Thermal management control method of data center, electronic equipment and readable storage medium

Abstract

The invention discloses a thermal management control method of a data center, electronic equipment and a readable storage medium, and belongs to the technical field of communication. The method comprises the following steps: acquiring first data, wherein the first data represents a plurality of actual operation conditions of the data center, and control parameters and operation modes under each actual operation condition; performing data supplement on the first data based on a pre-trained mechanism model to obtain second data, and training an energy consumption prediction data model according to the second data to obtain a trained energy consumption prediction data model; and based on the trained energy consumption prediction data model, predicting to obtain a target control parameter and a target operation mode for the data center, and performing thermal management control on the data center according to the target control parameter and the target operation mode. According to the invention, thermal management control with high calculation efficiency and high accuracy can be carried out on the data center, and the stability and reliability of thermal management control on the data center are improved.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a thermal management control method, electronic equipment, and readable storage medium for a data center. Background Art

[0002] With the rise of information industries such as 5G, big data, cloud computing, artificial intelligence, and the industrial Internet, it is crucial to reduce the PUE (Power Usage Effectiveness) of data center systems, build new green data centers, and use advanced models and algorithms to optimize and control the thermal management system of data centers in real time.

[0003] To achieve real-time energy conservation in data center thermal management control systems, domestic and international scholars primarily employ single-model approaches, either mechanistic or data-based models. Currently, mechanistic models are used to model data center thermal management system components, resulting in highly accurate, modified models. This approach optimizes system operating parameters based on the mechanistic model, significantly reducing system energy consumption. However, these modeling methods are complex, require specialized thermodynamics expertise, and are computationally expensive and inefficient, making them difficult to implement in real-time computational optimization systems.

[0004] In recent years, a growing number of researchers have explored new approaches, employing data modeling to optimize and control data center thermal management systems for energy conservation. Data-driven models offer high computational efficiency for predicting data center temperatures, providing insights for subsequent energy conservation and optimization. However, these modeling approaches rely entirely on the quality of collected data, resulting in high data costs and a lack of thermodynamic support. This can lead to counterintuitive behavior and even system failures, as well as limited initial data collection, low model accuracy, and poor energy conservation results.

[0005] That is, although both the mechanism model and the data model can describe the thermodynamic state of the data center thermal management system and provide mathematical expressions for system optimization and control, the two models have their own advantages and disadvantages. The mechanism model has high accuracy, but low computational efficiency and is difficult to optimize in real time. Although the data model has high computational efficiency, it has no thermodynamic knowledge support and depends on the quality and quantity of data. The model has low accuracy and is prone to "anti-logic" situations. Summary of the Invention

[0006] The main purpose of this application is to provide a thermal management control method, electronic device and readable storage medium for a data center, aiming to achieve high-computational efficiency and high-accuracy thermal management control of the data center, and to improve the stability and reliability of thermal management control of the data center.

[0007] To achieve the above objectives, the present application provides a data center thermal management control method, comprising:

[0008] Acquire first data, where the first data represents multiple actual operating conditions of the data center, and control parameters and operating modes under each actual operating condition;

[0009] Based on a pre-trained mechanism model, the first data is supplemented to obtain second data, and an energy consumption prediction data model is trained based on the second data to obtain a trained energy consumption prediction data model;

[0010] Based on the trained energy consumption prediction data model, target control parameters and target operating modes for the data center are predicted, and thermal management control of the data center is performed according to the target control parameters and target operating modes.

[0011] In addition, to achieve the above objectives, the present application further provides a thermal management control system, which is applied to the thermal management control method of the data center as described above, and includes:

[0012] A liquid cooling system comprising a cooling device, a heat exchanger, a first three-way valve, a first water pump, and a second water pump; the heat exchanger comprising a first pipeline and a second pipeline; the first pipeline having a first water inlet and a first water outlet; the second pipeline having a second water inlet and a second water outlet; the first three-way valve comprising a liquid inlet, a first liquid outlet, and a second liquid outlet; and the cooling device being a cooling tower or an air cooler;

[0013] An air cooling system includes a refrigeration device, a fan, a second three-way valve, and a third water pump, wherein the second three-way valve includes a water inlet, a first water outlet, and a second water outlet, and the refrigeration device is a chiller or a refrigeration unit;

[0014] server;

[0015] The outlet end of the cooling device, the first water inlet, the first water outlet, the liquid inlet, the first liquid outlet, and the inlet end of the cooling device form a liquid cooling circuit, and the first water pump is provided in the liquid cooling circuit; the coolant outlet of the server, the second water inlet, the second water outlet, and the coolant inlet of the server form a heat exchange circuit, and the second water pump is provided in the heat exchange circuit. The fluid in the heat exchange circuit is used to perform heat exchange with the fluid in the liquid cooling circuit in the heat exchanger;

[0016] The liquid outlet of the refrigeration device, the water inlet, the first water outlet, the liquid cooling inlet of the fan, the liquid cooling outlet of the fan, and the liquid inlet of the refrigeration device form an air cooling circuit, the third water pump is provided in the air cooling circuit, and the fan is used to blow cold air to the server;

[0017] The air cooling circuit is connected to the liquid cooling circuit through the second water outlet, and the liquid cooling circuit is connected to the air cooling circuit through the second liquid outlet.

[0018] In addition, to achieve the above-mentioned purpose, the present application also provides an electronic device, which includes: a memory, a processor, and a thermal management control program for a data center stored on the memory and runnable on the processor. When the thermal management control program for the data center is executed by the processor, the thermal management control method for the data center as described above is implemented.

[0019] In addition, to achieve the above-mentioned purpose, the present application also provides a storage medium, which is a computer-readable storage medium, and the computer-readable storage medium stores a thermal management control program of a data center. When the thermal management control program of the data center is executed by a processor, the thermal management control method of the data center as described above is implemented.

[0020] The present application proposes a thermal management control method, electronic device and readable storage medium for a data center. In the thermal management control method for a data center, the technical solution of the embodiment of the present application is to obtain first data, which represents multiple actual operating conditions of the data center, as well as control parameters and operating modes under each actual operating condition, and based on a pre-trained mechanism model, the first data is supplemented to obtain second data, so that the present application supplements data to the interval without valid measured data through the mechanism model, and establishes a concise and effective data sample library. By fusing the mechanism model sample library, for the blank operating condition area without measured data, super-regional data can be supplemented to improve the effective coverage of the sample library. That is, the embodiment of the present application can form a fusion sample library of field data and mechanism model with high data quality, wide range, small quantity and good effect. , and according to the second data, train the energy consumption prediction data model to obtain a trained energy consumption prediction data model; based on the trained energy consumption prediction data model, predict the target control parameters and target operation mode for the data center, so as to achieve the fitting of a comprehensive and accurate system prediction energy consumption model. The embodiment of the present application constructs a mechanism model according to the system operation mechanism, improves the accuracy of the mechanism model according to the on-site collected data, and supplements the sample library data through the mechanism model with improved accuracy, so as to facilitate real-time optimization and calculation of the optimal control parameters and optimal operation mode of the thermal management control system for the data center, so as to determine the thermal management control strategy for the server of the data center, and then realizes thermal management control of the data center with high computational efficiency and high accuracy, and improves the stability and reliability of thermal management control of the data center. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0022] Figure 1 This is a flow chart of an embodiment of a thermal management control method for a data center according to the present application;

