WO2021042368A1 - 功耗控制与方案生成方法、设备、系统及存储介质 - Google Patents

功耗控制与方案生成方法、设备、系统及存储介质 Download PDF

Info

Publication number
WO2021042368A1
WO2021042368A1 PCT/CN2019/104686 CN2019104686W WO2021042368A1 WO 2021042368 A1 WO2021042368 A1 WO 2021042368A1 CN 2019104686 W CN2019104686 W CN 2019104686W WO 2021042368 A1 WO2021042368 A1 WO 2021042368A1
Authority
WO
WIPO (PCT)
Prior art keywords
power consumption
target device
index data
consumption control
control scheme
Prior art date
Application number
PCT/CN2019/104686
Other languages
English (en)
French (fr)
Inventor
奉有泉
卢毅军
李栈
陶原
赵旭
Original Assignee
阿里巴巴集团控股有限公司
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 阿里巴巴集团控股有限公司 filed Critical 阿里巴巴集团控股有限公司
Priority to PCT/CN2019/104686 priority Critical patent/WO2021042368A1/zh
Priority to CN201980095638.9A priority patent/CN113728294B/zh
Publication of WO2021042368A1 publication Critical patent/WO2021042368A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
    • G06F1/26—Power supply means, e.g. regulation thereof
    • G06F1/32—Means for saving power
    • G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
    • G06F1/3234—Power saving characterised by the action undertaken
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Definitions

