WO2024016586A1 - 机房温度控制方法及装置、电子设备、存储介质 - Google Patents

机房温度控制方法及装置、电子设备、存储介质 Download PDF

Info

Publication number
WO2024016586A1
WO2024016586A1 PCT/CN2022/140624 CN2022140624W WO2024016586A1 WO 2024016586 A1 WO2024016586 A1 WO 2024016586A1 CN 2022140624 W CN2022140624 W CN 2022140624W WO 2024016586 A1 WO2024016586 A1 WO 2024016586A1
Authority
WO
WIPO (PCT)
Prior art keywords
temperature control
computer room
decision
parameters
temperature
Prior art date
Application number
PCT/CN2022/140624
Other languages
English (en)
French (fr)
Inventor
任宏丹
曾宇
王涛
Original Assignee
中国电信股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中国电信股份有限公司 filed Critical 中国电信股份有限公司
Publication of WO2024016586A1 publication Critical patent/WO2024016586A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D23/00—Control of temperature
    • G05D23/19—Control of temperature characterised by the use of electric means
    • G05D23/30—Automatic controllers with an auxiliary heating device affecting the sensing element, e.g. for anticipating change of temperature
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/50—Air quality properties
    • F24F2110/64—Airborne particle content

Definitions

  • the present disclosure relates to the field of temperature control, and specifically to methods, devices, equipment and storage media for temperature control in computer rooms.
  • the existing smart computer room temperature control strategy needs to be manually combined with IT (Internet Technology) load (power consumption of data center IT equipment) changes, ambient temperature, and indoor and outdoor temperature conditions to determine whether it needs to be reformulated during the offline update strategy stage.
  • IT Internet Technology
  • the strategy is updated iteratively; in the automatic control stage, the strategy formulation and updates are executed at a fixed frequency, and then handed over to the operation and maintenance personnel to judge whether to execute the strategy; in this stage, there are a large number of repeated strategies that do not need to be executed, and the strategy update is not considered in combination with other indicators. .
  • the purpose of this disclosure is to provide a computer room temperature control method, device, equipment and storage medium.
  • Embodiments of the present disclosure provide a computer room temperature control method, including:
  • the current temperature control strategy is used to control the computer room temperature.
  • calculating the comprehensive characteristic index based on the temperature parameter and the machine room load parameter includes:
  • the decision model is trained based on historical temperature control strategies, historical temperature parameters, computer room load parameters, and decision labels.
  • the decision labels include that the current temperature control strategy does not require iteration and that the current temperature control strategy requires Iterate.
  • the decision of the current temperature control strategy in response to the decision model requires iteration, and generating an updated temperature control strategy based on the output of the temperature prediction model includes:
  • the comprehensive characteristic index is input into the temperature prediction model.
  • the temperature prediction model adopts a Transformer model, wherein the self-attention layer of the Transformer model uses convolution kernels of different sizes to perform convolution operations during the calculation of Query, Key, and Value.
  • the convolution kernel is obtained by solving the particle swarm algorithm, and the Transformer model will also stack self-attention layers within a set period.
  • the temperature prediction model includes six serially connected encoders and six serially connected decoders, wherein the outputs of the last serially connected encoder are respectively output to six decoders. .
  • the temperature parameter includes one or more of the computer room temperature, cabinet temperature, weather temperature, and control temperature of the computer room temperature control system.
  • a machine room temperature control device including:
  • the parameter collection module is configured to collect temperature parameters and computer room load parameters
  • the comprehensive characteristic index module is configured to calculate the comprehensive characteristic index according to the temperature parameter and the computer room load parameter;
  • a decision input module configured to input the current temperature control strategy and the comprehensive characteristic index into the decision model
  • Decision output module configured as:
  • the current temperature control strategy is used to control the computer room temperature.
  • a machine room temperature control processing device including:
  • a memory in which executable instructions of the processor are stored
  • the processor is configured to execute the steps of the computer room temperature control method as described above by executing the executable instructions.
  • Embodiments of the present disclosure also provide a computer-readable storage medium for storing a program that implements the steps of the above computer room temperature control method when the program is executed.
  • Figure 1 is a flow chart of an embodiment of the computer room temperature control method of the present disclosure.
  • Figure 2 is a flow chart of another embodiment of the computer room temperature control method of the present disclosure.
  • Figure 3 is a flow chart of system execution of the computer room temperature control method of the present disclosure.
  • Figure 4 is a flow chart of another embodiment of the computer room temperature control method of the present disclosure.
  • Figure 5 is a schematic diagram of the temperature prediction model of the present disclosure.
  • Figure 6 is a module diagram of an embodiment of the computer room temperature control device of the present disclosure.
  • Figure 7 is a module diagram of another embodiment of the computer room temperature control device of the present disclosure.
  • Figure 8 is a schematic structural diagram of the computer room temperature control device of the present disclosure.
  • Figure 9 is a schematic structural diagram of a computer-readable storage medium according to an embodiment of the present disclosure.
  • Example embodiments will now be described more fully with reference to the accompanying drawings.
  • Example embodiments may, however, be embodied in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of example embodiments to those skilled in the art.
  • the same reference numerals in the drawings represent the same or similar structures, and thus their repeated description will be omitted.
  • Figure 1 is a flow chart of an embodiment of the computer room temperature control method applied to the calling terminal of the present disclosure.
  • An embodiment of the present disclosure provides a computer room temperature control method, which includes the following steps:
  • Step S110 Collect temperature parameters and machine room load parameters.
  • the temperature parameter may include one or more of the computer room temperature, cabinet temperature, weather temperature, and control temperature of the computer room temperature control system.
  • Step S120 Calculate the comprehensive characteristic index according to the temperature parameter and the machine room load parameter.
  • Step S130 Input the current temperature control strategy and the comprehensive characteristic index into the decision model.
  • the decision model can be trained based on the historical temperature control strategy, historical temperature parameters, computer room load parameters, and decision labels.
  • the decision labels include that the current temperature control strategy does not require iteration and that the current temperature control strategy requires iteration. Therefore, the decision-making model provided by the present disclosure is a supervised decision-making model.
  • the data set of historical temperature control strategies, historical temperature parameters, computer room load parameters and decision labels can be divided into training sets and test sets.
  • the training set is used to train the decision model
  • the test set is used to test the trained decision model. test.
  • Step S140 In response to the decision model deciding that the current temperature control strategy requires iteration, generate an updated temperature control strategy based on the output of the temperature prediction model.
  • the temperature parameters and machine room load parameters can be input into the temperature prediction model to ensure multi-maintenance and integrity of the input data of the model.
  • the comprehensive characteristic index can also be input into the temperature prediction model to achieve reuse of the comprehensive characteristic index.
  • the temperature prediction model predicts and outputs temperature parameters in the computer room.
  • the required output control temperature of the computer room temperature control system (such as the refrigeration system) in the future set time period is determined.
  • the temperature prediction model prediction may also directly output the control temperature of the computer room temperature control system.
  • Step S150 In response to the decision model deciding that the current temperature control strategy does not require iteration, use the current temperature control strategy to control the computer room temperature.
  • the temperature control strategy can give the temperature target value of the temperature control system and the upper and lower limits of the set value.
  • the computer room refrigeration control system can intelligently make temperature control strategy iterations based on changes in multi-dimensional temperature parameters and computer room load. decision-making, thereby reducing manual intervention in decision-making, reducing the number of temperature control strategy iterations, and improving the energy-saving efficiency of artificial intelligence algorithms for computer room cooling.
  • FIG. 2 is a flow chart of another embodiment of the computer room temperature control method of the present disclosure.
  • Another embodiment of a computer room temperature control method includes the following steps:
  • Step S201 Collect temperature parameters and computer room load parameters.
