Method and system for data center it equipment energy consumption and energy cost minimization regulation
By constructing a multi-energy flow collaborative optimization model, the problem of coordinated scheduling of power, heat and computing power among multiple data centers was solved, which reduced the total energy cost and made efficient use of resources, supporting the low-carbon operation of data center clusters.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NORTH CHINA ELECTRIC POWER UNIV
- Filing Date
- 2026-05-08
- Publication Date
- 2026-07-14
AI Technical Summary
Existing technologies struggle to achieve coordinated optimization of power, heat, and computing power among multiple data centers. The lack of a unified scheduling framework leads to a simplification of the physical coupling relationship between energy flows, making it difficult to reflect the true mechanism of power changes. Furthermore, existing solutions only optimize energy or computing power within a local scope, failing to achieve overall power optimization and coordinated resource utilization.
By constructing a multi-energy flow collaborative optimization model for electrical energy, thermal energy, and computing energy, and combining the scheduling costs between data centers and operators, the total scheduling cost is optimized, and joint scheduling of electrical energy, thermal energy, and computing energy is achieved. A resource collaborative optimization framework under a centralized architecture is adopted.
It enables coordinated scheduling of electrical, thermal, and computing power among multiple data centers, reduces total energy costs, supports efficient, low-carbon, and sustainable operation of data center clusters, and achieves energy sharing and minimizes operating costs at the system level.
Smart Images
Figure CN122390367A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data center technology, and more specifically, to a method and system for minimizing and controlling the energy consumption and energy costs of data center IT equipment. Background Technology
[0002] Due to the rapid advancement of the digital economy and the continuous increase in computing power demand, data centers, as critical infrastructure, are experiencing exponential growth in scale and power, leading to frequent peak power loads and escalating operating costs. Simultaneously, the types of tasks within data centers are becoming increasingly diverse, with some exhibiting deferred execution characteristics, providing potential opportunities for power management and scheduling optimization. Most studies focus on scheduling within a single data center, lacking collaborative modeling and unified scheduling mechanisms for computing power and energy resources across multiple centers, making it difficult to achieve cross-regional energy sharing and overall cost optimization. To address power growth and carbon constraint pressures, it is necessary to construct a collaborative optimization model for multiple energy flows (electricity, heat, and computing power) from the perspective of coordinating computing power allocation, power supply and demand, and thermal management, while ensuring Quality of Service (QoS). Existing research largely focuses on improving the economy and operational efficiency of data centers through task allocation or power regulation, but these methods still have shortcomings in system modeling and global coordination. These shortcomings are mainly reflected in the following two aspects: firstly, insufficient model coupling. Existing methods typically treat computing load, power consumption, and cooling power separately, failing to establish a unified model of IT power consumption, cooling systems, and thermodynamic processes. This simplifies the physical coupling relationships between energy flows, making it difficult to accurately reflect the true mechanisms of power changes. Secondly, the optimization hierarchy is fragmented. For example, the existing technology CN117880292A discloses a data center energy optimization method based on power-computing-communication collaborative scheduling. This method addresses the task scheduling problem of multi-geographical data centers, comprehensively considering the temporal logic relationship between data transmission latency and task computation, and achieves a scheduling scheme aimed at minimizing the overall power cost of the data center through optimization. However, this method only focuses on the computing power scheduling capability between data centers, without further addressing the scheduling of thermal and electrical energy. Existing technology CN118966702A discloses a data center task load optimization scheduling method and system based on power-computing-thermal collaboration. This method models the electrical, computing, and thermal energy consumption in each area of the data center, obtains electrical, computing, and thermal energy consumption models, calculates the total power of each area of the data center, and optimizes the data center operating cost by combining real-time electricity prices. However, this method only focuses on the optimal scheduling of a single data center and cannot perform collaborative optimization for data center clusters. In summary, existing solutions only optimize energy or computing power within a local scope, lacking a unified scheduling framework covering multiple data center clusters, making it difficult to achieve overall power optimization and coordinated resource utilization. Summary of the Invention
[0003] The purpose of this invention is to provide a method and system for minimizing the energy consumption and energy cost of data center IT equipment. It optimizes the overall scheduling cost by optimizing the allocation of power, heat and computing power between data centers and operators, and between data centers, through the scheduling costs of power, heat and computing power. This solves the problem that existing technologies do not fully consider the multiple energy flows of power, heat and computing power, and realizes an integrated optimization framework that can coordinate various resources and constraints under a centralized architecture.
