Server fan control method, system, equipment and medium

Through the combination of support vector regression model and adaptive weighted particle swarm optimization algorithm, the problem that traditional fan control systems cannot cope with complex load and temperature changes is solved, and intelligent control of server fan speed is realized, reducing noise and energy consumption, and achieving performance Balance with mute.

CN118934710APending Publication Date: 2024-11-12西安超越申泰信息科技有限公司
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411252686.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-06
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

Traditional server fan speed control systems cannot effectively deal with complex and variable workloads and temperature changes, resulting in waste of resources or insufficient heat dissipation, affecting server performance and energy consumption.

Method used

Support vector regression (SVR) model is used to combine adaptive weighted particle swarm optimization (AWPSO) and grid search (GS) algorithm (AWPGS), and intelligent control of fan speed is used by the historical data of server hardware and real-time temperature and load data.

Benefits of technology

It realizes refined control of server fan speed, minimizes noise and energy consumption, and achieves an ideal balance between performance and silence effect.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118934710A_ABST
    Figure CN118934710A_ABST
Patent Text Reader

Abstract

The invention discloses a server fan control method. The method comprises the following steps: training an SVR model by using a GS algorithm and an AWPSO algorithm based on historical data of multiple pieces of hardware of a server; acquiring real-time temperature and real-time load data of a plurality of pieces of hardware of the server; inputting the real-time temperature and the real-time load data of the plurality of pieces of hardware into a trained SVR model to obtain the rotating speed of a fan; and controlling the fan based on the rotating speed of the fan. The invention further discloses a system, computer equipment and a readable storage medium. According to the scheme, the temperature sensor in the server is used for monitoring the temperature and working load information of the key assembly in real time, the SVR model is combined, the AWPSO algorithm and the GS algorithm are combined to form the AWPGS algorithm to optimize the SVR hyper-parameters, intelligent and refined control over the rotating speed of the server fan is achieved, and on the basis that the stable performance of the server is guaranteed, the service life of the server is prolonged. And the noise level and the energy consumption are reduced to the maximum extent, so that the ideal balance between the performance and the mute effect is achieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of servers, and in particular to a server fan control method, system, device and storage medium. Background Art

[0002] With the continuous expansion of data center scale and the rapid development of cloud computing, big data and other fields, the reliability and energy efficiency of servers, as an important part of the core infrastructure, have become increasingly prominent. Among them, the heat dissipation management inside the server is a crucial task, which directly affects the performance stability, life span and overall energy consumption of the server. The key components inside the server, such as the central processing unit (CPU), memory, hard disk and power supply, will generate a lot of heat when operating under high load. If it is not dissipated in a timely and effective manner, it may cause overheating failure and affect the normal operation of the system.

[0003] Traditional fan speed control systems mostly use fixed thresholds or simple linear rules to adjust fan speed. This strategy often cannot cope with complex and changing workloads and temperature changes, and can easily lead to resource waste or insufficient heat dissipation. For example, over-rotating the fan under low load or low temperature conditions not only increases energy consumption, but also generates unnecessary noise. Failure to increase the fan speed in time under high load or high temperature conditions may cause the risk of system overheating.

[0004] In recent years, the application of machine learning technology in the field of server hardware management and intelligent control has gradually attracted attention. In particular, Support Vector Regression (SVR), as an effective nonlinear prediction model that can flexibly capture complex input-output relationships, is widely used to predict and control the dynamic behavior of the system.

[0005] However, the performance of the SVR model depends largely on the choice of its hyperparameters. Traditional hyperparameter optimization methods such as grid search are comprehensive but inefficient, while Particle Swarm Optimization (PSO), as a heuristic global optimization algorithm, has high search efficiency but may have premature convergence problems in some cases. Summary of the invention

[0006] In view of this, in order to overcome at least one aspect of the above problems, an embodiment of the present invention provides a server fan control method, comprising the following steps: Based on the historical data of multiple hardware of the server, the SVR model is trained using the GS algorithm and the AWPSO algorithm; Get the real-time temperature and load data of multiple hardware of the server; Inputting the real-time temperature and real-time load data of the plurality of hardware into the trained SVR model to obtain the fan speed; The fan is controlled based on a rotational speed of the fan.

