Intelligent optimization control system for data center energy management

Abstract

The present application relates to the technical field of data center management, and in particular to an intelligent optimization control system for data center energy management, which realizes cross-domain dynamic linkage by constructing energy, network and business three-side collaborative twins, balances energy and computing power resources while ensuring business service level, avoids global loss of local optimization, uses twin pre-play technology, the system can simulate the whole process influence of different optimization strategies in virtual space, through the cooperative fault handling mechanism of business and network side twins, the system constructs a closed loop of "fault early warning-load transfer-energy compensation", greatly reduces the risk of core business interruption, strengthens the system resilience and stable operation ability of optimization control in extreme scenarios, the contribution, reputation and benefit evaluation method proposed converts abstract optimization effect into a comparable numerical system, providing objective and interpretable decision basis for optimization control strategy selection, avoiding subjective bias driven by experience.

Description

Technical Field

[0001] This invention relates to the field of data center management technology, and more specifically, to an intelligent optimization control system for data center energy management. Background Technology

[0002] In the field of data center energy management, traditional management models often focus on the local optimization of a single system, either by regulating energy consumption or solely on managing business operational stability, lacking a systematic consideration of the inherent connections between energy, network, and business systems. This isolated optimization approach often leads to the problem of "local optima at the expense of global imbalance." For example, improving energy efficiency may come at the cost of sacrificing business responsiveness, while ensuring business continuity may result in excessive energy consumption, making it difficult to achieve a synergistic balance among multiple objectives.

[0003] Existing technologies have significant shortcomings in policy verification and optimization iteration. Traditional policy effectiveness evaluation relies heavily on actual trial operation of physical systems, which not only requires a significant investment of time and resources for debugging but also makes it difficult to comprehensively predict the potential impact of policies on the entire data center chain (such as from energy supply to service output). Furthermore, due to the lack of a virtual space pre-simulation mechanism, the stability and adaptability of policies cannot be verified in advance, often resulting in insufficient adaptability in practical applications and affecting overall management efficiency.

[0004] Meanwhile, existing solutions have limitations in risk response and effectiveness evaluation. Faced with complex scenarios such as equipment failures and load fluctuations, they rely heavily on manual intervention for emergency handling, lacking automated collaborative response mechanisms and exhibiting insufficient system resilience. Furthermore, the evaluation of optimization effects lacks unified and objective quantitative standards, relying heavily on experience-based judgments. This leads to subjective biases in strategy selection, making it difficult to meet the dynamic management needs of data centers in different scenarios and hindering the achievement of long-term, efficient operation and maintenance goals.

[0005] Based on the above, this application proposes an intelligent optimization control system for data center energy management. Summary of the Invention

[0006] In view of the shortcomings of existing technologies, the purpose of this invention is to provide an intelligent optimization control system for data center energy management.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] The intelligent optimization control system for data center energy management includes a three-sided twin building module for constructing the energy-side twin, network-side twin, and service-side twin of the data center.

[0009] The twin collaboration module is used to coordinate the energy-side twin, network-side twin, and business-side twin to form a three-in-one collaborative twin.

[0010] The energy optimization control analysis module updates the data involved in the energy-side twin, network-side twin, and service-side twin in real time, and periodically determines the energy optimization control index of various optimization control strategies in the optimization control strategy library;

[0011] The energy optimization control execution module executes the optimization control strategy that maximizes the energy optimization control index value to optimize and control the data center.

[0012] Further, the process for determining the energy optimization control index of the optimized control strategy is as follows: Select an optimized control strategy from the optimized control strategy library, use the selected optimized control strategy to simulate the control of the cooperative twin, divide the entire simulation control into i time periods of equal duration, obtain the three-dimensional feedback value for each time period, sum and average the three-dimensional feedback values ​​for each time period to calculate the average three-dimensional feedback value, sort all the three-dimensional feedback values ​​in chronological order according to the time periods, calculate the absolute difference between two adjacent three-dimensional feedback values ​​after sorting to calculate the three-dimensional feedback fluctuation value, sum and average all the three-dimensional feedback fluctuation values ​​to calculate the average three-dimensional feedback fluctuation value, and calculate the difference between the average three-dimensional feedback value and the average three-dimensional feedback fluctuation value to obtain the energy optimization control index of the optimized control strategy.

