Energy management method and device for electrothermal coupling system of cross-domain green data center based on spatiotemporal task migration

Abstract

The present invention discloses a method and device for energy management of a cross-domain green data center electrothermal coupling system based on spatiotemporal task migration. First, the day-ahead forecast data of the cross-domain green data center electrothermal coupling system is obtained, including renewable energy output, electric / thermal load demand, and data center task arrival rate. Second, a cross-domain green data center electrothermal coupling system model is established, including data center servers, cooling systems, heat recovery systems, distributed renewable energy generators, electric / thermal energy storage systems, gas boilers, and electric / thermal load demand. Third, based on the day-ahead forecast data and the system model, a strategy based on spatiotemporal task migration is adopted to schedule the energy of the cross-domain green data center electrothermal coupling system to obtain a scheduling plan. Finally, the scheduling plan is executed, and the results are evaluated from multiple perspectives, including economic, environmental, and power supply reliability.

Description

Technical Field

[0001] The present invention relates to the field of electric energy, and in particular to an energy management method and device for an electric-thermal coupling system of a cross-domain green data center based on spatiotemporal task migration. Background Art

[0002] Data centers are currently considered a critical piece of information infrastructure in modern society. With the rapid development of compute-intensive technologies such as artificial intelligence (AI) and cloud computing, data centers are becoming increasingly large, leading to significant power consumption. Since current power generation is primarily driven by carbon-intensive fossil fuels such as natural gas and coal, the massive power consumption of data centers can directly lead to significant carbon emissions. Therefore, energy conservation and emission reduction in data centers have become increasingly urgent and are receiving increasing attention.

[0003] With the global growth of photovoltaic (PV) and wind power in recent years, cloud service providers are expected to adopt renewable energy to power their data centers. A data center powered entirely by wind turbines (WT) in the United States has been reported, and Facebook claims that its global data centers will be 100% powered by renewable energy within a few years. Furthermore, computing tasks can be migrated in both time and space to manage workloads within data centers, thereby improving the efficiency of renewable energy utilization. Furthermore, much of the energy consumed by data centers can be reused to supply regional heating loads. Therefore, approximate energy management based on flexible task migration and waste heat recovery within data centers can improve energy efficiency and reduce carbon emissions. Summary of the Invention

[0004] The purpose of the present invention is to address the deficiencies of the existing technology and propose a method and device for energy management of an electrothermal coupling system of a cross-domain green data center based on spatiotemporal task migration. The rapid development of computing-intensive infrastructure has significantly increased the number and scale of data centers, resulting in a substantial increase in energy consumption and carbon emissions. The present invention designs an energy management solution based on spatiotemporal task migration for geographically distributed green data centers. Taking into account different forms of entities, such as distributed renewable energy generators, direct current, gas boilers, thermal loads, integrated cooling and heat recovery systems, an integrated electrothermal system (IEHS) model of geographically distributed direct current is developed; taking into account electricity prices and the complementarity of renewable energy in different regions, energy management based on spatiotemporal task migration is implemented to improve the utilization efficiency of renewable energy and reduce carbon emissions and operating costs.

[0005] The object of the present invention is achieved through the following technical solutions: In a first aspect, the present invention provides an energy management method for a cross-domain green data center electrothermal coupling system based on spatiotemporal task migration, the method comprising the following steps:

[0006] 1) Obtain day-ahead forecast data for the cross-domain green data center electric and thermal coupling system, including renewable energy output, electric / thermal load demand, and data center task arrival rate;

[0007] 2) Establish a cross-domain green data center electrothermal coupling system model. Data center tasks are divided into delay-sensitive tasks that need to be processed immediately and delay-tolerant tasks that need to be completed before the deadline. If the microgrid where the data center is located does not have sufficient renewable energy, the tasks it receives can be transferred to other data centers with renewable energy or lower electricity prices for processing.

[0008] 3) Based on the day-ahead forecast data and system model, preliminary results are obtained for the scheduling of delay-sensitive tasks. The power of unused renewable energy in each time slot is obtained. A strategy based on spatiotemporal task migration is used to schedule delay-tolerant tasks, and a scheduling plan is obtained and executed.

[0009] Furthermore, step 2) is specifically as follows:

[0010] 2.1) Cross-domain green data center

[0011] Cross-domain green data centers are distributed across different geographic regions and connected via communication networks. Each microgrid is grid-connected, and there is complementarity between different microgrids, allowing tasks to be transferred to other data centers for processing.