[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:
[0005] A method for minimizing and controlling the energy consumption and energy costs of data center IT equipment includes the following steps: Based on the power of the IT equipment in the data center, obtain the power dispatch costs between all data centers and power operators, and between different data centers; based on the power of the chiller units in the data center, obtain the thermal dispatch costs between all data centers and thermal operators, and between different data centers; based on the schedulable computing power tasks mapped by the CPU utilization rate of the supplier, obtain the computing power dispatch costs between all data centers and computing power operators, and between different data centers; based on the power dispatch costs, thermal dispatch costs, and computing power dispatch costs, and combined with the current power, thermal, and computing power demands of the data center, introduce SLA deadlines, IT equipment power, chiller unit power, and CPU utilization constraints, and with the minimum dispatch cost as the optimization objective, obtain a joint dispatch plan for power, thermal, and computing power, and construct a total energy cost model accordingly.
[0006] The power of IT equipment based on data centers is used to obtain the power dispatch cost between all data centers and power operators, as well as between different data centers. Specifically, this includes: the power dispatch unit dispatch cost of the power operator, multiplied by the power of the IT equipment to be dispatched for the task of the data center, to obtain the power dispatch cost between the data center and the power operator; and the power dispatch cost between different data centers is obtained by multiplying the power dispatch unit transmission cost between different data centers by the power of the IT equipment to be dispatched for the task.
[0007] The power of IT equipment that needs to be scheduled for the task between different data centers is represented as the power of at least one rack in the scheduled data center. The maximum power consumption increment is calculated by subtracting the power of the rack when it is idle from the power of the rack when it is fully loaded. The maximum power consumption increment is multiplied by the rack's CPU utilization rate to calculate the dynamic incremental power consumption occupied by the current business load. Then, the power of the rack when it is idle is added to obtain the power of each rack.
[0008] The heat dispatch cost between all data centers and heat operators, as well as between different data centers, is obtained based on the chiller power of the data center. Specifically, this includes multiplying the heat dispatch unit cost of the heat operator by the chiller power required to be dispatched for the data center's task to obtain the heat dispatch cost between the data center and the heat operator; and multiplying the heat dispatch unit transmission cost between different data centers by the chiller power required to be dispatched for the task to obtain the heat dispatch cost between different data centers.
[0009] The schedulable computing power task volume mapped by the CPU utilization of the supplier is used to obtain the computing power scheduling cost between all data centers and computing power operators, as well as between different data centers. Specifically, this includes: multiplying the computing power operator's unit scheduling cost by the computing power scheduling amount mapped by the computing power operator's CPU utilization to obtain the computing power scheduling cost between the data center and the computing power operator; and multiplying the unit transmission cost of computing power scheduling between different data centers by the computing power scheduling amount mapped by the CPU utilization of the supplier's data center to obtain the computing power scheduling cost between different data centers.
[0010] Based on the power dispatch cost, thermal dispatch cost, and computing power dispatch cost, and combined with the current power, thermal, and computing power demands of the data center, constraints such as SLA deadlines, IT equipment power, chiller power, and CPU utilization are introduced. With the minimum dispatch cost as the optimization objective, a joint dispatch plan for power, thermal, and computing power is obtained. Based on this, a total energy cost model is constructed. Specifically, this includes using the power allocation, thermal allocation, and computing power allocation between the power operator, thermal operator, computing power operator, and each data center in the joint dispatch plan as variables. Coupled with the unit dispatch cost, unit dispatch cost, unit dispatch cost, unit transmission cost, unit transmission cost, and unit transmission cost of different energy transmission paths, as well as the loss coefficient corresponding to each energy transmission path, and combined with SLA deadline constraints, a total energy cost model covering power, thermal, and computing power is constructed.
