet temperature T _out_i Calculate the inlet and outlet liquid temperature difference ΔT for each cold plate 2. _i =T _out_i -T _in_i ; Flow and pressure data: The actual flow rate Q of the i-th inlet branch 4 is collected by each flow sensor 7. _act_i The actual total flow rate Q is collected by the total flow sensor 13 of the main inlet channel 9. _total The actual system pressure P is collected by pressure sensor 11 on the main outlet channel. _act The actual flow rate Q of the spray cooling module is collected by the spray flow sensor 20 and the spray pressure sensor 19 on the spray main pipe 18. _spray_act and actual pressure P _spray_act ; Temperature data of secondary heating point 13: The actual temperature T of each secondary heating point 13 is collected by the patch temperature sensor 15. _spray_act_j ; Component status data: Collect the actual opening θ of the flow regulating valve 6 in each inlet branch 4. _act_i and the actual speed n of the main circulation pump 8 _act .

[0081] If all components pass the self-test, the device enters normal operation. If a component failure is detected, such as a valve not responding, a sensor not sending a signal, or a pump failing to start, the control module triggers a fault alarm, transmits the fault signal to the host computer, displays the fault type on the host computer, and the device suspends operation, awaiting maintenance personnel to handle it.

[0082] After the device enters normal operation, the target flow rate Q of each cold plate 2 is calculated based on the heat dissipation requirements of each GPU and each secondary heat source 13. _obj_i And the target flow rate of each spray branch pipe 17 is Q _spray And based on the actual flow rate Q of each cold plate 2 _act_i and the actual flow rate Q of each spray branch pipe 17 _spray_act Adjust the flow rate of the heat exchange module and the spray cooling module to bring the device into a stable operating state.

[0083] After the device enters stable operation, the control module continuously monitors each parameter at a 100ms acquisition cycle and performs dynamic feedback correction: GPU state change handling: If the state of one or more GPUs changes from running to stopped, immediately reduce the target flow Q of the corresponding cold plate 2. _obj_i Adjusted to Q min To avoid traffic redundancy; if the GPU is detected to change from stopped to running, immediately reduce the target traffic Q corresponding to cold plate 2. _obj_i Adjust to Q0, and based on the inlet and outlet liquid temperature difference ΔT _i Make adjustments to ensure adequate cooling for the GPU after startup.

[0084] Parameter drift correction: If, after a long period of operation, the flow deviation of some inlet branches 4 continues to exceed 5%, the control module will automatically correct the PID parameters of that branch, or transmit the deviation information to the host computer to prompt the maintenance personnel to check the device.

[0085] Energy efficiency optimization: When multiple GPUs are not running, the total target bandwidth Q is reduced. _obj_total The control module reduces the speed of the main circulation pump 8, thereby reducing its energy consumption. When all GPUs are stopped, the control module switches the heat dissipation device to energy-saving mode, controls the main circulation pump 8 to run at the lowest speed, and keeps the valves of each branch at the minimum opening to maintain the basic circulation of the heat exchange medium.

[0086] Throughout the heat dissipation process, the control module monitors the device's operating status in real time. If any of the following fault conditions occur, the fault handling procedure is immediately executed: if a flow sensor 7 fails to register a signal, valve opening feedback is abnormal, the main circulation pump 8 cannot adjust its speed, or the spray pump 14 malfunctions, the spray solenoid valve 16 does not respond, or the spray flow sensor 20 or spray pressure sensor 19 fails to register a signal, the control module marks the faulty component and triggers a fault alarm. If the number of faulty inlet branches 4 exceeds 50% of the total number of inlet branches 4, or the number of faulty spray branch pipes 17 exceeds 50% of the total number, the entire heat exchange module and the spray heat dissipation module will be shut down.

[0087] 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 covered. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A server heat dissipation device, the server comprising a motherboard and multiple processors, characterized in that, include: A heat exchange module is installed at the heat points of each processor. It contains a heat exchange medium, which absorbs the heat from the heat points to dissipate heat from the processor. The status monitoring module includes a processor status acquisition unit, which is connected to the motherboard and acquires the operating status signals of each processor in real time. The control module determines the real-time heat dissipation requirements of the heat exchange modules corresponding to each processor based on the collected processor operating status signals and outputs flow adjustment commands. The flow regulation module receives the flow regulation command and adjusts the flow rate of the heat exchange medium to regulate the heat dissipation rate of each processor.

2. The server heat dissipation device according to claim 1, characterized in that, The heat exchange module includes a cold plate, which is disposed at the heat-generating points of each processor for heat dissipation. The status monitoring module includes a temperature sensor, which is installed in the inlet and outlet branches of the cold plate to collect the inlet and outlet temperatures of the heat exchange medium in real time. The control module is configured to: preset a first temperature threshold T1 and a second temperature threshold T2, and calculate the inlet and outlet temperature difference ΔT of each cold plate based on the collected inlet and outlet liquid temperatures. _i Based on the first temperature threshold T1, the second temperature threshold T2, and the actual temperature difference ΔT _i Calculate the target flow rate Q for each cold plate _obj_i : When ΔT _i At time T1, the target flow Q _obj_i =Q0×[1+(ΔT _i -T1) / T1]; When ΔT _i < is less than T2, the target flow rate Q _obj_i = Q0 × [1 - (T2 - ΔT _i ) / T1]; When T2≤ΔT_i≤T1, the target flow rate Q _obj_i =Q0; Where Q0 is the preset base flow rate of the cold plate.