[0012] 2.2) Grid-connected green data center model

[0013] A data center is considered a grid-connected microgrid with the data center as the primary load. It consists of photovoltaics, wind turbines, electric / thermal energy storage systems, and different types of electric / thermal loads. The data center contains servers, an integrated cooling and heat recovery system (ICHRS), and some thermal loads. The servers process user tasks and provide cloud computing services, while the ICHRS cools the servers, recovers their waste heat, and supplies it to the thermal loads.

[0014] 2.3) Data Center Energy Consumption Model

[0015] Data center energy consumption Use the following two formulas to calculate:

[0016]

[0017]

[0018] Where i is the index number of the data center, is the server power consumption of the i-th data center at time t, pPUE i,t is the energy efficiency of the ith data center at time t, expressed as the ratio of the total power consumption of the data center to the power consumption of the server; represents the number of servers powered on in the i-th data center at time t, P idle is the no-load power consumption of a single server, P peak is the full load power consumption of a single server, μ i,t is the average CPU utilization of the servers in the i-th data center at time t, is the number of management nodes in the i-th data center, P i ms is the power of a single management node in the i-th data center; energy efficiency index pPUE i,t It is related to the outdoor air temperature of the data center and is calculated by the following formula:

[0019]

[0020] in, represents the outdoor temperature of the i-th data center at time t, a, b and c are coefficients;

[0021] 2.4) Data Center Task Model

[0022] There are two types of tasks in data centers: a. Delay-sensitive tasks, also known as interactive tasks, which need to be processed immediately upon arrival; b. Delay-tolerant tasks, also known as batch tasks, which can be deferred until a deadline.

[0023] task j It is expressed as the following formula:

[0024] task j ={arr j ,ddl j ,st j ,ft j},j=1,2,,N task

[0025] Among them, j is the index number of the task, arr j Indicates the task arrival time, ddl j Indicates the deadline of the task, st j Indicates the time when the task starts executing, ft j Indicates the completion time of the task, N task Indicates the number of tasks;

[0026] 2.5) Waste heat recovery system model

[0027] The power consumption and waste heat recovery power of ICHRS are calculated by the following two formulas:

[0028]

[0029]

[0030] in, represents the electric power consumed by the ICHRS of the i-th data center at time t, represents the heat power recovered by the ICHRS of the i-th data center at time t, ω hr is the heat recovery coefficient;

[0031] 2.6) Carbon trading model

[0032] The carbon emissions of the microgrid are calculated using the following formula:

[0033]

[0034] in, represents the carbon emission cost of the i-th data center, c is the benchmark carbon price, λ is the reward coefficient, is the carbon quota of the data center, v is the length of the carbon emission interval, is the total carbon emissions of the data center, and ξ is the growth rate of the step-by-step carbon price.

[0035] Furthermore, step 3) is specifically as follows:

[0036] 3.1) Preliminary dispatch of distributed power generation and grid power purchase

[0037] Based on the day-ahead forecast data for photovoltaic power generation, wind power generation, and power / heat load, as well as the number of delay-sensitive tasks arriving at each data center, the Gurobi solver is used to obtain the hourly optimal output power of each power source and the hourly power purchase from the grid, enabling preliminary scheduling.

[0038] 3.2) Consuming local renewable energy in each data center

[0039] Based on the preliminary scheduling results, the power of unused renewable energy in each time slot is obtained. Delay-tolerant tasks are assigned to time slots with abandoned renewable energy, thereby improving renewable energy utilization efficiency. Delay-tolerant tasks in each data center should first consume local renewable energy. If the data center does not abandon renewable energy before the deadline, the task will be migrated according to the spatiotemporal task migration strategy.

[0040] 3.3) Dual-dimensional migration of time and space tasks

[0041] For delay-tolerant tasks that were not assigned in the previous step, they are migrated to other data centers with renewable energy for execution according to the deadline. If other data centers have no remaining renewable energy output before the deadline, the task is migrated to the data center with the lowest electricity price for execution, and the power grid purchases electricity to meet the power consumption required for the task execution.

[0042] Furthermore, preliminary scheduling aims to minimize the total cost within the scheduling cycle. The costs include the operation and maintenance costs of system equipment, fuel costs, grid electricity purchase costs, and carbon emission costs. It also needs to ensure that each task must be completed before the deadline. Optimal scheduling needs to meet QoS constraints, task migration needs to meet communication network bandwidth constraints, and task scheduling needs to meet power balance constraints and equipment operation constraints.