[0011] The SLA period includes the task start time, the latest execution time, and the task execution time.
[0012] A data center IT equipment energy consumption and energy cost minimization control system includes: an electricity dispatch cost calculation module: based on the IT equipment power of the data center, it obtains the electricity dispatch cost between all data centers and electricity operators, as well as between different data centers; a thermal dispatch cost calculation module: based on the chiller power of the data center, it obtains the thermal dispatch cost between all data centers and thermal operators, as well as between different data centers; a computing power dispatch cost calculation module: based on the schedulable computing power tasks mapped by the CPU utilization rate of the supplier, it obtains the computing power dispatch cost between all data centers and computing power operators, as well as between different data centers; and an energy cost model construction module: based on the electricity dispatch cost, the thermal dispatch cost, and the computing power dispatch cost, combined with the current electricity, thermal, and computing power requirements of the data center, it introduces SLA deadlines, IT equipment power, chiller power, and CPU utilization constraints, and obtains a joint dispatch plan for electricity, thermal, and computing power with the minimum dispatch cost as the optimization objective, thereby constructing a total energy cost model.
[0013] An electronic device includes a memory, a processor, and a computer program running on the processor, wherein the processor executes the computer program to implement a method for minimizing and controlling the energy consumption and energy costs of the data center IT equipment.
[0014] A computer-readable storage medium storing a computer program, wherein when executed by a processor, the computer program implements the steps of the method for minimizing and controlling the energy consumption and energy cost of data center IT equipment.
[0015] Compared with the prior art, the present invention has at least the following advantages or beneficial effects:
[0016] This application provides a method for minimizing the energy consumption and energy cost of IT equipment in data centers. This method aims to achieve coordinated scheduling optimization of electrical, thermal, and computing power among different data centers to reduce total energy costs. Specifically, this application obtains the electrical scheduling cost based on the power of IT equipment between all data centers and power operators, and between different data centers; the thermal scheduling cost based on the chiller power between all data centers and thermal operators, and between different data centers; and the computing power scheduling cost based on the CPU utilization between all data centers and computing power operators, and between different data centers. Based on the electrical, thermal, and computing power scheduling costs, and with the minimum scheduling cost as the optimization objective, and according to the current electrical, thermal, and computing power demands of the data center, combined with corresponding SLA deadlines and operational constraints, a joint scheduling plan for electrical, thermal, and computing power for different operators and data centers is obtained. A total energy cost model is constructed, which can effectively reduce the total energy cost of data centers. By adopting a centralized optimization method to coordinate the power, thermal, and computing power resources among multiple data centers, energy sharing and operating cost minimization can be achieved at the system level, supporting the efficient, low-carbon, and sustainable operation of data center clusters. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0018] Figure 1 This is a schematic diagram of the heat flow principle of the rack in the data center IT equipment energy consumption and energy cost minimization control method of this application embodiment. Detailed Implementation
[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0020] Example
[0021] The flowchart of the data center IT equipment energy consumption and energy cost minimization control method provided in this application embodiment includes the following steps: Based on the IT equipment power of the data center, obtain the power dispatch cost between all data centers and power operators, and between different data centers; Based on the chiller power of the data center, obtain the thermal dispatch cost between all data centers and thermal operators, and between different data centers; Based on the schedulable computing power task volume mapped by the CPU utilization of the supplier, obtain the computing power dispatch cost between all data centers and computing power operators, and between different data centers; Based on the power dispatch cost, the thermal dispatch cost, and the computing power dispatch cost, combined with the current power, thermal, and computing power requirements of the data center, introduce SLA deadlines, IT equipment power, chiller power, and CPU utilization constraints, and with the minimum dispatch cost as the optimization objective, obtain a joint dispatch plan for power, thermal, and computing power, and construct a total energy cost model accordingly.