  • the temperature parameter may include one or more of the computer room temperature, cabinet temperature, weather temperature, and control temperature of the computer room temperature control system.
  • Step S202 Calculate the correlation between the temperature parameter and the machine room load parameter.
  • Step S203 Keep parameters that are irrelevant or weakly correlated with other parameters.
  • Step S204 Generate a comprehensive feature index based on the retained parameters.
  • PCA principal component analysis
  • Step S205 Input the current temperature control strategy and the comprehensive characteristic index into the decision-making model.
  • the decision model can be trained based on the historical temperature control strategy, historical temperature parameters, computer room load parameters, and decision labels.
  • the decision labels include that the current temperature control strategy does not require iteration and that the current temperature control strategy requires iteration. Therefore, the decision-making model provided by the present disclosure is a supervised decision-making model.
  • the data set of historical temperature control strategies, historical temperature parameters, computer room load parameters and decision labels can be divided into training sets and test sets.
  • the training set is used to train the decision model
  • the test set is used to test the trained decision model. test.
  • Step S206 In response to the decision model deciding that the current temperature control strategy requires iteration, generate an updated temperature control strategy based on the output of the temperature prediction model.
  • the temperature parameters and machine room load parameters can be input into the temperature prediction model to ensure multi-maintenance and integrity of the input data of the model.
  • the comprehensive characteristic index can also be input into the temperature prediction model to achieve reuse of the comprehensive characteristic index.
  • the temperature prediction model predicts and outputs temperature parameters in the computer room.
  • the required output control temperature of the computer room temperature control system (such as the refrigeration system) in the future set time period is determined.
  • the temperature prediction model prediction may also directly output the control temperature of the computer room temperature control system.
  • Step S207 In response to the decision model deciding that the current temperature control strategy does not require iteration, use the current temperature control strategy to control the computer room temperature.
  • Figure 3 is a flow chart of system execution of the computer room temperature control method of the present disclosure.
  • the temperature control system 300 mainly performs three branches of calculations.
  • step S311 can be performed: collect and store the computer room temperature, cabinet temperature, cooling output, and weather temperature required for the computer room energy-saving algorithm.
  • step S312 Collect and store computer room IT load and other data.
  • Step S313 Preprocess the collected data into a standard format.
  • step S321 perform correlation calculation based on the preprocessed data
  • step S322 retain irrelevant or weakly relevant parameters
  • step S323 calculate the comprehensive decision-making index of the algorithm iteration based on the retained parameters.
  • step S331 Based on the calculation results of the comprehensive decision-making index, intelligently perform the decision-making behavior of whether to iterate the computer room temperature energy-saving control algorithm; step S332: If iteration is required, use the temperature prediction model to perform temperature prediction. ; Step S333: Control the computer room refrigeration system based on the predicted computer room temperature.
  • the present disclosure involves weather temperature and computer room IT load data in the calculation and evaluation of decision-making indicators, and can provide policy update suggestions based on changes, improve the level of automation, and reduce repetitive work.
  • FIG. 4 is a flow chart of yet another embodiment of the computer room temperature control method of the present disclosure.
  • data such as temperature and IT load change data 401, computer room temperature 402, cabinet temperature 403, and computer room historical control strategy 404 are constructed into a vector set, and principal component analysis is used in step S410 ( PCA) performs dimensionality reduction on the data and extracts a set of comprehensive indicators that are unrelated to each other.
  • PCA principal component analysis
  • step S410 PCA
  • the principal component analysis algorithm is introduced to perform dimensionality reduction processing on the original data, while minimizing the loss of information contained in the original indicators, and at the same time achieving the purpose of comprehensive analysis of the collected temperature and load data.
  • Step S420 The calculation results based on the comprehensive indicators are input into the decision-making model for intelligent decision-making.
  • the decision-making model can perform supervised learning based on artificial experience to improve the accuracy of intelligent decision-making judgment.
  • the temperature prediction model is improved from the RNN/LSTM algorithm used in traditional solutions to the Transformer algorithm based on the attention model, which greatly improves the efficiency of parallel computing when the algorithm is running, saves computing resources, and improves prediction accuracy.
  • convolution kernels of different sizes are introduced for convolution operations.
  • the size of the convolution kernel is >1.
  • the calculation of the convolution kernel can introduce the particle swarm algorithm to find the optimal solution.
  • the algorithm goal of the particle swarm algorithm is to find the fastest fitting convolution kernel size setting. Therefore, compared with the original self-attention algorithm's Query, Key, and Value calculations that only reflect the correlation between single time points, the improved Query, Key, and Value calculation can reflect more dimensional correlations.
  • the present disclosure can also analyze the time-varying attention score distribution of different layers, thereby selecting specific self-attention layers within a certain period for stacking through expert experience settings, thereby reducing the space complexity of each layer to O( Llog 2 L).
  • L is the sequence length.
  • I k l is the index set of units to be accessed when unit l is calculated at the k to k+1 layers.
  • its space complexity is O(L*2), while in this disclosure it is reduced After the space complexity of each layer, the overall space complexity of the model is reduced to O(L(log 2 L) 2 ).
  • FIG. 5 is a schematic diagram of the temperature prediction model of the present disclosure.
  • the temperature prediction model includes six serially connected encoders 521 to 526 and six serially connected decoders 531 to 536, wherein the output of the last serially connected encoder 526 is respectively output to the six decoders 531 to 536.
  • the input features are converted into word vectors 510 through the embedding algorithm, and then input to the first encoder 521.
  • the six encoders 521 to 526 may have the same structure, as shown in the encoder 521, which first multiplies the word vector input therein by the trained weight matrix, and then calculates the query matrix, key matrix and value matrix; by query vector and The vector point integral is used to calculate the score.
  • the score is divided by 8 and then softmax normalized to obtain the weight value. Then all value vectors are weighted and summed, and finally input into the feedforward neural network.
  • the six decoders may have the same structure, and the structure of each decoder is as shown in decoder 536, which includes a feedforward neural network, an encoding-decoding attention layer, and a self-attention layer.
  • data such as temperature and load are constructed into a vector set, principal component analysis (PCA) is used to perform dimensionality reduction processing on the data, and a set of comprehensive indicators that are unrelated to each other are extracted to calculate decision-making indicators;
  • PCA principal component analysis
  • supervised learning is carried out on the basis of artificial experience to improve the accuracy of intelligent judgment in decision-making, thereby realizing intelligent judgment in decision-making;
  • the temperature prediction model is also improved from the RNN/LSTM algorithm used in traditional solutions to one based on
  • the Transformer algorithm of the attention model greatly improves the efficiency of parallel computing when the algorithm is running, saves computing resources, and improves prediction accuracy.
  • Figure 6 is a module diagram of an embodiment of the computer room temperature control device of the present disclosure.
  • the computer room temperature control device 600 of the present disclosure includes but is not limited to: a parameter collection module 610, a comprehensive feature index module 620, a decision input module 630, and a decision output module 640.
  • the parameter collection module 610 is configured to collect temperature parameters and computer room load parameters
  • the comprehensive characteristic index module 620 is configured to calculate the comprehensive characteristic index according to the temperature parameter and the computer room load parameter;
  • the decision input module 630 is configured to input the current temperature control strategy and the comprehensive characteristic index into the decision model
  • the decision output module 640 is configured to:
  • the current temperature control strategy is used to control the computer room temperature.
  • Figure 7 is a schematic module diagram of another embodiment of the computer room temperature control device of the present disclosure.
  • the computer room temperature control device 700 of the present disclosure includes but is not limited to: parameter collection module 710, correlation calculation module 720, screening module 730, comprehensive characteristic index calculation module 740, decision input module 750 and decision output module 760.
  • the parameter collection module 710 is configured to collect temperature parameters and computer room load parameters