  • This application relates to the technical field of data centers, and in particular to a method, equipment, system, and storage medium for power consumption control and scheme generation.
  • Data Center DC
  • power consumption control methods have become the first choice for data centers to reduce energy consumption.
  • Various aspects of the present application provide a method, device, system, and storage medium for power consumption control and scheme generation, so as to realize dynamic power consumption control and improve the energy-saving effect of power consumption control.
  • An embodiment of the present application provides a method for generating a power consumption control scheme, including: obtaining relevant parameters of a target device, the relevant parameters including power consumption index data of the target device in a historical period; according to the target device in the historical period Predict the power consumption index data of the target device in the future period; according to the power consumption index data of the target device in the future period, generate the power consumption of the target device in the future period Control plan.
  • An embodiment of the present application also provides a method for generating a power consumption control scheme, including: acquiring related parameters of a device group, the related parameters including power consumption index data of the device group in a historical period, the device group including at least one A physical device; predict the power consumption index data of the device group in the future time period according to the power consumption index data of the device group in the historical time period; generate according to the power consumption index data of the device group in the future time period The power consumption control scheme used by the device group in the future period.
  • An embodiment of the present application also provides a power consumption control method, including: receiving a newly arrived first power consumption control scheme; replacing the currently used second power consumption control scheme with the first power consumption control scheme; The first power consumption control scheme controls the power consumption of the device.
  • An embodiment of the present application also provides a power consumption control method, including: receiving a newly arrived first power consumption control scheme; obtaining a third power control scheme according to the first power consumption control scheme and the currently used second power consumption control scheme Power consumption control scheme; replacing the currently used second power consumption control scheme with the third power consumption control scheme, and control the power consumption of the device according to the third power consumption control scheme.
  • An embodiment of the application also provides a power consumption prediction method, including: obtaining relevant parameters of a target device, the relevant parameters including power consumption index data of the target device in a historical period; according to the target device in the historical period The power consumption index data of the target device predicts the power consumption index data of the target device in the future period.
  • An embodiment of the present application also provides a computing device, including: a memory and a processor; the memory is used to store a computer program; the processor is coupled with the memory and is used to execute the computer program for : Acquire relevant parameters of the target device, the relevant parameters include the power consumption index data of the target device in the historical period; according to the power consumption index data of the target device in the historical period, predict the target device in the future period According to the power consumption index data of the target device in the future period, generate the power consumption control scheme used by the target device in the future period.
  • An embodiment of the present application also provides a computing device, including: a memory and a processor; the memory is used to store a computer program; the processor is coupled with the memory and is used to execute the computer program for :
  • Obtain the relevant parameters of the device group include the power consumption index data of the device group in the historical period, the device group includes at least one physical device; according to the power consumption of the device group in the historical period
  • the index data predicts the power consumption index data of the device group in the future time period; according to the power consumption index data of the device group in the future time period, the power consumption control scheme used by the device group in the future time period is generated.
  • An embodiment of the present application also provides a physical device, including: a memory, a processor, and a communication component; the communication component is used to receive a newly arrived first power consumption control scheme; the memory is used to store computer programs, The first power consumption control scheme and the currently used second power consumption control scheme; the processor, coupled with the memory, is configured to execute and the computer program for: controlling the currently used second power consumption The solution is replaced with the first power consumption control solution; the power consumption control of the physical device is performed according to the first power consumption control solution.
  • An embodiment of the present application also provides a physical device, including: a memory, a processor, and a communication component; the communication component is used to receive a newly arrived first power consumption control scheme; the memory is used to store computer programs, The first power consumption control scheme and the currently used second power consumption control scheme; the processor, coupled with the memory, is configured to execute the computer program for: according to the first power consumption control scheme And the currently used second power consumption control scheme to obtain the third power consumption control scheme; replace the currently used second power consumption control scheme with the third power consumption control scheme, and compare according to the third power consumption control scheme The physical device performs power consumption control.
  • An embodiment of the present application also provides a computing device, including: a memory and a processor; the memory is used to store a computer program; the processor is coupled with the memory and is used to execute the computer program for : Acquire relevant parameters of the target device, the relevant parameters include the power consumption index data of the target device in the historical period; according to the power consumption index data of the target device in the historical period, predict the target device in the future period Power consumption indicator data within.
  • An embodiment of the application also provides a data center system, including: at least one computer room and power consumption control equipment; wherein each computer room includes at least one physical device; the power consumption control device is used to obtain relevant parameters of the target device ,
  • the related parameters include the power consumption index data of the target device in the historical period; according to the power consumption index data of the target device in the historical period, predict the power consumption index data of the target device in the future period; According to the power consumption index data of the target device in the future period, generate the power consumption control scheme used by the target device in the future period and provide it to the target device for the target device to perform power consumption in the future period Control; wherein the target device is any one of the at least one physical device.
  • An embodiment of the present application also provides a computer room system, including: a computer room containing at least one physical device and a power consumption control device; the power consumption control device is used to obtain relevant parameters of the target device, and the relevant The parameters include the power consumption index data of the target device in the historical period; according to the power consumption index data of the target device in the historical period, predict the power consumption index data of the target device in the future period; according to the target The power consumption index data of the device in the future time period is generated, and the power consumption control scheme used by the target device in the future time period is generated and provided to the target device for the target device to perform power consumption control in the future time period; wherein, The target device is any one of the at least one physical device.
  • the embodiments of the present application also provide a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the processor is caused to implement the steps in the method embodiments of the present application.
  • power control is abstracted as a solution, and the power control solution of each physical device or device group is dynamically updated through artificial intelligence to achieve dynamic power consumption control, and the power consumption control can be accurate to each physical device or device group.
  • Device groups, different physical devices or device groups can perform power consumption control schemes adapted to their own power consumption conditions, which can not only improve the accuracy of power consumption control, but also improve the overall energy-saving effect of power consumption control.
  • Fig. 1a is a schematic structural diagram of a computer room system provided by an exemplary embodiment of this application;
  • FIG. 1b is a block diagram of the working principle of a power consumption control device provided by an exemplary embodiment of this application;
  • FIG. 1c is a schematic diagram of an internal structure of a power consumption control device provided by an exemplary embodiment of this application;
  • FIG. 2 is a schematic structural diagram of a data center system provided by an exemplary embodiment of this application;
  • FIG. 3a is a schematic flowchart of a power consumption control method provided by an exemplary embodiment of this application.
  • FIG. 3b is a schematic flowchart of another power consumption control method provided by an exemplary embodiment of this application.
  • FIG. 4a is a schematic flowchart of a method for generating a power consumption control scheme according to an exemplary embodiment of this application;
  • FIG. 4b is a schematic flowchart of another method for generating a power consumption control scheme according to an exemplary embodiment of this application;
  • 4c is a schematic flowchart of another method for generating a power consumption control scheme according to an exemplary embodiment of this application;
  • FIG. 4d is a schematic flowchart of a method for power consumption prediction according to an exemplary embodiment of this application.
  • Fig. 5a is a schematic structural diagram of a computing device provided by an exemplary embodiment of this application.
  • FIG. 5b is a schematic structural diagram of another computing device provided by an exemplary embodiment of this application.
  • FIG. 6a is a schematic structural diagram of a physical device provided by an exemplary embodiment of this application.
  • FIG. 6b is a schematic structural diagram of another physical device provided by an exemplary embodiment of this application.
  • FIG. 7 is a schematic structural diagram of yet another computing device provided by an exemplary embodiment of this application.
  • power control is abstracted as a scheme, and the power control scheme of each physical device or device group is dynamically updated through artificial intelligence to realize dynamic power.
  • Power consumption control, and the power consumption control can be accurate to each physical device or device group.
  • Different physical devices or device groups can carry out power consumption control schemes adapted to their own power consumption conditions, which can not only improve the accuracy of power consumption control, but also The overall energy saving effect of power consumption control can be improved.
  • Fig. 1a is a schematic structural diagram of a computer room system provided by an exemplary embodiment of this application.
  • the computer room system 100 of this embodiment includes: a computer room, which refers to a physical place where machinery and equipment are stored, for example, a room or a factory building. Further, as shown in FIG. 1a, the computer room system 100 further includes: at least one physical device 101 and a power consumption control device 102 located in the computer room. This embodiment does not limit the number of physical devices 101 in the computer room, and it may be one or multiple.
  • the device form of the physical device 101 is not limited.
  • the physical device 101 may be an IT-type device in a computer room and a refrigeration device for cooling the IT-type device, such as an air-conditioning device.
  • the at least one physical device 101 may include, but is not limited to: cabinet equipment, server equipment, computer equipment, printers, hubs, power supply equipment, storage equipment, network switching equipment, air conditioning equipment, and the like.
  • the server device may include, but is not limited to, a conventional server, a server array, or a cloud server.
  • the power supply equipment can be storage battery equipment, dry battery equipment, or uninterruptible power supply (UPS).
  • Storage devices may include, but are not limited to: disks, disk arrays, hard disks, network storage devices (NAS), and so on.
  • the normal operation of at least one physical device 101 consumes power.
  • the physical device 101 may be configured with a power consumption control scheme, and the physical device 101 performs power consumption control according to the power consumption control scheme.
  • the power consumption control scheme is a scheme for instructing the physical device 101 to perform power consumption control, and includes a trigger condition and a power consumption control method. When the trigger condition in the power consumption control scheme is met, the power consumption control method in the power consumption control scheme will be executed.
  • the power control method can be understood as a power control action, and when the trigger condition is met, the power control action corresponding to the trigger condition will be executed.
  • the power consumption control method is not limited, and any method with a power consumption control, limitation, or adjustment function is applicable to the embodiment of the present application.
  • power control methods include but are not limited to: power capping (Power Capping), dynamic voltage and power adjustment (Dynamic Voltage and Frequency Scaling, DVFS), and C-mode (C-State).
  • power capping is a method to limit the power consumption of the device. It can ensure that the actual power consumption of the device is lower than the maximum power available, and can ensure that the device uses lower power consumption when the device load is low.
  • DVFS is a dynamic technology that dynamically adjusts the operating frequency and voltage of the chip according to the different needs of the computing power of the application program running on the chip (such as CPU), so as to achieve the purpose of energy saving. It can control various processors and control devices on the device. The power and voltage of the chip and peripheral equipment are adjusted.
  • C-state is a low-power mechanism that allows the CPU to enter a low-power state when it is idle. The C-states included in C-states start from C0 to Cn.
  • C0 is the normal operating mode of the CPU, and the CPU is at 100% Running state; the higher the value of n after C, the deeper the CPU sleeps, the lower the power consumption of the CPU, and of course it takes more time to return to the C0 mode; where n is a positive integer.
  • At least one physical device 101 in the computer room has the same or substantially the same power consumption at any time, then at least one physical device 101 can be configured with a unified power consumption control scheme, and all physical devices 101 use the same
  • the power consumption control scheme is relatively simple and easy to implement.
  • the load conditions of different physical devices 101 in the computer room are quite different. Even at the same time, the loads of different physical devices 101 are different, so the power consumption of different physical devices 101 are different. In this case, if the power consumption control scheme is still uniformly configured for at least one physical device 101, and the physical devices 101 with different power consumption are controlled according to the same power consumption control scheme, obviously, the overall energy saving effect is not ideal. In addition, even for the same physical device 101, its load conditions will be different at different times, which means that the power consumption of the same physical device 101 at different times will also be different.
  • the power consumption control scheme of each physical device 101 is dynamically updated through artificial intelligence to realize dynamic power consumption control and improve the energy-saving effect of power consumption control.
  • a power consumption control device 102 is added in the computer room.
  • the power consumption control device 102 is mainly based on artificial intelligence, and dynamically updates the power consumption control scheme of each physical device 101, so that each physical device 101 performs power consumption control according to the dynamically changing power consumption control scheme.
  • the power consumption control device 102 dynamically updates the power consumption control scheme for each physical device 101 in the same or similar manner.
  • the process of dynamically updating the power consumption control scheme for the physical device 101 by the power consumption control device 102 is described.
  • this physical device is referred to as a target device, and the target device is any one of at least one physical device 101.
  • the power consumption control device 102 can obtain the relevant parameters of the target device, the relevant parameters include: the power consumption index data of the target device in the historical period; according to the power consumption index data of the target device in the historical period, predict the target device in the future The number of power consumption indicators in the time period; further, according to the power consumption indicator data of the target device in the future time period, the power consumption control scheme used by the target device in the future time period is generated and provided to the target device for the target device in the future time period Power consumption is controlled according to the power consumption control scheme.
  • the time length of the future period is not limited.
  • the future period can be the next half hour (the upcoming half hour), or the next hour (the upcoming hour), or the next few hours (the upcoming few hours), or the future day (the upcoming day) ), or the next week (the upcoming week), or a certain period of time in the next hour, or a certain period of the day in the future, or a certain day or several days in the next week, etc.
  • the length of time in the future period can be flexibly set according to application requirements.
  • the time length of the future period may be the same or different.
  • the power consumption control device 102 will predict the target device’s power consumption index data in the future period according to the target device’s power consumption index data in the corresponding historical period, and according to the target device’s power consumption index data in the future period
  • the power consumption index data is used to generate the power consumption control scheme used by the target device in the future period.
  • the power consumption control device 102 will dynamically update the power consumption control scheme used by the target device in each future period according to changes in the future time period; in the long run, the power consumption control scheme used by the target device is dynamically changing , Can realize dynamic power consumption control.
  • the historical period refers to a period of time before the current moment, such as yesterday, yesterday morning, yesterday afternoon, last Tuesday, last Monday to Wednesday, last Saturday, last Sunday, etc., relative to the current time. Historical period.
  • the number of historical periods used is not limited.
  • the historical period can be one or multiple; in addition, the time length of each historical period is not limited, and it can be set adaptively according to application requirements. .
  • this embodiment does not limit the correspondence between historical time periods and future time periods, and can be set adaptively according to application requirements. The following example illustrates:
  • the power consumption index data of the target device For example, based on the power consumption index data of the target device between 10 am and 12 am every day last week, predict the power consumption index data of the target device between 10 am and 12 am every day in the next week; according to the target device in the future The power consumption index data between 10 am and 12 am every day of the week, and the power consumption control scheme used by the target device in the next week is generated.
  • the target device For example, according to the target device’s daily power consumption index data and every night’s power consumption index data in the last few days, it is possible to predict the target device’s daytime power consumption index data and night power consumption index data in the next day; according to the target device The power consumption index data during the day and the night power consumption index data in the next day are used to generate the power consumption control scheme used by the target device in the next day.
  • the power consumption index data refers to some data related to power consumption, and may include one type of data or multiple types of data.
  • the type of power consumption indicator data used in the future time period is the same as the type of power consumption indicator data used in the historical time period; or, the type of power consumption indicator data used in the future time period is used in the historical time period Part of the types of power consumption index data.
  • the power consumption index data related to the equipment will also be different.
  • the internal temperature, power consumption, CPU frequency, and CPU load of IT equipment are all related to the power consumption of IT equipment. Therefore, the internal temperature, power consumption, CPU frequency and CPU At least one type of data among the load and the like is used as power consumption index data.
  • the power consumption control device 102 can obtain the historical information of the target equipment.
  • the power consumption and CPU frequency in the time period are used as the power consumption index data of the target device in the historical time period; according to the power consumption and CPU frequency of the target device in the historical time period, the power consumption and CPU frequency of the target device in the future time period are predicted; The power consumption and CPU frequency of the target device in the future period to generate the power consumption control scheme used by the target device in the future period.
  • the power consumption control device 102 can obtain The power consumption, CPU frequency and CPU load of the target device in the historical period are used as the power consumption index data of the target device in the historical period; according to the power consumption, CPU frequency and CPU load of the target device in the historical period, the target device is predicted in the future Power consumption, CPU frequency and CPU load in the time period; according to the power consumption, CPU frequency and CPU load of the target device in the future time period, generate the power consumption control scheme used by the target device in the future time period.
  • the power consumption control device 102 can obtain the internal temperature, power consumption, CPU frequency and CPU load of the target device in the historical period as the power consumption index data of the target device in the historical period; according to the internal temperature, power consumption, and CPU frequency of the target device in the historical period And CPU load, predict the power consumption and CPU load of the target device in the future period; according to the power consumption and CPU load of the target device in the future period, generate the power consumption control scheme used by the target device in the future period.
  • the compressor frequency and fan speed of the air conditioning equipment can be At least various of the return air temperature and the outlet air temperature are used as power consumption index data.
  • the power consumption control device 102 can obtain the target device
  • the compressor frequency and fan speed in the historical period are used as the power consumption index data of the target device in the historical period; according to the compressor frequency and fan speed of the target device in the historical period, the target device's performance in the future period is predicted
  • the frequency of the compressor and the speed of the fan according to the frequency of the compressor and the speed of the fan of the target device in the future period, the power consumption control scheme used by the target device in the future period is generated.
  • the power consumption control device 102 can obtain the compressor frequency, fan speed, return air temperature, and outlet air temperature of the target device in the historical period as the power consumption index data of the target device in the historical period; according to the target device’s historical period Compressor frequency, fan speed, return air temperature, and outlet air temperature, predict the compressor frequency, fan speed, return air temperature, and outlet air temperature of the target device in the future period; according to the target device's future period of time.
  • the frequency of the compressor, the speed of the fan, the temperature of the return air and the temperature of the outlet air generate the power consumption control scheme used by the target device in the future period.
  • the power The consumption control device 102 can obtain the compressor frequency, fan speed, return air temperature, and outlet air temperature of the target device in the historical period as the power consumption index data of the target device in the historical period; according to the target device’s historical period Compressor frequency, fan speed, return air temperature and outlet air temperature, predict the return air temperature and outlet air temperature of the target device in the future period; generate according to the return air temperature and outlet air temperature of the target device in the future period The power consumption control scheme used by the target device in the future period.
  • the target device may also be determined according to the topological structure between the target device and other devices.
  • Other devices associated with the device obtain the power consumption of other devices associated with the target device in the historical period, which is also used as a kind of power consumption index data of the target device in the historical period.
  • other devices associated with the target device refer to devices that have an impact on the power consumption of the target device.
  • the manner in which the power consumption control device 102 obtains the power consumption index data of the target device in the historical period is not limited.
  • the power consumption control device 102 may provide a human-computer interaction interface through which the manager of the computer room can input the power consumption indicator data of the target device in a historical period to the power consumption control device 102 through the human-computer interaction interface.
  • the human-computer interaction interface can be implemented in multiple ways.
  • the human-computer interaction interface can be implemented as a web page or application page containing an input box, and the manager can input the power consumption indicator data of the target device in the historical period in the input box.
  • the human-computer interaction interface can be implemented as a web page or application page containing a power consumption indicator data option drop-down box.
  • the manager can use the power consumption indicator data option drop-down box to select the power consumption indicator data that needs to be input from the drop-down box.
  • the power consumption control device 102 supports configuration files, and the administrator can send the configuration file to the power consumption control device 102 through other devices (for example, the terminal device used by the administrator), and the configuration file contains the target The power consumption index data of the device in the historical period; the power consumption control device 102 can read the power consumption index data of the target device in the historical period from the configuration file.
  • the power consumption control solution provided by the power consumption control device 102 can be obtained, and the power consumption control can be performed according to the power consumption control solution.
  • the process of the target device performing power control according to the power control scheme includes: monitoring whether the trigger condition in the power control scheme is met; when the trigger condition is met, according to the power control method in the power control scheme Perform power consumption control.
  • the trigger condition in the power consumption control scheme is the condition required to trigger the target device to start the power consumption control.
  • the trigger condition may be a time range, which means that within the time range, a corresponding power consumption control method needs to be activated for power consumption control.
  • the trigger condition is the power consumption index data range, which means that when the power consumption index data is in this range, a corresponding power consumption control method needs to be activated to perform power consumption control.
  • the power consumption index data range as the trigger condition will also be different.
  • the power consumption index data range correspondingly includes: power consumption range.
  • the corresponding power consumption control method can be activated for power consumption control. Further optionally, if the actual power consumption of the target device in a short period of time is lower than the lower limit of the power consumption range, the power consumption control can be ended.
  • the power consumption index data range correspondingly includes: power consumption range and CPU frequency range.
  • both trigger conditions for example, when the actual power consumption of the target device within a short period of time is within the power consumption range and the actual CPU frequency of the target device within a short period of time is within the CPU range , Start the corresponding power consumption control method for power consumption control.
  • the power consumption control can be ended.
  • the reason why the actual power consumption or actual CPU frequency within a short period of time is selected is to maintain the stability of power consumption control and minimize the probability of the ping-pong effect.
  • the time length of "a short period of time" here is not limited, it can be 10 seconds, 1 minute, 5 minutes, etc.
  • real-time power consumption or power consumption index data such as CPU frequency can also be used as a basis for power consumption control. .
  • the power control is abstracted as a scheme, and the power control scheme of each physical device is dynamically updated through artificial intelligence to realize dynamic power consumption control, and the power consumption control can be accurate to each physical device and different physical devices.
  • the power consumption control scheme that can be adapted to its own power consumption situation can not only improve the accuracy of power consumption control, but also improve the overall energy-saving effect of power consumption control.
  • the power consumption control device 102 after the power consumption control device 102 obtains the power consumption index data of the target device in the historical period, it can directly predict the target device’s power consumption in the future period according to the power consumption index data of the target device in the historical period.
  • the power consumption index data After the power consumption control device 102 obtains the power consumption index data of the target device in the historical period, it can directly predict the target device’s power consumption in the future period according to the power consumption index data of the target device in the historical period.
  • the power consumption index data after the power consumption control device 102 obtains the power consumption index data of the target device in the historical period, it can directly predict the target device’s power consumption in the future period according to the power consumption index data of the target device in the historical period.
  • the target device in order to improve the accuracy of the prediction result, can be classified; if the target device is a device whose power consumption changes meet the regular requirements, then the target device will be executed according to the power consumption of the target device in the historical period. Index data, the operation of predicting the power consumption index data of the target device in the future period; if the target device is not a device whose power consumption changes meet the regular requirements, the operation of generating a power consumption control scheme for the target device is ended, or the target device is set The default power control scheme, such as DVFS.
  • the device parameters of the target device can be added.
  • the device parameters here mainly refer to some static parameters related to the device itself, such as device type, device model, device serial number, and device specifications.
  • the device parameters of the target device can be collected from the target device, or can be obtained in an out-of-band manner.
  • the power consumption control device 102 in addition to obtaining the target device’s power consumption index data in the historical period, the power consumption control device 102 also needs to obtain the device parameters of the target device, and according to the device parameters of the target device and the target device’s current
  • the power consumption index data in the historical period of time identifies whether the target device is a device whose power consumption changes meet the regular requirements.
  • the manner in which the power consumption control device 102 obtains the device parameters of the target device refer to the manner in which the power consumption control device 102 obtains the power consumption index data of the target device in a historical period, which is not repeated here.
  • an internal structure of the power consumption control device 102 includes: a classifier, a predictor, and a decision maker.
  • the classifier is mainly responsible for classifying the target device according to the device parameters of the target device and the power consumption index data of the target device in the historical period, to identify whether the target device is a device whose power consumption changes meet the regular requirements, and output the classification result ; If the target device is a device whose power consumption change meets the law requirement, it enters the predictor; if the target device is not a device whose power consumption change meets the law requirement, it enters the decision maker.
  • the predictor is mainly used to predict the power consumption index data of the target device in the future period according to the power consumption index data of the target device in the historical period, and output to the decision maker.
  • the decision maker is used to generate the power consumption control scheme used by the target device in the future period based on the target device's power consumption index data in the future period when the target device is a device whose power consumption changes meet the regular requirements; or, in the target device If it is not a device whose power consumption changes meet the regular requirements, a default power consumption control scheme can be set for the target device.
  • a classification model may be used to classify the target device.
  • the device parameters of the target device and the power consumption index data of the target device in the historical period can be input into the classification model to obtain the classification result of whether the target device is a device whose power consumption changes meet the regular requirements.
  • the benchmark distribution feature corresponding to the device type of the target device can be determined.
  • the benchmark distribution feature is the distribution feature of power consumption index data that meets the requirements of the law; according to the benchmark distribution feature and the target device The distribution characteristics of the power consumption index data in the historical period are used to classify the target device to obtain the classification result of whether the target device is a device whose power consumption changes meet the regular requirements.
  • the similarity between the baseline distribution feature and the distribution feature of the target device’s power consumption index data in the historical period can be directly calculated, and the target device can be classified according to the similarity; if the similarity is greater than the set similarity threshold, determine The target device is a device whose power consumption change meets the law requirements; if the similarity is less than or equal to the set similarity threshold, it is determined that the target device is not a device whose power consumption change meets the law requirements.
  • the logistic regression algorithm or random forest algorithm can be used to classify the target device according to the baseline distribution characteristics and the distribution characteristics of the target device’s power consumption index data in the historical period, so as to obtain whether the target device belongs to the power consumption change satisfies The classification result of the equipment required by the law.
  • the output is y ⁇ 0,1 ⁇
  • the value of z is converted to 0 or 1 through the Sigmoid function. For example, if the value calculated by the Sigmoid function is greater than or equal to 0.5, it is classified as category 1; if the value calculated by the Sigmoid function is less than 0.5, it is classified as category 0.
  • category 1 indicates that the target device is a device whose power consumption changes meet the requirements of the law
  • category 0 indicates that the target device is not a device whose power consumption changes meet the requirements of the law.
  • the benchmark distribution feature can be used as the training sample to construct multiple decision trees, and then use the multiple decision trees to analyze the distribution characteristics of the power consumption index data of the target device in the historical period. Classification, and finally vote according to multiple decision trees to determine whether the target device is a device whose power consumption changes meet the requirements of the law.
  • the process of constructing multiple decision trees please refer to the prior art, which will not be repeated here.
  • a machine learning model in artificial intelligence such as a prediction model, may be used to predict the power consumption index data of the target device in the future period.
  • the power consumption index data of the target device in the historical period can be input into the prediction model to obtain the power consumption index data of the target device in the future period.
  • linear regression algorithms or deep learning algorithms can be used to predict the power consumption index data of the target device in the future period.
  • the training process and the prediction process of the prediction model using the linear regression algorithm or the deep learning algorithm it is the same or similar to the prior art, and will not be repeated here.
  • the performance index data in the process of predicting the power consumption index data of the target device in the future period, in addition to considering the power consumption index data of the target device in the historical period, it can also be combined with the target device in the historical period.
  • the performance index data is shown in Figure 1b.
  • the performance index data here refers to data that can reflect the service performance of the target device.
  • the target device is an IT device
  • at least one application or service may be running on the target device, such as cloud computing service, game service, instant messaging service, mail service, or online transaction service, etc.
  • the performance index data of the target device may be QoS data of an application or service running on the target device.
  • QoS data can be response time, TPS, QPS, or the number of concurrent users.
  • the performance index data of the target device may be collected and stored in the database. Based on this, the performance index data of the target device in the historical period can be obtained from the database.
  • the service provider can provide the target device to a user for use.
  • a mutually recognized agreement is defined between the service provider and the user, that is, the service Service Level Agreement (SLA), in which the performance requirements that the target device needs to meet are agreed upon.
  • SLA Service Level Agreement
  • the performance index data that the target device needs to meet can be obtained from the SLA agreement corresponding to the target device, as the performance index data of the target device in the historical period.
  • the power consumption index data of the target device in the historical period and the performance index data of the target device in the historical period can be input into the prediction model to obtain the target device in the future.