[0022] Real-time power demands (e.g., 120kW, 100kW, 150kW), thermal demands (e.g., 90kW, 80kW, 110kW), and computing power demands (e.g., 70kW, 60kW, 80kW computing power loads, corresponding to CPU utilization rates of 70%, 60%, and 80%, respectively, with CPU utilization not exceeding the 80% limit during scheduling) are collected from three data centers. This joint scheduling aims to minimize total scheduling cost. It explicitly introduces scheduling cost parameters (energy scheduling / transmission costs between operators and data centers) and real-time demand parameters (real-time power, thermal, and computing power demands of data centers). A complete objective function is first established, and then, combined with constraints (time constraints for each energy scheduling function), the solution is iteratively obtained to finally arrive at the optimal joint scheduling plan. The specific optimization process and explanation are as follows:
[0023] With the goal of minimizing total scheduling cost, the scheduling plan uses the allocation of each energy source (i.e., the electrical / thermal / computing energy obtained by the data center from each energy supplier) as the scheduling plan. The scheduling unit cost and / or transmission unit cost of electrical energy, thermal energy, and computing energy for different energy transmission paths in this plan are used as decision variables. These are coupled with corresponding energy transmission path loss coefficients (e.g., electrical energy transmission loss coefficient 0.05, thermal energy transmission loss coefficient 0.03) to construct a total energy cost model. The total scheduling cost encompasses both energy procurement cost and the additional cost corresponding to transmission losses, specifically expressed as: Total Energy Cost = Σ(Energy Procurement Quantity from Each Supplier × Loss Coefficient × Corresponding Unit Price) + Σ(Energy Transmission Quantity from Each Supplier × Loss Coefficient × Corresponding Unit Price), achieving the core objective of cost minimization. The time constraints for each energy dispatch include the Service Level Agreement (SLA) deadlines for the corresponding service agreements between power operators, thermal operators, computing power operators, and data centers, as well as equipment operating limit constraints based on task scheduling (IT equipment power, chiller power, and CPU utilization). SLA deadlines require the dispatch scheme to complete energy allocation and transmission within the specified service level agreement period to ensure service continuity. IT equipment power constraints, chiller power constraints, and CPU utilization constraints can be the upper limits of data center reception and the carrying capacity limits of energy transmission equipment, avoiding equipment overload or service breaches. Minimizing the total dispatch cost is used as the objective function. The decision variables can be encoded into chromosomes that can be processed by a genetic algorithm, with each chromosome corresponding to a complete joint dispatch scheme. Several initial dispatch schemes (initial population) that meet the time constraints (SLA deadlines and equipment operating limit constraints) are randomly generated. The total energy cost corresponding to the current optimal dispatch plan is directly calculated, ultimately achieving the core objective of minimizing the total dispatch cost.
[0024] The power of IT equipment based on data centers is used to obtain the power dispatch cost between all data centers and power operators, as well as between different data centers. Specifically, this includes: the power dispatch unit dispatch cost of the power operator, multiplied by the power of the IT equipment to be dispatched for the task of the data center, to obtain the power dispatch cost between the data center and the power operator; and the power dispatch cost between different data centers is obtained by multiplying the power dispatch unit transmission cost between different data centers by the power of the IT equipment to be dispatched for the task.
[0025] The heat dispatch cost between all data centers and heat operators, as well as between different data centers, is obtained based on the chiller power of the data center. Specifically, this includes multiplying the heat dispatch unit cost of the heat operator by the chiller power required to be dispatched for the data center's task to obtain the heat dispatch cost between the data center and the heat operator; and multiplying the heat dispatch unit transmission cost between different data centers by the chiller power required to be dispatched for the task to obtain the heat dispatch cost between different data centers.
[0026] The schedulable computing power task volume mapped by the CPU utilization of the supplier is used to obtain the computing power scheduling cost between all data centers and computing power operators, as well as between different data centers. Specifically, this includes: multiplying the computing power operator's unit scheduling cost by the computing power scheduling amount mapped by the computing power operator's CPU utilization to obtain the computing power scheduling cost between the data center and the computing power operator; and multiplying the unit transmission cost of computing power scheduling between different data centers by the computing power scheduling amount mapped by the CPU utilization of the supplier's data center to obtain the computing power scheduling cost between different data centers.