  • the correlation calculation module 720 is configured to calculate the correlation between the temperature parameter and the computer room load parameter
  • the filtering module 730 is configured to retain parameters that are unrelated or weakly related to other parameters
  • the comprehensive feature index calculation module 740 is configured to generate a comprehensive feature index according to the retained parameters.
  • the decision input module 750 is configured to input the current temperature control strategy and the comprehensive characteristic index into the decision model
  • the decision output module 760 is configured to:
  • the current temperature control strategy is used to control the computer room temperature.
  • the disclosed computer room temperature control device participates in the calculation and evaluation process of decision-making indicators by involving multi-dimensional temperature parameters and computer room IT load parameters, thereby enabling the computer room refrigeration control system to intelligently make temperature control strategies based on changes in multi-dimensional temperature parameters and computer room load. Iterate or not decide, thereby reducing manual intervention in decision-making, reducing the number of temperature control strategy iterations, and improving the energy-saving efficiency of artificial intelligence algorithms for computer room cooling.
  • FIGS 6 and 7 are only schematically showing the computer room temperature control devices 600 and 700 provided by the present disclosure. Without violating the concept of the present disclosure, the splitting, merging, and addition of modules are all within the scope of the present disclosure. within.
  • the computer room temperature control devices 600 and 700 provided by the present disclosure can be implemented by software, hardware, firmware, plug-ins, and any combination thereof, and the present disclosure is not limited thereto.
  • An embodiment of the present disclosure also provides a computer room temperature control processing device, including a processor.
  • Memory which stores the executable instructions of the processor.
  • the processor is configured to perform the steps of the computer room temperature control method by executing executable instructions.
  • the computer room temperature control processing equipment of this embodiment of the present disclosure participates in the calculation and evaluation process of decision-making indicators by involving multi-dimensional temperature parameters and computer room IT load parameters, thereby enabling the computer room refrigeration control system to intelligently adjust the temperature according to the multi-dimensional temperature parameters.
  • Changes in parameters and computer room load make decisions about whether to iterate the temperature control strategy, thereby reducing manual intervention in decision-making, reducing the number of iterations of the temperature control strategy, and improving the energy-saving efficiency of artificial intelligence algorithms for computer room cooling.
  • FIG. 8 is a schematic structural diagram of the computer room temperature control processing equipment of the present disclosure.
  • An electronic device 700 according to this embodiment of the present disclosure is described below with reference to FIG. 8 .
  • the electronic device 800 shown in FIG. 8 is only an example and should not bring any limitations to the functions and usage scope of the embodiments of the present disclosure.
  • electronic device 800 is embodied in the form of a general computing device.
  • the components of the electronic device 800 may include, but are not limited to: at least one processing unit 810, at least one storage unit 820, a bus 830 connecting different platform components (including the storage unit 820 and the processing unit 810), a display unit 840, and the like.
  • the storage unit stores program code, and the program code can be executed by the processing unit 810, so that the processing unit 810 performs the steps according to various exemplary embodiments of the present disclosure described in the above-mentioned machine room temperature control method section of this specification.
  • processing unit 810 may perform steps as shown in FIG. 1 .
  • the storage unit 820 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 8201 and/or a cache storage unit 8202, and may further include a read-only storage unit (ROM) 8203.
  • RAM random access storage unit
  • ROM read-only storage unit
  • Storage unit 820 may also include a program/utility 8204 having a set of (at least one) program modules 8205 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, Each of these examples, or some combination, may include the implementation of a network environment.
  • program/utility 8204 having a set of (at least one) program modules 8205 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, Each of these examples, or some combination, may include the implementation of a network environment.
  • Bus 830 may be a local area representing one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or using any of a variety of bus structures. bus.
  • Electronic device 800 may also communicate with one or more external devices 8001 (e.g., keyboard, pointing device, Bluetooth device, etc.), may also communicate with one or more devices that enable a user to interact with electronic device 800, and/or with Any device that enables the electronic device 800 to communicate with one or more other computing devices (eg, router, modem, etc.). This communication may occur through an input/output (I/O) interface 850.
  • the electronic device 800 may also communicate with one or more networks (eg, a local area network (LAN), a wide area network (WAN), and/or a public network, such as the Internet) through a network adapter 860.
  • Network adapter 860 may communicate with other modules of electronic device 800 via bus 830.
  • electronic device 800 may be used in conjunction with electronic device 800, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives And data backup storage platform, etc.
  • Embodiments of the present disclosure also provide a computer-readable storage medium for storing a program and steps of a computer room temperature control method implemented when the program is executed.
  • various aspects of the present disclosure can also be implemented in the form of a program product, which includes program code.
  • the program product is run on a terminal device, the program code is used to cause the terminal device to execute the above described instructions.
  • the steps according to various exemplary embodiments of the present disclosure are described in the Machine Room Temperature Control Method section.
  • the computer-readable storage medium used to perform computer room temperature control in this embodiment participates in the calculation and evaluation process of decision-making indicators by involving multi-dimensional temperature parameters and computer room IT load parameters, thereby enabling the computer room refrigeration control system to intelligently adjust according to the Changes in multi-dimensional temperature parameters and computer room loads make decisions about whether to iterate the temperature control strategy, thereby reducing manual intervention in decision-making, reducing the number of iterations of the temperature control strategy, and improving the energy-saving efficiency of artificial intelligence algorithms for computer room cooling.
  • Figure 9 is a schematic structural diagram of a computer-readable storage medium of the present disclosure.
  • a program product 900 for implementing the above method according to an embodiment of the present disclosure is described, which can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be used on a terminal device, For example, run on a personal computer.
  • CD-ROM portable compact disk read-only memory
  • the program product of the present disclosure is not limited thereto.
  • a readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.
  • the Program Product may take the form of one or more readable media in any combination.
  • the readable medium may be a readable signal medium or a readable storage medium.
  • the readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: electrical connection with one or more conductors, portable disk, hard disk, 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 device, magnetic storage device, or any suitable combination of the above.
  • a computer-readable storage medium may include a data signal propagated in baseband or as part of a carrier wave carrying the readable program code therein. Such propagated data signals may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above.
  • a readable storage medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transport the program for use by or in connection with an instruction execution system, apparatus, or device.
  • Program code contained on a readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above.
  • Program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, C++, etc., as well as conventional procedural programming. Language—such as "C” or a similar programming language.
  • the program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server execute on.
  • the remote computing device may be connected to the user computing device through any kind of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device, such as provided by an Internet service. (business comes via Internet connection).
  • LAN local area network
  • WAN wide area network
  • the present disclosure involves the multi-dimensional temperature parameters and computer room IT load parameters in the calculation and evaluation process of decision-making indicators, thereby enabling the computer room refrigeration control system to intelligently make temperature control strategy iterations and adjustments based on changes in multi-dimensional temperature parameters and computer room load. decision-making, thereby reducing manual intervention in decision-making, reducing the number of temperature control strategy iterations, and improving the energy-saving efficiency of artificial intelligence algorithms for computer room cooling.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Automation & Control Theory (AREA)
  • Feedback Control In General (AREA)

