Active power distribution network optimization scheduling method and system considering edge data center cluster

Abstract

The application discloses an active power distribution network optimization scheduling method and system considering an edge data center cluster, which can fully utilize the space-time flexibility of the edge data center under the operation optimization technical framework of the power distribution network, effectively reduce the overload of the power distribution network line, improve the capacity of the active power distribution network to accommodate new energy, and reduce the energy consumption of the edge data center and the significant time and space changes of the computing load of the edge data center, so that the edge data center cluster has strong load regulation capacity. The method can optimize and schedule the large computing load of the edge data center cluster, so that the regulation capacity of the edge data center cluster can be fully utilized, thereby reducing the regulation pressure of the power generation side, the edge data center cluster participates in demand response, the line congestion problem of the power distribution network can be effectively alleviated, the load level can be changed by using the space-time migration of the computing load, the new energy output of the region can be matched, and the renewable energy can be maximally utilized.

Description

Technical Field

[0001] This invention belongs to the field of power distribution network operation optimization technology, specifically relating to an active power distribution network optimization scheduling method and system that considers edge data center clusters. Background Technology

[0002] With the widespread adoption of the Internet of Things (IoT), massive amounts of data are generated anytime, anywhere by numerous internet-connected terminal devices. To address the problems of network congestion, high latency, and low data security inherent in traditional cloud computing, edge computing—the concept of migrating computing resources from the remote cloud to the network edge—has been proposed. Edge data centers consume enormous amounts of energy, becoming one of the major power loads. Reducing data center energy consumption and achieving efficient energy management has become a pressing challenge for the data center industry. IT equipment accounts for the majority of energy consumption in edge data centers, and the energy consumption of IT equipment is positively correlated with the workload of the edge data center. Therefore, optimized scheduling of computing load is crucial for adjusting the load power of edge data centers and reducing their energy consumption. Edge data center clusters can participate in power grid scheduling through demand response, which can effectively reduce data center energy costs and play an important role in the safe and economical operation of the power system.

[0003] Existing research on data center cluster scheduling strategies focuses on energy consumption optimization for large-scale cloud data center clusters spanning multiple regions within power transmission networks. In edge computing environments, research on edge computing resources widely present in distribution networks primarily focuses on load balancing techniques to improve edge computing performance, utilizing the collaboration of edge computing nodes to reduce congestion probability and service latency, and improve resource utilization. Research on energy consumption optimization of edge computing resources and their collaborative optimization with the distribution network is relatively limited. The computing load of edge data centers is related to computing tasks and often exhibits significant temporal and spatial variations, providing ample flexibility for optimized scheduling and giving edge data center clusters strong load regulation capabilities. Responding to regional electricity price differences is a primary form of demand response for data center clusters. By optimizing the allocation of computing load, migrating it to data centers in low-electricity-price areas and during low-electricity-price periods, and simultaneously optimizing computing resource scheduling to minimize the number of active servers, the energy consumption costs of data center clusters can be significantly reduced. Specifically, by leveraging the low latency requirements of batch processing tasks, data centers can dynamically adjust their workload in response to electricity price differences at different times. Interactive tasks, which have higher latency requirements, can be addressed by migrating computing loads from overloaded edge data centers to underloaded edge data centers via optical networks in response to electricity price differences in different regions. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide an active distribution network optimization scheduling method and system that considers edge data center clusters in order to address the shortcomings of the prior art and solve the technical problem of coordinated operation between edge data center clusters and active distribution networks.

[0005] The present invention adopts the following technical solution:

[0006] Considering the active power distribution network optimization scheduling method for edge data center clusters, the following steps are included:

[0007] S1. Construct an energy consumption model for an edge data center cluster, including a load spatiotemporal migration model and an edge data center energy consumption model;

[0008] S2. Construct an active distribution network operation optimization model that takes into account the spatiotemporal flexibility of the edge data center cluster, and construct the constraints of the active distribution network operation optimization model based on the energy consumption model of the edge data center cluster obtained in step S1.