[0043] Furthermore, for the delay-tolerant tasks that are not assigned, first compare arr j +Δt l and ddl j -Δt l The size between arr j Indicates the task arrival time, ddl j represents the deadline of the task, Δt l is the transmission time of the task between different data centers;

[0044] If arr j +Δt l Greater than ddl j -Δt l , it means that there is enough time to migrate the delay-tolerant task to other data centers for execution and return the execution results, so the other data centers are sequentially queried for unused renewable energy until it is found in [arr j +Δt l ,ddl j -Δt l ] period, there are still wind and solar curtailment in other data centers, and then the delay-tolerant task is migrated to the data center with curtailment to execute, and the execution time is set to [arr j +Δt l ,ddl j -Δt l ]The moment when the wind and solar power curtailment is highest during the period;

[0045] If [arr j +Δt l ,ddl j -Δt l ] period, if no other data centers have wind and solar curtailment available, then the delay-tolerant task is migrated to [arr j +Δt l ,ddl j -Δt l ] during the period when the data center with the lowest electricity price has the lowest electricity price, and at the same time, the power purchased from the power grid is increased to meet the power consumption required to execute the delay-tolerant task.

[0046] If arr j +Δt l Less than ddl j -Δt l , then put the delay tolerance task on the local electricity price [arr j +Δt l ,ddl j -Δt l ] will be executed at the time when the electricity price is lowest during the period; and so on, until all tasks are assigned.

[0047] In the second aspect, the present invention also provides an energy management device for an electro-thermal coupling system of a cross-domain green data center based on spatio-temporal task migration, comprising a memory and one or more processors, wherein the memory stores executable code, and when the processor executes the executable code, it implements the energy management method for an electro-thermal coupling system of a cross-domain green data center based on spatio-temporal task migration.

[0048] In a third aspect, the present invention also provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, the energy management method of the cross-domain green data center electrothermal coupling system based on spatiotemporal task migration is implemented.

[0049] In a fourth aspect, the present invention also provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the energy management method of a cross-domain green data center electrothermal coupling system based on spatiotemporal task migration.

[0050] Beneficial effects of the present invention:

[0051] This invention coordinates the scheduling of delay-tolerant and delay-sensitive tasks in data centers, along with the output of energy equipment, fully leveraging the spatiotemporal flexibility of data center tasks. This improves the environmental friendliness, economy, low carbon footprint, and reliability of data center energy use. Furthermore, this invention explores the benefits of waste heat recovery in data centers, further contributing to energy conservation and environmental protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0053] Figure 1 This is a schematic diagram of a cross-domain green data center with multiple microgrids;

[0054] Figure 2It is a schematic diagram of a single data center system model;

[0055] Figure 3 This is a diagram of data center task types;

[0056] Figure 4 This is a schematic diagram of the relationship between carbon price and carbon emissions in the ladder carbon trading model;

[0057] Figure 5 This is a flow chart of the energy management method for the electrothermal coupling system of a cross-domain green data center based on spatiotemporal task migration;

[0058] Figure 6 This is a structural diagram of an energy management device for an electrothermal coupling system of a cross-domain green data center based on spatiotemporal task migration according to the present invention. DETAILED DESCRIPTION

[0059] The specific embodiments of the present invention are further described in detail below with reference to the accompanying drawings.

[0060] Reference Figure 1 、 Figure 2 、 Figure 3 、 Figure 4 and Figure 5 The present invention proposes a cross-domain green data center electrothermal coupling system energy management method based on spatiotemporal task migration, which specifically includes the following steps:

[0061] 1) Obtain day-ahead forecast data for the cross-domain green data center electric and thermal coupling system, including renewable energy output, electric / thermal load demand, and data center task arrival rate;

[0062] 2) Establish a cross-domain green data center electrothermal coupling system model

[0063] 2.1) Cross-domain green data center

[0064] Cross-domain green data center model Figure 1 As shown in the figure, cross-domain green data centers are typically distributed across different geographic regions and connected via communication networks. Each microgrid is grid-connected and consists of distributed generators, battery energy storage systems (BESS), heat storage systems (HSS), base power / heat loads, and direct current. Due to differences in natural environment and economic development levels, market electricity prices and renewable energy resources in these geographical microgrids often differ, which means that there is complementarity between different microgrids. When a microgrid where a data center is located does not have sufficient renewable energy, the tasks it receives can be transferred to other data centers with renewable energy or lower electricity prices.