[0027] The mapping relationship between CPU utilization and computing power scheduling is known and can be calculated similarly to the power calculation of IT equipment. When the supplier is a computing power operator or data center, the following mapping formula for computing power scheduling based on CPU utilization is used: Scheduling capacity = Supplier's rack computing power idle load + Supplier's real-time CPU utilization * (Supplier's rack computing power full load - Supplier's rack computing power idle load).
[0028] CPU utilization in a data center refers to the weighted average of the CPU utilization of all business racks within a single data center. It is aggregated with the computing capacity of each rack as the weight and represents the overall load level of the computing resources of the entire data center.
[0029] Based on the power dispatch cost, thermal dispatch cost, and computing power dispatch cost, and combined with the current power, thermal, and computing power demands of the data center, constraints such as SLA deadlines, IT equipment power, chiller power, and CPU utilization are introduced. With the minimum dispatch cost as the optimization objective, a joint dispatch plan for power, thermal, and computing power is obtained. Based on this, a total energy cost model is constructed. Specifically, this includes using the power allocation, thermal allocation, and computing power allocation between the power operator, thermal operator, computing power operator, and each data center in the joint dispatch plan as variables. Coupled with the unit dispatch cost, unit dispatch cost, unit dispatch cost, unit transmission cost, unit transmission cost, and unit transmission cost of different energy transmission paths, as well as the loss coefficient corresponding to each energy transmission path, and combined with SLA deadline constraints, a total energy cost model covering power, thermal, and computing power is constructed.
[0030] The SLA period includes the task start time, the latest execution time, and the task execution time.
[0031] The power of IT equipment that needs to be scheduled for the task between different data centers is represented as the power of at least one rack in the scheduled data center. The maximum power consumption increment is calculated by subtracting the power of the rack when it is idle from the power of the rack when it is fully loaded. The maximum power consumption increment is multiplied by the rack's CPU utilization rate to calculate the dynamic incremental power consumption occupied by the current business load. Then, the power of the rack when it is idle is added to obtain the power of each rack.
[0032] Rack CPU utilization refers to the instantaneous average load percentage of all CPU processors within a single server rack, representing the degree to which current business tasks consume server computing resources. When rack CPU utilization is 0%, the corresponding rack has no business task load and is only maintaining idle hardware operation; when rack CPU utilization is 100%, the corresponding rack's computing resources are operating at full capacity, reaching the limit of business capacity. Rack CPU utilization determines the dynamic power consumption of a single rack, and thus determines the power of schedulable IT equipment.
[0033] Rack CPU utilization is the percentage of CPU resources occupied within a single server rack, while data center CPU utilization is a weighted aggregate of all rack CPU utilization rates. The CPU utilization rate requiring scheduling is the percentage of surplus CPU load that can be migrated externally after deducting local essential service loads from the data center. This invention first calculates the power consumption of a single rack by combining the difference between full-load and idle power and rack CPU utilization, and then aggregates this power to obtain the schedulable IT equipment power in the data center. Simultaneously, based on the schedulable CPU utilization rate, it maps the computing power-power consumption linear relationship to standardized computing power scheduling quantities, eliminating the direct use of CPU utilization percentages in cost accounting. By multiplying the unit scheduling cost and unit transmission cost of computing power by the corresponding mapped computing power scheduling quantities, the computing power scheduling costs between the data center and computing power operators, and between the data center itself, are calculated sequentially, achieving unified and collaborative optimization modeling of the three types of scheduling costs: electrical energy, thermal energy, and computing power.
[0034] The power consumption of IT equipment in each data center can be expressed as the total power consumption of all racks. The power consumption of each rack = (full load power - idle power) × CPU utilization + idle power. CPU utilization varies between 0 and 100%, and power consumption is linearly related to CPU utilization. IT equipment power consumption is divided into two parts: basic idle power consumption and dynamic load power consumption. Basic idle power consumption refers to the power consumption of basic hardware such as motherboards, memory, hard drives, power supplies, and fans even when the CPU is completely idle and there is no business load. This is a fixed base power consumption and is unrelated to business load. Dynamic load power consumption is the maximum power consumption increment that a rack can increase from idle to full load. Multiplying the maximum power consumption increment by the CPU utilization represents the proportion of incremental power consumption currently consumed by the business load. That is, the actual rack power consumption = fixed idle base power consumption + dynamic incremental power consumption that changes linearly with CPU load.
