CN114650190B - Energy-saving method, system, terminal equipment and storage medium for data center network - Google Patents
Energy-saving method, system, terminal equipment and storage medium for data center network Download PDFInfo
- Publication number
- CN114650190B CN114650190B CN202011498034.XA CN202011498034A CN114650190B CN 114650190 B CN114650190 B CN 114650190B CN 202011498034 A CN202011498034 A CN 202011498034A CN 114650190 B CN114650190 B CN 114650190B
- Authority
- CN
- China
- Prior art keywords
- energy
- data
- saving
- energy consumption
- evaluation
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L12/00—Data switching networks
- H04L12/02—Details
- H04L12/12—Arrangements for remote connection or disconnection of substations or of equipment thereof
-
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/08—Configuration management of networks or network elements
- H04L41/0803—Configuration setting
- H04L41/0823—Configuration setting characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability
- H04L41/0833—Configuration setting characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability for reduction of network energy consumption
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to the technical field of energy conservation, and provides an energy conservation method of a data center network, which comprises the following steps: acquiring energy consumption factor data of a data center network; performing energy-saving decision processing according to the energy consumption factor data, and outputting energy-saving optimization adjustment actions; and adjusting the energy consumption nodes of the data center network according to the energy saving optimization adjustment action. Correspondingly, the application also provides an energy-saving system, terminal equipment and a readable storage medium of the data center network. By implementing the method and the device, efficient energy-saving adjustment of the data center network is realized.
Description
Technical Field
The application relates to the technical field of energy conservation, and in particular provides an energy conservation method, an energy conservation system, terminal equipment and a storage medium of a data center network.
Background
The energy-saving mode of the current data center network mainly comprises the following three modes:
(1) Energy saving for server
The server is a core component of the data center and is also one of the main objects of energy consumption. Consumption reduction is performed on a server under the condition that the processing capacity of a data center is not affected, and the consumption reduction is one of important modes for saving energy of a large-scale data center at present. The occurrence of virtualization technology brings new directions for energy conservation of servers, and by changing physical hardware of the servers into resources for management through virtualization, the use of configuration can be optimized, the number of the servers is reduced, and therefore energy consumption generated by the servers is reduced. Meanwhile, in the network data transmission process, the network equipment also accompanies a large amount of energy consumption, and the switching times and the frequency of the equipment can be adjusted by formulating an efficient equipment dormancy strategy, so that the energy conservation of the equipment level is realized; in the network-level energy-saving technology, the energy-saving routing path is calculated and configured according to the real-time network load, so that the network energy consumption can be reduced.
(2) Energy saving for air conditioning system
Traditional data centers cool equipment primarily through a cooling environment, while the energy consumption produced by air conditioning systems is enormous. Therefore, intelligent control of the air conditioning system is also one of the important ways of saving energy for the data center. The energy-saving optimization of the air conditioning refrigerating unit is always an important research direction of an air conditioning system, and the air conditioning refrigerating unit is subjected to energy-saving management in several modes of equipment energy saving, system energy saving and natural cooling, so that the energy saving and consumption reduction of the air conditioning system are finally realized.
(3) Temperature control energy saving
The refrigeration strategy of the data center is mainly to reasonably allocate and regulate according to the temperature of a machine room, and the judgment of the environmental temperature is particularly important. There are two main modes commonly used at present, one is to use a temperature collector to perform multipoint measurement on the temperature of a data center, and then to perform intelligent analysis and control through a controller. And the other is to predict the temperature of the machine room according to the historical parameters and control the temperature in advance.
The three energy-saving modes respectively have the following defects:
(1) The energy-saving mode of the server is as follows: along with the update and upgrade of the hardware equipment of the server, although many components can directly measure the real-time energy consumption data generated by the equipment, the energy consumption condition of a future machine room cannot be effectively predicted and perceived by means of the physical means.
(2) The energy-saving mode of the air conditioning system is as follows: in order to ensure stable operation of the air conditioning system of the data center, an uninterrupted set of ensuring system is required. However, in order to meet the temperature requirement of the machine room, most data centers can be refrigerated through an excessive cooling strategy, and truly effective energy saving and consumption reduction are still not realized.
(3) Temperature control energy-saving mode: because the indoor environment of the data center machine room is complex, a certain difficulty is caused in temperature control and energy saving, and the traditional control method is difficult to obtain a satisfactory control effect.
Disclosure of Invention
The purpose of the application is to provide an energy-saving method, an energy-saving system, a terminal device and a storage medium of a data center network, and aims to solve the existing problem that the prior art cannot adapt to a complex machine room environment and perform efficient energy-saving adjustment.
In order to achieve the above purpose, the technical scheme adopted in the application is as follows:
in a first aspect, the present application provides a method for saving energy in a data center network, including:
acquiring energy consumption factor data of a data center network;
performing energy-saving decision processing according to the energy consumption factor data, and outputting energy-saving optimization adjustment actions;
adjusting the energy consumption nodes of the data center network according to the energy saving optimization adjustment action
In a second aspect, the present application further provides an energy saving system of a data center network, including:
the acquisition module is used for acquiring the energy consumption factor data of the data center network;
the energy-saving decision module is used for carrying out energy-saving decision processing according to the energy consumption factor data and outputting energy-saving optimization adjustment actions;
and the node adjusting module is used for adjusting the energy consumption nodes of the data center network according to the energy saving optimization adjusting action.
In a third aspect, the present application further provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor implements the above-mentioned energy saving method when executing the computer program.
In a fourth aspect, the present application also provides a computer readable storage medium storing a computer program which, when executed by a processor, implements the above-described energy saving method.
