Intelligent heat dissipation control optimization method and system for server

Abstract

The invention relates to the technical field of server heat dissipation optimization, in particular to an intelligent heat dissipation control optimization method and system for a server, and the method comprises the steps: obtaining the temperature information of each region of the server and the working load data of the server; based on the temperature information, constructing multi-dimensional thermal field topological mapping; executing group theory base heat flow dynamic analysis according to the multi-dimensional thermal field topological mapping; performing spectral decomposition heat dissipation efficiency evaluation based on a group theory basis heat flow dynamic analysis result; according to the spectral decomposition heat dissipation efficiency evaluation result and the workload data, nonlinear heat transfer optimization is executed; a self-adaptive feedback control strategy is generated and used for adjusting a heat dissipation system of the server, accurate description of internal temperature distribution of the server is provided by a multi-dimensional thermal field topology mapping technology, and a foundation is laid for subsequent optimization; the group theory-based heat flow dynamic analysis method deeply reveals the essential law of heat transfer; the spectrum decomposition heat dissipation efficiency evaluation provides a brand new performance quantitative index, and the heat dissipation efficiency of the system is improved by more than 20%.

Description

Technical Field

[0001] The present invention relates to the technical field of server heat dissipation optimization, and more specifically, to a server intelligent heat dissipation control optimization method and system thereof. Background Art

[0002] With the rapid development of information technology, the scale of data centers continues to expand, server performance continues to improve, and heat dissipation issues are increasingly becoming a key bottleneck restricting the development of data centers. Traditional server heat dissipation control methods mainly rely on simple temperature threshold control or PID (proportional-integral-differential) control strategies. These methods perform well under stable conditions, but their limitations are increasingly prominent in the face of the complex and changeable load environment of modern data centers.

[0003] At present, the mainstream server heat dissipation control methods in the industry can be summarized into the following categories:

[0004] First, simple control strategies based on temperature thresholds are still widely used. This method triggers the start-up or speed regulation of heat dissipation equipment by setting a fixed temperature threshold. Although simple to implement, it is slow to respond and has difficulty coping with rapidly changing heat loads, often leading to problems of over-cooling or insufficient heat dissipation.

[0005] Secondly, PID control, as a classic feedback control method, has been introduced into the field of server cooling control. PID control can improve the dynamic response of the system to a certain extent by adjusting the proportional, integral and differential parameters. However, PID control is a linear control strategy and it is difficult to effectively handle the nonlinear characteristics and multivariable coupling problems in the server cooling system.

[0006] In recent years, with the development of artificial intelligence technology, some researchers have tried to apply machine learning methods to server cooling control. This method collects a large amount of historical data and trains models to predict thermal loads and optimize control strategies. Although this method performs well in some scenarios, its generalization ability is limited, making it difficult to cope with unseen working conditions, and requires a large amount of training data and computing resources.

[0007] In addition, some researchers have proposed a heat dissipation optimization method based on model predictive control (MPC). MPC builds a system model to predict the system behavior in the future and optimizes the control decision based on this. This method can achieve good control effects in theory, but in practical applications, it faces problems such as insufficient model accuracy and high computational complexity, making it difficult to be widely used in real-time control.

[0008] Although existing technologies have made some progress in the field of server heat dissipation control, the following key issues still exist:

[0009] 1. The description of the complex thermal field distribution inside the server is insufficient, making it difficult to accurately capture hot spots and heat flow paths.

[0010] 2. Lack of in-depth understanding of the dynamic characteristics of the cooling system makes it difficult to achieve precise heat transfer control.

[0011] 3. The existing heat dissipation efficiency evaluation method is too simple and cannot fully reflect the heat dissipation performance of the system.

[0012] 4. The control strategy is often a local optimization, which makes it difficult to achieve the global optimal heat dissipation effect.

[0013] 5. The system lacks adaptability and is unable to respond quickly to load changes and environmental disturbances.

[0014] These problems result in existing server cooling control systems often having problems such as excessive energy consumption, low cooling efficiency, uneven temperature distribution, and poor system stability, which seriously restricts the energy efficiency and reliability improvement of data centers. Summary of the invention

[0015] In view of the above problems, the present invention proposes an innovative server intelligent heat dissipation control optimization method and system. This method aims to achieve global optimization and intelligent control of the server heat dissipation system by introducing a series of innovative algorithms such as multi-dimensional thermal field topology mapping, group theory-based heat flow dynamic analysis, spectral decomposition heat dissipation efficiency evaluation, nonlinear heat transfer optimization and adaptive feedback control strategy.

[0016] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0017] A server intelligent heat dissipation control optimization method, comprising:

[0018] The acquisition steps include:

[0019] Obtain temperature information of each area of ​​the server and server workload data;

[0020] Processing steps include:

[0021] Based on the temperature information, construct a multi-dimensional thermal field topology map;

[0022] According to the multi-dimensional thermal field topological mapping, performing group theory-based thermal flow dynamic analysis;

[0023] Based on the group theory-based heat flow dynamic analysis results, spectral decomposition heat dissipation efficiency evaluation is performed;

[0024] performing nonlinear heat transfer optimization according to the spectral decomposition heat dissipation efficiency evaluation result and the workload data;

[0025] Output steps include:

[0026] Generate an adaptive feedback control strategy for regulating the server's cooling system.

[0027] Preferably, the multi-dimensional thermal field topology mapping specifically includes:

[0028] Map the temperature distribution of the server to a high-dimensional manifold and obtain the mapping function:

[0029] ,

[0030] Select the Morse function as the basis function and construct the mapping function:

[0031] ,

[0032] in, is a mapping function used to transform the spatial coordinates in the server Mapping to the target topological manifold It is a three-dimensional real space, representing the spatial coordinate system inside the server; is the target topological manifold, which represents the temperature distribution model after mapping; are spatial coordinates, representing the three dimensions inside the server; is the number of basis functions, indicating the total number of basis functions used to describe the thermal field distribution; is the weight coefficient, indicating The contribution of each basis function to the overall mapping; is the basis function used to describe the temperature distribution around the heat source.

[0033] Preferably, the Morse function is defined as:

[0034] ,

[0035] is the Morse function, which is used to describe the Temperature distribution around a heat source; are spatial coordinates, representing the three dimensions inside the server; is the heat source position coordinate, indicating the The specific location of the heat source inside the server.

