Calculation-electricity-heat coupled collaborative optimization method for integrated energy system of data center

Abstract

The invention provides a computing-electric-thermal coupled data center integrated energy system collaborative optimization method, and relates to the field of data center energy-saving management and control, and the method comprises the steps: firstly obtaining real-time data collected by a data center; secondly, a data center load model is constructed based on the relation among real-time data quantification calculation power, electric power and heating power; constructing a collaborative optimization model according to the data center load model, and solving the collaborative optimization model to obtain an optimization result; and finally, adjusting calculation task distribution, a power supply strategy and cooling system operation parameters according to an optimization result to form a calculation-electricity-heat coupled data center integrated energy system collaborative optimization strategy. According to the calculation-electricity-heat coupled data center comprehensive energy system collaborative optimization method, collaborative optimization is carried out on the data center energy system through multi-energy complementation, and energy conservation and efficiency improvement of the data center are achieved.

Description

Technical Field

[0001] The present invention relates to the field of energy-saving control of data centers, and specifically relates to a collaborative optimization method for an integrated energy system of a computing-electricity-thermal coupling data center. Background Art

[0002] With the rapid development of the new generation of information technology, the data center industry has developed rapidly, and its energy consumption problem has become increasingly prominent. The International Energy Agency (IEA) pointed out that the energy consumption of global data centers was approximately 460 TWh in 2022, accounting for about 1.3% of the global total electricity consumption. It is expected that this figure will increase to more than 1000 TWh by 2026. The energy consumption of data centers mainly comes from IT equipment, refrigeration systems, and power supply and distribution systems. The energy consumption of IT equipment accounts for about 50% - 60% of the total energy consumption of data centers, and the energy consumption of refrigeration systems accounts for about 30% - 40%.

[0003] However, the computing power scheduling, power distribution, and cooling systems of data centers usually operate independently, resulting in resource mismatches. When the servers are running at high loads, the cooling system responds laggingly, which is likely to cause local overheating. When the power supply does not dynamically adjust with the computing power demand, redundant power consumption is likely to occur. By performing collaborative optimization of computing-electricity-thermal coupling in data centers, energy efficiency can be improved, costs can be reduced, and carbon emissions can be reduced through collaborative management of computing resources, energy supply, and heat dissipation systems.

[0004] Therefore, how to perform collaborative optimization of the energy system of data centers through multi-energy complementarity to achieve energy conservation and efficiency improvement in data centers has become an urgent problem to be solved. Summary of the Invention

[0005] (1) Technical Problems to be Solved

[0006] Aiming at the deficiencies of the prior art, the present invention provides a collaborative optimization method for an integrated energy system of a computing-electricity-thermal coupling data center, which solves the technical problems of existing computing-electricity-thermal coupling collaborative optimization in data centers.

[0007] (2) Technical Solutions

[0008] To achieve the above objectives, the present invention is realized through the following technical solutions:

[0009] In a first aspect, the present invention provides a collaborative optimization method for an integrated energy system of a computing-electricity-thermal coupling data center, including the following steps:

[0010] Obtain the real-time data collected by the data center, where the real-time data includes load data, power consumption data, external energy price data, environmental parameter data, workload type data, and cooling system operating status data;

[0011] Based on the real-time data, quantify the relationship among computing power, electricity, and heat, and construct a data center load model;

[0012] Construct a collaborative optimization model according to the data center load model, and solve the collaborative optimization model to obtain an optimization result;

[0013] Adjust the computing task allocation, power supply strategy, and cooling system operation parameters according to the optimization result to form a collaborative optimization strategy for the computing-electricity-heat coupled data center integrated energy system.

[0014] Preferably, the constructing a data center load model based on the real-time data to quantify the relationship among computing power, electricity, and heat includes:

[0015] Construct an initial data center load model in the form of a mixed-integer nonlinearity based on the real-time data collected by the data center;

[0016] Obtain a linear data center load model by decomposing the initial data center load model.

[0017] Preferably, the initial data center load model is:

[0018]

[0019] where p it represents the power consumption of the data center at time t; represents the power consumption of the IT equipment with the lowest QoS in the data center at time t; represents the power consumption of the redundant IT equipment for higher QoS in the data center at time t; represents the power consumption of the cooling system in the data center at time t; represents the power consumption of other equipment in the data center; represents the number of edge switches in the data center; represents the number of aggregation switches in the data center; represents the number of core switches in the data center; represents the number of servers in the data center; represents the number of servers in the data center; represents the rated power of each active edge switch in the data center; represents the rated power of each active aggregation switch in the data center; represents the rated power of each active core switch in the data center; represents the idle power of each active server in the data center; represents the peak power of each active server in the data center; Denotes the minimum number of active servers serving interactive workloads in the data center at time t; Denotes the minimum number of active servers serving batch workload q in the data center at time t; Denotes the number of redundant active servers serving interactive workloads in the data center at time t; Denotes the number of redundant active servers serving batch workload q in the data center at time t; u i Denotes the average service rate of servers in the data center; λ iδt Denotes the interactive workload δ allocated from FS converted to the data center at time t; χ iqt Denotes the amount of batch workload q transferred to the data center at time t; Denotes the number of servers serving interactive workloads operating at peak power, ignoring their workloads in the data center at time t; Denotes the number of servers serving batch workload q operating at peak power, ignoring their workloads in the data center at time t; Denotes the power consumption of the cooling system; b 1i 、b 2i Denotes the empirical constant of the cooling system in the data center; h it Denotes the thermal cooling power of the data center at time t;