The beneficial effects of this application:
according to the energy-saving method, the system, the terminal equipment and the storage medium for the data center network, various energy consumption factors are acquired in real time to master complex machine room environments, then energy-saving decision is made on energy consumption factor data, prospective and adaptive energy-saving optimization adjustment actions are obtained, and then all nodes of the data center network are adjusted according to the energy-saving optimization adjustment actions, so that efficient energy-saving adjustment of the data center network is achieved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required for the embodiments or the description of the prior art will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a first embodiment of a method for conserving energy in a data center network of the present application;
FIG. 2 is a flowchart of an energy saving decision process according to the energy consumption factor data according to an embodiment of an energy saving method of the data center network of the present application;
FIG. 3 is an expanded flow chart of a first embodiment of a method for conserving energy in a data center network of the present application;
FIG. 4 is a flow chart of a second embodiment of an energy saving method for a data center network of the present application;
FIG. 5 is a flow chart of a third embodiment of a method for conserving energy in a data center network of the present application;
FIG. 6 is a flowchart of an energy saving decision process according to the energy consumption factor data according to a third embodiment of the energy saving method of the data center network of the present application;
FIG. 7 is a block diagram of an embodiment of an energy saving system of a data center network of the present application;
FIG. 8 is a block diagram of an energy conservation decision module of an embodiment of an energy conservation system of a data center network of the present application;
fig. 9 is a block diagram of a terminal device of the present application;
wherein, each reference sign in the figure:
Detailed Description
Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals refer to the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and intended for the purpose of explaining the present application and are not to be construed as limiting the present application.
In the description of the present application, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," and the like indicate orientations or positional relationships based on the orientation or positional relationships shown in the drawings, merely to facilitate description of the present application and simplify the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be configured and operated in a specific orientation, and therefore should not be construed as limiting the present application.
Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include one or more such feature. In the description of the present application, the meaning of "a plurality" is two or more, unless explicitly defined otherwise.
In this application, unless specifically stated and limited otherwise, the terms "mounted," "connected," "secured," and the like are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally formed; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communicated with the inside of two elements or the interaction relationship of the two elements. The specific meaning of the terms in this application will be understood by those of ordinary skill in the art as the case may be.
The energy saving method of the data center network provided by the embodiment of the application can be applied to terminal equipment such as desktop computers, servers, mobile phones, tablet computers, notebook computers, ultra-Mobile Personal Computer (UMPC), netbooks, personal digital assistants (Personal Digital Assistant, PDA) and the like, and the embodiment of the application does not limit the specific types of the terminal equipment.
In order to illustrate the technical solutions described in the present application, the following examples are provided.
Example 1
Referring to fig. 1, an energy saving method for a data center network according to an embodiment of the present application includes:
step S101, energy consumption factor data of a data center network are obtained.
In one embodiment, the data source of the energy consumption factor data may include three modules, that is, a system linkage module, a weather data module, and an energy monitoring platform, which store data center subsystem data, weather environment data, and energy data, respectively. Specifically, the system linkage module mainly connects and integrates all subsystems of the energy consumption node, so that sharing and unified processing and analysis of data are realized, all subsystems of the data center are combined into an organic whole, and all data of the energy consumption node are better acquired. The weather data module is mainly used for monitoring and analyzing the network operation environment of the data center according to data such as weather conditions, outdoor temperature, illumination, humidity and the like, and adjusting the network operation environment according to different weather conditions so as to adapt to weather changes. The energy supervision platform is mainly used for monitoring the energy use condition of each system of the data center in real time and intelligently analyzing the rationality of energy use. The three functional modules are used for collecting comprehensive real-time data and historical data to form various data influencing energy consumption efficiency, and the data are used as a data base of the energy saving method.
And step S102, performing energy-saving decision processing according to the energy consumption factor data, and outputting an energy-saving optimization adjustment action.
Referring to fig. 2, in one embodiment, the step of performing energy-saving decision processing according to the energy consumption factor data and outputting an energy-saving optimization adjustment action includes:
step S1020, generating an initial adjustment action according to the energy consumption factor data.
In one embodiment, the initial adjustment may be a parameter adjustment based on energy consumption factor data. For example, a certain server of the data center network is used as an energy consumption node to dissipate heat, a machine room needs to be kept at a specific temperature, at this time, the refrigeration temperature value of an air conditioner or other refrigeration equipment needs to be adjusted, at this time, the energy consumption factor data is the temperature of the machine room, for example, 20 ℃, and the initial adjustment action is an execution action of adjusting the refrigeration temperature of the refrigeration equipment to a target temperature value, for example, adjusting the refrigeration temperature of the refrigeration equipment to 18 ℃.
Step S1021, the energy consumption factor data and the initial adjustment action are input into an evaluation neural network so as to drive the evaluation neural network to output a first evaluation value.
In one embodiment, the initial adjustment action is input as a parameter into the evaluation neural network along with energy consumption factor data that may reflect the energy consumption state, and then the evaluation value is calculated by the evaluation neural network, and the evaluation function may be:
Q eval (s，a)＝Q(s，a)+α*[Q target (s，a)-Q(s，a)]
Q target (s，a)＝r+γmaxQ(s，a)
Wherein s is state data corresponding to the initial adjustment action, a is the initial adjustment action, and alpha, r and gamma are evaluation parameters.
In application, the parameters of the estimated neural network can be trained by back propagation with a loss function of Q target Subtracting Q eval The formula is as follows:
L(θ)＝E[(r+γmaxQ(s′，a′，θ)-Q(s，a，θ)) 2 ]
wherein s 'is the next acquired energy consumption factor data, a' is the next initial adjustment action, s is the energy consumption factor data evaluation value, a is the initial adjustment action evaluation value, and θ is the parameter for evaluating the neural network.