[0009] S3. Solve the active distribution network operation optimization model that takes into account the spatiotemporal flexibility of the edge data center cluster obtained in step S2 to obtain the active distribution network optimization scheduling result considering the edge data center cluster.

[0010] Specifically, in step S1, the constraints of the load spatiotemporal migration model are as follows:

[0011] Load balancing constraints that should be satisfied during interactive task migration:

[0012]

[0013] Where E is the set of nodes connecting the edge data center, and T is the set of time period values. This represents the interactive task load that migrates from edge data center j to edge data center i during time period t. This indicates an interactive task that has not been migrated. The arrival rate of interactive tasks unloaded to edge data center j during time period t;

[0014] The amount of interactive task migration cannot exceed the link bandwidth limit. Bandwidth constraints:

[0015]

[0016] in, This represents the upper limit of the link bandwidth between edge data centers i and j;

[0017] The total number of interactive tasks processed per unit time in edge data center j:

[0018]

[0019] in, This represents the interactive task load carried by edge data center j during time period t after migration.

[0020] Batch processing task workload constraints:

[0021]

[0022] in, The maximum allowable latency for batch task l to be offloaded to edge data center j. To offload the batch processing task load to the edge data center j, This represents the batch processing task load of the j-th edge data center during time period i after time migration;

[0023] The total number of batch processing tasks processed per unit time in the edge data center:

[0024]

[0025] in, The total number of batch processing tasks processed per unit time in the edge data center j.

[0026] Specifically, in step S1, the energy consumption model for the edge data center is as follows:

[0027] P EDC =η(P Idle (x I +x B )+(P Peak -P Idle )(λ I +λ B ) / μ)

[0028] Among them, P EDC Let P be the total energy consumption of the edge data center, η be the energy efficiency of the edge data center, and P be the total energy consumption of the edge data center. Idle x represents the server's idle power consumption. I x is the number of active servers that handle interactive tasks. B P is the number of active servers that can handle batch processing tasks. Peak For the peak power consumption of the server, λ I λ represents the amount of interactive task load it can handle. B The batch processing workload is represented by μ, which is the number of customers served per unit of time.

[0029] Specifically, in step S2, the active distribution network operation optimization model takes minimizing the total operating cost as its optimization objective, as follows:

[0030]

[0031] Where s is the scene number, Ω is the scene set, and π s Let be the probability of scenario s occurring. The energy cost for edge data center clusters in scenario s. The cost of migrating interactive tasks in scenario s. For the distribution network line loss cost in scenario s, To reduce penalties for new energy vehicles in scenario S. The penalty cost for overload of distribution network lines in scenario s.

[0032] Specifically, the constraints of the active distribution network operation optimization model include constraints related to edge data center clusters, active and reactive power output constraints of new energy generator units, and power flow constraints of the distribution network.

[0033] Furthermore, constraints related to edge data center clusters include:

[0034] Interactive task constraints:

[0035]

[0036]

[0037]

[0038]

[0039] in, This represents the number of active servers in edge data center j that process interactive tasks during time period t in scenario s. μ represents the interactive task load carried by edge data center j during time period t in scenario s. j τ represents the number of customers served per unit time in edge data center j. I The maximum allowable latency for interactive tasks is defined by E, where E is the set of edge data centers or the set of nodes connecting to edge data centers, T is the set of time period values, and Ω is the set of scenarios. This represents the workload of interactive tasks migrating from edge data center i to edge data center j during time period t in scenario s. The arrival rate of interactive tasks unloaded to edge data center j during time period t in scenario s. Let be the interactive task load carried by edge data center i during time period t in scenario s. This represents the upper limit of the link bandwidth between edge data centers i and j;

[0040] Batch processing task constraints:

[0041]

[0042]

[0043]

[0044] in, The maximum allowable latency for batch task l to be offloaded to edge data center j. Let l be the load of batch processing tasks carried by edge data center j during time period t after migration in scenario s. This refers to the load of batch processing task l that is offloaded to edge data center j in scenario s. This represents the total batch processing workload carried by edge data center j during time period t after migration in scenario s. μ represents the number of active servers in edge data center j carrying batch processing tasks during time period t in scenario s. j Let j be the number of customers served per unit time in the edge data center;

[0045] Edge data center computing capacity constraints:

[0046]

[0047] Power consumption constraints for edge data centers:

[0048]

[0049] in, Let η be the load power of edge data center j during time period t in scenario s. j For the power utilization efficiency of edge data center j, The idle power consumption of the server in edge data center j. This represents the number of active servers in edge data center j that process interactive tasks during time period t in scenario s. This represents the number of active servers in edge data center j that handle batch processing tasks during time period t in scenario s. The peak power consumption of the server in edge data center j. This represents the interactive task load carried by edge data center j during time period t in scenario s.

