Construction method of active distribution network planning model considering data centers

Abstract

The present invention discloses a method for constructing an active distribution network planning model taking into account a data center, including: obtaining basic parameters; using a two-stage random optimization method to establish an active distribution network planning model taking into account a data center; substituting the basic parameters into the model, with the goal of minimizing the investment and operation cost of the active distribution network while improving the utilization rate of renewable energy and reducing carbon emissions, using an improved group search optimization algorithm to solve the model, and outputting the selected model of the transformation line, the installation location of the smart meter, the installation location of the wind turbine, the number of servers started in each period of the data center, the task migration amount in the time dimension of the data center, the power purchase of the distribution network, the actual output of the wind turbine, the energy storage charging power, and the energy storage discharging power scheme. The present invention takes into account the transferable load in the time dimension of the data center and the flexible charging and discharging of the energy storage equipment equipped, which can effectively reduce the cost of active distribution network planning, promote distributed energy consumption, and reduce carbon emissions.

Description

Technical Field

[0001] The present invention relates to the field of power systems, and in particular to a method, device and computing equipment for constructing an active distribution network planning model taking into account a data center. Background Art

[0002] In recent years, the industrial Internet and the digital revolution have driven the construction of a new generation of power systems, and the demand for big data and cloud computing has exploded. In the past five years, the growth rate of China's data center (DC) construction has remained at around 20%, with annual electricity consumption accounting for more than 2% of the country's power generation and contributing about 0.3% of global carbon emissions. With the development of communication technology, the number of data processing tasks in data centers has continued to increase, causing the power consumption of data centers to rise rapidly.

[0003] Active Distribution Network (ADN) is a feasible technical solution for effectively regulating and using distributed renewable energy resources (DER).

[0004] In the existing technology, in order to reduce the power consumption of data centers, many research results have been achieved in the energy management and optimized operation of data centers. For example, the coordinated scheduling of data centers, energy storage and electric vehicles has effectively reduced the operating costs of data centers by formulating corresponding optimization strategies. Another example is the real-time energy management method of data centers that comprehensively considers factors such as data load, server sleep, coordinated operation of multiple energy storage, and interaction with active distribution networks. However, in the current joint planning method of data centers and active distribution networks, the energy planning effect is not ideal, resulting in high energy consumption. Summary of the invention

[0005] To this end, the present invention provides a device and a computing apparatus to try to solve or at least alleviate the above problems.

[0006] According to one aspect of the present invention, a method for constructing an active distribution network planning model taking into account a data center is provided, which is suitable for execution in a computing device, and the method comprises the steps of: obtaining basic parameters; establishing an active distribution network planning model taking into account a data center using a two-stage stochastic optimization method, wherein the model comprises an objective function and constraints; substituting the basic parameters into the model, and solving the model using an improved group search optimization algorithm with the goal of minimizing the investment and operation cost of the active distribution network while improving the utilization rate of renewable energy and reducing carbon emissions, and outputting the selected model of the transformation line, the installation location of the smart meter, the installation location of the wind turbine, the number of servers started in each period of the data center, the amount of task migration in the time dimension of the data center, the power purchase of the distribution network, the actual output of the wind turbine, the energy storage charging power, and the energy storage discharging power formulation plan; wherein the objective function is:

[0007] minC=C INV +C OPT

[0008] C INV =C line +C SM +C WG

[0009] C OPT =C grid +C DER +C DC-DR

[0010] In the formula, C INV represents the investment cost in the planning stage, C OPT represents the investment cost in the operation phase, C line represents the transmission line expansion cost of the data center, C SM represents the installation cost of smart meters in data centers, C WG represents the installation cost of the wind turbine in the data center, C grid represents the cost of active distribution network purchasing electricity from the upper grid, C DER represents the maintenance cost of energy storage equipment, C DC-DR represents the demand response incentive cost paid by the active distribution network to the data center.

[0011] Optionally, the objective function includes:

[0012]

[0013]

[0014]

[0015]

[0016]

[0017]

[0018] In the formula, δ Line ,δ SM ,δ WG Respectively represent the annual factors of the data center's transmission lines, smart meters, and wind turbine investment, Ω Line ,Ω M ,Ω DC ,Ω WG They represent the set of lines to be transformed, the set of line models, the set of active distribution network nodes including data centers, and the set of nodes where wind turbines are installed. represents the unit length reconstruction cost of line type m, Indicates the feeder length, represents the state variable, c SM represents the installation cost of a single smart meter, χ i represents the state of smart meter installation in the i-th data center, P i WG represents the installed capacity of wind turbines in the ith data center, c WG-inv represents the installation cost of wind turbines in the system, represents the 0-1 decision variable for installing wind turbines, α represents the number of typical days in a year, and ρ s represents the expected probability of scene s, Ω T ,Ω S They represent the time period set and scene set respectively, c buy It represents the unit price of electricity purchased by the data center from the active distribution network. represents the amount of electricity purchased from the main grid in time period t under scenario s, Δt represents the unit dispatching period, c WG-opt represents the maintenance cost per unit power of the fan unit, represents the actual output of the wind turbine, c DR represents the demand response price per unit data capacity, represents the batch processing load of the data center numbered k in scenario s that migrates from time period t to time period t′, where the load is the number of data tasks to be processed, σ ch , σ dis They represent the loss costs corresponding to charging and discharging of power equipment in the data center, They respectively represent the charging and discharging power of the uninterruptible power supply of the data center.

[0019] Optionally, the constraints include: one or more of: power constraints, data center load response constraints, data center load delay processing constraints, data center transmission line modification and equipment installation constraints, data center voltage constraints, adjacent transmission line flow constraints, and data center power equipment constraints.

[0020] Optionally, the power constraint includes a server power constraint, and the server power constraint includes:

[0021] Total power constraints for data center operations:

[0022]

[0023] Server power consumption constraints in data centers:

[0024]

[0025] The server opens the quantity constraint in real time:

[0026]

[0027] CPU utilization constraints when the server is working:

[0028]

[0029] In the formula, Indicates the total power of the data center operation, Indicates the power consumption of the server. Indicates the power consumption of the refrigeration equipment. Indicates the power consumption of other load devices. represents the actual number of servers required to handle the load, which represents the data processing tasks to be performed in the data center, Indicates the static power consumption of the server in idle state. represents the power consumption when the server is fully loaded, f s,t,k represents the total amount of processing load, μ k Indicates the amount of data that a single server can handle. represents the number of servers, σ max Indicates the maximum CPU utilization of the server.

[0030] Optionally, the power constraints also include power balance constraints of data centers, interactive power constraints of data centers and active distribution networks, cross-power balance constraints of data centers and active distribution networks, and power constraints sent to the active distribution network from the upper power grid:

[0031] Data center power balance constraints:

[0032]

[0033] Interaction power constraints between data centers and active distribution networks:

[0034]

[0035] Power balance constraints between data center and active distribution network:

[0036]

[0037] The power constraints of the active distribution network sent from the upper power grid are:

[0038]

[0039] In the formula, Indicates the power purchased by the main distribution network, Ω N represents the active distribution network node set, P ji,s,t represents the active power flowing from node j to node i in time period t under scenario s, Indicates the actual output of the wind turbine, P ik,s,t represents the active power flowing from node i to node k in time period t under scenario s, Indicates the total power of the data center operation, represents the active power of other loads connected to node i except the data center at time t, Q ji,s,t represents the reactive power flowing from node j to node i in time period t under scenario s, Represents the reactive power output of the wind turbine, Q ik,s,t represents the reactive power flowing from node i to node k in time period t under scenario s, represents the reactive power of other loads connected to node i except the data center at time t, represents the transmission power between the data center and the active distribution network, represents the maximum transmission power, represents the total power of the data center at node k in time period t under scenario s, Indicates the charging power of the data center power supply equipment. represents the transmission power between the data center and the distribution network at node k in time period t under scenario s, Indicates the discharge power of the data center energy storage equipment, They respectively represent the minimum and maximum values ​​of the interaction power between the active distribution network and the upper-level power grid.