Step S1022, adjusting the energy consumption node according to the initial adjustment action.
Step S1023, obtaining operation data and state data generated by the operation of the energy consumption node in the environment, and calculating rewarding parameters according to the operation data.
In one embodiment, the operation data may be data such as a completion time, a cache space, and calculation accuracy when the energy consumption node performs data processing. The state data may be new energy consumption factor data generated by the energy consumption node after being adjusted by the initial adjustment action and operated in the environment for a certain time.
Due to time and accuracy constraints, the prize R may be awarded t Defined as a weighted sum of the completion time ratio, the accuracy ratio, and the server resource occupancy ratio. The formula for calculating the bonus parameters from the operational data may be:
Wherein L is t To complete time, T t For the task state, T c Caching for edge server, A t For video block analysis, L α For accuracy, S t Caching space for remaining edge servers, remaininS t For the remaining buffer space in the edge server, α, β, γ represent the weights of the three parts, respectively.
As can be seen from the above equation, exceeding a deadline or failing to meet the accuracy requirements will result in a negative rewards, the greater the cache space remaining in the edge server, the higher the rewards.
Step S1024, selecting the initial adjustment motion with the bonus parameter being positive as the sample adjustment motion.
In one embodiment, the reward parameter is a measure of whether the initial adjustment action has a beneficial effect on the adjustment of the energy consumption node, for example, after the adjustment of the initial adjustment action, the new energy consumption factor data reflects the energy consumption efficiency to be increased, so that the energy saving effect is achieved, and if the reward parameter is a positive number, the reward parameter is a negative number. Therefore, when the reward parameter is positive, the initial adjustment action can obtain energy saving effect at least in the current adjustment, belongs to beneficial experience, and can be sample adjustment action.
Step S1025, the sample adjusting action and the corresponding state data are taken as samples to be put into an experience replay pool.
In one real-time example, the empirical replay pool may improve algorithm efficiency, avoiding trapping in a locally optimal solution. The system is a cache pool and is used for recording state data, execution actions and execution results of the state data, the execution actions and the execution results of the state data, the execution actions and the execution results of the state data. Of course, the bonus parameter values themselves may also be placed together in an experience replay pool. The empirical replay pool is used for subsequent random sampling of training data by which correlations between samples are obtained.
In application, the mechanism of the experience replay pool can be improved by sample screening and priority-based probability sampling. In order to improve the validity of the sample, a certain screening can be performed when the sample is recorded so as to meet the constraint condition of task scheduling. Constraint can be directly added to the loss function, such as completion time L t Task state T t And edge server cache T c Build constraint L t -T t +T c L t The loss function of the added constraint is:
L’(θ)＝E[(r+γmaxQ(s′，a′，θ)-Q(s，a，θ)) 2 ]+L t -T t +T c L t
the variables and parameters in the above formula are the same as those of the loss function L (θ), and will not be described in detail here.
At this time, the condition of the loss function convergence of the added constraint condition is satisfied, so that the sample can be recorded in the experience replay pool.
In addition, to avoid losing experience of long tail distribution, a punishment mechanism can be added, so that people who do not meet constraint conditions update experience playback pools under preset probability, rather than updating every time.
Step S1026, randomly sampling in the experience replay pool, generating random samples.
In one embodiment, all sample conditioning actions and their corresponding status data that are verified to have a power saving effect during a particular conditioning process are collected by the experience replay pool. These samples have value for further evaluation, validation.
Step S1027, the random samples are respectively input into the evaluation neural network and the target neural network to drive the evaluation neural network to generate a first evaluation value and drive the target neural network to generate a second evaluation value.
In one embodiment, the evaluation formulas for evaluating the neural network and the target neural network are the same, except that the parameters of the target neural network are relatively fixed and the frequency of variation is less.
Step S1028, performing gradient ascent method evaluation on the first evaluation value and the second evaluation value respectively, and generating a first gradient evaluation result.
Step S1029, when the first gradient evaluation result is that a local maximum value is obtained, using the sample adjustment action corresponding to the local maximum value as the energy-saving optimization adjustment action.
In one embodiment, the first evaluation value and the second evaluation value are evaluated by a gradient ascent method, that is, a sequence is formed by randomly extracted samples and a local maximum value in the sequence is found, a corresponding sample adjustment action is determined by the local maximum value, and the corresponding sample adjustment action is used as an energy-saving optimization adjustment action to obtain a local maximum evaluation value, that is, a local optimal energy-saving scheme.
And step S103, adjusting the energy consumption nodes of the data center network according to the energy saving optimization adjustment action.
In one embodiment, the energy consumption nodes can be energy consumption nodes in all subsystems of a fire protection system, a safety system, a building control system, a power environment monitoring and operation and maintenance management of the data center, and the energy consumption nodes are comprehensively adjusted, so that the overall energy conservation and consumption reduction of the data center are realized.
Referring to fig. 3, in one embodiment, the energy saving method further includes:
step S104, after the sample adjusting action corresponding to the local maximum value is taken as the energy-saving optimizing adjusting action, respectively performing gradient descent method evaluation on the first evaluation value and the second evaluation value to generate a second gradient evaluation result;
step S105, updating the network parameters of the estimated neural network according to the second gradient estimation result, and performing soft update on the network parameters of the target neural network.
The gradient descent method is opposite to the gradient ascent method, and is to randomly extract samples to form a sequence and search local minimum values in the sequence, and the parameters of the evaluation neural network and the parameters of the target neural network trained by the local minimum values are required to be increased continuously, so that the evaluation values corresponding to the local minimum values are increased, the evaluation values corresponding to the whole sequence are driven to be increased, and the parameters of the trained evaluation neural network and the target neural network can calculate higher evaluation values, so that a more efficient energy-saving optimization adjustment action is obtained.