  • Power consumption indicator data during the time period considering the influence of the performance index data on the power consumption index data, it is helpful to improve the accuracy of the prediction result, thereby generating a more reasonable power consumption control scheme for the target device, and improving the accuracy of the power consumption control.
  • the power consumption index data of the target device in the future time period can be used to generate the power consumption of the target device in the future time period. Consumption control scheme.
  • a power consumption control scheme may be generated based on rules, that is, some trigger conditions and power consumption control methods corresponding to the trigger conditions are predefined.
  • the process of generating the power consumption control scheme used by the target device in the future period includes: according to the power consumption index data of the target device in the future period, matching with preset trigger conditions, and determining that the preset trigger conditions are matched The target trigger condition of the target; obtain the preset power control method corresponding to the target trigger condition, such as DVFS or C-State, as the target power control method; according to the target trigger condition and the target power control method, generate the target device in the future The power consumption control scheme used during the time period.
  • the power consumption control scheme used by the target device in the future period includes: a target trigger condition and a target power consumption control method.
  • the power consumption control scheme can be generated by combining rules and artificial intelligence, that is, some trigger conditions can be predefined, and when the trigger conditions are matched, artificial intelligence is used to generate and The power consumption control method adapted to the trigger condition.
  • the process of generating the power consumption control scheme used by the target device in the future period includes: according to the power consumption index data of the target device in the future period, matching with preset trigger conditions, and determining that the preset trigger conditions are matched
  • the target trigger condition of the target simulate the energy-saving effect of various power control methods under the target trigger condition; according to the energy-saving effect of various power control methods under the target triggering condition, select the power control method that meets the energy-saving requirements with the energy-saving effect,
  • a target power consumption control method As a target power consumption control method; according to the target trigger condition and the target power consumption control method, a power consumption control scheme used by the target device in the future period of time is generated.
  • the power consumption control scheme used by the target device in the future period includes: a target trigger condition and a target power consumption control method, for example, from 6:00 to 8:00 in the morning, DVFS is executed.
  • the performance index data range corresponding to the trigger condition can be understood as a constraint condition for power consumption control, and a performance requirement that needs to be guaranteed during the power consumption control process. It is worth noting that not every trigger condition corresponds to a performance index data range.
  • the target trigger condition After the target trigger condition is determined, it can also be judged whether the target trigger condition has a corresponding performance index data range; if so, the preset performance index data range corresponding to the target trigger condition is obtained as the target performance index Data range; Then, according to the target trigger condition, the target power consumption control method, and the target performance index data range, the power consumption control scheme used by the target device in the future period is generated.
  • the power consumption control scheme used by the target device in the future period includes: a target trigger condition, a target power consumption control method, and a target performance index data range.
  • the preset trigger condition may include at least one of the following: The trigger condition of the time range and the trigger condition of the data range of the power consumption indicator.
  • the trigger condition indicating the time range means: within the time range indicated by the trigger condition, the corresponding power consumption control method is started to perform power consumption control.
  • the trigger condition indicating the data range of the power consumption index means: when the actual power consumption index data of the target device within a short period of time is within the power consumption index data range indicated by the trigger condition, start the corresponding power consumption control method to perform the function Consumption control.
  • the trigger condition indicating the data range of the power consumption index means: when the actual power consumption index data of the target device within a short period of time is within the power consumption index data range indicated by the trigger condition, start the corresponding power consumption control method to perform the function Consumption control.
  • matching the preset trigger condition according to the power consumption index data of the target device in the future period to determine the matched target trigger condition in the preset trigger condition may include at least one of the following methods:
  • Manner 1 According to the time range corresponding to the future time period, from the preset trigger conditions representing the time range, determine the trigger condition that the represented time range falls within the time range corresponding to the future time period as the target trigger condition. For example, suppose the future time period refers to the morning of the next day, the time range of a trigger condition is 6:00-8:00 am, and the time range of another trigger condition is 2:00-4:00 pm, it means The trigger condition of 6:00-8:00 am is the target trigger condition.
  • Method 2 According to the power consumption index data of the target device in the future time period, determine the power consumption index of the target device in the future time period from the preset trigger condition indicating the power consumption index data range.
  • the trigger condition that has appeared in the data is used as the target trigger condition. For example, suppose that the power consumption index data of the target device in the future period includes power consumption, and the power consumption value fluctuates between 10w-30w, the power consumption range represented by a trigger condition is greater than 20w, and the power consumption represented by another trigger condition If the consumption range is greater than 50w, it means that the trigger condition greater than 20w is the target trigger condition.
  • the power consumption control device 102 after obtaining the power consumption control scheme used by the target device in the future period, the power consumption control device 102 provides the power consumption control scheme used by the target device in the future period to the target device. For the target device, every time a new power consumption control scheme provided by the power consumption control device 102 is received, power consumption control can be performed according to the new power consumption control scheme.
  • the target device receives the newly arrived power consumption control scheme; replaces the currently used power consumption control scheme with the newly arrived power consumption control scheme, thereby performing power consumption control according to the newly arrived power consumption control scheme .
  • the newly arrived power consumption control scheme is referred to as the first power consumption control scheme
  • the currently used power consumption control scheme is referred to as the second power consumption control scheme.
  • the old power consumption control scheme will be overwritten with the new power consumption control scheme, which is simple and easy to implement.
  • the first power consumption control scheme includes: trigger conditions and power consumption control methods.
  • the process for the target device to perform power consumption control according to the first power consumption control scheme is: monitoring whether the trigger condition in the first power consumption control scheme is met; the situation where the trigger condition in the first power consumption control scheme is met Next, use the power consumption control method corresponding to the met trigger condition in the first power consumption control scheme to control the power consumption of the target device.
  • the trigger condition as 6:00-8:00 in the morning
  • the power consumption control method as DVFS as an example, during 6:00-8:00 in the morning, DVFS can be turned on for power consumption control.
  • the process of controlling the power consumption of the target device according to the first power consumption control scheme monitor the actual performance index data of the target device, such as actual QoS data; and adjust according to the actual performance index data of the target device The control intensity of the first power consumption control scheme.
  • the actual performance data of the target device can be compared with the specified performance lower limit; if the actual performance data of the target device is less than the specified performance lower limit, the first power consumption control scheme will be shut down until the actual performance index of the target device The data is greater than or equal to the specified performance lower limit.
  • the actual performance data of the target device can be compared with the specified performance lower limit; if the actual performance data of the target device is less than the specified performance lower limit, the power consumption control method in the first power consumption control scheme is adjusted to Control the power consumption control method with lower intensity until the actual performance index data of the target device is greater than or equal to the specified performance lower limit.
  • the specified lower limit of performance may be the lower limit of performance acceptable to the target device.
  • the designated performance lower limit value may also be the lower limit value of the performance index data range in the first power consumption control scheme.
  • the actual performance index data of the target device is combined to control the power consumption of the target device, which can not only save energy consumption, but also ensure the quality of service of the target device.
  • the target device receives the newly arrived first power consumption control scheme; obtains the third power consumption control scheme according to the first power consumption control scheme and the currently used second power consumption control scheme; The second power consumption control scheme used is replaced with the third power consumption control scheme, and the power consumption of the device is controlled according to the third power consumption control scheme.
  • the target device may directly merge the first power consumption control scheme and the second power consumption control scheme to obtain the third power consumption control scheme.
  • the target device can compare the trigger condition in the first power consumption control scheme with the trigger condition in the second power consumption control scheme; if the trigger condition in the first power consumption control scheme is compared with the second power
  • the types of trigger conditions in the power consumption control scheme are different, and the first power control scheme and the second power consumption control scheme are combined to obtain the third control scheme; if the trigger conditions in the first power consumption control scheme are the same as the second power consumption If the trigger conditions in the control scheme are of the same type, the old power consumption control scheme is covered by the new power consumption control scheme, that is, the first power consumption control scheme is used as the third control scheme.
  • the first power consumption control scheme and the second power consumption control scheme can be The two power consumption control schemes are combined as the third control scheme. If the trigger condition in the first power consumption control scheme is a trigger condition representing the time range, and the trigger condition in the second power consumption control scheme is also a trigger condition representing the time range, then the second power consumption control scheme can be discarded and the first The power consumption control scheme, that is, the first power consumption control scheme as the third control scheme.
  • the third power consumption control scheme includes trigger conditions and power consumption control methods, and there may be a combination of multiple sets of trigger conditions and power consumption control methods. Based on this, the power consumption control of the device according to the third power consumption control scheme includes: when the trigger conditions in the third power consumption control scheme are met, using the trigger conditions that are met in the third power consumption control scheme The corresponding power consumption control method controls the power consumption of the target device.
  • the process of controlling the power consumption of the target device according to the third power consumption control scheme monitor the actual performance index data of the target device, such as actual QoS data; and adjust according to the actual performance index data of the target device The control intensity of the third power consumption control scheme.
  • the actual performance data of the target device can be compared with the specified performance lower limit; if the actual performance data of the target device is less than the specified performance lower limit, the third power consumption control scheme will be shut down until the actual performance index of the target device The data is greater than or equal to the specified performance lower limit.
  • the actual performance data of the target device can be compared with the specified performance lower limit; if the actual performance data of the target device is less than the specified performance lower limit, the power consumption control method in the third power consumption control scheme is adjusted to Control the power consumption control method with lower intensity until the actual performance index data of the target device is greater than or equal to the specified performance lower limit.
  • the specified lower limit of performance may be the lower limit of performance acceptable to the target device.
  • the designated performance lower limit value may also be the lower limit value of the performance index data range in the third power consumption control scheme.
  • the actual performance index data of the target device is combined to control the power consumption of the target device, which can not only save energy consumption, but also ensure the quality of service of the target device.
  • the power consumption control device 102 uses one physical device as a unit, and can dynamically update the power consumption control scheme of each physical device, but it is not limited to this.
  • the power consumption control device 102 may also use the device group as a unit to dynamically update the power consumption control scheme of each device group.
  • the device group includes at least one physical device, and each physical device in the device group forms a binding relationship and can share the same power consumption control scheme.
  • the method of forming the device group is not limited.
  • physical devices that are relatively close in the computer room can be divided into one device group, or physical devices with the same or similar load conditions can be divided into one device group, or physical devices of the same type can be divided into one device group, or The physical devices of the same model are divided into one device group, or the physical devices of the same manufacturer are divided into one device group, or the physical devices in the entire computer room are divided into one device group, and so on.
  • the related parameters of the device group can be obtained.
  • the related parameters of the device group include the power consumption index data of the device group in the historical period; according to the power consumption index data of the device group in the historical period, the equipment is predicted The power consumption index data of the group in the future time period; according to the power consumption index data of the device group in the future time period, the power consumption control scheme used by the device group in the future time period is generated.
  • the power consumption index data of the device group in the historical period includes: the power consumption index data of each physical device in the device group in the historical period; correspondingly, the predicted device group is in the future period
  • the power consumption index data of includes: the power consumption index data of each physical device in the equipment group in the future period.
  • a prediction method is: a separate prediction method, that is, for each physical device in the device group, separately predict its power consumption index data in the future time period according to its power consumption index data in the historical time period.
  • another prediction method is: joint prediction method, that is, joint prediction based on the power consumption index data of all physical devices in the device group in the historical period to obtain the power consumption index of each physical device in the future period data.
  • the power consumption index data of the device group in the historical period is power consumption index data calculated based on the power consumption index data of each physical device in the device group in the historical period.
  • the power consumption index data of the device group in the historical period can be some discrete values or continuous values, including the power consumption index data at each historical moment in the historical period.
  • the power consumption index data of a device group at a certain historical moment may be the average of the power consumption index data of each physical device in the device group at the same historical moment, or the maximum value, or the minimum value. Or the mean of the maximum and minimum among them, and so on.
  • the power consumption index data of the device group in the future period is predicted based on the power consumption index data of the device group in the historical period, which may be some discrete values or continuous values.
  • the device group in order to improve the accuracy of the prediction result, can be classified; if the device group belongs to the device group whose power consumption changes meet the regular requirements, then the device group will be executed according to the power consumption of the device group in the historical period. Index data, the operation of predicting the power consumption index data of the device group in the future period; if the device group does not belong to the device group whose power consumption changes meet the regular requirements, the operation of generating a power consumption control plan for the device group is ended, or the device group is the device group Set the default power control scheme.
  • the device parameters of the device group can be added.
  • the device parameters of a device group mainly refer to some static parameters related to the physical devices in the device group, such as device type, device model, device serial number, and device specifications.
  • the power consumption control device 102 also needs to obtain the device parameters of the device group, and according to the device parameters of the device group and the power consumption of the device group in the historical period Index data to identify whether the device group belongs to the device group whose power consumption changes meet the regular requirements.
  • the device group belongs to the device group whose power consumption changes meet the regular requirements, according to the power consumption index data of the device group in the historical period, predict the power consumption index data of the device group in the future period, and then, according to the device group in the future period Power consumption index data to generate the power consumption control scheme used by the device group in the future period. If the device group does not belong to the device group whose power consumption changes meet the regular requirements, a default power consumption control scheme can be set for the device group.
  • a classification model when classifying the device group, a classification model can be used.
  • the classification model may use a logistic regression algorithm or a random forest algorithm to classify the device group, so as to obtain the classification result of whether the device group belongs to the device group whose power consumption changes meet the regular requirements.
  • a prediction model when predicting the power consumption index data of the device group in the future period, a prediction model may be used.
  • the power consumption index data of the device group in the historical period can be input into the prediction model to obtain the power consumption index data of the device group in the future period.
  • the prediction model can use a linear regression algorithm or a deep learning algorithm, but it is not limited to this.
  • the process of generating a power control scheme by the power control device 102 for each device group is similar to the process of generating a power control scheme for each physical device.
  • the difference is that some data used in the scheme generation process will be somewhat different.
  • the related process may be relatively more complicated, but the basic implementation or principle is the same, so it will not be repeated here, and it can be implemented by analogy with the related content in the foregoing embodiment.
  • the technical solution of the embodiment of the present application is described by taking the computer room system 100 as an example, but it is not limited to this.
  • the technical solutions of the embodiments of the present application can also be applied to environments including multiple physical devices, such as data center systems and clusters.
  • the technical solutions in the embodiments of the present application are also applicable to individual physical devices.
  • Fig. 2 is a schematic structural diagram of a data center system provided by an exemplary embodiment of this application.
  • the data center system 200 includes: at least one computer room 201 and a power consumption control device 202.
  • each computer room 201 includes at least one physical device 203.
  • the number of physical devices contained in each computer room 201 may be one or multiple.
  • the computer room 201 in this embodiment is similar to the computer room in the embodiment shown in FIG. 1a.
  • the description in the embodiment shown in FIG. 1a please refer to the description in the embodiment shown in FIG. 1a, which will not be repeated here.
  • the power consumption control device 202 does not belong to a certain computer room, but belongs to the entire data center system 200, and a power consumption control solution needs to be provided for the physical devices 203 in each computer room 201. .
  • the power consumption control device 202 does not belong to any computer room, in terms of physical deployment, it can be deployed in a certain computer room or independently deployed outside each computer room.
  • the power consumption control device 202 of this embodiment can dynamically update the power consumption control scheme of each physical device in units of physical devices, or dynamically update the power consumption control of each device group in units of device groups. Program.
  • the power consumption control device 202 dynamically updates the power consumption control scheme for each physical device 203 in the same or similar manner. Therefore, taking the target device as an example, the function of the power consumption control device 202 will be described.
  • the power consumption control device 202 is used to obtain relevant parameters of the target device.
  • the relevant parameters include the power consumption index data of the target device in the historical period; according to the power consumption index data of the target device in the historical period, predict the target device in the future period Power consumption index data within the target device; according to the power consumption index data of the target device in the future period, generate the power consumption control scheme used by the target device in the future period and provide it to the target device for the target device to perform power consumption control in the future period ;
  • the target device is any one of at least one physical device.
  • the power consumption control device 202 is used to obtain the relevant parameters of the device group.
  • the relevant parameters of the device group include the power consumption index data of the device group in the historical period; according to the power consumption index data of the device group in the historical period, it is predicted that the device group is in Power consumption index data in the future period; according to the power consumption index data of the device group in the future period, generate the power consumption control scheme used by the device group in the future period.
  • the physical devices in a device group can come from the same computer room or from different computer rooms (ie, across computer rooms), which is not limited.
  • Fig. 3a is a schematic flowchart of a power consumption control method provided by an exemplary embodiment of the application. This method is described from the perspective of devices that require power consumption control. As shown in Figure 3a, the method includes:
  • the power consumption control scheme used by the device is dynamically updated.
  • these dynamically updated power consumption control schemes may come from power consumption control devices, which are devices that are responsible for dynamically providing power consumption control schemes for physical devices that require power consumption control. For example, they may be those in the foregoing embodiments.
  • the old power consumption control scheme is covered with the new power consumption control scheme, which is simple and easy to implement.
  • the newly arrived power consumption control scheme is referred to as the first power consumption control scheme
  • the power consumption control scheme currently used by the device is referred to as the second power consumption control scheme.
  • the first power consumption control scheme and the second power consumption control scheme both include: trigger conditions and power consumption control methods.
  • the trigger conditions and power control methods in different power control schemes are different.
  • an implementation manner of step 33a includes: monitoring whether the trigger condition in the first power consumption control scheme is met; in the case that the trigger condition in the first power consumption control scheme is met, using the first power consumption The power consumption control method corresponding to the met trigger condition in the control scheme controls the power consumption of the device.
  • the actual performance index data of the device such as actual QoS data
  • the first power consumption can be adjusted according to the actual performance index data of the device.
  • the control intensity of a power consumption control scheme is not limited to actual QoS data.
  • the actual performance data of the device can be compared with the specified performance lower limit; if the actual performance data of the device is less than the specified performance lower limit, the first power consumption control scheme will be shut down until the actual performance index data of the device is greater than or Equal to the lower limit of the specified performance.
  • the actual performance data of the device can be compared with the specified performance lower limit; if the actual performance data of the device is less than the specified performance lower limit, the power consumption control method in the first power consumption control scheme is adjusted to the control intensity Lower power consumption control method until the actual performance index data of the device is greater than or equal to the specified performance lower limit.
  • the specified lower limit of performance may be the lower limit of acceptable performance of the device.
  • the designated performance lower limit value may also be the lower limit value of the performance index data range in the first power consumption control scheme.
  • the power consumption control of the device is combined with the actual performance index data of the device, which can not only save energy consumption, but also ensure the quality of service of the device.
  • FIG. 3b is a schematic flowchart of another power consumption control method provided by an exemplary embodiment of this application. This method is described from the perspective of devices that require power consumption control. As shown in Figure 3b, the method includes:
  • the power consumption control scheme used by the device is dynamically updated.
  • these dynamically updated power consumption control schemes may come from power consumption control devices, which are devices that are responsible for dynamically providing power consumption control schemes for physical devices that require power consumption control. For example, they may be those in the foregoing embodiments.
  • the new power consumption control scheme is merged with the old power consumption control scheme, and the power consumption control is performed based on the merged power consumption control scheme.
  • the power consumption control scheme in this way is more complete, which is conducive to improving power consumption control and improving the energy-saving effect of power consumption control.
  • step 32b the first power consumption control scheme and the second power consumption control scheme may be directly combined to obtain the third power consumption control scheme.
  • the trigger condition in the first power consumption control scheme can be compared with the trigger condition in the second power consumption control scheme; if the trigger condition in the first power consumption control scheme is compared with the second power consumption control scheme The types of trigger conditions in the power control scheme are different, and the first power control scheme and the second power control scheme are combined to obtain the third control scheme; if the trigger condition in the first power control scheme is the same as the second power control scheme If the trigger conditions in the consumption control scheme are of the same category, the old power consumption control scheme is covered with the new power consumption control scheme, that is, the first power consumption control scheme is used as the third control scheme.
  • the third power consumption control scheme includes trigger conditions and power consumption control methods, and there may be a combination of multiple sets of trigger conditions and power consumption control methods. Based on this, the power consumption control of the device according to the third power consumption control scheme includes: when the trigger conditions in the third power consumption control scheme are met, using the trigger conditions that are met in the third power consumption control scheme The corresponding power consumption control method controls the power consumption of the device.
  • the actual performance index data of the device such as actual QoS data
  • the first performance index data can be adjusted according to the actual performance index data of the device. 3. Control intensity of the power consumption control scheme.
  • the actual performance data of the device can be compared with the specified performance lower limit; if the actual performance data of the device is less than the specified performance lower limit, the third power consumption control scheme will be shut down until the actual performance index data of the device is greater than or Equal to the lower limit of the specified performance.
  • the actual performance data of the device can be compared with the specified performance lower limit; if the actual performance data of the device is less than the specified performance lower limit, the power consumption control method in the third power consumption control scheme is adjusted to the control intensity Lower power consumption control method until the actual performance index data of the device is greater than or equal to the specified performance lower limit.
  • the specified lower limit of performance may be the lower limit of acceptable performance of the device.
  • the designated performance lower limit value may also be the lower limit value of the performance index data range in the third power consumption control scheme.
  • combining the actual performance index data of the device and controlling the power consumption of the device can not only save energy consumption, but also ensure the quality of service of the device.
  • FIG. 4a is a schematic flowchart of a method for generating a power consumption control scheme according to an exemplary embodiment of this application. As shown in Figure 4a, the method includes:
  • the target device in this embodiment can be any physical device in the computer room system, or any physical device in the data center system, or any physical device in other clusters, or an independent one. Physical devices.
  • Fig. 4b is a schematic flowchart of another method for generating a power consumption control scheme according to an exemplary embodiment of the application. As shown in Figure 4b, the method includes:
  • the relevant parameters include power consumption index data of the target device in a historical period of time and device parameters of the target device.
  • step 42b According to the device parameters of the target device and the power consumption index data of the target device in the historical period, identify whether the target device is a device whose power consumption change meets the regular requirements; if the identification result is yes, go to step 43b; if the identification result is no , Go to step 45b.
  • step 44b According to the power consumption index data of the target device in the historical period, predict the power consumption index data of the target device in the future period, and perform step 44b.
  • this embodiment introduces the device parameters of the target device to classify the target device; if the target device is a device whose power consumption changes meet the regular requirements, it will be based on the target device’s historical period
  • the power consumption index data of the target device predicts the operation of the power consumption index data of the target device in the future period; if the target device is not a device whose power consumption changes meet the regular requirements, a default power consumption control scheme, such as DVFS, can be set for the target device.
  • DVFS power consumption control scheme
  • FIG. 4c is a schematic flowchart of another method for generating a power consumption control scheme according to an exemplary embodiment of this application. As shown in Figure 4c, the method includes:
  • step 42c According to the device parameters of the target device and the power consumption index data of the target device in the historical period, identify whether the target device is a device whose power consumption change meets the regular requirements; if the identification result is yes, go to step 43c; if the identification result is no , Go to step 45c.
  • step 44c According to the power consumption index data and performance index data of the target device in the historical period, predict the power consumption index data of the target device in the future period, and perform step 44c.
  • this embodiment introduces the performance index data of the target device in the historical period, and combines the power consumption index data of the target device in the future period to predict the power consumption of the target device in the future period.
  • the index data can consider the impact of performance index data on the power consumption index data, which is beneficial to improve the accuracy of the prediction result, and then can generate a more reasonable power consumption control scheme for the target device, and improve the accuracy of power consumption control.
  • an implementation of step 42b or step 42c includes: inputting the device parameters of the target device and the power consumption index data of the target device in the historical period into the classification model to obtain the target Whether the device belongs to the classification result of devices whose power consumption changes meet the regular requirements.
  • the reference distribution feature is the distribution feature of power consumption index data that meets the requirements of the law; according to the reference distribution feature According to the distribution characteristics of the power consumption index data of the target device in the historical period, the target device is classified to obtain the classification result of whether the target device is a device whose power consumption changes meet the requirements of the law.
  • an implementation manner for classifying the target device includes: according to the baseline distribution characteristics and the target device's performance in the historical period.
  • a logistic regression algorithm or a random forest algorithm is used to classify the target device to obtain the classification result of whether the target device is a device whose power consumption changes meet the requirements of the law.
  • an implementation manner of step 42a, step 43b, or step 43c includes: the power consumption index data of the target device in the historical period, or the target device’s power consumption index data in the historical period
  • the power consumption index data and the performance index data are used as input parameters, and the prediction model is input to obtain the power consumption index data of the target device in the future period.
  • a linear regression algorithm or a deep learning algorithm is used to predict the power consumption index data of the target device in the future period.
  • an implementation manner of step 42c, step 44b, or step 44c includes: determining that the preset trigger condition is matched according to the power consumption index data of the target device in the future period The target trigger condition of the target device; obtain the target power consumption control method adapted to the target trigger condition; according to the target trigger condition and the target power consumption control method, generate the power consumption control scheme used by the target device in the future period.
  • the foregoing determination of the matched target trigger condition among the preset trigger conditions according to the power consumption index data of the target device in the future period includes at least one of the following methods:
  • the preset trigger conditions representing the time range determine the trigger condition that the represented time range falls within the time range corresponding to the future time period as the target trigger condition;
  • the preset trigger conditions indicating the power consumption index data range it is determined that the power consumption index data range indicated appears in the power consumption index data of the target device in the future time period
  • the passed trigger condition is used as the target trigger condition.
  • an implementation manner of the foregoing method for acquiring a target power consumption control adapted to a target trigger condition includes: acquiring a preset power consumption control method corresponding to the target trigger condition as the target power consumption control method.
  • another implementation manner of obtaining the target power consumption control method adapted to the target trigger condition includes: simulating the energy saving effect of various power consumption control methods under the target trigger condition; according to various power consumption control methods The energy-saving effect under the target trigger condition is selected from the power consumption control method whose energy-saving effect meets the energy-saving requirement as the target power consumption control method.
  • an implementation manner of generating the power consumption control scheme used by the target device in the future period includes: obtaining a preset target performance index corresponding to the target trigger condition Data range: According to the target trigger condition, target power consumption control method, and target performance index data range, generate the power consumption control scheme used by the target device in the future period.
  • an implementation of step 41a, step 41b, or step 41c includes: if the target device is an IT device, acquiring the internal temperature and power consumption of the target device in the historical period At least one of CPU frequency and CPU load is used as the power consumption index data of the target device in the historical period; if the target device is an air-conditioning device for cooling IT equipment, the compressor frequency of the target device in the historical period is obtained , At least one of the fan speed, return air temperature, and outlet air temperature is used as the power consumption index data of the target device in the historical period.
  • the method further includes: acquiring the power consumption of other devices associated with the target device in a historical period as the power consumption index data of the target device in the historical period.
  • the power consumption control scheme used by the target device in the future period can also be provided to the target device to For the target device to perform power consumption control in the future period.
  • the process of the target device performing power consumption control according to the new power consumption control scheme refer to the description of the embodiment shown in FIG. 3a or FIG. 3b, which will not be repeated here.