[0050] Furthermore, the specific constraints on the active and reactive power output of new energy generator sets are as follows:

[0051]

[0052]

[0053] in, These represent the active and reactive power outputs of the new energy generator unit j during time period t in scenario s. For the day-ahead predicted output of new energy generating unit j in time period t under scenario s, Ω represents the upper limit of reactive power output of the new energy generator set during time period t, Ω represents the set of scenarios, and RES represents the set of nodes connecting the new energy generator set.

[0054] Furthermore, power flow constraints in the distribution network include:

[0055] Nodal active power balance constraints

[0056]

[0057]

[0058]

[0059]

[0060] Where Θ(j) represents the set of receiving nodes with node j as the sending node. Let be the active power flowing through line (i,j) during time period t in scenario s. Inject active power into the root node during time period t in scenario s, r ij Let l be the resistance of line (i,j). ij,t,s Let be the square of the current amplitude flowing through line (i,j) during time period t in scenario s. Let be the load power of edge data center j during time period t in scenario s. Let N be the load power of node j in scenario s during time period t, and N be the set of node numbers.

[0061] Nodal reactive power balance constraints

[0062]

[0063]

[0064]

[0065] in, Let be the reactive power flowing through line (i,j) during time period t in scenario s. Inject reactive power into the root node during time period t in scenario s, x ij Let (i,j) be the reactance of the line. Let be the load power of node j in scenario s during time period t;

[0066] Line voltage drop constraints:

[0067]

[0068] Among them, l ij,t,s Let v be the square of the current amplitude flowing through line (i, j) during time period t in scenario s. i,t,sLet be the square of the voltage amplitude of node j in time period t under scenario s;

[0069] Constraints on the relationship between line power, current and node voltage:

[0070]

[0071] Node voltage amplitude constraints:

[0072]

[0073] in, These represent the upper and lower limits of the allowable voltage amplitude at node j, respectively.

[0074] Line current constraints:

[0075]

[0076] Line power constraints:

[0077]

[0078] in, This represents the maximum active power allowed to be transmitted by line (i, j).

[0079] Specifically, in step S3, the scheduling results include:

[0080] The operational status and energy consumption of edge data centers, the optimal scheduling and operation mode of the power distribution network, and the economic indicators of power distribution network operation including edge data centers.

[0081] Secondly, embodiments of the present invention provide an active power distribution network optimization scheduling system considering edge data center clusters, characterized in that it includes:

[0082] Build modules to construct an edge data center cluster energy consumption model, including a load spatiotemporal migration model and an edge data center energy consumption model;

[0083] The optimization module constructs an active distribution network operation optimization model that takes into account the spatiotemporal flexibility of edge data center clusters, and constructs the constraints of the active distribution network operation optimization model based on the energy consumption model of edge data center clusters obtained from the construction module.

[0084] The scheduling module solves the active distribution network operation optimization model that takes into account the spatiotemporal flexibility of the edge data center cluster, obtained by the optimization module, and obtains the active distribution network optimization scheduling result considering the edge data center cluster.