Because the evaluation neural network performs back propagation training and gradient descent training, the network parameter fluctuation frequency is very high, short-time fluctuation of energy consumption factor data is easily conducted to the network parameter, and further fluctuation of an evaluation value is caused, so that energy-saving adjustment is influenced. Thus, the parameters of the target neural network need to be maintained relatively stable. Because the evaluation neural network has the same structure as the target neural network, the parameters of the target neural network can be updated in a soft mode, namely, the parameters of the target neural network are not updated every time the second gradient evaluation result is obtained, but the parameters of the evaluation neural network are directly copied according to a preset period, so that the parameters of the target neural network are relatively stable.
In the application, since the parameter update frequencies of the evaluation neural network and the target neural network are different, in the steps S1028-S1029, when the first evaluation result is obtained, if the local maximum value of the evaluation neural network is inconsistent with the local maximum value of the target neural network, one of the local maximum values may be selected according to the need, for example, the local maximum value of the target neural network may be selected to maintain the evaluation stability, or a larger local maximum value may be adopted to obtain a better energy saving effect.
It will be appreciated that the training of steps S104-S105 needs to be performed a number of times, and therefore, after step S105 is completed, the process will return to step S101, and the process will continue until the preset training stop condition is met.
According to the energy-saving method embodiment of the data center network, various energy consumption factors are acquired in real time to master complex machine room environments, then energy-saving decision is made on energy consumption factor data, prospective and adaptive energy-saving optimization adjustment actions are obtained, and then all nodes of the data center network are adjusted according to the energy-saving optimization adjustment actions, so that efficient energy-saving adjustment of the data center network is achieved.
Example two
The embodiment of the present application provides an energy saving method for a data center network, including steps S101 to S103 in the first embodiment, which is further described in the first embodiment, and the same or similar parts as those in the first embodiment can be referred to in the description related to the first embodiment, and will not be repeated here.
The specific steps of step S202 of the second embodiment are different from those of step S102 of the first embodiment, and are described in detail below. Meanwhile, the second embodiment may further include steps S204 to S207.
Specifically, referring to fig. 4, the energy saving method of the data center network in the present embodiment includes:
Step S201, energy consumption factor data of the data center network is obtained.
And step S202, performing energy-saving decision processing according to the energy consumption factor data, and outputting an energy-saving optimization adjustment action.
In one embodiment, the step of performing energy-saving decision processing according to the energy consumption factor data and outputting an energy-saving optimization adjustment action includes:
and inputting the energy consumption factor data into a convolutional neural network to drive the convolutional neural network to output energy-saving optimization adjustment actions.
In one embodiment, the convolutional neural network is trained for a plurality of times, has good energy-saving optimization and adjustment action prediction capability, has a simple structure, and is suitable for deployment of a lightweight energy-saving platform.
In addition, the steps S1021-S1029 are two mutually independent methods, can process the energy consumption factor data in parallel and output corresponding energy-saving optimization adjustment actions, and can be selected according to the needs if the energy-saving optimization adjustment actions output by the two methods are inconsistent.
And step S203, adjusting the energy consumption nodes of the data center network according to the energy saving optimization adjustment action.
In this embodiment, the steps S201, S202 and S203 are the same as or similar to the steps S101, S102 and S103 of the first embodiment, respectively, and the detailed description of the steps S101 to S103 will be omitted herein.
In one embodiment, the energy saving method further comprises:
and step S204, performing simulation adjustment according to the energy-saving optimization adjustment action to obtain simulation operation data and simulation state data.
In one embodiment, the simulation adjustment is designed to solve the problem that the number of training times of the convolutional neural network is large, and in practice, historical data is insufficient, and a large amount of simulation operation data and simulation state data can be obtained through the simulation adjustment to serve as the historical data, so that the training efficiency of the convolutional neural network is improved.
Step S205, updating the data source of the energy consumption factor data according to the simulation operation data and the simulation state data.
In one embodiment, updating the data source may be adding newly added simulation run data and simulation state data to the data source, thereby accumulating historical data continuously.
Step S206, calculating the overall power consumption and corresponding rewarding value of all energy consumption nodes of the data center network according to the updated energy consumption factor data.
In one embodiment, the energy saving effect is learned from historical data of the overall power consumption by comparing, and the prize value is used to evaluate the energy saving effect of the single energy saving optimization adjustment action, both data reflecting the simulated adjustment energy saving effect from different aspects.
And step S207, when the overall power consumption is reduced and/or the rewarding value is positive, outputting the updated energy consumption factor data to the convolutional neural network so as to drive the convolutional neural network to carry out the next energy saving optimization scheme decision.
In one embodiment, the analog adjustment may be performed by a simulator that integrates various data sources of energy consumption factor data, as well as a control interface. The control interface is used for reading the energy consumption factor data of the data source and inputting the energy consumption factor data into the convolutional neural network, and simultaneously performing simulation adjustment and updating the data source of the energy consumption factor data according to the simulation operation data and the simulation state data. In application, the simulator may be a dedicated server or may be a dedicated software module in an existing server.
It can be appreciated that the simulated training is a continuously cyclical process, with the goal of efficiently training the convolutional neural network through the simulated training to achieve good predictive power. Therefore, after the end of step S207, the process returns to step S201 to acquire new energy consumption factor data again for the next training.
According to the energy-saving method embodiment of the data center network, various energy consumption factors are acquired in real time to master complex machine room environments, then energy-saving decision is made on energy consumption factor data, prospective and adaptive energy-saving optimization adjustment actions are obtained, and then all nodes of the data center network are adjusted according to the energy-saving optimization adjustment actions, so that efficient energy-saving adjustment of the data center network is achieved.
