High-toughness coordinated operation method for data center and power distribution network

Abstract

The invention discloses a high-toughness coordinated operation method for a data center and a power distribution network, in a load recovery process of the power distribution network, a load with space-time flexibility, namely the data center, is introduced, a transfer strategy of standby energy storage in the data center is considered, and in addition, a distributed generator which can be used as a voltage source is also introduced. The introduction of the two is helpful for improving the toughness of the power distribution system, helping the power distribution system to quickly recover key loads, and reducing the economic loss in a power failure period. Under the background that the requirement of the data center for the power supply stability is high, the flexible transformer substations are introduced, the flexible transformer substations serve as novel power electronic converters and are installed on the two sides of the data center, the data center operates on a direct-current bus, efficient and stable power supply is provided, and meanwhile the operation cost of the data center can be reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of post-disaster recovery of power grids, and in particular to a method for high-resilience coordinated operation of a data center and a distribution network. Background Art

[0002] In recent years, as global climate change intensifies, large-scale power outages caused by extreme disasters have occurred frequently, causing serious economic losses. Compared with the transmission network, the distribution network has a weaker ability to respond to extreme disasters and is more prone to power outages due to sudden disasters. In addition, the distribution network is connected to industrial, commercial and residential terminal loads, so once the distribution network fails, it will have a huge impact on economic and social benefits.

[0003] Among them, Internet data centers, as key distribution network infrastructure, are different from other loads in the power system. They have special time-space load transfer characteristics and are more important flexible loads in the power system. Data centers can improve the resilience of the power system through the time-space scheduling of workloads and the interaction between the backup energy storage in the data center park and the power system. From the perspective of data center operations, data centers are driven by electricity. Therefore, power outages in the distribution network have always been a huge threat to data center operations. Therefore, considering the heavy dependence on electricity and its key position in the local economy, data centers have become new critical loads in the distribution system. The effective recovery of the distribution system after natural disasters is crucial to reducing the economic losses of data centers. Summary of the invention

[0004] The purpose of the present invention is to provide a highly resilient coordinated operation method between a data center and a distribution network to solve the problems raised in the above-mentioned background technology. In order to improve the resilience of the distribution system, the present invention introduces a data center as a critical load of the distribution network, and establishes an emergency operation model for the data center park during the distribution network recovery phase. Among them, the workload of the data center and the backup energy storage jointly assist the distribution network in restoring critical loads. In order to ensure the operational stability of the data center, the present invention introduces a flexible substation, which enables the data center to operate more stably and efficiently on the DC bus, and formulates a voltage support strategy for distribution network restoration based on the flexible substation and distributed generators. With the goal of minimizing the loss of the key load recovery process of the distribution network and minimizing the workload loss of the data center, a coordinated power supply restoration method is proposed. Without violating physical feasibility, spatiotemporal coordination between the source, network, storage and demand side can be achieved.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for high-resilience coordinated operation of a data center and a distribution network comprises the following steps:

[0007] Establish an emergency operation model for the data center park during the distribution network recovery phase. After the distribution network is powered off, the insufficient power supply will have a significant impact on the operation of the data center. Therefore, in the technical solution of the present invention, it is necessary to first establish an emergency operation model for the data center. Figure 1 The basic structure of a data center park includes the data center body, backup energy storage, and a flexible substation. During the emergency operation of a data center park, the power demand of the data center can be changed by the time and space transfer of the data center workload. The backup energy storage can be reasonably deployed to provide power support for the data center, and can help the distribution system restore critical loads when power is restored. Therefore, this section establishes a data center emergency data flow scheduling model and a data center park backup energy storage deployment strategy. Specifically, a data center load scheduling model based on data flow scheduling and its constraints are considered for two load regulation methods: geographic load balancing and delay-tolerant load time scheduling; a three-stage scheduling strategy for backup energy storage in a data center park. In the first stage, the backup energy storage in the park only supplies power to the data center. In the second stage, the distribution network provides power support for charging the data center and the backup energy storage. In the third stage, the backup energy storage responds to the deployment of the distribution system.

[0008] The emergency data flow scheduling model of the data center mainly includes the data center power consumption model and the data center workload time-space scheduling model. The data center power consumption model uses the CPU utilization model and uses the power utilization efficiency to characterize the proportional relationship between the IT equipment power and the total power of the data center. The IT equipment power considers the processing power consumption of delay-tolerant workloads and delay-sensitive workloads. The data center workload time-space scheduling model considers the spatial scheduling and workload removal constraints of delay-sensitive loads, as well as the temporal scheduling and workload removal constraints of delay-tolerant loads.