[0065] 2.2) Grid-connected green data center model

[0066] The structure of the electric-thermal coupling system of a single green data center is as follows: Figure 2 As shown in Figure 1, a data center can be considered a grid-connected microgrid with the data center as the primary load. It consists of photovoltaic (PV) equipment, wind turbines, electric / thermal energy storage systems, and various types of electric / thermal loads. Wind turbines, PV equipment, and energy storage devices are connected to the DC bus via power electronics such as rectifiers and inverters. The data center's electrical loads and other loads are connected to the AC bus, while heat sources and loads are connected to the thermal bus. The data center contains servers, an integrated cooling and heat recovery system (ICHRS), and several thermal loads. Servers, as the most critical infrastructure in a data center, process user tasks and provide cloud computing services. Because servers generate significant heat during computing, ICHRSs are used to cool the servers, recover waste heat, and supply it to the thermal loads. While ensuring quality of service (QoS), data centers need to consume as much renewable energy as possible, reduce carbon emissions, and make the electrothermal coupling system more economical and environmentally friendly. To minimize total costs and improve the environmental performance of the electrothermal coupling system, the reliability of the data center's power supply should be considered and a reasonable energy scheduling strategy should be developed.

[0067] 2.3) Data Center Energy Consumption Model

[0068] Data center energy consumption Usually it can be calculated using the following two formulas:

[0069]

[0070]

[0071] Where i is the index number of the data center, is the server power consumption of the i-th data center at time t, pPUE i,t is the energy efficiency of the i-th data center at time t, which is usually expressed as the ratio of the total power consumption of the data center to the power consumption of the server. represents the number of servers powered on in the i-th data center at time t, P idle is the no-load power consumption of a single server, P peak is the full load power consumption of a single server, μ i,t is the average CPU utilization of the servers in the i-th data center at time t, N i ms is the number of management nodes in the i-th data center, P i ms is the power of a single management node in the i-th data center. Energy efficiency index pPUE i,t It is usually related to the outdoor air temperature of the data center and is calculated as follows:

[0072]

[0073] in, represents the outdoor temperature of the i-th data center at time t, and a, b, and c are coefficients.

[0074] 2.4) Data Center Task Model

[0075] The task model of the data center is as follows: Figure 3 There are two types of tasks in data centers: a. Delay-sensitive tasks, also known as interactive tasks, which need to be processed immediately upon arrival; b. Delay-tolerant tasks, also known as batch tasks, which can often be postponed but should be completed before the deadline at the latest.

[0076] A task j It can be expressed as follows:

[0077] task j ={arr j ,ddl j ,st j ,ft j},j=1,2,,N task

[0078] Among them, j is the index number of the task, arr j Indicates the task arrival time, ddl j Indicates the deadline of the task, st j Indicates the time when the task starts executing, ft j Indicates the completion time of the task, N task Represents the number of tasks. The relationship between the attributes of data center tasks can be expressed by the following three formulas:

[0079]

[0080] Δt d =Δt w +Δt l

[0081] ft j =st j +Δt p

[0082] Where, Δt d Denotes the delay time of the delay-tolerant task, Δt w Indicates the waiting time before the task is executed, Δt p represents the processing time of the task, Δt l is the transmission time of the task between different data centers. It is easy to obtain that a delay-tolerant task in [arr j ,ddlj -Δt p ] can be scheduled within the time range.

[0083] Use A i,t To represent the number of tasks received by the i-th data center at time t, use E i,t represents the number of tasks being processed by the i-th data center at time t, then A i,t and E i,t It can be calculated by the following two formulas:

[0084]

[0085]

[0086] in, represents the set of tasks arriving at the i-th data center at time t, represents the set of tasks being processed by the i-th data center at time t. If task j arrives at data center i, then otherwise Similarly, if task j is moved to data center i for execution, then otherwise

[0087] 2.5) Waste heat recovery system model

[0088] An integrated cooling and heat recovery system (ICHRS) is installed in the data center to cool the servers and recover waste heat to supply the heat loads in the microgrid (such as hot water and heat sources for offices). The power consumption and waste heat recovery power of the ICHRS are calculated as follows:

[0089]

[0090]

[0091] in, represents the electric power consumed by the ICHRS of the i-th data center at time t, represents the heat power recovered by the ICHRS of the i-th data center at time t, ω hr is the heat recovery coefficient.

[0092] 2.6) Carbon trading model

[0093] The reward-penalty ladder carbon trading model is as follows: Figure 4 As shown. In order to limit carbon emissions, a reward-penalty ladder carbon trading mechanism is introduced. When the total carbon emissions are lower than the government's free quota for carbon emissions, the government will provide certain rewards and subsidies. The carbon emissions of the microgrid are calculated using the following formula:

[0094]

[0095] in, represents the carbon emission cost of the i-th data center, c is the benchmark carbon price, λ is the reward coefficient, and E c i is the carbon quota of the data center, v is the length of the carbon emission interval, is the total carbon emissions of the data center, and ξ is the growth rate of the step-by-step carbon price.