Example III
The embodiment of the present application provides an energy saving method for a data center network, including steps S101 to S103 in the first embodiment, which is further described in the first embodiment, and the same or similar parts as those in the first embodiment can be referred to in the description related to the first embodiment, and will not be repeated here.
Unlike the first embodiment, the step S302 may further include steps S3021 to S3022, while the third embodiment further includes steps S304 to S307.
Specifically, referring to fig. 5, the energy saving method of the data center network in the present embodiment includes:
step S301, energy consumption factor data of the data center network is obtained.
Step S302, energy-saving decision processing is carried out according to the energy consumption factor data, and energy-saving optimization adjustment actions are output
Referring to fig. 6, in one embodiment, the step of performing energy-saving decision processing according to the energy consumption factor data and outputting an energy-saving optimization adjustment action includes:
step S3021, inputting the energy consumption factor data into the machine learning model to drive the machine learning model to output a predicted energy efficiency value.
In one embodiment, the machine learning model may be a decision tree that outputs the predicted energy efficiency value by making a classification decision on the energy consumption factor data. The energy efficiency value refers to an energy efficiency ratio, such as a refrigeration energy efficiency ratio EER, a heating energy efficiency ratio COP, and the like. The energy efficiency value is an important index for measuring energy saving efficiency. Therefore, the energy efficiency value is accurately predicted, and a basis can be provided for the follow-up determination execution action.
And step S3022, regulating the load of the cooling unit according to the predicted energy efficiency value and the cooling demand.
In one embodiment, the execution of the cooling unit load is energy saving optimization, such as adjusting the outlet water temperature and/or the number of cooling units turned on. The machine learning model does not directly output the execution action, but rather ultimately determines a specific execution action by outputting an intermediate quantity, i.e., the predicted energy efficiency value, and other intermediate quantities, such as the cooling demand. The determined rule may be to minimize the energy efficiency value while ensuring that the cooling demand is satisfied. At this time, the machine learning model needs to be trained, so that the machine learning model outputs an accurate prediction energy efficiency value, and the prediction energy efficiency value is used as an accurate execution action to improve the energy saving efficiency.
Step S303, adjusting the energy consumption nodes of the data center network according to the energy saving optimization adjustment action.
In this embodiment, the steps S301, S302, and S303 are the same as or similar to the steps S101, S102, and S103 in the first embodiment, and the detailed description of the steps S101 to S103 is omitted here.
In one embodiment, the energy saving method further comprises:
Step S304, converting the energy consumption factor data into historical data and storing the historical data.
Step S305, inputting the history data into the machine learning model to drive the machine learning model to output a predicted energy efficiency value;
step S306, calculating an actual energy efficiency value according to the historical data;
step S307, adjusting parameters of the machine learning model according to the predicted energy efficiency value and the actual energy efficiency value.
In one embodiment, the machine learning model comprises 500 decision trees connected in sequence, historical data is input at a starting end, and a predicted energy efficiency value is output at an output end. In addition, the actual energy efficiency value in a specific period can be calculated through the historical data, and the node parameters of the decision tree are reversely adjusted according to the comparison result by comparing the predicted energy efficiency value and the actual energy efficiency value, so that the prediction accuracy of the decision tree is improved.
It will be appreciated that decision tree training is a constantly recurring process with the goal of improving the accuracy of decision tree predictions through training. Therefore, after the end of step S307, the process returns to step S301 to re-acquire new energy consumption factor data for the next training.
In addition, the three methods of step S1021-step S1029, step S2021, and step S3021-step S3023 are three mutually independent methods, and can process the energy consumption factor data in parallel, output the corresponding energy-saving optimization adjustment action, and if the energy-saving optimization adjustment actions output by the three methods are inconsistent, the selection can be performed according to the needs. For convenience of explanation, fig. 6 shows only a part of steps S3021 to S3023 in step S302.
According to the energy-saving method embodiment of the data center network, various energy consumption factors are acquired in real time to master complex machine room environments, then energy-saving decision is made on energy consumption factor data, prospective and adaptive energy-saving optimization adjustment actions are obtained, and then all nodes of the data center network are adjusted according to the energy-saving optimization adjustment actions, so that efficient energy-saving adjustment of the data center network is achieved.
Example IV
Fig. 7 shows a block diagram of an energy saving system 100 of a data center network according to an embodiment of the present application, which may be a virtual device (virtual appliance) in a terminal device, and may be executed by a processor of the terminal device or may be integrated in the terminal device itself. For convenience of explanation, only portions relevant to the embodiments of the present application are shown.
The energy saving system 100 of the data center network according to the embodiment of the present application includes:
and the acquisition module 1 is used for acquiring the energy consumption factor data of the data center network.
And the energy-saving decision module 2 is used for carrying out energy-saving decision processing according to the energy consumption factor data and outputting energy-saving optimization adjustment actions.
Referring to fig. 8, in one embodiment, the energy saving decision module 2 includes:
An initial action unit 201, configured to generate an initial adjustment action according to the energy consumption factor data;
an evaluation-value generating unit 202, configured to input the energy consumption factor data and the initial adjustment action to an evaluation neural network, so as to drive the evaluation neural network to output a first evaluation value;
an initial adjustment unit 203, configured to adjust the energy consumption node according to the initial adjustment action;
an environmental parameter obtaining unit 204, configured to obtain operation data and status data generated by the energy consumption node operating in the environment, and calculate a reward parameter according to the operation data;
a sample action unit 205 for selecting an initial adjustment action with a positive bonus parameter as a sample adjustment action;
a sample pool unit 206, configured to put the sample adjustment actions and corresponding status data as samples into an experience replay pool;
a sampling unit 207 for randomly sampling in the empirical playback pool to generate random samples;
a two-network value calculation unit 208, configured to input the random samples to the evaluation neural network and the target neural network, respectively, so as to drive the evaluation neural network to generate a first evaluation value and drive the target neural network to generate a second evaluation value;
A gradient raising unit 209, configured to perform gradient raising method evaluation on the first evaluation value and the second evaluation value, respectively, to generate a first gradient evaluation result;
and the extremum action output unit 210 is configured to, when the first gradient evaluation result is that a local maximum value is obtained, take the sample adjustment action corresponding to the local maximum value as the energy-saving optimization adjustment action.