[0035] Chiller units can employ liquid cooling equipment and chillers. The corresponding chiller unit power includes the power of the liquid cooling equipment's water pumps and the chiller's fan power for the data center. The total electricity sold and purchased between data centers are the same. The heat processed and the heat already processed between data centers are the same.
[0036] The principle of heat dissipation inside a data center is as follows: Figure 1 As shown, cold air from the cold aisle within the server rack cools the hot air inside the rack. After heat exchange with the hot air, the cold air flows through a heat exchanger, which transfers its heat to the chiller unit, thus cooling the air. The cooled air then re-enters the cold aisle for recirculation. Optionally, the server rack can dissipate heat through rear air vents, exchanging heat with the wall and being affected by the outside air temperature.
[0037] In summary, the embodiments of this application provide a method and system for minimizing the energy consumption and energy cost of data center IT equipment: the method aims to achieve coordinated scheduling optimization of electrical energy, thermal energy and computing energy to reduce the total energy cost. Specifically, this application derives the power dispatch cost based on the power of IT equipment between all data centers and power operators, as well as between different data centers; it derives the thermal dispatch cost based on the chiller power between all data centers and thermal operators, as well as between different data centers; and it derives the computing power dispatch cost based on the CPU utilization between all data centers and computing power operators, as well as between different data centers. Based on the power dispatch cost, thermal dispatch cost, and computing power dispatch cost, and with the minimum dispatch cost as the optimization objective, this application, according to the current power, thermal, and computing power requirements of the data centers, and in conjunction with the corresponding SLA deadlines and operational constraints, derives a joint dispatch plan for power, thermal, and computing power of the current data centers for different operators and data centers, and constructs a total energy cost model. This model can effectively reduce the total energy cost of data centers. By adopting a centralized optimization method to coordinate the power, thermal, and computing power resources among multiple data centers, it can achieve energy mutual assistance and minimize operating costs at the system level, supporting the efficient, low-carbon, and sustainable operation of data center clusters.
[0038] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.
Claims
1. A method for minimizing and controlling the energy consumption and energy costs of data center IT equipment, characterized in that, Includes the following steps: Based on the IT equipment power of data centers, the power dispatch costs between all data centers and power operators, as well as between different data centers, are obtained. Based on the chiller power of the data center, the thermal energy dispatch cost between all data centers and thermal energy operators, as well as between different data centers, is obtained. Based on the schedulable computing power tasks mapped by the CPU utilization of the supplier, the computing power scheduling costs between all data centers and computing power operators, as well as between different data centers, are obtained. Based on the power scheduling cost, the thermal scheduling cost, and the computing power scheduling cost, and combined with the current power, thermal, and computing power requirements of the data center, constraints such as SLA deadline, IT equipment power, chiller power, and CPU utilization are introduced. With the minimum scheduling cost as the optimization objective, a joint scheduling plan for power, thermal, and computing power is obtained, and a total energy cost model is constructed accordingly.
2. The method for minimizing and controlling the energy consumption and energy cost of data center IT equipment as described in claim 1, characterized in that, The power consumption of IT equipment based on data centers is used to obtain the power dispatch costs between all data centers and power operators, as well as between different data centers, specifically including: The power dispatch cost between the data center and the power operator is obtained by multiplying the power dispatch cost of the power dispatch unit of the power operator and the power of the IT equipment that the data center needs to dispatch for its tasks. The power scheduling cost between different data centers is obtained by multiplying the unit transmission cost of power scheduling between different data centers by the power of the IT equipment that needs to be scheduled for the task.
3. The method for minimizing and controlling the energy consumption and energy cost of data center IT equipment as described in claim 2, characterized in that, The power of IT equipment that needs to be scheduled for the task between different data centers is represented as the power of at least one rack in the scheduled data center. The maximum power consumption increment is calculated by subtracting the power of the rack when it is idle from the power of the rack when it is fully loaded. The maximum power consumption increment is multiplied by the rack's CPU utilization rate to calculate the dynamic incremental power consumption occupied by the current business load. Then, the power of the rack when it is idle is added to obtain the power of each rack.