[0007] In some embodiments, obtaining real-time temperature and real-time load data of multiple hardware of the server further includes: The CPU temperature, memory temperature, hard disk temperature and power supply temperature of the server are obtained, as well as the voltage of the CPU, the voltage of the memory and the current of the hard disk.

[0008] In some embodiments, based on the historical data of multiple hardware of the server, the SVR model is trained using the GS algorithm and the AWPSO algorithm, further comprising: Constructing a sample set using historical temperatures, historical load data, and corresponding fan speeds of multiple hardware of the server; Dividing the sample set into a training set, a validation set and a test set; The SVR model is trained using the training set and the GS algorithm and the AWPSO algorithm, the SVR model is tested using the test set, and the SVR model is verified using the verification set.

[0009] In some embodiments, constructing a sample set using historical temperatures, historical load data, and corresponding fan speeds of multiple hardware of the server further includes: Normalizing historical temperatures, historical load data, and corresponding fan speeds of multiple hardware of the server; A sample set is constructed using normalized historical temperatures, historical load data, and corresponding fan speeds of multiple hardware.

[0010] In some embodiments, training the SVR model using the training set and the GS algorithm and the AWPSO algorithm further includes: Determining an initial candidate set of several parameters to be optimized in the SVR model; Constructing a hyperparameter space and a hyperparameter grid based on the initial candidate set of parameters to be optimized; Using the GS algorithm to traverse the hyperparameter space to obtain multiple parameter combinations to be optimized; The plurality of parameter combinations to be optimized are used as initial particles, and the AWPSO algorithm is used to update the speed and position of the iterative particles in the hyperparameter grid to obtain new particles; In response to the performance of the SVR model corresponding to the new particle reaching a preset requirement, the training is stopped.

[0011] In some embodiments, the plurality of parameters to be optimized are combined as initial particles, and the AWPSO algorithm is used to update the speed and position of the iterative particles in the hyperparameter grid to obtain new particles, further comprising: In response to the number of update iterations reaching a preset number, a number of first particles having fitness greater than a threshold are selected from the obtained particle population; The hyperparameter space and the hyperparameter grid are reconstructed based on the candidate sets corresponding to the plurality of first particles, and the step of traversing the hyperparameter space by using the GS algorithm to obtain a plurality of parameter combinations to be optimized is returned.

[0012] In some embodiments, determining an initial candidate set of several parameters to be optimized in the SVR model further includes: The γ parameter of the penalty term coefficient kernel function and the penalty factor C are taken as parameters to be optimized.

[0013] Based on the same inventive concept, according to another aspect of the present invention, an embodiment of the present invention further provides a server fan control system, including: The training module is configured to train the SVR model using the GS algorithm and the AWPSO algorithm based on the historical data of multiple hardware of the server; An acquisition module, configured to acquire real-time temperature and real-time load data of multiple hardware of the server; A calculation module, configured to input the real-time temperature and real-time load data of the plurality of hardware into the trained SVR model to obtain the rotation speed of the fan; The control module is configured to control the fan based on the rotation speed of the fan.

[0014] Based on the same inventive concept, according to another aspect of the present invention, an embodiment of the present invention further provides a computer device, including: at least one processor; and A memory, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the program, the processor performs the steps of any one of the server fan control methods described above.

[0015] Based on the same inventive concept, according to another aspect of the present invention, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of any one of the server fan control methods described above are performed.

[0016] The present invention has one of the following beneficial technical effects: the solution proposed by the present invention utilizes the internal temperature sensor of the server to monitor the temperature and workload information of key components in real time, combines the SVR model and uses the AWPSO algorithm and the GS algorithm to form the AWPGS algorithm to optimize the SVR hyperparameters, thereby realizing intelligent and refined control of the server fan speed, so as to minimize the noise level and energy consumption while ensuring the stability of the server performance, thereby achieving an ideal balance between performance and quietness. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other embodiments can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A schematic flow chart of a server fan control method provided by an embodiment of the present invention; Figure 2 A flow chart of a server fan control method provided by an embodiment of the present invention; Figure 3 A schematic diagram of the structure of a server fan control system provided by an embodiment of the present invention; Figure 4 A schematic diagram of the structure of a computer device provided by an embodiment of the present invention; Figure 5 A schematic diagram of the structure of a computer-readable storage medium provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the embodiments of the present invention are further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0020] It should be noted that all expressions using "first" and "second" in the embodiments of the present invention are for distinguishing two non-identical entities with the same name or non-identical parameters. It can be seen that "first" and "second" are only for the convenience of expression and should not be understood as limitations on the embodiments of the present invention. The subsequent embodiments will not explain this one by one.

