Data center server heat dissipation optimization method, equipment and medium
The Kalman filtering and weighted averaging method combined with PID algorithm optimized fan control, which solves the problems of high energy consumption and noise of server fans, realizes the reduction of fan power consumption and temperature stability, and reduces the total fan power under the conditions of normal operation of the server chip. Consumption.
Patent Information
- Application Number
- CN202411105393.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-13
- Publication Date
- 2024-12-03
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art has problems of high energy consumption and poor noise when reducing the energy consumption of server fan cooling systems. Especially during the temperature regulation process, the PID control method is prone to oscillation and low-frequency vibration.
The real-time shell temperature of the temperature sensor is fused in real time with the Kalman filtering algorithm. After processing with the weighted average method, the PID algorithm is used to adjust the PWM duty cycle and rotation speed of the fan to achieve optimization of fan power consumption.
在满足服务器芯片正常工作的条件下,降低风扇总功耗,减少发热部件壳温震荡,降低风扇噪音,实现风扇功耗的降低2.47%以上。
Smart Images

Figure CN119065924A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of heat dissipation of data center servers, and in particular to a method, equipment and medium for optimizing heat dissipation of data center servers. Background Art
[0002] With the development of the digital economy, data centers not only create economic value but also bring high energy consumption, aggravating carbon emissions and air pollution. The energy consumption of server hardware cannot be significantly reduced in the short term. Therefore, how to reduce the energy consumption of the fan cooling system has become the primary task of server energy saving and consumption reduction.
[0003] In a data center, IT equipment such as servers accounts for more than 50% of the total energy consumption of the data center, and the power consumption of fans used for heat dissipation inside the server accounts for more than 20% of the total energy consumption of the server. Therefore, under the premise of ensuring the reliable operation of the server chip, effectively reducing the speed of the fan can effectively reduce part of the heat dissipation power consumption.
[0004] At present, the basis for the speed control of server fans is the shell temperature or air temperature of the CPU or other high-power devices. Generally, a temperature sensor and a BMC controller are installed in the shell or accessory position of the high-power components of the server motherboard, and the temperature sensor value is read in real time. The fan speed is controlled according to the control strategy. There are mainly two types: one is the PWM segmented speed control method, which sets the segmented temperature threshold in advance, selects the one with the smallest distance threshold from multiple temperature sensors, usually the one with the largest temperature value), and determines the duty cycle of the corresponding fan PWM signal. The disadvantage of this method is that if the temperature value is larger than the target value, the duty cycle of the PWM control signal needs to be achieved step by step through segments, rather than starting the maximum speed. In this process, more energy is consumed and the noise lasts for a long time; the other is to use the PID control method to control the fan according to the real-time change of the CPU shell temperature. This method realizes the control of fast fans. However, the problem is that in the final convergence stage, oscillation often occurs, so the fan will have low-frequency vibration, poor noise quality, and affect the user experience.
[0005] The patent document with document number CN113761758B discloses a method for optimizing the heat dissipation performance of a water-cooled head radiator, a radiator and a server. This application sets the water inlet temperature and microchannel fin parameters as parameters to be optimized, sets the adjustment value range of each parameter to be optimized and establishes a heat dissipation model of the water-cooled head radiator. When one parameter is stable, another parameter is adjusted and the corresponding heat dissipation performance is tested, and finally the optimized value of the parameter is determined. This application can realize the effective selection of the optimized value of the water-cooled head radiator parameter, and the water-cooled head radiator designed by the optimized value of the parameter can meet the required heat dissipation performance requirements. However, its solution also has the problem that the software system for optimizing server heat dissipation is not optimized. Summary of the invention
[0006] In view of the above problems, the purpose of the present invention is to provide a data center server heat dissipation optimization method, device and medium to optimize the server cooling fan control, while ensuring the normal operation of the server chip, while reducing the total power consumption of the fan.
[0007] The embodiments of the present invention provide a method, a device and a medium for optimizing heat dissipation of a data center server.
[0008] A first aspect: A method for optimizing heat dissipation of a data center server, comprising:
[0009] S1. Determine the heat generating components of the server, install temperature sensors on each heat generating component, and obtain the real-time shell temperature T of the heat generating component. n ;
[0010] S2, using Kalman filter algorithm to measure the real-time shell temperature T of the temperature sensor n Perform real-time fusion to obtain shell fusion temperature;
[0011] S3. Use the weighted average method to process the shell temperature fusion temperature and find the optimal temperature value X i ,
[0012] S4, based on the optimal temperature value X i , combined with the PID algorithm to adjust the fan's PWM duty cycle, control the fan speed and the fan's real-time power consumption.