[0020] The linear data center load model is as follows:

[0021]

[0022] Where, β 1it Denotes the power consumption baseline with the lowest QoS in the data center at time t; α 1i Denotes that when the power consumption of unit IT equipment increases, in the absence of TSTI, the total power consumption in the data center increases; Denotes the adjusted power demand contributed by GWB in the data center at time t compared to the set baseline; Denotes the adjusted power demand contributed by BWS in the data center at time t compared to the set baseline; Denotes the adjusted power demand contributed by TSTI in the data center at time t compared to the set baseline;

[0023] The constraint conditions of the linear data center load model include:

[0024]

[0025] Where, β 2iRepresents the difference in the energy storage level between time t and time t-1 in the data center, without heat dissipation and power generation, where the indoor temperature at time t-1 is the indoor temperature baseline; β 3i Represents the energy storage level of the initial data center; β 4it Represents the increase in the power consumption of the data center when processing the baseline amount of interactive workloads at the lowest QoS and without TSTI during time slot t; β 5it Represents the increase in the power consumption of the data center when processing the baseline amount of batch workload q at the lowest QoS and without TSTI during time slot t; β 6it Represents the number of servers not in the active state in the data center; β 7it Represents the margin of the IT device power in the data center at time t; β 8it The cooling power deficit in the data center at time t; β 9it The cooling power margin in the data center at time t; α 2i Represents the increase in the power consumption of the data center when processing a unit amount of interactive workloads at the lowest QoS and without TSTI; α 3i Represents that when the unit energy storage level at time t-1 increases, the data center energy storage level at time t also increases; α 4i Represents that when reducing the unit adjusted power demand contributed by TSTI, the energy storage level of the data center at time t increases; α 5i Represents that when the active servers in the data center process the maximum interactive workload, the power consumption increases and the QoS is the lowest; α 6i Represents that when the active servers in the data center process the maximum batch workload, the power consumption increases and the QoS is the lowest; α 7i Represents that when the unit amount of the total increased power consumption of the data center increases, the power consumption of the cooling system increases without TSTI; α 8i Represents that when the unit IT device power consumption increases, the power consumption of the cooling system in the data center increases without TSTI; Represents the energy storage level of the data center at time t.

[0026] Preferably, constructing the collaborative optimization model according to the data center load model specifically includes:

[0027] Establishing an objective function with the goal of minimizing the total operating cost of the integrated energy system of the data center that realizes the coupling of computing - electricity - heat;

[0028] Establishing the constraint conditions of the collaborative optimization model, and the constraint conditions include: power balance constraint, thermal balance constraint, power balance constraint, renewable energy constraint, and generator set constraint.

[0029] Preferably, the objective function is:

[0030]

[0031] In the formula, C DA represents the scheduling cost in the day-ahead stage, C RT represents the expected balancing cost in the real-time stage, C carbon represents the cost generated by carbon trading;

[0032]

[0033]

[0034] Among them, represents the traditional thermal power cost coefficient, the cost coefficient of the combined heat and power unit, represents the thermal power of the thermal power unit in the day-ahead market, represents the thermal power of the thermal power unit in the real-time market, represents the power of the combined heat and power unit in the day-ahead market, represents the power of the combined heat and power unit in the real-time market, + represents the positive regulation in the scenario, - represents the negative regulation in the scenario, γ s represents the probability of the real-time scenario, λ C represents the price of carbon emissions, and φ represents the set of DC and other variables in the system operation.

[0035] Preferably, the power balance constraint is:

[0036]

[0037] In the formula, represents the local load electrical power in the day-ahead stage, represents the local load electrical power in the real-time stage, represents the total electrical power of the data center in the day-ahead stage, represents the total electrical power of the data center in the real-time stage, represents the renewable energy power in the day-ahead stage, represents the renewable energy power in the real-time stage, represents the power of the transmission line in the day-ahead stage, represents the power of the transmission line in the real-time stage;

[0038] The thermal balance constraint is:

[0039]

[0040] Among them, represents the local load thermal power in the day-ahead stage, represents the local load thermal power in the real-time stage; Represents the total thermal power of the data center on the day-ahead basis; Represents the total thermal power of the data center in real-time;

[0041] The power balance constraint is:

[0042]

[0043] Wherein, Respectively represent the transmission line powers in the day-ahead stage and the real-time stage, A l,n Represents the power grid incidence matrix, Represents the voltage angle of the starting bus, Represents the voltage angle of the ending bus;

[0044] The renewable energy constraint is:

[0045]

[0046] Wherein, Represents the maximum power generation capacity of the renewable energy unit;

[0047] The generator set constraint is:

[0048]

[0049] Wherein, η represents the thermal power ratio of the combined heat and power unit.

[0050] In a second aspect, the present invention provides a computing-electricity-thermal coupling data center integrated energy system collaborative optimization system, including:

[0051] An acquisition module, which acquires real-time data collected by the data center;

[0052] A model construction module, which quantifies the relationship between computing power, electricity and heat based on the real-time data and constructs a data center load model;

[0053] A model optimization module, which constructs a collaborative optimization model according to the data center load model and solves the collaborative optimization model to obtain an optimization result;

[0054] A scheduling output module, which adjusts the computing task allocation, power supply strategy and cooling system operation parameters according to the optimization result to form a computing-electricity-thermal coupling data center integrated energy system collaborative optimization strategy.

