Multi-scenario shared energy storage planning method and system for multi-data center alliance

Abstract

The present invention discloses a multi-scenario shared energy storage planning method and system for a multi-data center alliance, wherein each data center belongs to an independent microgrid and shares an energy storage system. The method comprises: determining N S Typical scenarios and the probability p of each scenario s appearing s ; Construct a cost function that includes energy storage investment cost and operating cost of shared energy storage in multiple scenarios. Energy storage investment cost includes power capacity cost and energy capacity cost. Operating cost #imgabs0# Construct constraints. Energy storage capacity allocation constraints include power capacity allocation constraints #imgabs1# and energy capacity allocation constraints #imgabs2# Energy storage operation constraints are the energy storage power capacity P allocated to each data center n under scenario s. BS,n,s and energy storage capacity E BS,n,s Perform energy storage operation; solve the objective function and obtain the total capacity configuration plan for multi-scenario shared energy storage. This application introduces the multi-scenario energy storage sharing concept to maximize the shared energy storage efficiency, reduce the energy storage investment cost, and improve the economic efficiency of the data center.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy storage planning, and more specifically, relates to a multi-scenario shared energy storage planning method and system for a multi-data center alliance. Background Art

[0002] As one of the seven major areas of "new infrastructure", data centers are booming in number and scale. However, high energy consumption and high pollution have always restricted the development of data centers. Since most of the operating costs of data centers come from electricity consumption, how to reduce the growing electricity costs of data centers is a problem and challenge facing the sustainable development of data centers.

[0003] Since many places are currently implementing time-of-use electricity price policies, data centers can reduce electricity costs by configuring energy storage, utilizing the time transfer capability of energy storage, and combining the price difference of electricity in different time periods. However, due to the immaturity of energy storage technology, the current investment cost of energy storage is high. In order to improve the utilization rate of energy storage, for multiple users in the same distribution network, there are currently design solutions to optimize the distribution network by sharing energy storage. However, for multiple data centers in different microgrids, each data center corresponds to an independent microgrid. When multiple independent microgrids share an energy storage system, there is currently no solution to optimize the energy storage configuration.

[0004] Therefore, when data centers are located in different microgrids, how to reasonably configure their energy storage, reduce energy storage investment costs, and improve the economy of data centers under the condition of uncertainty of new energy and data load is an urgent problem that researchers need to solve. Summary of the invention

[0005] In response to the above defects or improvement needs of the prior art, the present invention provides a multi-scenario shared energy storage planning method and system for a multi-data center alliance, which aims to reasonably configure energy storage according to the situation where data centers are located in different microgrids to reduce energy storage investment costs and improve the economy of data centers.

[0006] To achieve the above object, according to one aspect of the present invention, a multi-scenario shared energy storage planning method for a multi-data center alliance is provided, wherein each data center belongs to an independent microgrid and shares an energy storage system, and the multi-scenario shared energy storage planning method includes:

[0007] Determine N based on historical data S Typical scenarios and the probability p of each scenario s appearing s ;

[0008] Construction includes energy storage investment cost C tz and the operating cost of shared energy storage in multiple scenarios C yxCost function, energy storage investment cost C tz Including power capacity cost and energy capacity cost, operating cost Among them, α t is the power grid price during period t, P buy,t,n,s is the amount of electricity purchased from the power grid by data center n in period t under scenario s, T is the optimized period of time-of-use electricity price, N DC is the number of the alliance’s data centers, and the objective function is to minimize the cost function;

[0009] Construct constraints, the constraints include data center energy storage capacity allocation constraints and data center energy storage operation constraints, wherein the energy storage capacity allocation constraints include: power capacity allocation constraints and energy capacity allocation constraints Among them, P BS,n,s and E BS,n,s are the energy storage power capacity and energy storage capacity allocated to data center n in any scenario s, and are the total energy storage power capacity and total energy storage capacity to be planned respectively; the energy storage operation constraint is the energy storage power capacity P allocated to each data center n under scenario s. BS,n,s and energy storage capacity E BS,n,s Perform energy storage operation;

[0010] Solve the objective function to obtain the total capacity configuration plan for multi-scenario shared energy storage.

[0011] In one embodiment, N is determined based on historical data. S Typical scenarios and the probability p of each scenario s appearing s ,include:

[0012] Step S11: Retrieving historical data of new energy output, data load, and temperature in a historical time period to obtain multiple historical scenes;

[0013] Step S12: Calculate the Euclidean distance between each historical scene, according to the objective function Determine the scenes to be deleted i , where p i For the scene i The probability of occurrence, d(ω i ,ω j ) is the scene ω i and scene j The Euclidean distance between them, S is the total number of scenes before scene reduction;

[0014] Step S13: Delete scene ω i , modify the total number of scenes S = S-1, find the deleted scene ω iThe closest scene k , the probability of the deleted scene appearing is p i Add to scene k The probability of occurrence p k superior;

[0015] Step S14: Repeat step S13 until S=N S , and get N S Typical scenarios and the probability p of each scenario s appearing s .

[0016] In one embodiment, the constraints further include:

[0017] Energy storage configuration constraints:

[0018]

[0019]

[0020]

[0021] Among them, x BS is the number of energy storage units to be planned, P is the maximum value of energy storage units allowed to be planned, BS and E BS are the power capacity and energy capacity of the energy storage unit, respectively;

[0022] Obtain the total capacity configuration plan for shared energy storage in multiple scenarios, including solving the number of energy storage units to be planned x BS .