[0013] Furthermore, the heat generating components include a CPU chip, a GPU chip and a PCIe switch chip.
[0014] The method for optimizing heat dissipation of a data center server according to claim 1 is characterized in that the Kalman filter algorithm is used to calculate the real-time shell temperature X of the temperature sensor. n Perform real-time fusion, including
[0015]
[0016] Y i (t) = HT i (t)+V(t)
[0017] Among them, X(t) represents the shell temperature value at the current moment, X(t-1) represents the shell temperature at the previous moment, W(t) is the process noise, V(t) is the measurement noise, and Y i (t) is the real-time fusion value of the i-th shell temperature at time t.
[0018] Furthermore, the shell temperature fusion temperature is processed by weighted average method to find the optimal temperature value X i (t), including:
[0019]
[0020] Among them, a1…a n is the weighted number, Y i (t) is the real-time fusion value of the i-th shell temperature at time t, 1≤i≤n.
[0021] Furthermore, the PWM signal duty ratio and the speed of the fan are at the lowest speed n min With the maximum speed n max There is a linear relationship between the stages.
[0022] Furthermore, the power consumption of the fan is in direct proportion to the cube of the rotation speed.
[0023] A second aspect: An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method provided in the first aspect when executing the program.
[0024] A third aspect: A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the method provided in the first aspect are implemented.
[0025] Beneficial effects of the present invention:
[0026] The present invention determines the server heating component and obtains the real-time shell temperature T of the heating component. n , the Kalman filter algorithm is used to calculate the real-time shell temperature T of the temperature sensor n Perform real-time fusion to obtain the shell temperature fusion temperature; use the weighted average method to process the shell temperature fusion temperature to find the optimal temperature value X i , according to the optimal temperature value X i , combined with the PID algorithm to adjust the fan's PWM duty cycle, control the fan speed and the fan's real-time power consumption, it can reduce the total power consumption of the fan while ensuring the normal operation of the server chip. At the same time, the shell temperature of the heat-generating component fluctuates little, so the fan is less likely to have low-frequency vibration and produce less noise. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] Figure 1 A schematic diagram of a process flow of a method for optimizing heat dissipation of a data center server according to the present invention;
[0028] Figure 2 This is a schematic diagram of the layout of the heat generating components according to an embodiment of the present invention;
[0029] Figure 3 It is a schematic diagram comparing the fan power consumption of Kalman filtering + least squares method and least squares method of the present invention;
[0030] Figure 4 It is a schematic diagram of the shell temperature change of the heat generating component of the Kalman filter + weighted average method and the weighted average method of the present invention;
[0031] Figure 5 It is a structural schematic diagram of the electronic device of the present invention; DETAILED DESCRIPTION
[0032] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar symbols throughout represent the same or similar elements or elements with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.
[0033] While creating economic value, servers in data centers also bring about high energy consumption, exacerbating carbon emissions and air pollution. The energy consumption of server hardware cannot be significantly reduced in the short term, so how to reduce the energy consumption of fan cooling systems has become the primary task of server energy conservation and consumption reduction.
[0034] In view of the above problems, the present invention provides a method for optimizing heat dissipation of a data center server. Figure 1 A schematic diagram of a process flow of a data center server heat dissipation optimization method provided by an embodiment of the present invention, the method comprising:
[0035] S1. Determine the heat generating components of the server, install temperature sensors on each heat generating component, and obtain the real-time shell temperature T of the heat generating component. n .
[0036] by Figure 2 For example, the server heat generating components include 2 CPUs and 2 GPUs, namely CPU1, CPU2, GPU3, and GPU4. The real-time power consumption value of each heat generating component is P n (n=1, 2, 3, 4), P n This can be obtained through system monitoring. A built-in fan is installed in the server to generate air flow, remove the heat generated by the CPU or GPU, and ensure that the chip shell temperature remains within the chip's operating temperature range.
[0037] For example, the maximum shell temperature of the processor of the XEON series server is generally 75°C, and the ideal operating temperature is less than 60°C. A temperature sensor is installed on each heat-generating component to detect the shell temperature in real time. The shell temperature of the heat-generating component is T n (n=1, 2, 3, 4).