[0013] Furthermore, the process for obtaining the three-dimensional feedback value over a time period is as follows: Select a time period, obtain the contribution quantification value, reputation quantification value, and benefit quantification value within that time period, and calculate the three-dimensional feedback value for that time period based on the contribution quantification value, reputation quantification value, and benefit quantification value.

[0014] Furthermore, the contribution metrics are obtained as follows: The standardized score for energy efficiency improvement (E), service reliability (B), and transmission efficiency (T) of the collaborative twin are obtained over a given time period. Calculate the contribution metric ;in, , , All are dynamic weights. .

[0015] Furthermore, the process for obtaining the standardized score E for energy efficiency improvement is as follows: obtain the time period PUE of the co-twin within a certain time period, calculate the difference between the time period PUE and the PUE baseline value to calculate the PUE improvement value, and obtain the standardized score E for energy efficiency improvement corresponding to the PUE improvement value based on the PUE mapping table.

[0016] The process for obtaining the business reliability standardization score B is as follows: Obtain the SLA compliance rate of the collaborative twin within a certain time period, and derive the business reliability standardization score B corresponding to the SLA compliance rate based on the SLA mapping table.

[0017] The process for obtaining the transmission efficiency standardized score T is as follows: Obtain the average transmission efficiency of the cooperating twin within a certain time period, and derive the transmission efficiency standardized score T corresponding to the average transmission efficiency based on the transmission efficiency mapping table.

[0018] Furthermore, reputation quantification Acquisition method: Obtain the standardized score L for low-carbon compliance and the standardized score G for the proportion of green networks of the collaborative twin within a certain time period, through... Calculate the reputation quantification value Where K is the certification multiplier; D is the violation attenuation factor; This is the reputation spillover coefficient.

[0019] Furthermore, the process for obtaining the low-carbon compliance standardization score L is as follows: obtain the carbon emissions of the collaborative twin within a certain time period, and derive the low-carbon compliance standardization score L corresponding to the carbon emissions based on the carbon emission mapping table.

[0020] The process for obtaining the standardized score G of green network proportion is as follows: obtain the green link transmission proportion of the collaborative twin within a certain time period, and obtain the standardized score G of green network proportion corresponding to the green link transmission proportion based on the green network mapping table.

[0021] Furthermore, the quantification of benefits Acquisition method: Obtain the standardized score of energy cost optimization of the co-twin over a time period. Standardized score for improving business revenue and network cost optimization standardization score ,pass Calculate the quantified value of benefits ;in, denoted as the scenario coefficient; s as the weakness score; and d as the fluctuation difference.

[0022] Furthermore, the standardized score for energy cost optimization. The acquisition process is as follows: Obtain the energy cost of the collaborative twin within a certain time period, set a baseline energy cost value, calculate the difference between the baseline energy cost value and the energy cost, calculate the energy cost savings, and obtain the standardized energy cost optimization score corresponding to the energy cost savings based on the energy cost optimization mapping table. ;

[0023] Business revenue improvement standardized score The acquisition process is as follows: Obtain the business revenue of the collaborative twin within a certain time period, set a benchmark value for business revenue, calculate the difference between the business revenue and the benchmark value to determine the business revenue improvement value, and derive the standardized score for the business revenue improvement value corresponding to the business revenue improvement value based on the business revenue improvement mapping table. ;

[0024] Network cost optimization standardized score The acquisition process is as follows: Obtain the network cost of the collaborative twin within a certain time period, set a baseline value for the network cost, calculate the difference between the baseline value and the network cost to obtain the optimized value for the network cost, and derive the standardized score for network cost optimization corresponding to the optimized value based on the network cost optimization mapping table. .