In one embodiment, the energy saving decision module 2 further comprises:
and the convolution action output unit 211 is used for inputting the energy consumption factor data into a convolution neural network so as to drive the convolution neural network to output an energy-saving optimization adjustment action.
In one embodiment, the energy saving decision module 2 further comprises:
an energy efficiency prediction unit 212, configured to input the energy consumption factor data into the machine learning model, so as to drive the machine learning model to output a predicted energy efficiency value;
and the chiller load adjustment unit 213 is configured to adjust the chiller load according to the predicted energy efficiency value and the cooling demand.
And the node adjusting module 3 is used for adjusting the energy consumption nodes of the data center network according to the energy saving optimization adjusting action.
In one embodiment, the energy saving system further comprises:
The gradient descent module 4 is configured to perform gradient descent method evaluation on the first evaluation value and the second evaluation value respectively after the sample adjustment action corresponding to the local maximum value is used as the energy-saving optimization adjustment action, so as to generate a second gradient evaluation result;
and the two-network updating module 5 is used for updating the network parameters of the evaluation neural network according to the second gradient evaluation result and carrying out soft update on the network parameters of the target neural network.
In one embodiment, the energy saving system further comprises:
the simulation adjustment module 6 is used for performing simulation adjustment according to the energy-saving optimization adjustment action to obtain simulation operation data and simulation state data;
the data source updating module 7 is used for updating the data source of the energy consumption factor data according to the simulation operation data and the simulation state data;
the power consumption rewarding calculation module 8 is used for calculating the overall power consumption and corresponding rewarding value of all the energy consumption nodes of the data center network according to the updated energy consumption factor data;
and the convolution retraining module 9 is used for outputting the updated energy consumption factor data to the convolution neural network when the overall power consumption is reduced and/or the reward value is positive, so as to drive the convolution neural network to carry out the next energy-saving optimization scheme decision.
In one embodiment, the energy saving system further comprises:
the history conversion storage module 10 is configured to convert the energy consumption factor data into history data and store the history data;
a machine prediction module 11 for inputting the history data into the machine learning model to drive the machine learning model to output a predicted energy efficiency value;
an actual energy efficiency calculation module 12, configured to calculate an actual energy efficiency value according to the historical data;
and the machine parameter adjustment module 13 is used for adjusting parameters of the machine learning model according to the predicted energy efficiency value and the actual energy efficiency value.
According to the energy-saving system embodiment of the data center network, various energy consumption factors are acquired in real time so as to master a complex machine room environment, then energy-saving decision is made on energy consumption factor data, prospective and adaptive energy-saving optimization adjustment actions are obtained, and each node of the data center network is adjusted according to the energy-saving optimization adjustment actions, so that efficient energy-saving adjustment of the data center network is achieved.
Example five
As shown in fig. 9, the present application also provides a terminal device 300 comprising a memory 301, a processor 302 and a computer program 303, such as a data center network power saving program, stored in and executable on said memory. The processor 302, when executing the computer program 303, implements the steps of the energy saving method embodiments of the data center networks described above, for example, the method steps of the first and/or second embodiments. The processor 302 implements the functions of the modules in the above-described embodiments of the apparatus, for example, the functions of the modules and units in the third embodiment, when executing the computer program 303.
Illustratively, the computer program 303 may be divided into one or more modules, which are stored in the memory 301 and executed by the processor 302 to accomplish the first, second, third, and/or fourth embodiments of the present application. The one or more modules may be a series of computer program instruction segments capable of performing specific functions for describing the execution of the computer program 303 in the terminal device 300. For example, the computer program 303 may be divided into an acquisition module 1, an energy saving decision module 2, a node adjustment module 3, and the like, where specific functions of each module are described in the fourth embodiment, and are not described herein.
The terminal device 300 may be a desktop computer, a mobile phone, a server, etc. The terminal device may include, but is not limited to, a memory 301, a processor 302. It will be appreciated by those skilled in the art that fig. 9 is merely an example of a terminal device 300 and is not limiting of the terminal device 300, and may include more or fewer components than shown, or may combine certain components, or different components, e.g., the terminal device may also include input and output devices, network access devices, buses, etc.
The memory 301 may be an internal storage unit of the terminal device 300, for example, a hard disk or a memory of the terminal device 300. The memory 301 may also be an external storage device of the terminal device 300, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card) or the like, which are provided on the terminal device 300. Further, the memory 301 may also include both an internal storage unit and an external storage device of the terminal device 300. The memory 301 is used for storing the computer program and other programs and data required by the terminal device. The memory 301 may also be used to temporarily store data that has been output or is to be output.
The processor 302 may be a central processing unit (Central Processing Unit, CPU), but may also be other general purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), field programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions. The functional units and modules in the embodiment may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit, where the integrated units may be implemented in a form of hardware or a form of a software functional unit. In addition, specific names of the functional units and modules are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present application. The specific working process of the units and modules in the above system may refer to the corresponding process in the foregoing method embodiment, which is not described herein again.