[0040] Optionally, the power constraint also includes wind turbine operating power constraint and line transmission power constraint:

[0041] Power constraints during wind turbine operation:

[0042]

[0043]

[0044] Line transmission power constraints:

[0045]

[0046]

[0047] In the formula, Indicates the actual output of the wind turbine. represents the predicted value of wind turbine output, Indicates the power factor angle, Indicates the reactive power output of the wind turbine. They represent the maximum capacity limit of the lth line, P l,s,t represents the active power transmitted by line l, Q l,s,t Represents the reactive power transmitted by line l.

[0048] Optionally, the data center load response constraints include:

[0049] The proportion of load that a data center can handle delayed is constrained by:

[0050]

[0051] The delay load constraint that a data center needs to handle at any time period t is:

[0052]

[0053] Data center load scheduling constraints:

[0054]

[0055] Load migration constraints:

[0056]

[0057] The total load constraint of the data center at any time:

[0058]

[0059] In the formula, represents the batch processing load of the kth data center at the initial time under scenario s, ζ represents the constant of the proportion of batch processing load in all loads, and f s,t,k,0 Indicates the amount of load that needs to be processed at the initial moment, represents the total amount of data that needs to be processed by the kth data center in time period t under scenario s, represents the batch processing load in the data center numbered k under scenario s that migrates from period t′ to period t, represents the batch processing load in the data center numbered k under scenario s that migrates from period t to period t', χ k represents the state of smart meter installation in the kth data center, f s,t,k It represents the batch processing load that migrates from period t to period t' in the data center numbered k under scenario s.

[0060] Optionally, data center power transmission line modification and equipment installation constraints include:

[0061] Line transformation selection model constraints:

[0062]

[0063] Constraints on the number of smart meters installed:

[0064] 0≤χ k ≤1

[0065] Constraints on the number of nodes for installing wind turbines:

[0066]

[0067] In the formula, Indicates the model of the line to be modified, Ω M represents the line model set of the lines to be selected, χ k represents the state of smart meter installation in the kth data center, Ω WG Represents the node set where wind turbines are installed. Indicates the installation location of the wind turbine, N WG Indicates the maximum number of nodes that the system allows to install power devices.

[0068] Optionally, the data center voltage constraint, the adjacent transmission line power flow constraint, and the data center power equipment constraint are respectively:

[0069] Data center voltage constraints:

[0070]

[0071] Power flow constraints of adjacent transmission lines:

[0072]

[0073] Data center power equipment constraints:

[0074]

[0075] In the formula, They represent the minimum and maximum voltage allowed in data center i, respectively. s,t,i represents the voltage value of node i in time period t under scenario s, U s,t,j represents the voltage value of node j in time period t under scenario s, P l,s,t , Q l,s,t Respectively represent the active power and reactive power transmitted on line l, Ω M Indicates the line model set to be selected. Indicates the model selected for the line that needs to be modified. Respectively represent the resistance and reactance of line l before transformation, R l,m , X l,m They represent the resistance and reactance of line l after transformation, They represent the charging and discharging state variables of the data center power supply equipment in time period t, E s,t,k represents the stored power of the power supply equipment in the kth data center in time period t, E s,t-1,k represents the storage power of the power supply equipment in the kth data center in time period t-1, η C , η D They represent the charging power and discharging efficiency of the power supply equipment respectively, △t represents the unit scheduling period, Indicates the discharge power of the power supply device. Indicates the charging power of the power supply device, P Emax Indicates the maximum charge and discharge power of the power supply device. Indicates the charge status of the data center power supply equipment. Indicates the capacity of the data center power supply equipment. Respectively represent the maximum and minimum values ​​of the power device's charge state.

[0076] Optionally, the basic parameters include: the active distribution network topology framework, the length of each line of the active distribution network, the impedance value per unit line length, the power load value of each node in a typical day, the data demand of the communication system in a typical day, the main network electricity purchase price, the rated installed capacity of a single wind turbine, the unit cost of the wind turbine, the maintenance price of the wind turbine, the daily output prediction curve of wind power generation, the amount of data that a single server in the data center can process, the installation cost of a single smart meter, the silent power consumption value of the server, the full load power consumption value of the server, the number of servers in a single data center, the maximum CPU utilization rate of the server, the demand response subsidy price, the rated capacity of the energy storage device installed in the data center, the maximum charging power of the energy storage device, the maximum discharging power of the energy storage device, the maximum charging and discharging efficiency of the energy storage device, the maximum and minimum charge states of the energy storage device, the resistance of the optional line, the reactance of the optional line, the current carrying capacity of the optional line, and the unit length price of the optional line. One or more.

[0077] According to one aspect of the present invention, there is provided a device for constructing an energy hub model taking into account renewable energy and demand response, which is suitable for execution in a computing device, and comprises: a parameter acquisition module, which is suitable for acquiring basic parameters; a model construction unit, which uses a two-stage stochastic optimization method to establish an active distribution network planning model taking into account a data center, and the model comprises an objective function and constraints; a model solving unit, which is suitable for substituting basic parameters into the model, and taking the active distribution network investment and operation cost as the minimum while improving the utilization rate of renewable energy and reducing carbon emissions as the goal, using an improved group search optimization algorithm to solve the model, and outputting the selected model of the transformation line, the installation location of the smart meter, the installation location of the wind turbine, the number of servers started in each period of the data center, the task migration amount in the time dimension of the data center, the power purchase of the distribution network, the actual output of the wind turbine, the energy storage charging power, and the energy storage discharging power formulation plan, wherein the objective function is:

[0078] minC=C INV +C OPT

[0079] C INV =C line +C SM +C WG

[0080] C OPT =Cg rid +C DER +C DC -DR

[0081] In the formula, C INV represents the investment cost in the planning stage, C OPT represents the investment cost in the operation phase, C line represents the transmission line expansion cost of the data center, C SM represents the installation cost of smart meters in data centers, C WG represents the installation cost of the wind turbine in the data center, C grid represents the cost of active distribution network purchasing electricity from the upper grid, C DER represents the maintenance cost of energy storage equipment, C DC-DR represents the demand response incentive cost paid by the active distribution network to the data center.

[0082] According to one aspect of the present invention, there is provided a computing device, comprising: at least one processor; and a memory storing program instructions, wherein the program instructions are configured to be suitable for being executed by the at least one processor, and the program instructions include instructions for executing the method as described above.

[0083] According to one aspect of the present invention, a readable storage medium storing program instructions is provided. When the program instructions are read and executed by a computing device, the computing device executes the method described above.

[0084] According to the technical solution of the present invention, with the goal of minimizing the investment and operation cost of the active distribution network while improving the utilization rate of renewable energy and reducing carbon emissions, the time-dimension transferable load of the data center and the flexible charging and discharging of the energy storage equipment are combined with the active distribution network, and a two-stage stochastic programming model is established. In the first stage, the plan for line upgrades, distributed power installation locations, and configuration of smart meters (SM) is optimized and determined with the goal of minimizing the system planning investment cost. In the second stage, the sum of the power purchase cost, demand response cost, and DER maintenance cost is minimized. Under the premise of meeting the user's information needs, considering the time-dimension transferable load of the data center and the flexible charging and discharging of the energy storage equipment, the active distribution network planning cost can be effectively reduced, the distributed energy consumption can be promoted, and carbon emissions can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 A schematic diagram of an active power distribution network system 100 according to an embodiment of the present invention is shown;

[0086] Figure 2 shows a structural block diagram of a computing device 200 according to an embodiment of the present invention;

[0087] Figure 3 A flow chart of a method 300 for constructing an active distribution network planning model taking data centers into account according to an embodiment of the present invention is shown;

[0088] Figure 4 A flowchart of solving the model by a hybrid swarm search algorithm of an opposition learning algorithm and a differential evolution algorithm is shown;

[0089] Figure 5 A schematic diagram shows a structural diagram of a device 500 for constructing an active distribution network planning model taking into account a data center according to an embodiment of the present invention;

[0090] Figure 6 A schematic diagram of an IEEE-33 node distribution network including an information domain is shown;

[0091] Figure 7 A schematic diagram showing the data flow of a data center in each period;

[0092] Figure 8 A schematic diagram showing the power load curve for each time period is shown;

[0093] Fig. 9 A schematic diagram of wind power abandonment balance in case 1 is shown;

[0094] Fig.10 A schematic diagram of wind power abandonment balance in case 4 is shown;

[0095] Fig.11 A schematic diagram showing the operation optimization results of each data center in case 4;

[0096] Fig.12 A schematic diagram showing transceiver scheduling results when data centers flexibly interact;

[0097] Fig.13 A schematic diagram showing the number of servers started up sequentially in any data center under different latency requirements. DETAILED DESCRIPTION

[0098] As a major carbon emitter, the power industry shoulders an important historical mission, and the planning of a new power system is an important prerequisite for leading the low-carbon development and transformation of the power system. It can be foreseen that the structural form of the power system will shift from a high-carbon power system to a deep low-carbon or zero-carbon power system.