[0023] Figure 2 This is the real-time energy-saving optimization and control process of the thermal management system of the data center in the embodiment of the present application;

[0024] Figure 3 This is a schematic diagram of the dual-drive process of the data center thermal management system according to an embodiment of the present application;

[0025] Figure 4 This is a schematic diagram of a scenario of thermal management control of a data center in an embodiment of the present application;

[0026] Figure 5 Schematic diagram of the operating mechanism of the system-level dual-drive mechanism model in the embodiment of the present application;

[0027] Figure 6 Schematic diagram of the operating mechanism of the device-level dual-drive mechanism model in the embodiment of the present application;

[0028] Figure 7 A schematic diagram of mutual calibration and optimization of the data model and the mechanism model in the embodiment of the present application;

[0029] Figure 8 This is a flow chart of the model calibration optimization of the model coefficient correction method in the embodiment of the present application;

[0030] Figure 9 This is a flow chart of the model calibration optimization of the error correction method in the embodiment of the present application;

[0031] Figure 10 This is an experimental data diagram of the model coefficient correction method in the embodiment of this application;

[0032] Figure 11 This is an experimental data diagram of the error correction method in the embodiment of this application;

[0033] Figure 12 This is a graph of predicted experimental data for dual-drive regulation in the embodiment of this application and basic regulation in related technologies;

[0034] Figure 13 This is an experimental data diagram of various operating conditions of the data center in the embodiment of this application;

[0035] Figure 14 A comparison chart of predicted data between the dual-drive control strategy of the embodiment of the present application and other control strategies in related technologies;

[0036] Figure 15 This is a schematic diagram of the structure of the thermal management control system of the data center in an embodiment of the present application;

[0037] Figure 16 This is a schematic diagram of the hardware structure of the electronic device involved in the embodiment of the present application.

[0038] Figure 15 Description of the accompanying figures:

[0039]

[0040] The realization of the objectives, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0041] It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application.

[0042] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0043] It should be noted that all directional indications in the embodiments of the present application (such as up, down, left, right, front, back, etc.) are only used to explain the relative position relationship, movement status, etc. between the various components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0044] In this application, unless otherwise specified or limited, the terms "connection" and "fixation" should be understood in a broad sense. For example, "fixation" can mean fixed connection, detachable connection, or integration; mechanical connection or electrical connection; direct connection or indirect connection through an intermediate medium; internal communication between two elements or interaction between two elements, unless otherwise specified. For those skilled in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0045] In addition, the descriptions of "first", "second", etc. in this application are for descriptive purposes only and should not be understood as indicating or implying their relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined as "first" or "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between the various embodiments can be combined with each other, but this must be based on the fact that they can be implemented by ordinary technicians in this field. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0046] In recent years, a growing number of researchers have explored new approaches, employing data modeling to optimize and control data center thermal management systems for energy conservation. Data-driven models offer high computational efficiency for predicting data center temperatures, providing insights for subsequent energy conservation and optimization. However, these modeling approaches rely entirely on the quality of collected data, resulting in high data costs and a lack of thermodynamic support. This can lead to counterintuitive behavior and even system failures, as well as limited initial data collection, low model accuracy, and poor energy conservation results.

[0047] That is, although both the mechanism model and the data model can describe the thermodynamic state of the data center thermal management system and provide mathematical expressions for system optimization and control, the two models have their own advantages and disadvantages. The mechanism model has high accuracy, but low computational efficiency and is difficult to optimize in real time. Although the data model has high computational efficiency, it has no thermodynamic knowledge support and depends on the quality and quantity of data. The model has low accuracy and is prone to "anti-logic" situations.

[0048] Based on this, the embodiment of the present application provides a thermal management control method for a data center, referring to Figure 1 , Figure 1 This is a flow chart of an embodiment of a thermal management control method for a data center according to the present application. In this embodiment, the thermal management control method for a data center includes:

[0049] Step S10: acquiring first data, where the first data represents a plurality of actual operating conditions of the data center, and control parameters and operating modes under each actual operating condition;

[0050] In this embodiment, the first data can be obtained by collecting operating data under various operating conditions of the data center (the operating data includes control parameters and operating modes) and preprocessing the operating data, wherein the preprocessing includes but is not limited to noise reduction processing or downsampling processing.

[0051] Exemplarily, obtaining the first data includes:

[0052] Step A10: collecting control parameters and operating modes under various actual operating conditions of the data center to obtain third data;

[0053] Step A20, performing data processing on the third data based on a preset data processing algorithm to obtain first data, wherein the preset data processing algorithm is at least one of Latin hypercube sampling method, wavelet threshold denoising method, empirical mode decomposition denoising method, k-nearest neighbor outlier removal method, Mahalanobis distance outlier removal method, linear interpolation data repair method, expectation maximization numerical repair method, Latin hypercube sampling algorithm interval partitioning method, and redundant data reduction method.

[0054] This embodiment obtains third data by collecting control parameters and operating modes under various actual operating conditions of the data center; and performs data processing on the third data based on a preset data processing algorithm to obtain first data, wherein the preset data processing algorithm is at least one of Latin hypercube sampling method, wavelet threshold denoising method, empirical mode decomposition denoising method, k-nearest neighbor outlier removal method, Mahalanobis distance outlier removal method, linear interpolation data repair method, expectation maximization numerical repair method, Latin hypercube sampling algorithm interval partitioning method, and redundant data reduction method, so that for the collected measured data, abnormal data can be excluded, abnormal data can be repaired, data reduction can be achieved, and the complexity of the sample library can be optimized.

[0055] After step S10, step S20 is executed to supplement the first data based on the pre-trained mechanism model to obtain second data, and to train an energy consumption prediction data model based on the second data to obtain a trained energy consumption prediction data model;

[0056] In this embodiment, those skilled in the art have conducted in-depth research on the training method or construction method of the mechanism model. For example, a mechanism model can be constructed according to the system operation mechanism, and the accuracy of the mechanism model can be improved based on the collected field data. The sample library data can be supplemented by the mechanism model with improved accuracy.

[0057] In one example, the mechanism model is a system-level dual-drive model, in which the components use a data model with a faster calculation speed. The system is built according to the actual mechanism, with an overall fast calculation speed and a certain degree of anti-logic.

[0058] In another example, the mechanism model is a device-level dual-drive model, which is constructed from the system down to the device level. This completely avoids "anti-logic" errors, but the simulation platform has a slower computational speed. These two examples are described in more detail in the following embodiments and will not be repeated here.

[0059] After step S20, step S30 is executed to predict target control parameters and target operating modes for the data center based on the trained energy consumption prediction data model, and perform thermal management control on the data center according to the target control parameters and target operating mode.

[0060] The target control parameter refers to a control parameter for achieving a thermal management control target for the data center. The target operating mode refers to an operating mode for achieving a thermal management control target for the data center. The thermal management control target is to minimize the system PUE or total system energy consumption of the data center.

[0061] Exemplarily, the types of operating modes include a first operating mode, a second operating mode and a third operating mode, wherein the first operating mode is an operating mode in which the refrigeration equipment is turned off and the cooling equipment supplies liquid to the liquid-cooled side and the air-cooled side, wherein the refrigeration equipment is a chiller or a refrigeration unit, and the cooling equipment is a cooling tower or an air cooler; the second operating mode is that the refrigeration equipment supplies liquid to the air-cooled side, and the cooling equipment supplies liquid to the liquid-cooled side; the third operating mode is that both the refrigeration equipment and the cooling equipment supply liquid to the air-cooled side and the liquid-cooled side.