[0021] According to one aspect of the present invention, an embodiment of the present invention provides a server fan control method, such as Figure 1 As shown, it may include the steps of: S1, based on the historical data of multiple hardware of the server, the SVR model is trained using the GS algorithm and the AWPSO algorithm; S2, obtains the real-time temperature and real-time load data of multiple hardware of the server; S3, inputting the real-time temperature and real-time load data of the plurality of hardware into the trained SVR model to obtain the fan speed; S4: Control the fan based on the rotation speed of the fan.

[0022] The solution proposed in the present invention utilizes the internal temperature sensor of the server to monitor the temperature and workload information of key components in real time, combines the SVR model and uses the AWPSO algorithm and the GS algorithm to form the AWPGS algorithm to optimize the SVR hyperparameters, thereby realizing intelligent and refined control of the server fan speed, so as to minimize the noise level and energy consumption while ensuring the stability of the server performance, thereby achieving an ideal balance between performance and quietness.

[0023] In some embodiments, obtaining real-time temperature and real-time load data of multiple hardware of the server further includes: The CPU temperature, memory temperature, hard disk temperature and power supply temperature of the server are obtained, as well as the voltage of the CPU, the voltage of the memory and the current of the hard disk.

[0024] In some embodiments, based on the historical data of multiple hardware of the server, the SVR model is trained using the GS algorithm and the AWPSO algorithm, further comprising: Constructing a sample set using historical temperatures, historical load data, and corresponding fan speeds of multiple hardware of the server; Dividing the sample set into a training set, a validation set and a test set; The SVR model is trained using the training set and the GS algorithm and the AWPSO algorithm, the SVR model is tested using the test set, and the SVR model is verified using the verification set.

[0025] In some embodiments, constructing a sample set using historical temperatures, historical load data, and corresponding fan speeds of multiple hardware of the server further includes: Normalizing historical temperatures, historical load data, and corresponding fan speeds of multiple hardware of the server; A sample set is constructed using normalized historical temperatures, historical load data, and corresponding fan speeds of multiple hardware.

[0026] Specifically, the temperature and load data of key components of the server are collected, and necessary normalization processing is performed to ensure that data of different dimensions are uniformly within a similar numerical range, which helps to enhance the training efficiency and accuracy of the SVR model. Then, the preprocessed data is input into the SVR model for training. At the same time, a dual optimization strategy is adopted, that is, the GS algorithm and the AWPSO algorithm are combined to jointly optimize and adjust the hyperparameters of the SVR model comprehensively and accurately. After the SVR model training is completed, it is deployed in a real-time environment to capture and process the system status monitoring data stream in real time. Whenever new temperature and load information is received, the trained SVR model is used for instant analysis and prediction to obtain the most suitable fan speed value for the current system condition, thereby achieving dynamic balance management between server cooling efficiency and operating noise.

[0027] The present invention establishes a sample set based on the temperature data of the server CPU, memory, hard disk, power supply, the load data of the CPU, memory, hard disk and the fan speed. ,in , represents the ideal fan speed under the corresponding state. Because the input data distribution is different and the dimensions are different, the normalization method shown in formula (1.1) is adopted, where is the original data value.

[0028]

[0029] In some embodiments, training the SVR model using the training set and the GS algorithm and the AWPSO algorithm further includes: Determining an initial candidate set of several parameters to be optimized in the SVR model; Constructing a hyperparameter space and a hyperparameter grid based on the initial candidate set of parameters to be optimized; Using the GS algorithm to traverse the hyperparameter space to obtain multiple parameter combinations to be optimized; The plurality of parameter combinations to be optimized are used as initial particles, and the AWPSO algorithm is used to update the speed and position of the iterative particles in the hyperparameter grid to obtain new particles; In response to the performance of the SVR model corresponding to the new particle reaching a preset requirement, the training is stopped.