In the foregoing embodiments, the descriptions of the embodiments are emphasized, and in part, not described or illustrated in any particular embodiment, reference is made to the related descriptions of other embodiments.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
In the embodiments provided in the present invention, it should be understood that the disclosed apparatus/terminal device and method may be implemented in other manners. For example, the apparatus/terminal device embodiments described above are merely illustrative, e.g., the division of the modules or units is merely a logical function division, and there may be additional divisions in actual implementation, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection via interfaces, devices or units, which may be in electrical, mechanical or other forms.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated modules, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the present invention may implement all or part of the flow of the method of the above embodiment, or may be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the computer program may implement the steps of each of the method embodiments described above. Wherein the computer program comprises computer program code which may be in source code form, object code form, executable file or some intermediate form etc. The computer readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer Memory, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), an electrical carrier signal, a telecommunications signal, a software distribution medium, and so forth. It should be noted that the computer readable medium contains content that can be appropriately scaled according to the requirements of jurisdictions in which such content is subject to legislation and patent practice, such as in certain jurisdictions in which such content is subject to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.
The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention.
Claims (9)
1. A method for conserving energy in a data center network, the method comprising:
acquiring energy consumption factor data of a data center network;
performing energy-saving decision processing according to the energy consumption factor data, and outputting energy-saving optimization adjustment actions; the step of carrying out energy-saving decision processing according to the energy consumption factor data and outputting energy-saving optimization adjustment action comprises the following steps:
generating initial adjustment actions according to the energy consumption factor data;
inputting the energy consumption factor data and the initial adjustment action into an evaluation neural network to drive the evaluation neural network to output a first evaluation value;
Adjusting the energy consumption node according to the initial adjustment action;
acquiring operation data and state data generated by the operation of the energy consumption node in the environment, and calculating rewarding parameters according to the operation data; wherein, the formula for calculating the rewarding parameter according to the operation data is as follows:
wherein R is t For the bonus parameters, L t To complete time, T t For the task state, T c Caching for edge server, A t For video block analysis, L α For accuracy, S t Caching space for remaining edge servers, remaininS t Alpha, beta and gamma are the residual cache space in the edge server, and respectively represent the weights of the three parts;
selecting an initial adjustment action with a positive reward parameter as a sample adjustment action;
the sample adjusting action and corresponding state data meeting the loss function convergence condition of adding constraint conditions are taken as samples to be put into an experience replay pool, the constraint conditions are obtained based on the operation data, and the loss function is used for training the evaluation neural network;
randomly sampling in the experience replay pool to generate random samples;
respectively inputting the random samples into the evaluation neural network and the target neural network to drive the evaluation neural network to generate a first evaluation value and drive the target neural network to generate a second evaluation value;
Performing gradient ascent method evaluation on the first evaluation value and the second evaluation value respectively to generate a first gradient evaluation result;
when the first gradient evaluation result is that a local maximum value is obtained, taking the sample adjusting action corresponding to the local maximum value as the energy-saving optimizing adjusting action;
and adjusting the energy consumption nodes of the data center network according to the energy saving optimization adjustment action.
2. The energy conservation method according to claim 1, characterized in that the energy conservation method further comprises:
after the sample adjusting action corresponding to the local maximum value is used as the energy-saving optimizing adjusting action, respectively carrying out gradient descent method evaluation on the first evaluation value and the second evaluation value to generate a second gradient evaluation result;
and updating the network parameters of the evaluation neural network according to the second gradient evaluation result, and carrying out soft update on the network parameters of the target neural network.
3. The energy saving method according to claim 1, wherein the step of performing energy saving decision processing according to the energy consumption factor data and outputting an energy saving optimization adjustment action includes:
and inputting the energy consumption factor data into a convolutional neural network to drive the convolutional neural network to output energy-saving optimization adjustment actions.
4. The energy conservation method of claim 3, wherein the energy conservation method further comprises:
performing simulation adjustment according to the energy-saving optimization adjustment action to obtain simulation operation data and simulation state data;
updating the data source of the energy consumption factor data according to the simulation operation data and the simulation state data;
calculating the overall power consumption and corresponding rewarding value of all energy consumption nodes of the data center network according to the updated energy consumption factor data;
and when the overall power consumption is reduced and/or the reward value is positive, outputting the updated energy consumption factor data to the convolutional neural network so as to drive the convolutional neural network to carry out the next energy saving optimization scheme decision.
5. The energy saving method according to claim 1, wherein the step of performing energy saving decision processing according to the energy consumption factor data and outputting an energy saving optimization adjustment action includes:
inputting the energy consumption factor data into a machine learning model to drive the machine learning model to output a predicted energy efficiency value;
and regulating the load of the cooling unit according to the predicted energy efficiency value and the cooling demand.
6. The energy conservation method of claim 5, further comprising:
Converting the energy consumption factor data into historical data and storing the historical data;
inputting the historical data into the machine learning model to drive the machine learning model to output a predicted energy efficiency value;
calculating an actual energy efficiency value according to the historical data;
and adjusting parameters of the machine learning model according to the predicted energy efficiency value and the actual energy efficiency value.