[0096] 3) Based on the day-ahead forecast data and the cross-domain green data center electro-thermal coupling system model, a strategy based on spatiotemporal task migration is used to schedule the energy of the cross-domain green data center electro-thermal coupling system and obtain a scheduling plan;

[0097] 3.1) Preliminary dispatch of distributed power generation and grid power purchase

[0098] Based on forecast data for photovoltaic power generation, wind power generation, power / heat load, and the number of delay-sensitive tasks arriving at each data center, the Gurobi solver is used to determine the hourly optimal output power for each generation source and the hourly power purchase from the grid. It is important to note that this step does not consider delay-tolerant tasks. The preliminary scheduling results are then used as the basis for subsequent scheduling.

[0099] 3.1.1) Determine the objective function

[0100] The objective function of the preliminary scheduling stage is:

[0101] min C=C om +C fuel +C grid +C CO2

[0102] Among them, C is the total cost in a scheduling cycle, C om ,C fuel ,C grid ,C CO2 They are respectively the operation and maintenance cost of system equipment, fuel cost, grid electricity purchase cost and carbon emission cost.

[0103] The operation and maintenance costs are as follows:

[0104]

[0105]

[0106]

[0107]

[0108]

[0109] in, are the operation and maintenance costs of photovoltaic, wind power, electric energy storage and thermal energy storage respectively. is the photovoltaic output power of the i-th data center at time t, is the wind power output of the i-th data center at time t, is the charging power of the energy storage system of the i-th data center at time t, is the discharge power of the energy storage system of the i-th data center at time t, is the heat absorption power of the thermal energy storage system of the i-th data center at time t, is the heat release power of the thermal energy storage system of the i-th data center at time t. are the unit operation and maintenance costs of photovoltaic, wind power, electric energy storage and thermal energy storage systems, respectively. Ndc represents the number of data centers, T represents the total time, and Δt represents the length of a single scheduling time slot.

[0110] The fuel cost is shown below:

[0111]

[0112] Among them, c fuel is the unit fuel cost, is the output power of the gas boiler of the i-th data center at time t.

[0113] The cost of purchasing electricity from the power grid is shown in the following formula:

[0114]

[0115] in, is the market electricity price of the i-th data center at time t, is the power purchased by the i-th data center at time t.

[0116] The cost of carbon emissions is shown in the following formula:

[0117]

[0118] Among them, the carbon emission cost of data center i is Depends on its carbon emissions The calculation method is as follows:

[0119]

[0120] Among them, ρ grid The unit carbon emission coefficient for supplying electricity to the grid, ρ gb is the unit carbon emission coefficient of the gas boiler.

[0121] 3.1.2) Determine the constraints

[0122] In order to ensure the reliability of the electrothermal coupling system of a cross-domain green data center, it is necessary to fully consider the constraints of the data center, power balance, and equipment operation.

[0123] To ensure the implementation of the cloud computing service level agreement, each task must be completed before the deadline, so SLA constraints are introduced:

[0124] ft j ≤ddl j

[0125] To ensure service quality and prevent data center overload or congestion, optimized scheduling must meet QoS constraints:

[0126]

[0127] in, represents the utilization rate of the xth server in data center i at time t, μ max The upper limit of server utilization.

[0128] Tasks are spatially migrated between different data centers through communication networks. Communication networks usually have bandwidth, and the amount of data transmitted per unit time cannot exceed its bandwidth. Therefore, the bandwidth constraint should be met:

[0129]

[0130] Among them, DS j Indicates task task j The amount of data, BW d→d' represents the communication bandwidth from data center d to data center d', represents the set of tasks from data center d to data center d'.

[0131] The power balance constraints are shown in the following two equations:

[0132]

[0133]

[0134] Where, represents the basic electric load power of the microgrid where data center i is located at time t (other power consumption except the data center), represents the heat load power of data center i at time t.

[0135] Equipment operation constraints include upper and lower power limits for wind power, photovoltaic power, gas boilers, and power grids, electric / thermal energy storage system constraints, and waste heat recovery system constraints.

[0136] The upper and lower power limits for wind power, photovoltaic power, gas boilers, and the power grid are as follows:

[0137]

[0138]

[0139]

[0140]

[0141] Where, and They represent the upper limits of photovoltaic and wind power output of data center i at time t, and where t represents the upper limit of the gas boiler power and the power purchased from the power grid for data center i, respectively. Due to the influence of natural factors such as wind speed and sunlight, the maximum output of wind power and photovoltaic power varies at different times. Therefore, the upper limits of wind power and photovoltaic power may vary at any moment. However, the gas boiler and power purchased from the power grid are not affected by the natural environment, and their upper limits are independent of time t.