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] The system of this invention achieves cross-domain dynamic linkage by constructing a collaborative twin of energy, network, and services. While ensuring the service level of services, it balances energy and computing resources, avoiding global losses from local optimization. Utilizing twin pre-simulation technology, the system can simulate the full-process impact of different optimization strategies in virtual space. By analyzing three-dimensional feedback values ​​over time periods, it can identify the stability of strategies in advance, significantly reducing offline debugging time. Through the collaborative fault handling mechanism of the service and network twins, the system constructs a closed loop of "fault early warning - load transfer - energy consumption compensation," greatly reducing the risk of core service interruption and enhancing the system's resilience and stable operation capability under extreme scenarios. The proposed contribution, reputation, and benefit assessment methods transform abstract optimization effects into a comparable numerical system, providing an objective and interpretable decision-making basis for the selection of optimization control strategies and avoiding subjective biases driven by experience. Attached Figure Description

[0027] Figure 1 System schematic diagram of an intelligent optimization control system for data center energy management;

[0028] Figure 2 Flowchart for determining the energy optimization control index to optimize the control strategy;

[0029] Figure 3 Flowchart for building an energy-side twin for a data center;

[0030] Figure 4 Flowchart for building a network-side twin for a data center;

[0031] Figure 5 Flowchart for building a business-side twin for the data center. Detailed Implementation

[0032] Reference Figures 1 to 5 The intelligent optimization control system for data center energy management includes a three-sided twin construction module, a twin collaboration module, an energy optimization control analysis module, and an energy optimization control execution module.

[0033] The three-sided twin construction module constructs the energy-side twin, network-side twin, and business-side twin of the data center.

[0034] Data center energy-side twin construction process:

[0035] S1: Define the construction scope; the construction scope includes the spatial scope and the equipment scope.

[0036] Spatial scope: Covers the core energy area of ​​the data center, including: power area; cooling area; IT load area.

[0037] Equipment scope: Screening core energy equipment (accounting for more than 90% of total energy consumption), including: power system: such as 1 transformer, 2 UPS, 1 set of lithium battery energy storage, 50 IT cabinets; refrigeration system: such as 2 water chillers, 10 precision air conditioners, 1 set of cooling tower; auxiliary equipment: such as lighting system, water pump.

[0038] S2: Physical entity modeling; Power system modeling: Draw the power link topology diagram: Generate a digital topology using Graphviz based on the actual connection relationship of mains power inlet → high-voltage switchgear → transformer → low-voltage switchgear → UPS → PDU → rack server; Mark key nodes: such as transformer output → UPS input as the core link, which requires key monitoring of voltage fluctuations; Assign physical attributes to equipment (e.g., transformer: rated capacity 1000kVA, no-load loss 2kW, load loss (5kW at 80% load rate), short-circuit impedance 4%; UPS: rated power 500kVA, efficiency curve (efficiency 94%-96% at 30%-80% load rate), battery pack capacity (200Ah / cell, 32 cells in series), switching time ≤5ms); Mathematical model construction (mathematical models include loss calculation model, UPS efficiency model, and IT load aggregation model).

[0039] Refrigeration system modeling: Drawing the refrigeration link topology: Generating a digital topology based on the actual flow direction of cooling tower → chiller unit → water pump → pipes → precision air conditioner → machine room space → return air duct, and labeling key nodes (such as chiller unit inlet / outlet, air conditioner return air outlet); Assigning physical attributes to equipment (e.g., chiller unit: rated cooling capacity 800kW, COP curve (COP=4.5 at inlet water temperature 12℃, COP=4.2 at inlet water temperature 15℃), compressor power (180kW at full load); precision air conditioner: rated air volume 3000m³ / h, cooling capacity 30kW at return air temperature 25℃, fan power 5kW, humidification capacity 3kg / h); Mathematical model construction (mathematical models include cooling capacity calculation model, temperature field CFD model, and air conditioning energy consumption model).

[0040] Energy storage system modeling: Draw the energy storage link: mains power / PV → energy storage inverter → lithium battery pack → load, and label the connection relationship between the inverter and the battery pack (e.g., 10 battery clusters in parallel, 100 cells in series in each cluster); Assign physical attributes to the equipment (e.g., lithium battery pack: total capacity 1000kWh, single cell voltage 3.2V, SOC (state of charge) 0-100%, cycle life (1500 times at 25℃); energy storage inverter: rated power 500kW, conversion efficiency 96% (during charging / discharging), response time ≤10ms); Mathematical model construction (mathematical models include charge / discharge efficiency model, temperature effect model, and remaining capacity prediction model).

[0041] S3: Use the Unity engine to build a 1:1 virtual energy system, import data center CAD drawings (accurate to the device size), and make the virtual device attributes correspond one-to-one with the physical devices (e.g., virtual transformer rated capacity = physical transformer 1000kVA, virtual air conditioner air volume = physical air conditioner 3000m³ / h), and establish a real-time correlation between physical sensor data and virtual device status.

[0042] Data center network-side twin construction process:

[0043] S1: Define the construction scope; the construction scope includes the spatial scope and the equipment scope.

[0044] Spatial scope: Covers network equipment rooms (switch / router deployment areas), fiber optic link routes (from data center equipment rooms to external fiber optic junction boxes), and 5G base station access points.

[0045] Equipment scope: Core network equipment (accounting for more than 90% of total network energy consumption), including: within the data center: such as 1 core switch, 3 aggregation switches, 20 access switches, and 5 routers; cross-data center links: such as 2 fiber optic links and 1 5G network slice link.

[0046] S2: Physical entity modeling; Network device modeling: Generate a digital topology using Visio or Graphviz based on the actual connection relationships of core switch → aggregation switch → access switch → server network card, router → 5G base station → cross-data center links (nodes are devices, edges are links, and port connection relationships are labeled, such as core switch port 1 → aggregation switch port 1); Assign physical attributes to devices (e.g., access switch: 48 ports, 10Gbps, rated power consumption 200W (idle) - 500W (full load), power consumption increases by 30W for every 10% increase in port utilization (e.g., power consumption is 290W at 30% utilization); Router: supports 5G network slicing, rated power consumption 300W, maximum forwarding rate 10Gbps, slice bandwidth adjustment response time ≤100ms); Mathematical model construction (mathematical models include load-power model and forwarding delay model).

[0047] Transmission link modeling: Draw the link topology: Mark the physical locations using a Geographic Information System (GIS) based on the actual route: Data center egress router → fiber optic link → peer data center router → 5G base station → core network (e.g., the fiber optic link runs from computer room A to computer room B, laid along XX highway, with a length of 10km). Assign link physical attributes (fiber optic link: length 10km, transmission rate 100Gbps, power consumption per GB data transmission 1.8W, maximum traffic 90Gbps (to avoid congestion), temperature impact coefficient (transmission rate decreases by 5% when ambient temperature > 35℃); 5G network slicing: core service slice (priority 1, bandwidth 1Gbps, latency ≤ 10ms, power consumption per GB 3W), non-core slice (priority 3, bandwidth 500Mbps, latency ≤ 100ms, power consumption per GB 2.5W)); Mathematical model construction (mathematical models include traffic-latency model, energy consumption model, and slice bandwidth model).

[0048] S3: Use Unreal Engine to build a 1:1 virtual network scene, import network topology CAD drawings and geographic information, place device models according to actual locations, and ensure that virtual device / link attributes correspond one-to-one with physical entities (e.g., virtual core switch port count = 48 ports, virtual fiber length = 10km), and establish a real-time association between physical data and virtual status.

[0049] Data center business-side twin construction process:

[0050] S1: Define the scope of construction; the scope of construction includes the business scope and the resource scope.

[0051] Business scope: Covering core and non-core businesses carried by the data center (accounting for more than 90% of the total computing power consumption), including: Core businesses: payment system (real-time transactions) and cloud computing services (virtual machine leasing).

[0052] Resource scope: Computing resources supporting business operations, including: 200 physical servers (each with 2 CPUs and 128GB of memory), 1000 containers (Docker / Kubernetes orchestration), and 5 database clusters (MySQL / Redis).

[0053] S2: Draw the business call chain: Reconstruct the call relationship of user request → load balancer → core business → database → cache according to the actual business process. For example: payment business chain: user payment request → NGINX load balancer → payment system server → MySQL database → Redis cache; offline analysis chain: data collection task → Hadoop cluster → Spark computing node → result storage server. Draw the computing resource topology: Reconstruct the resource allocation chain according to the hierarchical relationship of physical server → virtual machine (VM) → container → business process (e.g., physical machine Server-01 → VM-01 → container C-01 → payment system process); entity attribute extraction (entity attributes include business attributes and computing resource attributes. Business attributes such as core business (payment system): type = real-time transaction, concurrency = 10,000-50,000, SLA requirement, resource dependency (requires 20% CPU, 10% memory); non-core business (offline analysis): type = batch processing, data volume = 1 0TB / time, SLA requirements (latency ≤ 2 hours, success rate ≥ 99%), resource dependencies (requires 50% CPU, 30% memory); computing resource attributes: such as physical servers: CPU model, number of cores, memory, idle power consumption, full load power consumption (300W); containers: single container resource quota (2 vCPU, 4GB memory), migration time (≤ 30 seconds), deployment location (bound to physical machine rack number)); mathematical model construction (mathematical models include business load and resource usage relationship model, SLA compliance rate model, migration energy consumption impact model).

[0054] S3: Use Blender+WebGL to build a lightweight 3D scene, import data center CAD drawings and business architecture diagrams, restore them at a 1:1 scale, and establish a one-to-one correspondence between virtual entity attributes and physical business, thus establishing a real-time association between physical data and virtual state.

[0055] The twin collaboration module coordinates the energy-side twin, network-side twin, and business-side twin to form a three-in-one collaborative twin.

[0056] The energy optimization control analysis module updates the data involved in the energy-side twin, network-side twin, and service-side twin in real time, and periodically determines the energy optimization control index of various optimization control strategies in the optimization control strategy library (the optimization control strategy library contains a variety of verified optimization control strategies, and the optimization control strategies in the optimization control strategy library are updated and iterated regularly).

[0057] The energy optimization control execution module executes the optimization control strategy that maximizes the energy optimization control index value to optimize and control the data center.

[0058] The process for determining the energy optimization control index of the optimized control strategy is as follows: Select an optimized control strategy from the optimized control strategy library, use the selected optimized control strategy to simulate the control of the cooperative twin, divide the entire simulation control into i time periods of equal duration, obtain the three-dimensional feedback value for each time period, sum and average the three-dimensional feedback values ​​for each time period to calculate the average three-dimensional feedback value, sort all the three-dimensional feedback values ​​in chronological order according to the time periods, calculate the absolute difference between two adjacent three-dimensional feedback values ​​after sorting to calculate the three-dimensional feedback fluctuation value, sum and average all the three-dimensional feedback fluctuation values ​​to calculate the average three-dimensional feedback fluctuation value, and calculate the difference between the average three-dimensional feedback value and the average three-dimensional feedback fluctuation value to obtain the energy optimization control index of the optimized control strategy.

[0059] The process of obtaining the three-dimensional feedback value over a time period is as follows: Select a time period and obtain the contribution metric value within that time period. Reputation Quantification Value and the quantification of benefits ,pass Calculate the three-dimensional feedback value for this time period, where u1 is the contribution quantification coefficient, u2 is the reputation quantification coefficient, and u3 is the benefit quantification coefficient. For example, u3=0.5, u1=0.3, u2=0.2 (benefit has the highest weight, with contribution and reputation as secondary).

[0060] Contribution metric Acquisition method: Obtain the standardized score E of energy efficiency improvement, the standardized score B of service reliability, and the standardized score T of transmission efficiency of the collaborative twin within a certain time period, through... Calculate the contribution metric ;in, , , All are dynamic weights. ;example: =0.35: Daily operation requires continuous optimization of energy efficiency (such as reducing PUE), but there is no need to sacrifice service or network stability for extreme energy efficiency; While daily operations are not at their peak, basic stability remains the core priority (such as the continuous availability of payment and data storage), and its importance is on par with energy efficiency. For daily network transmission, the goal is "stability and no congestion", so there is no need to pursue the ultimate utilization rate (such as a link load rate of 60%), and its weight is slightly lower than the former two.

[0061] The process for obtaining the standardized score E for energy efficiency improvement is as follows: Step 1: Obtain the energy data of the collaborative twin within a certain time period (energy data includes: total energy consumption of IT load (kWh, denoted as IT), total energy consumption (kWh, denoted as Total, including cooling, lighting, etc.), and calculate the PUE for the time period = Total / IT; Step 2: Set a PUE baseline value (the PUE baseline value is based on the average PUE of the same season and time period over the past n months without implementing optimization strategies), calculate the difference between the time period PUE and the PUE baseline value, calculate the PUE improvement value, and obtain the standardized score E for energy efficiency improvement corresponding to the PUE improvement value based on the PUE mapping table.

[0062] PUE mapping table:

[0063]

[0064] The process for obtaining the Business Reliability Standardization Score B is as follows: Step 1: Obtain the business data of the collaborative twin within a certain time period (the business data includes: the total number of requests (N), the number of requests with a latency ≤ the threshold (e.g., 10ms) (N_ok)), and calculate the SLA compliance rate = N_ok / N × 100%; Step 2: Obtain the Business Reliability Standardization Score B corresponding to the SLA compliance rate based on the SLA mapping table.

[0065] SLA Mapping Table:

[0066]

[0067] The process for obtaining the transmission efficiency standardized score T is as follows: Obtain the average transmission efficiency of the collaborative twin within a certain time period, and derive the transmission efficiency standardized score T corresponding to the average transmission efficiency based on the transmission efficiency mapping table (the transmission efficiency standardized score T ranges from 0 to 10). The transmission efficiency mapping table will not be shown in detail again.

[0068] Reputation Quantification Acquisition method: Obtain the standardized score L for low-carbon compliance and the standardized score G for the proportion of green networks of the collaborative twin within a certain time period, through... Calculate the reputation quantification value Where K is the certification multiplier (base value 1, increased when key certifications are triggered): basic certifications (such as ISO14001, data compliance certification) → K=1.2; advanced certifications (such as carbon neutrality certification, global privacy protection certification) → K=1.5; no core certification → K=1; D is the violation attenuation factor (positive value 1, negative correction): no violation → D=1; 1 minor violation (such as a warning) → D=0.9; 2 or more serious violations (such as penalties) → D=0.7 (simplified calculation for multiple violation scenarios); Reputation spillover coefficient (synergistic reward, reflecting "extra points for overall excellence"): If both L and G are ≥ 8 points (excellent and balanced) → Otherwise → .

[0069] The process for obtaining the low-carbon compliance standardization score L is as follows: Obtain the carbon emissions of the collaborative twin within a certain time period, and derive the low-carbon compliance standardization score L corresponding to the carbon emissions based on the carbon emission mapping table (the low-carbon compliance standardization score L ranges from 0 to 10). The carbon emission mapping table will not be shown in detail again.

[0070] The process for obtaining the standardized score G of green network share is as follows: Obtain the green link transmission share (%, such as the transmission share of fiber optic links / energy-saving 5G slices) of the collaborative twin within a certain time period. Based on the green network mapping table, obtain the standardized score G of green network share corresponding to the green link transmission share (the standardized score G of green network share is between 0 and 10). The green network mapping table will not be shown in detail again.

[0071] Quantitative value of benefits Acquisition method: Obtain the standardized score of energy cost optimization of the co-twin over a time period. Standardized score for improving business revenue and network cost optimization standardization score ,pass Calculate the quantified value of benefits ;in, The scenario coefficient (0.3-0.7) reflects the priority of energy costs, such as the cost control period. Business expansion period ); s is the score for the weakest link (s=min( , , ), the minimum value among the three indicators, 0-10 points); d is the fluctuation difference (d=max( , , )-min( , , (This measures the fluctuation range of the indicator, from 0 to 10 points).