[0023] In one embodiment,

[0024] The energy storage operation constraints include:

[0025] Energy storage system charging and discharging power constraints:

[0026] 0≤P dis,t,n,s ≤Z dis,t,n,s P BS,n,s ;

[0027] 0≤P ch,t,n,s ≤Z ch,t,n,s P BS,n,s ;

[0028] Among them, P dis,t,n,s is the charging power of the energy storage allocated to data center n in time period t under scenario s, Z dis,t,n,s is the charging status of the energy storage allocated to data center n in time period t under scenario s, P ch,t,n,s is the discharge power of the energy storage allocated to data center n in time period t under scenario s, Zch,t,n,s is the discharge state of the energy storage allocated to data center n in time period t under scenario s;

[0029] Energy storage system charging and discharging state constraints:

[0030] 0≤Z ch,t,n,s +Z dis,t,n,s ≤1;

[0031] Energy storage end energy constraint:

[0032] E BS,T,n,s =E BS,0,n,s ;

[0033] Among them, E BS,T,n,s is the final energy level of the energy storage allocated to data center n under scenario s, E BS,0,n,s is the initial energy level of the energy storage allocated to data center n under scenario s;

[0034] Energy storage energy level constraints:

[0035]

[0036] 0.1E BS,n,s ≤E BS,t,n,s ≤E BS,n,s ;

[0037] in, is the energy storage charging and discharging efficiency, E BS,t,n,s is the energy level of the energy storage allocated to data center n in period t under scenario s.

[0038] In one embodiment, the constraints further include:

[0039] Power balance constraints in data centers:

[0040] P IT,t,n,s +P cooling,t,n,s +P ch,t,n,s =P wind,t,n,s +P solar,t,n,s +P dis,t,n,s +P buy,t,n,s ;

[0041] Among them, P IT,t,n,s is the total energy consumption of IT equipment in data center n during period t under scenario s, P cooling,t,n,s is the total energy consumption of the air conditioner in the data center n during period t under scenario s, P ch,t,n,s is the discharge power of the energy storage allocated to data center n in time period t under scenario s, P wind,t,n,s is the wind power output of data center n in period t under scenario s, P solar,t,n,s is the photovoltaic power output of data center n in time period t under scenario s, Pbuy,t,n,s is the electricity purchased by data center n from the grid during period t under scenario s.

[0042] In one embodiment, the constraint condition further includes: IT equipment constraints in the data center, and the IT equipment constraints in the data center include:

[0043] Service Level Agreement Constraints:

[0044]

[0045]

[0046] Among them, q t,n,s is the batch processing load of data center n in the queue during period t under scenario s, B t,n,s is the sum of the batch loads arriving at data center n before time period t in scenario s, D t,n,s is the sum of the batch processing loads executed by data center n before time period t in scenario s, b t,n,s is the batch processing load arriving at data center n in time period t under scenario s, d t,n,s is the batch processing load executed by data center n in period t under scenario s, The maximum delay allowed for batch loads;

[0047] Data center capacity and server quantity constraints:

[0048] 0≤μ t,n,s ≤N n ;

[0049] μ t,n,s =a t,n,s +d t,n,s ;

[0050]

[0051] Among them, μ t,n,s is the data load executed by data center n in time period t under scenario s, N n is the capacity of data center n, a t,n,s is the interactive load executed by data center n in period t under scenario s, m t,i,n,s is the number of servers activated in room i of data center n during period t under scenario s, The number of servers in each computer room;

[0052] CPU utilization constraint:

[0053] u t,n,s ≤1-σ;

[0054]

[0055] Among them, u t,n,s is the CPU utilization of the server in data center n during period t under scenario s, σ is the capacity margin of the server, F is the capacity of each server, N room is the number of computer rooms in data center n;

[0056] IT equipment energy consumption constraints:

[0057] P IT,t,i,n,s =m t,i,n,s (p idle +(p peak -p idle ) t,n,s );

[0058]

[0059] Among them, p idle and p peak are the air conditioning power consumption and peak power consumption of the server, P IT,t,i,n,s is the power consumption of IT equipment in room i of data center n during period t under scenario s.

[0060] In one embodiment, the constraint condition further includes an air conditioning constraint in a computer room in a data center, and the air conditioning constraint in the computer room in a data center includes:

[0061] Temperature constraints in the equipment room:

[0062]

[0063] Among them, T in,t,i,n,s is the indoor temperature of room i in data center n at time t under scenario s, and They are the upper and lower limits of the temperature in the equipment room respectively;

[0064] Thermal dynamic model constraints:

[0065]

[0066]

[0067] Among them, T out,t,n,s is the outdoor temperature of data center n at time t under scenario s, Q t,i,n,s is the cooling capacity of room i in data center n in scenario s during period t, R and C are the equivalent impedance and equivalent heat capacity of the air conditioning load, Δt is the step length of the optimization period, Q cooling,t,i,n,s is the cooling capacity of the air conditioner in room i of data center n in scenario s during period t, It is the ratio of IT equipment heat load to IT equipment electrical load;

[0068] Energy consumption constraints of computer room air conditioners:

[0069]

[0070]

[0071]

[0072] Among them, P cooling,t,i,n,s is the air conditioning energy consumption of room i in data center n in scenario s during period t, k1, k2, l1 and l2 are constant coefficients, The maximum power consumption of the computer room air conditioner.

[0073] In one embodiment, after obtaining the total energy storage power capacity to be planned and total storage energy capacity After that, it also includes:

[0074] Calculate the power capacity cost and energy capacity cost in the energy storage investment cost;

[0075] Calculate the energy storage power capacity P allocated to each data center n under a specific scenario s BS,n,s and energy storage capacity E BS,n,s ;

[0076] Calculate the energy storage capacity power P allocated to each data center n in the current scenario s BS,n,s The proportion of the total energy storage capacity allocated to all data centers And the energy storage capacity E allocated to each data center n in the current scenario s BS,n,s The proportion of the total energy storage capacity allocated to all data centers

[0077] The power capacity cost and energy capacity cost in the energy storage investment cost are allocated according to the energy storage capacity ratio of each data center n in the current scenario s.