7. An energy saving system for a data center network, the energy saving system comprising:
the acquisition module is used for acquiring the energy consumption factor data of the data center network;
the energy-saving decision module is used for carrying out energy-saving decision processing according to the energy consumption factor data and outputting energy-saving optimization adjustment actions; the energy-saving decision module comprises:
the initial action unit is used for generating initial adjustment action according to the energy consumption factor data;
the evaluation one-value generation unit is used for inputting the energy consumption factor data and the initial adjustment action into an evaluation neural network so as to drive the evaluation neural network to output a first evaluation value;
the initial adjusting unit is used for adjusting the energy consumption node according to the initial adjusting action;
the environment parameter acquisition unit is used for acquiring operation data and state data generated by the operation of the energy consumption node in the environment and calculating rewarding parameters according to the operation data; wherein, the formula for calculating the rewarding parameter according to the operation data is as follows:
Wherein R is t For the bonus parameters, L t To complete time, T t For the task state, T c Caching for edge server, A t For video block analysis, L α For accuracy, S t Caching space for remaining edge servers, remaininS t Alpha, beta and gamma are the residual cache space in the edge server, and respectively represent the weights of the three parts;
a sample action unit for selecting an initial adjustment action with a positive reward parameter as a sample adjustment action;
the sample pool unit is used for taking the sample adjusting action and corresponding state data which meet the loss function convergence condition of the added constraint condition as samples and putting the samples into the experience replay pool, the constraint condition is obtained based on the operation data, and the loss function is used for training the evaluation neural network;
a sampling unit for randomly sampling in the experience replay pool to generate random samples;
the two-network value calculation unit is used for inputting the random samples into the evaluation neural network and the target neural network respectively so as to drive the evaluation neural network to generate a first evaluation value and drive the target neural network to generate a second evaluation value;
the gradient rising unit is used for respectively carrying out gradient rising method evaluation on the first evaluation value and the second evaluation value to generate a first gradient evaluation result;
The extremum action output unit is used for taking the sample adjusting action corresponding to the local maximum value as the energy-saving optimizing adjusting action when the first gradient evaluation result is that the local maximum value is obtained;
and the node adjusting module is used for adjusting the energy consumption nodes of the data center network according to the energy saving optimization adjusting action.
8. A terminal device, characterized in that it comprises a memory, a processor and a computer program stored in the memory and executable on the processor, which processor, when executing the computer program, implements the energy saving method according to any of claims 1 to 6.
9. A computer readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the energy saving method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011498034.XA CN114650190B (en) | 2020-12-17 | 2020-12-17 | Energy-saving method, system, terminal equipment and storage medium for data center network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011498034.XA CN114650190B (en) | 2020-12-17 | 2020-12-17 | Energy-saving method, system, terminal equipment and storage medium for data center network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114650190A CN114650190A (en) | 2022-06-21 |
| CN114650190B true CN114650190B (en) | 2023-06-09 |
Family
ID=81990604
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011498034.XA Active CN114650190B (en) | 2020-12-17 | 2020-12-17 | Energy-saving method, system, terminal equipment and storage medium for data center network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114650190B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110555057A (en) * | 2019-08-19 | 2019-12-10 | 武汉世纪楚林科技有限公司 | energy-saving big data analysis method and device, terminal equipment and storage medium |
| CN111609534A (en) * | 2020-05-25 | 2020-09-01 | 珠海拓芯科技有限公司 | Temperature control method and device and central temperature control system |
-
2020
- 2020-12-17 CN CN202011498034.XA patent/CN114650190B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110555057A (en) * | 2019-08-19 | 2019-12-10 | 武汉世纪楚林科技有限公司 | energy-saving big data analysis method and device, terminal equipment and storage medium |
| CN111609534A (en) * | 2020-05-25 | 2020-09-01 | 珠海拓芯科技有限公司 | Temperature control method and device and central temperature control system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114650190A (en) | 2022-06-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112633316B (en) | Load prediction method and device based on boundary estimation theory | |
| CN106991504B (en) | Building energy consumption prediction method and system based on subentry measurement time sequence and building | |
| CN114422381A (en) | Communication network flow prediction method, system, storage medium and computer equipment | |
| CN115796393B (en) | Energy management optimization method, system and storage medium based on multi-energy interaction | |
| CN110826784B (en) | Method and device for predicting energy use efficiency, storage medium and terminal equipment | |
| CN104331737A (en) | Office building load prediction method based on particle swarm neural network | |
| CN112329997A (en) | Power demand load prediction method and system, electronic device, and storage medium | |
| CN110162390A (en) | A kind of method for allocating tasks and system of mist computing system | |
| CN111179108A (en) | Method and device for predicting power consumption | |
| CN115238599A (en) | Energy-saving method for refrigerating system and model reinforcement learning training method and device | |
| CN111445065A (en) | Energy consumption optimization method and system for refrigeration group control of data center | |
| Zhang et al. | Forecasting building plug load electricity consumption employing occupant-building interaction input features and bidirectional LSTM with improved swarm intelligent algorithms | |
| Tashiro et al. | Application of convolutional neural network to prediction of temperature distribution in data centers | |
| CN114650190B (en) | Energy-saving method, system, terminal equipment and storage medium for data center network | |
| CN117390550A (en) | Low-carbon park carbon emission dynamic prediction method and system considering emission training set | |
| CN116578134A (en) | Universal base station temperature control method and system based on reinforcement learning | |
| CN114065646B (en) | Energy consumption prediction method based on hybrid optimization algorithm, cloud computing platform and system | |
| Singh et al. | Grey Wolf Optimization Based CNN-LSTM Network for the Prediction of Energy Consumption in Smart Home Environment | |
| Sun et al. | Aledar: An attentions-based encoder-decoder and autoregressive model for workload forecasting of cloud data center | |
| CN115130788A (en) | Big data analysis-based power utilization prediction system | |
| Pérez-Vilarelle et al. | Modeling and simulation of wind energy production in the smart-grid scenario | |
| CN111626485A (en) | Load prediction system and method for regional building energy system | |
| CN111046321A (en) | Photovoltaic power station operation and maintenance strategy optimization method and device | |
| CN110807599A (en) | Method, device, server and storage medium for deciding electrochemical energy storage scheme | |
| CN115348179B (en) | Network traffic boundary prediction method, device, equipment and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |