CN-121984199-A - Data center intelligent power supply method and system based on real-time load dynamic adjustment

CN121984199ACN 121984199 ACN121984199 ACN 121984199ACN-121984199-A

Abstract

The application relates to the technical field of power supply control, and discloses an intelligent power supply method and system for a data center based on real-time load dynamic adjustment. And carrying out association processing, and determining a power supply topological relation and a load corresponding relation. And obtaining a load change trend, and determining the current minimum power supply capacity by combining the power supply topological relation, the load corresponding relation, the power supply state data and the energy storage state data to generate a candidate adjustment action set. And carrying out risk judgment on the operation data. When the load change trend meets the pre-wake-up triggering condition or the power supply risk state meets the adjustment triggering condition, performing executable verification on the candidate adjustment action set, and generating a power supply linkage adjustment strategy. And implementing the power supply linkage adjustment strategy in stages according to the preparation execution and the submission execution. According to the application, the linkage adjustment of the energy-saving effect, the power supply reliability and the dynamic response capability of the power supply module is realized under the scene of the data center.

Inventors

LI LI
LI ZHEN
SHEN CHENG
ZHAO JING
KONG XIANGFEI
XU BOWEN

Assignees

润泽科技发展有限公司

Dates

Publication Date: 20260505
Application Date: 20260409

Claims (10)

1. The intelligent power supply method for the data center based on the real-time load dynamic adjustment is characterized by comprising the following steps of: Collecting operation data, wherein the operation data comprises power supply state data, switch state data, energy storage state data and load state data, carrying out association processing on the operation data, and determining a power supply topological relation and a load corresponding relation; Trend processing is carried out according to the load state data to obtain a load change trend, and the current minimum power supply capacity is determined by combining the power supply topological relation, the load corresponding relation, the power supply state data and the energy storage state data to generate a candidate regulation action set, wherein the candidate regulation action set comprises a power supply module dormancy action and a power supply module pre-awakening action; When the load change trend meets a pre-awakening trigger condition or the power supply risk state meets an adjusting trigger condition, carrying out executable verification on the candidate adjusting action set, and generating a power supply linkage adjusting strategy according to a verification result, wherein the executable verification comprises power supply path continuity, key load redundancy, energy storage emergency reservation capacity and action frequency budget; and implementing the power supply linkage adjustment strategy according to the preparation execution and the submitting execution in stages, and pre-waking the power supply module to be put into operation in the preparation execution stage.
2. The data center intelligent power supply method based on real-time load dynamic adjustment according to claim 1, wherein when determining the current minimum power supply capacity, comprising: And determining the load demand in a preset time period according to the load change trend, and superposing the key load reserved capacity, the notch compensation quantity corresponding to the energy storage emergency reserved capacity and the capacity compensation quantity in the pre-awakening preparation period of the power supply module to obtain the current minimum power supply capacity.
3. The method for intelligent power supply to a data center based on real-time load dynamic adjustment according to claim 2, wherein when determining the load demand, critical load reserve capacity, gap compensation and capacity compensation, the method comprises: Distinguishing a critical load from a non-critical load according to the load state data and the load corresponding relation, and carrying out trend processing on the load state data to obtain the load demand in the preset time period; determining the available power supply capacity of a power supply path corresponding to a key load according to the power supply topological relation, the load corresponding relation and the power supply state data, and determining the key load reserved capacity according to key load requirements; determining the current energy storage available capacity and the emergency energy storage reserved capacity according to the energy storage state data and the power supply state data, and determining the gap compensation quantity according to the insufficient part; and determining a new load demand in the pre-wake-up preparation period according to the load change trend, the pre-wake-up preparation time of the power supply module and the power supply state data, and determining the new load demand as the capacity compensation amount in the pre-wake-up preparation period of the power supply module.
4. The intelligent power supply method of the data center based on real-time load dynamic adjustment according to claim 1 is characterized in that risk judgment comprises the steps of respectively carrying out risk judgment of power supply path continuity, key load redundancy, energy storage emergency reservation capability and action frequency budget according to operation data, comparing judgment results with corresponding adjustment triggering conditions and pre-warning conditions, determining a power supply risk state to be an adjustment triggering state when any judgment result meets the corresponding adjustment triggering conditions, determining the power supply risk state to be a pre-warning state when the corresponding adjustment triggering conditions are not met and at least one item meets the corresponding pre-warning conditions, determining the power supply risk state to be a normal state when the corresponding pre-warning conditions are not met, entering a power supply linkage adjustment strategy generation flow when the power supply risk state is not the adjustment triggering state and the load change trend meets the pre-wake triggering conditions, and keeping the current power supply state and continuing to collect the operation data when the power supply risk state is not the adjustment triggering state and the load change trend does not meet the pre-wake triggering conditions.
5. The intelligent power supply method for the data center based on the real-time load dynamic adjustment according to claim 1 is characterized in that when a candidate adjustment action set is generated, power supply modules carrying exclusive power supply paths of key loads are removed according to the power supply topological relation and the load corresponding relation, and power supply modules with current load rates lower than a sleep threshold value and continuous operation time length reaching an operation threshold value are selected from the rest power supply modules to generate power supply module sleep actions, and corresponding power supply module pre-awakening actions are generated.
6. The intelligent power supply method for the data center based on real-time load dynamic adjustment according to claim 5, wherein the pre-wake triggering condition comprises the steps of determining a load demand arrival time according to the load change trend, and triggering a pre-wake action of the power supply module when a time difference between the load demand arrival time and a pre-wake preparation time of the power supply module is smaller than a preset advance and the state continuously reaches a preset duration.
7. The intelligent power supply method for the data center based on the real-time load dynamic adjustment according to claim 1, wherein the preliminary execution stage comprises the steps of pre-waking up a power supply module to be powered on and collecting a preliminary execution result, implementing a commit execution stage when the preliminary execution result meets a stable condition, and carrying out rollback according to the sequence of first canceling a newly added sleep action and then keeping the continuously supplied power of the operated power supply module when the preliminary execution result does not meet the stable condition or the commit execution stage is abnormal.
8. The intelligent power supply method for the data center based on the real-time load dynamic adjustment according to claim 7, wherein when the power supply linkage adjustment strategy comprises a power supply module dormancy action and a power supply module pre-awakening action at the same time, after the executable verification is passed, the power supply module pre-awakening action is implemented in a preliminary execution stage and a preliminary execution result is acquired, the power supply module dormancy action is implemented in a submitting execution stage only when the preliminary execution result meets a stable condition and a critical load redundancy check is kept passed, and when the preliminary execution result does not meet the stable condition or the critical load redundancy check is not passed, the implementation of a newly added dormancy action is forbidden and the running power supply module is kept to supply power continuously.
9. The intelligent power supply method for the data center based on the real-time load dynamic adjustment according to claim 1 is characterized in that the executable verification is performed in the following sequence that the power supply path continuity verification and the key load redundancy verification are performed first, then the energy storage emergency retention capacity verification and the action frequency budget verification are performed, and when the power supply linkage adjustment strategy comprises discrete switching actions, the switching permission verification is performed on the power supply states at the two sides of the target switch.
10. A data center intelligent power supply system based on real-time load dynamic adjustment is used for applying the data center intelligent power supply method based on real-time load dynamic adjustment as set forth in any one of claims 1-9, The system comprises an acquisition unit, a load state data acquisition unit, a power supply topology relation acquisition unit and a load state data acquisition unit, wherein the acquisition unit is configured to acquire operation data, and the operation data comprises power supply state data, switch state data, energy storage state data and load state data; The processing unit is configured to perform trend processing according to the load state data to obtain a load change trend, and determine the current minimum power supply capacity by combining the power supply topological relation, the load corresponding relation, the power supply state data and the energy storage state data to generate a candidate regulation action set, wherein the candidate regulation action set comprises a power supply module dormancy action and a power supply module pre-awakening action; When the load change trend meets the pre-awakening triggering condition or the power supply risk state meets the adjusting triggering condition, carrying out executable verification on the candidate adjusting action set, and generating a power supply linkage adjusting strategy according to a verification result, wherein the executable verification comprises power supply path continuity, key load redundancy, energy storage emergency reservation capability and action frequency budget; The execution unit is configured to implement the power supply linkage adjustment strategy in stages according to the preliminary execution and the submitted execution, and pre-awaken the power supply module to be put into operation in the preliminary execution stage.