[0142] The operating constraints of the electric energy storage system are as follows:

[0143] Soc min,i ≤Soc i,t ≤Soc max,i

[0144]

[0145]

[0146]

[0147] Where, Soc i,t represents the charge of the battery energy storage system of data center i at time t, Soc min,i and Soc max,i represent the lower and upper limits of the charge of the battery energy storage system of data center i, and They represent the charging and discharging states of the battery energy storage system of data center i at time t. It should be noted that and It is a 0-1 variable. When the battery energy storage system is in the charging state and When the battery energy storage system is in the discharge state and and denote the charging and discharging power of the battery energy storage system of data center i at time t, and They represent the upper limits of charging and discharging power of the battery energy storage system of data center i respectively.

[0148] The operating constraints of the thermal energy storage system are as follows:

[0149] Soh min.i ≤Soh i,t ≤Soh max.i

[0150]

[0151]

[0152]

[0153] Where, Soh i,t represents the heat storage capacity of the heat storage system of data center i at time t, Soh min,i and Soh max,i represent the lower and upper limits of the heat storage capacity of the heat storage system of data center i, and They represent the charging and releasing states of the heat storage system of data center i at time t. It should be noted that and It is a 0-1 variable. When the heat storage system is in the charging state and When the heat storage system is in the heat release state and and denote the charging and discharging powers of the heat storage system of data center i at time t, and They represent the upper limits of charging and discharging power of the heat storage system of data center i respectively.

[0154] The operating constraints of the waste heat recovery system are as follows:

[0155]

[0156]

[0157]

[0158] Where, represents the electrical power consumed by the ICHRS system of data center i at time t, represents the upper limit of the power that the ICHRS system of data center i can consume, represents the amount of heat successfully recovered by the ICHRS system of data center i at time t, It represents the upper limit of heat that can be recovered by the ICHRS system of data center i.

[0159] After obtaining the objective function and constraints, the Gurobi solver is used to solve the problem and obtain a preliminary scheduling plan. The amount of wind and solar power curtailment in each data center at each moment is calculated. Based on the curtailment situation, the next step of task migration and energy scheduling is carried out.

[0160] 3.2) Consuming local renewable energy in each data center

[0161] Based on the preliminary scheduling results, the power of unused renewable energy (photovoltaic and WT power generation) in each time slot can be obtained. Delay-tolerant tasks can be assigned to time slots with depleted renewable energy, thereby improving renewable energy utilization efficiency. Delay-tolerant tasks in each data center should initially consume local renewable energy. If the data center does not deplete renewable energy before the deadline, the task will be migrated according to the policy.

[0162] 3.3) Dual-dimensional migration of time and space tasks

[0163] For delay-tolerant tasks that were not assigned in the previous step, check whether the deadline for the task is sufficient. If the deadline is sufficient, migrate the task to other data centers with renewable energy for execution. If other data centers have no remaining renewable energy output before the deadline, migrate the task to the data center with the lowest electricity price for execution, and purchase electricity from the power grid to meet the electricity consumption required for the task execution.

[0164] Specifically, for a delay-tolerant task that was not assigned in the previous step j , first compare arr j +Δt l and ddl j -Δt l The size between, where Δt l is the transmission time of the task between different data centers. j +Δt l Greater than ddl j -Δt l , it means there is enough time to complete the task j Migrate to other data centers to execute and return the execution results, and then sequentially query other data centers to see if there is unused renewable energy until it is found in [arr j +Δt l ,ddl j -Δtl ] During this period, there is another data center that still has wind and solar power curtailment. Then the task j It is migrated to the data center with abandoned wind and solar power for execution, and the execution time is set to [arr j +Δt l ,ddl j -Δt l ] is the moment when the wind and solar power curtailment is highest. j +Δt l ,ddl j -Δt l ] period, if no other data centers have wind and solar curtailment resources available, then the task will be migrated to [arr j +Δt l ,ddl j -Δt l ] during the period with the lowest electricity price of the data center to execute the task at the lowest electricity price, while increasing the power purchase power of the power grid to meet the power consumption required to execute the task. In this step, if arr j +Δt l Less than ddl j -Δt l , then put the task task j Put it in the local electricity price [arr j +Δt l ,ddl j -Δt l ] during the period with the lowest electricity price. And so on, until all tasks are assigned.

[0165] After all tasks are assigned, the energy scheduling plan for each data center can be obtained.

[0166] 4) Implement the dispatch plan and evaluate the results from multiple perspectives including economy, environmental protection, and power supply reliability.

[0167] Corresponding to the aforementioned embodiment of the energy management method of the cross-domain green data center electro-thermal coupling system based on spatiotemporal task migration, the present invention also provides an embodiment of the energy management device of the cross-domain green data center electro-thermal coupling system based on spatiotemporal task migration.