[0085] Compared with the prior art, the present invention has at least the following beneficial effects:

[0086] This invention considers an active distribution network optimization scheduling method for edge data center clusters. First, a spatiotemporal migration model of the computing load is constructed to model the allocation and migration process of the computing load. Then, combined with the basic model of the computing load, the coupling relationship between computing demand and power demand is modeled, constructing an energy consumption model for the edge data center. Next, integrating the basic models established in the preceding steps, an active distribution network operation optimization model considering the spatiotemporal flexibility of the edge data center cluster is constructed. Finally, a second-order cone programming solver is used to solve the model, obtaining the optimized scheduling result of the active distribution network. This invention models the spatiotemporal migration flexibility of edge data centers, and the constructed active distribution network optimization scheduling model can be directly and quickly calculated with the support of a second-order cone programming solver. The results calculated by this method can be directly applied to the scheduling practice of active distribution networks containing edge data centers, optimizing the operation mode of the distribution system, making full use of network paths for computing load migration, reducing distribution network line loss costs and penalties for renewable energy curtailment.

[0087] Furthermore, constructing a load spatiotemporal migration model aims to fully exploit the spatiotemporal flexibility of computational loads and is a necessary step in establishing the active distribution network optimization scheduling model in subsequent processes. Modeling the temporal mobility of computational loads prepares the necessary groundwork for the time transfer of computational loads in distribution network optimization scheduling, making it possible to absorb redundant renewable energy generation through the time transfer of computational loads. Modeling the spatial mobility of computational loads prepares the necessary groundwork for the spatial transfer of computational loads in distribution network optimization scheduling, making it possible to reduce line power losses by utilizing network links to achieve spatial transfer of computational loads.

[0088] Furthermore, in the actual dispatching process of the power system, it is necessary to first characterize the load energy consumption in the system, and then calculate the operation mode of the power system through an optimized dispatching model. As a special load in the active distribution network, the edge data center needs to establish a basic energy consumption model so that the optimized dispatching calculation results of the distribution network are close to the actual operation of the power network.

[0089] Furthermore, the main purpose of power system regulation and operation management is to ensure the economic and safe operation of the system. Therefore, minimizing the total operating cost of the system as the optimization objective of active distribution network operation optimization calculation is in line with the actual needs of power system dispatch.

[0090] Furthermore, the constraints of the active distribution network operation optimization model need to include constraints related to the edge data center cluster, active and reactive power output constraints of new energy generator units, and power flow constraints of the distribution network. The purpose is to ensure that the scheduling and operation of the active distribution network meets the basic requirements of power system operation safety, make the model calculation results closer to the actual operation of the active distribution network, avoid the overload operation of power equipment, and ensure the reliability of power supply from the distribution network.

[0091] Furthermore, edge data center clusters handle computing loads and serve as connecting elements between power networks and communication networks. Constraining the migration and processing of computing loads ensures that the operation of edge data centers meets the spatiotemporal migration characteristics of the computing loads; constraining the power consumption of the computing loads quantifies the energy requirements of edge data centers, providing a reference for the scheduling and operation of active distribution networks, and making the calculation results more closely reflect the actual operating conditions of edge data centers.

[0092] Furthermore, constructing constraints on the active and reactive power outputs of new energy generator units is a necessary step in executing subsequent calculations. In actual production, the active and reactive power outputs of new energy power units are limited by power generation resources. Therefore, it is necessary to reflect the operating characteristics of new energy generator units in the unit operating constraints of active distribution network optimization scheduling, so that the calculation results are close to the actual situation of the power system and meet actual scheduling needs.

[0093] Furthermore, constructing the constraints related to power flow in the distribution network is a necessary step in executing subsequent calculations. In actual production, the physical characteristics of power transmission in the power network can be characterized by a power flow model. Each node satisfies the constraints of active power balance, reactive power balance, and node voltage amplitude. The power flow of each distribution line is related to the voltage of its two ends and is limited by the line's power transmission capacity, as well as the relationship between line power, current, and node voltage. These power network transmission characteristics need to be reflected in the network constraints of the distribution network dispatching and operation optimization calculations to ensure that the calculation results closely approximate the actual distribution of power flow in the power network and meet dispatching requirements.