Abstract

一种机房温度控制方法、装置、设备及存储介质，该方法包括:采集温度参数以及机房负载参数（S110）；根据温度参数以及机房负载参数，计算综合特征指标（S120）；将当前温度控制策略以及综合特征指标输入决策模型（S130）；响应于决策模型决策当前温度控制策略需要迭代，基于温度预测模型的输出，生成更新的温度控制策略（S140）；响应于决策模型决策当前温度控制策略不需要迭代，使用当前温度控制策略控制机房温度（S150）。

Description

机房温度控制方法及装置、电子设备、存储介质
相关申请的交叉引用
本公开要求于2022年07月18日提交的申请号为202210845348.5、名称为“机房温度控制方法、装置、设备及存储介质”的中国专利申请的优先权，该中国专利申请的全部内容通过引用全部并入本文。
技术领域
本公开涉及温度控制领域，具体地说，涉及机房温度控制方法、装置、设备及存储介质。
背景技术
目前，机房智慧节能场景下构建基于大数据、机器学习的数据中心能耗优化模型极为重要。现有的智能机房温度控制策略，在离线更新策略阶段，需人工结合IT(internet Technology，互联网技术)负载(数据中心IT设备耗电量)变化、环境温度及室内外温度情况判断是否需要重新制定策略并进行迭代更新；在自动控制阶段，策略制定更新按照固定频率执行，之后交由运维人员判断是否执行该策略；该阶段存在大量重复、不需执行的策略，没有结合其他指标考虑策略更新。
由此，如何减少人工在温度控制决策中的干预，减少算法迭代次数，提高人工智能算法对机房制冷的节能效率，是本领域技术人员亟待解决的技术问题。
需要说明的是，上述背景技术部分公开的信息仅用于加强对本公开的背景的理解，因此可以包括不构成对本领域普通技术人员已知的现有技术的信息。
发明内容
针对现有技术中的问题，本公开的目的在于提供机房温度控制方法、装置、设备及存储介质。
本公开的实施例提供一种机房温度控制方法，包括：
采集温度参数以及机房负载参数；
根据所述温度参数以及机房负载参数，计算综合特征指标；
将当前温度控制策略以及所述综合特征指标输入决策模型；
响应于所述决策模型决策当前温度控制策略需要迭代，基于温度预测模型的输出，生成更新的温度控制策略；
响应于所述决策模型决策当前温度控制策略不需要迭代，使用当前温度控制策略控制机房温度。
在本公开的一些实施例中，所述根据所述温度参数以及机房负载参数，计算综合特征指标包括：
计算所述温度参数以及机房负载参数之间的相关性；
保留与其它参数无关或这弱相关的参数；
根据保留的参数生成综合特征指标。
在本公开的一些实施例中，所述决策模型基于历史温度控制策略、历史温度参数、机房负载参数以及决策标签进行训练，所述决策标签包括当前温度控制策略不需要迭代以及当前温度控制策略需要迭代。
在本公开的一些实施例中，所述响应于所述决策模型决策当前温度控制策略需要迭代，基于温度预测模型的输出，生成更新的温度控制策略包括：
将所述综合特征指标输入所述温度预测模型。
在本公开的一些实施例中，所述温度预测模型采用Transformer模型，其中，所述Transformer模型的自注意力层在计算Query，Key，Value的过程中采用不同大小的卷积核进行卷积操作，所述卷积核采用粒子群算法求解获得，所述Transformer模型还将设定周期内的自注意力层进行堆叠。
在本公开的一些实施例中，所述温度预测模型包括六个依次相连的编码器，以及六个依次相连的解码器，其中，依次相连的最后一个编码器的分别输出至六个解码器中。
在本公开的一些实施例中，所述温度参数包括机房温度、机柜温度、天气温度、机房温度控制系统的控制温度中的一项或多项。
根据本公开的又一方面，还提供一种机房温度控制装置，包括：
参数采集模块，配置成采集温度参数以及机房负载参数；
综合特征指标模块，配置成根据所述温度参数以及机房负载参数，计算综合特征指标；
决策输入模块，配置成将当前温度控制策略以及所述综合特征指标输入决策模型；
决策输出模块，配置成：
响应于所述决策模型决策当前温度控制策略需要迭代，基于温度预测模型的输出，生成更新的温度控制策略；
响应于所述决策模型决策当前温度控制策略不需要迭代，使用当前温度控制策略控制机房温度。
根据本公开的又一方面，还提供一种机房温度控制处理设备，包括：
处理器；
存储器，其中存储有所述处理器的可执行指令；
其中，所述处理器配置为经由执行所述可执行指令来执行如上所述机房温度控制方法的步骤。
本公开的实施例还提供一种计算机可读存储介质，用于存储程序，所述程序被执行时 实现上述机房温度控制方法的步骤。
附图说明
通过阅读参照以下附图对非限制性实施例所作的详细描述，本公开的其它特征、目的和优点将会变得更明显。
图1是本公开的机房温度控制方法的一种实施例的流程图。
图2是本公开的机房温度控制方法的另一种实施例的流程图。
图3是本公开的机房温度控制方法的系统执行的流程图。
图4是本公开的机房温度控制方法的又一种实施例的流程图。
图5是本公开的温度预测模型的示意图。
图6是本公开的机房温度控制装置的一种实施例的模块图。
图7是本公开的机房温度控制装置的另一种实施例的模块图。
图8是本公开的机房温度控制设备的结构示意图。
图9是本公开一实施例的计算机可读存储介质的结构示意图。
具体实施方式
现在将参考附图更全面地描述示例实施方式。然而，示例实施方式能够以多种形式实施，且不应被理解为限于在此阐述的实施方式。相反，提供这些实施方式使得本公开将全面和完整，并将示例实施方式的构思全面地传达给本领域的技术人员。在图中相同的附图标记表示相同或类似的结构，因而将省略对它们的重复描述。
参见图1，图1是本公开的应用于主叫终端的机房温度控制方法的一种实施例的流程图。本公开的实施例提供一种机房温度控制方法，包括以下步骤：
步骤S110：采集温度参数以及机房负载参数。
具体而言，温度参数可以包括机房温度、机柜温度、天气温度、机房温度控制系统的控制温度中的一项或多项。
步骤S120：根据所述温度参数以及机房负载参数，计算综合特征指标。
步骤S130：将当前温度控制策略以及所述综合特征指标输入决策模型。
具体而言，决策模型可以基于历史温度控制策略、历史温度参数、机房负载参数以及决策标签进行训练，所述决策标签包括当前温度控制策略不需要迭代以及当前温度控制策略需要迭代。由此，本公开提供的决策模型为有监督的决策模型。
进一步地，可以将历史温度控制策略、历史温度参数、机房负载参数以及决策标签的数据集划分为训练集和测试集，训练集用以训练决策模型，测试集用以对训练好的决策模型进行测试。
步骤S140：响应于所述决策模型决策当前温度控制策略需要迭代，基于温度预测模型的输出，生成更新的温度控制策略。
具体而言，可以将所述温度参数以及机房负载参数输入所述温度预测模型，以保证模型的输入数据的多维护化和完整性。在另一些实施例中，也可以将所述综合特征指标输入所述温度预测模型，以实现综合特征指标的复用。
具体而言，所述温度预测模型预测输出机房内的温度参数。由此，根据预测的未来设定时间段的机房内的温度参数，确定未来设定时间段的机房温度控制系统(诸如制冷系统)所需输出的控制温度。
在另一些实施例中，所述温度预测模型预测也可以直接输出机房温度控制系统的控制温度。
步骤S150：响应于所述决策模型决策当前温度控制策略不需要迭代，使用当前温度控制策略控制机房温度。
具体而言，温度控制策略可以给出温度控制系统温度目标值以及设定值的上下限。
由此，通过将多维温度参数及机房IT负载参数参与到决策指标的计算与评估过程，从而使得机房制冷控制系统能够智能地根据多维温度参数和机房负载的变化做出温度控制策略迭代与否的决策，从而减少了人工在决策中的干预，减少了温度控制策略迭代次数，提高了人工智能算法对机房制冷的节能效率。
下面参见图2，图2是本公开的机房温度控制方法的另一种实施例的流程图。另一实施例的机房温度控制方法，包括以下步骤：
步骤S201：采集温度参数以及机房负载参数。