[0078] In one embodiment, after obtaining the total energy storage power capacity to be planned and total storage energy capacity After that, it also includes: determining the comprehensive cost of each data center n after participating in the data center alliance based on the Nash bargaining profit distribution method, and the Nash bargaining profit distribution method includes:

[0079] Determine the objective function:

[0080]

[0081] Among them, C DC,n,0 is the comprehensive cost when data center n does not participate in the alliance, CDC,n is the comprehensive cost of data center n after joining the data center alliance;

[0082] Build constraints:

[0083] C DC,n ≤C DC,n,0 ;

[0084] C DC,n =C DC,n ′-π n

[0085]

[0086]

[0087] Among them, C DC,n and C DC,n ′ is the comprehensive cost of data center n before and after bargaining in the data center alliance co-construction and sharing energy storage model, π n is the subsidy that data center n receives from the alliance after negotiation, C P,n,s and C E,n,s are the energy storage power capacity cost and energy capacity cost of data center n under scenario s, respectively.

[0088] According to another aspect of the present invention, a multi-scenario shared energy storage planning system for a multi-data center alliance is provided, wherein each data center belongs to an independent microgrid and shares an energy storage system, and the multi-scenario shared energy storage planning system comprises:

[0089] A scene determination unit is used to determine N based on historical data. S Typical scenarios and the probability p of each scenario s appearing s ;

[0090] The objective function building unit is used to build a function that includes the energy storage investment cost C tz and the operating cost of shared energy storage in multiple scenarios C yx Cost function, energy storage investment cost C tz Including power capacity cost and energy capacity cost, operating cost Among them, α t is the power grid price during period t, P buy ,t,n,s is the amount of electricity purchased from the power grid by data center n in period t under scenario s, T is the optimized period of time-of-use electricity price, N DC is the number of the alliance’s data centers, and the objective function is to minimize the cost function;

[0091] A constraint condition construction unit is used to construct constraint conditions, wherein the constraint conditions include data center energy storage capacity allocation constraints and data center energy storage operation constraints, wherein the energy storage capacity allocation constraints include: power capacity allocation constraints and energy capacity allocation constraints Among them, P BS,n,s and E BS,n,s are the energy storage power capacity and energy storage capacity allocated to data center n in any scenario s, and are the total energy storage power capacity and total energy storage capacity to be planned respectively; the energy storage operation constraint is the energy storage power capacity P allocated to each data center n under scenario s. BS,n,s and energy storage capacity E BS,n,s Perform energy storage operation;

[0092] The solving unit is used to solve the objective function and obtain a total capacity configuration plan for multi-scenario shared energy storage.

[0093] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:

[0094] This application builds an energy storage planning model with objective functions and constraints, and obtains the decision variables to be planned by solving the model. This is a conventional framework for planning and design. In this application, for specific application scenarios, that is, each data center is in an independent microgrid, and there is no transfer of energy, power, or load between microgrids, it is necessary to design specific objective functions and constraints for solution.

[0095] On the one hand, the objective function constructed by this application includes the investment cost and operating cost of the energy storage system. When constructing the operating cost, it is considered that although there is no transfer between different data centers, in different scenarios, the energy storage system with a fixed capacity can adjust the energy storage allocation of each data center according to different scenarios. Therefore, the multi-scenario shared energy storage thinking is introduced, and the weighted sum of the operating cost of each data center in each typical scenario is taken to obtain the operating cost of multi-scenario shared energy storage. In traditional technology, either the sharing in different time periods under the same scenario is considered, or multiple scenario inputs are introduced only to reflect the randomness of the scenario when optimizing the same distribution network with tidal transfer function. For different microgrids, when planning energy storage capacity, the sharing of energy storage between different scenarios has not been considered. This application uses the differences and complementarities of users' demand for energy storage resources under different operating environments to plan shared energy storage, which can maximize the efficiency of shared energy storage.

[0096] On the other hand, the constraints constructed in this application are also different from the conventional constraints in the same distribution network. The constraints in this application include data center energy storage capacity allocation constraints and data center energy storage operation constraints, among which the data center energy storage capacity allocation constraints are constraints on the total capacity of all data centers. Since this application considers multi-scenario sharing, the energy storage allocated to the data center in different scenarios is variable. Therefore, when planning, the energy storage allocation amount of each data center is not used as a decision variable, and only the total energy storage capacity is constrained. Moreover, when constraining the data center energy storage operation, its operation is subject to the energy storage capacity allocated to the corresponding data center in the current scenario, and the energy storage capacity allocated to the data center in different scenarios is different. Therefore, in different scenarios, the upper and lower limits in the data center energy storage operation constraints also change accordingly, and are not fixed values ​​in conventional designs.

[0097] In summary, this application designs a set of energy storage optimization solutions for the above-mentioned specific occasions. The multi-scenario energy storage sharing concept is introduced during the optimization. This is the first time it has been proposed in energy storage planning. Corresponding constraints are designed for the specific occasions and optimization goals. By solving the energy storage planning model, the shared energy storage efficiency can be maximized, the energy storage investment cost can be reduced, and the economy of the data center can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0098] Figure 1 A schematic diagram of a data center alliance structure according to an embodiment;

[0099] Figure 2 A flowchart of the steps of a multi-scenario shared energy storage planning method for a multi-data center alliance according to an embodiment;

[0100] Figure 3 A schematic diagram of a two-stage optimization model according to an embodiment;

[0101] Figure 4 Schematic diagram of time-of-use electricity price according to an embodiment. DETAILED DESCRIPTION