  • FIG. 4d is a schematic flowchart of a method for power consumption prediction according to an exemplary embodiment of this application. As shown in Figure 4d, the method includes:
  • artificial intelligence is introduced, and the power consumption index data of the target device can be dynamically predicted through artificial intelligence, which can provide an accurate data basis for other operations that rely on the performance index data of the target device, and help improve the effects of other operations.
  • the relevant parameters of the target device also include device parameters of the target device. Based on this, before predicting the power consumption index data of the target device in the future period according to the power consumption index data of the target device in the historical period, the method further includes: according to the device parameters of the target device and the target device in the historical period To identify whether the target device is a device whose power consumption change meets the law requirements; if the target device is a device whose power consumption change meets the law requirements, execute the target device based on the power consumption indicator data of the target device in the historical period to predict the target The operation of the device's power consumption indicator data in the future period.
  • an implementation manner of classifying the target device according to the device parameters of the target device and the power consumption index data of the target device in the historical period includes: classifying the device parameters of the target device and the target device in The power consumption index data in the historical period is input into the classification model to obtain the classification result of whether the target device belongs to the device whose power consumption change meets the regular requirements.
  • predicting the power consumption index data of the target device in the future time period according to the power consumption index data of the target device in the historical time period includes: inputting the power consumption index data of the target device in the historical time period into the prediction model to Obtain the power consumption index data of the target device in the future period.
  • the relevant parameters of the target device also include performance index data of the target device in a historical period of time. Based on this, the power consumption index data of the target device in the historical period is input into the prediction model to obtain the power consumption index data of the target device in the future period, including: the power consumption index data of the target device in the historical period and the target device The performance index data in the historical period is input into the prediction model to obtain the power consumption index data of the target device in the future period.
  • the prediction model can use a linear regression algorithm or a deep learning algorithm, which is not limited.
  • the power consumption control scheme used by the target device in the future period may be generated according to the power consumption index data of the target device in the future period.
  • the power consumption control scheme used by the target device in the future period may be generated according to the power consumption index data of the target device in the future period.
  • the execution subject of each step of the method provided in the foregoing embodiment may be the same device, or different devices may also be the execution subject of the method.
  • the execution subject of steps 31a to 33a can be device A; for example, the execution subject of steps 31a and 32a can be device A, and the execution subject of step 33a can be device B; and so on.
  • Fig. 5a is a schematic structural diagram of a computing device provided by an exemplary embodiment of this application. As shown in Fig. 5a, the computing device includes a memory 51a and a processor 52a.
  • the memory 51a is used to store computer programs, and can be configured to store various other data to support operations on the computing device. Examples of such data include instructions, messages, pictures, videos, etc. for any application or method operating on the computing device.
  • the processor 52a coupled with the memory 51a, is configured to execute the computer program in the memory 51a to obtain relevant parameters of the target device, the relevant parameters including the power consumption index data of the target device in the historical period; according to the historical period of the target device
  • the power consumption index data in the time period predicts the power consumption index data of the target device in the future time period; according to the power consumption index data of the target device in the future time period, the power consumption control scheme used by the target device in the future time period is generated.
  • the target device refers to the physical device that needs to be controlled for power consumption. It can be any physical device in the computer room system, or any physical device in the data center system, or any physical device in the cluster, or An independent physical device.
  • the relevant parameters of the target device also include device parameters of the target device.
  • the processor 52a is also used to: before predicting the power consumption index data of the target device in the future period, according to the device parameters of the target device and the power consumption index data of the target device in the historical period, identify whether the target device is a function. Equipment whose power consumption changes meet the regular requirements; and when the target device is a device whose power consumption changes meet the regular requirements, perform the prediction of the power consumption index of the target device in the future based on the power consumption index data of the target device in the historical period Data manipulation.
  • the processor 52a when the processor 52a classifies the target device, it is specifically configured to: input the device parameters of the target device and the power consumption index data of the target device in the historical period into the classification model to obtain whether the target device is It belongs to the classification result of devices whose power consumption changes meet the regular requirements.
  • the processor 52a when the processor 52a obtains the classification result of whether the target device belongs to a device whose power consumption change meets the regular requirements, it is specifically configured to: in the classification model, determine the device type to which the target device belongs according to the device parameters of the target device.
  • the benchmark distribution characteristics are the distribution characteristics of the power consumption index data that meet the requirements of the law; according to the benchmark distribution characteristics and the distribution characteristics of the power consumption index data of the target device in the historical period, the target device is classified to obtain Whether the target device belongs to the classification result of devices whose power consumption changes meet the regular requirements.
  • the processor 52a classifies the target device according to the baseline distribution characteristics and the distribution characteristics of the target device's power consumption index data in the historical period, it is specifically configured to: according to the baseline distribution characteristics and the target device's historical period According to the distribution characteristics of the power consumption index data, the logistic regression algorithm or the random forest algorithm is used to classify the target device to obtain the classification result of whether the target device is a device whose power consumption changes meet the requirements of the law.
  • the processor 52a when predicting the power consumption index data of the target device in the future period, is specifically configured to: input the power consumption index data of the target device in the historical period into the prediction model to obtain the target device Power consumption indicator data in the future period.
  • the processor 52a is specifically configured to: in the prediction model, according to the power consumption index data of the target device in the historical period, adopt a linear regression algorithm or a deep learning algorithm to predict the power consumption index of the target device in the future period. data.
  • the relevant parameters of the target device also include performance index data of the target device in a historical period of time.
  • the processor 52a is specifically configured to: input the power consumption index data of the target device in the historical period and the performance index data of the target device in the historical period into the prediction model to obtain the power consumption index data of the target device in the future period. .
  • the processor 52a when the processor 52a generates the power consumption control scheme used by the target device in the future period, it is specifically configured to: determine the preset trigger condition according to the power consumption index data of the target device in the future period The target trigger condition being matched; the target power consumption control method adapted to the target trigger condition is acquired; the power consumption control scheme used by the target device in the future period is generated according to the target trigger condition and the target power consumption control method.
  • the processor 52a is specifically configured to perform at least one of the following operations when determining the target trigger condition that is matched in the preset trigger condition:
  • the preset trigger conditions representing the time range determine the trigger condition that the represented time range falls within the time range corresponding to the future time period as the target trigger condition;
  • the preset trigger conditions indicating the power consumption index data range it is determined that the power consumption index data range indicated appears in the power consumption index data of the target device in the future time period
  • the passed trigger condition is used as the target trigger condition.
  • the processor 52a when the processor 52a obtains the target power consumption control method adapted to the target trigger condition, it is specifically configured to obtain a preset power consumption control method corresponding to the target trigger condition as the target power consumption control method.
  • the processor 52a acquires the target power consumption control method adapted to the target trigger condition, it is specifically configured to: simulate the energy saving effect of various power consumption control methods under the target trigger condition; and control according to various power consumption The method selects the energy-saving effect under the target trigger condition, and selects the power control method whose energy-saving effect meets the energy-saving requirement as the target power control method.
  • the processor 52a when the processor 52a generates the power consumption control scheme used by the target device in the future period, it is specifically configured to: obtain a preset target performance index data range corresponding to the target trigger condition; The power consumption control method and the target performance index data range are used to generate the power consumption control scheme used by the target device in the future period.
  • the processor 52a when the processor 52a obtains the power consumption indicator data of the target device in the historical period, it is specifically configured to: if the target device is an IT-type device, obtain the internal temperature, At least one of power consumption, CPU frequency, and CPU load is used as the power consumption index data of the target device in the historical period; if the target device is an air-conditioning device for cooling IT equipment, the compressor of the target device in the historical period is obtained At least one of the frequency, fan speed, return air temperature, and outlet air temperature is used as the power consumption index data of the target device in the historical period.
  • the processor 52a is further configured to: obtain the power consumption of other devices associated with the target device in the historical period as the power consumption indicator of the target device in the historical period data.
  • the processor 52a is further configured to: after obtaining the power consumption control scheme used by the target device in the future time period, provide the power consumption control scheme used by the target device in the future time period to the target device, so as to For the target device to perform power consumption control in the future period.
  • the computing device further includes: a communication component 53a, a display 54a, a power supply component 55a, an audio component 56a and other components. Only part of the components are schematically shown in FIG. 5a, which does not mean that the computing device only includes the components shown in FIG. 5a. In addition, according to different implementation forms of computing devices, the components in the dashed box in FIG. 5a are optional components, not mandatory components.
  • the computing device when the computing device is implemented as a terminal device such as a smart phone, a tablet computer, or a desktop computer, it can include the components in the dashed box in Figure 5a; when the computing device is implemented as a server such as a conventional server, a cloud server, a data center, or a server array When the device is used, the components in the dashed box in Figure 5a may not be included.
  • the computing device abstracts power control as a solution, and dynamically updates the power control solution of each physical device through artificial intelligence to achieve dynamic power consumption control, and the power consumption control can be accurate to each physical device.
  • the physical device can implement a power consumption control scheme adapted to its own power consumption, which not only improves the accuracy of power consumption control, but also improves the overall energy-saving effect of power consumption control.
  • an embodiment of the present application also provides a computer-readable storage medium storing a computer program, which can implement each step in the method embodiment shown in FIGS. 4a to 4c when the computer program is executed.
  • Fig. 5b is a schematic structural diagram of another computing device provided by an exemplary embodiment of this application. As shown in Fig. 5b, the computing device includes a memory 51b and a processor 52b.
  • the memory 51b is used to store computer programs, and can be configured to store various other data to support operations on the computing device. Examples of such data include instructions, messages, pictures, videos, etc. for any application or method operating on the computing device.
  • the processor 52b is coupled to the memory 51b, and is configured to execute the computer program in the memory 51b to obtain related parameters of the device group, the related parameters include power consumption index data of the device group in a historical period, and the device group includes at least one A physical device; according to the power consumption index data of the device group in the historical period, predict the power consumption index data of the device group in the future period; according to the power consumption index data of the device group in the future period, generate the device group in the future period The power consumption control scheme used.
  • the device group includes at least one physical device, and each physical device in the device group forms a binding relationship and can share the same power consumption control scheme.
  • the method of forming the device group is not limited.
  • physical devices that are relatively close in the computer room can be divided into one device group, or physical devices with the same or similar load conditions can be divided into one device group, or physical devices of the same type can be divided into one device group, or The physical devices of the same model are divided into one device group, or the physical devices of the same manufacturer are divided into one device group, or the physical devices in the entire computer room are divided into one device group, and so on.
  • the power consumption index data of the device group in the historical period includes: power consumption index data of each physical device in the device group in the historical period.
  • the predicted power consumption index data of the device group in the future time period includes: the power consumption index data of each physical device in the device group in the future time period.
  • the power consumption index data of the device group in the historical period is power consumption index data calculated based on the power consumption index data of each physical device in the device group in the historical period.
  • the power consumption index data of the device group in the historical period may be some discrete values or continuous values, including the power consumption index data at each historical moment in the historical period.
  • the power consumption index data of a device group at a certain historical moment may be the average of the power consumption index data of each physical device in the device group at the same historical moment, or the maximum value, or the minimum value. Or the mean of the maximum and minimum among them, and so on.
  • the power consumption index data of the device group in the future period is predicted based on the power consumption index data of the device group in the historical period, which may be some discrete values or continuous values.
  • the related parameters of the device group further include: device parameters of the device group.
  • the device parameters of a device group mainly refer to some static parameters related to the physical devices in the device group, such as device type, device model, device serial number, and device specifications.
  • the processor 52b is also used to: obtain the device parameters of the device group, and according to the device parameters of the device group and the power consumption index data of the device group in the historical period, identify whether the device group belongs to the device whose power consumption changes meet the regular requirements. Group; and in the case that the device group belongs to the device group whose power consumption changes meet the regular requirements, perform the operation of predicting the power consumption index data of the device group in the future period based on the power consumption index data of the device group in the historical period.
  • the processor 52b is specifically configured to: use a classification model to classify the device group.
  • the classification model may use a logistic regression algorithm or a random forest algorithm to classify the device group, so as to obtain the classification result of whether the device group belongs to the device group whose power consumption changes meet the regular requirements.
  • the processor 52b is specifically configured to input the power consumption index data of the device group in the historical period into the prediction model to obtain the power consumption index data of the device group in the future period.
  • the predictive model can use a linear regression algorithm or a deep learning algorithm, but it is not limited to this.
  • the computing device further includes: a communication component 53b, a display 54b, a power supply component 55b, an audio component 56b and other components. Only some of the components are schematically shown in FIG. 5b, which does not mean that the computing device only includes the components shown in FIG. 5b. In addition, according to different implementation forms of computing devices, the components in the dashed box in FIG. 5b are optional components, not mandatory components.
  • the computing device when the computing device is implemented as a terminal device such as a smart phone, a tablet computer, or a desktop computer, it can include the components in the dashed box in Figure 5b; when the computing device is implemented as a server such as a conventional server, a cloud server, a data center, or a server array, etc. When the device is used, the components in the dashed box in Figure 5b may not be included.
  • the computing device provided in this embodiment abstracts power control as a solution, and dynamically updates the power control solution for each device group through artificial intelligence to achieve dynamic power consumption control, and the power consumption control can be accurate to the device group and different device groups.
  • the power consumption control scheme that can be adapted to its own power consumption situation can not only improve the accuracy of power consumption control, but also improve the overall energy-saving effect of power consumption control.
  • an embodiment of the present application also provides a computer-readable storage medium storing a computer program, which can implement the operations in the above-mentioned device group-related embodiments when the computer program is executed.
  • Fig. 6a is a schematic structural diagram of a physical device provided by an exemplary embodiment of this application. As shown in FIG. 6a, the physical device includes: a memory 61a, a processor 62a, and a communication component 63a.
  • the memory 61a is used to store computer programs, and can be configured to store other various data to support operations on the physical device. Examples of these data include instructions, messages, pictures, videos, various power consumption schemes, etc. for any application or method operating on the physical device.
  • the processor 62a coupled with the memory 61a, is configured to execute the computer program in the memory 61a, so as to: receive the newly arrived first power consumption control scheme through the communication component 63a; replace the currently used second power consumption control scheme with The first power consumption control scheme; power consumption control of the device according to the first power consumption control scheme.
  • the first power consumption control scheme includes trigger conditions and power consumption control methods.
  • the processor 62a when the processor 62a controls the power consumption of the device according to the first power consumption control scheme, it is specifically configured to: use the first power consumption control scheme when the trigger condition in the first power consumption control scheme is met.
  • a power consumption control method corresponding to the met trigger condition in a power consumption control scheme controls the power consumption of the device.
  • the processor 62a is further configured to: in the process of controlling the power consumption of the device according to the first power consumption control scheme, monitor the actual performance index data of the device; adjust according to the actual performance index data of the device The control intensity of the first power consumption control scheme.
  • the processor 62a when the processor 62a adjusts the control intensity of the first power consumption control scheme, it is specifically configured to: if the actual performance index data of the device is less than the specified performance lower limit, turn off the first power consumption control Plan until the actual performance index data of the device is greater than or equal to the specified lower limit of performance; or, if the actual performance index data of the device is less than the specified lower limit of performance, adjust the power consumption control method in the first power control plan In order to control low-intensity power consumption control methods, until the actual performance index data of the device is greater than or equal to the specified lower limit of performance; among them, the specified lower limit of performance is the lower limit of performance acceptable to the device, or is the first The lower limit of the performance index data range in the power control scheme.
  • the physical device further includes: a display 64a, a power supply component 66a, an audio component 66a and other components. Only part of the components are schematically shown in FIG. 6a, which does not mean that the physical device only includes the components shown in FIG. 6a. In addition, according to the different implementation forms of the physical device, the components in the dashed box in FIG. 6a are optional components, not mandatory components.
  • the physical device when the physical device is implemented as a terminal device such as a smart phone, a tablet computer, or a desktop computer, it can include the components in the dashed box in Figure 6a; when the physical device is implemented as a server such as a conventional server, a cloud server, a data center, or a server array When the device is used, the components in the dashed box in Figure 6a may not be included.
  • the power consumption control scheme used by it will be dynamically updated. Whenever a new power consumption control scheme arrives, the old power consumption control scheme is covered with the new power consumption control scheme. This method is simple and easy. Implement.
  • an embodiment of the present application also provides a computer-readable storage medium storing a computer program, which can implement each step in the method embodiment shown in FIG. 3a when the computer program is executed.
  • Fig. 6b is a schematic structural diagram of a physical device provided by an exemplary embodiment of this application.
  • the physical device includes: a memory 61b, a processor 62b, and a communication component 63b.
  • the memory 61b is used to store computer programs, and can be configured to store other various data to support operations on the physical device. Examples of these data include instructions for any application or method operating on the physical device, messages, pictures, videos, various power consumption schemes, etc.
  • the processor 62b is coupled with the memory 61b and is configured to execute the computer program in the memory 61b for: receiving the newly arrived first power consumption control scheme through the communication component 63b; according to the first power consumption control scheme and the currently used first power consumption control scheme
  • the second power consumption control scheme is to obtain the third power consumption control scheme; the currently used second power consumption control scheme is replaced with the third power consumption control scheme, and the power consumption of the device is controlled according to the third power consumption control scheme.
  • the processor 62b when the processor 62b obtains the third power consumption control scheme, it is specifically configured to: if the trigger condition in the first power consumption control scheme is of a different type from the trigger condition in the second power consumption control scheme , Combine the first power control scheme and the second power control scheme to obtain the third control scheme; if the trigger conditions in the first power control scheme are of the same type as the trigger conditions in the second power control scheme, The first power consumption control scheme is taken as the third control scheme.
  • the processor 62b when the processor 62b controls the power consumption of the device according to the third power consumption control scheme, it is specifically configured to: use the third power consumption control scheme when the trigger condition in the third power consumption control scheme is met.
  • the power consumption control method corresponding to the met trigger condition in the power consumption control scheme controls the power consumption of the device.
  • the processor 62b is further configured to: in the process of controlling the power consumption of the device according to the third power consumption control scheme, monitor the actual performance index data of the device; adjust according to the actual performance index data of the device The control intensity of the third power consumption control scheme.
  • the processor 62b when the processor 62b adjusts the control intensity of the third power consumption control scheme, it is specifically configured to: if the actual performance index data of the device is less than the specified performance lower limit, shut down the third power consumption control Plan until the actual performance index data of the device is greater than or equal to the specified lower limit of performance; or, if the actual performance index data of the device is less than the specified lower limit of performance, adjust the power consumption control method in the third power control plan In order to control low-intensity power consumption control methods, until the actual performance index data of the device is greater than or equal to the specified lower limit of performance; among them, the specified lower limit of performance is the lower limit of acceptable performance for the device, or the third The lower limit of the performance index data range in the power control scheme.
  • the physical device further includes: a display 64b, a power supply component 66b, an audio component 66b and other components. Only part of the components are schematically shown in FIG. 6b, which does not mean that the physical device only includes the components shown in FIG. 6b. In addition, according to different implementation forms of physical devices, the components in the dashed box in FIG. 6b are optional components, not mandatory components.
  • the physical device when the physical device is implemented as a terminal device such as a smart phone, a tablet computer, or a desktop computer, it can include the components in the dashed box in Figure 6b; when the physical device is implemented as a server such as a conventional server, a cloud server, a data center, or a server array When the device is used, the components in the dashed box in Figure 6b may not be included.
  • the power consumption control scheme used will be dynamically updated. Whenever a new power consumption control scheme arrives, the new power consumption control scheme is merged with the old power consumption control scheme, based on the merged The power consumption control scheme performs power consumption control.
  • the power consumption control scheme in this way is more complete, which is conducive to improving power consumption control and improving the energy-saving effect of power consumption control.
  • the embodiment of the present application also provides a computer-readable storage medium storing a computer program, which can implement each step in the method embodiment shown in FIG. 3b when the computer program is executed.
  • Fig. 7 is a schematic structural diagram of yet another computing device provided by an exemplary embodiment of the present application. As shown in FIG. 7, the computing device includes: a memory 71 and a processor 72.
  • the memory 71 is used to store computer programs, and can be configured to store various other data to support operations on the computing device. Examples of such data include instructions, messages, pictures, videos, etc. for any application or method operating on the computing device.
  • the processor 72 is coupled with the memory 71, and is configured to execute the computer program in the memory 71 to obtain relevant parameters of the target device, the relevant parameters including the power consumption index data of the target device in the historical period;
  • the power consumption index data in the time period predicts the power consumption index data of the target device in the future time period.
  • the target device refers to the physical device that needs to be predicted for power consumption. It can be any physical device in the computer room system, or any physical device in the data center system, or any physical device in the cluster, or An independent physical device.
  • the relevant parameters of the target device also include device parameters of the target device.
  • the processor 72 is also used to: before predicting the power consumption index data of the target device in the future period, according to the device parameters of the target device and the power consumption index data of the target device in the historical period, identify whether the target device is a function. Equipment whose power consumption changes meet the regular requirements; and when the target device is a device whose power consumption changes meet the regular requirements, perform the prediction of the power consumption index of the target device in the future based on the power consumption index data of the target device in the historical period Data manipulation.
  • the processor 72 when the processor 72 classifies the target device, it is specifically configured to: input the device parameters of the target device and the power consumption index data of the target device in the historical period into the classification model to obtain whether the target device is It belongs to the classification result of devices whose power consumption changes meet the regular requirements.
  • the processor 72 when it obtains the classification result of whether the target device belongs to the device whose power consumption change meets the regular requirements, it is specifically configured to: in the classification model, determine the device type to which the target device belongs according to the device parameters of the target device.
  • the benchmark distribution characteristics are the distribution characteristics of the power consumption index data that meet the requirements of the law; according to the benchmark distribution characteristics and the distribution characteristics of the power consumption index data of the target device in the historical period, the target device is classified to obtain Whether the target device belongs to the classification result of devices whose power consumption changes meet the regular requirements.
  • the processor 72 classifies the target device according to the baseline distribution characteristics and the distribution characteristics of the target device's power consumption index data in the historical period, it is specifically configured to: according to the baseline distribution characteristics and the target device's historical period According to the distribution characteristics of the power consumption index data, the logistic regression algorithm or the random forest algorithm is used to classify the target device to obtain the classification result of whether the target device is a device whose power consumption changes meet the requirements of the law.
  • the processor 72 when predicting the power consumption index data of the target device in the future period, is specifically configured to: input the power consumption index data of the target device in the historical period into the prediction model to obtain the target device Power consumption indicator data in the future period.
  • the processor 72 is specifically configured to: in the prediction model, according to the power consumption index data of the target device in the historical period, adopt a linear regression algorithm or a deep learning algorithm to predict the power consumption index of the target device in the future period. data.
  • the relevant parameters of the target device also include performance index data of the target device in a historical period of time.
  • the processor 72 is specifically configured to: input the power consumption index data of the target device in the historical period and the performance index data of the target device in the historical period into the prediction model to obtain the power consumption index data of the target device in the future period. .
  • the processor 72 when the processor 72 obtains the power consumption index data of the target device in the historical period, it is specifically configured to: if the target device is an IT-type device, obtain the internal temperature, At least one of power consumption, CPU frequency, and CPU load is used as the power consumption index data of the target device in the historical period; if the target device is an air-conditioning device for cooling IT equipment, the compressor of the target device in the historical period is obtained At least one of the frequency, fan speed, return air temperature, and outlet air temperature is used as the power consumption index data of the target device in the historical period.
  • the processor 72 is further configured to: obtain the power consumption of other devices associated with the target device in the historical period as the power consumption indicator of the target device in the historical period data.
  • the computing device further includes: a communication component 73, a display 74, a power supply component 75, an audio component 76 and other components. Only some components are schematically shown in FIG. 7, which does not mean that the computing device only includes the components shown in FIG. 7.
  • the components in the dashed box in FIG. 7 are optional components, not mandatory components.
  • the computing device when the computing device is implemented as a terminal device such as a smart phone, a tablet computer, or a desktop computer, it can include the components in the dashed box in Figure 7; when the computing device is implemented as a server such as a conventional server, a cloud server, a data center, or a server array, etc.
  • the components in the dashed box in Figure 7 may not be included.
  • the computing device provided in this embodiment introduces artificial intelligence, which can dynamically predict the power consumption index data of the target device through artificial intelligence, which can provide an accurate data basis for other operations that rely on the performance index data of the target device, and help improve the performance of other operations. effect.
  • the embodiment of the present application also provides a computer-readable storage medium storing a computer program, which can implement each step in the method embodiment shown in FIG. 4d when the computer program is executed.
  • the above-mentioned communication components in Figures 5a-7 are configured to facilitate wired or wireless communication between the device where the communication component is located and other devices.
  • the device where the communication component is located can access wireless networks based on communication standards, such as WiFi, 2G, 3G, 4G/LTE, 5G and other mobile communication networks, or a combination of them.
  • the communication component receives a broadcast signal or broadcast related information from an external broadcast management system via a broadcast channel.
  • the communication component may further include a near field communication (NFC) module, radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra wideband (UWB) technology, Bluetooth (BT) technology, and the like.
  • NFC near field communication
  • RFID radio frequency identification
  • IrDA infrared data association
  • UWB ultra wideband
  • Bluetooth Bluetooth
  • the above-mentioned display in FIGS. 5a-7 includes a screen, and the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user.
  • the touch panel includes one or more touch sensors to sense touch, sliding, and gestures on the touch panel. The touch sensor can not only sense the boundary of the touch or slide action, but also detect the duration and pressure related to the touch or slide operation.
  • the power supply components in Figures 5a-7 above provide power for various components of the equipment where the power supply component is located.
  • the power supply component may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for the device where the power supply component is located.
  • the audio components in Figs. 5a-7 may be configured to output and/or input audio signals.
  • the audio component includes a microphone (MIC).
  • the microphone When the device where the audio component is located is in an operating mode, such as call mode, recording mode, and voice recognition mode, the microphone is configured to receive external audio signals.
  • the received audio signal can be further stored in a memory or sent via a communication component.
  • the audio component further includes a speaker for outputting audio signals.
  • the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.
  • computer-usable storage media including but not limited to disk storage, CD-ROM, optical storage, etc.
  • These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing equipment to work in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including the instruction device.
  • the device implements the functions specified in one process or multiple processes in the flowchart and/or one block or multiple blocks in the block diagram.
  • These computer program instructions can also be loaded on a computer or other programmable data processing equipment, so that a series of operation steps are executed on the computer or other programmable equipment to produce computer-implemented processing, so as to execute on the computer or other programmable equipment.
  • the instructions provide steps for implementing the functions specified in one process or multiple processes in the flowchart and/or one block or multiple blocks in the block diagram.
  • the computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
  • processors CPUs
  • input/output interfaces network interfaces
  • memory volatile and non-volatile memory
  • the memory may include non-permanent memory in a computer readable medium, random access memory (RAM) and/or non-volatile memory, such as read-only memory (ROM) or flash memory (flash RAM). Memory is an example of computer readable media.
  • RAM random access memory
  • ROM read-only memory
  • flash RAM flash memory
  • Computer-readable media include permanent and non-permanent, removable and non-removable media, and information storage can be realized by any method or technology.
  • the information can be computer-readable instructions, data structures, program modules, or other data.
  • Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, CD-ROM, digital versatile disc (DVD) or other optical storage, Magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media can be used to store information that can be accessed by computing devices. According to the definition in this article, computer-readable media does not include transitory media, such as modulated data signals and carrier waves.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