[0009] In addition, through the reasonable planning and deployment of backup energy storage, it is possible to flexibly interact with the distribution network while ensuring the stability of the power supply of the data center, so as to improve the resilience of the distribution network. The invention divides the dispatching strategy of backup energy storage into three stages. In the first stage, the backup energy storage in the park only supplies power to the data center. In the second stage, the distribution network provides power support for the charging of the data center and the backup energy storage. In the third stage, the backup energy storage can respond to the deployment of the distribution system.

[0010] Establish a voltage support strategy based on flexible substations and distributed generators: After the distribution network fault is isolated, consider the voltage support role that flexible substations and distributed generators can play on the bus. Consider that flexible substations and distributed generators are in two working states, namely power regulation mode and voltage support mode. For flexible substations, when the buses on both sides are in normal operation, the flexible substation operates in power regulation mode to achieve power interaction between AC and DC buses; when one side of the bus is operating normally, the flexible substation operates in voltage support mode to provide voltage support for the power-off bus. For distributed generators, when adjacent nodes are in a power-off state, they will operate in voltage support mode to provide voltage support for the power-off nodes; if they are all operating normally, they will be in power regulation mode, and the flexible substation terminal and distributed generators will provide voltage support and assist available substations;

[0011] Consider the post-disaster recovery modeling of the distribution network of the data center: After the emergency operation model of the data center and the voltage support strategy based on flexible substations and distributed generators are established, the model parameters should be delivered to the distribution network operator for integration, and the distribution network flow constraints and the distribution network radial topology constraints should be added to merge into a recovery strategy model for the entire distribution network, which will be jointly solved by the distribution network operator to ultimately achieve the recovery of key loads in the distribution network; the objective function and constraints in the distribution network recovery process will minimize the total power outage losses of the data center data workload and other unrestored power loads during the power supply restoration process.

[0012] Compared with the prior art, the present invention has the following beneficial effects:

[0013] (1) In the process of load recovery in the distribution network, the present invention introduces a load with spatiotemporal flexibility, namely, a data center, and considers the mobilization strategy of backup energy storage in the data center. In addition, a distributed generator that can be used as a voltage source is introduced. The introduction of these two helps to improve the resilience of the distribution system, help the distribution system quickly restore key loads, and reduce economic losses during power outages;

[0014] (2) In the context of data centers having high requirements for power supply stability, the present invention introduces a flexible substation. As a new type of power electronic converter, the flexible substation is installed on both sides of the data center, enabling the data center to operate on a DC bus. While providing efficient and stable power supply, it can reduce the operating costs of the data center. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is the basic structure diagram of the AC / DC hybrid distribution network that includes the data center park;

[0016] Figure 2 Schematic diagram of the coupling mechanism between data center and distribution network and the high-resilience coordinated operation method;

[0017] Figure 3 Schematic diagram of the process of provisioning workloads for a data center. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] See also Figures 1 to 3 In an embodiment of the present invention, a method for high-resilience coordinated operation of a data center and a distribution network includes the following steps:

[0020] (1) The emergency operation model of the data center park during the distribution network restoration phase is described in detail as follows:

[0021] The emergency operation model of the data center park in the distribution network recovery phase of the present invention is mainly divided into a data center data flow scheduling model and a three-stage planning strategy for backup energy storage.

[0022] This section assumes that the N data centers managed by the same data center operator are distributed in different locations of the same distribution network. Figure 3 As shown in Figure 1, data center operators aggregate workload requests from network users based on the principle of proximity through front-end proxies. After aggregation, the front-end proxies spatially allocate delay-sensitive workloads according to the power availability information of each data center, and delay-tolerant workloads are distributed to each data center using a balancer. If it is not worth using computing resources or exceeds the computing capacity or latency requirements of the data center, the data load will be deleted, thereby maximizing the use of limited power and reducing the losses of data center operators. In this chapter, all nodes mentioned are data center nodes (i∈Ω IDC ).

[0023] The technical solution of the present invention assumes that all servers in each data center are homogeneous, that is, the performance, rated power, etc. of the servers are the same.