[0030] In some embodiments, the plurality of parameters to be optimized are combined as initial particles, and the AWPSO algorithm is used to update the speed and position of the iterative particles in the hyperparameter grid to obtain new particles, further comprising: In response to the number of update iterations reaching a preset number, a number of first particles having fitness greater than a threshold are selected from the obtained particle population; The hyperparameter space and the hyperparameter grid are reconstructed based on the candidate sets corresponding to the plurality of first particles, and the step of traversing the hyperparameter space by using the GS algorithm to obtain a plurality of parameter combinations to be optimized is returned.

[0031] In some embodiments, determining an initial candidate set of several parameters to be optimized in the SVR model further includes: The γ parameter of the penalty term coefficient kernel function and the penalty factor C are taken as parameters to be optimized.

[0032] Specifically, SVR is selected as the prediction model because of its excellent generalization ability and good fitting characteristics for nonlinear relationships. Temperature and load characteristics are used as input variables, and fan speed is used as output variable. Since the problem designed by the present invention is a nonlinear prediction problem and there is no clear linear relationship, the kernel function is a radial basis kernel function (RBF).

[0033] In order to improve the generalization ability of the model and ensure that the model can achieve good prediction results when facing new and unseen data, the data set is randomly divided into 70% as a training set, 15% as a validation set, and 15% as a test set. And other parameters.

[0034] Adaptive weighted particle swarm optimization (AWPSO) is proposed to improve the problem of PSO being prone to premature maturity and falling into local optimality. In AWPSO, the concept of adaptive weight is introduced, that is, the inertia weight of different particles is dynamically adjusted according to different states or dimensions, so that the algorithm can achieve a better balance between global search and local fine search, thereby improving the algorithm's search ability and the probability of ultimately finding the global optimal solution.

[0035] The grid search algorithm is a machine learning hyperparameter optimization method that traverses pre-defined hyperparameter combinations and uses cross-validation to evaluate the model performance under each set of hyperparameters. Its advantages are its comprehensiveness and systematicity, ensuring that the optimal solution is not missed within the preset hyperparameter range. However, its disadvantage is that when the number of hyperparameters is large or the range of values ​​is wide, the computational cost is very high.

[0036] The present invention combines the characteristics of the AWPSO and GS algorithms to design a dynamic adjustment search strategy, that is, in the early stage of the search, it mainly relies on AWPSO for rapid convergence. As the search progresses, the proportion of grid search is gradually increased. In particular, when multiple candidate optimal solutions appear in the search area, the search of the area is refined by the GS algorithm. The algorithm realizes the organic unity of global optimization and local refined search, and has fast convergence speed and low computational cost.

[0037] like Figure 2 As shown, first, the hyperparameters and their value ranges that need to be optimized in the SVR model are clarified. In the present invention, the hyperparameters are the γ parameter and the penalty factor C of the penalty coefficient kernel function. The "grid" formed by the candidate value set of the parameters γ and C is used as the hyperparameter space; then, the algorithm initialization phase is entered, which mainly includes the use of creating a hyperparameter grid and particle swarm initialization. In the particle swarm algorithm, each solution is regarded as a "particle" in the search space. In the present invention, the particle is a combination of each γ and C. Among them, each particle represents a hyperparameter combination, the position of the particle represents the hyperparameter value currently searched, and the speed represents the search direction and step size. Finally, the hyperparameter optimization phase is entered to use the GS algorithm and the AWPSO algorithm to complete the hyperparameter determination of the SVR model.

[0038] In the optimization stage, the GS algorithm is first used to traverse the hyperparameter space to obtain an initial performance evaluation table, which contains the model performance indicators corresponding to different hyperparameter combinations; then, based on some excellent hyperparameter combinations obtained by grid search, the AWPSO algorithm is started as the population position of the AWPSO algorithm. Each generation of particles updates its position (i.e., hyperparameter combination) and calculates the corresponding model performance. The fitness function is used to evaluate the fitness of the particles, and the speed and position of the particles are updated according to the adaptive weights, gradually approaching the optimal solution area. The particle update rules of the AWPSO algorithm are as follows:

[0039] in, It is a particle In Dimension The speed on The value at the iteration; It is a particle In Dimension The speed on The value at the iteration; is the inertia weight, , is the current iteration number, T is the predetermined maximum number of iterations; and are cognitive and social learning factors, respectively, controlling the degree to which particles respond to their individual best positions and group best positions; and is a random number between [0,1]; It is a particle The best position found in its history is at the dimension The value on The entire group is in dimension The best position found on the grid is finally integrated into the AWPSO iteration process. The integration method is: every 100 AWPSO algorithm iterations, the excellent individuals with higher fitness are optimized from the population to enter the next round of GS+AWPSO optimization process. This method promotes the expansion of the search space and the improvement of the convergence speed. When the preset maximum number of iterations is reached or the performance improvement tends to be stable, the search process is stopped. At this time, the hyperparameter combination is identified as the most optimal parameter of the SVR model.