[0055] In a third aspect, the present invention provides a computer-readable storage medium, which stores a computer program for the collaborative optimization of the computing-electricity-thermal coupling data center integrated energy system, wherein the computer program enables a computer to execute the computing-electricity-thermal coupling data center integrated energy system collaborative optimization method as described above.

[0056] Fourthly, the present invention provides an electronic device, including:

[0057] One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the programs include a method for collaborative optimization of an integrated energy system of a data center for performing the calculation-electricity-thermal coupling as described above.

[0058] (3) Beneficial effects

[0059] The present invention provides a method for collaborative optimization of an integrated energy system of a data center for calculation-electricity-thermal coupling. Compared with the prior art, it has the following beneficial effects:

[0060] The present invention establishes a data center load model by acquiring data collected in real time by the data center. Through the collaborative optimization of coupling computing power, electricity, and heat, it realizes the flexible scheduling of energy in time and space in the data center, significantly improves the overall energy efficiency, then introduces the carbon trading cost, establishes a collaborative optimization model, optimizes the energy scheduling plan, reduces the dependence on traditional thermal power, and reduces carbon emissions. Description of the drawings

[0061] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0062] Figure 1 It is a schematic flowchart of the method for collaborative optimization of an integrated energy system of a data center for calculation-electricity-thermal coupling of the present invention. Specific implementation manners

[0063] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0064] The embodiments of the present application solve the problem of the single management method of the existing data center by providing a method for collaborative optimization of an integrated energy system of a data center for calculation-electricity-thermal coupling, realizing the coupling of multiple scheduling methods, and achieving energy conservation and efficiency improvement of the data center.

[0065] The technical solutions in the embodiments of the present application for solving the above technical problems are generally as follows:

[0066] With the rapid development of artificial intelligence technology, the number and scale of data centers have increased explosively, and the energy consumption problem has become increasingly prominent. While consuming a large amount of electric energy, data centers also generate a large amount of waste heat. If this waste heat is directly discharged into the environment, it will not only cause energy waste but also have a negative impact on the environment. Therefore, from the perspective of energy utilization efficiency, the coupling of electric energy and heat energy can significantly improve the comprehensive energy utilization rate of data centers.

[0067] On the other hand, IT equipment and the like in the data center consume a large amount of electricity for data processing during operation, generating a large amount of heat while generating computing power. The power supply provides power support for the generation of computing power, and the size of the computing power and the operating state of the equipment directly determine the amount of heat generated, thus forming a tight coupling relationship among electricity - computing power - heat. The coupling of electricity, computing power, and heat in this data center can further meet the requirements of different scenarios, ensure the stable operation of the system, improve the energy utilization rate of the data center, and better achieve energy conservation and emission reduction.

[0068] With the increase in workload processing, the traditional operation mode of data centers faces a series of challenges. The old energy-saving technologies can no longer meet the increasingly complex operating environment of data centers, and the normalized workload scheduling mode makes the operation flexibility of data centers unable to be fully exerted. Traditional data center energy management systems often only focus on energy conservation in a certain aspect, such as the optimization of power supply or the improvement of cooling systems, and lack the comprehensive coordination and optimization of the entire energy system. This single management method is difficult to achieve energy conservation and efficiency improvement in data centers.

[0069] Therefore, it is necessary to couple computing power, electricity, and heat to achieve the collaborative optimization scheduling of the integrated energy system of data centers. Existing scheduling methods are generally divided into time scheduling and space scheduling. There are also scheduling divisions for electricity and heat in different time and space scheduling dimensions. The coupling of computing power, electricity, and heat in the data center, as well as the implementation of waste heat recovery in the data center, makes it necessary to consider the collaborative matching of computing power, electricity, and heat simultaneously when conducting the collaborative optimization scheduling of the data center. This requires coupling multiple scheduling methods to give full play to the time and space flexibility of the data center.

[0070] The collaborative optimization method for the integrated energy system of a computing - electricity - heat - coupled data center used in this application is mainly used to solve the following technical problems existing in traditional methods:

[0071] 1) Existing collaborative optimization methods for data centers often only focus on energy conservation in a single aspect and lack the comprehensive coordination and optimization of the entire energy system, failing to fully achieve energy conservation and efficiency improvement in data centers.

[0072] 2) The existing data center collaborative optimization methods generally only adopt the scheduling methods of a single energy type for collaborative optimization, and cannot couple multiple types of energy for optimal scheduling.

[0073] 3) The existing data center collaborative optimization methods mainly focus on the operational optimization within the data center and fail to fully utilize the synergy between the data center and the power system.

[0074] To better understand the above technical solutions, the following will describe the above technical solutions in detail in combination with the accompanying drawings of the specification and specific implementation manners.

[0075] As Figure 1 shown, in one embodiment, a collaborative optimization method for an arithmetic-electric-thermal coupled data center integrated energy system includes the following steps:

[0076] Step S110, obtaining the real-time data collected by the data center.

[0077] Step S120, quantifying the relationship between computing power, electricity, and heat based on the real-time data, and constructing a data center load model.

[0078] Step S130, constructing a collaborative optimization model according to the data center load model, and solving the collaborative optimization model to obtain an optimization result.

[0079] Step S140, adjusting the computing task allocation, power supply strategy, and cooling system operation parameters according to the optimization result to form a collaborative optimization strategy for the arithmetic-electric-thermal coupled data center integrated energy system.