[0102] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0103] like Figure 1The diagram shows the operational framework of the shared energy storage of the data center alliance targeted by this application, which includes several data centers, shared energy storage equipment, city power, etc. Each data center belongs to an independent microgrid, and there is no power flow transfer. Each data center can be powered by self-built wind power and photovoltaic power, city power and shared energy storage, and its energy consumption mainly comes from IT equipment and computer room air conditioners. Since wind power and photovoltaic power output are intermittent and volatile, user requests from each data center are also different in time. Therefore, the supply and demand balance within each data center changes over time, and its demand for energy storage also changes accordingly. For example, in a scenario where new energy has low output and user requests are large, in order to purchase less city power during the peak electricity price period, the data center needs a larger capacity of energy storage to achieve the transfer of electricity in time; in a scenario where new energy has high output, the data center's demand for energy storage is greatly reduced. If the data center is configured with energy storage alone, it may lead to a shortage of energy storage resources in some scenarios and idle energy storage resources in some scenarios. This also provides ample opportunities for multiple data centers to jointly build and share energy storage: multiple data centers can form a data center alliance, jointly invest in the construction of energy storage equipment, and reallocate capacity based on the differences and complementarities of each data center's energy storage needs under different operating scenarios, thereby improving the utilization rate of energy storage equipment and, at the same time, improving the operating economy of each data center.

[0104] Based on this, this application designs a multi-scenario shared energy storage planning method for a multi-data center alliance, such as Figure 2 As shown, the method includes the following steps.

[0105] Step S100: Determine N based on historical data S Typical scenarios and the probability p of each scenario s appearing s .

[0106] Since this application involves energy storage sharing among multiple scenarios, multiple typical scenarios are required as input and reserve sharing among multiple scenarios is considered when building the model.

[0107] In one embodiment, based on the historical data of new energy output, data load, and temperature, the synchronous back-elimination method is used to perform scenario reduction to obtain a small number of representative scenarios and their probabilities of occurrence, and the scenario set obtained by scenario reduction is used as the scenario input for multi-scenario shared energy storage.

[0108] Specifically, the synchronous back-generation elimination method includes the following steps:

[0109] 1) Calculate the Euclidean distance between each scene and determine the scene to be deleted ω i ;

[0110]

[0111] Among them, D minis the objective function of scene reduction, p i For the scene i The probability of occurrence, d(ω i ,ω j ) is the scene ω i and scene j The Euclidean distance between them is , and S is the total number of scenes before scene reduction.

[0112] 2) Modify the total number of scenes S = S-1, find the deleted scene ω i The closest scene k , the probability of the deleted scene appearing is p i Add to scene k The probability of occurrence p k superior;

[0113] p k =p k +p i

[0114] 3) Repeat steps 1)-2) until S=N S , save N S scene data and the probability data of its occurrence.

[0115] Step S200: Construct a system including energy storage investment cost C tz and the operating cost of shared energy storage in multiple scenarios C yx Cost function, energy storage investment cost C tz Including power capacity cost and energy capacity cost, operating cost The objective function is to minimize the cost function.

[0116] With the goal of minimizing the sum of the energy storage investment cost and operating cost of the data center alliance, the influence of the new energy output, data load and temperature uncertainty of each data center is considered, and a two-stage stochastic optimization model for multi-data center shared energy storage planning is established. Figure 3 As shown, the first phase is to plan the shared energy storage of the data center alliance and determine the capacity of the shared energy storage. The second phase is to allocate and optimize the shared energy storage power capacity and energy capacity of each data center while considering the output of new energy, data load, and temperature uncertainty.

[0117] Specifically, the objective function of the two-stage stochastic optimization model for multi-data center shared energy storage planning is:

[0118]

[0119] Where γ is the discount rate, T LT is the service life of the energy storage device, κ P and κ Eare the unit power capacity cost and unit energy capacity cost of energy storage equipment, and are the shared energy storage power capacity and energy capacity invested by the data center alliance, N S The number of scenes input in the second stage, p s is the probability of scene s appearing, N DC is the number of data centers in the alliance, T is the optimization period, α t is the power grid price during period t, P buy,t,n,s This is the situation where data center n purchases electricity from the power grid during period t under scenario s.

[0120] Step S300: constructing constraint conditions, where the constraint conditions include data center energy storage capacity allocation constraints and data center energy storage operation constraints.

[0121] First, constraints can be set for energy storage configuration. Shared energy storage configuration constraints are used to determine the capacity of energy storage equipment invested by the data center alliance. If the shared energy storage capacity is too large, it may lead to low utilization of energy storage resources and low economic benefits; if the shared energy storage capacity is too small, the optimization space for data center scheduling and operation is limited. Specifically, the following constraints can be set:

[0122]

[0123]

[0124]

[0125] Among them, x BS is the planned number of energy storage units, is the maximum value of energy storage units allowed to be planned, is the planned energy storage power capacity, P BS is the power capacity of the energy storage unit, is the planned energy storage capacity, E BS is the energy capacity of the energy storage unit.

[0126] In this application, the following relevant operating constraints are set in combination with the special occasions targeted.

[0127] 1. Constraints on energy storage capacity allocation in data centers.

[0128] In the data center alliance co-construction and sharing energy storage mode, one day can be considered as a scheduling operation cycle, and the capacity of shared energy storage is allocated once in each scheduling operation cycle. Each data center can obtain the right to use energy storage with a certain power capacity and energy capacity according to its own needs. Therefore, the total power capacity and energy capacity allocated to all data centers in any scenario are constrained as follows:

[0129] 1.1 Power capacity allocation constraints:

[0130]

[0131] Among them, P BS,n,s is the energy storage power capacity allocated to data center n in scenario s.

[0132] 1.2 Energy capacity allocation constraints:

[0133]

[0134] Among them, E BS,n,s is the energy storage capacity allocated to data center n under scenario s.

[0135] The above energy storage capacity allocation constraint is a constraint on the total allocated capacity of all data centers. Because the present application involves energy storage sharing between multiple scenarios, the allocation to each data center is different in different scenarios. Therefore, the present application does not use the allocation of a single data center as a decision-making quantity.