一种功耗控制与方案生成方法、设备、系统及存储介质，将功率控制抽象为方案，通过人工智能可动态更新每台物理设备或设备组的功率控制方案，实现动态功耗控制，而且可使功耗控制精确到每台物理设备或设备组，不仅可提高功耗控制的精度，还可以提高功耗控制的整体节能效果。

Description

功耗控制与方案生成方法、设备、系统及存储介质 技术领域
本申请涉及数据中心技术领域，尤其涉及一种功耗控制与方案生成方法、设备、系统及存储介质。
背景技术
随着云计算技术的发展，各类数据中心(Data Center，DC)不断被部署，数据中心的能耗问题也日益突出。其中，各类功耗控制方法成为数据中心用于降低能耗的首选方案。
在数据中心中，为了降低能耗，会预置一些功耗控制方法，例如DVFS等，当功耗控制方法对应的条件被触发时，功耗控制方法会被启动，以降低设备能耗。但是，现有功耗控制方案的节能效果较差。
发明内容
本申请的多个方面提供一种功耗控制与方案生成方法、设备、系统及存储介质，用以实现动态功耗控制，提高功耗控制的节能效果。
本申请实施例提供一种功耗控制方案生成方法，包括：获取目标设备的相关参数，所述相关参数包括所述目标设备在历史时段内的功耗指标数据；根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据；根据所述目标设备在未来时段内的功耗指标数据，生成所述目标设备在未来时段内使用的功耗控制方案。
本申请实施例还提供一种功耗控制方案生成方法，包括：获取设备组的相关参数，所述相关参数包括所述设备组在历史时段内的功耗指标数据，所 述设备组包括至少一台物理设备；根据所述设备组在历史时段内的功耗指标数据，预测所述设备组在未来时段内的功耗指标数据；根据所述设备组在未来时段内的功耗指标数据，生成所述设备组在未来时段内使用的功耗控制方案。
本申请实施例还提供一种功耗控制方法，包括：接收新到达的第一功耗控制方案；将当前使用的第二功耗控制方案替换为所述第一功耗控制方案；根据所述第一功耗控制方案对设备进行功耗控制。
本申请实施例还提供一种功耗控制方法，包括：接收新到达的第一功耗控制方案；根据所述第一功耗控制方案和当前使用的第二功耗控制方案，得到第三功耗控制方案；将当前使用的第二功耗控制方案替换为所述第三功耗控制方案，根据所述第三功耗控制方案对设备进行功耗控制。
本申请实施例还提供一种功耗预测方法，包括：获取目标设备的相关参数，所述相关参数包括所述目标设备在历史时段内的功耗指标数据；根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据。
本申请实施例还提供一种计算设备，包括：存储器和处理器；所述存储器，用于存储计算机程序；所述处理器，与所述存储器耦合，用于执行与所述计算机程序以用于：获取目标设备的相关参数，所述相关参数包括所述目标设备在历史时段内的功耗指标数据；根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据；根据所述目标设备在未来时段内的功耗指标数据，生成所述目标设备在未来时段内使用的功耗控制方案。
本申请实施例还提供一种计算设备，包括：存储器和处理器；所述存储器，用于存储计算机程序；所述处理器，与所述存储器耦合，用于执行与所述计算机程序以用于：获取设备组的相关参数，所述相关参数包括所述设备组在历史时段内的功耗指标数据，所述设备组包括至少一台物理设备；根据所述设备组在历史时段内的功耗指标数据，预测所述设备组在未来时段内的 功耗指标数据；根据所述设备组在未来时段内的功耗指标数据，生成所述设备组在未来时段内使用的功耗控制方案。
本申请实施例还提供一种物理设备，包括：存储器、处理器以及通信组件；所述通信组件，用于接收新到达的第一功耗控制方案；所述存储器，用于存储计算机程序、所述第一功耗控制方案以及当前使用的第二功耗控制方案；所述处理器，与所述存储器耦合，用于执行与所述计算机程序以用于：将当前使用的第二功耗控制方案替换为所述第一功耗控制方案；根据所述第一功耗控制方案对所述物理设备进行功耗控制。
本申请实施例还提供一种物理设备，包括：存储器、处理器以及通信组件；所述通信组件，用于接收新到达的第一功耗控制方案；所述存储器，用于存储计算机程序、所述第一功耗控制方案以及当前使用的第二功耗控制方案；所述处理器，与所述存储器耦合，用于执行与所述计算机程序以用于：根据所述第一功耗控制方案和当前使用的第二功耗控制方案，得到第三功耗控制方案；将当前使用的第二功耗控制方案替换为所述第三功耗控制方案，根据所述第三功耗控制方案对所述物理设备进行功耗控制。
本申请实施例还提供一种计算设备，包括：存储器和处理器；所述存储器，用于存储计算机程序；所述处理器，与所述存储器耦合，用于执行与所述计算机程序以用于：获取目标设备的相关参数，所述相关参数包括所述目标设备在历史时段内的功耗指标数据；根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据。
本申请实施例还提供一种数据中心系统，包括：至少一个机房和功耗控制设备；其中，每个机房包括至少一台物理设备；所述功耗控制设备，用于获取目标设备的相关参数，所述相关参数包括所述目标设备在历史时段内的功耗指标数据；根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据；根据所述目标设备在未来时段内的功耗指标数据，生成所述目标设备在未来时段内使用的功耗控制方案并提供给所述目标设备，以供所述目标设备在未来时段进行功耗控制；其中，所述 目标设备是所述至少一台物理设备中任一台设备。
本申请实施例还提供一种机房系统，包括：机房，所述机房内包含至少一台物理设备和功耗控制设备；所述功耗控制设备，用于获取目标设备的相关参数，所述相关参数包括所述目标设备在历史时段内的功耗指标数据；根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据；根据所述目标设备在未来时段内的功耗指标数据，生成所述目标设备在未来时段内使用的功耗控制方案并提供给所述目标设备，以供所述目标设备在未来时段进行功耗控制；其中，所述目标设备是所述至少一台物理设备中任一台设备。
本申请实施例还提供一种存储有计算机程序的计算机可读存储介质，当所述计算机程序被处理器执行时，致使所述处理器实现本申请各方法实施例中的步骤。
在本申请实施例中，将功率控制抽象为方案，通过人工智能动态更新每台物理设备或设备组的功率控制方案，实现动态功耗控制，而且可使功耗控制精确到每台物理设备或设备组，不同物理设备或设备组可进行与自身功耗情况适配的功耗控制方案，不仅可提高功耗控制的精度，还可以提高功耗控制的整体节能效果。
附图说明
此处所说明的附图用来提供对本申请的进一步理解，构成本申请的一部分，本申请的示意性实施例及其说明用于解释本申请，并不构成对本申请的不当限定。在附图中：
图1a为本申请示例性实施例提供的一种机房系统的结构示意图；
图1b为本申请示例性实施例提供的功耗控制设备的工作原理框图；
图1c为本申请示例性实施例提供的功耗控制设备的一种内部结构示意图；
图2为本申请示例性实施例提供的一种数据中心系统的结构示意图；
图3a为本申请示例性实施例提供的一种功耗控制方法的流程示意图；
图3b为本申请示例性实施例提供的另一种功耗控制方法的流程示意图；
图4a为本申请示例性实施例提供的一种功耗控制方案生成方法的流程示意图；
图4b为本申请示例性实施例提供的另一种功耗控制方案生成方法的流程示意图；
图4c为本申请示例性实施例提供的又一种功耗控制方案生成方法的流程示意图；
图4d为本申请示例性实施例提供的一种功耗预测方法的流程示意图。
图5a为本申请示例性实施例提供的一种计算设备的结构示意图；
图5b为本申请示例性实施例提供的另一种计算设备的结构示意图；
图6a为本申请示例性实施例提供的一种物理设备的结构示意图；
图6b为本申请示例性实施例提供的另一种物理设备的结构示意图；
图7为本申请示例性实施例提供的又一种计算设备的结构示意图。
具体实施方式
为使本申请的目的、技术方案和优点更加清楚，下面将结合本申请具体实施例及相应的附图对本申请技术方案进行清楚、完整地描述。显然，所描述的实施例仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
针对现有功耗控制方案节能效果较差的技术问题，在本申请一些实施例中，将功率控制抽象为方案，通过人工智能动态更新每台物理设备或设备组的功率控制方案，实现动态功耗控制，而且可使功耗控制精确到每台物理设备或设备组，不同物理设备或设备组可进行与自身功耗情况适配的功耗控制方案，不仅可提高功耗控制的精度，还可以提高功耗控制的整体节能效果。
以下结合附图，详细说明本申请各实施例提供的技术方案。
图1a为本申请示例性实施例提供的一种机房系统的结构示意图。如图1a所示，本实施例的机房系统100包括：机房，机房是指存放机器设备的物理场所，例如可以是一个房间或厂房等。进一步，如图1a所示，机房系统100还包括：位于机房内的至少一个物理设备101和功耗控制设备102。本实施例并不限定机房内物理设备101的数量，可以是一台，也可以是多台。
在本实施例中，并不限定物理设备101的设备形态。如图1a所示，物理设备101可以是机房内的IT类设备和为IT类设备降温的制冷设备，例如空调设备。举例说明，至少一个物理设备101可以包括但不限于：机柜设备、服务器设备、计算机设备、打印机、集线器、电源设备、存储设备、网络交换设备以及空调设备等。服务器设备可以是包括但不限于：常规服务器、服务器阵列或云服务器等。电源设备可以是蓄电池设备、干电池设备、或不间断电源(UPS)等。存储设备可以包括但不限于：磁盘、磁盘阵列、硬盘、网络存储设备(NAS)等。
其中，至少一台物理设备101正常运行会消耗电能，为了降低机房系统100的能耗，有必要对至少一台物理设备101进行功耗控制。在本实施例中，可以为物理设备101配置功耗控制方案，物理设备101根据功耗控制方案进行功耗控制。其中，功耗控制方案是指示物理设备101进行功耗控制的方案，包括触发条件和功耗控制方法。当功耗控制方案中的触发条件被满足时，功耗控制方案中的功耗控制方法会被执行。其中，可以将功率控制方法理解为功率控制动作，则当触发条件被满足时，与该触发条件对应的功率控制动作会被执行。
在本实施例中，不对功耗控制方法进行限定，凡是具有功耗控制、限制或调整作用的方法均适用于本申请实施例。举例说明，功率控制方法包括但不限于：功率封顶(Power Capping)、动态电压功率调整(Dynamic Voltage and Frequency Scaling，DVFS)和C模式(C-State)等。其中，功率封顶是一种限制设备功耗的方法，它可以确保设备的实际功耗低于可用的最大功率，并 且在设备负载较低时可确保设备使用较低功耗。DVFS是根据芯片(如CPU)所运行的应用程序对计算能力的不同需要，动态调节芯片的运行频率和电压，从而达到节能目的的一种动态技术，可对设备上的各种处理器、控制芯片和外围设备的功率和电压进行调整。C-state是一种可以让CPU在空闲状态时进入低功耗状态的低功耗机制，C-states包含的C模式从C0开始一直到Cn，C0是CPU的正常工作模式，CPU处于100％运行状态；C后n的取值越高，CPU睡眠得越深，CPU的功耗越小，当然也就需要更多的时间返回到C0模式；其中，n是正整数。
在一种极端情况下，若机房内至少一台物理设备101在任何时候的功耗情况相同或基本相同，则可以为至少一台物理设备101统一配置功耗控制方案，所有物理设备101使用相同的功耗控制方案进行功耗控制，统一配置功耗控制方案的方式相对简单，易于实施。
但是，在大多数情况下，机房内不同物理设备101的负载情况差异较大，即使是在同一时刻，不同物理设备101的负载也是不同的，所以不同物理设备101的功耗情况是不同的。对于这种情况，如果依旧为至少一台物理设备101统一配置功耗控制方案，让功耗不同的物理设备101按照相同的功耗控制方案进行功耗控制，显然，整体的节能效果不够理想。另外，即使是同一物理设备101，其负载情况在不同时刻也会有所不同，这意味着，同一物理设备101在不同时刻的功耗也会有所不同。
基于上述分析，在本实施例中，引入人工智能，通过人工智能动态更新每台物理设备101的功耗控制方案，实现动态功耗控制，提高功耗控制的节能效果。为了实现动态更新每台物理设备101的功耗控制方案的目的，在机房内增设功耗控制设备102。功耗控制设备102主要以人工智能为基础，动态更新每台物理设备101的功耗控制方案，以供每台物理设备101根据动态变化的功耗控制方案进行功耗控制。
其中，功耗控制设备102为每台物理设备101动态更新功耗控制方案的方式相同或相似。在本实施例中，结合图1b所示原理框图，以其中一台物理 设备101为例，对功耗控制设备102为物理设备101动态更新功耗控制方案的过程进行说明。为便于描述，在下述内容中，将该台物理设备称为目标设备，目标设备是至少一台物理设备101中任一台物理设备。
其中，功耗控制设备102可以获取目标设备的相关参数，该相关参数包括：目标设备在历史时段内的功耗指标数据；根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数；进而，根据目标设备在未来时段内的功耗指标数据，生成目标设备在未来时段内使用的功耗控制方案，并提供给目标设备，以供目标设备在未来时段根据该功耗控制方案进行功耗控制。
在本实施例中，并不限定未来时段的时间长度。例如，未来时段可以是未来半小时(即将到来的半小时)，或未来一个小时(即将到来的一小时)，或未来几小时(即将到来的几个小时)，或未来一天(即将到来的一天)，或未来一周(即将到来的一周)，或未来一小时内的某个时段，或未来一天内的某个时段，或未来一周内的某天或某几天，等等。未来时段的时间长度，可根据应用需求灵活设置。另外，对不同物理设备101来说，未来时段的时间长度可以相同，也可以不相同。
对于每个未来时段，功耗控制设备102都会根据目标设备在相应历史时段内的功耗指标数据，预测目标设备在该未来时段内的功耗指标数据，并根据目标设备在该未来时段内的功耗指标数据，生成目标设备在该未来时段内使用的功耗控制方案。也就是说，功耗控制设备102会根据未来时段的变化，动态更新目标设备在每个未来时段内使用的功耗控制方案；从长期来看，目标设备使用的功耗控制方案是动态变化的，可实现动态功耗控制。
在本实施例中，历史时段是指当前时刻之前的一段时间，例如昨天、昨天上午、昨天下午、上周二、上周一到周三、上周六，上周日等，相对当前时刻都属于历史时段。在本实施例中，并不限定所使用的历史时段的数量，历史时段可以是一个，也可以是多个；另外，也不限定每个历史时段的时间长度，均可根据应用需求适应性设置。同样，本实施例也不限定历史时段与 未来时段之间的对应关系，可根据应用需求适应性设置。下面举例说明：
例如，可以根据目标设备在最近三个月内每个周末的功耗指标数据，预测目标设备在本周周末的功耗指标数据，进而根据目标设备在本周周末的功耗指标数据，生成目标设备在本周周末内的使用的功耗控制方案。
例如，可以根据目标设备在最近一个月内每天的功耗指标数据，预测目标设备在未来一周内的功耗指标数据，进而根据目标设备在未来一周内的功耗指标数据，生成目标设备在未来一周内使用的功耗控制方案。
例如，可以根据目标设备在上周每天上午10点-12点之间的功耗指标数据，预测目标设备在未来一周每天上午10点-12点之间的功耗指标数据；根据目标设备在未来一周每天上午10点-12点之间的功耗指标数据，生成目标设备在未来一周内使用的功耗控制方案。
例如，可以根据目标设备在最近若干天每天白天的功耗指标数据和每天晚上的功耗指标数据，预测目标设备在未来一天内白天的功耗指标数据和晚上的功耗指标数据；根据目标设备在未来一天内白天的功耗指标数据和晚上的功耗指标数据，生成目标设备在未来一天内使用的功耗控制方案。
在本实施例中，功耗指标数据是指与功耗相关的一些数据，可以包括一种数据，也可以包括多种数据。其中，在未来时段内使用的功耗指标数据的种类与在历史时段内使用的功耗指标数据的种类相同；或者，在未来时段内使用的功耗指标数据的种类是在历史时段内使用的功耗指标数据的种类中的一部分。对不同类型的设备，与设备相关的功耗指标数据也会有所不同。
例如，对于IT类设备，IT类设备的内部温度、功耗、CPU频率、CPU负载等都与IT类设备的功耗相关，故可以将IT类设备的内部温度、功耗、CPU频率以及CPU负载等中至少一种数据作为功耗指标数据。
在一可选实施例中，以选择IT类设备的功耗和CPU频率作为功耗指标数据为例，则在目标设备是IT类设备的情况下，功耗控制设备102可以获取目标设备在历史时段内的功耗和CPU频率作为目标设备在历史时段内的功耗指标数据；根据目标设备在历史时段内的功耗和CPU频率，预测目标设备在未 来时段内的功耗和CPU频率；根据目标设备在未来时段内的功耗和CPU频率，生成目标设备在未来时段内使用的功耗控制方案。
在另一可选实施例中，以选择IT类设备的功耗、CPU频率和CPU负载作为功耗指标数据为例，则在目标设备是IT类设备的情况下，功耗控制设备102可以获取目标设备在历史时段内的功耗、CPU频率和CPU负载作为目标设备在历史时段内的功耗指标数据；根据目标设备在历史时段内的功耗、CPU频率和CPU负载，预测目标设备在未来时段内的功耗、CPU频率和CPU负载；根据目标设备在未来时段内的功耗、CPU频率和CPU负载，生成目标设备在未来时段内使用的功耗控制方案。
在又一可选实施例中，以选择IT类设备的内部温度、功耗、CPU频率和CPU负载作为功耗指标数据为例，则在目标设备是IT类设备的情况下，功耗控制设备102可以获取目标设备在历史时段内的内部温度、功耗、CPU频率和CPU负载作为目标设备在历史时段内的功耗指标数据；根据目标设备在历史时段内的内部温度、功耗、CPU频率和CPU负载，预测目标设备在未来时段内的功耗和CPU负载；根据目标设备在未来时段内的功耗和CPU负载，生成目标设备在未来时段内使用的功耗控制方案。
例如，对于空调设备，空调设备的压缩机的频率、风机的转速、回风温度以及出风温度等数据与空调设备的功耗相关，故可以将空调设备的压缩机的频率、风机的转速、回风温度以及出风温度等中至少各种作为功耗指标数据。
在一可选实施例中，以选择空调设备的压缩机的频率和风机的转速作为功耗指标数据为例，则在目标设备是空调设备的情况下，功耗控制设备102可以获取目标设备在历史时段内的压缩机的频率和风机的转速作为目标设备在历史时段内的功耗指标数据；根据目标设备在历史时段内的压缩机的频率和风机的转速，预测目标设备在未来时段内的压缩机的频率和风机的转速；根据目标设备在未来时段内的压缩机的频率和风机的转速，生成目标设备在未来时段内使用的功耗控制方案。
在另一可选实施例中，以选择空调设备的压缩机的频率、风机的转速、回风温度以及出风温度作为功耗指标数据为例，则在目标设备是空调设备的情况下，功耗控制设备102可以获取目标设备在历史时段内的压缩机的频率、风机的转速、回风温度以及出风温度作为目标设备在历史时段内的功耗指标数据；根据目标设备在历史时段内的压缩机的频率、风机的转速、回风温度以及出风温度，预测目标设备在未来时段内的压缩机的频率、风机的转速、回风温度以及出风温度；根据目标设备在未来时段内的压缩机的频率、风机的转速、回风温度以及出风温度，生成目标设备在未来时段内使用的功耗控制方案。
在又一可选实施例中，以选择空调设备的压缩机的频率、风机的转速、回风温度以及出风温度作为功耗指标数据为例，则在目标设备是空调设备的情况下，功耗控制设备102可以获取目标设备在历史时段内的压缩机的频率、风机的转速、回风温度以及出风温度作为目标设备在历史时段内的功耗指标数据；根据目标设备在历史时段内的压缩机的频率、风机的转速、回风温度以及出风温度，预测目标设备在未来时段内的回风温度以及出风温度；根据目标设备在未来时段内的回风温度以及出风温度，生成目标设备在未来时段内使用的功耗控制方案。
进一步可选地，对于IT类设备，有些情况下，其功耗还会受与其关联的其它设备的影响，例如对于位于机架上的服务器，机架的功耗会有一部分传递给服务器。基于此，在目标设备为IT类设备的情况下，除了获取目标设备自身与功耗相关的各种功耗指标数据之外，还可以根据目标设备与其它设备之间的拓扑结构，确定与目标设备关联的其它设备，获取与目标设备关联的其它设备在历史时段内的功耗，也作为目标设备在历史时段内的一种功耗指标数据。其中，与目标设备关联的其它设备是指对目标设备的功耗有影响的设备。
在本实施例中，并不限定功耗控制设备102以何种方式来获取目标设备在历史时段内的功耗指标数据。在一些可选实施例中，功耗控制设备102可 以提供人机交互界面，机房的管理人员可以通过该人机交互界面向功耗控制设备102输入目标设备在历史时段内的功耗指标数据。其中，人机交互界面可以有多种实现方式。例如，人机交互界面可实现为包含输入框的web页面或应用页面，管理人员可以在输入框内输入目标设备在历史时段内的功耗指标数据。或者，人机交互界面可实现为包含功耗指标数据选项下拉框的web页面或应用页面，管理人员可以通过功耗指标数据选项下拉框，从下拉框中选择需要输入的功耗指标数据。在另一些可选实施例中，功耗控制设备102支持配置文件，则管理人员可以通过其它设备(例如管理人员使用的终端设备)向功耗控制设备102发送配置文件，该配置文件中包含目标设备在历史时段内的功耗指标数据；功耗控制设备102可以从配置文件中读取目标设备在历史时段内的功耗指标数据。
对目标设备来说，可获取功耗控制设备102提供的功耗控制方案，并根据该功耗控制方案进行功耗控制。其中，目标设备根据功耗控制方案进行功耗控制的过程包括：监测功耗控制方案中的触发条件是否被满足；在触发条件被满足的情况下，根据功耗控制方案中的功耗控制方法进行功耗控制。
其中，功耗控制方案中的触发条件是触发目标设备启动功耗控制所需的条件。例如，触发条件可以是一时间范围，表示在该时间范围内，需要启动相应的功耗控制方法进行功耗控制。或者，触发条件是功耗指标数据范围，表示当功耗指标数据位于该范围时，需要启动相应的功耗控制方法进行功耗控制。其中，根据功耗指标数据的不同，作为触发条件的功耗指标数据范围也会有所不同。
例如，假设功耗指标数据包括功耗，则功耗指标数据范围相应包括：功耗范围。当目标设备在一小段时间内的实际功耗位于该功耗范围内时，可启动相应功耗控制方法进行功耗控制。进一步可选地，若目标设备在一小段时间内的实际功耗低于该功耗范围的下限值，可以结束功耗控制。
例如，假设功耗指标数据包括功耗和CPU频率，则功耗指标数据范围相应包括：功耗范围和CPU频率范围。在一种实现中，当两个触发条件都满足 时，例如当目标设备在一小段时间内的实际功耗位于功耗范围内且目标设备在一小段时间内的实际CPU频率位于CPU范围内时，启动相应功耗控制方法进行功耗控制。相应地，在两个触发条件中任意一个不再满足时，可以结束功耗控制。在另一种实现中，当任意一个条件被满足时，即当目标设备在一小段时间内的实际功耗位于功耗范围时，或者目标设备在一小段时间内的实际CPU频率位于CPU频率范围时，启动相应功耗控制方法进行功耗控制。相应地，在两个触发条件中均不被满足时，可以结束功耗控制。
在上述实施例中，之所以选用一小段时间内的实际功耗或实际CPU频率，是为了维护功耗控制的稳定性，尽量减少出现乒乓效应的概率。这里的“一小段时间”的时间长度不做限定，可以是10秒钟、1分钟、5分钟等。当然除了利用一小段时间内的实际功耗或实际CPU频率等功耗指标数据作为依据进行功耗控制之外，也可以采用实时的功耗或CPU频率等功耗指标数据作为依据进行功耗控制。
在本实施例中，将功率控制抽象为方案，通过人工智能动态更新每台物理设备的功率控制方案，实现动态功耗控制，而且可使功耗控制可精确到每台物理设备，不同物理设备可进行与自身功耗情况适配的功耗控制方案，不仅可提高功耗控制的精度，还可以提高功耗控制的整体节能效果。
在本申请一些实施例中，功耗控制设备102在获取目标设备在历史时段内的功耗指标数据之后，可以直接根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据。
在本申请另一些实施例中，为了提高预测结果的准确度，可对目标设备进行分类；若目标设备属于功耗变化满足规律要求的设备，则再执行根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据的操作；若目标设备不属于功耗变化满足规律要求的设备，则结束为目标设备生成功耗控制方案的操作，或者为目标设备设置默认的功耗控制方案，例如DVFS。
可选地，为了更加准确地对目标设备进行分类，可以加入目标设备的设备参数。这里的设备参数主要是指一些与设备自身相关的静态参数，例如设备类型、设备型号、设备序列号、设备规格等。其中，目标设备的设备参数可以从目标设备上采集到，也可以通过带外方式获取。基于此，如图1b所示，功耗控制设备102除了获取目标设备在历史时段内的功耗指标数据之外，还需获取目标设备的设备参数，并根据目标设备的设备参数和目标设备在历史时段内的功耗指标数据，识别目标设备是否属于功耗变化满足规律要求的设备。关于功耗控制设备102获取目标设备的设备参数的方式，可参见前述功耗控制设备102获取目标设备在历史时段内的功耗指标数据的方式，在此不再赘述。
进一步，如图1c所示，功耗控制设备102的一种内部结构包括：分类器、预测器和决策器。其中，分类器主要负责根据目标设备的设备参数和目标设备在历史时段内的功耗指标数据，对目标设备进行分类，以识别目标设备是否属于功耗变化满足规律要求的设备，并输出分类结果；若目标设备属于功耗变化满足规律要求的设备，则进入预测器；若目标设备不属于功耗变化满足规律要求的设备，则进入决策器。预测器主要用于根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据，并输出至决策器。决策器用于在目标设备属于功耗变化满足规律要求的设备情况下，根据目标设备在未来时段内的功耗指标数据，生成目标设备在未来时段内使用的功耗控制方案；或者，在目标设备不属于功耗变化满足规律要求的设备的情况下，可为目标设备设置默认的功耗控制方案。
在一可选实施例中，可利用分类模型对目标设备进行分类。可以将目标设备的设备参数和目标设备在历史时段内的功耗指标数据输入分类模型，以得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。
在分类模型内部，可根据目标设备的设备参数，确定与目标设备所属设备类型对应的基准分布特征，该基准分布特征是满足规律要求的功耗指标数据的分布特征；根据基准分布特征和目标设备在历史时段内的功耗指标数据 的分布特征，对目标设备进行分类，以得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。
可选地，可以直接计算基准分布特征和目标设备在历史时段内的功耗指标数据的分布特征的相似度，根据相似度对目标设备进行分类；如果相似度大于设定的相似度阈值，确定目标设备属于功耗变化满足规律要求的设备；如果相似度小于或等于设定的相似度阈值，确定目标设备不属于功耗变化满足规律要求的设备。
可选地，还可以根据基准分布特征和目标设备在历史时段内的功耗指标数据的分布特征，采用逻辑回归算法或随机森林算法对目标设备进行分类，以得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。
以采用逻辑回归算法对目标设备进行分类为例，可以将目标设备是否属于功耗变化满足规律要求的设备作为一个二分类问题，输出为y∈{0,1}，二线性回归模型z＝wTx+b是个实数值，并通过Sigmoid函数将z的值向0或1转化。例如，若Sigmoid函数计算得到的值大于或等于0.5，则归为类别1；若Sigmoid函数计算得到的值小于0.5，则归为类别0。在本实施例中，类别1表示目标设备属于功耗变化满足规律要求的设备；类别0表示目标设备不属于功耗变化满足规律要求的设备。
在上述逻辑回归过程中，可以定义一个代价函数，将基准分布特征作为训练集，以所定义的代价函数为目标，训练(拟合)得到wT的取值；然后，将目标设备在历史时段内的功耗指标数据的分布特征作为x的取值，代入二线性回归模型z＝wTx+b得到z的取值；进而，通过Sigmoid函数将z的值转化为0或1，得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。其中，b是一个常数，一般符合均值为0的正态分布。
以采用随机森林算法对目标设备进行分类为例，可以将基准分布特征作为训练样本，构建多棵决策树，然后利用多棵决策树对目标设备在历史时段内的功耗指标数据的分布特征进行分类，最终根据多棵决策树投票决定目标设备是否属于功耗变化满足规律要求的设备。关于构建多棵决策树的过程可 参见现有技术，在此不再赘述。
在本申请上述实施例或下述实施例中，可以利用人工智能中的机器学习模型，例如预测模型，预测目标设备在未来时段内的功耗指标数据。其中，可将目标设备在历史时段内的功耗指标数据输入预测模型，以得到目标设备在未来时段内的功耗指标数据。
其中，在预测模型内部，可以根据目标设备在历史时段内的功耗指标数据，采用线性回归算法或深度学习算法，预测目标设备在未来时段内的功耗指标数据。关于采用线性回归算法或深度学习算法的预测模型的训练过程与预测过程，与现有技术相同或类似，在此不再赘述。