具体而言，温度参数可以包括机房温度、机柜温度、天气温度、机房温度控制系统的控制温度中的一项或多项。
步骤S202：计算所述温度参数以及机房负载参数之间的相关性。
步骤S203：保留与其它参数无关或者弱相关的参数。
步骤S204：根据保留的参数生成综合特征指标。
具体而言，步骤S202至步骤S204可以利用主成分分析(PCA)对数据进行降维处理，提取出一组相互无关的综合指标。无关和弱相关可以根据相关性阈值进行筛选。
步骤S205：将当前温度控制策略以及所述综合特征指标输入决策模型。
具体而言，决策模型可以基于历史温度控制策略、历史温度参数、机房负载参数以及决策标签进行训练，所述决策标签包括当前温度控制策略不需要迭代以及当前温度控制策略需要迭代。由此，本公开提供的决策模型为有监督的决策模型。
进一步地，可以将历史温度控制策略、历史温度参数、机房负载参数以及决策标签的数据集划分为训练集和测试集，训练集用以训练决策模型，测试集用以对训练好的决策模型进行测试。
步骤S206：响应于所述决策模型决策当前温度控制策略需要迭代，基于温度预测模型的输出，生成更新的温度控制策略。
具体而言，可以将所述温度参数以及机房负载参数输入所述温度预测模型，以保证模型的输入数据的多维护化和完整性。在另一些实施例中，也可以将所述综合特征指标输入所述温度预测模型，以实现综合特征指标的复用。
具体而言，所述温度预测模型预测输出机房内的温度参数。由此，根据预测的未来设定时间段的机房内的温度参数，确定未来设定时间段的机房温度控制系统(诸如制冷系统)所需输出的控制温度。
在另一些实施例中，所述温度预测模型预测也可以直接输出机房温度控制系统的控制温度。
步骤S207：响应于所述决策模型决策当前温度控制策略不需要迭代，使用当前温度控制策略控制机房温度。
下面参见图3，图3是本公开的机房温度控制方法的系统执行的流程图。
温度控制系统300主要执行三个分支的计算。对于数据采集分支，其可以执行步骤S311：采集并储存机房节能算法所需的机房温度、机柜温度、制冷输出、天气温度。步骤S312：采集并储存机房IT负载等数据。步骤S313：将所采集的数据预处理为标准格式。
对于特征构建分支，其可以执行步骤S321：基于预处理后的数据，进行相关性计算；步骤S322：保留无关或者弱相关的参数；步骤S323：基于保留的参数，计算出算法迭代的综合决策指标。
对于智能决策分支，其可以执行步骤S331：基于综合决策指标的计算结果，智能地对机房温度节能控制算法进行迭代与否的决策行为；步骤S332：若需要迭代，则利用温度预测模型进行温度预测；步骤S333：根据预测的机房温度，进行控制机房制冷系统的工作。
由此，本公开相较于现有技术将天气温度及机房IT负载数据参与到决策指标的计算与评估，可根据变化情况给出策略更新建议，提高自动化水平，减少重复工作。
下面参见图4，图4是本公开的机房温度控制方法的又一种实施例的流程图。如图4所示，本实施例中，将气温及IT负载变化数据401、机房温度402、机柜温度403以及机房历史调控策略404等数据构造成一个向量集，在步骤S410中利用主成分分析(PCA)对数据进行降维处理，提取出一组相互无关的综合指标。由于机房节能优化涉及到的变量较多，历史数据较为庞杂，许多变量之间可能存在相关性，从而增加了问题分析的复杂性，同时对分析带来不便。因此引入主成分分析算法，对原数据进行降维处理，在尽量减少原指标包含信息的损失的同时，达到对所收集的温度、负载数据进行全面综合分析的目的。
步骤S420：基于综合指标的计算结果输入决策模型进行智能决策。决策模型可以在人工经验的基础上进行监督学习，提高决策智能判断的准确度。步骤S430：响应于决策模型 决策需要更新迭代温度控制策略时，利用温度预测模型进行温度预测。步骤S440：根据温度预测模型的预测结果进行智能温度控制。
进一步地，本公开中，将温度预测模型从传统方案中使用的RNN/LSTM算法改进为基于注意力模型的Transformer算法，大大提高算法运行时并行计算的效率，节约计算资源，提高预测准确度。
具体而言，本公开对Transformer模型进行了如下改进：
在计算Transformer模型的自注意力层的Query、Key和Value的过程中引入不同大小的卷积核进行卷积操作。其中，卷积核的大小>1。卷积核的计算可以引入粒子群算法求解最优解，粒子群算法的算法目标为求解最快拟合的卷积核大小设置。由此，相较于原有自注意力算法的Query、Key和Value计算仅体现单时间点之间的关联，改进的Query、Key和Value计算可以体现更多维的关联。
本公开还可以分析不同层的随时间变化的注意力得分分布，从而通过专家经验设置，来选取一定周期内特定自注意层进行堆叠，由此，可以降低每一层的空间复杂度至O(Llog 2L)。其中，L是序列长度。在I k l为单元l在第k至k+1层计算时要访问的单元的索引集合，在现有的Transformer模型中，其空间复杂度为O(L*2)，而在本公开降低每一层的空间复杂度后，模型整体空间复杂度降低为O(L(log 2L) 2)。
下面参见图5，图5是本公开的温度预测模型的示意图。温度预测模型包括六个依次相连的编码器521至526，以及六个依次相连的解码器531至536，其中，依次相连的最后一个编码器526的分别输出至六个解码器531至536中。输入特征通过嵌入算法转为词向量510，然后输入到第一个编码器521。六个编码器521至526可以具有相同的结构，如编码器521所示，其首先使输入其中的词向量乘以训练的权重矩阵，然后计算查询矩阵、键矩阵和值矩阵；通过查询向量和向量点积分计算打分，分数除以8再进行softmax归一化得到权重值，然后对所有值向量进行加权求和，最后输入到前馈神经网络中。六个解码器可以具有相同的结构，每个解码器的结构如解码器536所示，其包括前馈神经网络、编码-解码注意力层以及自注意力层。
由此，本公开中，将温度、负载等数据构造成一个向量集，利用主成分分析(PCA)对数据进行降维处理，提取出一组相互无关的综合指标，以进行决策指标的计算；基于综合指标的计算结果，在人工经验的基础上进行监督学习，提高决策智能判断的准确度，从而实现决策的智能判断；还将温度预测模型从传统方案中使用的RNN/LSTM算法改进为基于注意力模型的Transformer算法，大大提高算法运行时并行计算的效率，节约计算资源，提高预测准确度。
以上仅仅是示意性地描述本公开的具体实现方式，本公开并非也以此为限制，步骤的拆分、合并、执行顺序的变化、模块的拆分、合并、信息传输的变化皆在本公开的保护范围之内。
图6是本公开的机房温度控制装置的一种实施例的模块图。本公开的机房温度控制装置600，如图6所示，包括但不限于：参数采集模块610、综合特征指标模块620、决策输入模块630以及决策输出模块640。
参数采集模块610配置成采集温度参数以及机房负载参数；
综合特征指标模块620配置成根据所述温度参数以及机房负载参数，计算综合特征指标；
决策输入模块630配置成将当前温度控制策略以及所述综合特征指标输入决策模型；
决策输出模块640配置成：
响应于所述决策模型决策当前温度控制策略需要迭代，基于温度预测模型的输出，生成更新的温度控制策略；
响应于所述决策模型决策当前温度控制策略不需要迭代，使用当前温度控制策略控制机房温度。
上述模块的实现原理参见机房温度控制方法中的相关介绍，此处不再赘述。
图7是本公开的机房温度控制装置的另一种实施例的模块示意图。本公开的机房温度控制装700包括但不限于：参数采集模块710、相关性计算模块720、筛选模块730、综合特征指标计算模块740、决策输入模块750以及决策输出模块760。
参数采集模块710配置成采集温度参数以及机房负载参数；
相关性计算模块720配置成计算所述温度参数以及机房负载参数之间的相关性；
筛选模块730配置成保留与其它参数无关或这弱相关的参数；