[0136] 2. Constraints on data center energy storage operation.

[0137] This application also needs to constrain the energy storage operation state. Since each data center belongs to an independent microgrid, there is no energy transfer involved in it. The operation state of each data center is constrained by its actual allocated capacity. Therefore, the energy storage operation constraint in this application is the energy storage power capacity P allocated to each data center n under scenario s. BS,n,s and energy storage capacity E BS,n,s Carry out energy storage operation. Specifically including:

[0138] 2.1 Energy storage system charging and discharging power constraints

[0139] 0≤P dis,t,n,s ≤Z dis,t,n,s P BS,n,s

[0140] 0≤P ch,t,n,s ≤Z ch,t,n,s P BS,n,s

[0141] Among them, P dis,t,n,s is the charging power of the energy storage allocated to data center n in time period t under scenario s, Z dis,t,n,s is the charging status of the energy storage allocated to data center n in time period t under scenario s, P ch,t,n,s is the discharge power of the energy storage allocated to data center n in time period t under scenario s, Z ch,t,n,s It is the discharge status of the energy storage allocated to data center n in period t under scenario s.

[0142] 2.2 Energy storage system charging and discharging state constraints:

[0143] 0≤Z ch,t,n,s +Z dis,t,n,s ≤1

[0144] 2.3 Energy storage end energy constraint:

[0145] E BS,T,n,s =E BS,0,n,s

[0146] Among them, E BS,T,n,s is the final energy level of the energy storage allocated to data center n under scenario s, E BS,0,n,s is the initial energy level of the energy storage allocated to data center n under scenario s.

[0147] 2.4 Energy storage energy level constraints:

[0148]

[0149] 0.1E BS,n,s ≤E BS,t,n,s ≤E BS,n,s

[0150] in, is the energy storage charging and discharging efficiency, E BS,t,n,s is the energy level of the energy storage allocated to data center n in period t under scenario s.

[0151] The above data center energy storage operation constraints constrain each data center in each scenario. The upper and lower limits of the constraints depend on the allocated capacity and do not involve energy transfer between data centers.

[0152] In one embodiment, the constraints in this case also include:

[0153] 3. Power balance constraints in data centers:

[0154] P IT,t,n,s +P cooling,t,n,s +P ch,t,n,s =P wind,t,n,s +P solar,t,n,s +P dis,t,n,s +P buy,t,n,s

[0155] Among them, P IT,t,n,s is the total energy consumption of IT equipment in data center n during period t under scenario s, P cooling,t,n,s is the total energy consumption of the air conditioner in the data center n during period t under scenario s, P wind,t,n,s is the wind power output of data center n in period t under scenario s, P solar,t,n,s is the photovoltaic power output of data center n in time period t under scenario s, P buy,t,n,sis the electricity purchased by data center n from the grid during period t under scenario s.

[0156] 4. Constraints on IT equipment in the data center.

[0157] In one embodiment, the IT equipment energy consumption model considering demand response mainly adjusts the execution amount of batch loads at each moment according to the fluctuation of grid electricity prices to reduce the electricity demand during peak electricity price periods. Its constraints include service level agreement constraints, data center capacity and server quantity constraints, CPU utilization constraints and IT equipment energy consumption constraints.

[0158] Specifically include:

[0159] 4.1 Service Level Agreement Constraints:

[0160]

[0161]

[0162] Among them, q t,n,s is the batch processing load of data center n in the queue during period t under scenario s, B t,n,s is the sum of the batch loads arriving at data center n before time period t in scenario s, D t,n,s is the sum of the batch processing loads executed by data center n before time period t in scenario s, b t,n,s is the batch processing load arriving at data center n in time period t under scenario s, d t,n,s is the batch processing load executed by data center n in period t under scenario s, The maximum delay allowed for batch loads.

[0163] 4.2 Data center capacity and server quantity constraints:

[0164] 0≤μ t,n,s ≤N n

[0165] μ t,n,s =a t,n,s +d t,n,s

[0166]

[0167] Among them, μ t,n,s is the data load executed by data center n in time period t under scenario s, N n is the capacity of data center n, a t,n,s is the interactive load executed by data center n in period t under scenario s, m t,i,n,s is the number of servers activated in room i of data center n during period t under scenario s, The number of servers in each computer room.

[0168] 4.3CPU utilization constraint:

[0169] u t,n,s ≤1-σ

[0170]

[0171] Among them, u t,n,s is the CPU utilization of the server in data center n during period t under scenario s, σ is the capacity margin of the server, F is the capacity of each server, N room is the number of computer rooms in data center n.

[0172] 4.4IT equipment energy consumption constraints:

[0173] P IT,t,i,n,s =m t,i,n,s (p idle +(p peak -p idle ) t,n,s )

[0174]

[0175] Among them, p idle and p peak are the air conditioning power consumption and peak power consumption of the server, P IT,t,i,n,s is the power consumption of IT equipment in room i of data center n during period t under scenario s.

[0176] In one embodiment, the constraints in this case also include:

[0177] 5. Air conditioning constraints in computer rooms within data centers.

[0178] Different from the cooling system model represented by PUE and the empirical model obtained by fitting, the proposed computer room air conditioning energy consumption model can take into account the thermal effect of IT equipment and the impact of ambient temperature on the computer room air conditioning energy consumption, and can perform pre-cooling (large-scale cooling to reduce the computer room temperature) during off-peak electricity price periods, thereby reducing the air conditioning power consumption during peak electricity price periods.

[0179] Specifically include:

[0180] 5.1 Equipment room temperature constraints:

[0181]

[0182] Among them, T in,t,i,n,s is the indoor temperature of room i in data center n at time t under scenario s, and They are the upper and lower limits of the temperature in the equipment room.