[0072] Energy cost optimization standardization score The acquisition process is as follows: Obtain the energy cost of the collaborative twin within a certain time period, set a benchmark energy cost (the benchmark energy cost is based on the total energy cost of the same historical period), calculate the difference between the benchmark energy cost and the energy cost, calculate the energy cost savings, and obtain the standardized energy cost optimization score corresponding to the energy cost savings based on the energy cost optimization mapping table. (Standardization score for energy cost optimization) (The range is between 0 and 10 points), and the energy cost optimization mapping table will not be shown in detail again.

[0073] Business revenue improvement standardized score The acquisition process is as follows: Obtain the business revenue (such as cloud computing service fees) of the collaborative twin within a certain time period; set a benchmark value for business revenue (the benchmark value is based on the business revenue of the same period in history); calculate the difference between the business revenue and the benchmark value to obtain the business revenue improvement value; and derive the standardized score of the business revenue improvement value corresponding to the business revenue improvement value based on the business revenue improvement mapping table. (Standardized score for improving business revenue) (Scores range from 0 to 10), the business revenue improvement mapping table will not be shown in detail again; network cost optimization standardized score. The acquisition process is as follows: Obtain the network cost of the collaborative twin within a certain time period (cost per unit transmission volume, RMB / GB, total network cost / total transmission volume); set a baseline value for the network cost (the baseline value is based on the network cost of the same historical period); calculate the difference between the baseline value and the network cost to obtain the optimized network cost value; and derive the standardized score for network cost optimization corresponding to the optimized network cost value based on the network cost optimization mapping table. (Network cost optimization standardization score) (The score ranges from 0 to 10), and the network cost optimization mapping table will not be shown in detail again.

[0074] The aforementioned system achieves cross-domain dynamic linkage by constructing a collaborative twin of energy, network, and services. While ensuring service levels, it balances energy and computing resources, avoiding global losses from local optimization. Utilizing twin pre-simulation technology, the system can simulate the full-process impact of different optimization strategies in virtual space. By analyzing three-dimensional feedback values ​​over time periods, it can identify strategy stability in advance, significantly reducing offline debugging time. Through a collaborative fault handling mechanism between the service and network twins, the system constructs a closed loop of "fault warning - load transfer - energy consumption compensation," greatly reducing the risk of core service interruptions and enhancing the system's resilience and stable operation under extreme scenarios. The proposed contribution, reputation, and benefit assessment methods transform abstract optimization effects into a comparable numerical system, providing objective and interpretable decision-making basis for the selection of optimization control strategies and avoiding subjective biases driven by experience.

[0075] The above formulas are all dimensionless calculations, and the preset parameters in the formulas should be set by those skilled in the art according to the actual situation.