[0062] For example, the control parameters refer to the operating control parameters of each power-consuming device. These power-consuming devices (i.e., the underlying controlled components described below) may include fans, water pumps, compressors, cooling equipment, and refrigeration equipment. Cooling equipment may be a cooling tower or air cooler, and refrigeration equipment may be a chiller or refrigeration unit. The operating control parameters include fan frequency, water pump frequency, compressor frequency, and valve opening.

[0063] The present application proposes a thermal management control method, electronic device and readable storage medium for a data center. In the thermal management control method for a data center, the technical solution of the embodiment of the present application is to obtain first data, which represents multiple actual operating conditions of the data center, as well as control parameters and operating modes under each actual operating condition, and based on a pre-trained mechanism model, the first data is supplemented to obtain second data, so that the present application supplements data to the interval without valid measured data through the mechanism model, and establishes a concise and effective data sample library. By fusing the mechanism model sample library, for the blank operating condition area without measured data, super-regional data can be supplemented to improve the effective coverage of the sample library. That is, the embodiment of the present application can form a fusion sample library of field data and mechanism model with high data quality, wide range, small quantity and good effect. , and according to the second data, train the energy consumption prediction data model to obtain a trained energy consumption prediction data model; based on the trained energy consumption prediction data model, predict the target control parameters and target operation mode for the data center, so as to achieve the fitting of a comprehensive and accurate system prediction energy consumption model. The embodiment of the present application constructs a mechanism model according to the system operation mechanism, improves the accuracy of the mechanism model according to the on-site collected data, and supplements the sample library data through the mechanism model with improved accuracy, so as to facilitate real-time optimization and calculation of the optimal control parameters and optimal operation mode of the thermal management control system for the data center, so as to determine the thermal management control strategy for the server of the data center, and then realizes thermal management control of the data center with high computational efficiency and high accuracy, and improves the stability and reliability of thermal management control of the data center.

[0064] It is worth mentioning that the embodiment of the present application integrates the mechanism model with the data model, processes the field data through multiple methods, and uses the mechanism model to supplement the intervals with invalid data, thereby forming a sample library (i.e., second data) that integrates the field data and the mechanism model with high data quality, wide range, small quantity, and good effect. The data model fits the sample library data to form a comprehensive and effective system data model (i.e., a dual-drive model that fits the data model and the mechanism model). The data model can be combined with the optimization algorithm to generate the best control variables and operating modes in real time.

[0065] The embodiment of the present application integrates the mechanism model and the data model in a systematic way, taking advantage of their strengths and overcoming their weaknesses, thereby producing a dual-drive model with low computational cost, strong model logic, and comprehensive functions.

[0066] The embodiment of the present application discloses an energy-saving control method for a data center cooling system based on a dual-driven model of mechanism and data, which belongs to the field of energy-saving control of data centers. The main steps include: excluding or repairing abnormal values ​​from the on-site collected data through the mechanism model; supplementing the data to the interval without valid measured data with the mechanism model to establish a concise and effective data sample library with comprehensive working condition coverage; fitting a comprehensive and accurate system PUE and energy consumption model by the data-driven method; combining the optimization algorithm (the optimization algorithm can be a genetic algorithm, particle swarm method, simulated annealing algorithm or ant colony algorithm and other optimization algorithms, which are not limited here) to calculate the optimal control parameters and operating mode of the thermal management system and store them in the knowledge base; according to the actual working conditions on site, the optimal control parameters and operating mode of the corresponding working conditions can be called in real time to avoid online optimization delays; the PLC controls the operating status of the equipment based on the optimization decision to achieve real-time energy-saving optimization control of the system. The embodiment of the present application has low computational cost, high precision, good logic, and low dependence on data quality for thermal management control of data centers, and can achieve comprehensive energy saving of the thermal management system of data centers in the short, medium and long term.

[0067] In order to help understand the technical principles of the embodiments of the present application, a specific embodiment is listed, referring to Figure 2 , Figure 2 This is a flow chart of real-time energy-saving optimization and control of the thermal management system of the data center in the embodiment of the present application. This embodiment connects the host computer intelligent computing platform, PLC (Programmable Logic Controller), and terminal equipment system to form a set of efficient control system processes. Specifically:

[0068] The host computer intelligent computing platform consists of a database and a dual-drive model. The database stores and manages data, while also implementing functions such as data cleaning, data classification, and data updates. The dual-drive model intelligently calculates the optimal operating status of power-consuming components such as cooling towers, chillers, pumps, and fans in real time and sends this information to the PLC controller. The PLC controller then uploads the collected data to the database.

[0069] PLC controller: controls the working status of the terminal equipment according to the instructions of the host computer, and collects real-time data on site through sensors.

[0070] Terminal equipment: The working status of the terminal equipment of the data center thermal management system changes in real time. While meeting the temperature requirements of the server in the computer room, the energy consumption of the thermal management system is reduced to achieve real-time energy-saving control function.

[0071] It should be noted that the above-mentioned specific embodiment 1 is only used to help understand the technical concept of the embodiment of the present application, and does not constitute a limitation of the present application. More simple transformations based on the technical concept should all be within the scope of protection of the present application.

[0072] As an example, in step S30, based on the trained energy consumption prediction data model, target control parameters and target operation modes for the data center are predicted, including:

[0073] Step B10: Based on the trained energy consumption prediction data model, predict the optimal control parameters and optimal operating mode for each operating condition of the data center;

[0074] Step B20, storing the optimal control parameters and optimal operating modes corresponding to each of the operating conditions in a knowledge base;

[0075] Step B30: dynamically obtain the real-time operating conditions of the data center, query and call from the knowledge base to obtain the target control parameters and target operating mode corresponding to the real-time operating conditions.

[0076] Exemplarily, in step B30, querying and calling the knowledge base to obtain target control parameters and target operation modes for the data center includes:

[0077] Step C10: querying and retrieving from the knowledge base control parameters corresponding to achieving a thermal management control target and an operating mode corresponding to achieving the thermal management control target, wherein the thermal management control target is minimizing the system PUE or minimizing the total system energy consumption of the data center;

[0078] Step C20, using the control parameters corresponding to the thermal management control targets under the real-time operating conditions as target control parameters for the data center; and

[0079] Step C30 : Using the operating mode that achieves the thermal management control target under the real-time operating conditions as the target operating mode for the data center.

[0080] In this embodiment, dynamically acquiring the real-time operating conditions of the data center refers to periodically acquiring the real-time operating conditions of the data center, or acquiring the real-time operating conditions of the data center in real time at preset time intervals.

[0081] The embodiment of the present application predicts the optimal control parameters and optimal operating mode for each operating condition of the data center based on a trained energy consumption prediction data model, and stores the optimal control parameters and optimal operating mode corresponding to each operating condition in a knowledge base. Then, the real-time operating conditions of the data center are dynamically obtained, and the target control parameters and target operating mode corresponding to the real-time operating conditions are queried and called from the knowledge base, thereby matching the real-time operating conditions with the full range of operating conditions in the knowledge base. The conditions matched by the knowledge base are quickly called in real time to select the corresponding optimal control parameters and optimal operating mode. This embodiment traverses all possible operating conditions of the system and can be combined with an optimization algorithm (the optimization algorithm can be a preset optimization algorithm such as a genetic algorithm, a particle swarm algorithm, a simulated annealing algorithm, or an ant colony algorithm) to pre-calculate the optimal control parameters and optimal operating mode corresponding to all operating conditions, store all of them in and establish a knowledge base, and according to the actual on-site operating conditions, the optimal control parameters and operating mode of the corresponding operating condition can be quickly called in real time from the knowledge base, avoiding online optimization delays, eliminating the algorithm optimization process, and achieving fast real-time optimization.