综合特征指标计算模块740配置成根据保留的参数生成综合特征指标。
决策输入模块750配置成将当前温度控制策略以及所述综合特征指标输入决策模型；
决策输出模块760配置成：
响应于所述决策模型决策当前温度控制策略需要迭代，基于温度预测模型的输出，生成更新的温度控制策略；
响应于所述决策模型决策当前温度控制策略不需要迭代，使用当前温度控制策略控制机房温度。
上述模块的实现原理参见机房温度控制方法中的相关介绍，此处不再赘述。
本公开的机房温度控制装置通过将多维温度参数及机房IT负载参数参与到决策指标的计算与评估过程，从而使得机房制冷控制系统能够智能地根据多维温度参数和机房负载的变化做出温度控制策略迭代与否的决策，从而减少了人工在决策中的干预，减少了温度控制策略迭代次数，提高了人工智能算法对机房制冷的节能效率。
图6和图7仅仅是示意性的分别示出本公开提供的机房温度控制装置600和700，在不违背本公开构思的前提下，模块的拆分、合并、增加都在本公开的保护范围之内。本公 开提供的机房温度控制装置600和700可以由软件、硬件、固件、插件及他们之间的任意组合来实现，本公开并非以此为限。
本公开实施例还提供一种机房温度控制处理设备，包括处理器。存储器，其中存储有处理器的可执行指令。其中，处理器配置为经由执行可执行指令来执行的机房温度控制方法的步骤。
如上所示，该实施例本公开的机房温度控制处理设备由此，通过将多维温度参数及机房IT负载参数参与到决策指标的计算与评估过程，从而使得机房制冷控制系统能够智能地根据多维温度参数和机房负载的变化做出温度控制策略迭代与否的决策，从而减少了人工在决策中的干预，减少了温度控制策略迭代次数，提高了人工智能算法对机房制冷的节能效率。
所属技术领域的技术人员能够理解，本公开的各个方面可以实现为系统、方法或程序产品。因此，本公开的各个方面可以具体实现为以下形式，即：完全的硬件实施方式、完全的软件实施方式(包括固件、微代码等)，或硬件和软件方面结合的实施方式，这里可以统称为“电路”、“模块”或“平台”。
图8是本公开的机房温度控制处理设备的结构示意图。下面参照图8来描述根据本公开的这种实施方式的电子设备700。图8显示的电子设备800仅仅是一个示例，不应对本公开实施例的功能和使用范围带来任何限制。
如图8所示，电子设备800以通用计算设备的形式表现。电子设备800的组件可以包括但不限于：至少一个处理单元810、至少一个存储单元820、连接不同平台组件(包括存储单元820和处理单元810)的总线830、显示单元840等。
其中，存储单元存储有程序代码，程序代码可以被处理单元810执行，使得处理单元810执行本说明书上述机房温度控制方法部分中描述的根据本公开各种示例性实施方式的步骤。例如，处理单元810可以执行如图1中所示的步骤。
存储单元820可以包括易失性存储单元形式的可读介质，例如随机存取存储单元(RAM)8201和/或高速缓存存储单元8202，还可以进一步包括只读存储单元(ROM)8203。
存储单元820还可以包括具有一组(至少一个)程序模块8205的程序/实用工具8204，这样的程序模块8205包括但不限于：操作系统、一个或者多个应用程序、其它程序模块以及程序数据，这些示例中的每一个或某种组合中可能包括网络环境的实现。
总线830可以为表示几类总线结构中的一种或多种，包括存储单元总线或者存储单元控制器、外围总线、图形加速端口、处理单元或者使用多种总线结构中的任意总线结构的局域总线。
电子设备800也可以与一个或多个外部设备8001(例如键盘、指向设备、蓝牙设备 等)通信，还可与一个或者多个使得用户能与该电子设备800交互的设备通信，和/或与使得该电子设备800能与一个或多个其它计算设备进行通信的任何设备(例如路由器、调制解调器等等)通信。这种通信可以通过输入/输出(I/O)接口850进行。并且，电子设备800还可以通过网络适配器860与一个或者多个网络(例如局域网(LAN)，广域网(WAN)和/或公共网络，例如因特网)通信。网络适配器860可以通过总线830与电子设备800的其它模块通信。应当明白，尽管图中未示出，可以结合电子设备800使用其它硬件和/或软件模块，包括但不限于：微代码、设备驱动器、冗余处理单元、外部磁盘驱动阵列、RAID系统、磁带驱动器以及数据备份存储平台等。
本公开实施例还提供一种计算机可读存储介质，用于存储程序，程序被执行时实现的机房温度控制方法的步骤。在一些可能的实施方式中，本公开的各个方面还可以实现为一种程序产品的形式，其包括程序代码，当程序产品在终端设备上运行时，程序代码用于使终端设备执行本说明书上述机房温度控制方法部分中描述的根据本公开各种示例性实施方式的步骤。
如上所示，该实施例的用以执行机房温度控制的计算机可读存储介质通过将多维温度参数及机房IT负载参数参与到决策指标的计算与评估过程，从而使得机房制冷控制系统能够智能地根据多维温度参数和机房负载的变化做出温度控制策略迭代与否的决策，从而减少了人工在决策中的干预，减少了温度控制策略迭代次数，提高了人工智能算法对机房制冷的节能效率。
图9是本公开的计算机可读存储介质的结构示意图。参考图9所示，描述了根据本公开的实施方式的用于实现上述方法的程序产品900，其可以采用便携式紧凑盘只读存储器(CD-ROM)并包括程序代码，并可以在终端设备，例如个人电脑上运行。然而，本公开的程序产品不限于此，在本文件中，可读存储介质可以是任何包含或存储程序的有形介质，该程序可以被指令执行系统、装置或者器件使用或者与其结合使用。
程序产品可以采用一个或多个可读介质的任意组合。可读介质可以是可读信号介质或者可读存储介质。可读存储介质例如可以为但不限于电、磁、光、电磁、红外线、或半导体的系统、装置或器件，或者任意以上的组合。可读存储介质的更具体的例子(非穷举的列表)包括：具有一个或多个导线的电连接、便携式盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦式可编程只读存储器(EPROM或闪存)、光纤、便携式紧凑盘只读存储器(CD-ROM)、光存储器件、磁存储器件、或者上述的任意合适的组合。
计算机可读存储介质可以包括在基带中或者作为载波一部分传播的数据信号，其中承载了可读程序代码。这种传播的数据信号可以采用多种形式，包括但不限于电磁信号、光信号或上述的任意合适的组合。可读存储介质还可以是可读存储介质以外的任何可读介质，该可读介质可以发送、传播或者传输用于由指令执行系统、装置或者器件使用或者与其结合使用的程序。可读存储介质上包含的程序代码可以用任何适当的介质传输，包括但不限 于无线、有线、光缆、RF等等，或者上述的任意合适的组合。
可以以一种或多种程序设计语言的任意组合来编写用于执行本公开操作的程序代码，程序设计语言包括面向对象的程序设计语言—诸如Java、C++等，还包括常规的过程式程序设计语言—诸如“C”语言或类似的程序设计语言。程序代码可以完全地在用户计算设备上执行、部分地在用户设备上执行、作为一个独立的软件包执行、部分在用户计算设备上部分在远程计算设备上执行、或者完全在远程计算设备或服务器上执行。在涉及远程计算设备的情形中，远程计算设备可以通过任意种类的网络，包括局域网(LAN)或广域网(WAN)，连接到用户计算设备，或者，可以连接到外部计算设备(例如利用因特网服务提供商来通过因特网连接)。
综上，本公开通过将多维温度参数及机房IT负载参数参与到决策指标的计算与评估过程，从而使得机房制冷控制系统能够智能地根据多维温度参数和机房负载的变化做出温度控制策略迭代与否的决策，从而减少了人工在决策中的干预，减少了温度控制策略迭代次数，提高了人工智能算法对机房制冷的节能效率。
以上内容是结合具体的优选实施方式对本公开所作的进一步详细说明，不能认定本公开的具体实施只局限于这些说明。对于本公开所属技术领域的普通技术人员来说，在不脱离本公开构思的前提下，还可以做出若干简单推演或替换，都应当视为属于本公开的保护范围。