[0076] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0077] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0078] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0079] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0080] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0081] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0082] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An intelligent optimization control system for data center energy management, characterized in that, It includes a three-sided twin building module for constructing energy-side twins, network-side twins, and service-side twins for data centers; The twin collaboration module is used to coordinate the energy-side twin, network-side twin, and business-side twin to form a three-in-one collaborative twin. The energy optimization control analysis module updates the data involved in the energy-side twin, network-side twin, and service-side twin in real time. The data involved in the service-side twin includes at least the total energy consumption of IT load, total energy consumption, time period PUE, energy cost, service SLA compliance rate, service revenue, average transmission efficiency, green link transmission ratio, and network cost. It also periodically determines the energy optimization control index of various optimization control strategies in the optimization control strategy library. The process for determining the energy optimization control index of the optimized control strategy is as follows: Select an optimized control strategy from the optimized control strategy library, use the selected optimized control strategy to simulate the control of the cooperative twin, divide the entire simulation control into i time periods of equal duration, obtain the three-dimensional feedback value for each time period, sum and average the three-dimensional feedback values ​​for each time period to calculate the average three-dimensional feedback value, sort all the three-dimensional feedback values ​​in chronological order according to the time periods, calculate the absolute difference between two adjacent three-dimensional feedback values ​​after sorting to calculate the three-dimensional feedback fluctuation value, sum and average all the three-dimensional feedback fluctuation values ​​to calculate the average three-dimensional feedback fluctuation value, and calculate the difference between the average three-dimensional feedback value and the average three-dimensional feedback fluctuation value to obtain the energy optimization control index of the optimized control strategy. The process of obtaining the three-dimensional feedback value for a time period is as follows: Select a time period, obtain the contribution quantification value, reputation quantification value, and benefit quantification value within that time period, and calculate the three-dimensional feedback value for that time period based on the contribution quantification value, reputation quantification value, and benefit quantification value. Method for obtaining contribution metrics: Obtain the standardized score E for energy efficiency improvement, the standardized score B for service reliability, and the standardized score T for transmission efficiency of the collaborative twin over a time period. Calculate the contribution metric ;in, , , All are dynamic weights. ; Reputation Quantification Acquisition method: Obtain the standardized score L for low-carbon compliance and the standardized score G for the proportion of green networks of the collaborative twin within a certain time period, through... Calculate the reputation quantification value Where K is the certification multiplier; D is the violation attenuation factor; Reputation spillover coefficient; Quantitative value of benefits Acquisition method: Obtain the standardized score of energy cost optimization of the co-twin over a time period. Standardized score for improving business revenue and network cost optimization standardization score ,pass Calculate the quantified value of benefits ;in, s is the scenario coefficient; s is the weakness score; d is the fluctuation difference; s=min( , , ); d=max( , , )-min( , , ); The energy optimization control execution module executes the optimization control strategy that maximizes the energy optimization control index value to optimize and control the data center.

2. The intelligent optimized control system for data center energy management of claim 1, wherein, The process of obtaining the standardized score E for energy efficiency improvement is as follows: Obtain the time period PUE of the co-twin within a time period, calculate the difference between the time period PUE and the PUE baseline value to calculate the PUE improvement value, and obtain the standardized score E for energy efficiency improvement corresponding to the PUE improvement value based on the PUE mapping table. The process for obtaining the business reliability standardization score B is as follows: Obtain the SLA compliance rate of the collaborative twin within a certain time period, and derive the business reliability standardization score B corresponding to the SLA compliance rate based on the SLA mapping table. The process for obtaining the transmission efficiency standardized score T is as follows: Obtain the average transmission efficiency of the cooperating twin within a certain time period, and derive the transmission efficiency standardized score T corresponding to the average transmission efficiency based on the transmission efficiency mapping table.

3. The intelligent optimized control system of data center energy management of claim 1, wherein, The process for obtaining the low-carbon compliance standardization score L is as follows: obtain the carbon emissions of the collaborative twin within a certain time period, and derive the low-carbon compliance standardization score L corresponding to the carbon emissions based on the carbon emission mapping table. The obtaining process of the green network proportion standardization score G is as follows: obtaining the proportion of the green link transmission of the cooperative twin in a time period, and obtaining the green network proportion standardization score G corresponding to the proportion of the green link transmission according to the green network mapping table.

4. The intelligent optimized control system of data center energy management of claim 1, wherein, Energy cost optimization standardization score The acquisition process is as follows: Obtain the energy cost of the collaborative twin within a certain time period, set a baseline energy cost value, calculate the difference between the baseline energy cost value and the energy cost, calculate the energy cost savings, and obtain the standardized energy cost optimization score corresponding to the energy cost savings based on the energy cost optimization mapping table. ; Business revenue improvement standardized score The acquisition process is as follows: Obtain the business revenue of the collaborative twin within a certain time period, set a benchmark value for business revenue, calculate the difference between the business revenue and the benchmark value to determine the business revenue improvement value, and derive the standardized score for the business revenue improvement value corresponding to the business revenue improvement value based on the business revenue improvement mapping table. ; Network cost optimization standardized score The acquisition process is as follows: Obtain the network cost of the collaborative twin within a certain time period, set a baseline value for the network cost, calculate the difference between the baseline value and the network cost to obtain the optimized value for the network cost, and derive the standardized score for network cost optimization corresponding to the optimized value based on the network cost optimization mapping table. .

Images