[0024] 1) Data Center Power Consumption Model

[0025] At present, data centers often use power usage effectiveness (PUE) to describe the proportional relationship between IT equipment power and the total power of the data center. The power of data center i at time t is:

[0026]

[0027] In the formula, is the total power consumption of data center i at time t, PUE i is the power utilization rate of data center i, is the power consumption of IT equipment in data center i at time t.

[0028] The IT equipment power includes server power and network transmission power. The IT equipment power at time t is:

[0029]

[0030] In the formula, is the power consumed by the server i in the data center at time t, is the network transmission power of data center i at time t. The network transmission power of the data center includes the transmission link power and the transmission equipment power. The dynamic range is very small, so the network transmission power Can be considered as a fixed value.

[0031] The utilization model describes the utilization of servers based on the ratio of workload arrival rate to service rate, and solves the data center server power by integrating parameters such as the number of servers, no-load power, and full-load power:

[0032]

[0033] Where P i idle and P i peak are the power of the active servers in data center i in idle and fully loaded states, respectively, and u is the CPU utilization of the active servers.

[0034] According to the different processing characteristics of workloads in data centers, the workloads processed by data centers can be divided into delay sensitive (DS) workloads and delay tolerant (DT) workloads. Delay sensitive workloads have high requirements for the processing time of service responses, usually requiring the delay time to be within 1 second. Delay tolerant workloads have low requirements for service response time. According to the deadline required by network users, the delay time of their responses is usually several hours to several days. This chapter considers the different processing characteristics of these two workloads to establish a server power consumption model.

[0035] Assume that data center i is turned on at time t The average CPU utilization of a server processing a latency-sensitive workload can be calculated as:

[0036]

[0037] In the formula, is the server CPU utilization of latency-sensitive workloads in data center i at time t, Δ is the total number of geographically dispersed front-end proxies, and λ i,δ,t is the number of arrivals of delay-sensitive workloads from front-end proxy δ at data center i in period t, The average service rate of latency-sensitive workloads handled by servers in data center i.

[0038] Assume that data center i is turned on at time t The average CPU utilization of a server processing a delay-tolerant workload h can be calculated as:

[0039]

[0040] In the formula, is the server CPU utilization of the delay-tolerant workload in data center i during period t, b i,h,t is the number of delay-tolerant workloads h processed by data center i in period t, The service rate for servers in data center i processing delay-tolerant workload h.

[0041] Then the power of the server in data center i during period t can be expressed as:

[0042]

[0043] Where H is the type of delay-tolerant workload.

[0044] 2) Data center spatiotemporal characteristics

[0045] For latency-sensitive workloads, the front-end agent spatially schedules the workload based on power availability information. If the power supply is insufficient to handle the workload, the workload will be deleted, which will cause losses to the data center operator.

[0046] The distribution constraints of the latency-sensitive workloads arriving at the front-end proxy at each time period are as follows:

[0047]

[0048] In the formula, The total number of latency-sensitive workload arrivals at the front-end proxy δ in period t.

[0049] The processing delay time of latency-sensitive workloads is calculated based on the M / M / 1 queuing model, and the following constraints must be met:

[0050]

[0051] Converted into linear form:

[0052]

[0053] Where D is the maximum processing delay time of the delay-sensitive workload;

[0054] For delay-tolerant workloads, since they generally require a large amount of background data, scheduling between different data centers will sacrifice a large amount of memory and cause bandwidth loss, so the main consideration is their time scheduling capabilities. Assume that the same type of delay-tolerant data load is evenly distributed to each data center by the balancer. Delay-tolerant workloads should be processed and completed before their deadline. If they are not completed within the deadline, the workload will be deleted:

[0055]

[0056] In the formula, τ h is the maximum delay time of the delay-tolerant workload h, is the time period for the delay-tolerant workload h to arrive at data center i, B i,h is the total number of delay-tolerant workloads h;

[0057] The latency-tolerant workload assigned to each time slot should all be processed to completion:

[0058]

[0059] 3) Interaction strategy between backup energy storage and distribution network in data center parks

[0060] ①First-stage deployment strategy

[0061] When the distribution network is out of power (i.e. ), the power supply of the data center will switch to the backup energy storage to ensure uninterrupted power supply of the data center. In order to maintain the high reliability of data center workload processing, the backup energy storage only supplies power to the data center during this stage, and does not provide energy storage services for the recovery of the distribution system. That is, during this stage, there is no power exchange between the data center and the distribution system.