[0099] In recent years, the industrial Internet and the digital revolution have driven the construction of a new generation of power systems, and the demand for big data, cloud computing, etc. has exploded. In the past five years, the growth rate of data center construction has remained at around 20%, and the annual electricity consumption accounts for more than 2% of the country's power generation, contributing about 0.3% of the world's carbon emissions. With the development of communication technology, the number of data processing tasks in data centers continues to increase, causing the power consumption of data centers to show a rapid upward trend.

[0100] Active Distribution Network (ADN) is considered to be a feasible technical solution for effectively regulating and utilizing distributed renewable energy resources (DER) with randomness and intermittency. However, introducing a high proportion of DER in the distribution system will bring great challenges to the power balance of the grid.

[0101] At present, a large number of achievements have been made in the research on AND planning issues to promote DER consumption. For example, considering the uncertainty of intermittent distributed generation (DG) planning layout, output power and load, a two-layer scenario planning model of ADN is established to promote the efficient utilization of intermittent distributed generation; another example is the three-layer planning model of active distribution network (ADN) that considers the interests of distribution companies, DG operators and users, and analyzes the relationship between each layer to coordinate the interests of "source", "network" and "load" and promote the optimal utilization of resources; another example is the uncertainty time series set with a certain degree of conservatism for DG and load, and the scenario screening rules are designed. Secondly, the hierarchical robust planning model of active distribution network is constructed, and the optimization goal is to minimize wind abandonment, solar abandonment and load loss, and then the investment layer and operation layer of the planning model are effectively associated. However, in the above studies, data centers are not taken into consideration.

[0102] As an important facility supporting the digital economy, in order to reduce the power consumption of data centers, many experts and scholars at home and abroad have made many achievements in the energy management and optimization of data centers. For example, the coordinated scheduling of data centers, energy storage and electric vehicles has effectively reduced the operating costs of data centers by formulating corresponding optimization strategies. Another example is the real-time energy management method of data centers that comprehensively considers factors such as data load, server sleep, coordinated operation of multiple energy storage, and interaction with active distribution networks. However, the current joint planning method of data centers and active distribution networks does not take into account the role of flexible demand response resources in data centers, and the energy planning effect is not ideal, resulting in high energy consumption.

[0103] Figure 1 A schematic diagram of an active distribution network system 100 composed of a data center and an active distribution network according to an embodiment of the present invention is shown. The active distribution network system 100 includes an active distribution network 110, a wind turbine 120 and a data center 130, and the active distribution network 110 is communicatively connected with the wind turbine 120 and the data center 130, respectively. There can be multiple active distribution networks 110 (respectively 1101, 1102, ..., 110n), each of which serves as a node of the system 100, there can be multiple wind turbines 120 (respectively 1201, 1202, ..., 120n), there can be multiple data centers 130 (respectively 1301, 1302, ..., 130n), each active distribution network may be installed with a wind turbine, or may not be installed. Figure 1 The system 100 shown is a situation where a wind turbine is installed at each active power distribution network node, that is, at this time, the number of wind turbines corresponds to the number of active power distribution networks.

[0104] Each data center 130 includes one or more servers, power supply equipment, and smart meters. The server is suitable for processing data processing tasks, the power supply equipment is suitable for powering the server and storing excess power in each active distribution network, and the smart meter is suitable for monitoring the power consumption of the data center. The above-mentioned servers, power supply equipment, and smart meters can be selected according to actual conditions, and the present invention does not limit this. For example, the power supply equipment can be an uninterruptible power supply (UPS).

[0105] In order to solve the problem that the energy planning effect of the existing technology is not ideal, the present invention provides a method for constructing an active distribution network planning model taking into account the data center based on the active distribution network system 100. In the process of constructing the model, the flexibility of the data center is fully considered, the migration of the entire data center load (the load is the data task to be processed by the data center) is reasonably optimized, and the active distribution network and the power supply equipment of the data center are reasonably optimized to perform charging and discharging, so as to minimize the investment and operation cost of the active distribution network system while improving the absorption of renewable energy and promoting carbon emission reduction.

[0106] The flexibility of the data center described in the present invention refers to the flexibility of the interaction between the data center and the active power distribution network, which is mainly reflected in two aspects.

[0107] On the one hand, data center operators can evaluate the urgency of task processing requirements of data user loads, fully tap the potential for time shifting of data loads, shift non-immediate processing loads, and change data load processing time, thereby causing time shifting of energy flow.

[0108] On the other hand, data centers can also mobilize their own power supply equipment to interact flexibly with the active distribution network, and participate in the optimization of power system operation through flexible charging and discharging. Specifically: in the active distribution network system, when the load of the active distribution network is at a low point or the output of renewable energy is high, data center operators can deploy power supply equipment for charging, thereby promoting the consumption of renewable energy and the realization of dual carbon goals. When the load of the active distribution network system is at a typical intraday peak, the power supply equipment is used to discharge the active distribution network, reducing the pressure on the distribution network and ensuring the quality of power.

[0109] The method for constructing an active distribution network planning model taking into account a data center provided by the present invention is suitable for execution in a computing device. Figure 2 FIG. 2 shows a block diagram of a computing device 200 according to an embodiment of the present invention. Figure 2As shown, in a basic configuration 202, computing device 200 typically includes a system memory 206 and one or more processors 204. A memory bus 208 may be used for communication between processor 204 and system memory 206.

[0110] Depending on the desired configuration, the processor 204 can be any type of processor, including but not limited to: a microprocessor (μP), a microcontroller (μC), a digital information processor (DSP), or any combination thereof. The processor 204 can include one or more levels of cache such as a primary cache 210 and a secondary cache 212, a processor core 214, and registers 216. An example processor core 214 can include an arithmetic logic unit (ALU), a floating point unit (FPU), a digital signal processing core (DSP core), or any combination thereof. An example memory controller 218 can be used with the processor 204, or in some implementations, the memory controller 218 can be an internal part of the processor 204.

[0111] Depending on the desired configuration, system memory 206 may be any type of memory, including but not limited to: volatile memory (such as RAM), non-volatile memory (such as ROM, flash memory, etc.), or any combination thereof. System memory 206 may include an operating system 220, one or more applications 222, and program data 224. In some implementations, application 222 may be arranged to operate on the operating system using program data 224.

[0112] The computing device 200 further includes a storage device 232, which includes a removable storage 236 and a non-removable storage 238, and both the removable storage 236 and the non-removable storage 238 are connected to the storage interface bus 234. In the present invention, data related to each event occurring during program execution and time information indicating the occurrence of each event can be stored in the storage device 232, and the operating system 220 is suitable for managing the storage device 232. The storage device 232 can be a disk.

[0113] The computing device 200 may also include an interface bus 240 that facilitates communication from various interface devices (e.g., output devices 242, peripheral interfaces 244, and communication devices 246) to the basic configuration 202 via the bus / interface controller 230. Example output devices 242 include an image processing unit 248 and an audio processing unit 250. They can be configured to facilitate communication with various external devices such as a display or speakers via one or more A / V ports 252. Example peripheral interfaces 244 may include a serial interface controller 254 and a parallel interface controller 256, which may be configured to facilitate communication with external devices such as input devices (e.g., keyboards, mice, pens, voice input devices, touch input devices) or other peripherals (e.g., printers, scanners, etc.) via one or more I / O ports 258. Example communication devices 246 may include a network controller 260, which may be arranged to facilitate communication with one or more other computing devices 262 via a network communication link via one or more communication ports 264.