[0094] Furthermore, the calculation results of this invention include the operational status and energy consumption of edge data centers, the optimal dispatch operation mode of the distribution network, and the economic indicators of the distribution network operation containing edge data centers, which is both necessary and reasonable. Analyzing the calculation results of the operational status and energy consumption of edge data centers allows us to obtain the operation mode and energy consumption level of the edge data center cluster, providing a reference for formulating the operation mode of the distribution network. The calculation results of the optimal dispatch mode of the distribution network represent the optimal dispatch operation of the active distribution network under specific scenarios, providing a reference for actual power system dispatchers. By comparing and analyzing the economic indicators of distribution network operation, the economics of different distribution network operation modes can be evaluated, resulting in the most economical and efficient power system operation mode.

[0095] It is understandable that the beneficial effects of the second aspect mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here.

[0096] In summary, this invention can fully leverage the spatiotemporal flexibility of energy consumption in edge data center clusters, reduce power distribution line overload, and improve the ability of active power distribution networks to accept new energy sources.

[0097] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0098] Figure 1 The flowchart for the active power distribution network optimization scheduling method considering edge data center clusters is shown below.

[0099] Figure 2 This is a structural diagram of the IEEE-33 node power distribution system after the modification.

[0100] Figure 3 This is a diagram of the link structure of an edge data center cluster.

[0101] Figure 4 The graph shows the output of wind turbines and photovoltaic units and the arrival rate of interactive tasks in typical scenarios. (a) represents the output of wind turbines, (b) represents the output of photovoltaic units, and (c) represents the arrival rate of interactive tasks.

[0102] Figure 5 A graph showing the cumulative power consumption exceeding the safety threshold in power distribution network lines under various scenarios;

[0103] Figure 6 This is a graph showing the reduction in renewable energy power generation under various scenarios. Detailed Implementation

[0104] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0105] In the description of this invention, it should be understood that the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0106] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0107] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this document generally indicates that the preceding and following objects have an "or" relationship.

[0108] It should be understood that although terms such as first, second, third, etc., may be used in the embodiments of the present invention to describe the preset range, these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from one another. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0109] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0110] The accompanying drawings illustrate various structural schematic diagrams according to embodiments disclosed in this invention. These drawings are not to scale, and some details have been enlarged for clarity, and some details may have been omitted. The shapes of the various regions and layers shown in the drawings, as well as their relative sizes and positional relationships, are merely exemplary and may deviate from reality due to manufacturing tolerances or technical limitations. Furthermore, those skilled in the art can design regions / layers with different shapes, sizes, and relative positions as needed.

[0111] To fully leverage the spatiotemporal flexibility of optimized computing load scheduling, edge data center clusters can participate in proactive power grid optimization scheduling through demand response, dynamically adjusting workloads to reduce data center energy costs and eliminate power grid congestion. Alternatively, they can dynamically adjust the computing load of edge data centers to change the load level of the data center's region, matching renewable energy output and maximizing the utilization of renewable energy.

[0112] Please see Figure 1 The present invention provides an active power distribution network optimization scheduling method considering edge data center clusters, comprising the following steps:

[0113] S1. Construct an energy consumption model for edge data center clusters, including a computing load spatiotemporal migration model and an edge data center energy consumption model;

[0114] First, relevant data is obtained from relevant departments, including: power distribution network system parameters, distributed generation equipment parameters and day-ahead forecast data, edge data center cluster data, computing load data, and electricity market data. Based on the power distribution network system parameters, distributed generation equipment parameters and day-ahead forecast data, edge data center cluster data, computing load data, and electricity market data, an energy consumption model for the edge data center cluster is constructed.

[0115] Distribution network system parameters: distribution network topology, impedance data of each branch of the distribution network system, active and reactive load data of each node of the distribution network system.

[0116] Distributed generation parameters and day-ahead forecast data: the connection location of new energy distributed generator units in the distribution network, the installed capacity and day-ahead forecast output of distributed photovoltaic power generation at each node; the installed capacity and day-ahead forecast output of distributed wind power generation at each node.

[0117] Edge data center cluster parameters: the access location of the edge data center in the distribution network, the number of servers equipped in each edge data center, the power utilization efficiency of each edge data center, the average service rate of each server, the idle power consumption of each server, the peak power consumption of each server, and the link bandwidth capacity.

[0118] Calculate relevant data for the workload: allowable latency of interactive tasks, arrival rate of interactive tasks, and migration cost coefficient of interactive tasks.