[0082] For example, the control strategy in the related art is: measured working conditions → system fitting model → genetic optimization algorithm → policy issuance, which has a time delay. However, the control strategy of the embodiment of the present application is: traversing working conditions → system mechanism supplements working condition data to obtain a sample library with more comprehensive working condition coverage → training an energy consumption prediction data model based on the more comprehensive sample library → genetic optimization algorithm → policy storage in knowledge base → measured working conditions → knowledge base call → policy issuance. Compared with the control strategy in the related art, the embodiment of the present application is established through the knowledge base, and can directly call the preferred control strategy corresponding to the knowledge base, eliminating the optimization delay.

[0083] Furthermore, the energy consumption prediction data model is trained and stored in a cloud server, and the knowledge base is located locally in the user terminal.

[0084] In the embodiment of the present application, the energy consumption prediction data model is trained and stored on the cloud server, and the knowledge base is set up locally in the user terminal, thereby realizing the integration of end, edge and cloud. A large amount of model training and verification are carried out on the cloud, and control strategy calculations are performed on various working conditions at the same time. The calculation results are output as a knowledge base, which is stored in the edge data center and quickly called on the user terminal. That is, by establishing a knowledge base locally, the corresponding preferred control strategy in the local knowledge base is directly called to shorten the control response time.

[0085] In one possible implementation, the method further includes:

[0086] Step D10, when the data center is in a risky condition, query and call the safety regulation strategy corresponding to the risky condition from the knowledge base, wherein the safety regulation strategy includes control parameters and operating modes for safety regulation of the risky condition.

[0087] In this embodiment, based on the energy consumption prediction data model, risky operating conditions can be predicted and analyzed for the data center. Safety control strategies for these risky conditions, namely, optimal control parameters and optimal operating modes, are obtained and stored in a knowledge base. When the data center is in a risky condition, the corresponding safety control strategy is retrieved and invoked from the knowledge base, allowing for rapid response and execution.

[0088] In the embodiment of the present application, when a data center is in a risky condition, the safety control strategy corresponding to the risky condition is queried and called from the knowledge base, wherein the safety control strategy includes control parameters and operating modes for safety control of the risky condition, thereby pre-analyzing the risky condition, abnormal condition or extreme condition, and obtaining the corresponding safe and reliable control strategy in advance, storing it in the knowledge base for reliable control and efficient emergency response when it actually occurs, and having reliable risk prediction and risk response functions.

[0089] In a possible implementation, training an energy consumption prediction data model based on the second data, and obtaining the trained energy consumption prediction data model includes:

[0090] Step E10: constructing a sample database based on the second data;

[0091] Step E20: performing feature selection on the sample database based on principal component analysis to obtain independent variable parameters that affect the thermal management control target;

[0092] Step E30 : training the energy consumption prediction data model according to the independent variable parameters that affect the thermal management control target to obtain a trained energy consumption prediction data model.

[0093] In this embodiment, the thermal management control target includes but is not limited to minimizing the system PUE or minimizing the total system energy consumption of the data center. Those skilled in the art can set it according to actual conditions, and this embodiment does not make any specific limitations.

[0094] The embodiment of the present application constructs a sample database based on the second data, and performs feature selection on the sample database based on the principal component analysis method to obtain independent variable parameters that affect the thermal management control target, and performs model training on the energy consumption prediction data model based on the independent variable parameters that affect the thermal management control target to obtain a trained energy consumption prediction data model, thereby realizing a feature parameter selection method based on the principal component analysis method and other methods, converging on the direct and indirect parameters that have the greatest impact on the target control quantity as the required feature quantity, thereby greatly reducing the system sensor layout, reducing the parameter complexity of the dual-drive model (i.e., the above-mentioned energy consumption prediction data model), and improving the fitting speed and accuracy of the dual-drive model.

[0095] Further, in order to help understand the technical principles of the embodiments of the present application, a specific embodiment 2 is listed, referring to Figure 3 , Figure 3 This is a schematic diagram of the dual-drive process of the data center thermal management system according to an embodiment of the present application, specifically including:

[0096] (1) Sample library construction:

[0097] First, after collecting field data, data preprocessing is performed: wavelet threshold denoising, empirical mode decomposition and other methods are used to reduce data noise, remove sensor data noise, shield interference, and ensure the authenticity of the signal; empirical mode decomposition denoising, k-nearest neighbor outlier removal or Mahalanobis distance outlier removal and other data processing methods are used to filter data, remove outliers, and retain valid data; linear interpolation, expectation maximization algorithm and other methods are used to learn and analyze historical information data to complete the repair of missing data or faulty data.

[0098] Furthermore, when the mechanism model has a high accuracy in the future, the mechanism model simulation data can be used to eliminate or repair abnormal values ​​in the field collected data.

[0099] Then, this embodiment uses Latin hypercube sampling to determine the sample library interval and remove redundant data to ensure that the data is simple and effective, and uses the mechanism model to supplement the data to the interval without valid data, and establishes a sample library that integrates field data and mechanism models with high data quality, wide range, small quantity and good effect.

[0100] (2) Data model training:

[0101] Principal component analysis (PCA) was used to perform feature selection on the dual-drive sample library data, selecting the parameters that most significantly impact the system's total power consumption and PUE as independent variables. The most representative parameters for this data center's thermal management system are ambient temperature and humidity, air-to-water heat load, and the control parameters and operating modes of various power-consuming devices. The output parameters are system power consumption and PUE.

[0102] After determining the input and output parameters of the neural network function, we iteratively fit the system model using data-driven algorithms such as BPNN, LSTM, SVMXGBoost, or LightGBM until the RMSE and MAPE meet the accuracy requirements. We then compare and analyze the accuracy and fitting performance of various data-driven algorithms to select the optimal data-driven algorithm for this data center and improve system control effectiveness.

[0103] (3) Optimization calculation:

[0104] The thermal environment within a data center varies with ambient temperature, humidity, and heat load. If the operating parameters of the data center thermal management system are improperly selected, servers can overheat or undercool. Severe overheating can cause server downtime, while severe undercooling can lead to inefficient cooling and significant power consumption. Therefore, to maintain servers within a safe temperature range and minimize energy consumption, it is necessary to calculate the optimal system control parameters and operating modes in real time under varying ambient temperatures, humidity, and heat loads.

[0105] Optimization algorithms such as genetic algorithms can achieve the above requirements. Taking the lowest system PUE as the thermal management control target as an example, the optimal control parameters and operating mode of the system under the ambient temperature, humidity and air-liquid heat load at any time are calculated.

[0106] (4) Knowledge base construction:

[0107] Traversing all possible operating conditions of the thermal management system, the optimal control parameters and optimal operating modes corresponding to all operating conditions can be pre-calculated using the optimization algorithm and stored in the knowledge base;

[0108] (5) Optimize strategy issuance and execution:

[0109] There are two modes for optimization strategy sources:

[0110] Mode 1: Based on the actual working conditions on site, the optimal control parameters and operating mode for the corresponding working conditions can be quickly and in real time called from the knowledge base, avoiding online optimization delays.