[0062] As in the above scenario, the reserve energy storage constraints in the first stage are as follows:

[0063] The backup energy storage is in a discharging state in the first stage, and its discharge power should be less than the maximum discharge power:

[0064]

[0065] in:

[0066] is the discharge power of the first stage backup energy storage at data center i at time t;

[0067] is a binary variable indicating whether data center i is in the first stage of discharge at time t;

[0068] P i dmax Indicates the maximum value of the backup energy storage charging and discharging power of data center i;

[0069] The indicator of whether the data center park is in the first stage of discharge is At this time, the distribution network does not supply power to the data center park, that is, the data center is not connected

[0070]

[0071] In the formula, is a binary variable indicating whether node i is connected at time t.

[0072] State of Charge (SOC) timing constraints for backup energy storage:

[0073]

[0074] In the formula, SOC i,t is the charge state of the backup energy storage of data center i at time t, E i Represented as the capacity of backup energy storage of data center i.

[0075] The charge state of the backup energy storage of the data center in the first phase should not be lower than the minimum charge state:

[0076]

[0077] In the formula, is the minimum capacity of backup energy storage required for the reliability of data center i.

[0078] When the backup energy storage works as a power source, its energy storage level is the initial energy storage level before the power outage:

[0079]

[0080] In the formula, is the initial charge state of the backup energy storage before data center i is shut down.

[0081] The backup energy storage is only used to power the data center at this time, and the discharge power is equal to the power consumed by the data center:

[0082]

[0083] Wherein, M is a positive integer M in the big-M method.

[0084] ② Second phase deployment strategy

[0085] When the data center park is initially powered on, the data center park will be powered by the distribution network. At this time, the backup energy storage level of the data center park is at a low level and has no power supply capacity (z i,t =0), so the backup energy storage in the data center park will be charged immediately after power is turned on. In this stage, the power provided by the distribution system is the sum of the data center operating power and the charging power of the backup energy storage. The charging time of the second stage depends on the power outage time. Therefore, the operation behavior of the data center is closely related to the load recovery decision.

[0086] In this article, it is assumed that when the data center is powered back on, The backup energy storage will be charged to its initial energy storage level as quickly as possible. Before the energy storage level of the backup energy storage reaches the initial level, the backup energy storage cannot provide energy storage services for the distribution system. And for the sake of simplicity, it is assumed in this stage that the charging power of the backup energy storage remains unchanged.

[0087] Therefore, the energy storage constraint in the second stage is:

[0088]

[0089] SOC i,t =[SOC i,1 ,SOC i,2 ,…SOC i,T ](19)

[0090]

[0091] In the formula, z i,t is the availability status of backup energy storage for distribution system restoration, P i cmax The maximum charging power of the backup energy storage of data center i, is the charging efficiency of backup energy storage in data center i, P i nd is the charging and discharging power of the second-stage backup energy storage of data center i.

[0092] ③The third stage interactive strategy

[0093] After the second stage, the charge state of the backup energy storage reaches the initial level through continuous charging. Compared with the minimum backup capacity, the additional part of the backup energy storage can be used to serve the distribution system (i.e., help restore the load). Therefore, at this stage, the data center has been fully restored, and the backup energy storage begins to provide energy storage services to the distribution system in the third stage (i.e., z i,t =1).

[0094] Therefore, the energy storage constraint in the third stage is:

[0095] Once backup storage becomes available, storage services can always be provided:

[0096] z i,t+1 ≥z i,t (twenty four)

[0097] This means that the available time of backup energy storage in this stage should be later than the power-on time of the data center:

[0098]

[0099] The power of backup energy storage in the third stage of data center i at time t is:

[0100]

[0101] In the formula, is the charge / discharge power of the backup energy storage of data center i at time t in the third stage, is the power of backup energy storage of data center i at time t in the third stage.

[0102] Restrictions on backup energy storage that provides energy storage services for distribution systems (limiting the size of charging and discharging power):

[0103]

[0104] In the formula, A binary variable representing the charge / discharge status of the backup energy storage of data center i at time t in the third stage.

[0105] Backup energy storage can only work in the third stage and cannot charge and discharge at the same time:

[0106]

[0107] When the backup energy storage works as a power source, its energy storage level is the initial energy storage level:

[0108]

[0109] In the formula, is the charge state of the backup energy storage of data center i at time t in the third stage.