[0119] Electricity market data: peak, flat, and valley electricity price data, demand response periods.

[0120] 1. Load Spatiotemporal Migration Model

[0121] In an edge data center, all servers are homogeneous, and both interactive and batch processing tasks have homogeneous computational loads. The arrival process of interactive tasks follows a Poisson flow, and the time of interactive tasks follows an exponential distribution. Since the homogeneous servers work independently, the arrival process of interactive tasks can be represented by an M / M / 1 queuing model.

[0122] The average dwell time of the M / M / 1 queuing model is expressed as:

[0123]

[0124] Where λ represents the average number of customers arriving per unit time; μ represents the number of customers served per unit time, i.e., the average service rate; and W represents the average time a customer stays in the system, which is the sum of waiting time and service time.

[0125] The computing power of edge data centers needs to meet QoS (Quality of Service) constraints, meaning that the computing load latency cannot exceed the maximum allowable latency. Here, the computing load latency is the average dwell time in the M / M / 1 queuing model. Therefore, the QoS constraints for interactive tasks are expressed as follows:

[0126]

[0127] Where E is the set of nodes connecting the edge data center, and T is the set of time period values. The number of active servers in edge data center j processing interactive tasks during time period t. τ represents the interactive task load carried by edge data center j during time period t after migration. I The maximum allowable latency for interactive tasks, Interactive tasks assigned to each active server.

[0128] Convert QoS into a constraint representing the number of active servers, i.e.:

[0129]

[0130] Edge data centers are connected via optical networks. When computing loads migrate between edge data centers, it is necessary to ensure that all loads are processed. The load balancing constraint is expressed as:

[0131]

[0132] in, This represents the interactive task load that migrates from edge data center j to edge data center i during time period t. This indicates an interactive task that has not been migrated. The arrival rate of interactive tasks unloaded to edge data center j during time period t.

[0133] The total number of interactive tasks processed per unit time in edge data center i is represented as:

[0134]

[0135] The amount of interactive task migration cannot exceed the link bandwidth limit. The bandwidth constraint is expressed as follows:

[0136]

[0137] in, This represents the upper limit of the link bandwidth between edge data centers i and j. If there is no link connection between the two edge data centers, then...

[0138] Compared to interactive tasks, batch processing tasks involve significantly more computation, but are less sensitive to latency and can be processed for several hours. Cloud service providers can flexibly distribute batch processing tasks across different time periods as needed. Considering the large data volume of batch processing tasks, spatial migration would consume a significant amount of network bandwidth; therefore, batch processing tasks are all processed locally.

[0139] The QoS constraints for batch processing tasks are expressed as follows:

[0140]

[0141] in, The maximum allowable latency for batch task l to be offloaded to edge data center j. To offload the batch processing task load to the edge data center j, Let be the batch processing task load of the j-th edge data center during time period i after time migration, where

[0142] The total number of batch processing tasks processed per unit time in the edge data center Represented as:

[0143]

[0144] If all servers carrying batch processing tasks are running at peak performance, then the number of active servers carrying batch processing tasks is represented as:

[0145]

[0146] The number of active servers handling interactive and batch processing tasks cannot exceed the total number of servers in the edge data center, i.e.:

[0147]

[0148] Among them, S j Let j be the total number of servers in the edge data center.

[0149] Edge data centers have a large number of servers, including a significant number of active servers handling both interactive and batch processing tasks. and Relaxation is a continuous variable.

[0150] 2. Edge Data Center Energy Consumption Model

[0151] Servers are the primary energy-consuming units in IT equipment. This invention treats server power consumption as the power consumption of IT equipment, employing a univariate linear regression model based on server utilization to model server power consumption as a function of CPU utilization, i.e.:

[0152] P Server =P Idle +(P Peak -P Idle )ρ CPU

[0153] Among them, P Server For server power consumption; P Idle P represents the server's idle power consumption. Peak This represents the server's peak power consumption; ρ CPU This refers to CPU utilization.