Claims (10)

  1. 一种机房温度控制方法，包括：
    采集温度参数以及机房负载参数；
    根据所述温度参数以及机房负载参数，计算综合特征指标；
    将当前温度控制策略以及所述综合特征指标输入决策模型；
    响应于所述决策模型决策当前温度控制策略需要迭代，基于温度预测模型的输出，生成更新的温度控制策略；
    响应于所述决策模型决策当前温度控制策略不需要迭代，使用当前温度控制策略控制机房温度。
  2. 根据权利要求1所述的机房温度控制方法，其中，所述根据所述温度参数以及机房负载参数，计算综合特征指标包括：
    计算所述温度参数以及机房负载参数之间的相关性；
    保留与其它参数无关或这弱相关的参数；
    根据保留的参数生成综合特征指标。
  3. 根据权利要求1所述的机房温度控制方法，其中，所述决策模型基于历史温度控制策略、历史温度参数、机房负载参数以及决策标签进行训练，所述决策标签包括当前温度控制策略不需要迭代以及当前温度控制策略需要迭代。
  4. 根据权利要求1所述的机房温度控制方法，其中，所述响应于所述决策模型决策当前温度控制策略需要迭代，基于温度预测模型的输出，生成更新的温度控制策略包括：
    将所述综合特征指标输入所述温度预测模型。
  5. 根据权利要求1所述的机房温度控制方法，其中，所述温度预测模型采用Transformer模型，其中，所述Transformer模型的自注意力层在计算Query，Key，Value的过程中采用不同大小的卷积核进行卷积操作，所述卷积核采用粒子群算法求解获得，所述Transformer模型还将设定周期内的自注意力层进行堆叠。
  6. 根据权利要求1所述的机房温度控制方法，其中，所述温度预测模型包括六个依次相连的编码器，以及六个依次相连的解码器，其中，依次相连的最后一个编码器的分别输出至六个解码器中。
  7. 根据权利要求1所述的机房温度控制方法，其中，所述温度参数包括机房温度、机柜温度、天气温度、机房温度控制系统的控制温度中的一项或多项。
  8. 一种机房温度控制装置，包括：
    参数采集模块，配置成采集温度参数以及机房负载参数；
    综合特征指标模块，配置成根据所述温度参数以及机房负载参数，计算综合特征指标；
    决策输入模块，配置成将当前温度控制策略以及所述综合特征指标输入决策模型；
    决策输出模块，配置成：
    响应于所述决策模型决策当前温度控制策略需要迭代，基于温度预测模型的输出，生成更新的温度控制策略；
    响应于所述决策模型决策当前温度控制策略不需要迭代，使用当前温度控制策略控制机房温度。
  9. 一种机房温度控制处理设备，包括：
    处理器；
    存储器，其中存储有所述处理器的可执行指令；
    其中，所述处理器配置为经由执行所述可执行指令来执行：
    权利要求1至7任意一项所述机房温度控制方法。
  10. 一种计算机可读存储介质，用于存储程序，所述程序被执行时实现：
    权利要求1至7任意一项所述机房温度控制方法。
PCT/CN2022/140624 2022-07-18 2022-12-21 机房温度控制方法及装置、电子设备、存储介质 WO2024016586A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210845348.5 2022-07-18
CN202210845348.5A CN116027829A (zh) 2022-07-18 2022-07-18 机房温度控制方法、装置、设备及存储介质