[0114] A network communication link can be an example of a communication medium. Communication media can generally be embodied as computer-readable instructions, data structures, program modules in a modulated data signal such as a carrier wave or other transmission mechanism, and can include any information delivery medium. A "modulated data signal" can be a signal in which one or more of its data sets or its changes can be carried out in a manner that encodes information in the signal. As non-limiting examples, communication media can include wired media such as a wired network or a dedicated line network, and various wireless media such as sound, radio frequency (RF), microwave, infrared (IR) or other wireless media. The term computer-readable medium used herein can include both storage media and communication media.

[0115] The computing device 200 can be implemented as a server, such as a file server, a database server, an application server, and a WEB server, etc., and can also be implemented as a part of a small-sized portable (or mobile) electronic device, such as a cellular phone, a personal digital assistant (PDA), a personal media player device, a wireless network browsing device, a personal head-mounted device, an application-specific device, or a hybrid device that can include any of the above functions. The computing device 200 can also be implemented as a personal computer including a desktop computer and a notebook computer configuration. In some embodiments, the computing device 200 is configured to execute the method 300 according to the present invention.

[0116] Figure 3 A schematic diagram of a method 300 for constructing an active distribution network planning model taking into account a data center according to an embodiment of the present invention is shown. The method is suitable for residing in Figure 2The method 300 is executed in the computing device 200 shown. The model includes an objective function and constraints. The method 300 includes steps S310 to S330.

[0117] In step S310, basic parameters are obtained. Basic data is used as input data for the model. Basic parameters include: active distribution network topology framework, length of each line of active distribution network, impedance value per unit line length, power load value of each node in a typical day, data demand of communication system in a typical day, main network electricity purchase price, rated installed capacity of a single wind turbine, unit cost of wind turbine, maintenance price of wind turbine, daily output forecast curve of wind power generation, amount of data that a single server in a data center can process, installation cost of a single smart meter, silent power consumption value of server, full load power consumption value of server, number of servers in a single data center, maximum CPU utilization of server, demand response subsidy price, rated capacity of energy storage equipment installed in data center, maximum charging power of energy storage equipment, maximum discharge power of energy storage equipment, maximum charge and discharge efficiency of energy storage equipment, maximum and minimum charge state of energy storage equipment, resistance of optional line, reactance of optional line, current carrying capacity of optional line and price per unit length of optional line.

[0118] The typical day mentioned above can be the spring equinox, summer solstice, autumnal equinox and winter solstice in a year. The typical day of the communication system is a day in a year, such as June 1. The active distribution network topology framework, such as the IEEE-33 node distribution network topology structure, such as Figure 6 The power load values ​​of each node in a typical day are shown in Table 1.

[0119] Table 1 Power load values ​​of each node in a typical day

[0120]

[0121] Then, in step S320, a two-stage stochastic optimization method is used to establish an active distribution network planning model taking into account the data center, where the input of the known model includes an objective function and constraints.

[0122] The objective function is:

[0123] minC=C INV +C OPT

[0124] C INV =C line +C SM +C WG

[0125] C OPT =C grid +C DER +C DC-DR

[0126] In the formula, C INV represents the investment cost in the planning stage, C OPT represents the investment cost in the operation phase, C line represents the transmission line expansion cost of the data center, C SM represents the installation cost of smart meters in data centers, C WG represents the installation cost of the wind turbine in the data center, C grid represents the cost of active distribution network purchasing electricity from the upper grid, C DER represents the maintenance cost of energy storage equipment, C DC-DR represents the demand response incentive cost paid by the active distribution network to the data center.

[0127] Among them, C line , C SM , C WG , C grid , C DER and C DC-DR They are:

[0128]

[0129]

[0130]

[0131]

[0132]

[0133]

[0134] In the formula, δ Line ,δ SM ,δ WG Respectively represent the annual factors of transmission lines, smart meters, and wind turbine investment, Ω Line ,Ω M ,Ω DC ,Ω WG They represent the set of lines to be transformed, the set of line models, the set of active distribution network nodes including data centers, and the set of nodes where wind turbines are installed. represents the unit length reconstruction cost of line type m, Indicates the feeder length, represents the state variable, c SM represents the installation cost of a single smart meter, χ i represents the state of smart meter installation in the i-th data center, P i WG represents the installed capacity of wind turbines in the ith data center, cWG-inv represents the installation cost of wind turbines in the system, represents the 0-1 decision variable for installing wind turbines, α represents the number of typical days in a year, and ρ s represents the expected probability of scene s, Ω T ,Ω S They represent the time period set and scene set respectively, c buy It represents the unit price of electricity purchased by the data center from the active distribution network. represents the amount of electricity purchased from the main grid in time period t under scenario s, Δt represents the unit dispatching period, c WG-opt represents the maintenance cost per unit power of the fan unit, represents the actual output of the wind turbine, c DR represents the demand response price per unit data capacity, represents the batch processing load of the data center numbered k in scenario s that migrates from time period t to time period t′. The load is the number of data tasks to be processed, σ ch , σ dis They represent the loss costs corresponding to charging and discharging of energy storage equipment in the data center, They respectively represent the charging and discharging power of the uninterruptible power supply of the data center.

[0135] According to one embodiment of the present invention, the constraints include: one or more of power constraints, data center load response constraints, data center load delay processing constraints, data center transmission line modification and equipment installation constraints, data center voltage constraints, adjacent transmission line flow constraints, and data center power equipment constraints.

[0136] 1. Power constraints are essentially constraints on data center power-related content, including server power constraints, data center power balance constraints, data center and active distribution network interaction power constraints, data center and active distribution network cross power balance constraints, active distribution network power constraints sent from the upper power grid, wind turbine operation power constraints, and line transmission power constraints.

[0137] 1) Server power constraints include the total power constraints of the data center operation, the power consumption constraints of the servers in the data center, the number of servers turned on in real time, and the CPU utilization constraints when the servers are working, which are:

[0138] The total power constraint for data center operation is:

[0139]

[0140] The power consumption of servers in a data center is related to the number of servers turned on and the data processing status of the servers. When the server processes data tasks, the power consumption of the server increases with the increase of data tasks processed. When there are no data tasks waiting to be processed, the server only consumes the static power required for operation. Therefore, the power consumption constraint of the server in the data center is:

[0141]

[0142] When performing data processing operations in a data center, the number of servers that are enabled in real time must meet certain constraints. The constraints on the number of servers that are enabled in real time are:

[0143]

[0144] Considering that the CPU utilization of the server should not be greater than a specific limit during actual operation, the CPU utilization constraint when the server is working is:

[0145]

[0146] In the formula, Indicates the total power of the data center operation, Indicates the power consumption of the server. Indicates the power consumption of the refrigeration equipment. Indicates the power consumption of other load devices. Indicates the actual number of servers required to handle the load. The load represents the data processing tasks to be performed in the data center. Indicates the static power consumption of the server in idle state. represents the power consumption when the server is fully loaded, f s,t,k represents the total amount of processing load, μ k Indicates the amount of data that a single server can handle. represents the number of servers, σ max Indicates the maximum CPU utilization of the server.

[0147] 2) The power balance constraint of the data center, the interactive power constraint between the data center and the active distribution network, the cross-power balance constraint between the data center and the active distribution network, and the power constraint sent to the active distribution network from the upper power grid are:

[0148] According to the law of conservation of energy, each node in the system 100 should satisfy power balance, so the power balance constraint of the data center is:

[0149]

[0150] For the data center in the system 100, as a separate energy body, in the process of interacting with the active distribution network, it is necessary to meet the constraints of the interactive power with the active distribution network and the real-time balance of the internal power. Therefore, the interactive power constraints between the data center and the active distribution network are:

[0151]

[0152] The power balance constraint between the data center and the active distribution network is:

[0153]

[0154] The power constraint of the active distribution network sent from the upper power grid is:

[0155]

[0156] In the formula, Indicates the power purchased by the main distribution network, Ω N represents the active distribution network node set, P ji,s,t represents the active power flowing from node j to node i in time period t under scenario s, Indicates the actual output of the wind turbine, P ik,s,t represents the active power flowing from node i to node k in time period t under scenario s, Indicates the total power of the data center operation, represents the active power of other loads connected to node i except the data center at time t, Q ji,s,t represents the reactive power flowing from node j to node i in time period t under scenario s, Represents the reactive power output of the wind turbine, Q ik,s,t represents the reactive power flowing from node i to node k in time period t under scenario s, represents the reactive power of other loads connected to node i except the data center at time t, represents the transmission power between the data center and the active distribution network, represents the maximum transmission power, represents the total power of the data center at node k in time period t under scenario s, Indicates the charging power of the data center energy storage equipment, represents the transmission power between the data center and the distribution network at node k in time period t under scenario s, Indicates the discharge power of the data center energy storage equipment, They respectively represent the minimum and maximum values ​​of the interaction power between the active distribution network and the upper-level power grid.

[0157] 3) When the installed wind turbine is actually in operation, its active output should not exceed the predicted output value. Assuming that the power factor of the wind turbine is constant during operation, the power constraints of the wind turbine and the line transmission power constraints during operation are:

[0158] The power constraint of wind turbines during operation is:

[0159]

[0160]

[0161] The line transmission power constraint is:

[0162]

[0163]

[0164] In the formula, Indicates the actual output of the wind turbine. represents the predicted value of wind turbine output, Indicates the power factor angle, Indicates the reactive power output of the wind turbine. They represent the maximum capacity limit of the lth line, P l,s,t represents the active power transmitted by line l, Q l,s,t Represents the reactive power transmitted by line l.

[0165] 2. The load response constraints of the data center include the proportion of the load that can be delayed in the data center, the delayed load constraint that the data center needs to handle in any period t, the load scheduling constraint of the data center, the load migration constraint and the total load constraint of the data center at any time, which are:

[0166] In a data center, the types of data requested for processing include immediate processing and deferrable types. For deferrable batch processing loads such as big data computing and data analysis, the proportion constraint of the deferrable processing load in the data center is:

[0167]

[0168] Assuming that the load can be migrated to the time period after this moment in the scheduling cycle, that is, the batch processing load at time t can be migrated to the period [t+1,T] for processing, then the delay load constraint that the data center needs to process in any period t is:

[0169]

[0170] During the data processing, the real-time transferable load of each data center is regulated by the configured smart meter, and the load scheduling constraints of the data center are satisfied as follows:

[0171]

[0172] At the same time, the total amount of data load that can be migrated in the entire data center should not exceed the total amount of data that can be migrated. Then the load migration amount constraint is:

[0173]

[0174] The total amount of data arriving at the DC numbered k at a certain moment is the sum of the interactive load and the batch load. Then the total load constraint of the data center at any moment is:

[0175]

[0176] In the formula, represents the batch processing load of the kth data center at the initial time under scenario s, ζ represents the constant of the proportion of batch processing load in all loads, and f s,t,k,0 Indicates the amount of load that needs to be processed at the initial moment, represents the total amount of data that needs to be processed by the kth data center in time period t under scenario s, represents the batch processing load in the data center numbered k under scenario s that migrates from period t′ to period t, represents the batch processing load in the data center numbered k under scenario s that migrates from period t to period t', χ k represents the state of smart meter installation in the kth data center, f s,t,k It represents the batch processing load migrated from time period t to time period t' in the data center numbered k under scenario s, that is, the number of batch processing data processing tasks migrated from time period t to time period t' in the data center.

[0177] 3. Transmission line transformation and equipment installation constraints for data centers include line transformation selection model constraints, smart meter installation quantity constraints, and wind turbine installation node quantity constraints (i.e., active distribution network quantity constraints for wind turbine installation):

[0178] During the line transformation and upgrade process, at most one model can be selected for line transformation, and the model selection constraints are:

[0179]

[0180] Each DC can install at most one smart meter. The number of smart meters to be installed is constrained as follows:

[0181] 0≤χ k ≤1

[0182] The number of nodes where wind turbines are installed in the system 100 should be less than the allowed number of nodes. Then the number of nodes where wind turbines are installed is constrained as follows:

[0183]

[0184] In the formula, Indicates the model of the line to be modified, Ω M represents the line model set of the lines to be selected, χ k represents the state of smart meter installation in the kth data center, Ω WG Represents the node set where wind turbines are installed. Indicates the installation location of the wind turbine, N WG Indicates the maximum number of nodes that the system allows to install power devices.

[0185] 4. The data center voltage constraints, adjacent transmission line power flow constraints, and data center power equipment constraints are:

[0186] To ensure the safe operation of the active distribution network containing the data center, the voltage of each node should be maintained within a certain range. Then the voltage constraint of the data center is:

[0187]

[0188] The power flow constraints of adjacent transmission lines are:

[0189]

[0190] The power supply equipment configured in the data center can participate in the active distribution network interaction as an energy storage device under the premise of meeting the power supply reliability. It needs to meet the constraints of singleness of charge and discharge transition, battery power conversion, battery maximum charge and discharge, battery state of charge, etc. The constraints of the power supply equipment in the data center are:

[0191]

[0192] In the formula, They represent the minimum and maximum voltage allowed in data center i, respectively. s,t,i represents the voltage value of node i in time period t under scenario s, U s,t,j represents the voltage value of node j in time period t under scenario s, P l,s,t , Q l,s,t Respectively represent the active power and reactive power transmitted on line l, Ω M Indicates the line model set to be selected. Indicates the model selected for the line that needs to be modified. Respectively represent the resistance and reactance of line l before transformation, R l,m , X l,m They represent the resistance and reactance of line l after transformation, They represent the charging and discharging state variables of the data center power supply equipment in time period t, Es,t,k represents the stored power of the power supply equipment in the kth data center in time period t, E s,t-1,k represents the storage power of the power supply equipment in the kth data center in time period t-1, η C , η D They represent the charging power and discharging efficiency of the power supply equipment respectively, △t represents the unit scheduling period, Indicates the discharge power of the power supply device. Indicates the charging power of the power supply device, P Emax Indicates the maximum charge and discharge power of the power supply device. Indicates the charge status of the data center power supply equipment. Indicates the capacity of the data center power supply equipment. Respectively represent the maximum and minimum values ​​of the power device's charge state.

[0193] After establishing the active distribution network planning model taking the data center into account, continue to execute step S330, substitute the basic parameters into the model, that is, use the basic parameters as the input data of the model, and take minimizing the investment and operation cost of the active distribution network while improving the utilization rate of renewable energy and reducing carbon emissions as the goal, use the improved group search optimization algorithm to solve the model, and output the selected model of the transformation line, the installation location of the smart meter, the installation location of the wind turbine, the number of servers started in each period of the data center, the task migration amount in the time dimension of the data center, the power purchase amount of the distribution network, the actual output of the wind turbine, the energy storage charging power and the energy storage discharging power.

[0194] From the above content, it can be seen that the input of the model is the basic parameters, as mentioned above, and will not be repeated here. The output of the model is the selected model of the transformation line, the installation location of the smart meter, the installation location of the wind turbine, the number of servers turned on in each period of the data center, the task migration amount in the time dimension of the data center, the power purchase of the distribution network, the actual output of the wind turbine, the energy storage charging power, and the energy storage discharge power formulation plan.

[0195] From the above content, we can also know that the goal of the model is to minimize the investment and operation costs of the active distribution network while increasing the consumption of renewable energy and promoting carbon emission reduction. The above model objectives are only general. The following is a detailed description of the model objectives:

[0196] The objectives of the model may include two objectives: the investment phase objective and the operation phase objective of the active distribution network system 100, which is equivalent to constructing the model using a two-stage planning method. In the planning phase of the active distribution network system 100, the goal is to minimize the system investment cost, optimize and determine the line upgrade, wind turbine installation location and configuration of smart meters. In the operation phase of the active distribution network system 100, the goal is to minimize the sum of the power purchase cost, demand response cost and DER maintenance cost, and reasonably optimize the information load migration and the charging and discharging status of the power supply equipment under the premise of meeting the user data requirements.

[0197] It should be understood that there are a variety of methods for solving the model, and the present invention is not limited to the specific implementation method. All methods that can solve the above model are within the protection scope of the present invention. According to one embodiment, the present invention uses a hybrid group search algorithm based on opposition learning and differential evolution to solve the above model. The hybrid group search algorithm based on opposition learning and differential evolution in the present invention combines the differential evolution algorithm (Differential evolution algorithm DE) and the opposition learning algorithm (OBL) into the group search optimizer (Group Search Optimizer, GSO).

[0198] The process of solving the above model by using a hybrid swarm search algorithm of the anti-learning algorithm and the differential evolution algorithm is as follows: Figure 4 As shown, the following steps are included:

[0199] 1) First, randomly initialize the population P, set the population size to N, the maximum number of iterations Tmax and the iteration counter t.

[0200] 2) Calculate the fitness of each individual.

[0201] 3) Randomly select 0.3N individuals from population P to construct subpopulation SP1, and randomly select 0.4N individuals from the remaining 0.7 population P to construct subpopulation SP2.

[0202] 4) The remaining 0.3N individuals are subjected to the group search optimization algorithm to generate population SP3.

[0203] 5) Apply OBL to population SP1 to generate opposition-based population OBP, combine SP1 and OBP, and sort the individuals in the population after combining SP1 and OBP in descending order of fitness values, select half of the individuals in order of fitness values ​​from high to low, and construct population P1.

[0204] 6) Apply DE to SP2 to generate a differential evolution population P2 of size 0.4N.

[0205] 7) The remaining 0.3N individuals are subjected to the group search optimization algorithm to generate the overall population P3, where P3 = 0.3N.

[0206] 8) Combine populations P1, P2, and P3 to form the next group, and add 1 to the iteration counter to obtain an updated iteration counter.

[0207] 9) Determine whether the updated iteration counter is greater than the maximum number of iterations Tmax. If not, return to step 2). If greater, end the solution process. That is, if the updated iteration counter is less than the maximum number of iterations Tmax, continue to execute steps 2) to 9). If greater, it means that the maximum number of iterations has been reached, and end the solution process.

[0208] Figure 5 FIG. 5 shows a structural block diagram of a device 500 for constructing an active distribution network planning model taking into account a data center according to an embodiment of the present invention. The device 500 may reside in a computing device 100. Figure 5 As shown, the device 500 includes: a parameter acquisition unit 510, a model construction unit 520 and a model solving unit 530.

[0209] A parameter acquisition unit 510, adapted to acquire basic parameters;

[0210] A model building unit 520 uses a two-stage stochastic optimization method to establish an active distribution network planning model taking into account the data center, the model including an objective function and constraints;

[0211] The model solving unit 530 is suitable for substituting basic parameters into the model, and taking the investment and operation cost of the active distribution network as the minimum while improving the utilization rate of renewable energy and reducing carbon emissions as the goal, using the improved group search optimization algorithm to solve the model, output the selected model of the transformation line, the installation location of the smart meter, the installation location of the wind turbine, the number of servers started in each period of the data center, the task migration amount of the internal time dimension of the data center, the power purchase of the distribution network, the actual output of the wind turbine, the energy storage charging power and the energy storage discharging power. Among them, the objective function is: As mentioned above, the steps here are redundant.

[0212] It should be noted that the working principle of the construction device 500 for the active distribution network planning model taking into account the data center is similar to the above-mentioned construction method 300 for the active distribution network planning model taking into account the data center. For relevant details, please refer to the description of the above-mentioned construction method 300 for the active distribution network planning model taking into account the data center, which will not be repeated here.

[0213] The following will use a specific case to verify the active distribution network planning model considering data centers constructed by the present invention and perform numerical simulation. Figure 1The active power distribution network system 100 shown in FIG. Figure 6 As shown in the figure, the present invention selects an IEEE-33 node active distribution network with information domain for example analysis. The voltage level is 10kV. The active distribution network includes 32 load nodes. Node 1 is a transformer node connected to the main grid. The time-sharing change curve of data demand load is shown in Figure 7 As shown in Figure 2, the power load curve of the active distribution network is as follows: Figure 8 shown.

[0214] The present invention assumes that the main grid electricity price is 0.38 yuan / (kWh), and the power factor of each node load is the same, which is 0.90. The installed capacity of a single wind turbine is 800kW, the unit cost of the wind turbine is 7000 yuan / kW, the maintenance cost of the wind turbine is 0.029 yuan / (kWh), and the daily output prediction curve of wind power generation (derived by substituting the predicted wind speed into the wind power output formula). The amount of data that a single server can process is 500 / s, and the installation cost of a single smart meter is c SM The static power consumption of the server in idle state is 10,000 yuan. Power consumption when the server is fully loaded The power consumption of the data center is 150W and 300W respectively. The total number of data center servers is 1000, and the maximum value of server CPU utilization is 0.9. The demand response subsidy price cDR is 0.1 yuan / Gbps, the rated capacity of a single data center energy storage battery is 1000kW·h, the maximum charging and discharging power is 200kW, the charging and discharging efficiency is set to 0.85, and the maximum / minimum state of charge SOC is 90% and 10% respectively. The relevant parameters of the available feeders are shown in Table 2.

[0215] Table 2 Parameters of the optional lines

[0216]

[0217]

[0218] We selected 1,000 typical daily data load forecasting scenarios, and used the Monte Carlo sampling method and K-means clustering method to reduce the data load scenarios to 10. In addition, the time to meet the user data loading response requirements was set to 100ms.

[0219] Based on the above parameter settings, the active distribution network planning problem considering data center demand response is optimized. In order to verify the impact of different data center participation in demand response and distribution network interaction modes on the planning results, the two indicators of time-transferable load proportion and whether the power supply equipment of the data center participates in the interaction are changed. The mode settings of the data center participating in demand response are shown in Table 3.

[0220] Table 3 Data center settings for different working conditions

[0221] situation Time-shiftable load ratio Does energy storage participate in the interaction? 1 0 no 2 10％ no 3 0 yes 4 10％ yes

[0222] The calculation results under different working modes are shown in Table 4. Comparing Situation 2 with Situation 1, it can be seen that when considering the time-transferable data load, the total system cost and investment cost decreased by 78,000 yuan and 86,000 yuan respectively, and the operating cost increased. This is because by reasonably adjusting the transferable data load in the time dimension, the time-transferable data load is delayed during the peak power consumption period through the data collected by the smart meter, and the power shortage of the power flow is reduced by delaying the data processing of the information flow, thereby reducing the requirements for the line's carrying capacity and reducing the line investment cost. At the same time, the data load that needs to be delayed can be processed during the high-generation period of wind power, thereby promoting the efficient use of DER, reducing the main grid purchase of electricity, and effectively reducing carbon emissions.

[0223] Comparing Situation 3 and Situation 1, it can be seen that by flexibly calling the energy storage devices equipped in the data center and participating in the operation of the active distribution network, the total system cost and planning cost can be reduced. Although the operating cost has increased by 53,000 yuan, the DER consumption has increased by 13.45% compared with Situation 1. This is because DER has the characteristics of anti-peak regulation. By flexibly calling the energy storage devices in the data center, the energy storage devices are charged and controlled during the high-occurrence period of DER, and the stored electric energy is transmitted to the active distribution network during the peak period of power load, thereby increasing the consumption of DER.

[0224] Table 4 Optimization results of different working conditions

[0225]

[0226]

[0227] Among them, C INV-Line represents the line investment cost, C INV-SM represents the cost of installing smart meters, C INV-RES represents the cost of investing in renewable energy, C grid represents the cost of active distribution network purchasing electricity from the upper grid, C DER represents the maintenance cost of distributed generation, C DC-DR represents the demand response incentive cost paid by the active distribution network to the data center, C INV represents the total planning cost after annualization, C OPT Represents the total cost of operation.

[0228] Scenario 4 is a comprehensive call for data load transfer in the time dimension and power equipment to participate in grid interaction. By comparison, it can be seen that the total system cost is reduced by 67,000 yuan compared with Scenario 1, and the total cost is not much different from that of Scenario 2. This is because the demand response costs for data load delay processing and calling power equipment in Scenario 4 are higher than those in Scenario 2. At the same time, Scenario 4 increases the consumption of DER and reduces the main grid's electricity purchase. Analysis shows that the call of the two types of data center demand response resources in Scenario 4 can reduce the cost of active distribution network line planning, improve DER utilization, and obtain better low-carbon benefits.