[0154] Precisely calculating the energy consumption of other components in an edge data center, such as cooling and power distribution systems, is very difficult. Generally, the energy consumption of IT equipment is used for indirect calculation. The Power Usage Effectiveness (PUE) standard, developed by The Green Grid, is commonly used to calculate the total energy consumption of a data center, defined as the ratio of total data center energy consumption to IT equipment energy consumption. A lower PUE indicates a higher energy efficiency ratio for the data center, meaning a higher proportion of IT equipment energy consumption in the overall data center energy consumption.

[0155] This invention uses PUE to calculate the power consumption of edge data centers, expressed as:

[0156] P EDC =ηP Server

[0157] Among them, P EDC Let η be the total energy consumption of the edge data center, and η be the power utilization efficiency of the edge data center.

[0158] Based on the above analysis, the overall power consumption of the edge data center can be derived by using the server power consumption model and the PUE of the edge data center.

[0159] The power consumption per unit time of an active server processing interactive tasks is expressed as:

[0160] P I =(P Idle +(P Peak -P Idle )ρ I )x I

[0161] Among them, P I Total power consumption of active servers for processing interactive tasks; ρ Iρ represents the CPU utilization of active servers handling interactive tasks. I =λ I / (μx I This refers to the ratio of server interaction task load to average service rate; x I The number of active servers handling interactive tasks. ρ I x I Replace with λ I / μ, then the power consumption per unit time of the active server processing interactive tasks can be converted to:

[0162] P I =P Idle x I +(P Peak -P Idle )λ I / μ

[0163] The servers handling batch processing tasks are all running at peak performance. The power consumption of these active servers per unit time is expressed as:

[0164] P B =P Peak x B =P Idle x B +(P Peak -P Idle )λ B / μ

[0165] Among them, P B Total power consumption of active servers for processing batch tasks.

[0166] The power consumption of the edge data center is expressed as:

[0167] P EDC =η(P Idle (x I +x B )+(P Peak -P Idle )(λ I +λ B ) / μ).

[0168] S2. Construct an active power distribution network operation optimization model that takes into account the spatiotemporal flexibility of edge data center clusters;

[0169] S201. Establish active distribution network operation optimization objectives;

[0170] The active distribution network operation, which takes into account the spatiotemporal flexibility of edge data center clusters, aims to minimize the total operating cost, including the energy consumption cost of edge data center clusters, the migration cost of interactive tasks, and the distribution network line loss cost. It also adds penalties for the reduction of renewable energy sources and the overload of distribution lines.

[0171]

[0172] Where s is the scene number, Ω is the scene set, and π s Let be the probability of scenario s occurring. The energy cost for edge data center clusters in scenario s. The cost of migrating interactive tasks in scenario s. For the distribution network line loss cost in scenario s, To reduce penalties for new energy vehicles in scenario S. The penalty cost for overload of distribution network lines in scenario s.

[0173] The various costs and expenses are represented as follows:

[0174]

[0175]

[0176]

[0177] Where T is the set of time period values, E is the set of nodes connecting the edge data center, and B is the set of branches. The wholesale market electricity price for time period t. Let c be the load power of edge data center j during time period t in scenario s. Tr The cost coefficient for migrating interactive tasks. Let r be the interaction task load during time period t when migrating from edge data center i to edge data center j in scenario s. ij Let l be the resistance of line (i,j). ij,t,s Let be the square of the current amplitude flowing through line (i,j) during time period t in scenario s.

[0178] The various penalty fees are listed as follows:

[0179]

[0180]

[0181] Where RES is the set of new energy generator sets or the set of nodes connecting new energy generator sets, and λ Curt Reduce the penalty coefficient for new energy sources; λ Cong This is the line overload penalty coefficient; Reduce the power output of the new energy generator unit j during time period t in scenario s; Let be the transmission power of line (i,j) exceeding the safety threshold during time period t in scenario s.

[0182] The power reduction in renewable energy generation is expressed as follows:

[0183]

[0184] in, For the predicted output of renewable energy power generation j days before time t in scenario s, The active power output of the new energy generator unit j during time period t.

[0185] Overload power is expressed as:

[0186]

[0187] in, Let be the active power flowing through line (i,j) during time period t. Let (i,j) be the safe threshold for transmission power of line. β represents the maximum transmission capacity of the line, and β represents the safety margin of the line.