[0080] In the embodiment of the present invention, a data center load model is established by obtaining the real-time data collected by the data center. Through the collaborative optimization of coupling computing power, electricity, and heat, the spatio-temporal flexible scheduling of energy within the data center is realized, significantly improving the overall energy efficiency. Subsequently, the carbon trading cost is introduced to establish a collaborative optimization model, optimize the energy scheduling scheme, reduce the dependence on traditional thermal power, and reduce carbon emissions. In step S110, the real-time data collected by the data center is obtained. Specifically, it includes: The real-time data collected by the data center includes the following aspects:

[0081] Computing load data: The computing load conditions of each server in the data center are collected in real time through the server monitoring system, including CPU utilization rate, memory usage rate, network bandwidth occupancy rate, etc.

[0082] Power consumption data: The power consumption conditions of each device in the data center are monitored in real time through intelligent electricity meters and power monitoring devices, including the electricity consumption of IT devices (such as servers, switches, etc.), cooling systems (such as air conditioners, liquid cooling systems, etc.), and other auxiliary devices.

[0083] External energy price data: Real-time external energy price data is obtained through the connection with the power market information system.

[0084] Environmental parameter data: Environmental parameters outside the data center, such as outdoor temperature, humidity, wind speed, etc., are collected in real time through environmental monitoring devices.

[0085] Workload type data: The distribution of different types of workloads in the data center, including interactive workloads and batch processing workloads, is collected in real time through the task scheduling system.

[0086] Cooling system operation status data: The operation status data of the cooling system is collected in real time through the monitoring devices of the cooling system.

[0087] The collaborative optimization method for the integrated energy system of the computing-electricity-thermal coupling data center in this embodiment collects multi-dimensional data in real time, providing input for subsequent modeling and optimization.

[0088] In step S120, based on the real-time data collected by the data center, the relationships among computing power, electricity, and heat are quantified, and a data center load model is constructed. Specifically, it includes:

[0089] In the embodiment of the present invention, an Internet service company with multiple geographically dispersed data centers is preferably selected. When each data center is powered by a dedicated substation in the power grid, all substations are operated by the same utility company and connected to different power system nodes. The embodiments of this application consider both interactive workloads and batch processing workloads. Compared with interactive workloads, the service delay tolerance and computing requirements of batch processing workloads are more important in practical applications. Therefore, batch processing workloads are selected for local processing. Within a time slot, decisions regarding workload allocation, cooling system control, and power system scheduling will be updated. At the same time, this time period (for example, once an hour) is much longer than the tolerable service delay of interactive workloads and much shorter than the tolerable service delay of batch processing workloads. In each time slot, the embodiments of the present invention consider coupling multiple adjustment methods to adjust the power and heat demands of the data center.

[0090] In the embodiment of the present invention, a mixed-integer non-linear form is used to describe the power consumption composition of the data center load, and a linear data center load model is obtained through model decomposition.

[0091] The initial data center load model is formulated into a mixed-integer non-linear form as follows:

[0092]

[0093] Equation (1) describes the power consumption composition of the data center load. Among them, the total power consumption of IT devices is divided into two parts. The first part is the power consumption to meet the minimum Quality of Service (QoS), and the second part is the additional power consumption to provide computing resource redundancy for higher QoS (Quality of Service). In the formula, p it represents the power consumption of the data center at time t, represents the power consumption of IT devices with the minimum QoS in the data center at time t, represents the power consumption of redundant IT devices for higher QoS in the data center at time t, represents the power consumption of the data center cooling system at time t, represents the power consumption of other devices in the data center.

[0094]

[0095] Equation (2) shows the first part of the IT device power consumption, which increases with the increase in the number of active servers and the workload being processed. Equation (3) shows the second part of the IT device power consumption, which includes two considerations. The first term is the increase in the power consumption of IT devices considering the additional use of active servers; the second term is the increase in the power consumption of IT devices because some active servers will run at a service rate higher than their minimum requirements. In the formula, represents the number of servers serving interactive workloads / batches of workload q running at peak power, regardless of how their workload is in the data center at time t.

[0096] In the formula, respectively represent the number of edge switches, aggregation switches, core switches, and servers in the data center, respectively represent the rated power of each active edge switch, aggregation switch, and core switch in the data center, and represent the idle power and peak power of each active server in the data center, represents the minimum number of active servers serving interactive workloads in the data center at time t, represents the minimum number of active servers serving batch workload q in the data center at time t. represents the number of redundant active servers serving interactive workloads in the data center at time t, represents the number of redundant active servers serving batch workload q in the data center at time t. u i represents the average service rate of servers in the data center, λ iδt represents the interactive workload δ allocated from FS converted to the data center at time t, χiqt Represents the amount of batch workload q transferred to time t in the data center. Represents the number of servers serving the interactive workload running at peak power, ignoring their workload in the data center at time t; Represents the number of servers serving the batch workload q running at peak power, ignoring their workload in the data center at time t.

[0097]

[0098] Equation (4) shows the power consumption of the cooling system, which is assumed to be a linear function of the thermoelectric cooling power generation. In the equation, Represents the power consumption of the cooling system, b 1i , b 2i Represents the empirical constant of the cooling system in the data center. h it Represents the thermoelectric cooling power of the data center at time t.

[0099]

[0100] Equation (5) describes the power consumption range of the cooling system, which cannot exceed its power limit.

[0101] Where, Represents the power consumption of the cooling system, Represents the maximum value of the cooling system power.