[0040] The present invention proposes a system and method for intelligent control of server fan speed, the core of which is to use the temperature sensor on the server motherboard to continuously monitor the temperature of key components (such as CPU, memory, hard disk and power supply), while considering the overall workload indicators of the server (including but not limited to CPU utilization, memory occupancy and I / O activity). The present invention first constructs a support vector regression (SVR) model to accurately map the relationship between the above parameters and the server fan speed, and then innovatively introduces the adaptive weighted particle swarm optimization (AWPSO) and grid search (GS) fusion algorithm - AWPGS, which is used to efficiently optimize the hyperparameters of the SVR model. This intelligent control system can automatically adjust the server fan speed, effectively suppress noise while meeting the server performance requirements, and achieve an ideal balance between server performance and silent effect.

[0041] Based on the same inventive concept, according to another aspect of the present invention, an embodiment of the present invention further provides a server fan control system 400, such as Figure 3 As shown, including: The training module 401 is configured to train the SVR model using the GS algorithm and the AWPSO algorithm based on the historical data of multiple hardware of the server; An acquisition module 402 is configured to acquire real-time temperature and real-time load data of multiple hardware of the server; The calculation module 403 is configured to input the real-time temperature and real-time load data of the plurality of hardware into the trained SVR model to obtain the rotation speed of the fan; The control module 404 is configured to control the fan based on the rotation speed of the fan.

[0042] Based on the same inventive concept, according to another aspect of the present invention, Figure 4 As shown, an embodiment of the present invention further provides a computer device 501, including: at least one processor 520; and The memory 510 stores a computer program 511 that can be run on the processor. When the processor 520 executes the program, the processor 520 performs the steps of any of the above server fan control methods.

[0043] Based on the same inventive concept, according to another aspect of the present invention, Figure 5 As shown, an embodiment of the present invention further provides a computer-readable storage medium 601, which stores a computer program 610. When the computer program 610 is executed by a processor, the steps of any of the above server fan control methods are performed.

[0044] Finally, it should be noted that a person skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods.

[0045] Furthermore, it should be appreciated that the computer-readable storage medium (eg, memory) herein may be either volatile memory or nonvolatile memory, or may include both volatile memory and nonvolatile memory.

[0046] It will also be appreciated by those skilled in the art that various exemplary logic blocks, modules, circuits and algorithm steps described in conjunction with the disclosure herein can be implemented as electronic hardware, computer software or a combination of the two. In order to clearly illustrate this interchangeability of hardware and software, a general description has been given to the functions of various schematic components, blocks, modules, circuits and steps. Whether this function is implemented as software or hardware depends on specific applications and the design constraints imposed on the entire system. Those skilled in the art can implement the function in various ways for each specific application, but this implementation decision should not be interpreted as causing a departure from the disclosed scope of the embodiments of the present invention.

[0047] The above are exemplary embodiments disclosed in the present invention, but it should be noted that various changes and modifications may be made without departing from the scope disclosed in the embodiments of the present invention as defined in the claims. The functions, steps and / or actions of the method claims according to the disclosed embodiments described herein do not need to be performed in any particular order. In addition, although the elements disclosed in the embodiments of the present invention may be described or required in individual form, they may also be understood as multiple unless explicitly limited to the singular.

[0048] It should be understood that, as used herein, the singular forms "a", "an" are intended to include the plural forms as well, unless the context clearly supports an exception. It should also be understood that, as used herein, "and / or" refers to any and all possible combinations including one or more of the associated listed items.

[0049] The serial numbers of the embodiments disclosed in the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0050] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.