[0036] Preferably, the group theory-based thermal flow dynamic analysis specifically includes:

[0037] Defining the permutation group , where each element Represents a possible heat flow path;

[0038] Define group action:

[0039] ,

[0040] Introducing group algebra Indicates the heat flux intensity:

[0041] ,

[0042] in, is the permutation group, which represents the set of all possible heat flow paths; is a group element, indicating heat flow paths; is the order of the group, indicating the number of heat flow paths; For group action, it means group Topological manifold The role of; is a group element, representing a specific heat flow path; is a point on the topological manifold, representing a certain position in the temperature field; is the group algebra, representing the linear combination of all heat flow paths; is the complex intensity, indicating the heat flow path The intensity and phase information of is the complex field, representing the set of complex numbers.

[0043] Preferably, the spectrum decomposition heat dissipation efficiency evaluation specifically includes:

[0044] Constructing group algebra The regular representation matrix

[0045] right Perform eigenvalue decomposition:

[0046] ,

[0047] Define the heat dissipation efficiency index:

[0048] ,

[0049] in, is a regular representation matrix, representing the group algebra The matrix form of is the eigenvector matrix, representing the matrix The matrix composed of the eigenvectors of ; is the eigenvalue matrix, which means the matrix A diagonal matrix composed of the eigenvalues ​​of ; is the eigenvalue, indicating The main heat transfer modes of the heat flow paths; is the number of eigenvalues, equal to the order of the group It is a heat dissipation efficiency index, which is used to quantitatively evaluate the efficiency of the current heat dissipation system; It is the sum of the absolute values ​​of all eigenvalues, reflecting the overall heat transfer capacity; is the largest eigenvalue, representing the dominant heat transfer mode.

[0050] Preferably, the nonlinear heat transfer optimization specifically includes:

[0051] Define the objective function:

[0052] ,

[0053] Solve the optimal control problem using the calculus of variations:

[0054] ,

[0055] We get the Euler-Lagrange equation:

[0056] ,

[0057] in, It is a comprehensive objective function used to evaluate the balance between heat dissipation efficiency and energy consumption; is the heat dissipation efficiency, which represents the heat dissipation effect under a given control variable u; It is a trade-off coefficient used to balance heat dissipation efficiency and energy consumption; is the trace of the matrix, representing the total heat of the system; is a regular representation matrix, which indicates that given the control variable The heat flow path matrix under ; u is the control variable, which represents the adjustable parameters of fan speed and cooling power; is the variational derivative, which represents the objective function The rate of change of the control variable u; is the Lagrangian function, which is used to describe the dynamic behavior of the system; is the partial derivative, which represents the Lagrangian function of the control variable The rate of change of is the partial derivative, which represents the rate of change of the time derivative of the Lagrangian function with respect to the control variable u.

[0058] Preferably, the adaptive feedback control strategy specifically includes:

[0059] Define the state-space model:

[0060] ,

[0061] Design an adaptive control law:

[0062] ,

[0063] in, is the state vector, which represents the state variables such as the temperature and heat flux intensity of each key point inside the server; u is the control input, which represents the adjustable parameters of fan speed and cooling power; y is the observation output, which represents the temperature value that can be directly measured; is the state dependency matrix, which represents the interaction between system states; is the control input matrix, which represents the influence of the control input on the system state; is the observation matrix, which represents the relationship between the system state and the observed output; is the time derivative of the state vector, which represents the change of the system state over time; is the state feedback gain matrix, which is used to quickly respond to changes in system states; is the adaptive gain matrix, which is used to eliminate steady-state errors and deal with the uncertainty of system parameters; is the actual output, indicating that the system The actual observed output at ; is the expected output, indicating that the system The expected output when .

[0064] Preferably, the adaptive feedback control strategy further comprises:

[0065] definition and The update law of:

[0066] ,

[0067] in, represents the state feedback gain matrix The time derivative of , that is, the rate of change of the gain matrix over time, is the state vector Transpose The outer product of forms a symmetric matrix, is the learning rate, used to update the state feedback gain matrix , is the learning rate, used to update the adaptive gain matrix , is the time derivative of the state feedback gain matrix, which represents the change of the gain matrix over time; is the time derivative of the adaptive gain matrix, which indicates the change of the gain matrix over time. is the solution of the Lyapunov equation, is the error vector, which represents the difference between the actual output and the expected output; is the transpose of the error vector.

[0068] Preferably, the Lyapunov equation is defined as:

[0069] ,

[0070] in, is the state dependency matrix The transpose of the matrix is ​​to swap the rows and columns. is a positive definite matrix.

[0071] A server intelligent heat dissipation control optimization system for executing the method includes:

[0072] Temperature collection module, used to obtain temperature information of each area of ​​the server;

[0073] A load monitoring module is used to obtain the workload data of the server;

[0074] A thermal field mapping module, used to construct a multi-dimensional thermal field topology map based on the temperature information;

[0075] A heat flow analysis module, used for performing group theory-based heat flow dynamic analysis according to the multi-dimensional thermal field topological mapping;

[0076] An efficiency evaluation module, used for evaluating the heat dissipation efficiency by spectrum decomposition based on the group theory-based heat flow dynamic analysis results;

[0077] an optimization calculation module, configured to perform nonlinear heat transfer optimization according to the spectral decomposition heat dissipation efficiency evaluation result and the workload data;

[0078] A control strategy generation module, used to generate an adaptive feedback control strategy;

[0079] The execution control module is used to adjust the heat dissipation system of the server according to the adaptive feedback control strategy.

[0080] Compared with the prior art, the beneficial effects of the present invention are embodied in the following aspects:

[0081] The core of the present invention is to construct a complete theoretical framework from thermal field analysis to control strategy generation, which effectively solves several key problems existing in the prior art. First, the multi-dimensional thermal field topology mapping technology provides an accurate description of the temperature distribution inside the server, laying the foundation for subsequent optimization. Secondly, the group theory-based thermal flow dynamic analysis method deeply reveals the essential laws of heat transfer and overcomes the problem that traditional methods are insufficient in describing complex heat flow processes. Furthermore, the spectral decomposition heat dissipation efficiency evaluation provides a new performance quantification indicator, which provides a reliable basis for the formulation of optimization goals. The nonlinear heat transfer optimization algorithm breaks through the limitations of traditional linear control and realizes precise control of nonlinear systems. Finally, the introduction of the adaptive feedback control strategy greatly improves the adaptability and robustness of the system.

[0082] These innovative algorithms are not simply superimposed, but form an organic whole, synergizing and complementing each other. For example, thermal field topology mapping provides the necessary input for thermal flow dynamic analysis, and the results of thermal flow analysis provide a theoretical basis for heat dissipation efficiency evaluation. Nonlinear optimization algorithms use the analysis results of the previous steps to generate optimal control strategies, while adaptive control ensures that these strategies can be effectively applied to actual systems.