[0038] The heat removed by the fan can be expressed by the following formula:
[0039]
[0040] Where, Q is the heat removed by the fan, in J; C p is the constant pressure specific heat capacity of air, in J / (kg·℃); CMM is the air volume provided by the fan per minute, in m3 / min; ρ is the air density, about 1200g / m3; ΔT C is the temperature difference between the inlet and outlet, in °C.
[0041] From the above formula, we can know that the power consumption value P of the heat-generating component is n When the shell temperature T is kept constant, n Part of it depends on the fan air volume value CMM. When the air volume value CMM is larger, the more heat Q is taken away, and the device shell temperature T n The faster the decrease, the more energy is consumed; the smaller the air volume value CMM, the higher the shell temperature T n The slower it decreases, the less energy it consumes.
[0042] The air volume value CMM is closely related to the fan speed. By controlling the fan speed, the air volume value CMM can be controlled, and then the heat Q can be controlled. However, the fan speed is closely related to the fan energy consumption. It is necessary to consider the balance between the air volume value CMM and the fan energy consumption to achieve the best real-time power consumption control of the fan.
[0043] S2, using Kalman filter algorithm to measure the real-time shell temperature T of the temperature sensor n Perform real-time fusion to obtain the shell fusion temperature.
[0044] The Kalman filter algorithm can be used for real-time fusion of temperature sensor information in a dynamic environment. It is an algorithm that uses the linear system state equation to optimally estimate the system state through system input and output observation data. Specifically, the Kalman filter algorithm is used to estimate the real-time shell temperature X n Perform real-time fusion, including
[0045]
[0046] Y i (t) = HT i (t)+V(t)
[0047] Among them, X(t) represents the shell temperature value at the current moment, X(t-1) represents the shell temperature at the previous moment, W(t) is the process noise, V(t) is the measurement noise, and Y i (t) is the real-time fusion value of the i-th shell temperature at moment t. The real-time fusion temperature value of each shell temperature at each moment can be obtained through the Kalman filter algorithm.
[0048] S3. Use the weighted average method to process the shell temperature fusion temperature and find the optimal temperature value Xi .
[0049] Weighted average method: It uses the real-time shell temperature fusion temperature value of the heating component at each moment to process the shell temperature fusion temperature value using the weighted average method to find the optimal temperature value X i (t), a method for finding the average value includes:
[0050]
[0051] Among them, a1…a n is the weighted number, Y i (t) is the real-time fusion value of the i-th shell temperature at time t, 1≤i≤n.
[0052] S4, based on the optimal temperature value X i , combined with the PID algorithm to adjust the fan's PWM duty cycle, control the fan speed and the fan's real-time power consumption.
[0053] The PID algorithm is a commonly used algorithm that can adjust the PWM duty cycle of the fan according to changes in temperature values. It will not be introduced in detail here.
[0054] The fan's PWM signal duty cycle and speed show an approximately linear relationship in the stages of minimum speed nmin and maximum speed nmax, and the fan's power consumption is directly proportional to the cube of the speed.
[0055] Therefore, the relationship between the PWM duty cycle, speed, and power consumption of a specific fan model can be calculated, and the relationship between the fan control PWM signal duty cycle and the fan speed, fan power consumption, and the heat Q removed by the fan and the shell temperature T of the heat-generating component is established. n The relationship between.
[0056] The results obtained by the above method are as follows Figure 3 As shown in the results, the cumulative heat of Kalman filtering + weighted averaging method is compared with that of simple weighted averaging method. The energy consumption of Kalman filtering + weighted averaging method is reduced by more than 2.47% compared with simple weighted averaging method.
[0057] Under the premise that the power consumption of server components remains unchanged, the method of the present invention reduces the power consumption of the fan only by optimizing the algorithm, and the reduction range of the power consumption is relatively ideal.
[0058] The method used in the present invention not only focuses on the energy-saving status of the fan power consumption, but also focuses on the reliability of the chip operation.
[0059] The combination of Kalman filtering and weighted averaging is used, compared with the simple weighted averaging method. Figure 4As shown in the figure, the combined method of Kalman filtering + weighted averaging has obvious energy-saving effect, the shell temperature fluctuation is also within the permitted temperature range, and has no effect on the reliability of the device. Compared with the traditional weighted average method, the combined method of Kalman filtering + weighted averaging can reduce the total power consumption of the fan by more than 2.47% while meeting the normal operation of the server chip, and has good promotion and application value.