[0051] A person skilled in the art should understand that the discussion of any of the above embodiments is only exemplary and is not intended to imply that the scope of the disclosure of the embodiments of the present invention (including the claims) is limited to these examples; under the concept of the embodiments of the present invention, the technical features in the above embodiments or different embodiments can also be combined, and there are many other changes in different aspects of the above embodiments of the present invention, which are not provided in detail for the sake of simplicity. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present invention should be included in the protection scope of the embodiments of the present invention.

Claims

1. A server fan control method, characterized in that: The following steps are involved: Based on the historical data of multiple hardware of the server, the SVR model is trained using the GS algorithm and the AWPSO algorithm; Acquire real-time temperature and real-time load data of the plurality of hardware of the server; Inputting the real-time temperature and real-time load data of the plurality of hardware into the trained SVR model to obtain the fan speed; The fan is controlled based on a rotational speed of the fan.

2. The method according to claim 1, characterized in that Acquiring real-time temperature and real-time load data of the plurality of hardware of the server further comprises: The CPU temperature, memory temperature, hard disk temperature and power supply temperature of the server are obtained, as well as the voltage of the CPU, the voltage of the memory and the current of the hard disk.

3. The method according to claim 1, characterized in that Based on the historical data of multiple hardware of the server, the SVR model is trained using the GS algorithm and the AWPSO algorithm, further including: Constructing a sample set using historical temperatures, historical load data, and corresponding fan speeds of multiple hardware of the server; Dividing the sample set into a training set, a validation set and a test set; The SVR model is trained using the training set and the GS algorithm and the AWPSO algorithm, the SVR model is tested using the test set, and the SVR model is verified using the verification set.

4. The method according to claim 3, characterized in that Constructing a sample set using historical temperatures, historical load data, and corresponding fan speeds of multiple hardware of the server further includes: Normalizing historical temperatures, historical load data, and corresponding fan speeds of multiple hardware of the server; A sample set is constructed using normalized historical temperatures, historical load data, and corresponding fan speeds of multiple hardware.

5. The method according to claim 3, characterized in that The SVR model is trained using the training set and the GS algorithm and the AWPSO algorithm, further comprising: Determining an initial candidate set of several parameters to be optimized in the SVR model; Constructing a hyperparameter space and a hyperparameter grid based on the initial candidate set of parameters to be optimized; Using the GS algorithm to traverse the hyperparameter space to obtain multiple parameter combinations to be optimized; The plurality of parameter combinations to be optimized are used as initial particles, and the AWPSO algorithm is used to update the speed and position of the iterative particles in the hyperparameter grid to obtain new particles; In response to the performance of the SVR model corresponding to the new particle reaching a preset requirement, the training is stopped.

6. The method according to claim 5, characterized in that The plurality of parameters to be optimized are combined as initial particles, and the AWPSO algorithm is used to update the speed and position of the iterative particles in the hyperparameter grid to obtain new particles, further comprising: In response to the number of update iterations reaching a preset number, a number of first particles having fitness greater than a threshold are selected from the obtained particle population; The hyperparameter space and the hyperparameter grid are reconstructed based on the candidate sets corresponding to the plurality of first particles, and the step of traversing the hyperparameter space by using the GS algorithm to obtain a plurality of parameter combinations to be optimized is returned.

7. The method according to claim 5, characterized in that Determining an initial candidate set of several parameters to be optimized in the SVR model further includes: The γ parameter of the penalty term coefficient kernel function and the penalty factor C are taken as parameters to be optimized.

8. A server fan control system, characterized in that: include: The training module is configured to train the SVR model using the GS algorithm and the AWPSO algorithm based on the historical data of multiple hardware of the server; An acquisition module, configured to acquire real-time temperature and real-time load data of the plurality of hardware of the server; A calculation module, configured to input the real-time temperature and real-time load data of the plurality of hardware into the trained SVR model to obtain the rotation speed of the fan; The control module is configured to control the fan based on the rotation speed of the fan.

9. A computer device comprising: at least one processor; as well as A memory storing a computer program executable on the processor, wherein the processor executes the steps of the method according to any one of claims 1 to 7 when executing the program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are performed.

Citation Information

Cited By

  • Control method and device of heat dissipation equipment

    CN120447701A

  • Dynamic control method and system for modularized energy-saving array of server fan

    CN121255005A