[0083] This systematic approach has resulted in significant technical results:

[0084] 1. Greatly improved heat dissipation efficiency: Through global optimization and precise control, the system heat dissipation efficiency has been improved by more than 20%.

[0085] 2. Significantly reduced energy consumption: Thanks to more precise control strategies, energy consumption is reduced by 15-25% while ensuring heat dissipation.

[0086] 3. More uniform temperature distribution: Multi-dimensional thermal field analysis and global optimization make the temperature distribution inside the server more uniform, and the hot spot temperature is reduced by 5-10°C.

[0087] 4. Improved system stability: The adaptive control strategy enables the system to respond quickly to load changes, and the control stability is improved by more than 50%.

[0088] 5. Enhanced adaptability: This method can automatically adapt to different server configurations and working environments, greatly improving the versatility and scalability of the system.

[0089] In general, the present invention has achieved a qualitative leap in server heat dissipation control through the organic combination of a series of innovative algorithms. It not only solves the key problems existing in the prior art, but also provides a new technical path for future green data centers and efficient computing. The application of this method will significantly improve the energy efficiency of data centers, reduce operating costs, and improve the reliability and performance of servers, which is of great significance to promoting the sustainable development of information technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] Figure 1 The present invention is a flow chart of the method.

[0091] Figure 2 It is the overall logic block diagram of the system of the present invention. DETAILED DESCRIPTION

[0092] like Figure 1-2As shown, the present invention relates to a server intelligent heat dissipation control optimization method and system thereof. With the continuous expansion of the scale of data centers and the continuous improvement of server performance, the heat dissipation problem has become increasingly prominent. Traditional heat dissipation methods are often difficult to cope with complex and changeable thermal environments, resulting in energy waste or poor heat dissipation effects. In order to solve this technical problem, the present invention proposes an innovative server intelligent heat dissipation control optimization method and system thereof.

[0093] First, the method of the present invention includes an acquisition step, a processing step, and an output step. In the acquisition step, the system acquires the temperature information of each area of ​​the server and the workload data of the server. These data are the basis for subsequent processing. The temperature information is usually collected by temperature sensors distributed at various key locations of the server, and the workload data may include indicators such as CPU usage, memory usage, and disk I / O.

[0094] In the processing step, the method first constructs a multi-dimensional thermal field topology map based on the acquired temperature information. The purpose of this step is to convert the complex three-dimensional temperature distribution into a mathematical model that is easier to analyze. Specifically, the present invention adopts an innovative mapping function:

[0095] ,

[0096] in, is a mapping function used to transform the spatial coordinates in the server Mapping to the target topological manifold It is a three-dimensional real space, representing the spatial coordinate system inside the server; is the target topological manifold, which represents the temperature distribution model after mapping; are spatial coordinates, representing the three dimensions inside the server, for example, length, width, and height; is the number of basis functions, indicating the total number of basis functions used to describe the thermal field distribution; is the weight coefficient, indicating The contribution of each basis function to the overall mapping; is the basis function used to describe the temperature distribution around the heat source. This mapping method can effectively capture the temperature distribution characteristics inside the server and lay the foundation for subsequent analysis.

[0097] In the server intelligent heat dissipation control optimization method, building a multi-dimensional thermal field topology map is the basis of the entire system. The purpose of this step is to convert the complex three-dimensional temperature distribution into a mathematical model that is easier to analyze, so as to facilitate subsequent dynamic analysis and optimization.

[0098] In a server environment, the temperature distribution is usually uneven, especially under high load conditions, key components such as the CPU, GPU, and memory will generate a lot of heat. Traditional temperature monitoring methods can only provide temperature information at discrete points, which is difficult to fully reflect the thermal field distribution inside the entire server. By constructing a multi-dimensional thermal field topology map, these discrete temperature data can be integrated into a continuous temperature field model, thereby more accurately describing the heat transfer process inside the server.

[0099] For example, in a standard 19-inch rack server, temperature sensors are set at key locations such as the CPU position (0.1, 0.2, 0.05), memory bar position (0.15, 0.25, 0.03), and power supply position (0.4, 0.1, 0.1). The data collected by these sensors are input into the mapping function to generate a multi-dimensional thermal field model. This model can not only capture the temperature distribution of each heat source, but also reflect the mutual influence between different heat sources, helping the system better understand the path and mode of heat transfer.

[0100] Through multi-dimensional thermal field topology mapping, the system can obtain the global temperature distribution inside the server, not just the local hot spots. This helps to find potential cooling bottlenecks and avoid system failures caused by local overheating. As the server load changes, the location and intensity of the heat source will also change. The weight coefficient in the mapping function Adjustments can be made based on real-time data, allowing the system to dynamically adapt to different working conditions. Based on the multi-dimensional thermal field model, the system can predict future temperature change trends and take heat dissipation measures in advance to prevent overheating.

[0101] The Morse function is a special basis function with good locality and smoothness, and is suitable for describing the temperature distribution around a heat source.

[0102] Preferably, the present invention selects Morse function as the basis function. The selection of Morse function is based on its advantages in describing critical points and topological features. The specific definition of Morse function is as follows:

[0103] ,

[0104] in, is the Morse function, which is used to describe the Temperature distribution around a heat source; are spatial coordinates, representing the three dimensions inside the server, for example; is the heat source position coordinate, indicating the The specific location of each heat source inside the server. It is especially suitable for situations where there are multiple heat sources inside the server.

[0105] The core idea of ​​the Morse function is to describe the temperature distribution around the heat source in the form of exponential decay. The closer to the heat source, the higher the temperature, and the farther away from the heat source, the lower the temperature gradually.

[0106] In the server cooling system, the choice of Morse function is of great significance. The heat sources inside the server (such as CPU, GPU, power supply unit) are usually point-shaped or small-area heat sources. The Morse function can well capture the temperature gradient changes around these heat sources. For example, when the CPU load is high, the temperature around the CPU will rise rapidly, while the temperature far away from the CPU will be relatively low. The Morse function can be adjusted by adjusting the parameters ( ) to accurately describe this temperature change.

[0107] In addition, the Morse function also has good locality, that is, it only has a significant temperature contribution near the heat source, and decays quickly away from the heat source. This feature enables the system to effectively separate the influence of each heat source in a complex multi-heat source environment and avoid interference between different heat sources. The Morse function can accurately describe the temperature distribution around the heat source, especially in the case of multiple heat sources, it can effectively separate the contribution of each heat source and avoid aliasing of the temperature field. Due to the locality of the Morse function, the computational complexity is low and it is suitable for application in real-time control systems. Compared with other complex basis functions, the Morse function can reduce the consumption of computing resources while ensuring accuracy. The Morse function has good robustness to noise. Even if there are certain errors in the data collected by the temperature sensor, the Morse function can still stably describe the temperature distribution around the heat source.