在本申请一些实施例中，在预测目标设备在未来时段内的功耗指标数据的过程中，除了考虑目标设备在历史时段内的功耗指标数据之外，还可以结合目标设备在历史时段内的性能指标数据，如图1b所示。其中，这里的性能指标数据是指可以反应目标设备服务性能的数据。在目标设备为IT类设备的情况下，目标设备上可能运行有至少一种应用或服务，例如云计算服务、游戏服务，即时通信服务、邮件服务或在线交易服务等等。其中，目标设备的性能指标数据可以是目标设备上运行的应用或服务的QoS数据。例如，对应用或服务来说，QoS数据可以是响应时间、TPS、QPS或并发用户数等。
在一可选实施例中，在目标设备运行过程中，可以采集目标设备的性能指标数据，并存储到数据库中。基于此，可以从数据库中获取目标设备在历史时段内的性能指标数据。
在另一可选实施例中，服务提供商可将目标设备提供给某个用户使用，则为了保障服务的性能和可用性，服务提供商与用户之间会定义一种双方认可的协定，即服务等级协议(Service Level Agreement，SLA)，在该协定中，会约定目标设备需要满足的性能要求。基于此，可以从目标设备对应的SLA协定中，获取目标设备需要满足的性能指标数据，作为目标设备在历史时段内的性能指标数据。
在结合目标设备在历史时段内的性能指标数据的情况下，可以将目标设备在历史时段内的功耗指标数据以及目标设备在历史时段内的性能指标数据输入预测模型，以得到目标设备在未来时段内的功耗指标数据。在本实施例中，考虑性能指标数据对功耗指标数据的影响，有利于提高预测结果的准确性，进而可为目标设备生成更加合理的功耗控制方案，提高功耗控制的准确度。
在本申请上述或下述实施例中，在得到目标设备在未来时段内的功耗指标数据之后，可根据目标设备在未来时段内的功耗指标数据，生成目标设备在未来时段内使用的功耗控制方案。
在本申请一些实施例中，可基于规则生成功耗控制方案，即预先定义一些触发条件，以及触发条件对应的功耗控制方法。基于此，生成目标设备在未来时段内使用的功耗控制方案的过程包括：根据目标设备在未来时段内的功耗指标数据，与预设触发条件进行匹配，确定预设触发条件中被匹配中的目标触发条件；获取预设的与目标触发条件对应的功耗控制方法，如DVFS或C-State，作为目标功耗控制方法；根据目标触发条件与目标功耗控制方法，生成目标设备在未来时段内使用的功耗控制方案。在该实施例中，目标设备在未来时段内使用的功耗控制方案包括：目标触发条件和目标功耗控制方法。
在本申请另一些实施例中，可采用规则与人工智能相结合的方式生成功耗控制方案，即可以预先定义一些触发条件，在触发条件被匹配中的情况下，采用人工智能的方式生成与该触发条件适配的功耗控制方法。基于此，生成目标设备在未来时段内使用的功耗控制方案的过程包括：根据目标设备在未来时段内的功耗指标数据，与预设触发条件进行匹配，确定预设触发条件中被匹配中的目标触发条件；模拟各种功耗控制方法在目标触发条件下的节能效果；根据各种功耗控制方法在目标触发条件下的节能效果，从中选择节能效果满足节能要求的功耗控制方法，作为目标功耗控制方法；根据目标触发条件与目标功耗控制方法，生成目标设备在未来时段内使用的功耗控制方案。 在该实施例中，目标设备在未来时段内使用的功耗控制方案包括：目标触发条件和目标功耗控制方法，例如上午6:00-8:00，执行DVFS。
进一步，无论是在基于规则生成功耗控制方案的实施例中，还是在采用规则与人工智能相结合的方式生成功耗控制方案的实施例中，除了预先定义触发条件之外，还可以预先定义与触发条件对应的性能指标数据范围，这里的性能指标数据范围可理解为是功耗控制的约束条件，是功耗控制过程需要保证的性能要求。值得说明的是，并不是每个触发条件都对应有性能指标数据范围。基于此，在确定目标触发条件之后，还可以判断该目标触发条件是否有对应的性能指标数据范围；若有，则获取预设的与该目标触发条件对应的性能指标数据范围，作为目标性能指标数据范围；然后，根据目标触发条件、目标功耗控制方法以及目标性能指标数据范围，生成目标设备在未来时段内使用的功耗控制方案。在该实施例中，目标设备在未来时段内使用的功耗控制方案包括：目标触发条件、目标功耗控制方法以及目标性能指标数据范围。
进一步，无论是在基于规则生成功耗控制方案的实施例中，还是在采用规则与人工智能相结合的方式生成功耗控制方案的实施例中，预设触发条件可以包括以下至少一种：表示时间范围的触发条件和表示功耗指标数据范围的触发条件。其中，表示时间范围的触发条件的意思是：在该触发条件所表示的时间范围内，启动对应的功耗控制方法进行功耗控制。根据所表示的时间范围的不同，表示时间范围的触发条件可以有多个。表示功耗指标数据范围的触发条件的意思是：在目标设备在一小段时间内的实际功耗指标数据位于该触发条件所表示的功耗指标数据范围时，启动对应的功耗控制方法进行功耗控制。同理，根据功耗指标数据的不同，表示功耗指标数据范围的触发条件也可以有多个。
基于上述，根据目标设备在未来时段内的功耗指标数据，与预设触发条件进行匹配，确定预设触发条件中被匹配中的目标触发条件，可以包括以下至少一种方式：
方式1：根据未来时段对应的时间范围，从预设表示时间范围的触发条件中，确定所表示的时间范围落在未来时段对应的时间范围内的触发条件，作为目标触发条件。例如，假设未来时段是指未来一天的上午，某个触发条件表示的时间范围是上午6:00-8:00，另一个触发条件表示的时间范围是下午2:00-4:00，则表示上午6:00-8:00的触发条件为目标触发条件。
方式2：根据目标设备在未来时段内的功耗指标数据，从预设表示功耗指标数据范围的触发条件中，确定所表示的功耗指标数据范围在目标设备在未来时段内的功耗指标数据中出现过的触发条件，作为目标触发条件。例如，假设目标设备在未来时段内的功耗指标数据包括功耗，且功耗值在10w-30w之间波动，某个触发条件表示的功耗范围是大于20w，另一个触发条件表示的功耗范围是大于50w，则表示大于20w的触发条件为目标触发条件。
在本申请各实施例中，在得到目标设备在未来时段内使用的功耗控制方案之后，功耗控制设备102将目标设备在未来时段内使用的功耗控制方案提供给目标设备。对目标设备来说，每次接收到功耗控制设备102提供的新的功耗控制方案后，可以根据新的功耗控制方案进行功耗控制。
在本申请一些实施例中，目标设备接收新到达的功耗控制方案；将当前使用的功耗控制方案替换为新到达的功耗控制方案，从而根据新到达的功耗控制方案进行功耗控制。为便于描述和区分，将新到达的功耗控制方案称为第一功耗控制方案，将当前使用的功耗控制方案称为第二功耗控制方案。在该实施例中，每当有新的功耗控制方案到达，会用新的功耗控制方案覆盖旧的功耗控制方案，这种方式简单，易于实施。
其中，第一功耗控制方案包括：触发条件和功耗控制方法。基于此，目标设备根据第一功耗控制方案进行功耗控制的过程为：监测第一功耗控制方案中的触发条件是否被满足；在第一功耗控制方案中的触发条件被满足的情况下，利用第一功耗控制方案中与被满足的触发条件对应的功耗控制方法对目标设备进行功耗控制。以触发条件为上午6:00-8:00，功耗控制方法为DVFS 为例，则在上午6:00-8:00期间，可开启DVFS进行功耗控制。
进一步可选地，在根据第一功耗控制方案对目标设备进行功耗控制的过程中，监控目标设备的实际性能指标数据，例如实际QoS数据；并可根据目标设备的实际性能指标数据，调整第一功耗控制方案的控制强度。
例如，可以将目标设备的实际性能数据与指定性能下限值进行比较；若目标设备的实际性能数据小于指定性能下限值，则关停第一功耗控制方案，直至目标设备的实际性能指标数据大于或等于指定性能下限值为止。
又例如，可以将目标设备的实际性能数据与指定性能下限值进行比较；若目标设备的实际性能数据小于指定性能下限值，则将第一功耗控制方案中的功耗控制方法调整为控制强度较低的功耗控制方法，直至目标设备的实际性能指标数据大于或等于指定性能下限值为止。
其中，指定性能下限值可以是目标设备所能接受的性能下限值。或者，在第一功耗控制方案包含与触发条件对应的性能指标数据范围的情况下，指定性能下限值也可以是第一功耗控制方案中性能指标数据范围的下限值。
在上述实施例中，结合目标设备的实际性能指标数据，对目标设备进行功耗控制，不仅可以节省能耗，还可以保证目标设备的服务质量。
在本申请另一些实施例中，目标设备接收新到达的第一功耗控制方案；根据第一功耗控制方案和当前使用的第二功耗控制方案，得到第三功耗控制方案；将当前使用的第二功耗控制方案替换为第三功耗控制方案，根据第三功耗控制方案对设备进行功耗控制。
在一种方式中，目标设备可以直接将第一功耗控制方案和第二功耗控制方案合并，得到第三功耗控制方案。
在另一种方式中，目标设备可以将第一功耗控制方案中的触发条件与第二功耗控制方案中的触发条件进行比较；若第一功耗控制方案中的触发条件与第二功耗控制方案中的触发条件的类别不同，将第一功耗控制方案和第二功耗控制方案进行合并，得到第三控制方案；若第一功耗控制方案中的触发 条件与第二功耗控制方案中的触发条件的类别相同，则用新的功耗控制方案覆盖旧的功耗控制方案，即将第一功耗控制方案作为第三控制方案。
例如，若第一功耗控制方案中的触发条件是表示时间范围的触发条件，第二功耗控制方案中的触发条件是表示功耗范围的触发条件，则可以第一功耗控制方案和第二功耗控制方案合并起来，作为第三控制方案。若第一功耗控制方案中的触发条件是表示时间范围的触发条件，第二功耗控制方案中的触发条件也是表示时间范围的触发条件，则可以丢弃第二功耗控制方案，保留第一功耗控制方案，即将第一功耗控制方案作为第三控制方案。
其中，第三功耗控制方案中包括触发条件和功耗控制方法，且可能有多组触发条件和功耗控制方法的组合。基于此，根据第三功耗控制方案对设备进行功耗控制，包括：在第三功耗控制方案中的触发条件被满足的情况下，利用第三功耗控制方案中与被满足的触发条件对应的功耗控制方法对目标设备进行功耗控制。
进一步可选地，在根据第三功耗控制方案对目标设备进行功耗控制的过程中，监控目标设备的实际性能指标数据，例如实际QoS数据；并可根据目标设备的实际性能指标数据，调整第三功耗控制方案的控制强度。
例如，可以将目标设备的实际性能数据与指定性能下限值进行比较；若目标设备的实际性能数据小于指定性能下限值，则关停第三功耗控制方案，直至目标设备的实际性能指标数据大于或等于指定性能下限值为止。
又例如，可以将目标设备的实际性能数据与指定性能下限值进行比较；若目标设备的实际性能数据小于指定性能下限值，则将第三功耗控制方案中的功耗控制方法调整为控制强度较低的功耗控制方法，直至目标设备的实际性能指标数据大于或等于指定性能下限值为止。
其中，指定性能下限值可以是目标设备所能接受的性能下限值。或者，在第三功耗控制方案包含与触发条件对应的性能指标数据范围的情况下，指定性能下限值也可以是第三功耗控制方案中性能指标数据范围的下限值。
在上述实施例中，结合目标设备的实际性能指标数据，对目标设备进行 功耗控制，不仅可以节省能耗，还可以保证目标设备的服务质量。
关于设备组的实施例：
在上述实施例中，功耗控制设备102以一台物理设备为单位，可动态更新每台物理设备的功耗控制方案，但并不限于此。在本申请下述实施例中，功耗控制设备102还可以以设备组为单位，动态更新每个设备组的功耗控制方案。其中，设备组包含至少一台物理设备，设备组内的各台物理设备形成一种绑定关系，可共用相同的功耗控制方案。
在本申请实施例中，并不限定设备组的形成方式。例如，可以将机房内距离相对较近的物理设备划分为一个设备组，或者将负载情况相同或类似的物理设备划分为一个设备组，或者将同一类型的物理设备划分为一个设备组，或者将同一型号的物理设备划分为一个设备组，或者将同一厂家的物理设备划分为一个设备组，或者将整个机房内的物理设备划分为一个设备组，等等。
对功耗控制设备102来说，可获取设备组的相关参数，设备组的相关参数包括设备组在历史时段内的功耗指标数据；根据设备组在历史时段内的功耗指标数据，预测设备组在未来时段内的功耗指标数据；根据设备组在未来时段内的功耗指标数据，生成设备组在未来时段内使用的功耗控制方案。
在一可选实施例中，设备组在历史时段内的功耗指标数据包括：设备组内各台物理设备在历史时段内的功耗指标数据；相应地，预测出的设备组在未来时段内的功耗指标数据包括：设备组内各台物理设备在未来时段内的功耗指标数据。可选地，一种预测方式为：单独预测方式，即对设备组内的每台物理设备，根据其在历史时段内的功耗指标数据单独预测其在未来时段内的功耗指标数据。可选地，另一种预测方式为：联合预测方式，即根据设备组内的所有物理设备在历史时段内的功耗指标数据进行联合预测，得到每台物理设备在未来时段内的功耗指标数据。
在另一可选实施例中，设备组在历史时段内的功耗指标数据是根据设备组内各台物理设备在历史时段内的功耗指标数据计算出的功耗指标数据。设备组在历史时段内的功耗指标数据可以是一些离散值，也可以是连续值，包 括在历史时段内各历史时刻的功耗指标数据。例如，设备组在某个历史时刻的功耗指标数据可以是设备组内各台物理设备在相同历史时刻的功耗指标数据的平均值，或者是其中的最大值，或者是其中的最小值，或者是其中最大值和最小值的均值，等等。相应地，设备组在未来时段内的功耗指标数据是根据设备组在历史时段内的功耗指标数据预测出的，可以是一些离散值，也可以是连续值。
在本申请一些实施例中，为了提高预测结果的准确度，可对设备组进行分类；若设备组属于功耗变化满足规律要求的设备组，则再执行根据设备组在历史时段内的功耗指标数据，预测设备组在未来时段内的功耗指标数据的操作；若设备组不属于功耗变化满足规律要求的设备组，则结束为设备组生成功耗控制方案的操作，或者为设备组设置默认的功耗控制方案。
可选地，为了更加准确地对设备组进行分类，可以加入设备组的设备参数。设备组的设备参数主要是指一些与设备组内物理设备自身相关的静态参数，例如设备类型、设备型号、设备序列号、设备规格等。基于此，功耗控制设备102除了获取设备组在历史时段内的功耗指标数据之外，还需获取设备组的设备参数，并根据设备组的设备参数和设备组在历史时段内的功耗指标数据，识别设备组是否属于功耗变化满足规律要求的设备组。若设备组属于功耗变化满足规律要求的设备组，根据设备组在历史时段内的功耗指标数据，预测设备组在未来时段内的功耗指标数据，进而，根据设备组在未来时段内的功耗指标数据，生成设备组在未来时段内使用的功耗控制方案。若设备组不属于功耗变化满足规律要求的设备组，可为设备组设置默认的功耗控制方案。
可选地，在对设备组进行分类时，可以采用分类模型。其中，分类模型可以采用逻辑回归算法或随机森林算法对设备组进行分类，以得到设备组是否属于功耗变化满足规律要求的设备组的分类结果。
可选地，在预测设备组在未来时段内的功耗指标数据时，可以采用预测模型。可以将设备组在历史时段内的功耗指标数据输入预测模型，以得到设 备组在未来时段内的功耗指标数据。其中，预测模型可以采用线性回归算法或深度学习算法，但不限于此。
其中，功耗控制设备102为每个设备组生成功耗控制方案的过程，与为每台物理设备的功耗控制方案的过程相似，区别在于方案生成过程中所使用的一些数据会有些不同，相关过程可能相对要复杂一些，但基本实现或原理相同，故在此不再赘述，可类比于前述实施例中的相关内容来实现。
在此说明，在上述实施例中，以机房系统100为例对本申请实施例的技术方案进行了说明，但并不限于此。本申请实施例的技术方案还可以应用于数据中心系统以及集群等包含多台物理设备的环境。当然，本申请实施例的技术方案也适用于单独的物理设备。
图2为本申请示例性实施例提供的一种数据中心系统的结构示意图。如图2所示，该数据中心系统200包括：至少一个机房201和功耗控制设备202。其中，每个机房201包括至少一台物理设备203。每个机房201包含的物理设备的数量，可以是一台，也可以是多台。
本实施例中的机房201与图1a所示实施例中的机房类似，关于物理设备203以及机房201的相关描述，可参见图1a所示实施例中的描述，在此不再赘述。
本实施例与图1a所示实施例的区别在于：功耗控制设备202不属于某个机房，而是隶属于整个数据中心系统200，需要为各机房201内的物理设备203提供功耗控制方案。虽然功耗控制设备202不属于任何机房，但是物理部署上，可以部署在某个机房内，也可以独立部署于各机房之外。
同理，本实施例的功耗控制设备202，可以以物理设备为单位，动态更新每台物理设备的功耗控制方案，也可以以设备组为单位，动态更新每个设备组的功耗控制方案。
对于以物理设备为单位的情况，功耗控制设备202为每台物理设备203动态更新功耗控制方案的方式相同或相似。故以目标设备为例，对功耗控制 设备202的功能进行说明。功耗控制设备202，用于获取目标设备的相关参数，该相关参数包括目标设备在历史时段内的功耗指标数据；根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据；根据目标设备在未来时段内的功耗指标数据，生成目标设备在未来时段内使用的功耗控制方案并提供给目标设备，以供目标设备在未来时段进行功耗控制；其中，目标设备是至少一台物理设备中任一台设备。
对于以设备组为单位的情况，对功耗控制设备202的功能进行说明。功耗控制设备202，用于获取设备组的相关参数，设备组的相关参数包括设备组在历史时段内的功耗指标数据；根据设备组在历史时段内的功耗指标数据，预测设备组在未来时段内的功耗指标数据；根据设备组在未来时段内的功耗指标数据，生成设备组在未来时段内使用的功耗控制方案。
需要说明的是，在数据中心系统200中，一个设备组中的物理设备可以来自同一机房，也可以来自不同机房(即跨机房)，对此不做限定。
关于功耗控制设备202动态更新目标设备或设备组的功耗控制方案的详细过程，可参见前述实施例，在此不再赘述。
除上述机房系统和数据中心系统实施例之外，本申请还提供了一些方法实施例，下面对这些方法实施例进行描述。
图3a为本申请示例性实施例提供的一种功耗控制方法的流程示意图。该方法是从需要功耗控制的设备的角度进行的描述。如图3a所示，该方法包括：
31a、接收新到达的第一功耗控制方案。
32a、将当前使用的第二功耗控制方案替换为第一功耗控制方案。
33a、根据第一功耗控制方案对设备进行功耗控制。
在本实施例中，设备使用的功耗控制方案会动态更新。可选地，这些动态更新的功耗控制方案可来自于功耗控制设备，功耗控制设备是负责为需要功耗控制的物理设备动态提供功耗控制方案的设备，例如可以是前述实施例中的功耗控制方案102或202。
在本实施例中，每当有新的功耗控制方案到达，用新的功耗控制方案覆盖旧的功耗控制方案，这种方式简单，易于实施。
为便于描述和区分，将新到达的功耗控制方案称为第一功耗控制方案，将设备当前使用的功耗控制方案称为第二功耗控制方案。其中，第一功耗控制方案和第二功耗控制方案均包括：触发条件和功耗控制方法。不同功耗控制方案中的触发条件和功耗控制方法不尽相同。
可选地，步骤33a的一种实施方式包括：监测第一功耗控制方案中的触发条件是否被满足；在第一功耗控制方案中的触发条件被满足的情况下，利用第一功耗控制方案中与被满足的触发条件对应的功耗控制方法对设备进行功耗控制。
进一步可选地，在根据第一功耗控制方案对设备进行功耗控制的过程中，还可以监控设备的实际性能指标数据，例如实际QoS数据；并可根据设备的实际性能指标数据，调整第一功耗控制方案的控制强度。
例如，可以将设备的实际性能数据与指定性能下限值进行比较；若设备的实际性能数据小于指定性能下限值，则关停第一功耗控制方案，直至设备的实际性能指标数据大于或等于指定性能下限值为止。
又例如，可以将设备的实际性能数据与指定性能下限值进行比较；若设备的实际性能数据小于指定性能下限值，则将第一功耗控制方案中的功耗控制方法调整为控制强度较低的功耗控制方法，直至设备的实际性能指标数据大于或等于指定性能下限值为止。
其中，指定性能下限值可以是设备所能接受的性能下限值。或者，在第一功耗控制方案包含与触发条件对应的性能指标数据范围的情况下，指定性能下限值也可以是第一功耗控制方案中性能指标数据范围的下限值。
在上述实施例中，结合设备的实际性能指标数据，对设备进行功耗控制，不仅可以节省能耗，还可以保证设备的服务质量。
图3b为本申请示例性实施例提供的另一种功耗控制方法的流程示意图。该方法是从需要功耗控制的设备的角度进行的描述。如图3b所示，该方法包 括：
31b、接收新到达的第一功耗控制方案。
32b、根据第一功耗控制方案和当前使用的第二功耗控制方案，得到第三功耗控制方案。
33b、将当前使用的第二功耗控制方案替换为第三功耗控制方案，根据第三功耗控制方案对设备进行功耗控制。
在本实施例中，设备使用的功耗控制方案会动态更新。可选地，这些动态更新的功耗控制方案可来自于功耗控制设备，功耗控制设备是负责为需要功耗控制的物理设备动态提供功耗控制方案的设备，例如可以是前述实施例中的功耗控制方案102或202。
在本实施例中，每当有新的功耗控制方案到达，将新的功耗控制方案与旧的功耗控制方案进行融合，基于融合后的功耗控制方案进行功耗控制。这种方式中的功耗控制方案更加完善，有利于提高功耗控制力度，提高功耗控制的节能效果。
在步骤32b的一种方式中，可以直接将第一功耗控制方案和第二功耗控制方案合并，得到第三功耗控制方案。
在步骤32b的另一种方式中，可以将第一功耗控制方案中的触发条件与第二功耗控制方案中的触发条件进行比较；若第一功耗控制方案中的触发条件与第二功耗控制方案中的触发条件的类别不同，将第一功耗控制方案和第二功耗控制方案进行合并，得到第三控制方案；若第一功耗控制方案中的触发条件与第二功耗控制方案中的触发条件的类别相同，则用新的功耗控制方案覆盖旧的功耗控制方案，即将第一功耗控制方案作为第三控制方案。
其中，第三功耗控制方案中包括触发条件和功耗控制方法，且可能有多组触发条件和功耗控制方法的组合。基于此，根据第三功耗控制方案对设备进行功耗控制，包括：在第三功耗控制方案中的触发条件被满足的情况下，利用第三功耗控制方案中与被满足的触发条件对应的功耗控制方法对设备进行功耗控制。
进一步可选地，在根据第三功耗控制方案对设备进行功耗控制的过程中，还可以监控设备的实际性能指标数据，例如实际QoS数据；并可根据设备的实际性能指标数据，调整第三功耗控制方案的控制强度。
例如，可以将设备的实际性能数据与指定性能下限值进行比较；若设备的实际性能数据小于指定性能下限值，则关停第三功耗控制方案，直至设备的实际性能指标数据大于或等于指定性能下限值为止。
又例如，可以将设备的实际性能数据与指定性能下限值进行比较；若设备的实际性能数据小于指定性能下限值，则将第三功耗控制方案中的功耗控制方法调整为控制强度较低的功耗控制方法，直至设备的实际性能指标数据大于或等于指定性能下限值为止。
其中，指定性能下限值可以是设备所能接受的性能下限值。或者，在第三功耗控制方案包含与触发条件对应的性能指标数据范围的情况下，指定性能下限值也可以是第三功耗控制方案中性能指标数据范围的下限值。
在本实施例中，结合设备的实际性能指标数据，对设备进行功耗控制，不仅可以节省能耗，还可以保证设备的服务质量。
图4a为本申请示例性实施例提供的一种功耗控制方案生成方法的流程示意图。如图4a所示，该方法包括：
41a、获取目标设备的相关参数，该相关参数包括所述目标设备在历史时段内的功耗指标数据。
42a、根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据。
43a、根据目标设备在未来时段内的功耗指标数据，生成目标设备在未来时段内使用的功耗控制方案。
关于该实施例中各步骤的描述，可参见前述实施例，在此不再赘述。
在本实施例中，引入人工智能，通过人工智能可动态更新目标设备的功耗控制方案，使得目标设备可实现动态功耗控制，提高功耗控制的节能效果。其中，本实施例中的目标设备可以是机房系统中任一台物理设备，也可是数 据中心系统中任一台物理设备，还可以是其它集群中任一台物理设备，也可以是独立的某个物理设备。
图4b为本申请示例性实施例提供的另一种功耗控制方案生成方法的流程示意图。如图4b所示，该方法包括：
41b、获取目标设备的相关参数，该相关参数包括所述目标设备在历史时段内的功耗指标数据和目标设备的设备参数。
42b、根据目标设备的设备参数和目标设备在历史时段内的功耗指标数据，识别目标设备是否属于功耗变化满足规律要求的设备；若识别结果为是，执行步骤43b；若识别结果为否，执行步骤45b。
43b、根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据，并执行步骤44b。
44b、根据目标设备在未来时段内的功耗指标数据，生成目标设备在未来时段内使用的功耗控制方案，结束此次操作。
45b、为目标设备设置默认的功耗控制方案，结束此次操作。
与图4a所示实施例相比，本实施例引入了目标设备的设备参数，用于对目标设备进行分类；若目标设备属于功耗变化满足规律要求的设备，则根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据的操作；若目标设备不属于功耗变化满足规律要求的设备，可为目标设备设置默认的功耗控制方案，例如DVFS。通过对目标设备进行分类，有利于提高预测结果的准确度，进而可为目标设备生成更为合适的功耗控制方案，提高功耗控制的节能效果。
图4c为本申请示例性实施例提供的又一种功耗控制方案生成方法的流程示意图。如图4c所示，该方法包括：
41c、获取目标设备的相关参数，该相关参数包括目标设备的设备参数、目标设备在历史时段内的功耗指标数据和性能指标数据。
42c、根据目标设备的设备参数和目标设备在历史时段内的功耗指标数据，识别目标设备是否属于功耗变化满足规律要求的设备；若识别结果为是， 执行步骤43c；若识别结果为否，执行步骤45c。
43c、根据目标设备在历史时段内的功耗指标数据和性能指标数据，预测目标设备在未来时段内的功耗指标数据，并执行步骤44c。
44c、根据目标设备在未来时段内的功耗指标数据，生成目标设备在未来时段内使用的功耗控制方案，结束此次操作。
45c、为目标设备设置默认的功耗控制方案，结束此次操作。
与图4b所示实施例相比，本实施例引入了目标设备在历史时段内的性能指标数据，同时结合目标设备在未来时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据，可考虑性能指标数据对功耗指标数据的影响，有利于提高预测结果的准确性，进而可为目标设备生成更加合理的功耗控制方案，提高功耗控制的准确度。
在上述图4b和图4c所示实施例中，步骤42b或步骤42c的一种实施方式包括：将目标设备的设备参数和目标设备在历史时段内的功耗指标数据输入分类模型，以得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。
进一步可选地，在分类模型内部，根据目标设备的设备参数，确定与目标设备所属设备类型对应的基准分布特征，基准分布特征是满足规律要求的功耗指标数据的分布特征；根据基准分布特征和目标设备在历史时段内的功耗指标数据的分布特征，对目标设备进行分类，以得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。
进一步可选地，根据基准分布特征和目标设备在历史时段内的功耗指标数据的分布特征，对目标设备进行分类的一种实施方式包括：根据基准分布特征和目标设备在历史时段内的功耗指标数据的分布特征，采用逻辑回归算法或随机森林算法对目标设备进行分类，以得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。
在上述图4a-图4c所示实施例中，步骤42a、步骤43b或步骤43c的一种实施方式包括：将目标设备在历史时段内的功耗指标数据，或者将目标设备 在历史时段内的功耗指标数据和性能指标数据作为入参，输入预测模型，以得到目标设备在未来时段内的功耗指标数据。
进一步可选地，在预测模型内部，根据目标设备在历史时段内的功耗指标数据，采用线性回归算法或深度学习算法，预测目标设备在未来时段内的功耗指标数据。
在上述图4a-图4c所示实施例中，步骤42c、步骤44b或步骤44c的一种实施方式包括：根据目标设备在未来时段内的功耗指标数据，确定预设触发条件中被匹配中的目标触发条件；获取与目标触发条件适配的目标功耗控制方法；根据目标触发条件和目标功耗控制方法，生成目标设备在未来时段内使用的功耗控制方案。
进一步可选地，上述根据目标设备在未来时段内的功耗指标数据，确定预设触发条件中被匹配中的目标触发条件，包括以下至少一种方式：
根据未来时段对应的时间范围，从预设表示时间范围的触发条件中，确定所表示的时间范围落在未来时段对应的时间范围内的触发条件，作为目标触发条件；
根据目标设备在未来时段内的功耗指标数据，从预设表示功耗指标数据范围的触发条件中，确定所表示的功耗指标数据范围在目标设备在未来时段内的功耗指标数据中出现过的触发条件，作为目标触发条件。
进一步可选地，上述获取与目标触发条件适配的目标功耗控制方法的一种实施方式包括：获取预设的与目标触发条件对应的功耗控制方法，作为目标功耗控制方法。
进一步可选地，上述获取与目标触发条件适配的目标功耗控制方法的另一种实施方式包括：模拟各种功耗控制方法在目标触发条件下的节能效果；根据各种功耗控制方法在目标触发条件下的节能效果，从中选择节能效果满足节能要求的功耗控制方法，作为目标功耗控制方法。
进一步可选地，上述根据目标触发条件和目标功耗控制方法，生成目标设备在未来时段内使用的功耗控制方案的一种实施方式包括：获取预设的与 目标触发条件对应的目标性能指标数据范围；根据目标触发条件、目标功耗控制方法以及目标性能指标数据范围，生成目标设备在未来时段内使用的功耗控制方案。
在上述图4a-图4c所示实施例中，步骤41a、步骤41b或步骤41c的一种实施方式包括：若目标设备是IT类设备，则获取目标设备在历史时段内的内部温度、功耗、CPU频率以及CPU负载中至少一种，作为目标设备在历史时段内的功耗指标数据；若目标设备是为IT类设备降温的空调设备，则获取目标设备在历史时段内的压缩机的频率、风机的转速、回风温度以及出风温度中至少一种，作为目标设备在历史时段内的功耗指标数据。
进一步可选地，在目标设备是IT类设备的情况下，还包括：获取与目标设备关联的其它设备在历史时段内的功耗，作为目标设备在历史时段内的功耗指标数据。
在上述图4a-图4c所示实施例中，在生成目标设备在未来时段内使用的功耗控制方案之后，还可以将目标设备在未来时段内使用的功耗控制方案提供给目标设备，以供目标设备在未来时段内进行功耗控制。其中，目标设备根据新的功耗控制方案进行功耗控制的过程可参见前述图3a或图3b所示实施例的描述，在此不再赘述。
图4d为本申请示例性实施例提供的一种功耗预测方法的流程示意图。如图4d所示，该方法包括：
41d、获取目标设备的相关参数，该相关参数包括目标设备在历史时段内的功耗指标数据。
42d、根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据。
在本实施例中，引入人工智能，通过人工智能可动态预测目标设备的功耗指标数据，可为其它依赖目标设备的功指标数据的操作提供准确的数据基础，有利于提高其它操作的效果。
在一可选实施例中，目标设备的相关参数还包括目标设备的设备参数。 基于此，所述方法在根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据之前，还包括：根据目标设备的设备参数和目标设备在历史时段内的功耗指标数据，识别目标设备是否属于功耗变化满足规律要求的设备；若目标设备属于功耗变化满足规律要求的设备，则执行根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据的操作。
在一可选实施例中，上述根据目标设备的设备参数和目标设备在历史时段内的功耗指标数据，对目标设备进行分类的一种实施方式包括：将目标设备的设备参数和目标设备在历史时段内的功耗指标数据输入分类模型，以得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。
进一步可选地，根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据，包括：将目标设备在历史时段内的功耗指标数据输入预测模型，以得到目标设备在未来时段内的功耗指标数据。
进一步可选地，目标设备的相关参数还包括目标设备在历史时段内的性能指标数据。基于此，将目标设备在历史时段内的功耗指标数据输入预测模型，以得到目标设备在未来时段内的功耗指标数据，包括：将目标设备在历史时段内的功耗指标数据以及目标设备在历史时段内的性能指标数据输入预测模型，以得到目标设备在未来时段内的功耗指标数据。
其中，预测模型可采用线性回归算法或深度学习算法，对此不做限定。
进一步可选地，在预测出目标设备在未来时段内的功耗指标数据之后，可根据目标设备在未来时段内的功耗指标数据，生成目标设备在未来时段内使用的功耗控制方案。关于如何根据目标设备在未来时段内的功耗指标数据，生成目标设备在未来时段内使用的功耗控制方案，可参见前述实施例的描述，在此不再赘述。
需要说明的是，上述实施例所提供方法的各步骤的执行主体均可以是同一设备，或者，该方法也由不同设备作为执行主体。比如，步骤31a至步骤33a的执行主体可以为设备A；又比如，步骤31a和32a的执行主体可以为设 备A，步骤33a的执行主体可以为设备B；等等。
另外，在上述实施例及附图中的描述的一些流程中，包含了按照特定顺序出现的多个操作，但是应该清楚了解，这些操作可以不按照其在本文中出现的顺序来执行或并行执行，操作的序号如31a、32a等，仅仅是用于区分开各个不同的操作，序号本身不代表任何的执行顺序。另外，这些流程可以包括更多或更少的操作，并且这些操作可以按顺序执行或并行执行。需要说明的是，本文中的“第一”、“第二”等描述，是用于区分不同的消息、设备、模块等，不代表先后顺序，也不限定“第一”和“第二”是不同的类型。
图5a为本申请示例性实施例提供的一种计算设备的结构示意图。如图5a所示，该计算设备包括：存储器51a和处理器52a。
存储器51a，用于存储计算机程序，并可被配置为存储其它各种数据以支持在计算设备上的操作。这些数据的示例包括用于在计算设备上操作的任何应用程序或方法的指令，消息，图片，视频等。
处理器52a，与存储器51a耦合，用于执行存储器51a中的计算机程序，以用于：获取目标设备的相关参数，相关参数包括目标设备在历史时段内的功耗指标数据；根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据；根据目标设备在未来时段内的功耗指标数据，生成目标设备在未来时段内使用的功耗控制方案。
其中，目标设备是指需要进行功耗控制的物理设备，可以是机房系统中任一台物理设备，或者是数据中心系统中任一台物理设备，或者是集群中任一台物理设备，或者是某台独立的物理设备。
在一可选实施例中，目标设备的相关参数还包括目标设备的设备参数。基于此，处理器52a还用于：在预测目标设备在未来时段内的功耗指标数据之前，根据目标设备的设备参数和目标设备在历史时段内的功耗指标数据，识别目标设备是否属于功耗变化满足规律要求的设备；并在目标设备属于功耗变化满足规律要求的设备的情况下，执行根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据的操作。
在一可选实施例中，处理器52a在对目标设备进行分类时，具体用于：将目标设备的设备参数和目标设备在历史时段内的功耗指标数据输入分类模型，以得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。
进一步可选地，处理器52a在得到目标设备是否属于功耗变化满足规律要求的设备的分类结果时，具体用于：在分类模型内部，根据目标设备的设备参数，确定与目标设备所属设备类型对应的基准分布特征，基准分布特征是满足规律要求的功耗指标数据的分布特征；根据基准分布特征和目标设备在历史时段内的功耗指标数据的分布特征，对目标设备进行分类，以得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。
进一步可选地，处理器52a在根据基准分布特征和目标设备在历史时段内的功耗指标数据的分布特征对目标设备进行分类时，具体用于：根据基准分布特征和目标设备在历史时段内的功耗指标数据的分布特征，采用逻辑回归算法或随机森林算法对目标设备进行分类，以得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。
在一可选实施例中，处理器52a在预测目标设备在未来时段内的功耗指标数据时，具体用于：将目标设备在历史时段内的功耗指标数据输入预测模型，以得到目标设备在未来时段内的功耗指标数据。
进一步可选地，处理器52a具体用于：在预测模型内部，根据目标设备在历史时段内的功耗指标数据，采用线性回归算法或深度学习算法，预测目标设备在未来时段内的功耗指标数据。
进一步可选地，目标设备的相关参数还包括目标设备在历史时段内的性能指标数据。基于此，处理器52a具体用于：将目标设备在历史时段内的功耗指标数据以及目标设备在历史时段内的性能指标数据输入预测模型，以得到目标设备在未来时段内的功耗指标数据。
在一可选实施例中，处理器52a在生成目标设备在未来时段内使用的功耗控制方案时，具体用于：根据目标设备在未来时段内的功耗指标数据，确定预设触发条件中被匹配中的目标触发条件；获取与目标触发条件适配的目 标功耗控制方法；根据目标触发条件和目标功耗控制方法，生成目标设备在未来时段内使用的功耗控制方案。
进一步可选地，处理器52a在确定预设触发条件中被匹配中的目标触发条件时，具体用于执行以下至少一种操作：
根据未来时段对应的时间范围，从预设表示时间范围的触发条件中，确定所表示的时间范围落在未来时段对应的时间范围内的触发条件，作为目标触发条件；
根据目标设备在未来时段内的功耗指标数据，从预设表示功耗指标数据范围的触发条件中，确定所表示的功耗指标数据范围在目标设备在未来时段内的功耗指标数据中出现过的触发条件，作为目标触发条件。
进一步可选地，处理器52a在获取与目标触发条件适配的目标功耗控制方法时，具体用于：获取预设的与目标触发条件对应的功耗控制方法，作为目标功耗控制方法。
进一步可选地，处理器52a在获取与目标触发条件适配的目标功耗控制方法时，具体用于：模拟各种功耗控制方法在目标触发条件下的节能效果；根据各种功耗控制方法在目标触发条件下的节能效果，从中选择节能效果满足节能要求的功耗控制方法，作为目标功耗控制方法。
进一步可选地，处理器52a在生成目标设备在未来时段内使用的功耗控制方案时，具体用于：获取预设的与目标触发条件对应的目标性能指标数据范围；根据目标触发条件、目标功耗控制方法以及目标性能指标数据范围，生成目标设备在未来时段内使用的功耗控制方案。
在一可选实施例中，处理器52a在获取目标设备在历史时段内的功耗指标数据时，具体用于：若目标设备是IT类设备，则获取目标设备在历史时段内的内部温度、功耗、CPU频率以及CPU负载中至少一种，作为目标设备在历史时段内的功耗指标数据；若目标设备是为IT类设备降温的空调设备，则获取目标设备在历史时段内的压缩机的频率、风机的转速、回风温度以及出风温度中至少一种，作为目标设备在历史时段内的功耗指标数据。
进一步可选地，在目标设备是IT类设备的情况下，处理器52a还用于：获取与目标设备关联的其它设备在历史时段内的功耗，作为目标设备在历史时段内的功耗指标数据。
在一可选实施例中，处理器52a还用于：在获取目标设备在未来时段内使用的功耗控制方案之后，将目标设备在未来时段内使用的功耗控制方案提供给目标设备，以供目标设备在未来时段内进行功耗控制。
进一步，如图5a所示，该计算设备还包括：通信组件53a、显示器54a、电源组件55a、音频组件56a等其它组件。图5a中仅示意性给出部分组件，并不意味着计算设备只包括图5a所示组件。另外，根据计算设备的实现形态的不同，图5a中虚线框内的组件为可选组件，而非必选组件。例如，当计算设备实现为智能手机、平板电脑或台式电脑等终端设备时，可以包括图5a中虚线框内的组件；当计算设备实现为常规服务器、云服务器、数据中心或服务器阵列等服务端设备时，可以不包括图5a中虚线框内的组件。
本实施例提供的计算设备，将功率控制抽象为方案，通过人工智能动态更新每台物理设备的功率控制方案，实现动态功耗控制，而且可使功耗控制可精确到每台物理设备，不同物理设备可进行与自身功耗情况适配的功耗控制方案，不仅可提高功耗控制的精度，还可以提高功耗控制的整体节能效果。
相应地，本申请实施例还提供一种存储有计算机程序的计算机可读存储介质，计算机程序被执行时能够实现上述图4a-图4c所示方法实施例中的各步骤。
图5b为本申请示例性实施例提供的另一种计算设备的结构示意图。如图5b所示，该计算设备包括：存储器51b和处理器52b。
存储器51b，用于存储计算机程序，并可被配置为存储其它各种数据以支持在计算设备上的操作。这些数据的示例包括用于在计算设备上操作的任何应用程序或方法的指令，消息，图片，视频等。
处理器52b，与存储器51b耦合，用于执行存储器51b中的计算机程序，以用于：获取设备组的相关参数，相关参数包括设备组在历史时段内的功耗 指标数据，设备组包括至少一台物理设备；根据设备组在历史时段内的功耗指标数据，预测设备组在未来时段内的功耗指标数据；根据设备组在未来时段内的功耗指标数据，生成设备组在未来时段内使用的功耗控制方案。
其中，设备组包含至少一台物理设备，设备组内的各台物理设备形成一种绑定关系，可共用相同的功耗控制方案。在本实施例中，并不限定设备组的形成方式。例如，可以将机房内距离相对较近的物理设备划分为一个设备组，或者将负载情况相同或类似的物理设备划分为一个设备组，或者将同一类型的物理设备划分为一个设备组，或者将同一型号的物理设备划分为一个设备组，或者将同一厂家的物理设备划分为一个设备组，或者将整个机房内的物理设备划分为一个设备组，等等。
在一可选实施例中，设备组在历史时段内的功耗指标数据包括：设备组内各台物理设备在历史时段内的功耗指标数据。相应地，预测出的设备组在未来时段内的功耗指标数据包括：设备组内各台物理设备在未来时段内的功耗指标数据。
在另一可选实施例中，设备组在历史时段内的功耗指标数据是根据设备组内各台物理设备在历史时段内的功耗指标数据计算出的功耗指标数据。设备组在历史时段内的功耗指标数据可以是一些离散值，也可以是连续值，包括在历史时段内各历史时刻的功耗指标数据。例如，设备组在某个历史时刻的功耗指标数据可以是设备组内各台物理设备在相同历史时刻的功耗指标数据的平均值，或者是其中的最大值，或者是其中的最小值，或者是其中最大值和最小值的均值，等等。相应地，设备组在未来时段内的功耗指标数据是根据设备组在历史时段内的功耗指标数据预测出的，可以是一些离散值，也可以是连续值。
在一可选实施例中，设备组的相关参数还包括：设备组的设备参数。设备组的设备参数主要是指一些与设备组内物理设备自身相关的静态参数，例如设备类型、设备型号、设备序列号、设备规格等。基于此，处理器52b还用于：获取设备组的设备参数，并根据设备组的设备参数和设备组在历史时 段内的功耗指标数据，识别设备组是否属于功耗变化满足规律要求的设备组；并在设备组属于功耗变化满足规律要求的设备组的情况下，执行根据设备组在历史时段内的功耗指标数据，预测设备组在未来时段内的功耗指标数据的操作。
进一步可选地，处理器52b具体用于：采用分类模型，对设备组进行分类。分类模型可以采用逻辑回归算法或随机森林算法对设备组进行分类，以得到设备组是否属于功耗变化满足规律要求的设备组的分类结果。
进一步可选地，处理器52b具体用于：将设备组在历史时段内的功耗指标数据输入预测模型，以得到设备组在未来时段内的功耗指标数据。预测模型可以采用线性回归算法或深度学习算法，但不限于此。
进一步，如图5b所示，该计算设备还包括：通信组件53b、显示器54b、电源组件55b、音频组件56b等其它组件。图5b中仅示意性给出部分组件，并不意味着计算设备只包括图5b所示组件。另外，根据计算设备的实现形态的不同，图5b中虚线框内的组件为可选组件，而非必选组件。例如，当计算设备实现为智能手机、平板电脑或台式电脑等终端设备时，可以包括图5b中虚线框内的组件；当计算设备实现为常规服务器、云服务器、数据中心或服务器阵列等服务端设备时，可以不包括图5b中虚线框内的组件。
本实施例提供的计算设备，将功率控制抽象为方案，通过人工智能动态更新每个设备组的功率控制方案，实现动态功耗控制，而且可使功耗控制可精确到设备组，不同设备组可进行与自身功耗情况适配的功耗控制方案，不仅可提高功耗控制的精度，还可以提高功耗控制的整体节能效果。
相应地，本申请实施例还提供一种存储有计算机程序的计算机可读存储介质，计算机程序被执行时能够实现上述与设备组相关的实施例中的操作。
图6a为本申请示例性实施例提供的一种物理设备的结构示意图。如图6a所示，该物理设备包括：存储器61a、处理器62a以及通信组件63a。
存储器61a，用于存储计算机程序，并可被配置为存储其它各种数据以支持在物理设备上的操作。这些数据的示例包括用于在物理设备上操作的任何 应用程序或方法的指令，消息，图片，视频，各种功耗方案等。
处理器62a，与存储器61a耦合，用于执行存储器61a中的计算机程序，以用于：通过通信组件63a接收新到达的第一功耗控制方案；将当前使用的第二功耗控制方案替换为第一功耗控制方案；根据第一功耗控制方案对设备进行功耗控制。其中，第一功耗控制方案包括触发条件和功耗控制方法。
在一可选实施例中，处理器62a在根据第一功耗控制方案对设备进行功耗控制时，具体用于：在第一功耗控制方案中的触发条件被满足的情况下，利用第一功耗控制方案中与被满足的触发条件对应的功耗控制方法对设备进行功耗控制。
在一可选实施例中，处理器62a还用于：在根据第一功耗控制方案对设备进行功耗控制的过程中，监控设备的实际性能指标数据；根据设备的实际性能指标数据，调整第一功耗控制方案的控制强度。
在一可选实施例中，处理器62a在调整第一功耗控制方案的控制强度时，具体用于：若设备的实际性能指标数据小于指定性能下限值，则关停第一功耗控制方案，直至设备的实际性能指标数据大于或等于指定性能下限值为止；或者，若设备的实际性能指标数据小于指定性能下限值，则将第一功耗控制方案中的功耗控制方法调整为控制强度较低的功耗控制方法，直至设备的实际性能指标数据大于或等于指定性能下限值为止；其中，指定性能下限值是设备所能接受的性能下限值，或者是第一功耗控制方案中性能指标数据范围的下限值。
进一步，如图6a所示，该物理设备还包括：显示器64a、电源组件66a、音频组件66a等其它组件。图6a中仅示意性给出部分组件，并不意味着物理设备只包括图6a所示组件。另外，根据物理设备的实现形态的不同，图6a中虚线框内的组件为可选组件，而非必选组件。例如，当物理设备实现为智能手机、平板电脑或台式电脑等终端设备时，可以包括图6a中虚线框内的组件；当物理设备实现为常规服务器、云服务器、数据中心或服务器阵列等服务端设备时，可以不包括图6a中虚线框内的组件。
本实施例的物理设备，其使用的功耗控制方案会动态更新，每当有新的功耗控制方案到达，用新的功耗控制方案覆盖旧的功耗控制方案，这种方式简单，易于实施。
相应地，本申请实施例还提供一种存储有计算机程序的计算机可读存储介质，计算机程序被执行时能够实现上述图3a所示方法实施例中的各步骤。
图6b为本申请示例性实施例提供的一种物理设备的结构示意图。如图6b所示，该物理设备包括：存储器61b、处理器62b以及通信组件63b。
存储器61b，用于存储计算机程序，并可被配置为存储其它各种数据以支持在物理设备上的操作。这些数据的示例包括用于在物理设备上操作的任何应用程序或方法的指令，消息，图片，视频，各种功耗方案等。
处理器62b，与存储器61b耦合，用于执行存储器61b中的计算机程序，以用于：通过通信组件63b接收新到达的第一功耗控制方案；根据第一功耗控制方案和当前使用的第二功耗控制方案，得到第三功耗控制方案；将当前使用的第二功耗控制方案替换为第三功耗控制方案，根据第三功耗控制方案对设备进行功耗控制。
在一可选实施例中，处理器62b在得到第三功耗控制方案时，具体用于：若第一功耗控制方案中的触发条件与第二功耗控制方案中的触发条件的类别不同，将第一功耗控制方案和第二功耗控制方案进行合并，得到第三控制方案；若第一功耗控制方案中的触发条件与第二功耗控制方案中的触发条件的类别相同，将第一功耗控制方案作为第三控制方案。
在一可选实施例中，处理器62b在根据第三功耗控制方案对设备进行功耗控制时，具体用于：在第三功耗控制方案中的触发条件被满足的情况下，利用第三功耗控制方案中与被满足的触发条件对应的功耗控制方法对设备进行功耗控制。
在一可选实施例中，处理器62b还用于：在根据第三功耗控制方案对设备进行功耗控制的过程中，监控设备的实际性能指标数据；根据设备的实际性能指标数据，调整第三功耗控制方案的控制强度。
在一可选实施例中，处理器62b在调整第三功耗控制方案的控制强度时，具体用于：若设备的实际性能指标数据小于指定性能下限值，则关停第三功耗控制方案，直至设备的实际性能指标数据大于或等于指定性能下限值为止；或者，若设备的实际性能指标数据小于指定性能下限值，则将第三功耗控制方案中的功耗控制方法调整为控制强度较低的功耗控制方法，直至设备的实际性能指标数据大于或等于指定性能下限值为止；其中，指定性能下限值是设备所能接受的性能下限值，或者是第三功耗控制方案中性能指标数据范围的下限值。
进一步，如图6b所示，该物理设备还包括：显示器64b、电源组件66b、音频组件66b等其它组件。图6b中仅示意性给出部分组件，并不意味着物理设备只包括图6b所示组件。另外，根据物理设备的实现形态的不同，图6b中虚线框内的组件为可选组件，而非必选组件。例如，当物理设备实现为智能手机、平板电脑或台式电脑等终端设备时，可以包括图6b中虚线框内的组件；当物理设备实现为常规服务器、云服务器、数据中心或服务器阵列等服务端设备时，可以不包括图6b中虚线框内的组件。
本实施例的物理设备，其使用的功耗控制方案会动态更新，每当有新的功耗控制方案到达，将新的功耗控制方案与旧的功耗控制方案进行融合，基于融合后的功耗控制方案进行功耗控制。这种方式中的功耗控制方案更加完善，有利于提高功耗控制力度，提高功耗控制的节能效果。
相应地，本申请实施例还提供一种存储有计算机程序的计算机可读存储介质，计算机程序被执行时能够实现上述图3b所示方法实施例中的各步骤。
图7本申请示例性实施例提供的又一种计算设备的结构示意图。如图7所示，该计算设备包括：存储器71和处理器72。
存储器71，用于存储计算机程序，并可被配置为存储其它各种数据以支持在计算设备上的操作。这些数据的示例包括用于在计算设备上操作的任何应用程序或方法的指令，消息，图片，视频等。
处理器72，与存储器71耦合，用于执行存储器71中的计算机程序，以 用于：获取目标设备的相关参数，相关参数包括目标设备在历史时段内的功耗指标数据；根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据。
其中，目标设备是指需要进行功耗预测的物理设备，可以是机房系统中任一台物理设备，或者是数据中心系统中任一台物理设备，或者是集群中任一台物理设备，或者是某台独立的物理设备。
在一可选实施例中，目标设备的相关参数还包括目标设备的设备参数。基于此，处理器72还用于：在预测目标设备在未来时段内的功耗指标数据之前，根据目标设备的设备参数和目标设备在历史时段内的功耗指标数据，识别目标设备是否属于功耗变化满足规律要求的设备；并在目标设备属于功耗变化满足规律要求的设备的情况下，执行根据目标设备在历史时段内的功耗指标数据，预测目标设备在未来时段内的功耗指标数据的操作。
在一可选实施例中，处理器72在对目标设备进行分类时，具体用于：将目标设备的设备参数和目标设备在历史时段内的功耗指标数据输入分类模型，以得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。
进一步可选地，处理器72在得到目标设备是否属于功耗变化满足规律要求的设备的分类结果时，具体用于：在分类模型内部，根据目标设备的设备参数，确定与目标设备所属设备类型对应的基准分布特征，基准分布特征是满足规律要求的功耗指标数据的分布特征；根据基准分布特征和目标设备在历史时段内的功耗指标数据的分布特征，对目标设备进行分类，以得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。
进一步可选地，处理器72在根据基准分布特征和目标设备在历史时段内的功耗指标数据的分布特征对目标设备进行分类时，具体用于：根据基准分布特征和目标设备在历史时段内的功耗指标数据的分布特征，采用逻辑回归算法或随机森林算法对目标设备进行分类，以得到目标设备是否属于功耗变化满足规律要求的设备的分类结果。
在一可选实施例中，处理器72在预测目标设备在未来时段内的功耗指标 数据时，具体用于：将目标设备在历史时段内的功耗指标数据输入预测模型，以得到目标设备在未来时段内的功耗指标数据。
进一步可选地，处理器72具体用于：在预测模型内部，根据目标设备在历史时段内的功耗指标数据，采用线性回归算法或深度学习算法，预测目标设备在未来时段内的功耗指标数据。
进一步可选地，目标设备的相关参数还包括目标设备在历史时段内的性能指标数据。基于此，处理器72具体用于：将目标设备在历史时段内的功耗指标数据以及目标设备在历史时段内的性能指标数据输入预测模型，以得到目标设备在未来时段内的功耗指标数据。
在一可选实施例中，处理器72在获取目标设备在历史时段内的功耗指标数据时，具体用于：若目标设备是IT类设备，则获取目标设备在历史时段内的内部温度、功耗、CPU频率以及CPU负载中至少一种，作为目标设备在历史时段内的功耗指标数据；若目标设备是为IT类设备降温的空调设备，则获取目标设备在历史时段内的压缩机的频率、风机的转速、回风温度以及出风温度中至少一种，作为目标设备在历史时段内的功耗指标数据。
进一步可选地，在目标设备是IT类设备的情况下，处理器72还用于：获取与目标设备关联的其它设备在历史时段内的功耗，作为目标设备在历史时段内的功耗指标数据。
进一步，如图7所示，该计算设备还包括：通信组件73、显示器74、电源组件75、音频组件76等其它组件。图7中仅示意性给出部分组件，并不意味着计算设备只包括图7所示组件。另外，根据计算设备的实现形态的不同，图7中虚线框内的组件为可选组件，而非必选组件。例如，当计算设备实现为智能手机、平板电脑或台式电脑等终端设备时，可以包括图7中虚线框内的组件；当计算设备实现为常规服务器、云服务器、数据中心或服务器阵列等服务端设备时，可以不包括图7中虚线框内的组件。
本实施例提供的计算设备，引入人工智能，通过人工智能可动态预测目标设备的功耗指标数据，可为其它依赖目标设备的功指标数据的操作提供准 确的数据基础，有利于提高其它操作的效果。
相应地，本申请实施例还提供一种存储有计算机程序的计算机可读存储介质，计算机程序被执行时能够实现上述图4d所示方法实施例中的各步骤。
上述图5a-图7中的通信组件被配置为便于通信组件所在设备和其他设备之间有线或无线方式的通信。通信组件所在设备可以接入基于通信标准的无线网络，如WiFi，2G、3G、4G/LTE、5G等移动通信网络，或它们的组合。在一个示例性实施例中，通信组件经由广播信道接收来自外部广播管理系统的广播信号或广播相关信息。在一个示例性实施例中，通信组件还可以包括近场通信(NFC)模块，射频识别(RFID)技术，红外数据协会(IrDA)技术，超宽带(UWB)技术，蓝牙(BT)技术等。
上述图5a-图7中的显示器包括屏幕，其屏幕可以包括液晶显示器(LCD)和触摸面板(TP)。如果屏幕包括触摸面板，屏幕可以被实现为触摸屏，以接收来自用户的输入信号。触摸面板包括一个或多个触摸传感器以感测触摸、滑动和触摸面板上的手势。触摸传感器可以不仅感测触摸或滑动动作的边界，而且还检测与触摸或滑动操作相关的持续时间和压力。
上述图5a-图7中的电源组件，为电源组件所在设备的各种组件提供电力。电源组件可以包括电源管理系统，一个或多个电源，及其他与为电源组件所在设备生成、管理和分配电力相关联的组件。
上述图5a-图7中的音频组件，可被配置为输出和/或输入音频信号。例如，音频组件包括一个麦克风(MIC)，当音频组件所在设备处于操作模式，如呼叫模式、记录模式和语音识别模式时，麦克风被配置为接收外部音频信号。所接收的音频信号可以被进一步存储在存储器或经由通信组件发送。在一些实施例中，音频组件还包括一个扬声器，用于输出音频信号。
本领域内的技术人员应明白，本发明的实施例可提供为方法、系统、或计算机程序产品。因此，本发明可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本发明可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘 存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本发明是参照根据本发明实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
在一个典型的配置中，计算设备包括一个或多个处理器(CPU)、输入/输出接口、网络接口和内存。
内存可能包括计算机可读介质中的非永久性存储器，随机存取存储器(RAM)和/或非易失性内存等形式，如只读存储器(ROM)或闪存(flash RAM)。内存是计算机可读介质的示例。
计算机可读介质包括永久性和非永久性、可移动和非可移动媒体可以由任何方法或技术来实现信息存储。信息可以是计算机可读指令、数据结构、程序的模块或其他数据。计算机的存储介质的例子包括，但不限于相变内存(PRAM)、静态随机存取存储器(SRAM)、动态随机存取存储器(DRAM)、其 他类型的随机存取存储器(RAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、快闪记忆体或其他内存技术、只读光盘只读存储器(CD-ROM)、数字多功能光盘(DVD)或其他光学存储、磁盒式磁带，磁带磁磁盘存储或其他磁性存储设备或任何其他非传输介质，可用于存储可以被计算设备访问的信息。按照本文中的界定，计算机可读介质不包括暂存电脑可读媒体(transitory media)，如调制的数据信号和载波。
还需要说明的是，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、商品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、商品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、商品或者设备中还存在另外的相同要素。
以上所述仅为本申请的实施例而已，并不用于限制本申请。对于本领域技术人员来说，本申请可以有各种更改和变化。凡在本申请的精神和原理之内所作的任何修改、等同替换、改进等，均应包含在本申请的权利要求范围之内。