[0183] 5.2 Thermal dynamic model constraints:

[0184]

[0185]

[0186] Among them, T out,t,n,s is the outdoor temperature of data center n at time t under scenario s, Q t,i,n,s is the cooling capacity of room i in data center n in scenario s during period t, R and C are the equivalent impedance and equivalent heat capacity of the air conditioning load, Δt is the step length of the optimization period, Q cooling,t,i,n,s is the cooling capacity of the air conditioner in room i of data center n in scenario s during period t, It is the ratio of IT equipment heat load to IT equipment electrical load.

[0187] 5.3 Energy consumption constraints of computer room air conditioning:

[0188]

[0189]

[0190]

[0191] Among them, P cooling,t,i,n,s is the air conditioning energy consumption of room i in data center n in scenario s during period t, k1, k2, l1 and l2 are constant coefficients, The maximum power consumption of the computer room air conditioner.

[0192] Step S400: Solve the objective function to obtain a total capacity configuration plan for multi-scenario shared energy storage.

[0193] Input the scenario set into the objective function and solve the objective function based on the above constraints to obtain the total capacity configuration plan of multi-scenario shared energy storage and the total energy storage power capacity to be planned. and total storage energy capacity

[0194] Furthermore, after obtaining the total capacity configuration plan for shared energy storage in multiple scenarios, it is also necessary to estimate the energy storage investment cost that each data center in the alliance needs to share. In this embodiment, the energy storage investment cost that each data center needs to share is shared in proportion to the energy storage power capacity and energy storage energy capacity used in each scenario:

[0195]

[0196]

[0197] Among them, C P,n,s and CE,n,s are the energy storage power capacity cost and energy capacity cost that data center n needs to share under scenario s, P BS,n,s and E BS,n,s They are respectively the energy storage power capacity and energy storage energy capacity allocated to data center n under scenario s.

[0198] Furthermore, after obtaining the total capacity configuration plan for multi-scenario shared energy storage, it is also necessary to estimate the comprehensive investment cost (including energy storage investment cost and operating cost) that each data center in the alliance needs to share. In this embodiment, the comprehensive cost of each data center n after participating in the data center alliance is determined based on the Nash bargaining profit distribution method. The Nash bargaining profit distribution method includes:

[0199] Determine the objective function:

[0200]

[0201] Among them, C DC,n,0 is the comprehensive cost when data center n does not participate in the alliance, C DC,n is the comprehensive cost of data center n after joining the data center alliance;

[0202] Build constraints:

[0203] C DC,n ≤C DC,n,0 ;

[0204] C DC,n =C DC,n ′-π n

[0205]

[0206]

[0207] Among them, C DC,n and C DC,n ′ is the comprehensive cost of data center n before and after bargaining in the data center alliance co-construction and sharing energy storage model, π n is the subsidy that data center n receives from the alliance after negotiation, C P,n,s and C E,n,s are the energy storage power capacity cost and energy capacity cost of data center n under scenario s, respectively.

[0208] In order to further illustrate the multi-data center shared energy storage planning method provided by the present invention, Figure 1The system shown in the figure is configured with shared energy storage, and a maximum of 2000 energy storage units are allowed to be configured. The capacity of the energy storage unit is 1kW / 6kWh, and the design service life of the energy storage is 10 years. The number of computer rooms in Data Center 1 and Data Center 2 is 12, and the number of computer rooms in Data Center 3 is 16. Each computer room is equipped with 400 servers. The wind power installed capacity of each data center is 1000kW, and the photovoltaic installed capacity is 800kW. The time-of-use electricity price is as follows Figure 4 .

[0209] Simulation results: In this embodiment, the planning scheme and the cost of each data center are shown in Table 1 (unit: yuan).

[0210] Table 1 Cost of each data center

[0211]

[0212] Accordingly, the present application also relates to a multi-scenario shared energy storage planning system for a multi-data center alliance, where each data center belongs to an independent microgrid and shares a shared energy storage system. The multi-scenario shared energy storage planning system includes:

[0213] A scene determination unit is used to determine N based on historical data. S Typical scenarios and the probability p of each scenario s appearing s ;

[0214] The objective function building unit is used to build a function that includes the energy storage investment cost C tz and the operating cost of shared energy storage in multiple scenarios C yx Cost function, energy storage investment cost C tz Including power capacity cost and energy capacity cost, operating cost Among them, α t is the power grid price during period t, P buy,t,n,s is the amount of electricity purchased from the power grid by data center n in period t under scenario s, T is the optimized period of time-of-use electricity price, N DC is the number of data centers in the alliance, and the objective function is to minimize the cost function;

[0215] The constraint condition construction unit is used to construct the constraint conditions, which include the data center energy storage capacity allocation constraint and the data center energy storage operation constraint. The energy storage capacity allocation constraint includes: power capacity allocation constraint and energy capacity allocation constraints Among them, P BS,n,s and E BS,n,s are the energy storage power capacity and energy storage capacity allocated to data center n in any scenario s, and are the total energy storage power capacity and total energy storage capacity to be planned respectively; the energy storage operation constraint is the energy storage power capacity P allocated to each data center n under scenario s. BS,n,s and energy storage capacity E BS,n,s Perform energy storage operation;

[0216] The solving unit is used to solve the objective function and obtain the total capacity configuration plan of shared energy storage in multiple scenarios.

[0217] It can be understood that the multi-scenario shared energy storage planning system of the multi-data center alliance is used to execute the multi-scenario shared energy storage planning method of the multi-data center alliance. The specific details can be referred to the above introduction and will not be repeated here.

[0218] In order to analyze the effectiveness and superiority of the proposed shared energy storage planning method, two comparison schemes are set according to the energy storage construction mode:

[0219] Solution 1: Each data center is equipped with separate energy storage;

[0220] Solution 2 (the solution proposed in this application): Data center alliance jointly builds and shares energy storage.