[0111] Mode 2: Based on a high-precision system energy consumption prediction data model and actual operating parameters, an optimization algorithm (or optimization algorithm) is used to perform real-time optimization. After the optimization is completed, the optimal control parameters and operating mode under the operating conditions can be output. A certain optimization time is required. Among them, the optimization algorithm can be, for example, a particle swarm algorithm, a simulated annealing algorithm, or an ant colony algorithm, etc. This embodiment does not make specific limitations on this.

[0112] The optimization results (optimization strategy) are sent down by the control device to the underlying controllable components to achieve system energy-saving optimization control, maintain the server within a safe temperature range and minimize energy consumption. Among them, the control device can be, for example, a PLC or DDC (Direct Digital Control).

[0113] It should be noted that the above-mentioned specific embodiment 2 is only used to help understand the technical concept of the embodiment of the present application, and does not constitute a limitation of the present application. More simple transformations based on the technical concept should all be within the scope of protection of the present application.

[0114] Based on the first embodiment of the present application, a second embodiment of the thermal management control method for a data center of the present application is proposed, wherein the method further includes:

[0115] Step F10, collecting actual operation data generated by each of the bottom-level components to be controlled based on the target control parameters and target operation mode;

[0116] Step F20: constructing a model prediction value deviation model based on the actual operation data and the prediction data of the energy consumption prediction data model under the operating conditions corresponding to the actual operation data;

[0117] Step F30 , using the model prediction value deviation model, performing error correction and optimization on the energy consumption prediction data model.

[0118] The embodiment of the present application collects the actual operation data generated by each bottom-level controlled component based on the target control parameters and the target operation mode, and constructs a model prediction value deviation model based on the actual operation data and the prediction data of the energy consumption prediction data model under the operating conditions corresponding to the actual operation data. The model prediction value deviation model is used to correct and optimize the energy consumption prediction data model, thereby proposing an error correction method, which improves the accuracy of the mechanism model by fitting the prediction error with the data model and superimposing the error data model on the mechanism model. Figure 9 shown.

[0119] To further understand the technical concept of the embodiments of the present application, please refer to Figure 7 , Figure 7This is a schematic diagram of the mutual calibration and optimization of the data model and the mechanism model in an embodiment of the present application. This embodiment constructs a model error function based on the target control parameters and target operating mode corresponding to the real-time operating conditions, as well as the predicted data of the energy consumption prediction data model under the real-time operating conditions, and iteratively optimizes the model coefficients of the energy consumption prediction data model based on the model error function, thereby effectively calibrating the accuracy of the data model and the mechanism model, continuously improving the model accuracy during operation, and further continuously improving the accuracy, reliability, and energy-saving effectiveness of the intelligent control strategy.

[0120] In addition, you can refer to Figure 8 and Figure 9 ,in, Figure 8 This is a flow chart of the model calibration optimization of the model coefficient correction method in the embodiment of this application. Figure 9 This is a flow chart for model calibration optimization using the error correction method in an embodiment of this application. This embodiment of this application, based on the model coefficient correction method, uses the model prediction error as the optimization target to optimize the model coefficients of the energy consumption prediction data model, improve the accuracy of the mechanism model, and safeguard the dual-drive sample library. In another example, this embodiment of this application can also improve the accuracy of the mechanism model by fitting the prediction error with the data model and superimposing the error data model on the mechanism model based on the error correction method.

[0121] Further, please refer to Figure 10 and Figure 11 , Figure 10 This is the experimental data diagram of the model coefficient correction method in the embodiment of this application. Figure 11 This is an experimental data diagram of the error correction method in the embodiment of this application. It can be seen from the diagram that compared with the model calibration optimization process of the model coefficient correction method, the model calibration optimization process of the error correction method has a higher prediction accuracy of the mechanism model.

[0122] As an implementable manner, based on a pre-trained mechanism model, the first data is supplemented to obtain the second data, including:

[0123] Step G10: based on the pre-trained mechanism model, exclude or correct abnormal data in the first data to obtain third data; and

[0124] Step G20: supplement the third data based on the pre-trained mechanism model to obtain the second data.

[0125] The embodiment of the present application eliminates or corrects abnormal data in the first data based on a pre-trained mechanism model to obtain third data, and supplements the third data based on the pre-trained mechanism model to obtain second data. Therefore, for the collected measured data, abnormal data can be eliminated and repaired, thereby achieving data simplicity and optimizing the complexity of the sample library.

[0126] In one possible implementation, performing thermal management control on the data center according to the target control parameter and the target operating mode includes:

[0127] Step H10: controlling the operating parameters of the underlying components to be controlled of the data center according to the target control parameters and the target operating mode, so as to perform thermal management control on the data center;

[0128] The bottom-level components to be controlled include fans, water pumps, compressors and valves, and the operating parameters of the bottom-level components to be controlled include fan frequency, water pump frequency, compressor frequency and valve opening.

[0129] As an example, the pre-trained mechanism model is a system-level dual-drive mechanism model, and the method further includes:

[0130] Step I10: Obtain the trained component data models corresponding to the underlying components to be controlled from a preset component model library;

[0131] Step I20: constructing a system-level mechanism model of the data center based on the component data models to obtain the system-level dual-drive mechanism model.

[0132] Please refer to Figure 5 , Figure 5 The following is a schematic diagram of the operating mechanism of the system-level dual-drive mechanism model in an embodiment of the present application. In this embodiment, the system-level dual-drive mechanism model is configured as follows: obtaining component data models corresponding to each of the underlying components to be controlled; constructing and analyzing a system-level mechanism model of the data center based on each of the component data models, namely the system-level dual-drive mechanism model; and obtaining the optimal operating parameters and optimal operating mode of the data center under various operating conditions based on the system-level dual-drive mechanism model.

[0133] The embodiment of the present application adopts a system-level dual-drive mechanism model as the mechanism model, in which the components adopt a data model with faster calculation speed. The system is built according to the actual mechanism, and the overall calculation speed is fast and has a certain degree of anti-logic.

[0134] Furthermore, the system-level mechanism model of the data center is constructed based on the component data models to obtain the system-level dual-drive mechanism model, including:

[0135] Step J10: Based on the component data models, the system-level dual-drive mechanism model is constructed in a first preset building block stacking manner.

[0136] The embodiment of the present application is based on the data models of each component and constructs a system-level dual-drive mechanism model in a first preset building block stacking manner, thereby performing mechanism modeling on typical components and equipment respectively, thereby forming a mechanism model component library. When building the system-level dual-drive mechanism model, corresponding elements are preferentially selected from the mechanism model component library to reduce the secondary construction time of the component / equipment model. The system-level dual-drive mechanism model is quickly formed in the form of building blocks through library selection. This embodiment simplifies the construction efficiency of the system-level dual-drive mechanism model through the component library and building block stacking form, and also improves the scalability of the system mechanism model.

[0137] As another example, the pre-trained mechanism model is a device-level dual-drive mechanism model, based on each component data model, wherein the method further includes:

[0138] Step K10: Obtain the trained component mechanism models corresponding to the underlying components to be controlled from a preset component model library;

[0139] Step K20 , constructing a system-level mechanism model of the data center based on the mechanism models of each component, and obtaining the device-level dual-drive mechanism model.