[0102]

[0103] Equation (6) is a constraint related to geographical load balancing (GWB), which shows that the total workload assigned from one front-end server (FS) to multiple data centers is equal to the workload arriving at that FS. Equation (7) is a constraint related to batch workload scheduling (BWS), which represents ensuring that the total amount of batch workload scheduled within each time window is equal to the total amount of batch workload arriving in the previous time window, for batch workload scheduling (BWS).

[0104] In the equation, jτ q Represents the tolerated service delay of the batch workload q, φ iqt Represents the amount of batch workload q arriving at the data center at time t, χ iδt Represents the interactive workload δ allocated from the FS converted to the data center at time t, Represents the interactive workload arriving at FSδ at time t, χ iqt Represents the amount of batch workload q transferred to time t in the data center.

[0105]

[0106] Equations (8)-(10) are the constraints related to the Thermal Storage Operation (TSTI). Equation (8) indicates that the internal environmental temperature of the data center is within an acceptable range. Equation (9) is the relationship between the indoor temperature, the generated thermal cooling power, the power consumption of IT equipment and other equipment, and the outdoor temperature. Equation (10) describes the indoor temperature at the initial time.

[0107] Wherein, b 3i and b 4i respectively represent the equivalent thermal resistance and equivalent thermal capacitance in the data center, represents the indoor temperature of the data center at time t, represents the indoor temperature at the initial time in the data center, represents the outdoor temperature of the data center at time t, respectively represent the minimum and maximum indoor temperatures in the data center.

[0108]

[0109] Equation (11) shows the relationship between the minimum active servers and the processed interactive workload, which ensures that the interactive workload is served with a certain probability before the deadline. Wherein, u i represents the average service rate of the servers in the data center, and v represents the service delay tolerating the interactive workload.

[0110]

[0111] Equation (12) indicates that the interactive workload allocated from one FS to a data center, the batch workload transferred to one of the time slots, and the active servers are all non-negative.

[0112]

[0113] Equation (13) describes that the total number of active servers cannot exceed the total number of servers.

[0114]

[0115] Equations (14)-(15) describe the range of active servers operating at peak power, and this range cannot exceed the number of active servers.

[0116] Separate the workload variables, redundant server variables, IT equipment power consumption, and cooling system power consumption from the data center load model, introduce power variables to replace non-power variables, and introduce variables representing regulation to replace variables representing demand, so as to complete the model decomposition. A linear data center load model is obtained through model decomposition, which can fully demonstrate the spatio-temporal load regulation potential of the data center, as shown in Equation (16):

[0117]

[0118] Wherein, p it represents the power consumption of the data center at time t, and β 1it represents the power consumption baseline with the lowest QoS of the data center at time t; α 1i represents that when the power consumption of a unit of IT equipment increases, without TSTI, the total power consumption in the data center increases; represents the adjusted power demand contributed by GWB in the data center at time t compared with the set baseline; represents the adjusted power demand contributed by BWS in the data center at time t compared with the set baseline; represents the adjusted power demand contributed by TSTI in the data center at time t compared with the set baseline, represents the power consumption of redundant IT equipment for higher QoS in the data center at time t

[0119]

[0120] Equations (17)-(27) represent the constraint conditions of the linear data center load model. Wherein, β 2i represents the difference in the energy storage level between time t and time t-1 in the data center, without heat dissipation and power generation, where the indoor temperature at time t-1 is the indoor temperature baseline; β 3i represents the energy storage level of the initial data center; β 4it represents that when processing the baseline amount of interactive workload at time slot t, without TSTI and with the lowest QoS, the power consumption of the data center increases; β 5it represents that when processing the baseline amount of batch workload q at time slot t, without TSTI and with the lowest QoS, the power consumption of the data center increases; β 6it represents the number of servers not in the active state in the data center; β 7it represents the margin of the IT equipment power in the data center at time t; β 8it the cooling power deficit in the data center at time t; β 9it the cooling power margin in the data center at time t; α 2i represents that when processing a unit amount of interactive workload, without TSTI and with the lowest QoS, the power consumption of the data center increases; α 3i represents that when the unit energy storage level at time t-1 increases, the data center energy storage level at time t also increases; α 4i represents that when reducing the unit adjusted power demand contributed by TSTI, the energy storage level of the data center at time t increases; α 4iIndicates that when the active servers in the data center handle the largest interactive workload, the power consumption increases and the QoS is the lowest; α 6i Indicates that when the active servers in the data center handle the largest batch workload, the power consumption increases and the QoS is the lowest; α 7i Indicates that when the unit amount of total increased power consumption in the data center increases, without TSTI, the power consumption of the cooling system increases; α 8i Indicates that when the power consumption of the unit IT equipment increases, without TSTI, the power consumption of the cooling system in the data center increases; Indicates the energy storage level of the data center at time t.

[0121] The calculation-electricity-thermal coupling data center integrated energy system collaborative optimization method of the embodiments of the present invention constructs a calculation-electricity-thermal coupling mathematical model, quantifies the energy relationship, and transfers a linearized load model to the collaborative optimization model.