[0221] The simulation calculation is carried out on the MATLAB platform, and the commercial solver GUROBI is called through the YALMIP toolbox to solve the optimization problem involved, considering the optimization error of 0.01%.

[0222] The energy storage configuration results and costs under each scheme are shown in Table 2.

[0223] Table 2 Optimization results under each scheme

[0224]

[0225] It can be seen that the comprehensive cost of the data center is the lowest under Scheme 2. Although the energy storage capacity configured in the data center alliance co-construction and sharing energy storage mode is slightly more than that configured in each data center individually, its electricity purchase cost is significantly reduced, indicating that the utilization rate of energy storage is higher after the data center alliance configures shared energy storage, and the comprehensive cost of the data center is reduced.

[0226] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-scenario shared energy storage planning method for a multi-data center alliance, characterized in that: Each data center belongs to an independent microgrid and shares a shared energy storage system. The multi-scenario shared energy storage planning method includes: Determine N based on historical data S Typical scenarios and the probability p of each scenario s appearing s ; Construction includes energy storage investment cost C tz and the operating cost of shared energy storage in multiple scenarios C yx Cost function, energy storage investment cost C tz Including power capacity cost and energy capacity cost, operating cost Among them, α t is the power grid price during period t, P buy,t,n,s is the amount of electricity purchased from the power grid by data center n in period t under scenario s, T is the optimized period of time-of-use electricity price, N DC is the number of the alliance’s data centers, and the objective function is to minimize the cost function; Construct constraints, including data center energy storage capacity allocation constraints, data center energy storage operation constraints, and power balance constraints within the data center; the energy storage capacity allocation constraints include: power capacity allocation constraints and energy capacity allocation constraints Among them, P BS,n,s and E BS,n,s are the energy storage power capacity and energy storage capacity allocated to data center n in any scenario s, and are the total energy storage power capacity and total energy storage capacity to be planned respectively; the energy storage operation constraint is the energy storage power capacity P allocated to each data center n under scenario s. BS,n,s and energy storage capacity E BS,n,s Perform energy storage operation; Solving the objective function to obtain a total capacity configuration plan for multi-scenario shared energy storage; in, Determine N based on historical data S Typical scenarios and the probability p of each scenario s appearing s ,include: Step S11: Retrieving historical data of new energy output, data load, and temperature in a historical time period to obtain multiple historical scenes; Step S12: Calculate the Euclidean distance between each historical scene, according to the objective function Determine the scenes to be deleted i , where p i For the scene i The probability of occurrence, d(ω i ,ω j ) is the scene ω i and scene j The Euclidean distance between them, S is the total number of scenes before scene reduction; Step S13: Delete scene ω i , modify the total number of scenes S = S-1, find the deleted scene ω i The closest scene k , the probability of the deleted scene appearing is p i Add to scene k The probability of occurrence p k superior; Step S14: Repeat step S13 until S=N S , and get N S Typical scenarios and the probability p of each scenario s appearing s ; The power balance constraints in the data center include: P IT,t,n,s +P cooling,t,n,s +P ch,t,n,s =P wind,t,n,s +P solar,t,n,s +P dis,t,n,s +P buy,t,n,s ; Among them, P IT,t,n,s is the total energy consumption of IT equipment in data center n during period t under scenario s, P cooling,t,n,s is the total energy consumption of the air conditioner in the data center n during period t under scenario s, P ch,t,n,s is the discharge power of the energy storage allocated to data center n in time period t under scenario s, P wind,t,n,s is the wind power output of data center n in period t under scenario s, P solar,t,n,s is the photovoltaic power output of data center n in time period t under scenario s, P buy,t,n,s is the electricity purchased by data center n from the grid during period t under scenario s; The data center energy storage operation constraints include: Calculate the power capacity cost and energy capacity cost in the energy storage investment cost; Calculate the energy storage power capacity P allocated to each data center n under a specific scenario s BS,n,s and energy storage capacity E BS,n,s ; Calculate the energy storage capacity power P allocated to each data center n in the current scenario s BS,n,s The proportion of the total energy storage capacity allocated to all data centers And the energy storage capacity E allocated to each data center n in the current scenario s BS,n,s The proportion of the total energy storage capacity allocated to all data centers The power capacity cost and energy capacity cost in the energy storage investment cost are allocated according to the energy storage capacity ratio of each data center n in the current scenario s.

2. The multi-scenario shared energy storage planning method according to claim 1, characterized in that: The constraints also include: Energy storage configuration constraints: Among them, x BS is the number of energy storage units to be planned, P is the maximum value of energy storage units allowed to be planned, BS and E BS are the power capacity and energy capacity of the energy storage unit, respectively; Obtain the total capacity configuration plan for shared energy storage in multiple scenarios, including solving the number of energy storage units to be planned x BS .

3. The multi-scenario shared energy storage planning method according to claim 1, characterized in that: The energy storage operation constraints include: Energy storage system charging and discharging power constraints: 0≤P dis,t,n,s ≤Z dis,t,n,s P BS,n,s ; 0≤P ch,t,n,s ≤Z ch,t,n,s P BS,n,s ; Among them, P dis,t,n,s is the charging power of the energy storage allocated to data center n in time period t under scenario s, Z dis,t,n,s is the charging status of the energy storage allocated to data center n in time period t under scenario s, P ch,t,n,s is the discharge power of the energy storage allocated to data center n in time period t under scenario s, Z ch,t,n,s is the discharge state of the energy storage allocated to data center n in time period t under scenario s; Energy storage system charging and discharging state constraints: 0≤Z ch,t,n,s +Z dis,t,n,s ≤1; Energy storage end energy constraint: AND BS,T,n,s =And BS,0,n,s ; Among them, E BS,T,n,s is the final energy level of the energy storage allocated to data center n under scenario s, E BS,0,n,s is the initial energy level of the energy storage allocated to data center n under scenario s; Energy storage energy level constraints: 0.1E BS,n,s ≤E BS,t,n,s ≤E BS,n,s ; in, is the energy storage charging and discharging efficiency, E BS,t,n,s is the energy level of the energy storage allocated to data center n in period t under scenario s.