[0108] In actual applications, the location of the heat source usually corresponds to the main heat-generating components in the server, such as the CPU, GPU, power supply unit, etc. For example, for a standard 19-inch rack-mounted server, we may set the heat source at the following locations: CPU location (0.1, 0.2, 0.05), memory location (0.15, 0.25, 0.03), power location (0.4, 0.1, 0.1), etc. These coordinate values ​​are usually in meters and are normalized based on the actual size of the server.

[0109] Group theory-based heat flow dynamic analysis is one of the core technologies of the present invention, which aims to systematically analyze the dynamic process of heat transfer through abstract modeling of heat flow paths.

[0110] Next, this method performs group-theoretic heat flow dynamic analysis based on multi-dimensional thermal field topology mapping. This step introduces the concept of group theory and abstracts the heat flow path into group elements, so that the dynamic process of heat transfer can be analyzed more systematically. Specifically, we define a permutation group:

[0111] ,

[0112] Each element Represents a possible heat flow path. This abstraction allows us to simplify the complex heat transfer process into the operation of group elements, greatly improving the efficiency and accuracy of the analysis. In order to describe the way the group acts on the topological manifold, we introduce the concept of group action:

[0113] ,

[0114] here, Representation Group Topological manifold In this way, we can study the effect of different heat flow paths on the overall temperature distribution. Furthermore, in order to quantify the intensity of heat flow, the present invention introduces group algebra Concept:

[0115] ,

[0116] in, is the permutation group, which represents the set of all possible heat flow paths; is a group element, indicating heat flow paths; is the order of the group, indicating the number of heat flow paths; For group action, it means group Topological manifold The role of; is a group element, representing a specific heat flow path; is a point on the topological manifold, representing a certain position in the temperature field; is the group algebra, representing the linear combination of all heat flow paths; is the complex intensity, indicating the heat flow path The intensity and phase information of The complex field represents the set of complex numbers. Using complex numbers to represent the intensity allows us to consider both the magnitude and phase information of the heat flow, which is particularly useful when analyzing periodic changes in heat.

[0117] This group-theory-based approach to heat flow dynamic analysis has several advantages. First, it provides a systematic way to describe and analyze complex heat transfer processes. Second, by abstracting the heat flow paths into group elements, we can leverage the rich theoretical knowledge of group theory to study the properties of heat flow, such as symmetries and invariants. Finally, the abstract nature of this approach makes it easily adaptable to server systems of different types and sizes.

[0118] In the server cooling system, the selection of heat flow path is crucial. Different heat flow paths will lead to different cooling effects, so it is necessary to systematically analyze and optimize the heat flow path. Group theory-based thermal flow dynamic analysis can systematically study the interaction between different paths by abstracting the heat flow path as a group element, so as to find the optimal cooling path.

[0119] In practical applications, the order m of the group usually depends on the complexity of the server and the required analysis accuracy. For a moderately complex server, m may be between 10 and 20. For example, we may consider the following typical heat flow paths:

[0120] 1. The path from the CPU directly to the heat sink;

[0121] 2. The path from the CPU through the memory to the heat sink;

[0122] 3. The path from the power supply unit to the chassis wall;

[0123] 4. The path from the hard disk through the air circulation inside the chassis to the fan; and so on.

[0124] Each path It can be represented by a vector, such as , where source is the heat source, intermediate is the intermediate transfer point (there may be multiple), and sink is the final heat dissipation point. Complex intensity The magnitude of can represent the intensity of heat flow, while its phase can represent the temporal characteristics of heat flow. For example, It may mean that the base strength of a heat flow path is 100 watts, with a 20-watt fluctuation component, and the phase information reflects the periodicity of this fluctuation.

[0125] Through group-theoretic thermal flow dynamic analysis, the system can evaluate the heat transfer efficiency of different paths and select the optimal path combination to achieve the best heat dissipation effect. Group-theoretic thermal flow dynamic analysis provides a systematic method to describe and analyze complex heat transfer processes, which can fully consider the interactions between different heat flow paths. By analyzing different heat flow paths, the system can select the optimal heat dissipation path to avoid excessive concentration of heat in certain areas, thereby improving the overall heat dissipation efficiency.

[0126] The group-theory-based heat flow dynamic analysis can adjust the heat flow path in real time according to the server's workload and environmental changes to ensure that the system is always in the best heat dissipation state.

[0127] In this way, the method of the present invention can comprehensively and accurately describe the heat transfer dynamics inside the server, providing a solid theoretical basis for subsequent heat dissipation optimization. This method based on advanced mathematics and group theory far exceeds the traditional simple temperature monitoring and control, and reflects the innovation and advancement of the present invention in the field of server heat dissipation. After completing the group theory-based heat flow dynamic analysis, the method of the present invention further performs spectral decomposition heat dissipation efficiency evaluation based on the analysis results. This step aims to quantitatively evaluate the efficiency of the current heat dissipation system and provide a basis for subsequent optimization.

[0128] Spectral decomposition heat dissipation efficiency evaluation is based on group algebra The regular representation matrix carried out.

[0129] Specifically, this method first constructs a group algebra The regular representation matrix The matrix contains comprehensive information about the heat flow dynamics, and its elements reflect the interactions between different heat flow paths. Preferably, It can be expressed as:

[0130] ,

[0131] in, Indicates the heat flow path On the path This matrix representation allows us to systematically analyze the complex interactions between heat flow paths.

[0132] Next, this method Perform eigenvalue decomposition:

[0133] ,

[0134] in, is the eigenvalue matrix, is the corresponding eigenvector matrix. The significance of this step is that the eigenvalues ​​and eigenvectors can reveal the essential characteristics of the system, such as the main heat transfer modes and their relative importance.

[0135] Based on the results of eigenvalue decomposition, the present invention innovatively defines a heat dissipation efficiency index:

[0136] ,

[0137] in, is a regular representation matrix, representing the group algebra The matrix form of is the eigenvector matrix, representing the matrix The matrix composed of the eigenvectors of ; is the eigenvalue matrix, which means the matrix A diagonal matrix composed of the eigenvalues ​​of ; is the eigenvalue, indicating The main heat transfer modes of the heat flow paths; is the number of eigenvalues, equal to the order of the group It is a heat dissipation efficiency index, which is used to quantitatively evaluate the efficiency of the current heat dissipation system; It is the sum of the absolute values ​​of all eigenvalues, reflecting the overall heat transfer capacity; is the maximum eigenvalue, representing the dominant heat transfer mode. This indicator cleverly combines the information of all eigenvalues. The numerator represents the sum of the absolute values ​​of all eigenvalues, reflecting the overall heat transfer capacity; the denominator uses the maximum eigenvalue, representing the dominant heat transfer mode. In this way, It not only takes into account the overall heat dissipation effect, but also reflects the balance of heat dissipation.

[0138] In practical applications, The value of is usually between 0 and 1. The value indicates better heat dissipation efficiency and more balanced heat distribution. For example, when When , the system can be considered to be in a state of efficient heat dissipation; when When the system cooling effect is average, it may need to be optimized; when , it indicates that the system cooling efficiency is low and immediate improvement measures are required.

[0139] This spectral decomposition-based approach to heat dissipation efficiency evaluation has several advantages. First, it provides a single, easy-to-understand metric to evaluate complex heat dissipation systems. Second, by considering all eigenvalues, it can fully reflect the heat dissipation characteristics of the system, rather than just focusing on the most significant heat transfer mode. Finally, the mathematical foundation of this method makes it well-founded and can provide a reliable basis for subsequent optimization decisions.

[0140] In the server cooling system, the evaluation of cooling efficiency is the key to optimizing the cooling strategy. Traditional cooling efficiency evaluation methods often rely on a single temperature indicator, which is difficult to fully reflect the cooling performance of the system. Through spectral decomposition cooling efficiency evaluation, the system can evaluate cooling efficiency from multiple dimensions (such as the contribution of different heat flow paths, the diversity of heat transfer modes, etc.), so as to more accurately identify problems and optimize them.

[0141] For example, suppose there are two main heat flow paths in the cooling system of a server: one is from the CPU to the radiator, and the other is from the power supply unit to the chassis wall. Through spectral decomposition heat dissipation efficiency evaluation, the system can find that the eigenvalue of the first path is larger, indicating that the heat transfer capacity of this path is stronger; while the eigenvalue of the second path is smaller, indicating that the heat transfer capacity of this path is weaker. Based on this result, the system can prioritize the optimization of the second path to improve the overall heat dissipation efficiency. Spectral decomposition heat dissipation efficiency evaluation can evaluate the heat dissipation efficiency from multiple dimensions (such as the contribution of different heat flow paths, the diversity of heat transfer modes, etc.), avoiding the one-sidedness of traditional methods. By analyzing the size of the eigenvalue, the system can identify the heat flow path with low heat dissipation efficiency and optimize it in a targeted manner, thereby improving the overall heat dissipation performance. Eigenvalue decomposition can also be used for system stability analysis to help the system identify potential unstable factors and ensure the long-term stable operation of the cooling system.

[0142] After obtaining the heat dissipation efficiency evaluation result, the method of the present invention further combines the workload data of the server to perform nonlinear heat transfer optimization. The goal of this step is to find the optimal heat dissipation control strategy to maximize the heat dissipation efficiency while ensuring the normal operation of the server.

[0143] To this end, this method first defines a comprehensive objective function:

[0144] ,

[0145] This objective function cleverly combines the heat dissipation efficiency and total system heat (via ), thereby achieving a balance between improving heat dissipation efficiency and reducing overall heat.

[0146] Preferably, The value range of can be between 0.1 and 0.5. The value will focus more on improving the heat dissipation efficiency, while the larger The specific choice can be determined based on the server's usage scenario and energy consumption requirements. For example, for a high-performance computing server, a smaller A value (such as 0.1-0.2) is selected to prioritize heat dissipation efficiency; for general-purpose servers, a larger value may be selected. value (such as 0.3-0.5) to better balance heat dissipation efficiency and energy consumption.

[0147] Next, this method uses the calculus of variations to solve the optimal control problem:

[0148] ,

[0149] This step aims to find the objective function Control variables that achieve extreme values . By solving this equation, we can obtain the Euler-Lagrange equation:

[0150] ,

[0151] in, It is a comprehensive objective function used to evaluate the balance between heat dissipation efficiency and energy consumption; is the heat dissipation efficiency, which represents the heat dissipation effect under a given control variable u; It is a trade-off coefficient used to balance heat dissipation efficiency and energy consumption; is the trace of the matrix, representing the total heat of the system; is a regular representation matrix, which indicates that given the control variable The heat flow path matrix under ; u is the control variable, which represents the adjustable parameters of fan speed and cooling power; is the variational derivative, which represents the objective function The rate of change of the control variable u; is the Lagrangian function, which is used to describe the dynamic behavior of the system; is the partial derivative, which represents the Lagrangian function of the control variable The rate of change of is a partial derivative, which represents the rate of change of the time derivative of the Lagrangian function with respect to the control variable u. This equation describes the conditions that the optimal control strategy should satisfy. In practical applications, solving the Euler-Lagrange equation may involve complex numerical methods. In a preferred embodiment of the present invention, an iterative method can be used to approximate the optimal solution.

[0152] In the server cooling system, nonlinear heat transfer optimization is a crucial link. The server's workload and environmental conditions are constantly changing, and traditional linear control methods are difficult to adapt to such dynamic changes. Through nonlinear heat transfer optimization, the system can adjust the cooling strategy in real time according to the current working status and cooling requirements to ensure that the server is always in the best cooling state.

[0153] For example, in a high-performance computing server, the CPU and GPU loads may increase dramatically in a short period of time, causing the temperature to rise rapidly. Through nonlinear heat transfer optimization, the system can quickly adjust the fan speed and cooling power to reduce the temperature in time and avoid overheating of the system. At the same time, the system will also consider energy consumption issues to avoid excessive consumption of power resources.

[0154] Specifically, an initial control strategy can be Start with, and then gradually adjust based on the gradient information:

[0155] ,

[0156] in, is the step size parameter, , usually between 0.01 and 0.1. This iterative process continues until the preset convergence condition is met, such as the change in the control variable between two consecutive iterations is less than a certain threshold (such as 1%). For the The control variable of the iteration is represented by The control parameters for the iteration. For the The control variable of the iteration is represented by The control parameter for the iteration. The value range is usually between 0.1 and 0.5. The value will focus more on improving the heat dissipation efficiency, while the larger The value is more focused on reducing the overall heat. The specific choice can be determined based on the server's usage scenario and energy consumption requirements. Step parameter The value of is usually between 0.01 and 0.1. Makes the system more responsive to changes but may cause instability; smaller This will make the system more stable, but the response speed may be slower.

[0157] The advantage of this nonlinear heat transfer optimization method is that it can fully consider the nonlinear characteristics of the system and multiple objectives (heat dissipation efficiency and energy consumption), thereby obtaining a more sophisticated and efficient heat dissipation control strategy. Compared with the traditional linear control method, this method can better adapt to the dynamic changes of server load and environment, and realize smarter and more energy-saving heat dissipation control. After completing the nonlinear heat transfer optimization, the method of the present invention further generates an adaptive feedback control strategy to realize real-time dynamic adjustment of the server cooling system. This step aims to convert the aforementioned optimization results into executable control instructions, and can be adaptively adjusted according to real-time feedback.

[0158] Specifically, this method first defines a state space model:

[0159] ,

[0160] The advantage of this state-space model is that it can fully describe the dynamic behavior of the server cooling system, including the interaction between the state variables and the impact of the control input on the system. At the same time, the introduction of the state-dependency matrix enables the model to adapt to the nonlinear characteristics of the system. Based on this state-space model, the present invention designs an innovative adaptive control law:

[0161]

[0162] This control law consists of two main parts: state feedback term and integral term. The state feedback term is used to quickly respond to changes in system state, while the integral term is used to eliminate steady-state errors and deal with the uncertainty of system parameters.

[0163] Preferably, the present invention also defines and The update law of:

[0164]

[0165] Among them, is the state vector, which represents the state variables such as the temperature and heat flux intensity of each key point inside the server; u is the control input, which represents the adjustable parameters of fan speed and cooling power; y is the observation output, which represents the temperature value that can be directly measured; is the state dependency matrix, which represents the interaction between system states; is the control input matrix, which represents the influence of the control input on the system state; is the observation matrix, which represents the relationship between the system state and the observed output; is the time derivative of the state vector, which represents the change of the system state over time; is the state feedback gain matrix, which is used to quickly respond to changes in system states; is the adaptive gain matrix, which is used to eliminate steady-state errors and deal with the uncertainty of system parameters; is the actual output, indicating that the system The actual observed output at ; is the expected output, indicating that the system These update laws enable the control system to adaptively adjust its parameters to adapt to changes in server workload and environment. In practical applications, the learning rate and The choice of has a significant impact on system performance. Typically, can be chosen between 0.01 and 0.1, and You can choose between 0.001 and 0.01. A larger learning rate will make the system respond faster to changes, but may cause instability; a smaller learning rate will make the system more stable, but may respond more slowly. The specific choice needs to be adjusted according to the characteristics of the server and the usage scenario.

[0166] In order to ensure the stability of the system, the present invention introduces the Lyapunov equation:

[0167]

[0168] in, is the state dependency matrix The transpose of the matrix is ​​to swap the rows and columns. is a positive definite matrix is a positive definite matrix. The solution to this equation is Used to construct Lyapunov function to prove the stability of the system. It is usually chosen to be the identity matrix or a diagonal matrix to simplify the calculation process.

[0169] The advantage of this adaptive feedback control strategy is that it can adjust the control parameters in real time to adapt to the dynamic changes of server load and environment. Compared with the traditional fixed parameter control method, this strategy can maintain good control effect under various working conditions, greatly improving the adaptability and robustness of the server cooling system.

[0170] In order to implement the above method, the present invention also proposes a corresponding server intelligent heat dissipation control optimization system. The system includes multiple functional modules, each module is responsible for a specific step in the method.

[0171] First, the temperature acquisition module 1 is used to obtain the temperature information of each area of ​​the server. The module may include multiple temperature sensors distributed in key positions of the server, and a microprocessor for collecting and preprocessing temperature data. Preferably, the sampling frequency of the temperature sensor can be set between 0.1Hz and 10Hz to balance data accuracy and system burden.

[0172] The load monitoring module 2 is responsible for obtaining the workload data of the server. This module may need to interact with the server's operating system and hardware components to collect indicators such as CPU usage, memory usage, and disk I / O. The data collection interval can be set between 1 second and 1 minute, depending on the purpose of the server and the speed of load changes.

[0173] The thermal field mapping module 3 constructs a multi-dimensional thermal field topology map based on the temperature information. This module needs to implement the mapping function and Morse function described above, and may require strong computing power to process a large amount of temperature data.

[0174] The heat flow analysis module 4 performs group theory-based heat flow dynamic analysis according to the multi-dimensional thermal field topology mapping. This module is one of the cores of the entire system and needs to implement complex group theory operations and heat flow dynamic models.

[0175] The efficiency evaluation module 5 performs spectral decomposition heat dissipation efficiency evaluation based on the group theory-based heat flow dynamic analysis results. This module needs to realize the eigenvalue decomposition of the matrix and the calculation of the heat dissipation efficiency index.

[0176] The optimization calculation module 6 performs nonlinear heat transfer optimization based on the spectral decomposition heat dissipation efficiency evaluation results and the workload data. This module may need to use a high-performance numerical calculation library to solve complex optimization problems.

[0177] The control strategy generation module 7 is used to generate an adaptive feedback control strategy. This module needs to implement a state space model and an adaptive control law, and may also need to perform real-time parameter estimation and updating.

[0178] Finally, the execution control module 8 adjusts the server's cooling system according to the adaptive feedback control strategy. This module directly interacts with the server's cooling hardware (such as fans, refrigeration devices) to execute specific control instructions.

[0179] These modules need to work closely together to form a closed-loop control system. For example, the output of the executive control module 8 will affect the reading of the temperature acquisition module 1, which will ultimately affect the generation of the control strategy through other modules. This closed-loop design ensures that the system can continuously optimize its performance and adapt to changing working conditions.

[0180] In general, the server intelligent heat dissipation control optimization method and system proposed in the present invention realizes accurate modeling, dynamic analysis and intelligent control of the server heat dissipation process by combining advanced concepts such as advanced mathematics, group theory, and nonlinear control theory. This method can not only improve heat dissipation efficiency and reduce energy consumption, but also adapt to various complex working scenarios, providing an innovative solution for the heat dissipation management of modern data centers.

[0181] In order to verify the effectiveness and superiority of the server intelligent heat dissipation control optimization method proposed in this invention, we designed a series of experiments and compared it with the traditional PID control method. These experiments are aimed at comprehensively evaluating the performance of the invention in terms of heat dissipation efficiency, energy consumption, temperature uniformity and system stability.

[0182] Embodiment 1: The server intelligent heat dissipation control optimization method of the present invention

[0183] In this embodiment, we use a typical two-way server as the test object. The server is equipped with two Intel Xeon Gold 6248R processors, 256GB DDR4 memory, 4 NVMe SSDs and a 1200W power supply. 20 temperature sensors are installed inside the server, distributed in key locations such as the CPU, memory, hard disk and power supply. The cooling system includes 6 variable speed fans and a liquid cooling radiator.

[0184] We implemented the full set of algorithms proposed in this paper, including multi-dimensional thermal field topology mapping, group theory-based heat flow dynamic analysis, spectral decomposition heat dissipation efficiency evaluation, nonlinear heat transfer optimization and adaptive feedback control strategy. The system runs on an embedded controller with a sampling frequency set to 1 Hz and a control update frequency of 0.1 Hz.

[0185] Comparative Example 1: Traditional PID Control Method

[0186] In the comparative experiment, we used the same model of server, but adopted the traditional PID (proportional-integral-derivative) control method to manage the cooling system. The PID controller takes the CPU temperature as input and controls the fan speed as output. The PID parameters are manually tuned to obtain good performance under general working conditions.

[0187] Experimental methods:

[0188] We designed a 24-hour load test scenario to simulate the typical working mode of a data center server. The test included the following phases:

[0189] 1. Idle state (2 hours);

[0190] 2. Moderate CPU load (4 hours);

[0191] 3. High CPU and memory load (6 hours);

[0192] 4. High I / O load (4 hours);

[0193] 5. Mixed load (6 hours);

[0194] 6. Cooling recovery (2 hours);

[0195] Throughout the testing process, we continuously monitored and recorded the following metrics:

[0196] 1. Heat dissipation efficiency index η (defined in this invention);

[0197] 2. Energy consumption (kWh);

[0198] 3. Maximum CPU temperature (°C);

[0199] 4. Temperature uniformity (standard deviation, °C);

[0200] 5. Control stability (standard deviation of fan speed variation, RPM).

[0201] The following table shows the experimental results of Example 1 and Comparative Example 1:

[0202] index Example 1 Comparative Example 1 Improvement percentage Heat dissipation efficiency index η 0.89 0.72 +23.6% Energy consumption (kWh) 18.7 23.5 -20.4% Max CPU temperature (°C) 76.3 82.1 -7.1% Temperature uniformity (°C) 2.8 5.6 -50.0% Control stability (RPM) 87 312 -72.1%

[0203] The analysis and discussion are as follows:

[0204] 1. Heat dissipation efficiency index η: The heat dissipation efficiency index of the method of the present invention is significantly higher than that of the traditional PID control method, which is increased by 23.6%. This shows that the multi-dimensional thermal field analysis and group-theory-based dynamic optimization of the present invention can more effectively utilize heat dissipation resources and achieve more efficient heat transfer.

[0205] 2. Energy consumption: The method of the present invention saves 20.4% of energy consumption compared with PID control in a 24-hour test. This significant energy saving effect is mainly due to the nonlinear heat transfer optimization algorithm of the present invention, which can minimize energy input while ensuring heat dissipation effect.

[0206] 3. Maximum CPU temperature: The method of the present invention controls the maximum CPU temperature at 76.3°C, which is 5.8°C lower than PID control. This not only improves the reliability of the system, but also provides greater room for CPU performance improvement.

[0207] 4. Temperature uniformity: The method of the present invention makes the temperature distribution inside the server more uniform, with a temperature standard deviation of only 2.8°C, which is 50% better than PID control. This is due to the multi-dimensional thermal field topology mapping and global optimization strategy of the present invention, which effectively reduces the generation of hot spots.

[0208] 5. Control stability: The method of the present invention exhibits excellent control stability, with the standard deviation of fan speed variation being only 87RPM, which is 72.1% lower than PID control. This smooth control not only reduces mechanical stress and prolongs fan life, but also reduces noise levels.

[0209] These experimental results fully demonstrate the superiority of the method of the present invention. It is particularly noteworthy that the present invention has achieved significant improvements in all test indicators, which shows that the method can achieve a good balance between multiple goals. For example, it not only reduces energy consumption, but also improves the temperature control effect at the same time, which is often a contradiction in traditional methods.

[0210] The advantages of the present invention mainly come from the following innovative points:

[0211] 1. Multi-dimensional thermal field topology mapping provides a more accurate temperature distribution model.

[0212] 2. Group theory based heat flow dynamic analysis captures the complex heat transfer patterns.

[0213] 3. Spectral decomposition heat dissipation efficiency evaluation provides a reliable quantitative indicator for optimization.

[0214] 4. Nonlinear heat transfer optimization achieves global optimal control.

[0215] 5. Adaptive feedback control strategy enables the system to respond quickly to load changes.

[0216] These innovations enable the present invention to more comprehensively and deeply understand and control the heat dissipation process of the server, thereby maintaining excellent performance under various working conditions.

[0217] Based on the above experimental results, we believe that Example 1 is the best embodiment of the present invention. In this embodiment, we fully implement all the key algorithms of the present invention and achieve the best balance between heat dissipation efficiency, energy consumption and temperature control through careful parameter tuning. In particular, we found that when the update frequency of the thermal field mapping is set to 0.1Hz and the learning rates ρ1 and ρ2 of the adaptive control are set to 0.05 and 0.005 respectively, the system exhibits the best overall performance.

[0218] These experimental results not only verify the effectiveness of the present invention, but also point out the direction for further optimization in the future. For example, we observed that the response speed of the system can be improved when the load changes sharply. This may require further improvement of the thermal flow dynamic model or the introduction of predictive control strategies. In addition, combining this method with machine learning technology may lead to a smarter and more efficient heat dissipation control system.

[0219] In general, the present invention brings a new methodology to the field of server heat dissipation control, and its superiority has been verified through rigorous experiments. This method can not only significantly improve heat dissipation efficiency and reduce energy consumption, but also provide a more stable and uniform temperature environment for the server, thereby improving the reliability and performance of the system. With the continuous expansion of the scale of data centers and the continuous improvement of server performance, the method of the present invention will play an important role in the future construction of green computing and efficient data centers.

[0220] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto; any person familiar with the art who, within the scope disclosed by the present invention, makes substitutions or changes based on the scheme and improved concepts of the present invention shall be covered within the protection scope of the present invention.

Claims

1. A server intelligent heat dissipation control optimization method, characterized in that: include: The acquisition steps include: Obtain temperature information of each area of ​​the server and server workload data; Processing steps include: Based on the temperature information, construct a multi-dimensional thermal field topology map; According to the multi-dimensional thermal field topological mapping, performing group theory-based thermal flow dynamic analysis; Based on the group theory-based heat flow dynamic analysis results, spectral decomposition heat dissipation efficiency evaluation is performed; performing nonlinear heat transfer optimization according to the spectral decomposition heat dissipation efficiency evaluation result and the workload data; Output steps include: Generate an adaptive feedback control strategy for regulating the server's cooling system.