[0140] Please refer to Figure 6 , Figure 6 The following is a schematic diagram of the operating mechanism of the device-level dual-drive mechanism model in an embodiment of this application. In this embodiment, the device-level dual-drive mechanism model can be configured to: obtain the component mechanism model corresponding to each of the underlying controlled components; based on each of the component mechanism models, construct and analyze the system-level mechanism model of the data center, namely the device-level dual-drive mechanism model; based on the device-level dual-drive mechanism model, obtain the optimal operating parameters and optimal operating mode of the data center under various operating conditions.

[0141] This embodiment of the application utilizes a device-level dual-drive mechanism model, breaking down the mechanism model from the system down to the device level. This model is constructed entirely from the mechanism model, completely avoiding "anti-logic" issues. However, this results in slower computational speeds on the simulation platform. Furthermore, the dual-drive mechanism model offers significant advantages, including high accuracy, low computational complexity, and low cost, independent of data quality and quantity, and low data costs, resulting in significant energy savings.

[0142] Furthermore, the pre-trained mechanism model is a device-level dual-drive mechanism model. Based on the mechanism models of each component, a system-level mechanism model of the data center is constructed to obtain the device-level dual-drive mechanism model, including:

[0143] Step L10: Based on the mechanism models of each component, the device-level dual-drive mechanism model is constructed in a second preset building block stacking manner.

[0144] The embodiment of the present application is based on the mechanism models of each component and constructs a device-level dual-drive mechanism model in a second preset building block stacking manner, thereby performing mechanism modeling on typical components and equipment respectively, thereby forming a mechanism model component library. When building the device-level dual-drive mechanism model, corresponding elements are preferentially selected from the mechanism model component library to reduce the secondary construction time of the component / equipment model, and a system-level mechanism model is quickly formed in the form of building blocks through library selection. This embodiment simplifies the construction efficiency of the device-level dual-drive mechanism model through the component library and building block stacking form, and also improves the scalability of the device-level dual-drive mechanism model.

[0145] In order to help understand the technical concept of the embodiment of the present application, the specific embodiment 3 is listed, referring to Figure 4 , Figure 4 The following is a schematic diagram of the thermal management control scenario of a data center in an embodiment of the present application, including the following main process steps:

[0146] In this embodiment, 1 is the field data (i.e., the first data) after multiple data processing such as data noise reduction, data filtering, and data repair. The data is redundant and messy, and the distribution is uneven, resulting in slow calculation and poor accuracy of the data-driven fitting function.

[0147] In this embodiment, the optimal data interval and size of this database are determined by Latin hypercube sampling, and the field data is filled in to form a sample library of form 2. The data in this sample library is still redundant, messy and unevenly distributed, and further operations are required.

[0148] Perform data simplification on each data interval, delete redundant data, and form a concise and effective form 3 sample library (i.e., third data). Although this database solves the data redundancy problem, the data distribution is still uneven and needs further operation.

[0149] The mechanism model supplements the intervals without valid data to form a comprehensive, concise and effective form 4 mechanism and data dual-driven fusion sample library (i.e., second data). This database is concise and uniform, which can greatly improve the calculation speed and result accuracy of the data-driven fitting function.

[0150] It should be noted that what is disclosed above is only a preferred embodiment of the present application, and it is certainly not intended to limit the scope of protection of the present application. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made in accordance with the claims of the present invention are still within the scope covered by the present application.

[0151] That is to say, the above-mentioned specific embodiment three is only used to help understand the technical concept of the embodiment of this application, and does not constitute a limitation of this application. More simple transformations based on this technical concept should all be within the scope of protection of this application.

[0152] Please refer to Figure 12 , Figure 12 This graph shows experimental data from the dual-drive control of this embodiment and basic control in related technologies. It shows that compared to the basic control predictions in related technologies, the power consumption prediction accuracy of the dual-drive control in this embodiment, which uses a fitted mechanism model and a data model, is higher, meaning the prediction effect is better. By supplementing data with a dual-drive model, this embodiment of the application achieves high security, good accuracy, high robustness, and good generalization of intelligent control.

[0153] In addition, please refer to Figure 13 and Figure 14 , Figure 13 This is an experimental data diagram of various operating conditions of the data center in the embodiment of this application. Figure 14 This is a comparison chart of the predicted data between the dual-drive control strategy of the embodiment of this application and other control strategies in the related art. It can be seen that the dual-drive control strategy of the embodiment of this application has a better optimization effect. Other control strategies, due to the lack of fitting the mechanism model supplementary data, have poor energy-saving effects in the data model, huge early fluctuations, and are prone to "anti-logic". Through data comparison and analysis, the data model control strategy saves 80.14% energy compared to the basic control strategy, and the dual-drive model control strategy of the embodiment of this application saves 73.31% energy compared to the data model control strategy.

[0154] In addition, please refer to Figure 15 The present application also provides a thermal management control system 100 for a data center. The thermal management control system 100 is applied to the thermal management control method for the data center described above. The thermal management control system 100 includes:

[0155] The liquid cooling system includes a cooling device 1, a heat exchanger 2, a first three-way valve 3, a first water pump 4, and a second water pump 5. The heat exchanger 2 includes a first pipeline and a second pipeline. The first pipeline is provided with a first water inlet 21 and a first water outlet 22. The second pipeline is provided with a second water inlet 23 and a second water outlet 24. The first three-way valve 3 includes a liquid inlet 31, a first liquid outlet 32, and a second liquid outlet 33. The cooling device 1 is a cooling tower or an air cooler.

[0156] The air cooling system includes a refrigeration device 6, a fan 7, a second three-way valve 8, and a third water pump 9. The second three-way valve 8 includes a water inlet end 81, a first water outlet end 82, and a second water outlet end 83. The refrigeration device 6 is a chiller or a refrigeration unit.

[0157] Server 10;

[0158] Among them, the outlet end of the cooling device 1, the first water inlet 21, the first water outlet 22, the liquid inlet 31, the first liquid outlet 32, and the inlet end of the cooling device 1 form a liquid cooling circuit, and the first water pump 4 is provided in the liquid cooling circuit; the coolant outlet of the server 10, the second water inlet 23, the second water outlet 24, and the coolant inlet of the server 10 form a heat exchange circuit, and the second water pump 5 is provided in the heat exchange circuit. The fluid in the heat exchange circuit is used to perform heat exchange with the fluid in the liquid cooling circuit in the heat exchanger 2;

[0159] The liquid outlet of the refrigeration device 6, the water inlet 81, the first water outlet 82, the liquid cooling inlet of the fan 7, the liquid cooling outlet of the fan 7, and the liquid inlet of the refrigeration device 6 form an air cooling circuit. The third water pump 9 is provided in the air cooling circuit. The fan 7 is used to blow cold air to the server 10.

[0160] The air cooling circuit is connected to the liquid cooling circuit through the second water outlet 83 , and the liquid cooling circuit is connected to the air cooling circuit through the second liquid outlet 33 .

[0161] For example, when the operating mode is the first operating mode, the cooling device 1 is turned on, the refrigeration device 6 is turned off, the valve of the second water outlet 83 and the valve of the second liquid outlet 33 are both opened, and the valve of the water inlet 81 is closed, so that the cooling device 1 supplies liquid to the liquid cooling circuit and the air cooling circuit at the same time;

[0162] When the operating mode is the second operating mode, the cooling device 1 and the refrigeration device 6 are both turned on, the valve of the second water outlet end 83 and the valve of the second liquid outlet 33 are both closed, and the valve of the water inlet end 81 is open, so that the cooling device 1 supplies liquid to the liquid cooling circuit and the refrigeration device 6 supplies liquid to the air cooling circuit.