[0168] See also Figure 6 An embodiment of the present invention provides an energy management device for an electro-thermal coupling system of a cross-domain green data center based on spatio-temporal task migration, comprising a memory and one or more processors, wherein the memory stores executable code. When the processor executes the executable code, it is used to implement the energy management method for an electro-thermal coupling system of a cross-domain green data center based on spatio-temporal task migration in the above embodiment.

[0169] The embodiment of the cross-domain green data center electrothermal coupling system energy management device based on spatiotemporal task migration provided by the present invention can be applied to any device with data processing capabilities, and the any device with data processing capabilities can be a device or apparatus such as a computer. The device embodiment can be implemented through software, or through hardware or a combination of software and hardware. Taking software implementation as an example, as a device in a logical sense, it is formed by the processor of any device with data processing capabilities in which it is located reading the corresponding computer program instructions in the non-volatile memory into the memory for execution. From the hardware level, if Figure 6 As shown, it is a hardware structure diagram of any device with data processing capability where the energy management device of the cross-domain green data center electrothermal coupling system based on spatiotemporal task migration provided by the present invention is located. Figure 6 In addition to the processor, memory, network interface, and non-volatile memory shown, any device with data processing capabilities in which the apparatus in the embodiment is located may also include other hardware, generally based on the actual functions of the device with data processing capabilities, which will not be described in detail.

[0170] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0171] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present invention. A person of ordinary skill in the art can understand and implement the present invention without inventive work.

[0172] An embodiment of the present invention also provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, the energy management method of the cross-domain green data center electrothermal coupling system based on spatiotemporal task migration in the above embodiment is implemented.

[0173] The computer-readable storage medium may be an internal storage unit of any device with data processing capabilities described in any of the aforementioned embodiments, such as a hard disk or memory. The computer-readable storage medium may also be an external storage device of any device with data processing capabilities, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), an SD card, a flash card, etc. equipped on the device. Furthermore, the computer-readable storage medium may also include both an internal storage unit and an external storage device of any device with data processing capabilities. The computer-readable storage medium is used to store the computer program and other programs and data required by any device with data processing capabilities, and may also be used to temporarily store data that has been output or is to be output.

[0174] The present invention also provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the energy management method of a cross-domain green data center electrothermal coupling system based on spatiotemporal task migration.

[0175] The above embodiments are used to illustrate the present invention rather than to limit the present invention. Any modifications and changes made to the present invention within the spirit of the present invention and the protection scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A cross-domain green data center electrothermal coupling system energy management method based on spatiotemporal task migration, characterized in that: The method comprises the following steps: 1) Obtain day-ahead forecast data for the cross-domain green data center electric and thermal coupling system, including renewable energy output, electric / thermal load demand, and data center task arrival rate; 2) Establish a cross-domain green data center electrical and thermal coupling system model. Data center tasks are divided into delay-sensitive tasks that need to be processed immediately and delay-tolerant tasks that need to be completed before a deadline. Delay-sensitive tasks are called interactive tasks, while delay-tolerant tasks are called batch tasks and can be deferred before a deadline. If the microgrid where the data center is located does not have sufficient renewable energy, the tasks it receives can be transferred to other data centers with renewable energy or lower electricity prices for processing; 3) Based on the day-ahead forecast data and system model, preliminary results are obtained for scheduling delay-sensitive tasks. The unused renewable energy power in each time slot is obtained. A strategy based on spatiotemporal task migration is used to schedule delay-tolerant tasks, resulting in a schedule plan and execution. Specifically: 3.1) Preliminary dispatch of distributed generation and grid power purchase Based on the day-ahead forecast data for photovoltaic power generation, wind power generation, and power / heat load, as well as the number of delay-sensitive tasks arriving at each data center, the Gurobi solver is used to obtain the hourly optimal output power of each power source and the hourly power purchase from the grid, enabling preliminary scheduling. 3.2) Consuming local renewable energy in each data center Based on the preliminary scheduling results, the power of unused renewable energy in each time slot is obtained. For delay-tolerant tasks, they are assigned to time slots with abandoned renewable energy for processing, thereby improving the utilization efficiency of renewable energy. Delay-tolerant tasks in each data center should first consume local renewable energy. If the data center does not abandon renewable energy before the deadline, the task will be migrated according to the spatiotemporal task migration strategy. 3.3) Dual-dimensional migration of tasks in time and space For the delay-tolerant tasks that are not assigned, first compare and The size between Indicates the task arrival time, Indicates the deadline of the task. is the transmission time of the task between different data centers; if Greater than , it means that there is enough time to migrate the delay-tolerant task to other data centers for execution and return the execution results, so the other data centers are sequentially queried for unused renewable energy until it is found in During this period, if there are still wind and solar power curtailment in other data centers, then the delay-tolerant task is migrated to the data center with wind and solar power curtailment for execution, and the execution time is set to The moment with the highest wind and solar power curtailment during the period; If in During this period, if no other data centers have wind or solar power available, the delay tolerance task will be migrated to The task is executed at the time when the electricity price of the data center with the lowest electricity price is the lowest during the period, and the power purchased from the power grid is increased to meet the power consumption required to execute the delay-tolerant task; if Less than , then put the delay tolerance task on the local electricity price The task will be executed at the time when the electricity price is the lowest during the period; and so on, until all tasks are assigned.