[0122] In step S130, a collaborative optimization model is constructed according to the data center load model, and the collaborative optimization model is solved to obtain an optimization result. Specifically, it includes:

[0123] Based on the linearized load model in step S120, step S130 further constructs a collaborative optimization model. An objective function is established with the goal of minimizing the total operating cost of the calculation-electricity-thermal coupling data center integrated energy system. The objective function includes three parts: the scheduling cost in the day-ahead stage, the expected balancing cost in the real-time stage, and the cost generated by carbon trading, as shown in Equation (28):

[0124]

[0125]

[0126] Equation (28) is a two-stage stochastic programming model, indicating that under all constraint conditions, the total cost of the system is minimized. The total cost includes three parts: the day-ahead scheduling cost, the real-time balancing cost, and the carbon trading cost. Equation (29) represents the day-ahead scheduling cost, Equation (30) represents the expected balancing cost in the real-time stage, and Equation (31) represents the cost generated by carbon trading, that is, the carbon trading cost calculated due to exceeding the carbon emission limit.

[0127] In the formula, C DA represents the day-ahead scheduling cost, C RT represents the expected balancing cost in the real-time stage, represents the traditional thermal power cost coefficient, The cost coefficient of the combined heat and power unit. φ represents the DC (Data Center data center) and other variable sets in the system operation. represents the thermal power of the day-ahead market, Represents the thermal power of the real-time market. Represents the power of the combined heat and power (CHP) unit in the day-ahead market, Represents the power of the combined heat and power (CHP) unit in the real-time market. + indicates positive regulation in the scenario, and - indicates negative regulation in the scenario. Indicates the carbon emissions at time t, i.e., the total carbon emissions within this time period; E cap Indicates the carbon emission limit, i.e., the maximum total carbon emissions allowed by the system; γ s Indicates the probability of the real-time scenario; C carbon Indicates the carbon emission cost; λ C Indicates the price of carbon emissions.

[0128]

[0129] Equation (32) represents the cumulative carbon emissions exceeding the emission limit.

[0130]

[0131] Constraint (33) represents the positive and negative adjustments of traditional thermal power, and constraint (34) represents the positive and negative adjustments of the combined heat and power (CHP) unit. + indicates positive regulation in the scenario, and - indicates negative regulation in the scenario.

[0132] Establish the constraint conditions of the collaborative optimization model, and the constraint conditions include: power balance constraint, thermal balance constraint, electricity balance constraint, renewable energy constraint, and generator set constraint.

[0133] Power balance constraint:

[0134]

[0135] Constraints (36) and (37) represent power balance, and the thermal balance therein is supported by the waste heat recovery of the CHP unit and the data center, as shown in equations (38) and (39).

[0136] In the formula, Represents the local load electric power in the day-ahead stage, Represents the local load electric power in the real-time stage, Represents the total electric power of the data center in the day-ahead stage, Represents the total electric power of the data center in the real-time stage, Represents the renewable energy power in the day-ahead stage, Represents the renewable energy power in the real-time stage, Represents the transmission line power in the day-ahead stage, Represents the transmission line power in the real-time stage.

[0137] Thermal balance constraint:

[0138]

[0139] In the formula, represents the local load thermal power in the day-ahead stage, represents the local load thermal power in the real-time stage; represents the total thermal power of the data center in the day-ahead; represents the total thermal power of the data center in the real-time.

[0140] Power balance constraint:

[0141]

[0142] Constraints (40)-(44) limit the power flow in the transmission line. In the formula, represent the transmission line power in the day-ahead stage and the real-time stage respectively, and A l,n represents the grid incidence matrix, represents the voltage angle of the starting bus, represents the voltage angle of the ending bus.

[0143] Renewable energy constraint:

[0144]

[0145] In the formula, constraint (45) limits the power of renewable energy, where represents the planned power generation of the renewable energy unit at time t in the early day-ahead stage, that is, the renewable energy power in the day-ahead stage; represents the maximum power generation capacity of the renewable energy unit.

[0146] Generator set constraint:

[0147]

[0148] The real-time feasible operation capabilities of conventional thermal and combined heat and power units are restricted by (46)-(49). η represents the thermal power ratio of the combined heat and power unit.

[0149] The calculation-electricity-thermal coupling data center integrated energy system collaborative optimization method according to the embodiment of the present invention constructs a two-stage stochastic programming model by simultaneously considering the current information in the first stage and the expected information in the second stage, realizes multi-objective optimization, and transmits the optimization result to the scheduling module.

[0150] In step S140, according to the optimization result, adjust the calculation task allocation, power supply strategy, and cooling system operation parameters to form a calculation-electricity-thermal coupling data center integrated energy system collaborative optimization strategy. Specifically, it includes:

[0151] This step converts the optimal strategy generated by the optimization algorithm into actual operation instructions to ensure that the data center operates in an efficient, energy-saving, and stable state.

[0152] First, perform computational task allocation control. According to the computational task allocation scheme generated by the collaborative optimization model, dynamically adjust the task loads of each server in the data center. Through intelligent scheduling algorithms, reasonably allocate computational tasks to different server clusters to ensure that high-priority tasks (such as interactive workloads) are processed in a timely manner, and at the same time, make full use of the latency tolerance characteristics of batch workloads to achieve load balancing.

[0153] Secondly, perform power supply strategy control. According to the power supply scheme generated by the collaborative optimization model, dynamically adjust the power demand of the data center.

[0154] Finally, according to the optimization results, control the operating state of the heat exchange equipment, recover the waste heat generated by the servers and use it for other systems that require heat energy to improve the comprehensive utilization rate of energy.

[0155] In the embodiment of the present invention, this step adjusts the computational task allocation, power supply strategy, and cooling system operating parameters according to the optimization results, converts the optimal strategy generated by the optimization algorithm into actual operation instructions, and the implementation instructions ensure the efficient, energy-saving, and stable operation of the data center.