4. The method for minimizing and controlling the energy consumption and energy cost of data center IT equipment as described in claim 1, characterized in that, The chiller unit power based on the data center is used to obtain the heat energy dispatch cost between all data centers and heat energy operators, as well as between different data centers, specifically including: The thermal energy dispatch cost between the data center and the thermal energy operator is obtained by multiplying the thermal energy dispatch cost of the thermal energy operator's dispatch unit by the power of the chiller units that the data center needs to dispatch for its tasks. The thermal energy scheduling cost between different data centers is obtained by multiplying the unit transmission cost of thermal energy scheduling between different data centers by the power of the chiller units that need to be scheduled for the task.
5. The method for minimizing and controlling the energy consumption and energy cost of data center IT equipment as described in claim 4, characterized in that, The schedulable computing power tasks mapped based on the CPU utilization of the supplier are used to obtain the computing power scheduling costs between all data centers and computing power operators, as well as between different data centers, specifically including: The computing power scheduling cost between the data center and the computing power operator is obtained by multiplying the unit scheduling cost of computing power scheduling by the computing power operator and the computing power scheduling amount mapped by the CPU utilization of the computing power operator. The computing power scheduling cost between different data centers is obtained by multiplying the unit transmission cost of computing power scheduling between different data centers by the computing power scheduling amount mapped by the CPU utilization of the supplier's data center.
6. The method for minimizing and controlling the energy consumption and energy cost of data center IT equipment as described in claim 5, characterized in that, Based on the power scheduling cost, thermal scheduling cost, and computing power scheduling cost, and considering the current power, thermal, and computing power requirements of the data center, constraints such as SLA deadlines, IT equipment power, chiller power, and CPU utilization are introduced. With minimum scheduling cost as the optimization objective, a joint scheduling plan for power, thermal, and computing power is obtained. Based on this, a total energy cost model is constructed, specifically including: In the joint dispatch plan, the allocation of electricity, heat, and computing power among electricity operators, heat operators, computing power operators, and data centers is used as variables. Coupled with the unit dispatch cost of electricity, heat, computing power, electricity, heat, and computing power for different energy transmission paths, as well as the loss coefficient for each energy transmission path, and combined with the SLA deadline constraints, a total energy cost model covering electricity, heat, and computing power is constructed.
7. The method for minimizing and controlling the energy consumption and energy cost of data center IT equipment as described in claim 1, characterized in that, The SLA period includes the task start time, the latest execution time, and the task execution time.
8. A data center IT equipment energy consumption and energy cost minimization control system, characterized in that, include: Power dispatch cost calculation module: Based on the power of IT equipment in data centers, it calculates the power dispatch costs between all data centers and power operators, as well as between different data centers; Thermal energy dispatch cost calculation module: Based on the chiller power of data centers, it calculates the thermal energy dispatch costs between all data centers and thermal energy operators, as well as between different data centers; Computing power scheduling cost calculation module: Based on the schedulable computing power task volume mapped by the CPU utilization of the supplier, it obtains the computing power scheduling cost between all data centers and computing power operators, as well as between different data centers. Energy cost model construction module: Based on the power scheduling cost, the thermal scheduling cost, and the computing power scheduling cost, combined with the current power, thermal, and computing power requirements of the data center, SLA deadlines, IT equipment power, chiller power, and CPU utilization constraints are introduced. With the minimum scheduling cost as the optimization objective, a joint scheduling plan for power, thermal, and computing power is obtained, and a total energy cost model is constructed accordingly.
9. An electronic device comprising a memory, a processor, and a computer program running on the processor, characterized in that: When the processor executes the computer program, it implements the steps of the data center IT equipment energy consumption and energy cost minimization control method as described in any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the processor, it implements the steps of the data center IT equipment energy consumption and energy cost minimization control method as described in any one of claims 1 to 7.