[0163] When the operating mode is the third operating mode, the cooling device 1 and the refrigeration device 6 are both turned on, and the valve of the second water outlet end 83, the valve of the second liquid outlet 33 and the valve of the water inlet end 81 are all opened, so that the cooling device 1 and the refrigeration device 6 supply liquid to the air cooling circuit and the liquid cooling circuit at the same time.

[0164] In a possible implementation, the thermal management control system 100 further includes a filter 11 , and each of the air cooling circuit, the liquid cooling circuit, and the heat exchange circuit is provided with at least one filter 11 .

[0165] The present application proposes a thermal management control system 100 for a data center. In the thermal management control method for a data center, the technical solution of the embodiment of the present application is to configure the thermal management control system 100 to include a liquid cooling system and an air cooling system. The liquid cooling system includes a cooling device 1, a heat exchanger 2, a first three-way valve 3, a first water pump 4, and a second water pump 5. The heat exchanger 2 includes a first pipeline (not shown) and a second pipeline (not shown). The first pipeline is provided with a first water inlet 21 and a first water outlet 22. The second pipeline is provided with a second water inlet 23 and a second water outlet 24. The first three-way valve 3 includes a liquid inlet 31, a first liquid outlet 32, and a second liquid outlet 33. The cooling device 1 is a cooling tower or an air cooler. The air cooling system includes a refrigeration device 6, a fan 7, a second three-way valve 8, and a third water pump 9. The second three-way valve 8 includes a water inlet 81, a first water outlet 82, and a second water outlet 83. The refrigeration device 6 is a chiller or a refrigeration unit.

[0166] The outlet of cooling device 1, first water inlet 21, first water outlet 22, liquid inlet 31, first liquid outlet 32, and the inlet of cooling device 1 form a liquid cooling circuit, in which first water pump 4 is provided. The coolant outlet of server 10, second water inlet 23, second water outlet 24, and the coolant inlet of server 10 form a heat exchange circuit, in which second water pump 5 is provided. The fluid in the heat exchange circuit is used to exchange heat with the fluid in the liquid cooling circuit in heat exchanger 2. The liquid outlet, water inlet 81, first water outlet 82, liquid cooling inlet of fan 7, liquid cooling outlet of fan 7, and the liquid inlet of cooling device 6 form an air cooling circuit. Third water pump 9 is provided in the air cooling circuit, and fan 7 is used to blow cold air to server 10. The air-cooling circuit is connected to the liquid-cooling circuit through the second water outlet 83, and the liquid-cooling circuit is connected to the air-cooling circuit through the second liquid outlet 33, so that in the low-temperature mode (i.e., the first operating mode), the refrigeration device 6 is turned off, and the cooling device 1 supplies liquid to the liquid-cooled side and the air-cooled side. In the normal temperature mode (i.e., the second operating mode), the refrigeration device 6 supplies liquid for air cooling, and the cooling device 1 supplies liquid for liquid cooling. In the high-temperature mode (i.e., the third operating mode), the refrigeration device 6 supplies cold water to the liquid-cooled side through the three-way valve, thereby realizing real-time changes in the working state of the terminal equipment of the thermal management system of the data center, reducing the energy consumption of the thermal management system while meeting the temperature requirements of the server 10 in the computer room, realizing real-time energy-saving control function, facilitating real-time optimization and calculation of the optimal control parameters and optimal operating mode of the thermal management control system for the data center, so as to determine the thermal management control strategy for the server 10 in the data center, thereby realizing high-efficiency and high-accuracy thermal management control of the data center, and improving the stability and reliability of the thermal management control of the data center.

[0167] The thermal management control system 100 of the data center provided in this embodiment and the thermal management control method of the data center provided in the above-mentioned embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be found in the embodiment of the thermal management control method of the data center. This embodiment has the same beneficial effects as the various embodiments of the thermal management control method of the data center, and will not be repeated here.

[0168] The system embodiment described above is merely illustrative. The units described as separate components may or may not be physically separate, i.e., they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0169] In addition, the present invention also provides an electronic device, referring to Figure 16 , Figure 16 This is a hardware structure diagram of an electronic device provided in an embodiment of the present application. Figure 16 As shown, the electronic device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the user interface 1003 may optionally include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (WIreless-FIdelity, WI-FI) interface). The memory 1005 may be a high-speed random access memory (Random Access Memory, RAM) memory, or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0170] Those skilled in the art will understand that Figure 16 The structure shown in the figure does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange components differently. Figure 16 As shown, the memory 1005 as a storage medium may include an operating system, a data storage module, a network communication module, a user interface module, and a thermal management control program for a data center.

[0171] exist Figure 16In the electronic device shown, the network interface 1004 is mainly used for data communication with other devices; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in this embodiment can be set in the communication device, and the communication device calls the thermal management control program of the data center stored in the memory 1005 through the processor 1001, and executes the thermal management control method applied to the data center provided in any of the above embodiments.

[0172] The terminal proposed in this embodiment and the thermal management control method applied to the data center proposed in the above embodiment belong to the same inventive concept. The technical details not described in detail in this embodiment can be referred to any of the above embodiments, and this embodiment has the same beneficial effects as executing the thermal management control method of the data center.

[0173] In addition, an embodiment of the present application also proposes a storage medium, which is a computer storage medium. The computer-readable storage medium can be a non-volatile computer-readable storage medium. The computer-readable storage medium stores a thermal management control program of a data center. When the thermal management control program of the data center is executed by a processor, the thermal management control method of the data center of the present application as described above is implemented.

[0174] The various embodiments of the electronic device and computer-readable storage medium of the present application may refer to the various embodiments of the thermal management control method for the data center of the present application, and will not be repeated here.

[0175] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or system comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or system. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or system comprising the element.

[0176] The serial numbers of the above embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0177] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for enabling an electronic device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0178] The above are only preferred embodiments of the present application and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A thermal management control method for a data center, characterized in that: include: Acquire first data, where the first data represents multiple actual operating conditions of the data center, and control parameters and operating modes under each actual operating condition; Based on a pre-trained mechanism model, the first data is supplemented to obtain second data, and an energy consumption prediction data model is trained based on the second data to obtain a trained energy consumption prediction data model; Based on the trained energy consumption prediction data model, target control parameters and target operating modes for the data center are predicted, and thermal management control of the data center is performed according to the target control parameters and target operating modes.

2. The thermal management control method for a data center according to claim 1, wherein: Based on the trained energy consumption prediction data model, the target control parameters and target operating mode for the data center are predicted, including: Based on the trained energy consumption prediction data model, the optimal control parameters and optimal operating mode for each operating condition of the data center are predicted; Storing the optimal control parameters and optimal operating modes corresponding to each of the operating conditions in a knowledge base; The real-time operating conditions of the data center are dynamically acquired, and the target control parameters and target operating mode corresponding to the real-time operating conditions are queried and retrieved from the knowledge base.

3. The thermal management control method for a data center according to claim 2, wherein: The method further comprises: When the data center is in a risky operating condition, a safety regulation strategy corresponding to the risky operating condition is queried and called from the knowledge base, wherein the safety regulation strategy includes control parameters and operating modes for safety regulation of the risky operating condition.