Description

Data center intelligent power supply method and system based on real-time load dynamic adjustment Technical Field The invention relates to the technical field of power supply control, in particular to an intelligent power supply method and system for a data center based on real-time load dynamic adjustment. Background With the increasing types of computing tasks carried by the data center, the service load presents obvious volatility and burstiness, so that the power supply system not only needs to meet the requirements of continuous power supply and quick response of the key load, but also needs to avoid the reduction of conversion efficiency and the waste of energy consumption caused by the fact that the number of on-line modules is kept too high for a long time under a low load. Therefore, the demand of the industry for intelligent power supply technology is controlled from simple module start-stop, the real-time linkage adjustment of the data center power supply and distribution scene is gradually turned to, namely, the power supply capacity is dynamically determined based on operation data, and the actions of dormancy, pre-awakening and adjustment are implemented on the premise of ensuring the power supply safety constraint, so that the energy efficiency, the response speed and the power supply reliability are considered. In the existing scheme, for example, the invention patent CN 111103962A-is used for dormancy, dormancy determining method and device, dormancy equipment determines according to power information and load information and power supply enters a dormancy mode, and exits dormancy when load traffic changes, so that the problem of power supply energy-saving dormancy control is mainly solved. The technical center of gravity of the scheme is mainly triggered by the sleep threshold value and the load change, the corresponding relation between the power supply topological relation and the load cannot be established for the power supply and distribution scene of the data center, whether the continuity of the power supply path and the redundancy of the key load are affected or not is difficult to judge before the sleep or wake-up action is executed, and meanwhile, the execution constraint of the energy storage emergency reservation capacity, the action frequency budget and the like is lack of verification, so that the problem that the energy-saving action can be calculated but not necessarily can be safely executed easily occurs in the scene of the data center. Therefore, it is necessary to design a data center intelligent power supply method and system based on real-time load dynamic adjustment to solve the problems in the prior art. Disclosure of Invention In view of the above, the invention provides a data center intelligent power supply method and system based on real-time load dynamic adjustment, which aims to solve the problems that in the existing data center power supply energy-saving control, the module is triggered to sleep or wake up only according to load change, and constraint verification is lacking, so that energy-saving actions can be calculated but safe landing is difficult. The invention provides a data center intelligent power supply method based on real-time load dynamic adjustment, which comprises the following steps: Collecting operation data, wherein the operation data comprises power supply state data, switch state data, energy storage state data and load state data, carrying out association processing on the operation data, and determining a power supply topological relation and a load corresponding relation; Trend processing is carried out according to the load state data to obtain a load change trend, and the current minimum power supply capacity is determined by combining the power supply topological relation, the load corresponding relation, the power supply state data and the energy storage state data to generate a candidate regulation action set, wherein the candidate regulation action set comprises a power supply module dormancy action and a power supply module pre-awakening action; When the load change trend meets a pre-awakening trigger condition or the power supply risk state meets an adjusting trigger condition, carrying out executable verification on the candidate adjusting action set, and generating a power supply linkage adjusting strategy according to a verification result, wherein the executable verification comprises power supply path continuity, key load redundancy, energy storage emergency reservation capacity and action frequency budget; and implementing the power supply linkage adjustment strategy according to the preparation execution and the submitting execution in stages, and pre-waking the power supply module to be put into operation in the preparation execution stage. Further, when determining the current minimum power supply capacity, the method includes: And determining the load demand in a preset time period according to the load change trend,