A method, apparatus, equipment, and medium for temperature control prediction of a liquid-cooled server.

Abstract

This invention relates to a method, apparatus, device, and medium for predicting temperature control in a liquid-cooled server, belonging to the field of server temperature control. The method includes: acquiring a historical monitoring dataset of the liquid-cooled server; constructing a temperature prediction model for the liquid-cooled server based on the historical monitoring dataset using an improved XGBOOST algorithm; acquiring a preset parameter set for the liquid-cooled server; inputting the preset parameter set into the temperature prediction model to obtain the predicted temperature and predicted temperature uniformity; constructing a temperature control model for the liquid-cooled server based on the predicted temperature and predicted temperature uniformity and outputting temperature control information; and providing feedback on parameter adjustment commands for the liquid-cooled server based on the temperature control information. This invention achieves scientific prediction of the possible operating temperature and temperature uniformity of the liquid-cooled server and takes measures to control both within a reasonable range.

Description

Technical Field

[0001] This invention belongs to the field of server temperature control technology, specifically relating to a method, device, equipment, and medium for predicting temperature control in a liquid-cooled server. Background Technology

[0002] As data centers expand and server density increases, traditional air cooling systems are struggling to meet the demands of high-density, high-performance computing. Liquid cooling technology, due to its efficient heat transfer capabilities, is widely used in high-performance servers. However, liquid cooling systems require precise control of liquid flow rate and temperature to prevent overheating. Most existing temperature control systems are based on static thresholds and empirical rules, lacking the ability to predict future conditions. Traditional temperature control systems are often reactive, adjusting only after detecting excessive temperatures, which may fail to prevent temperature anomalies in a timely manner. Secondly, they are inefficient: the lack of predictive capabilities means the system cannot proactively adjust its cooling strategy according to changes in server load, potentially leading to energy waste. Furthermore, traditional temperature control systems often ignore temperature uniformity among servers, making it prone to overheating of certain components and system failure.

[0003] XGBoost (Gradient Boosting Algorithm) excels at handling large-scale datasets and solving complex prediction problems. It's an algorithm that can handle various machine learning tasks such as classification, regression, ranking, and custom objective functions. Before training, it sorts the data based on features and uses a sparse matrix storage format in each block structure. Subsequent training processes reuse these block structures, significantly reducing computational cost. XGBoost supports not only gradient-boosted decision trees but also linear classifiers, providing a valuable option for a wide range of scenarios. The XGBoost objective function consists of two parts: a loss function and a regularization term. The regularization term controls model complexity, simplifying the model and preventing overfitting. It employs a forward distribution algorithm for greedy training. In each iteration, a CART tree is learned, and the prediction residuals of the first t-1 trees are fitted with the actual values ​​of the training samples. Summary of the Invention

[0004] To address the aforementioned problems in the prior art, this invention provides a method, apparatus, device, and medium for temperature control prediction of a liquid-cooled server.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A temperature control prediction method for a liquid-cooled server, the implementation of which includes the following steps:

[0007] Obtain historical monitoring datasets for liquid-cooled servers, which include historical temperature, historical computing power, historical power consumption, historical primary oil pump speed, historical secondary water pump speed, and historical ambient temperature of the liquid-cooled servers.

[0008] A temperature prediction model for the liquid-cooled server was constructed based on the historical monitoring dataset of the liquid-cooled server by improving the XGBOOST algorithm.

[0009] Obtain a preset parameter set for the liquid-cooled server, which includes the liquid-cooled server's computing power, operating power, primary oil pump speed, secondary water pump speed, and ambient temperature.

[0010] The preset parameter set of the liquid-cooled server is input into the temperature prediction model of the liquid-cooled server to obtain the predicted temperature and the predicted temperature uniformity.

[0011] A liquid-cooled server temperature control model is constructed based on the predicted temperature and the predicted temperature uniformity, and temperature control information is output. The temperature control information includes temperature control anomaly information and temperature control normal information. Based on the temperature control information, liquid-cooled server parameter adjustment instructions are fed back. The liquid-cooled server parameter adjustment instructions include primary oil pump speed adjustment instructions, secondary water pump speed adjustment instructions, and computing power adjustment instructions.

[0012] Preferably, the step of constructing a liquid-cooled server temperature prediction model based on the historical monitoring dataset of the liquid-cooled server using the improved XGBOOST algorithm includes:

[0013] The historical temperature uniformity of the liquid-cooled server is obtained based on the historical temperature of the liquid-cooled server, which includes the historical temperature of the CPU, the historical temperature of the GPU, and the historical temperature of the memory.

[0014] Given the historical monitoring dataset of the liquid-cooled server and the historical temperature uniformity of the liquid-cooled server, the improved XGBOOST algorithm is used to learn k decision trees additively to construct the temperature prediction model of the liquid-cooled server.

[0015] Preferably, obtaining the historical temperature uniformity of the liquid-cooled server based on the historical temperature of the liquid-cooled server includes:

[0016] The number of heat-generating components in the liquid-cooled server is obtained, including the number of CPUs, GPUs, and memory.

[0017] The historical temperature uniformity of the liquid-cooled server is calculated based on the historical temperature of the liquid-cooled server and the number of heat-generating components.

[0018] Preferably, the step of constructing the liquid-cooled server temperature prediction model by additively learning k decision trees using the improved XGBOOST algorithm includes:

[0019] Obtain the loss function of the liquid-cooled server temperature prediction model, the loss function including a temperature loss function and a temperature uniformity loss function;

[0020] Obtain the model regularization term of the liquid-cooled server temperature prediction model;

[0021] The model objective function is obtained based on the loss function and the model regularization term;

[0022] The objective function of the model is expanded using a second-order Taylor series.

[0023] The objective function of the model is simplified by simplifying the constant term, and the liquid-cooled server temperature prediction model is obtained.

[0024] Preferably, the expression for the temperature loss function is as follows: ,in, l Let y be the temperature loss function. j Historical temperatures of liquid-cooled servers Let be the predicted historical temperature output by the j-th decision tree in the t-th iteration of the model. This represents the predicted historical temperature output by the j-th decision tree during the (t-1)th iteration of the model. Let k be the predicted historical temperature of the j-th decision tree in the model at the t-th iteration, and k be the number of decision trees in the model; the expression for the temperature uniformity loss function is: Where w is the temperature uniformity loss function, r j To ensure the historical temperature uniformity of liquid-cooled servers, Let the predicted historical temperature uniformity be the output of the j-th decision tree in the t-th iteration of the model. To determine the predicted historical temperature uniformity output by the j-th decision tree in the (t-1)th iteration of the model, Let represent the predicted historical temperature uniformity of the j-th decision tree in the t-th iteration of the model.

[0025] Preferably, the step of constructing a liquid-cooled server temperature control model based on the predicted temperature and the predicted temperature uniformity, and outputting temperature control information, includes:

[0026] A preset predicted temperature threshold and a predicted temperature uniformity threshold are defined. When the predicted temperature is greater than or equal to the predicted temperature threshold or the predicted temperature uniformity is less than or equal to the predicted temperature uniformity threshold, the temperature control abnormality information is output. When the predicted temperature is less than the predicted temperature threshold and the predicted temperature uniformity is greater than the predicted temperature uniformity threshold, the temperature control normality information is output.

[0027] Preferably, the parameter adjustment instructions for the liquid-cooled server based on the temperature control information feedback include:

[0028] When the temperature control abnormality information is output, the speed of the liquid cooling server's oil pump is determined. When the speed of the liquid cooling server's oil pump is less than the maximum speed of the oil pump, the oil pump speed adjustment command is fed back to adjust the speed of the liquid cooling server's oil pump until the temperature control normality information is output.

[0029] When the liquid-cooled server runs the oil pump at a speed equal to the maximum speed, the liquid-cooled server runs the secondary water pump at a speed that is determined. When the liquid-cooled server runs the secondary water pump at a speed that is less than the maximum speed of the secondary water pump, the liquid-cooled server sends back a secondary water pump speed adjustment command to adjust the secondary water pump speed until the temperature control normal information is output.

[0030] When the secondary water pump speed of the liquid-cooled server is equal to the maximum speed, the operating computing power adjustment command is fed back to reduce the operating computing power of the liquid-cooled server until the temperature control normal information is output.

[0031] A temperature control prediction device for a liquid-cooled server, used to execute the temperature control prediction method described above, includes a data acquisition module, a model building module, and a temperature control module;

[0032] The data acquisition module is used to acquire the historical monitoring dataset of the liquid-cooled server. The historical monitoring dataset of the liquid-cooled server includes the historical temperature of the liquid-cooled server, the historical computing power of the liquid-cooled server, the historical power of the liquid-cooled server, the historical speed of the primary oil pump of the liquid-cooled server, the historical speed of the secondary water pump of the liquid-cooled server, and the historical ambient temperature of the liquid-cooled server.

[0033] The model building module is used to construct a liquid-cooled server temperature prediction model based on the historical monitoring dataset of the liquid-cooled server using an improved XGBOOST algorithm; obtain a preset parameter set for the liquid-cooled server, which includes the liquid-cooled server's computing power, operating power, primary oil pump speed, secondary water pump speed, and ambient temperature; and input the preset parameter set into the liquid-cooled server temperature prediction model to obtain the predicted temperature and predicted temperature uniformity.

[0034] The temperature control module is used to construct a liquid-cooled server temperature control model based on the predicted temperature and the predicted temperature uniformity, and output temperature control information. The temperature control information includes temperature control abnormal information and temperature control normal information. Based on the temperature control information, the module feeds back liquid-cooled server parameter adjustment instructions, which include primary oil pump speed adjustment instructions, secondary water pump speed adjustment instructions, and computing power adjustment instructions.

[0035] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the temperature control prediction method described above.

[0036] A storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the temperature control prediction method described above.

[0037] The beneficial effects of this invention are as follows:

[0038] (1) By improving the XGBOOST algorithm, a liquid cooling server temperature prediction model is constructed based on the historical monitoring dataset of the liquid cooling server. This model can effectively predict the server temperature and temperature uniformity, and make reasonable temperature regulation and control measures. This not only ensures the safety and stability of the server operation, but also optimizes energy efficiency and reduces operating costs.

[0039] (2) The liquid-cooled server temperature control model is constructed based on the predicted temperature and the predicted temperature uniformity and the temperature control information is output, instead of considering temperature alone. The constructed model is more practical and the subsequent adjustment can be more accurate.

[0040] (3) By simplifying the constant term, the objective function of the model is simplified and the liquid-cooled server temperature prediction model is obtained, which simplifies the calculation process and improves the system operating efficiency. Attached Figure Description

[0041] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0042] Figure 1 This is a flowchart of the temperature control prediction method for a liquid-cooled server according to the present invention. Detailed Implementation

[0043] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0044] Working principle and usage process of this invention:

[0045] Please see Figure 1 A method for predicting temperature control in a liquid-cooled server, comprising:

[0046] S1: Obtain the historical monitoring dataset of the liquid-cooled server, which includes the historical temperature of the liquid-cooled server, the historical computing power of the liquid-cooled server, the historical power of the liquid-cooled server, the historical speed of the primary oil pump of the liquid-cooled server, the historical speed of the secondary water pump of the liquid-cooled server, and the historical ambient temperature of the liquid-cooled server.

[0047] S2: Construct a liquid cooling server temperature prediction model based on the historical monitoring dataset of the liquid cooling server by improving the XGBOOST algorithm;

[0048] S3: Obtain the preset parameter set of the liquid-cooled server, which includes the liquid-cooled server's operating computing power, liquid-cooled server's operating power, liquid-cooled server's primary oil pump speed, liquid-cooled server's secondary water pump speed, and liquid-cooled server's operating ambient temperature.

[0049] S4: Input the preset parameter set of the liquid-cooled server into the liquid-cooled server temperature prediction model to obtain the predicted temperature and predicted temperature uniformity, instead of considering temperature alone. The constructed model is more practical and the subsequent adjustment can be more accurate.

[0050] S5: Construct a liquid-cooled server temperature control model based on the predicted temperature and the predicted temperature uniformity, and output temperature control information. The temperature control information includes temperature control abnormality information and temperature control normal information. Feedback liquid-cooled server parameter adjustment instructions based on the temperature control information includes primary oil pump speed adjustment instructions, secondary water pump speed adjustment instructions, and computing power adjustment instructions.

[0051] In this embodiment, a liquid-cooled server temperature prediction model is constructed based on the historical monitoring dataset of the liquid-cooled server using an improved XGBOOST algorithm. This can be implemented through the following steps:

[0052] S201: Obtain the historical temperature uniformity of the liquid-cooled server based on the historical temperature of the liquid-cooled server, wherein the historical temperature of the liquid-cooled server includes the historical temperature of the CPU, the historical temperature of the GPU, and the historical temperature of the memory.

[0053] S202: Given the historical monitoring dataset of the liquid-cooled server and the historical temperature uniformity of the liquid-cooled server, the improved XGBOOST algorithm is used to learn k decision trees additively to construct the temperature prediction model of the liquid-cooled server. This model effectively predicts the server temperature and temperature uniformity and makes reasonable temperature regulation and control measures, which not only ensures the safety and stability of server operation, but also optimizes energy efficiency and reduces operating costs.

[0054] In this embodiment, the historical temperature uniformity of the liquid-cooled server is obtained based on the historical temperature of the liquid-cooled server, which can be implemented through the following steps:

[0055] S201-1: Obtain the number of heat-generating components in the liquid-cooled server, including the number of CPUs, GPUs, and memory.

[0056] S201-2: Calculate the historical temperature uniformity of the liquid-cooled server based on the historical temperature of the liquid-cooled server and the number of heat-generating components. The calculation formula is as follows: Where S represents the historical temperature uniformity of the liquid-cooled server, N1 represents the number of CPUs, N2 represents the number of GPUs, N3 represents the amount of memory, and T represents the number of CPUs. i1 Let T be the historical CPU temperature of the i-th CPU. ave1 T represents the historical average temperature of N1 CPUs. i2 Let T be the historical GPU temperature of the i-th GPU. ave2 T represents the historical average temperature of N2 GPUs. i3 Let T be the historical temperature of the i-th memory location. ave3 This represents the historical average temperature of N3 memory modules.

[0057] In this embodiment, the improved XGBOOST algorithm is used to learn k decision trees additively to construct the liquid-cooled server temperature prediction model. This can be implemented through the following steps:

[0058] S202-1: Obtain the loss function of the liquid-cooled server temperature prediction model. The loss function includes a temperature loss function and a temperature uniformity loss function. The expression for the temperature loss function is as follows: ,in, l Let y be the temperature loss function. j Historical temperatures of liquid-cooled servers Let be the predicted historical temperature output by the j-th decision tree in the t-th iteration of the model. This represents the predicted historical temperature output by the j-th decision tree during the (t-1)th iteration of the model. Let k be the predicted historical temperature of the j-th decision tree in the model at the t-th iteration, and k be the number of decision trees in the model; the expression for the temperature uniformity loss function is: Where w is the temperature uniformity loss function, r j To ensure the historical temperature uniformity of liquid-cooled servers, Let the predicted historical temperature uniformity be the output of the j-th decision tree in the t-th iteration of the model. To determine the predicted historical temperature uniformity output by the j-th decision tree in the (t-1)th iteration of the model, Let k be the uniformity of the predicted historical temperature of the j-th decision tree in the model at the t-th iteration, and k be the number of decision trees in the model.

[0059] S202-2: Obtain the model regularization term of the liquid-cooled server temperature prediction model, the expression of which is: ,in, Let α be the model regularization term, λ be the parameter of the first quantization regularization term, λ be the parameter of the second quantization regularization term, S be the number of leaf nodes in the model decision tree, and T be the number of leaves in the decision tree. Output the result for the t-th leaf of the decision tree;

[0060] S202-3: Based on the loss function and the model regularization term, the model objective function is obtained, and the expression of the model objective function is: Where R is the model objective function;

[0061] S202-4: By expanding the objective function of the model using a second-order Taylor series, we can obtain... , where h i p is the first derivative of the temperature loss function. i Let q be the second derivative of the temperature loss function. i z is the first derivative of the temperature uniformity loss function. i This is the second derivative of the temperature uniformity loss function;

[0062] S202-5: By simplifying the objective function of the model through the simplification of constant terms, the liquid-cooled server temperature prediction model can be obtained. This simplifies the calculation process and improves system operating efficiency.

[0063] In this embodiment, a liquid-cooled server temperature control model is constructed based on the predicted temperature and the predicted temperature uniformity, and temperature control information is output. This can be implemented through the following steps:

[0064] S501: Preset a predicted temperature threshold and a predicted temperature uniformity threshold. When the predicted temperature is greater than or equal to the predicted temperature threshold or the predicted temperature uniformity is less than or equal to the predicted temperature uniformity threshold, output the temperature control abnormal information. When the predicted temperature is less than the predicted temperature threshold and the predicted temperature uniformity is greater than the predicted temperature uniformity threshold, output the temperature control normal information.

[0065] In this embodiment, the parameter adjustment command for the liquid-cooled server is fed back based on the temperature control information, which can be implemented through the following steps:

[0066] S502-1: When the temperature control abnormal information is output, determine the oil pump speed of the liquid cooling server. When the oil pump speed of the liquid cooling server is less than the maximum speed of the oil pump, feed back the oil pump speed adjustment command and adjust the oil pump speed of the liquid cooling server until the temperature control normal information is output.

[0067] S502-2: When the speed of the primary oil pump of the liquid-cooled server is equal to the maximum speed, the speed of the secondary water pump of the liquid-cooled server is determined. When the speed of the secondary water pump of the liquid-cooled server is less than the maximum speed of the secondary water pump, the speed adjustment command of the secondary water pump is fed back to adjust the speed of the secondary water pump of the liquid-cooled server until the temperature control normal information is output.

[0068] S502-3: When the secondary water pump speed of the liquid-cooled server is equal to the maximum speed, the operating computing power adjustment command is fed back to reduce the operating computing power of the liquid-cooled server until the temperature control normal information is output.

[0069] A temperature control and prediction device for a liquid-cooled server includes a data acquisition module, a model building module, and a temperature control module;

[0070] The data acquisition module is used to acquire the historical monitoring dataset of the liquid-cooled server. The historical monitoring dataset of the liquid-cooled server includes the historical temperature of the liquid-cooled server, the historical computing power of the liquid-cooled server, the historical power of the liquid-cooled server, the historical speed of the primary oil pump of the liquid-cooled server, the historical speed of the secondary water pump of the liquid-cooled server, and the historical ambient temperature of the liquid-cooled server.

[0071] The model building module is used to construct a liquid-cooled server temperature prediction model based on the historical monitoring dataset of the liquid-cooled server using an improved XGBOOST algorithm; obtain a preset parameter set for the liquid-cooled server, which includes the liquid-cooled server's computing power, operating power, primary oil pump speed, secondary water pump speed, and ambient temperature; and input the preset parameter set into the liquid-cooled server temperature prediction model to obtain the predicted temperature and predicted temperature uniformity.

[0072] The temperature control module is used to construct a liquid-cooled server temperature control model based on the predicted temperature and the predicted temperature uniformity, and output temperature control information. The temperature control information includes temperature control abnormal information and temperature control normal information. Based on the temperature control information, the module feeds back liquid-cooled server parameter adjustment instructions, which include primary oil pump speed adjustment instructions, secondary water pump speed adjustment instructions, and computing power adjustment instructions.

[0073] The computer storage medium of this invention can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. Computer-readable storage media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0074] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0075] The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof. The computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0076] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for predicting temperature control in a liquid-cooled server, characterized in that, The implementation of the temperature control prediction method includes the following steps: Obtain historical monitoring datasets for liquid-cooled servers, which include historical temperature, historical computing power, historical power consumption, historical primary oil pump speed, historical secondary water pump speed, and historical ambient temperature of the liquid-cooled servers. A temperature prediction model for the liquid-cooled server was constructed based on the historical monitoring dataset of the liquid-cooled server by improving the XGBOOST algorithm. Obtain a preset parameter set for the liquid-cooled server, which includes the liquid-cooled server's computing power, operating power, primary oil pump speed, secondary water pump speed, and ambient temperature. The preset parameter set of the liquid-cooled server is input into the temperature prediction model of the liquid-cooled server to obtain the predicted temperature and the predicted temperature uniformity. A liquid-cooled server temperature control model is constructed based on the predicted temperature and the predicted temperature uniformity, and temperature control information is output. The temperature control information includes temperature control anomaly information and temperature control normal information. Based on the temperature control information, liquid-cooled server parameter adjustment instructions are fed back. The liquid-cooled server parameter adjustment instructions include primary oil pump speed adjustment instructions, secondary water pump speed adjustment instructions, and computing power adjustment instructions.

2. The temperature control prediction method according to claim 1, characterized in that, The step of constructing a liquid-cooled server temperature prediction model based on the historical monitoring dataset of the liquid-cooled server using an improved XGBOOST algorithm includes: The historical temperature uniformity of the liquid-cooled server is obtained based on the historical temperature of the liquid-cooled server, which includes the historical temperature of the CPU, the historical temperature of the GPU, and the historical temperature of the memory. Given the historical monitoring dataset of the liquid-cooled server and the historical temperature uniformity of the liquid-cooled server, the improved XGBOOST algorithm is used to learn k decision trees additively to construct the temperature prediction model of the liquid-cooled server.

3. The temperature control prediction method according to claim 2, characterized in that, The step of obtaining the historical temperature uniformity of the liquid-cooled server based on the historical temperature of the liquid-cooled server includes: The number of heat-generating components in the liquid-cooled server is obtained, including the number of CPUs, GPUs, and memory. The historical temperature uniformity of the liquid-cooled server is calculated based on the historical temperature of the liquid-cooled server and the number of heat-generating components.

4. The temperature control prediction method according to claim 2, characterized in that, The step of constructing the liquid-cooled server temperature prediction model by additively learning k decision trees using the improved XGBOOST algorithm includes: Obtain the loss function of the liquid-cooled server temperature prediction model, the loss function including a temperature loss function and a temperature uniformity loss function; Obtain the model regularization term of the liquid-cooled server temperature prediction model; The model objective function is obtained based on the loss function and the model regularization term; The objective function of the model is expanded using a second-order Taylor series. The objective function of the model is simplified by simplifying the constant term, and the liquid-cooled server temperature prediction model is obtained.

5. The temperature control prediction method according to claim 4, characterized in that, The expression for the temperature loss function is as follows: ,in, l Let y be the temperature loss function. j Historical temperatures of liquid-cooled servers Let be the predicted historical temperature output by the j-th decision tree in the t-th iteration of the model. This represents the predicted historical temperature output by the j-th decision tree during the (t-1)th iteration of the model. Let k be the predicted historical temperature of the j-th decision tree in the model at the t-th iteration, and k be the number of decision trees in the model; the expression for the temperature uniformity loss function is: Where w is the temperature uniformity loss function, r j To ensure the historical temperature uniformity of liquid-cooled servers, Let the predicted historical temperature uniformity be the output of the j-th decision tree in the t-th iteration of the model. To determine the predicted historical temperature uniformity output by the j-th decision tree in the (t-1)th iteration of the model, Let represent the predicted historical temperature uniformity of the j-th decision tree in the t-th iteration of the model.

6. The temperature control prediction method according to claim 1, characterized in that, The step of constructing a liquid-cooled server temperature control model based on the predicted temperature and the predicted temperature uniformity, and outputting temperature control information, includes: A preset predicted temperature threshold and a predicted temperature uniformity threshold are defined. When the predicted temperature is greater than or equal to the predicted temperature threshold or the predicted temperature uniformity is less than or equal to the predicted temperature uniformity threshold, the temperature control abnormality information is output. When the predicted temperature is less than the predicted temperature threshold and the predicted temperature uniformity is greater than the predicted temperature uniformity threshold, the temperature control normality information is output.

7. The temperature control prediction method according to claim 1, characterized in that, The parameter adjustment instructions for the liquid-cooled server based on the temperature control information feedback include: When the temperature control abnormality information is output, the speed of the liquid cooling server's oil pump is determined. When the speed of the liquid cooling server's oil pump is less than the maximum speed of the oil pump, the oil pump speed adjustment command is fed back to adjust the speed of the liquid cooling server's oil pump until the temperature control normality information is output. When the liquid-cooled server runs the oil pump at a speed equal to the maximum speed, the liquid-cooled server runs the secondary water pump at a speed that is determined. When the liquid-cooled server runs the secondary water pump at a speed that is less than the maximum speed of the secondary water pump, the liquid-cooled server sends back a secondary water pump speed adjustment command to adjust the secondary water pump speed until the temperature control normal information is output. When the secondary water pump speed of the liquid-cooled server is equal to the maximum speed, the operating computing power adjustment command is fed back to reduce the operating computing power of the liquid-cooled server until the temperature control normal information is output.

8. A temperature control and prediction device for a liquid-cooled server, characterized in that, The device is applied to the temperature control prediction method as described in any one of claims 1-7, and includes a data acquisition module, a model building module, and a temperature control module; The data acquisition module is used to acquire the historical monitoring dataset of the liquid-cooled server. The historical monitoring dataset of the liquid-cooled server includes the historical temperature of the liquid-cooled server, the historical computing power of the liquid-cooled server, the historical power of the liquid-cooled server, the historical speed of the primary oil pump of the liquid-cooled server, the historical speed of the secondary water pump of the liquid-cooled server, and the historical ambient temperature of the liquid-cooled server. The model building module is used to construct a liquid-cooled server temperature prediction model based on the historical monitoring dataset of the liquid-cooled server using an improved XGBOOST algorithm; obtain a preset parameter set for the liquid-cooled server, which includes the liquid-cooled server's computing power, operating power, primary oil pump speed, secondary water pump speed, and ambient temperature; and input the preset parameter set into the liquid-cooled server temperature prediction model to obtain the predicted temperature and predicted temperature uniformity. The temperature control module is used to construct a liquid-cooled server temperature control model based on the predicted temperature and the predicted temperature uniformity, and output temperature control information. The temperature control information includes temperature control abnormal information and temperature control normal information. Based on the temperature control information, the module feeds back liquid-cooled server parameter adjustment instructions, which include primary oil pump speed adjustment instructions, secondary water pump speed adjustment instructions, and computing power adjustment instructions.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the temperature control prediction method as described in any one of claims 1-7.

10. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform the temperature control prediction method as described in any one of claims 1-7.

Images