4. The thermal management control method for a data center according to claim 2, wherein: The energy consumption prediction data model is trained and stored in a cloud server, and the knowledge base is located locally in the user terminal.

5. The thermal management control method for a data center according to claim 1, wherein: Obtaining first data, including: Collecting control parameters and operating modes under various actual operating conditions of the data center to obtain third data; The third data is processed based on a preset data processing algorithm to obtain the first data, wherein the preset data processing algorithm is at least one of a Latin hypercube sampling method, a wavelet threshold denoising method, an empirical mode decomposition denoising method, a k-nearest neighbor outlier removal method, a Mahalanobis distance outlier removal method, a linear interpolation data repair method, an expectation maximization numerical repair method, a Latin hypercube sampling algorithm interval partitioning method, and a de-redundant data reduction method.

6. The thermal management control method for a data center according to claim 1, wherein: Training an energy consumption prediction data model based on the second data to obtain a trained energy consumption prediction data model includes: constructing a sample database according to the second data; Based on the principal component analysis method, feature selection is performed on the sample database to obtain independent variable parameters that affect the thermal management control target; An energy consumption prediction data model is trained according to independent variable parameters that affect thermal management control targets to obtain a trained energy consumption prediction data model.

7. The thermal management control method for a data center according to claim 1, wherein: Performing thermal management control on the data center according to the target control parameter and the target operation mode includes: Controlling the operating parameters of the underlying components to be controlled of the data center according to the target control parameters and the target operating mode, so as to perform thermal management control on the data center; The bottom-level components to be controlled include fans, water pumps, compressors and valves, and the operating parameters of the bottom-level components to be controlled include fan frequency, water pump frequency, compressor frequency and valve opening.

8. The thermal management control method for a data center according to claim 7, wherein: The pre-trained mechanism model is a system-level dual-drive mechanism model, and the method further includes: Obtain the trained component data models corresponding to each underlying component to be controlled from the preset component model library; Based on the component data models, a system-level mechanism model of the data center is constructed to obtain the system-level dual-drive mechanism model.

9. The thermal management control method for a data center according to claim 8, wherein: The system-level mechanism model of the data center is constructed based on the component data models to obtain the system-level dual-drive mechanism model, including: Based on the component data models, the system-level dual-drive mechanism model is constructed in a first preset building block stacking manner.

10. The thermal management control method for a data center according to claim 7, wherein: The pre-trained mechanism model is a device-level dual-drive mechanism model, wherein the method further includes: Obtain the trained component mechanism models corresponding to each underlying component to be controlled from the preset component model library; Based on the mechanism models of each component, a system-level mechanism model of the data center is constructed to obtain the device-level dual-drive mechanism model.

11. The thermal management control method for a data center according to claim 10, wherein: Based on the mechanism models of each component, a system-level mechanism model of the data center is constructed to obtain the device-level dual-drive mechanism model, including: Based on the mechanism models of each component, the device-level dual-drive mechanism model is constructed in a second preset building block stacking manner.

12. The thermal management control method for a data center as claimed in claims 7 to 11, characterized in that: The method further comprises: Collecting actual operation data generated by each of the underlying components to be controlled based on the target control parameters and the target operation mode; Constructing a model prediction value deviation model based on the actual operation data and the prediction data of the energy consumption prediction data model under the operating conditions corresponding to the actual operation data; The model prediction value deviation model is used to perform error correction and optimization on the energy consumption prediction data model.

13. The thermal management control method for a data center according to any one of claims 7 to 11, characterized in that: The types of operation modes include the first operation mode, the second operation mode and the third operation mode, wherein: The first operating mode is when the refrigeration equipment is turned off and the cooling equipment is in an operating mode where liquid is supplied to the liquid cooling side and the air cooling side, wherein the refrigeration equipment is a chiller or a refrigeration unit, and the cooling equipment is a cooling tower or an air cooler; The second operation mode is that the refrigeration device supplies liquid to the air-cooled side, and the cooling device supplies liquid to the liquid-cooled side; The third operation mode is that both the refrigeration device and the cooling device supply liquid on the air-cooled side and the liquid-cooled side.

14. A thermal management control system, characterized in that the thermal management control system is applied to the thermal management control method for a data center according to any one of claims 1 to 13, the thermal management control system comprising: A liquid cooling system comprising a cooling device, a heat exchanger, a first three-way valve, a first water pump, and a second water pump; the heat exchanger comprising a first pipeline and a second pipeline; the first pipeline having a first water inlet and a first water outlet; the second pipeline having a second water inlet and a second water outlet; the first three-way valve comprising a liquid inlet, a first liquid outlet, and a second liquid outlet; and the cooling device being a cooling tower or an air cooler; An air cooling system includes a refrigeration device, a fan, a second three-way valve, and a third water pump, wherein the second three-way valve includes a water inlet, a first water outlet, and a second water outlet, and the refrigeration device is a chiller or a refrigeration unit; server; The outlet end of the cooling device, the first water inlet, the first water outlet, the liquid inlet, the first liquid outlet, and the inlet end of the cooling device form a liquid cooling circuit, and the first water pump is provided in the liquid cooling circuit; the coolant outlet of the server, the second water inlet, the second water outlet, and the coolant inlet of the server form a heat exchange circuit, and the second water pump is provided in the heat exchange circuit. The fluid in the heat exchange circuit is used to perform heat exchange with the fluid in the liquid cooling circuit in the heat exchanger; The liquid outlet of the refrigeration device, the water inlet, the first water outlet, the liquid cooling inlet of the fan, the liquid cooling outlet of the fan, and the liquid inlet of the refrigeration device form an air cooling circuit, the third water pump is provided in the air cooling circuit, and the fan is used to blow cold air to the server; The air cooling circuit is connected to the liquid cooling circuit through the second water outlet, and the liquid cooling circuit is connected to the air cooling circuit through the second liquid outlet.

15. The thermal management control system according to claim 14, wherein: include: When the operating mode is the first operating mode, the cooling device is turned on, the refrigeration device is turned off, the valve at the second water outlet and the valve at the second liquid outlet are both opened, and the valve at the water inlet is closed, so that the cooling device supplies liquid to the liquid cooling circuit and the air cooling circuit at the same time; When the operating mode is the second operating mode, the cooling device and the refrigeration device are both turned on, the valve at the second water outlet and the valve at the second liquid outlet are both closed, and the valve at the water inlet is open, so that the cooling device supplies liquid to the liquid cooling circuit and the refrigeration device supplies liquid to the air cooling circuit.

16. The thermal management control system according to claim 14, wherein: include: When the operating mode is the third operating mode, the cooling device and the refrigeration device are both turned on, and the valve at the second water outlet, the valve at the second liquid outlet and the valve at the water inlet are all opened, so that the cooling device and the refrigeration device supply liquid to the air cooling circuit and the liquid cooling circuit at the same time.

17. An electronic device, characterized in that: include: A memory, a processor, and a thermal management control program for a data center stored in the memory and executable on the processor, wherein the thermal management control program for the data center, when executed by the processor, implements the thermal management control method for the data center according to any one of claims 1 to 13.

18. A storage medium, characterized in that The storage medium is a computer-readable storage medium, which stores a thermal management control program for a data center. When the thermal management control program for the data center is executed by a processor, the thermal management control method for the data center according to any one of claims 1 to 13 is implemented.

Images