[0060] The present invention also provides an electronic device, Figure 5 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, such as Figure 5 As shown, the electronic device may include: a processor, a communications interface, a memory, and a communications bus, wherein the processor, the communications interface, and the memory communicate with each other via the communications bus. The processor may call the logic instructions in the memory, for example, to execute the following method:
[0061] S1. Determine the heat generating components of the server, install temperature sensors on each heat generating component, and obtain the real-time shell temperature T of the heat generating component. n ;
[0062] S2, using Kalman filter algorithm to measure the real-time shell temperature T of the temperature sensor n Perform real-time fusion to obtain shell fusion temperature;
[0063] S3. Use the weighted average method to process the shell temperature fusion temperature and find the optimal temperature value X i ,
[0064] S4, based on the optimal temperature value X i , combined with the PID algorithm to adjust the fan's PWM duty cycle, control the fan speed and the fan's real-time power consumption.
[0065] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.
[0066] An embodiment of the present invention further provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method provided in each of the above embodiments is implemented, for example, including:
[0067] S1. Determine the heat generating components of the server, install temperature sensors on each heat generating component, and obtain the real-time shell temperature T of the heat generating component. n ;
[0068] S2, using Kalman filter algorithm to measure the real-time shell temperature T of the temperature sensor n Perform real-time fusion to obtain shell fusion temperature;
[0069] S3. Use the weighted average method to process the shell temperature fusion temperature and find the optimal temperature value X i ,
[0070] S4, based on the optimal temperature value X i , combined with the PID algorithm to adjust the fan's PWM duty cycle, control the fan speed and the fan's real-time power consumption.
[0071] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.
[0072] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.
[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for optimizing heat dissipation of a data center server. It is characterized in that include: S1. Determine the heat generating components of the server, install temperature sensors on each heat generating component, and obtain the real-time shell temperature T of the heat generating component. n ; S2, using Kalman filter algorithm to measure the real-time shell temperature T of the temperature sensor n Perform real-time fusion to obtain shell fusion temperature; S3. Use the weighted average method to process the shell temperature fusion temperature and find the optimal temperature value X i , S4, based on the optimal temperature value X i , combined with the PID algorithm to adjust the fan's PWM duty cycle, control the fan speed and the fan's real-time power consumption.
2. A method for optimizing heat dissipation of a data center server according to claim 1, It is characterized in that The heat generating components include a CPU chip, a GPU chip and a PCIe switching chip.
3. A method for optimizing heat dissipation of a data center server according to claim 1, It is characterized in that The Kalman filter algorithm is used to calculate the real-time shell temperature X of the temperature sensor. n Perform real-time fusion, including Y i (t)=HT i (t)+V(t) Among them, X(t) represents the shell temperature value at the current moment, X(t-1) represents the shell temperature at the previous moment, W(t) is the process noise, V(t) is the measurement noise, and Y i (t) is the real-time fusion value of the i-th shell temperature at time t.
4. A method for optimizing heat dissipation of a data center server according to claim 3, It is characterized in that The shell temperature fusion temperature is processed by weighted average method to find the optimal temperature value X i (t), including: Among them, a 1 …a n is the weighted number, Y i (t) is the real-time fusion value of the i-th shell temperature at time t, 1≤i≤n.
5. A method for optimizing heat dissipation of a data center server according to claim 4, It is characterized in that The PWM signal duty ratio of the fan and the speed are at the minimum speed n min With the maximum speed n max There is a linear relationship between the stages.
6. A method for optimizing heat dissipation of a data center server according to claim 5, It is characterized in that The power consumption of the fan is in direct proportion to the cube of the rotation speed.
7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, It is characterized in that When the processor executes the program, the steps of a data center server heat dissipation optimization method as described in any one of claims 1 to 6 are implemented.
8. A non-transitory computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of a data center server heat dissipation optimization method as described in any one of claims 1 to 6 are implemented.
Citation Information
Patent Citations
A method for optimizing the heat dissipation performance of a water-cooled head radiator, the radiator itself, and a server.
CN113761758B
Server fan control method and device, computer equipment and storage medium
CN116339468A
Server heat dissipation abnormity detection method and device, equipment and medium
CN118093290A
Intelligent door lock control method and device based on multi-dimensional sensor
CN118447610A
Server regulation and control method and device, electronic equipment and readable storage medium
CN118466720A