[0156] The embodiment of the present invention also provides a collaborative optimization system for an integrated energy system of a computing-electricity-thermal coupling data center, including:

[0157] An acquisition module that acquires real-time data collected by the data center.

[0158] A model construction module that quantifies the relationship between computing power, electricity, and heat based on real-time data and constructs a data center load model.

[0159] A model optimization module that constructs a collaborative optimization model based on the data center load model and solves the collaborative optimization model to obtain optimization results.

[0160] A scheduling and output module that adjusts the computational task allocation, power supply strategy, and cooling system operating parameters according to the optimization results to form a collaborative optimization strategy for the integrated energy system of the computing-electricity-thermal coupling data center.

[0161] It can be understood that the collaborative optimization system for the integrated energy system of the computing-electricity-thermal coupling data center provided by the embodiment of the present invention corresponds to the above-mentioned collaborative optimization method for the integrated energy system of the computing-electricity-thermal coupling data center. For the explanations, examples, beneficial effects, etc. of the relevant content, reference can be made to the corresponding content in the collaborative optimization method for the integrated energy system of the computing-electricity-thermal coupling data center, which will not be elaborated here.

[0162] An embodiment of the present invention also provides a computer-readable storage medium, which stores a computer program for the collaborative optimization method of the computing-electricity-thermal coupling data center integrated energy system. Among them, the computer program enables a computer to execute the collaborative optimization method of the computing-electricity-thermal coupling data center integrated energy system as described above.

[0163] An embodiment of the present application also provides an electronic device, including: one or more processors; a memory; and one or more programs, where the one or more programs are stored in the memory and configured to be executed by the one or more processors. The program includes a method for executing the collaborative optimization method of the computing-electricity-thermal coupling data center integrated energy system as described above.

[0164] In summary, compared with the prior art, the following beneficial effects are achieved:

[0165] 1. Through the collaborative optimization of coupling computing power, electricity, and heat, the flexible scheduling of energy in time and space within the data center is realized, significantly improving the overall energy efficiency, reducing the redundant energy consumption of the refrigeration system, and at the same time recovering the waste heat of the server for other systems to reduce energy waste.

[0166] 2. Build a collaborative operation strategy for the power system and the data center to reduce the impact of uncertainties such as external energy price fluctuations and load changes on the system, reduce operating costs, and improve resource scheduling efficiency.

[0167] 3. Introduce the carbon trading cost, optimize the energy scheduling plan, reduce the dependence on traditional thermal power, reduce carbon emissions, and contribute to the realization of the "dual carbon" goal.

[0168] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0169] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A collaborative optimization method for the integrated energy system of a computing-electricity-thermal coupling data center, characterized in that including: Obtain real-time data collected by the data center, where the real-time data includes load data, power consumption data, external energy price data, environmental parameter data, workload type data, and cooling system operation status data; Quantify the relationship between computing power, electricity, and heat based on the real-time data, and construct a data center load model; Construct a collaborative optimization model according to the data center load model, and solve the collaborative optimization model to obtain an optimization result; Adjust the computing task allocation, power supply strategy, and cooling system operation parameters according to the optimization result to form a collaborative optimization strategy for the computing-electricity-heat coupled data center integrated energy system.

2. The collaborative optimization method for the integrated energy system of the computing-electricity-thermal coupling data center according to claim 1, characterized in that The quantifying the relationship between computing power, electricity, and heat based on the real-time data and constructing a data center load model includes: Construct an initial data center load model in a mixed-integer non-linear form based on the real-time data collected by the data center; Obtain a linear data center load model by decomposing the initial data center load model.