[0229] In order to analyze the impact of data center participation in interactive response on the benefits of active distribution network, case 1 and case 4 are selected for comparison and their dispatching and operation schemes are analyzed. The system power balance diagrams under the two cases are shown in the following figure: Fig. 9 and Fig.10 As shown. Fig. 9 and Fig.10 It can be seen that compared with the situation 1 where the data center does not participate in demand response, the data center's participation in interaction can effectively improve the system load curve, play a role in peak load reduction and valley filling, and improve the contradiction between power supply and demand during peak and valley load periods. At the same time, the flexible interaction of the data center significantly reduces the abandoned wind power, and greatly increases the amount of wind power connected to the grid during the peak wind power generation period.

[0230] Fig.11 The operation results of case 4 data center participating in demand response are given. Fig.11 It can be seen that after the data center participates in the interaction, during the peak hours of electricity load (10-13, 15-18), the time-transferable data load in these periods will be transferred and processed, and delayed to the high-incidence period of DER, thereby changing the energy consumption of the data center, balancing the load curve, and promoting DER absorption. In addition, the power supply equipment is charged during the high-incidence period of wind power generation, which increases the system's absorption of DER. During the low-incidence period of wind power generation, the power storage equipment can be used to supply power to the data center or the power grid, which can reduce the amount of power purchased by the active distribution network from the external market, thereby reducing carbon emissions on the power generation side.

[0231] To further analyze the transfer of data load in the time dimension, we select node 13 and node 30 data center nodes for analysis. Fig.12 As shown in the figure, during the peak period of power load, the data center, under the premise of meeting the rigid data load, delays the processing of the load that can be transferred in the time dimension to the period with lower data processing demand (0-7 o'clock). From the comparison before and after, it can be inferred that the higher the proportion of transferable load, the better the effect of data center participating in grid demand response.

[0232] After that, a sensitivity analysis is performed. The present invention sets different delay times between 5ms and 100ms, and optimizes the solution when other parameters are set unchanged in case 4 of Table 1. In order to reflect the seasonal characteristics of the system, four typical days in spring, summer, autumn and winter are taken for solution, and the change curve of the number of server startups in each period under different delays is obtained, as shown in Figure 1. Fig.13 shown.

[0233] Depend on Fig.13 It can be seen that when the delay requirement increases from 10ms to 100ms, the number of servers turned on first decreases and then remains unchanged as the delay increases, and the reduction in the number of servers turned on also decreases. Limited by the maximum utilization of the server, when the data load delay requirement is 22ms, the number of servers turned on drops to the minimum, and when the delay is increased, the number of servers turned on no longer changes. This derivation result also further verifies the validity of the simulation results.

[0234] From the above analysis, we can see that the delay requirements of data users can affect the power regulation effect and demand response capability of data centers. Therefore, the active distribution network can reasonably adjust the delay requirements of data loads within the range allowed by users, thereby achieving the purpose of reducing the power consumption of data centers, and providing a new idea for more fully tapping the demand response potential of data centers.

[0235] This paper analyzes the use of data centers as flexible resources in active distribution networks, deeply explores the operating characteristics of data centers in the communication domain, takes into account the energy consumption characteristics of data centers and the load transfer process in the time dimension, and proposes a method for constructing an active distribution network planning model taking into account data centers. By migrating the data load time dimension in the information domain and calling the energy storage device to charge and discharge, the power flow in the energy domain is improved, the utilization rate of renewable energy is increased, and carbon emission reduction is promoted. Through example analysis, the conclusions are as follows:

[0236] 1) The new load data center can participate in the active distribution network demand response as a flexible resource. By calling the transferable data load and data center power equipment in the time dimension, the active distribution network source-grid-load coordinated planning can be realized to achieve optimal economic efficiency. At the same time, it can promote DER consumption and reduce carbon emissions.

[0237] 2) The energy consumption characteristics of the data center are related to the number of servers that are turned on and the data load that needs to be processed. By installing smart meters in the data center to obtain processing load information, the amount of data information in each time period can be flexibly scheduled. While meeting the data delay request, the time domain distribution of the information flow can be changed, thereby adjusting the energy consumption of the data center and participating in the active distribution network demand response.

[0238] 3) The relationship between the planning and operation stages is taken into consideration, and the resulting decision-making solution has good practical value.

Claims

1. A method for constructing an active distribution network planning model taking into account a data center, suitable for execution in a computing device, the method comprising the steps of: Get basic parameters, including: The active distribution network topology framework, the length of each line of the active distribution network, the impedance value per unit line length, the power load value of each node in a typical day, the data demand of the communication system in a typical day, the main network electricity purchase price, the rated installed capacity of a single wind turbine, the unit cost of a wind turbine, the maintenance price of a wind turbine, the daily output prediction curve of wind power generation, the amount of data that a single server in a data center can process, the installation cost of a single smart meter, the silent power consumption value of the server, the full load power consumption value of the server, the number of servers in a single data center, the maximum CPU utilization rate of the server, the demand response subsidy price, the rated capacity of the energy storage equipment installed in the data center, the maximum charging power of the energy storage equipment, the maximum discharging power of the energy storage equipment, the maximum charging and discharging efficiency of the energy storage equipment, the maximum and minimum charge states of the energy storage equipment, the resistance of the optional line, the reactance of the optional line, the current carrying capacity of the optional line, and the price per unit length of the optional line. A two-stage stochastic optimization method is used to establish an active distribution network planning model considering data centers. The model includes objective functions and constraints. Substitute the basic parameters into the model, and aim to minimize the investment and operation cost of the active distribution network while improving the utilization rate of renewable energy and reducing carbon emissions. Use an improved group search optimization algorithm to solve the model, output the selected model of the transformation line, the installation location of the smart meter, the installation location of the wind turbine, the number of servers powered on in each period of the data center, the amount of task migration in the time dimension of the data center, the power purchase of the distribution network, the actual output of the wind turbine, the energy storage charging power, and the energy storage discharging power. The improved group search optimization algorithm combines the differential evolution algorithm and the anti-learning algorithm into the group search optimization algorithm; Wherein, the objective function is: minC=C INV +C OPT C INV =Cline+C SM +C WG C OPT =Cgrid+C DER +C DC -DR In the formula, C INV represents the investment cost in the planning stage, C OPT represents the investment cost in the operation phase, C line represents the transmission line expansion cost of the data center, C SM represents the installation cost of smart meters in data centers, C WG represents the installation cost of the wind turbine in the data center, C grid represents the cost of active distribution network purchasing electricity from the upper grid, C DER represents the maintenance cost of energy storage equipment, C DC-DR represents the demand response incentive cost paid by the active distribution network to the data center; The constraints include: one or more of power constraints, data center load response constraints, data center load delay processing constraints, data center transmission line modification and equipment installation constraints, data center voltage constraints, adjacent transmission line flow constraints, and data center power supply equipment constraints.

2. The method of claim 1, wherein: The objective function includes: In the formula, δ Line ,δ SM ,δ WG Respectively represent the annual factors of the data center's transmission lines, smart meters, and wind turbine investment, Ω Line ,Ω M ,Ω DC ,Ω WG They represent the set of lines to be transformed, the set of line models, the set of active distribution network nodes including data centers, and the set of nodes where wind turbines are installed. represents the unit length reconstruction cost of line type m, Indicates the feeder length, represents the state variable, c SM represents the installation cost of a single smart meter, χ i represents the state of smart meter installation in the i-th data center, P i WG represents the installed capacity of wind turbines in the ith data center, c WG-inv represents the installation cost of wind turbines in the system, represents the 0-1 decision variable for installing wind turbines, α represents the number of typical days in a year, and ρ s represents the expected probability of scene s, Ω T ,Ω S They represent the time period set and scene set respectively, c buy It represents the unit price of electricity purchased by the data center from the active distribution network. represents the amount of electricity purchased from the main grid in time period t under scenario s, Δt represents the unit dispatching period, c WG-opt represents the maintenance cost per unit power of the fan unit, represents the actual output of the wind turbine, c DR represents the demand response price per unit data capacity, represents the batch processing load of the data center numbered k in scenario s that migrates from time period t to time period t', where the batch processing load is the number of data tasks to be processed, σ ch , σ dis They represent the loss costs corresponding to charging and discharging of power equipment in the data center, They respectively represent the charging and discharging power of the uninterruptible power supply of the data center.

3. The method of claim 1, wherein: The power constraint includes a server power constraint, and the server power constraint includes: Total power constraints for data center operations: Server power consumption constraints in data centers: The server opens the quantity constraint in real time: CPU utilization constraints when the server is working: In the formula, Indicates the total power of the data center operation, Indicates the power consumption of the server. Indicates the power consumption of the refrigeration equipment. Indicates the power consumption of other load devices. represents the actual number of servers required to handle the load, which represents the data processing tasks to be performed in the data center, Indicates the static power consumption of the server in idle state. represents the power consumption when the server is fully loaded, f s,t,k represents the total amount of processing load, μ k Indicates the amount of data that a single server can handle. represents the number of servers, σ max Indicates the maximum CPU utilization of the server.

4. The method of claim 1, wherein: The power constraints also include power balance constraints of data centers, interactive power constraints between data centers and active distribution networks, cross-power balance constraints between data centers and active distribution networks, and power constraints sent to the active distribution network from the upper power grid: Data center power balance constraints: Interaction power constraints between data centers and active distribution networks: Power balance constraints between data center and active distribution network: The power constraints of the active distribution network sent from the upper power grid are: In the formula, Indicates the power purchased by the main distribution network, Ω N represents the active distribution network node set, P ji,s,t represents the active power flowing from node j to node i in time period t under scenario s, Indicates the actual output of the wind turbine, P ik,s,t represents the active power flowing from node i to node k in time period t under scenario s, Indicates the total power of the data center operation, represents the active power of other loads connected to node i except the data center at time t, Q ji,s,t represents the reactive power flowing from node j to node i in time period t under scenario s, Represents the reactive power output of the wind turbine, Q ik,s,t represents the reactive power flowing from node i to node k in time period t under scenario s, represents the reactive power of other loads connected to node i except the data center at time t, represents the transmission power between the data center and the active distribution network, represents the maximum transmission power, represents the total power of the data center at node k in time period t under scenario s, Indicates the charging power of the data center power supply equipment. represents the transmission power between the data center and the distribution network at node k in time period t under scenario s, Indicates the discharge power of the data center energy storage equipment, They respectively represent the minimum and maximum values ​​of the interaction power between the active distribution network and the upper-level power grid.

5. The method of claim 1, wherein: The power constraints also include wind turbine operating power constraints and line transmission power constraints: Power constraints during wind turbine operation: Line transmission power constraints: In the formula, Indicates the actual output of the wind turbine. represents the predicted value of wind turbine output, Indicates the power factor angle, Indicates the reactive power output of the wind turbine. They represent the maximum output limit and maximum capacity limit of the lth line respectively, P l,s,t represents the active power transmitted by line l, Q l,s,t Represents the reactive power transmitted by line l.

6. The method of claim 1, wherein: The data center load response constraints include: The proportion of load that a data center can handle delayed is constrained by: The delay load constraint that a data center needs to handle at any time period t is: Data center load scheduling constraints: Load migration constraints: The total load constraint of the data center at any time: In the formula, represents the batch processing load of the kth data center at the initial time under scenario s, ζ represents the constant of the proportion of batch processing load in all loads, and f s,t,k,0 Indicates the amount of load that needs to be processed at the initial moment, represents the total amount of data that needs to be processed by the kth data center in time period t under scenario s, represents the batch processing load that migrates from period t' to period t in the data center numbered k under scenario s, represents the batch processing load in the data center numbered k under scenario s that migrates from period t to period t', χ k represents the state of smart meter installation in the kth data center, f s,t,k It represents the batch processing load that migrates from period t to period t' in the data center numbered k under scenario s.

7. The method of claim 1, wherein: The data center power transmission line transformation and equipment installation constraints include: Line transformation selection model constraints: Constraints on the number of smart meters installed: 0≤χ k ≤1 Constraints on the number of nodes for installing wind turbines: In the formula, Indicates the model of the line to be modified, Ω M represents the line model set of the lines to be selected, χ k represents the state of smart meter installation in the kth data center, Ω WG Represents the node set where wind turbines are installed. Indicates the installation location of the wind turbine, N WG Indicates the maximum number of nodes that the system allows to install power devices.

8. The method according to any one of claims 1 to 7, wherein: The data center voltage constraints, adjacent transmission line power flow constraints, and data center power equipment constraints are: Data center voltage constraints: Power flow constraints of adjacent transmission lines: Data center power equipment constraints: In the formula, They represent the minimum and maximum voltage allowed in data center i, respectively. s,t,i represents the voltage value of node i in time period t under scenario s, U s,t,j represents the voltage value of node j in time period t under scenario s, P l,s,t , Q l,s,t Respectively represent the active power and reactive power transmitted on line l, Ω M Indicates the line model set to be selected. Indicates the model selected for the line that needs to be modified. Respectively represent the resistance and reactance of line l before transformation, R l,m , X l,m They represent the resistance and reactance of line l after transformation, They represent the charging and discharging state variables of the data center power supply equipment in time period t, E s,t,k represents the stored power of the power supply equipment in the kth data center in time period t, E s,t-1,k represents the storage power of the power supply equipment in the kth data center in time period t-1, η C , η D They represent the charging power and discharging efficiency of the power supply equipment respectively, Δt represents the unit scheduling period, Indicates the discharge power of the power supply device. Indicates the charging power of the power supply device, P E max Indicates the maximum charge and discharge power of the power supply device. Indicates the charge status of the data center power supply equipment. Indicates the capacity of the data center power supply equipment. Respectively represent the maximum and minimum values ​​of the power device's charge state.

9. A device for constructing an energy hub model taking into account renewable energy and demand response, suitable for execution in a computing device, the device comprising: A parameter acquisition module is suitable for acquiring basic parameters, including: active distribution network topology framework, length of each line of the active distribution network, impedance value per unit line length, power load value of each node in a typical day, data demand of the communication system in a typical day, main network electricity purchase price, rated installed capacity of a single wind turbine, unit cost of a wind turbine, maintenance price of a wind turbine, daily output prediction curve of wind power generation, amount of data that a single server in a data center can process, installation cost of a single smart meter, silent power consumption value of a server, full load power consumption value of a server, number of servers in a single data center, maximum CPU utilization of a server, demand response subsidy price, rated capacity of energy storage equipment installed in a data center, maximum charging power of energy storage equipment, maximum discharge power of energy storage equipment, maximum charge and discharge efficiency of energy storage equipment, maximum and minimum charge states of energy storage equipment, resistance of an optional line, reactance of an optional line, current carrying capacity of an optional line, and price per unit length of an optional line; A model building unit, using a two-stage stochastic optimization method to establish an active distribution network planning model taking into account data centers, the model includes an objective function and constraints; The model solving unit is suitable for substituting the basic parameters into the model, and taking the investment and operation cost of the active distribution network as the minimum while improving the utilization rate of renewable energy and reducing carbon emissions as the goal, using an improved group search optimization algorithm to solve the model, output the selected model of the transformation line, the installation location of the smart meter, the installation location of the wind turbine, the number of servers started in each period of the data center, the task migration amount in the time dimension of the data center, the power purchase of the distribution network, the actual output of the wind turbine, the energy storage charging power and the energy storage discharging power formulation plan, the improved group search optimization algorithm is to combine the differential evolution algorithm and the anti-learning algorithm into the group search optimization algorithm, wherein the objective function is: my C=C INV +C OPT C INV =C line +C SM +C WG C OPT =C grid +C DER +C DC-DR In the formula, C INV represents the investment cost in the planning stage, C OPT represents the investment cost in the operation phase, C line represents the transmission line expansion cost of the data center, C SM represents the installation cost of smart meters in data centers, C WG represents the installation cost of the wind turbine in the data center, C grid represents the cost of active distribution network purchasing electricity from the upper grid, C DER represents the maintenance cost of energy storage equipment, C DC-DR represents the demand response incentive cost paid by the active distribution network to the data center; The constraints include: one or more of power constraints, data center load response constraints, data center load delay processing constraints, data center transmission line modification and equipment installation constraints, data center voltage constraints, adjacent transmission line flow constraints, and data center power supply equipment constraints.

10. A computing device comprising: at least one processor; and A memory storing program instructions, wherein the program instructions are configured to be executed by the at least one processor, and the program instructions include instructions for executing the method according to any one of claims 1 to 8.

11. A readable storage medium storing program instructions, when the program instructions are read and executed by a computing device, the computing device executes the method according to any one of claims 1 to 8.

Images