3. The server heat dissipation device according to claim 2, characterized in that, The flow regulation module includes a flow regulation valve, which is installed on the liquid inlet branch of each cold plate. The flow regulation valve adjusts its opening degree according to the flow regulation command to regulate the flow rate of the heat exchange medium in each cold plate. The status monitoring module includes flow sensors installed on the liquid inlet branches of each cold plate, and the flow sensors are used to collect the actual flow rate of the heat exchange medium. The control module is configured to: acquire the actual flow rate in the inlet branch through the flow sensor, and determine the actual flow rate Q of each cold plate's inlet branch. _act_i With target traffic Q _obj_i Calculate the target opening θ of the flow control valve in each inlet branch. _obj_i A pulse signal is sent to the corresponding flow control valve to control the valve to adjust the valve opening to the target value θ. _obj_i .

4. The server heat dissipation device according to claim 3, characterized in that, The control module is configured to: preset the minimum sustaining flow rate Q of the cold plate. min The minimum sustaining flow rate Q min The target flow rate Q of the cold plate is determined when the processor is stopped and T2 ≤ ΔT_i ≤ T1, and the processor's running status is less than the base flow rate Q0. _obj_i =Q min .

5. The server heat dissipation device according to claim 2, characterized in that, The heat exchange module also includes a main circulation pump. The output end of the main circulation pump is connected to each liquid inlet branch through the main liquid inlet channel, and its input end is connected to each liquid outlet branch through the main liquid outlet channel. The main circulation pump is used to drive the heat exchange medium in the cold plate to circulate and dissipate heat. The status monitoring module includes a pressure sensor, which is installed on the main liquid outlet line to collect the actual pressure P of the heat exchange medium in the main liquid outlet line. _act ; The control module is configured to preset a maximum pressure threshold P. max and minimum pressure threshold P min In P _act >P max or P _act <P min At that time, the target speed of the main circulation pump is calculated, and a speed adjustment command is output to the main circulation pump; The formula for calculating the target rotational speed is: n _obj =n _act -ΔP K, ΔP=P _act -[(P min +P max [) / 2]; where n _act The actual speed of the main circulation pump, n _objt K is the target speed of the main circulation pump, and K is the preset pressure-speed adjustment coefficient.

6. The server heat dissipation device according to any one of claims 1-5, characterized in that, It also includes a host computer, which is connected to the server; The control module is configured as follows: Preset safe temperature threshold T max and safety duration threshold t max ; The collected liquid outlet temperature T of each cold plate was determined. _out_i greater than the safe temperature threshold T max At that time, the processor's overheating fault information is sent to the host computer; At the liquid outlet temperature T of each cold plate _out_i greater than the safe temperature threshold T max And the duration exceeds the safe duration threshold t. max At that time, a shutdown signal is sent to the server.

7. The server heat dissipation device according to claim 4, characterized in that, Also includes: A spray cooling module includes spray heads arranged toward each secondary heat-generating point, which dissipate heat by spraying cooling medium onto the secondary heat-generating points. The status monitoring module includes a spray temperature sensor, which is used to collect the real-time temperature of each secondary heating point. The control module determines the real-time heat dissipation requirements of each secondary heating point based on the collected real-time temperature of each secondary heating point, and outputs flow rate adjustment commands for each spray head. The flow regulation module includes a spray solenoid valve, which opens or closes according to the flow regulation command of each spray head, thereby controlling the start and stop of spraying of each spray head to regulate the heat dissipation rate of each secondary heat-generating point. The secondary heat sources include other heat sources within the server besides the processor.

8. The server heat dissipation device according to claim 7, characterized in that, The control module is configured to: preset the spray start-up temperature threshold T _spray Spray stop temperature threshold T _stop According to the temperature T of the secondary heating point _ j and spray start-up temperature threshold T _spray Spray stop temperature threshold T _stop Output spray adjustment commands: When T _j >T _spray At that time, an opening command is output to the spray solenoid valve corresponding to the secondary heating point; When T _j <T _stop At that time, a shut-off command is output to the spray solenoid valve corresponding to the secondary heating point; When T _stop <T _j <T _spray When it is determined that no spray adjustment command needs to be output, it is clear that no such command is required.

9. The server heat dissipation device according to claim 8, characterized in that, The spray cooling module also includes spray branch pipes, spray main pipes, spray power units, and spray liquid storage units; The input end of the main spray pipe is connected to the spray storage unit via the spray power unit. After the output end of the main spray pipe extends to the secondary heating point area, it forms a spray branch pipe corresponding to each secondary heating point. The end of each spray branch pipe is connected to the spray head of each secondary heating point. The spray solenoid valve is installed on each spray branch pipe and is used to control the start and stop of spraying on the corresponding spray branch pipe. The status monitoring module includes a spray flow sensor and a spray pressure sensor installed on the spray main pipe. The spray flow sensor and the spray pressure sensor are used to collect the flow rate and pressure of the cooling medium in the spray main pipe, respectively. The control module is configured to adjust the output power of the spray power unit based on the flow rate and pressure data of the cooling medium collected by the spray flow sensor and the spray pressure sensor, thereby adjusting the flow rate and pressure of the cooling medium.

10. The server heat dissipation device according to claim 9, characterized in that, The spray power unit includes a spray pump, and the flow rate of the cooling medium is adjusted by adjusting the speed of the spray pump. The control module is configured to: preset the target flow rate of each spray branch pipe, and calculate the target flow rate Q of the main spray pipe based on the number of opening spray solenoid valves. _spray ; Compare the actual flow rate Q of the main sprinkler pipe _spray_act and target traffic Q _spray Size: when Q _spray_act <Q _spray When Q is reached, the rotational speed of the spray pump is increased; when Q is reached... _spray_act >Q _spray When this is the case, reduce the speed of the spray pump.

Images