4. The multi-scenario shared energy storage planning method according to claim 1, characterized in that: The constraint conditions also include: IT equipment constraints in the data center, and the IT equipment constraints in the data center include: Service Level Agreement Constraints: Among them, q t,n,s is the batch processing load of data center n in the queue during period t under scenario s, B t,n,s is the sum of the batch loads arriving at data center n before time period t in scenario s, D t,n,s is the sum of the batch processing loads executed by data center n before time period t in scenario s, b t,n,s is the batch processing load arriving at data center n in time period t under scenario s, d t,n,s is the batch processing load executed by data center n in period t under scenario s, The maximum delay allowed for batch loads; Data center capacity and server quantity constraints: 0≤μ t,n,s ≤N n ; m t,n,s =a t,n,s +d t,n,s ; Among them, μ t,n,s is the data load executed by data center n in time period t under scenario s, N n is the capacity of data center n, a t,n,s is the interactive load executed by data center n in period t under scenario s, m t,i,n,s is the number of servers activated in room i of data center n during period t under scenario s, The number of servers in each computer room; CPU utilization constraints: in t,n,s ≤1-s; Among them, u t,n,s is the CPU utilization of the server in data center n during period t under scenario s, s is the capacity margin of the server, F is the capacity of each server, and N room is the number of computer rooms in data center n; IT equipment energy consumption constraints: P IT,t,i,n,s =m t,i,n,s (p idle +(p peak -p idle )u t,n,s ); Among them, p idle and p peak are the air conditioning power consumption and peak power consumption of the server, P IT,t,i,n,s is the power consumption of IT equipment in room i of data center n during period t under scenario s.

5. The multi-scenario shared energy storage planning method according to claim 1, characterized in that: The constraint conditions also include air conditioning constraints in the computer room of the data center, and the air conditioning constraints in the computer room of the data center include: Temperature constraints in the equipment room: Among them, T in,t,i,n,s is the indoor temperature of room i in data center n at time t under scenario s, and They are the upper and lower limits of the temperature in the equipment room respectively; Thermal dynamic model constraints: Among them, T out,t,n,s is the outdoor temperature of data center n at time t under scenario s, Q t,i,n,s is the cooling capacity of room i in data center n in scenario s during period t, R and C are the equivalent impedance and equivalent heat capacity of the air conditioning load, △t is the step length of the optimization period, Q cooling,t,i,n,s is the cooling capacity of the air conditioner in room i of data center n in scenario s during period t, It is the ratio of IT equipment heat load to IT equipment electrical load; Energy consumption constraints of computer room air conditioners: Among them, P cooling,t,i,n,s is the air conditioning energy consumption of room i in data center n in scenario s during period t, k1, k2, l1 and l2 are constant coefficients, The maximum power consumption of the computer room air conditioner.

6. The multi-scenario shared energy storage planning method according to claim 1, characterized in that: After obtaining the total energy storage power capacity to be planned and total storage energy capacity After that, it also includes: determining the comprehensive cost of each data center n after participating in the data center alliance based on the Nash bargaining profit distribution method, and the Nash bargaining profit distribution method includes: Determine the objective function: Among them, C DC,n,0 is the comprehensive cost when data center n does not participate in the alliance, C DC,n is the comprehensive cost of data center n after joining the data center alliance; Build constraints: C DC,n ≤C DC,n,0 ; C DC,n =C DC,n ′-π n Among them, C DC,n and C DC,n ′ is the comprehensive cost of data center n before and after bargaining in the data center alliance co-construction and sharing energy storage model, π n is the subsidy that data center n receives from the alliance after negotiation, C P,n,s and C E,n,s are the energy storage power capacity cost and energy capacity cost of data center n under scenario s, respectively.

7. A multi-scenario shared energy storage planning system for a multi-data center alliance, characterized in that: Used to execute the multi-scenario shared energy storage planning method according to any one of claims 1 to 6, each data center belongs to an independent microgrid and shares an energy storage system, and the multi-scenario shared energy storage planning system includes: A scene determination unit is used to determine N based on historical data. S Typical scenarios and the probability p of each scenario s appearing s ; The objective function building unit is used to build a function that includes the energy storage investment cost C tz and the operating cost of shared energy storage in multiple scenarios C yx Cost function, energy storage investment cost C tz Including power capacity cost and energy capacity cost, operating cost Among them, α t is the power grid price during period t, P buy,t,n,s is the amount of electricity purchased from the power grid by data center n in period t under scenario s, T is the optimized period of time-of-use electricity price, N DC is the number of the alliance’s data centers, and the objective function is to minimize the cost function; A constraint condition construction unit is used to construct constraint conditions, wherein the constraint conditions include data center energy storage capacity allocation constraints and data center energy storage operation constraints, wherein the energy storage capacity allocation constraints include: power capacity allocation constraints and energy capacity allocation constraints Among them, P BS,n,s and E BS,n,s are the energy storage power capacity and energy storage capacity allocated to data center n in any scenario s, and are the total energy storage power capacity and total energy storage capacity to be planned respectively; the energy storage operation constraint is the energy storage power capacity P allocated to each data center n under scenario s. BS,n,s and energy storage capacity E BS,n,s Perform energy storage operation; The solving unit is used to solve the objective function and obtain a total capacity configuration plan for multi-scenario shared energy storage.

Images