[0110] The SOC constraints of backup energy storage are as follows:

[0111]

[0112] The charge level of the backup energy storage should be restored to its initial level when the load restoration ends:

[0113]

[0114] (2) The voltage support strategy during the distribution network recovery phase is described in detail as follows:

[0115] 1) Voltage support for AC bus

[0116] Available substations, flexible substations and distributed generators can provide voltage support for AC buses in a power outage state. For the safe operation of the distribution system, each AC bus has only one voltage source to stabilize the bus voltage.

[0117]

[0118] In the formula, u g,t is the square of the AC support voltage provided, u ref is the square of the AC support reference voltage, is a binary variable indicating whether to provide AC support voltage (1 means providing voltage support), Ω FS,ac is the set of AC nodes connected to FS, Ω sub is the set of AC nodes connected to the substation, Ω DG is the set of AC nodes connected to the distributed generators.

[0119] According to the operating characteristics of the flexible substation, when the AC bus lacks voltage support and the DC bus is in normal operation, that is, when the data center park on the DC bus is powered on, The flexible substation can provide voltage support for the AC bus When the node is located on a powered branch and connected to the flexible substation, that is, the AC bus is in normal operation, the flexible substation operates in power adjustment mode. That is, the AC terminal of the flexible substation does not provide voltage support for the connected AC nodes. When all nodes connected to the distributed generator are in a power-off state, the distributed generator will provide voltage support. When a node is connected to a substation node, even if it is located on an energized branch line, the substation should give priority to providing voltage support. At this time, the flexible substation and distributed generator switch to power adjustment mode.

[0120]

[0121] In the formula, Ω ac,off is the set of AC nodes that are powered off, is a node connected to the same flexible substation m, is the node connected to the same distributed generator m, ζ ac is the set of communication branch nodes, ζ ac,off It is the set of AC branch nodes that are de-energized.

[0122] 2) DC low voltage bus voltage support

[0123] Since only the flexible substation is related to the DC node, this section only considers the voltage support of the flexible substation. During the recovery period, the DC terminal of the flexible substation can stabilize the DC voltage. For the radial topology, there is a unique voltage source on the DC bus. According to the operating characteristics of the flexible substation, when the DC bus lacks voltage support and the AC bus is in normal operation, the flexible substation will provide voltage support, and the data center in the first stage will enter the second and third stages of operation.

[0124]

[0125] In the formula, is a binary variable indicating whether to provide DC support voltage (1 means providing voltage support), Ω dc is the set of all DC nodes.

[0126] For DC nodes, when the AC nodes connected by the same flexible substation are powered on, they will definitely receive voltage support from the flexible substation.

[0127]

[0128] 3) Cooperation between AC and DC busbars

[0129] Only when any terminal of the flexible substation is connected to a normally operating busbar, the flexible substation terminal can actively maintain the node voltage. Since only two ports of the flexible substation are considered in this paper, the flexible substation can only provide DC voltage support or AC voltage support, not both at the same time.

[0130]

[0131] (3) Consider the post-disaster recovery modeling of the data center distribution network, as detailed below:

[0132] 1) Objective function

[0133] The main objective of the present invention is to minimize the total power outage losses caused by the decline in data workload in the data center and other unrestored power loads during the power restoration process.

[0134] min{f IDC +f L +f FS} (39)

[0135] In the formula, f IDC is the economic loss of the data center, f L is the economic loss of other loads, f FSThe economic loss of the flexible substation.

[0136] The economic losses of a data center are as follows:

[0137]

[0138] In the formula, The price of data received by the data center.

[0139] The loss of other loads is measured in electricity prices:

[0140]

[0141] In the formula, π E is the price of active power purchased from the upper power grid, α i is the load recovery rate, ω i represents the load weight, is the active power of the node i load at time t.

[0142]

[0143] 2) Constraints

[0144] ① Flow constraints

[0145] This section adopts the Distflow power flow branch model, where the AC node power flow constraints are shown in (43)-(44):

[0146] In this part, the lines are all AC lines, that is, ij,ki∈ζ ac , representing the flow from i to j and from k to i respectively.

[0147] The node voltage amplitude constraints are as follows:

[0148]

[0149] In the formula, u i,t 、u j,t represents the square of the voltage at the corresponding node, l ij,t Represents the square of the line current, P ij,t , Q ij,t represents the active / reactive power of the line at time t, r ij 、x ij Indicates the resistance and reactance of the line, P ki,t , Q ki,t represents the active / reactive power transmitted from node k to node i, P ij,t , Q ki,t represents the active / reactive power transmitted from node i to node j, is the active / reactive power of the distributed generator at node i at time t, is the active / reactive power of the node i load at time t, is a binary variable indicating whether the load is put into use. is the square of the upper / lower limit of the node voltage, is the upper / lower limit of the active power that the line can bear, It is the upper / lower limit of the reactive power that the line can bear.

[0150] The DC node power flow constraints are shown in (52)-(53):

[0151]

[0152] In the formula, z i,t A binary variable indicating the availability status of backup energy storage for recovery dispatch.

[0153] ②Power recovery constraints of power loads

[0154] The load recovery rate during a power outage is limited by the operating status of the node where it is located. The load of the node can be restored. The load restoration rate depends on the level of power availability and the weight of the load.

[0155]

[0156] ③Operation constraints of distributed generators

[0157] Distributed generators have capacity and ramp constraints, which are described as follows:

[0158]

[0159]

[0160] Where P i DGmax , P i DGmin are the upper and lower limits of the active power output of the distributed generators, are the active power of the distributed generator at node i at time t, are the upper and lower limits of the active power output of the distributed generators, are the active power of the distributed generator at node i at time t, P i r It is the ramp rate limit of the distributed generator power.

[0161] ④ Emergency operation constraints of data center parks

[0162] The emergency operation constraints of the data center campus are as described in Section 1. In addition, there are constraints on workload and the number of active servers:

[0163] λ i,δ,t ≥0 (65)

[0164] b i,h,t ≥0 (66)

[0165]

[0166] Where M i is the total number of servers in data center i.

[0167] ⑤ Operation constraints of flexible substation

[0168] The basic structure of a flexible substation is as follows: Figure 1 As shown in the figure, according to the schematic diagram of the internal converter structure of the four-port flexible substation, g∈Ω in this section FS,ac ,i∈Ω FS,dc , the operation model of the flexible substation is as follows:

[0169] The power balance constraints of the flexible substation are as follows:

[0170]

[0171] In the formula, is the active power output from port 1 of the flexible substation on the AC side at node g at time t, is the active power output of DC port 2 of flexible substation at node i at time t, P t ac,l is the active power loss of the AC / DC converter in FS, is the active power loss of the DC / DC converter in FS.

[0172] The losses caused by the AC / DC converter are as follows:

[0173]

[0174] Where P t ac,l is the active power loss of FS AC port 1, and μ1 is the loss coefficient of the AC / DC converter in the flexible substation.

[0175] The losses caused by the DC / DC converter are as follows:

[0176]

[0177] Where μ2 is the loss coefficient of the DC / DC converter in the flexible substation.

[0178] The capacity constraints of the DC terminals are as follows:

[0179]

[0180] In the formula, S FS,DD is the capacity of the DC / DC converter in FS.

[0181] Since FS achieves electrical isolation of the connected feeders, the reactive power output on the AC side is independent and needs to meet the following capacity constraints:

[0182]

[0183] In the formula, Q AD,min , Q AD,max is the upper and lower limits of reactive power of AC / DC converter in flexible substation, is the reactive power of DC port 1 in the flexible substation.

[0184] The capacity of the AC port is constructed as a cone constraint:

[0185]

[0186] In the formula, is the reactive power output of the AC port in the flexible substation, S FS,AD is the capacity of the AC / DC converter in the flexible substation.

[0187] ⑥ Radial topology constraints of flexible distribution networks

[0188] The following formula is the timing constraint of the node power supply state. Once the node is powered on, it will remain in the connected state:

[0189]

[0190] The following formula limits the switching capacity of the branches. Only the switched line can change state. The non-switched branches keep the original state. The following consideration can reduce the scale of variables to meet the limitations of the actual distribution network switch installation:

[0191]

[0192] In the formula, s ij,t a binary indicator representing the connection status of the distribution line between node i and node j at time t (1 if connected), A binary indicator indicating whether a physical connection (distribution line) exists between nodes i and j (1 if so).

[0193]

[0194] In the formula, is a binary indicator variable used to indicate whether a remote switch is provided in the line between node i and node j (if yes, it is 1).

[0195] For a connected branch, the connection direction is unique:

[0196] L ij +L ji =s ij,t (77)

[0197] Where, L ij , L ji is the connectivity indicator. When node j is a child node of node i, then L ij =1 and L ji =0.

[0198] 0≤L ij ≤1,0≤L ji ≤1(78)

[0199] For a connected node There is only one parent node (i.e., there is no ring network in this recovery process). To ensure the radiance, if the node is directly connected to an available substation or is supported by a flexible substation or distributed generator, then the associated direction indicator will be forced to zero:

[0200]

[0201] ⑦ Radial topology constraints based on modified virtual networks

[0202] Based on the radial topology of the modified virtual network, a modified virtual network is constructed to describe the flexible radial restoration structure with flexible substations and distributed generators. A set of virtual energy variables is introduced to reveal the connection order of branches and nodes. The following formula maps the physical connections and operation states between nodes, branches, flexible substations, distributed generators and substations to the virtual network.

[0203] The virtual energy flow limits on the line are as follows:

[0204] L ij ≤F ij ≤M·L ij (80)

[0205] L ji ≤F ji ≤M·L ji (81)

[0206] i,j∈Ω ac

[0207] In the formula, Fij is the virtual energy flowing from node i to node j.

[0208] The virtual energy limit of FS connected nodes is as follows:

[0209]

[0210] In the formula, F i FS is the virtual energy flowing from the flexible substation to node i.

[0211] The virtual energy limits of the substation connection nodes are as follows:

[0212]

[0213] In the formula, F i sub is the virtual energy flowing from the substation to node i.

[0214] The virtual energy limit of the distributed generator connection node is as follows:

[0215]

[0216] In the formula, F i DG is the virtual energy flowing from the substation to node i.

[0217] The voltage source node starts the restoration path as a virtual source, and the source node can provide virtual energy to other nodes. The branch and flexible substation transfer the virtual energy to each non-source subnode with virtual load:

[0218]

[0219] In the formula, represents the parent node set of node i, Represents the set of child nodes of node i.

[0220] Available substations and distributed generators can be virtual sources for the entire system. There may be multiple voltage source selections. Nodes and branches can be divided into several restoration paths, but there should be at least one voltage source for the entire network.

[0221]

[0222] In order to ensure the radial conditions of each path, the following constraints are added based on graph theory:

[0223]

[0224] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0225] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A method for high-resilience coordinated operation of a data center and a distribution network, characterized in that: The steps include: (1) Establish an emergency operation model for the data center park during the distribution network recovery phase: Establish an emergency data flow scheduling model for the data center and a backup energy storage allocation strategy for the data center park. The basic structure of the data center park includes the data center body, backup energy storage, and flexible substation. During the emergency operation of the data center park, the power demand of the data center is changed through the spatiotemporal transfer of the data center workload. The backup energy storage is deployed to provide power support for the data center, and the distribution system is helped to restore critical loads when power is restored. The emergency data flow scheduling model for the data center includes a data center power consumption model and a data center workload time-space scheduling model. The data center power consumption model uses the CPU utilization model and uses power utilization efficiency to characterize the proportional relationship between IT equipment power and the total power of the data center. The IT equipment power considers the processing power consumption of delay-tolerant workloads and delay-sensitive workloads. The data center workload time-space scheduling model considers the spatial scheduling and workload removal constraints of delay-sensitive loads, as well as the temporal scheduling and workload removal constraints of delay-tolerant loads. (2) Establish a voltage support strategy based on flexible substations and distributed generators: After the distribution network fault is isolated, the voltage support role that the flexible substation and distributed generator can play on the bus is considered. The flexible substation and distributed generator are considered to be in two working states, namely power regulation mode and voltage support mode. For the flexible substation, when the buses on both sides are in normal operation, the flexible substation operates in power regulation mode to realize the power interaction of the AC and DC buses; when one side of the bus is operating normally, the flexible substation operates in voltage support mode to provide voltage support for the power-off bus. For distributed generators, when the adjacent nodes are in a power-off state, they will operate in voltage support mode to provide voltage support for the power-off nodes; if they are all operating normally, they will be in power regulation mode, and the flexible substation terminal and distributed generators will provide voltage support and assist the available substations; (3) Modeling of post-disaster recovery of distribution networks considering data centers: After the emergency operation model of the data center and the voltage support strategy based on flexible substations and distributed generators are established, the model parameters are delivered to the distribution network operator for integration, and the distribution network flow constraints and distribution network radial topology constraints are added to form a recovery strategy model for the entire distribution network. The distribution network operator will jointly solve the model to achieve the recovery of key loads in the distribution network.

Images