3. The collaborative optimization method for the integrated energy system of the computing-electric-thermal coupling data center according to claim 2, characterized in that, The initial data center load model is: Among them, p it represents the power consumption of the data center at time t; represents the power consumption of the IT equipment with the lowest QoS in the data center at time t; represents the power consumption of the redundant IT equipment for higher QoS in the data center at time t; represents the power consumption of the cooling system of the data center at time t; represents the power consumption of other devices in the data center; represents the number of edge switches in the data center; represents the number of aggregation switches in the data center; represents the number of core switches in the data center; represents the number of servers in the data center; represents the rated power of each active edge switch in the data center; represents the rated power of each active aggregation switch in the data center; represents the rated power of each active core switch in the data center; represents the idle power of each active server in the data center; represents the peak power of each active server in the data center; represents the minimum number of active servers serving interactive workloads in the data center at time t; represents the minimum number of active servers serving batch workload q in the data center at time t; represents the number of redundant active servers serving interactive workloads in the data center at time t; represents the number of redundant active servers serving batch workload q in the data center at time t; u i represents the average service rate of the servers in the data center; λ iδt represents the conversion of the interactive workload δ allocated from FS to the data center at time t; χ iqt represents the amount of batch workload q transferred to the data center at time t; represents the number of servers serving interactive workloads operating at peak power, ignoring their workloads in the data center at time t; represents the number of servers serving batch workload q operating at peak power, ignoring their workloads in the data center at time t; represents the power consumption of the cooling system; b 1i 、b 2i represents the empirical constant of the cooling system in the data center; h it represents the thermal cooling power of the data center at time t; The linear data center load model is: where β 1it represents the power consumption baseline with the lowest QoS in the data center at time t; α 1i represents that when the power consumption of a unit of IT equipment increases, without TSTI, the total power consumption in the data center increases; represents the adjusted power demand contributed by GWB in the data center at time t compared to the set baseline; represents the adjusted power demand contributed by BWS in the data center at time t compared to the set baseline; represents the adjusted power demand contributed by TSTI in the data center at time t compared to the set baseline; The constraint conditions of the linear data center load model include: Among them, β 2i represents the difference in the energy storage level between time t and time t-1 in the data center, without heat dissipation and power generation, where the indoor temperature at time t-1 is the indoor temperature baseline; β 3i represents the energy storage level of the initial data center; β 4it represents the increase in the power consumption of the data center under the lowest QoS and without TSTI when processing the baseline amount of interactive workloads at time slot t; β 5it represents the increase in the power consumption of the data center under the lowest QoS and without TSTI when processing the baseline amount of batch workload q at time slot t; β 6it represents the number of servers not in an active state in the data center; β 7it represents the margin of IT device power in the data center at time t; β 8it The cooling power deficit in the data center at time t; β 9it The cooling power margin in the data center at time t; α 2i represents the increase in the power consumption of the data center under the lowest QoS and without TSTI when processing a unit amount of interactive workloads; α 3i represents that when the unit energy storage level at time t-1 increases, the data center energy storage level at time t also increases; α 4i represents that when reducing the unit adjusted power demand contributed by TSTI, the energy storage level of the data center at time t increases; α 5i represents that when the active servers in the data center process the maximum interactive workload, the power consumption increases and the QoS is the lowest; α 6i represents that when the active servers in the data center process the maximum batch workload, the power consumption increases and the QoS is the lowest; α 7i represents that when the unit amount of the total increased power consumption of the data center increases, the power consumption of the cooling system increases without TSTI; α 8i represents that when the unit IT device power consumption increases, the power consumption of the cooling system in the data center increases without TSTI; represents the energy storage level of the data center at time t.

4. The collaborative optimization method for the integrated energy system of the computing-electricity-thermal coupling data center according to any one of claims 1 to 3, characterized in that The constructing a collaborative optimization model according to the data center load model specifically includes: Establish an objective function with the goal of minimizing the total operating cost of the computing-electricity-heat coupled data center integrated energy system; Establish the constraint conditions of the collaborative optimization model, and the constraint conditions include: power balance constraint, heat balance constraint, electricity balance constraint, renewable energy constraint, and generator set constraint.

5. The collaborative optimization method for the integrated energy system of the computing-electricity-thermal coupling data center according to claim 4, characterized in that The objective function is: Among them, C DA represents the scheduling cost in the day-ahead stage, C RT represents the expected balancing cost in the real-time stage, C carbon represents the cost generated by carbon trading; Among them, represents the traditional thermal power cost coefficient, the cost coefficient of the combined heat and power unit, represents the thermal power in the day-ahead market, represents the thermal power in the real-time market, represents the power of the combined heat and power unit in the day-ahead market, represents the power of the combined heat and power unit in the real-time market. + represents positive regulation in the scenario, - represents negative regulation in the scenario, and γ s represents the probability of the real-time scenario, and λ C represents the price of carbon emissions. φ represents the set of DC and other variables in the system operation.

6. The collaborative optimization method for the integrated energy system of the computing-electricity-thermal coupling data center according to claim 4, wherein The power balance constraint is: Among them, represents the local load electric power in the day-ahead stage, represents the local load electric power in the real-time stage, represents the total electric power of the data center in the day-ahead stage, represents the total electric power of the data center in the real-time stage, represents the renewable energy power in the day-ahead stage, represents the renewable energy power in the real-time stage, represents the transmission line power in the day-ahead stage, represents the transmission line power in the real-time stage; The heat balance constraint is: Among them, represents the local load thermal power in the day-ahead stage, represents the local load thermal power in the real-time stage; represents the total thermal power of the data center in the day-ahead; represents the total thermal power of the data center in the real-time; The electricity balance constraint is: Among them, respectively represent the transmission line power in the pre-dispatch stage and the real-time stage, A l,n represents the power grid incidence matrix, represents the voltage angle of the starting bus, represents the voltage angle of the ending bus; The renewable energy constraint is: Among them, represents the maximum power generation capacity of the renewable energy unit; The generator set constraint is: where η represents the heat power ratio of the combined heat and power unit.

7. A collaborative optimization system for an integrated energy system of a computing-electricity-thermal coupling data center, characterized in that, including: An acquisition module that acquires real-time data collected by the data center; A model construction module that quantifies the relationship between computing power, electricity, and heat based on the real-time data and constructs a data center load model; A model optimization module that constructs a collaborative optimization model according to the data center load model and solves the collaborative optimization model to obtain an optimization result; A scheduling output module that adjusts the computing task allocation, power supply strategy, and cooling system operation parameters according to the optimization result to form a collaborative optimization strategy for the computing-electricity-heat coupled data center integrated energy system.

8. A computer-readable storage medium, characterized in that, It stores a computer program for the collaborative optimization of the computing-electricity-heat coupled data center integrated energy system, where the computer program causes the computer to execute the collaborative optimization method for the computing-electricity-heat coupled data center integrated energy system according to any one of claims 1 to 6.

9. An electronic device, characterized in that, including: One or more processors, a memory, and one or more programs, where the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the programs include those for executing the collaborative optimization method for the computing-electricity-heat coupled data center integrated energy system according to any one of claims 1 to 6.

Images