2. The energy management method for the cross-domain green data center electrothermal coupling system based on spatiotemporal task migration according to claim 1 is characterized in that: Step 2) is as follows: 2.1) Cross-domain green data center Cross-domain green data centers are distributed across different geographic regions and connected via communication networks. Each microgrid is grid-connected, and there is complementarity between different microgrids, allowing tasks to be transferred to other data centers for processing. 2.2) Grid-connected green data center model A data center is considered a grid-connected microgrid with the data center as the primary load. It consists of photovoltaic (PV) equipment, wind turbines, electric / thermal energy storage systems, and various types of electric / thermal loads. The data center contains servers, an integrated cooling and heat recovery system (ICHRS), and some thermal loads. The servers process user tasks and provide cloud computing services, while the ICHRS cools the servers, recovers their waste heat, and supplies it to the thermal loads. 2.3) Data Center Energy Consumption Model Data center energy consumption Use the following two formulas to calculate: Where, is the index number of the data center, For the Data centers in The server power consumption at the moment, For the Data centers in Energy efficiency at the moment, expressed as the ratio of the total power consumption of the data center to the power consumption of the server; Indicates the Data centers in The number of servers powered on at the moment, is the no-load power consumption of a single server, is the full load power consumption of a single server, For the Data centers in The average CPU utilization of the server at the time, For the The number of data center management nodes, For the The power of a single management node in a data center; energy efficiency index It is related to the outdoor air temperature of the data center and is calculated by the following formula: in, Indicates the Data centers in The outdoor temperature at the time, a, b and c are coefficients; 2.4) Data Center Task Model There are two types of tasks in data centers: a. Delay-sensitive tasks, also known as interactive tasks, which need to be processed immediately upon arrival; b. Delay-tolerant tasks, also known as batch tasks, which can be deferred until their deadline. Task It is expressed as the following formula: in, is the index number of the task, Indicates the task arrival time, Indicates the deadline of the task. Indicates the time when the task starts executing. Indicates the completion time of the task. Indicates the number of tasks; 2.5) Waste heat recovery system model The power consumption and waste heat recovery power of ICHRS are calculated by the following two formulas: in, Indicates the Data centers in The electrical power consumed by the ICHRS at the moment, Indicates the Data centers in The thermal power recovered by the ICHRS at the moment, is the heat recovery coefficient; 2.6) Carbon trading model The carbon emissions of the microgrid are calculated using the following formula: in, Indicates the The carbon emission cost of a data center, is the benchmark carbon price, is the reward coefficient, Carbon quota for data centers, is the length of the carbon emission interval, is the total carbon emissions of the data center, is the growth rate of the step carbon price.

3. The energy management method for the cross-domain green data center electrothermal coupling system based on spatiotemporal task migration according to claim 1 is characterized in that: The goal of preliminary scheduling is to minimize the total cost within the scheduling cycle. The costs include the operation and maintenance costs of system equipment, fuel costs, grid electricity purchase costs, and carbon emission costs. It is also necessary to ensure that each task must be completed before the deadline. Optimal scheduling needs to meet QoS constraints, task migration needs to meet communication network bandwidth constraints, and task scheduling needs to meet power balance constraints and equipment operation constraints.

4. A cross-domain green data center electrothermal coupling system energy management device based on spatiotemporal task migration, comprising a memory and one or more processors, wherein the memory stores executable code, characterized in that: When the processor executes the executable code, it implements a cross-domain green data center electrothermal coupling system energy management method based on spatiotemporal task migration according to any one of claims 1 to 3.

5. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, an energy management method for a cross-domain green data center electrothermal coupling system based on spatiotemporal task migration according to any one of claims 1 to 3 is implemented.

6. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, an energy management method for a cross-domain green data center electrothermal coupling system based on spatiotemporal task migration is implemented as described in any one of claims 1-3.

Images