Publications (1)

Publication Number Publication Date
WO2024016586A1 true WO2024016586A1 (zh) 2024-01-25

Family

ID=86075031

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/140624 WO2024016586A1 (zh) 2022-07-18 2022-12-21 机房温度控制方法及装置、电子设备、存储介质

Country Status (2)

Country Link
CN (1) CN116027829A (zh)
WO (1) WO2024016586A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116806814B (zh) * 2023-06-30 2024-04-12 广东美赛尔细胞生物科技有限公司 一种间充质干细胞外泌体保存温度控制方法及系统

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180089582A1 (en) * 2016-09-28 2018-03-29 International Business Machines Corporation Ensemble model policy generation for prediction systems
CN111795488A (zh) * 2020-07-10 2020-10-20 哈尔滨工程大学 一种分布式机房温度的智能调控系统及方法
CN112050397A (zh) * 2020-08-27 2020-12-08 浙江省邮电工程建设有限公司 一种机房温度调控方法及系统
CN113190974A (zh) * 2021-04-09 2021-07-30 广东工业大学 基于深度学习的辊道窑温度场多点预测方法
CN113537459A (zh) * 2021-06-28 2021-10-22 淮阴工学院 一种药品库房温湿度预测方法
CN113935534A (zh) * 2021-10-20 2022-01-14 华能大理风力发电有限公司洱源分公司 一种基于卷积transformer架构的光伏日前预测方法、系统及设备
CN114282727A (zh) * 2021-12-27 2022-04-05 天津卓朗科技发展有限公司 机房温度预测方法、系统及电子设备
CN114353872A (zh) * 2021-12-30 2022-04-15 中国电信股份有限公司 一种机房温度的预测方法及装置
CN114757086A (zh) * 2021-12-17 2022-07-15 北京航空航天大学 多旋翼无人机实时剩余使用寿命预测方法和系统

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180089582A1 (en) * 2016-09-28 2018-03-29 International Business Machines Corporation Ensemble model policy generation for prediction systems
CN111795488A (zh) * 2020-07-10 2020-10-20 哈尔滨工程大学 一种分布式机房温度的智能调控系统及方法
CN112050397A (zh) * 2020-08-27 2020-12-08 浙江省邮电工程建设有限公司 一种机房温度调控方法及系统
CN113190974A (zh) * 2021-04-09 2021-07-30 广东工业大学 基于深度学习的辊道窑温度场多点预测方法
CN113537459A (zh) * 2021-06-28 2021-10-22 淮阴工学院 一种药品库房温湿度预测方法
CN113935534A (zh) * 2021-10-20 2022-01-14 华能大理风力发电有限公司洱源分公司 一种基于卷积transformer架构的光伏日前预测方法、系统及设备
CN114757086A (zh) * 2021-12-17 2022-07-15 北京航空航天大学 多旋翼无人机实时剩余使用寿命预测方法和系统
CN114282727A (zh) * 2021-12-27 2022-04-05 天津卓朗科技发展有限公司 机房温度预测方法、系统及电子设备
CN114353872A (zh) * 2021-12-30 2022-04-15 中国电信股份有限公司 一种机房温度的预测方法及装置

Also Published As

Publication number Publication date
CN116027829A (zh) 2023-04-28

Similar Documents

Publication Publication Date Title
WO2021078027A1 (zh) 构建网络结构优化器的方法、装置及计算机可读存储介质
CN113905391B (zh) 集成学习网络流量预测方法、系统、设备、终端、介质
CN102624865B (zh) 集群负载预测方法及分布式集群管理系统
CN114898121B (zh) 基于图注意力网络的混凝土坝缺陷图像描述自动生成方法
WO2024016586A1 (zh) 机房温度控制方法及装置、电子设备、存储介质
WO2024087468A1 (zh) 类别预测模型的训练方法、预测方法、设备和存储介质
WO2023168838A1 (zh) 语句文本的识别方法和装置、存储介质及电子装置
Tan et al. Multi-node load forecasting based on multi-task learning with modal feature extraction
CN115170344A (zh) 调控系统运行事件智能处理方法及装置、介质及设备
CN111445065A (zh) 用于数据中心制冷群控的能耗优化方法及系统
CN109946968B (zh) 一种楼宇内分布式能源匹配系统和方法
WO2020227983A1 (en) Hybrid-learning neural network architecture
WO2019227273A1 (en) Hierarchical concept based neural network model for data center power usage effectiveness prediction
CN114648103A (zh) 用于处理深度学习网络的自动多目标硬件优化
CN114679899B (zh) 机房空调自适应节能控制方法及装置、介质及设备
CN111624874A (zh) 用于城市污水处理的泵站集群智能化预测方法及系统
CN116975434A (zh) 内容推荐方法以及相关设备
CN112308594A (zh) 选址系统、方法、电子设备及存储介质
Yu [Retracted] Research on Optimization Strategy of Task Scheduling Software Based on Genetic Algorithm in Cloud Computing Environment
CN111046321B (zh) 光伏电站运维策略优化方法和装置
CN113988267A (zh) 用户意图识别模型的生成方法、用户意图识别方法和设备
CN113591391A (zh) 一种电力负荷控制装置、控制方法、终端、介质及应用
CN113076217A (zh) 基于国产平台的磁盘故障预测方法
CN114650190B (zh) 数据中心网络的节能方法、系统、终端设备及存储介质
Yu et al. Fast reconfiguration method of low-carbon distribution network based on convolutional neural network

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22951843

Country of ref document: EP

Kind code of ref document: A1