Claims (40)

  1. 一种功耗控制方案生成方法，其特征在于，包括：
    获取目标设备的相关参数，所述相关参数包括所述目标设备在历史时段内的功耗指标数据；
    根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据；
    根据所述目标设备在未来时段内的功耗指标数据，生成所述目标设备在未来时段内使用的功耗控制方案。
  2. 根据权利要求1所述的方法，其特征在于，所述相关参数还包括所述目标设备的设备参数；
    在根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据之前，还包括：
    根据所述目标设备的设备参数和所述目标设备在历史时段内的功耗指标数据，识别所述目标设备是否属于功耗变化满足规律要求的设备；
    若所述目标设备属于功耗变化满足规律要求的设备，则执行根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据的操作。
  3. 根据权利要求1所述的方法，其特征在于，根据所述目标设备的设备参数和所述目标设备在历史时段内的功耗指标数据，对所述目标设备进行分类，包括：
    将所述目标设备的设备参数和所述目标设备在历史时段内的功耗指标数据输入分类模型，以得到所述目标设备是否属于功耗变化满足规律要求的设备的分类结果。
  4. 根据权利要求3所述的方法，其特征在于，将所述目标设备的设备参数和所述目标设备在历史时段内的功耗指标数据输入分类模型，以得到所述目标设备是否属于功耗变化满足规律要求的设备的分类结果，包括：
    在所述分类模型内部，根据所述目标设备的设备参数，确定与所述目标设备所属设备类型对应的基准分布特征，所述基准分布特征是满足规律要求的功耗指标数据的分布特征；
    根据所述基准分布特征和所述目标设备在历史时段内的功耗指标数据的分布特征，对所述目标设备进行分类，以得到所述目标设备是否属于功耗变化满足规律要求的设备的分类结果。
  5. 根据权利要求4所述的方法，其特征在于，根据所述基准分布特征和所述目标设备在历史时段内的功耗指标数据的分布特征，对所述目标设备进行分类，以得到所述目标设备是否属于功耗变化满足规律要求的设备的分类结果，包括：
    根据所述基准分布特征和所述目标设备在历史时段内的功耗指标数据的分布特征，采用逻辑回归算法或随机森林算法对所述目标设备进行分类，以得到所述目标设备是否属于功耗变化满足规律要求的设备的分类结果。
  6. 根据权利要求1所述的方法，其特征在于，根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据，包括：
    将所述目标设备在历史时段内的功耗指标数据输入预测模型，以得到所述目标设备在未来时段内的功耗指标数据。
  7. 根据权利要求6所述的方法，其特征在于，所述相关参数还包括所述目标设备在历史时段内的性能指标数据；
    将所述目标设备在历史时段内的功耗指标数据输入预测模型，以得到所述目标设备在未来时段内的功耗指标数据，包括：
    将所述目标设备在历史时段内的功耗指标数据以及所述目标设备在历史时段内的性能指标数据输入预测模型，以得到所述目标设备在未来时段内的功耗指标数据。
  8. 根据权利要求6所述的方法，其特征在于，将所述目标设备在历史时段内的功耗指标数据输入预测模型，以得到所述目标设备在未来时段内 的功耗指标数据，包括：
    在所述预测模型内部，根据所述目标设备在历史时段内的功耗指标数据，采用线性回归算法或深度学习算法，预测所述目标设备在未来时段内的功耗指标数据。
  9. 根据权利要求1-8任一项所述的方法，其特征在于，根据所述目标设备在未来时段内的功耗指标数据，生成所述目标设备在未来时段内使用的功耗控制方案，包括：
    根据所述目标设备在未来时段内的功耗指标数据，确定预设触发条件中被匹配中的目标触发条件；
    获取与所述目标触发条件适配的目标功耗控制方法；
    根据所述目标触发条件和所述目标功耗控制方法，生成所述目标设备在未来时段内使用的功耗控制方案。
  10. 根据权利要求9所述的方法，其特征在于，根据所述目标设备在未来时段内的功耗指标数据，确定预设触发条件中被匹配中的目标触发条件，包括以下至少一种方式：
    根据所述未来时段对应的时间范围，从预设表示时间范围的触发条件中，确定所表示的时间范围落在所述未来时段对应的时间范围内的触发条件，作为目标触发条件；
    根据所述目标设备在未来时段内的功耗指标数据，从预设表示功耗指标数据范围的触发条件中，确定所表示的功耗指标数据范围在所述目标设备在未来时段内的功耗指标数据中出现过的触发条件，作为目标触发条件。
  11. 根据权利要求10所述的方法，其特征在于，获取与所述目标触发条件适配的目标功耗控制方法，包括：
    获取预设的与所述目标触发条件对应的功耗控制方法，作为所述目标功耗控制方法。
  12. 根据权利要求10所述的方法，其特征在于，获取与所述目标触发条件适配的目标功耗控制方法，包括：
    模拟各种功耗控制方法在所述目标触发条件下的节能效果；
    根据各种功耗控制方法在所述目标触发条件下的节能效果，从中选择节能效果满足节能要求的功耗控制方法，作为所述目标功耗控制方法。
  13. 根据权利要求9所述的方法，其特征在于，根据所述目标触发条件和所述目标功耗控制方法，生成所述目标设备在未来时段内使用的功耗控制方案，包括：
    获取预设的与所述目标触发条件对应的目标性能指标数据范围；
    根据所述目标触发条件、所述目标功耗控制方法以及所述目标性能指标数据范围，生成所述目标设备在未来时段内使用的功耗控制方案。
  14. 根据权利要求1-8任一项所述的方法，其特征在于，获取目标设备在历史时段内的功耗指标数据，包括：
    若所述目标设备是IT类设备，则获取所述目标设备在历史时段内的内部温度、功耗、CPU频率以及CPU负载中至少一种，作为所述目标设备在历史时段内的功耗指标数据；
    若所述目标设备是为IT类设备降温的空调设备，则获取所述目标设备在历史时段内的压缩机的频率、风机的转速、回风温度以及出风温度中至少一种，作为所述目标设备在历史时段内的功耗指标数据。
  15. 根据权利要求14所述的方法，其特征在于，在所述目标设备是IT类设备的情况下，还包括：获取与所述目标设备关联的其它设备在历史时段内的功耗，作为所述目标设备在历史时段内的功耗指标数据。
  16. 根据权利要求1-8任一项所述的方法，其特征在于，还包括：
    将所述目标设备在未来时段内使用的功耗控制方案提供给所述目标设备，以供所述目标设备在未来时段内进行功耗控制。
  17. 一种功耗控制方案生成方法，其特征在于，包括：
    获取设备组的相关参数，所述相关参数包括所述设备组在历史时段内的功耗指标数据，所述设备组包括至少一台物理设备；
    根据所述设备组在历史时段内的功耗指标数据，预测所述设备组在未 来时段内的功耗指标数据；
    根据所述设备组在未来时段内的功耗指标数据，生成所述设备组在未来时段内使用的功耗控制方案。
  18. 根据权利要求17所述的方法，其特征在于，所述设备组在历史时段内的功耗指标数据包括：所述设备组内各台物理设备在历史时段内的功耗指标数据；或者，所述设备组在历史时段内的功耗指标数据是根据所述设备组内各台物理设备在历史时段内的功耗指标数据计算出的功耗指标数据。
  19. 一种功耗控制方法，其特征在于，包括：
    接收新到达的第一功耗控制方案；
    将当前使用的第二功耗控制方案替换为所述第一功耗控制方案；
    根据所述第一功耗控制方案对设备进行功耗控制。
  20. 根据权利要求19所述的方法，其特征在于，根据所述第一功耗控制方案对设备进行功耗控制，包括：
    在所述第一功耗控制方案中的触发条件被满足的情况下，利用所述第一功耗控制方案中与所述被满足的触发条件对应的功耗控制方法对所述设备进行功耗控制。
  21. 根据权利要求19或20所述的方法，其特征在于，还包括：
    在根据所述第一功耗控制方案对所述设备进行功耗控制的过程中，监控所述设备的实际性能指标数据；
    根据所述设备的实际性能指标数据，调整所述第一功耗控制方案的控制强度。
  22. 根据权利要求21所述的方法，其特征在于，根据所述设备的实际性能指标数据，调整所述第一功耗控制方案的控制强度，包括：
    若所述设备的实际性能指标数据小于指定性能下限值，则关停所述第一功耗控制方案，直至所述设备的实际性能指标数据大于或等于指定性能下限值为止；或者
    若所述设备的实际性能指标数据小于指定性能下限值，则将所述第一功耗控制方案中的功耗控制方法调整为控制强度较低的功耗控制方法，直至所述设备的实际性能指标数据大于或等于指定性能下限值为止；
    其中，指定性能下限值是所述设备所能接受的性能下限值，或者是所述第一功耗控制方案中性能指标数据范围的下限值。
  23. 一种功耗控制方法，其特征在于，包括：
    接收新到达的第一功耗控制方案；
    根据所述第一功耗控制方案和当前使用的第二功耗控制方案，得到第三功耗控制方案；
    将当前使用的第二功耗控制方案替换为所述第三功耗控制方案，根据所述第三功耗控制方案对设备进行功耗控制。
  24. 根据权利要求23所述的方法，其特征在于，根据所述第一功耗控制方案和当前使用的第二功耗控制方案，得到第三功耗控制方案，包括：
    若所述第一功耗控制方案中的触发条件与所述第二功耗控制方案中的触发条件的类别不同，将所述第一功耗控制方案和所述第二功耗控制方案进行合并，得到所述第三控制方案；
    若所述第一功耗控制方案中的触发条件与所述第二功耗控制方案中的触发条件的类别相同，将所述第一功耗控制方案作为所述第三控制方案。
  25. 根据权利要求23所述的方法，其特征在于，根据所述第三功耗控制方案对设备进行功耗控制，包括：
    在所述第三功耗控制方案中的触发条件被满足的情况下，利用所述第三功耗控制方案中与所述被满足的触发条件对应的功耗控制方法对所述设备进行功耗控制。
  26. 根据权利要求23-25任一项所述的方法，其特征在于，还包括：
    在根据所述第三功耗控制方案对所述设备进行功耗控制的过程中，监控所述设备的实际性能指标数据；
    根据所述设备的实际性能指标数据，调整所述第三功耗控制方案的控 制强度。
  27. 根据权利要求26所述的方法，其特征在于，根据所述设备的实际性能指标数据，调整所述第三功耗控制方案的控制强度，包括：
    若所述设备的实际性能指标数据小于指定性能下限值，则关停所述第三功耗控制方案，直至所述设备的实际性能指标数据大于或等于指定性能下限值为止；或者
    若所述设备的实际性能指标数据小于指定性能下限值，则将所述第三功耗控制方案中的功耗控制方法调整为控制强度较低的功耗控制方法，直至所述设备的实际性能指标数据大于或等于指定性能下限值为止；
    其中，指定性能下限值是所述设备所能接受的性能下限值，或者是所述第三功耗控制方案中性能指标数据范围的下限值。
  28. 一种功耗预测方法，其特征在于，包括：
    获取目标设备的相关参数，所述相关参数包括所述目标设备在历史时段内的功耗指标数据；
    根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据。
  29. 根据权利要求28所述的方法，其特征在于，所述相关参数还包括所述目标设备的设备参数；
    在根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据之前，还包括：
    根据所述目标设备的设备参数和所述目标设备在历史时段内的功耗指标数据，识别所述目标设备是否属于功耗变化满足规律要求的设备；
    若所述目标设备属于功耗变化满足规律要求的设备，则执行根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据的操作。
  30. 根据权利要求29所述的方法，其特征在于，根据所述目标设备的设备参数和所述目标设备在历史时段内的功耗指标数据，对所述目标设备 进行分类，包括：
    将所述目标设备的设备参数和所述目标设备在历史时段内的功耗指标数据输入分类模型，以得到所述目标设备是否属于功耗变化满足规律要求的设备的分类结果。
  31. 根据权利要求28-30任一项所述的方法，其特征在于，根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据，包括：
    将所述目标设备在历史时段内的功耗指标数据输入预测模型，以得到所述目标设备在未来时段内的功耗指标数据。
  32. 根据权利要求31所述的方法，其特征在于，所述相关参数还包括所述目标设备在历史时段内的性能指标数据；
    将所述目标设备在历史时段内的功耗指标数据输入预测模型，以得到所述目标设备在未来时段内的功耗指标数据，包括：
    将所述目标设备在历史时段内的功耗指标数据以及所述目标设备在历史时段内的性能指标数据输入预测模型，以得到所述目标设备在未来时段内的功耗指标数据。
  33. 一种计算设备，其特征在于，包括：存储器和处理器；
    所述存储器，用于存储计算机程序；
    所述处理器，与所述存储器耦合，用于执行与所述计算机程序以用于：
    获取目标设备的相关参数，所述相关参数包括所述目标设备在历史时段内的功耗指标数据；
    根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据；
    根据所述目标设备在未来时段内的功耗指标数据，生成所述目标设备在未来时段内使用的功耗控制方案。
  34. 一种计算设备，其特征在于，包括：存储器和处理器；
    所述存储器，用于存储计算机程序；
    所述处理器，与所述存储器耦合，用于执行与所述计算机程序以用于：
    获取设备组的相关参数，所述相关参数包括所述设备组在历史时段内的功耗指标数据，所述设备组包括至少一台物理设备；
    根据所述设备组在历史时段内的功耗指标数据，预测所述设备组在未来时段内的功耗指标数据；
    根据所述设备组在未来时段内的功耗指标数据，生成所述设备组在未来时段内使用的功耗控制方案。
  35. 一种物理设备，其特征在于，包括：存储器、处理器以及通信组件；
    所述通信组件，用于接收新到达的第一功耗控制方案；
    所述存储器，用于存储计算机程序、所述第一功耗控制方案以及当前使用的第二功耗控制方案；
    所述处理器，与所述存储器耦合，用于执行与所述计算机程序以用于：
    将当前使用的第二功耗控制方案替换为所述第一功耗控制方案；
    根据所述第一功耗控制方案对所述物理设备进行功耗控制。
  36. 一种物理设备，其特征在于，包括：存储器、处理器以及通信组件；
    所述通信组件，用于接收新到达的第一功耗控制方案；
    所述存储器，用于存储计算机程序、所述第一功耗控制方案以及当前使用的第二功耗控制方案；
    所述处理器，与所述存储器耦合，用于执行与所述计算机程序以用于：
    根据所述第一功耗控制方案和当前使用的第二功耗控制方案，得到第三功耗控制方案；
    将当前使用的第二功耗控制方案替换为所述第三功耗控制方案，根据所述第三功耗控制方案对所述物理设备进行功耗控制。
  37. 一种计算设备，其特征在于，包括：存储器和处理器；
    所述存储器，用于存储计算机程序；
    所述处理器，与所述存储器耦合，用于执行与所述计算机程序以用于：
    获取目标设备的相关参数，所述相关参数包括所述目标设备在历史时段内的功耗指标数据；
    根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据。
  38. 一种数据中心系统，其特征在于，包括：至少一个机房和功耗控制设备；其中，每个机房包括至少一台物理设备；
    所述功耗控制设备，用于获取目标设备的相关参数，所述相关参数包括所述目标设备在历史时段内的功耗指标数据；根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据；根据所述目标设备在未来时段内的功耗指标数据，生成所述目标设备在未来时段内使用的功耗控制方案并提供给所述目标设备，以供所述目标设备在未来时段进行功耗控制；其中，所述目标设备是所述至少一台物理设备中任一台设备。
  39. 一种机房系统，其特征在于，包括：机房，所述机房内包含至少一台物理设备和功耗控制设备；
    所述功耗控制设备，用于获取目标设备的相关参数，所述相关参数包括所述目标设备在历史时段内的功耗指标数据；根据所述目标设备在历史时段内的功耗指标数据，预测所述目标设备在未来时段内的功耗指标数据；根据所述目标设备在未来时段内的功耗指标数据，生成所述目标设备在未来时段内使用的功耗控制方案并提供给所述目标设备，以供所述目标设备在未来时段进行功耗控制；其中，所述目标设备是所述至少一台物理设备中任一台设备。
  40. 一种存储有计算机程序的计算机可读存储介质，其特征在于，当所述计算机程序被处理器执行时，致使所述处理器实现权利要求1-30任一项所述方法中的步骤。
PCT/CN2019/104686 2019-09-06 2019-09-06 功耗控制与方案生成方法、设备、系统及存储介质 WO2021042368A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2019/104686 WO2021042368A1 (zh) 2019-09-06 2019-09-06 功耗控制与方案生成方法、设备、系统及存储介质
CN201980095638.9A CN113728294B (zh) 2019-09-06 2019-09-06 功耗控制与方案生成方法、设备、系统及存储介质

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2019/104686 WO2021042368A1 (zh) 2019-09-06 2019-09-06 功耗控制与方案生成方法、设备、系统及存储介质

Publications (1)

Publication Number Publication Date
WO2021042368A1 true WO2021042368A1 (zh) 2021-03-11

Family

ID=74852992

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/104686 WO2021042368A1 (zh) 2019-09-06 2019-09-06 功耗控制与方案生成方法、设备、系统及存储介质

Country Status (2)

Country Link
CN (1) CN113728294B (zh)
WO (1) WO2021042368A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113778211A (zh) * 2021-08-24 2021-12-10 联想(北京)有限公司 一种电源电路的控制方法、装置及电子设备

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116450446A (zh) * 2022-01-06 2023-07-18 长鑫存储技术有限公司 存储器功耗确定方法及装置、存储介质及电子设备

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104049716A (zh) * 2014-06-03 2014-09-17 中国科学院计算技术研究所 一种结合温度感知的计算机节能方法及系统
CN104534617A (zh) * 2014-12-08 2015-04-22 北京华电方胜技术发展有限公司 一种基于能耗监测的冷源集中数字控制方法
CN105577796A (zh) * 2015-12-25 2016-05-11 曙光信息产业(北京)有限公司 集群的功耗控制方法及装置
CN105659188A (zh) * 2013-11-27 2016-06-08 英特尔公司 上下文功率管理
CN109324902A (zh) * 2018-09-21 2019-02-12 深圳市中科明望通信软件有限公司 一种调整移动终端工作频率的方法、移动终端及存储介质
CN109388224A (zh) * 2018-09-26 2019-02-26 广东小天才科技有限公司 一种智能设备的功耗优化方法、系统及智能设备

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120144219A1 (en) * 2010-12-06 2012-06-07 International Business Machines Corporation Method of Making Power Saving Recommendations in a Server Pool
CN102736725B (zh) * 2012-05-18 2016-03-30 华为技术有限公司 一种硬盘节能控制方法、装置及中央处理器
CN103605418B (zh) * 2013-10-23 2017-01-04 曙光信息产业(北京)有限公司 集群服务器的功耗调节方法和装置
CN103902016A (zh) * 2014-04-28 2014-07-02 浪潮电子信息产业股份有限公司 一种面向场景预测的服务器功耗管理方法
CN109725702A (zh) * 2018-12-28 2019-05-07 三星电子（中国）研发中心 一种基于ai预测的智能终端节能方法和设备

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105659188A (zh) * 2013-11-27 2016-06-08 英特尔公司 上下文功率管理
CN104049716A (zh) * 2014-06-03 2014-09-17 中国科学院计算技术研究所 一种结合温度感知的计算机节能方法及系统
CN104534617A (zh) * 2014-12-08 2015-04-22 北京华电方胜技术发展有限公司 一种基于能耗监测的冷源集中数字控制方法
CN105577796A (zh) * 2015-12-25 2016-05-11 曙光信息产业(北京)有限公司 集群的功耗控制方法及装置
CN109324902A (zh) * 2018-09-21 2019-02-12 深圳市中科明望通信软件有限公司 一种调整移动终端工作频率的方法、移动终端及存储介质
CN109388224A (zh) * 2018-09-26 2019-02-26 广东小天才科技有限公司 一种智能设备的功耗优化方法、系统及智能设备

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113778211A (zh) * 2021-08-24 2021-12-10 联想(北京)有限公司 一种电源电路的控制方法、装置及电子设备

Also Published As

Publication number Publication date
CN113728294A (zh) 2021-11-30
CN113728294B (zh) 2023-03-31

Similar Documents

Publication Publication Date Title
CN109960395B (zh) 资源调度方法和计算机设备
Khan et al. Machine learning (ML)-centric resource management in cloud computing: A review and future directions
US10554786B2 (en) Dynamic adjustment of mobile device based on peer event data
US9509632B2 (en) Workload prediction for network-based computing
US9462965B2 (en) Dynamic adjustment of mobile device based on system events
US20200057771A1 (en) Users campaign for peaking energy usage
US9465679B2 (en) Dynamic adjustment of mobile device based on adaptive prediction of system events
WO2019062417A1 (zh) 应用清理方法、装置、存储介质及电子设备
US20200225995A1 (en) Application cleaning method, storage medium and electronic device
US9813990B2 (en) Dynamic adjustment of mobile device based on voter feedback
WO2021042368A1 (zh) 功耗控制与方案生成方法、设备、系统及存储介质
EP3690603A1 (en) Application cleaning method and apparatus, storage medium and electronic device
WO2019085754A1 (zh) 应用清理方法、装置、存储介质及电子设备
US9911063B1 (en) Identifying images using pre-loaded image identifiers
CN114598665A (zh) 资源调度方法、装置和计算机可读存储介质及电子设备
Mahmud et al. Power profiling of context aware systems: a contemporary analysis and framework for power conservation
US20140122403A1 (en) Loading prediction method and electronic device using the same
JP7121251B2 (ja) 分析装置、分析方法およびプログラム
CN109062396B (zh) 用于控制设备的方法和装置
CN112486683B (zh) 处理器控制方法、控制设备以及计算机可读存储介质
US11336424B1 (en) Clock drift estimation
US20180373316A1 (en) Target Based Power Management
WO2021042373A1 (zh) 数据处理与任务调度方法、设备、系统及存储介质
WO2019085742A1 (zh) 后台应用清理方法、装置、存储介质及电子设备
Hasegawa et al. An experimental result of estimating an application volume by machine learning techniques

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: 19944051

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19944051

Country of ref document: EP

Kind code of ref document: A1