2. The server intelligent heat dissipation control optimization method according to claim 1 is characterized in that: The multi-dimensional thermal field topology mapping specifically includes: Map the temperature distribution of the server to a high-dimensional manifold and obtain the mapping function: , Select the Morse function as the basis function and construct the mapping function: , in, is a mapping function used to transform the spatial coordinates in the server Mapping to the target topological manifold It is a three-dimensional real space, representing the spatial coordinate system inside the server; is the target topological manifold, which represents the temperature distribution model after mapping; are spatial coordinates, representing the three dimensions inside the server; is the number of basis functions, indicating the total number of basis functions used to describe the thermal field distribution; is the weight coefficient, indicating The contribution of each basis function to the overall mapping; is the basis function used to describe the temperature distribution around the heat source.

3. The server intelligent heat dissipation control optimization method according to claim 2 is characterized in that: The Morse function is defined as: , in, is the Morse function, which is used to describe the Temperature distribution around a heat source; are spatial coordinates, representing the three dimensions inside the server; is the heat source position coordinate, indicating the The specific location of the heat source inside the server.

4. The server intelligent heat dissipation control optimization method according to claim 1 is characterized in that: The group theory-based heat flow dynamic analysis specifically includes: Defining the permutation group , where each element Represents a possible heat flow path; Define group action: , Introducing group algebra Indicates the heat flux intensity: , in, is the permutation group, which represents the set of all possible heat flow paths; is a group element, indicating heat flow paths; is the order of the group, indicating the number of heat flow paths; For group action, it means group Topological manifold The role of; is a group element, representing a specific heat flow path; is a point on the topological manifold, representing a certain position in the temperature field; is the group algebra, representing the linear combination of all heat flow paths; is the complex intensity, indicating the heat flow path The intensity and phase information of is the complex field, representing the set of complex numbers.

5. The server intelligent heat dissipation control optimization method according to claim 1 is characterized in that: The spectrum decomposition heat dissipation efficiency evaluation specifically includes: Constructing group algebra The regular representation matrix right Perform eigenvalue decomposition: , Define the heat dissipation efficiency index: , in, is a regular representation matrix, representing the group algebra The matrix form of is the eigenvector matrix, representing the matrix The matrix composed of the eigenvectors of ; is the eigenvalue matrix, which means the matrix A diagonal matrix composed of the eigenvalues ​​of ; is the eigenvalue, indicating the The main heat transfer modes of the heat flow paths; is the number of eigenvalues, equal to the order of the group It is a heat dissipation efficiency index, which is used to quantitatively evaluate the efficiency of the current heat dissipation system; It is the sum of the absolute values ​​of all eigenvalues, reflecting the overall heat transfer capacity; is the largest eigenvalue, representing the dominant heat transfer mode.

6. The server intelligent heat dissipation control optimization method according to claim 1 is characterized in that: The nonlinear heat transfer optimization specifically includes: Define the objective function: , Solve the optimal control problem using the calculus of variations: , We get the Euler-Lagrange equation: , in, It is a comprehensive objective function used to evaluate the balance between heat dissipation efficiency and energy consumption; is the heat dissipation efficiency, which represents the heat dissipation effect under a given control variable u; It is a trade-off coefficient used to balance heat dissipation efficiency and energy consumption; is the trace of the matrix, representing the total heat of the system; is a regular representation matrix, which indicates that given the control variable The heat flow path matrix under ; u is the control variable, which represents the adjustable parameters of fan speed and cooling power; is the variational derivative, which represents the objective function The rate of change of the control variable u; is the Lagrangian function, which is used to describe the dynamic behavior of the system; is the partial derivative, which represents the Lagrangian function of the control variable The rate of change of is the partial derivative, which represents the rate of change of the time derivative of the Lagrangian function with respect to the control variable u.

7. The server intelligent heat dissipation control optimization method according to claim 1 is characterized in that: The adaptive feedback control strategy specifically includes: Define the state-space model: , Design an adaptive control law: , in, is the state vector, which represents the temperature and heat flux intensity state variables of each key point inside the server; u is the control input, which represents the adjustable parameters of fan speed and cooling power; y is the observation output, which represents the temperature value that can be directly measured; is the state dependency matrix, which represents the interaction between system states; is the control input matrix, which represents the influence of the control input on the system state; is the observation matrix, which represents the relationship between the system state and the observed output; is the time derivative of the state vector, which represents the change of the system state over time; is the state feedback gain matrix, which is used to quickly respond to changes in system states; is an adaptive gain matrix used to eliminate steady-state errors and deal with the uncertainty of system parameters; is the actual output, indicating that the system The actual observed output at ; is the expected output, indicating that the system The expected output when .

8. The server intelligent heat dissipation control optimization method according to claim 7, characterized in that: The adaptive feedback control strategy also includes: definition and The update law of: , in, represents the state feedback gain matrix The time derivative of , that is, the rate of change of the gain matrix over time, is the state vector Transpose The outer product of forms a symmetric matrix, is the learning rate, used to update the state feedback gain matrix , is the learning rate, used to update the adaptive gain matrix , is the time derivative of the state feedback gain matrix, which represents the change of the gain matrix over time; is the time derivative of the adaptive gain matrix, which indicates the change of the gain matrix over time. is the solution of the Lyapunov equation, is the error vector, which represents the difference between the actual output and the expected output; is the transpose of the error vector.

9. The server intelligent heat dissipation control optimization method according to claim 8, characterized in that: The Lyapunov equation is defined as: ， in, is the state dependency matrix The transpose of the matrix is ​​to swap the rows and columns. is a positive definite matrix.

10. A server intelligent heat dissipation control optimization system for executing the method according to any one of claims 1 to 9, characterized in that: include: Temperature collection module, used to obtain temperature information of each area of ​​the server; A load monitoring module is used to obtain the workload data of the server; A thermal field mapping module, used to construct a multi-dimensional thermal field topology map based on the temperature information; A heat flow analysis module, used for performing group theory-based heat flow dynamic analysis according to the multi-dimensional thermal field topological mapping; An efficiency evaluation module, used for evaluating the heat dissipation efficiency by spectrum decomposition based on the group theory-based heat flow dynamic analysis results; an optimization calculation module, configured to perform nonlinear heat transfer optimization according to the spectral decomposition heat dissipation efficiency evaluation result and the workload data; A control strategy generation module, used to generate an adaptive feedback control strategy; The execution control module is used to adjust the heat dissipation system of the server according to the adaptive feedback control strategy.

Images