[0188] S202, Construct constraints;

[0189] 1. Constraints related to edge data center clusters, including: interactive task constraints, batch processing task constraints, edge data center computing power constraints, and edge data center power consumption constraints;

[0190] Interactive task constraints:

[0191]

[0192]

[0193]

[0194]

[0195] Batch processing task constraints:

[0196]

[0197]

[0198]

[0199] Edge data center computing capacity constraints:

[0200]

[0201] Power consumption constraints for edge data centers:

[0202]

[0203] 2. Constraints on active and reactive power output of new energy generator sets;

[0204] The upper limit of active power output of new energy generator sets is the day-ahead predicted output. Since the operating state of the inverter is limited by the maximum output current, it is necessary to consider the apparent power upper limit of wind turbines and photovoltaic units. Therefore, the active and reactive power output constraints of new energy generator sets are expressed as follows:

[0205]

[0206]

[0207] in, These represent the active and reactive power outputs of the new energy generator unit j during time period t in scenario s. For the day-ahead predicted output of new energy generating unit j in time period t under scenario s, This represents the upper limit of reactive power output of the new energy generator units during time period t, and has... in, These are the upper limits of the active power and apparent power of the new energy generator set j, respectively.

[0208] 3. Distribution network power flow constraints, including: node active power balance constraints, node reactive power balance constraints, line voltage drop constraints, second-order cone relaxation constraints on the relationship between line power, current and node voltage, node voltage amplitude constraints, line current constraints, and line power constraints.

[0209] Nodal active power balance constraints

[0210]

[0211]

[0212]

[0213]

[0214] Where Θ(j) represents the set of receiving nodes with node j as the sending node. Let be the active power flowing through line (i,j) during time period t in scenario s. Inject active power into the root node during time period t in scenario s, r ij Let l be the resistance of line (i,j). ij,t,s Let be the square of the current amplitude flowing through line (i,j) during time period t in scenario s. Let be the load power of edge data center j during time period t in scenario s. Let N be the load power of node j in scenario s during time period t, and N be the set of node numbers.

[0215] Nodal reactive power balance constraints

[0216]

[0217]

[0218]

[0219] in, Let be the reactive power flowing through line (i,j) during time period t in scenario s. Inject reactive power into the root node during time period t in scenario s, x ij Let (i,j) be the reactance of the line. Let be the load power of node j during time period t in scenario s.

[0220] Edge data center clusters utilize reactive power compensation equipment for local reactive power compensation, resulting in zero reactive power load for all data center clusters.

[0221] Line voltage drop constraint

[0222]

[0223] Among them, l ij,t,s Let v be the square of the current amplitude flowing through line (i,j) during time period t in scenario s. i,t,s Let be the square of the voltage amplitude of node j in time period t under scenario s.

[0224] Constraints on the Relationship between Line Power, Current and Node Voltage

[0225]

[0226] The above equation will result in a non-convex optimization problem, which needs to be further relaxed to:

[0227]

[0228] Transforming the inequality into a second-order cone constraint and converting the quadratic term into a linear term, we get:

[0229]

[0230] Node voltage amplitude constraints

[0231]

[0232] in, These represent the upper and lower limits of the allowable voltage amplitude at node j, respectively.

[0233] Line current constraint

[0234]

[0235] Line power constraints

[0236]

[0237] in, This represents the maximum active power allowed to be transmitted by line (i, j).

[0238] S3. Solve the active distribution network operation optimization model that takes into account the spatiotemporal flexibility of the edge data center cluster obtained in step S2 to obtain the active distribution network optimization scheduling result considering the edge data center cluster.

[0239] 1. The operational status and energy consumption of the edge data center;

[0240] 2. Optimal dispatching and operation mode of distribution network;

[0241] 3. Economic indicators for the operation of power distribution networks including edge data centers.

[0242] In another embodiment of the present invention, an active distribution network optimization scheduling system considering edge data center clusters is provided. This system can be used to implement the above-mentioned active distribution network optimization scheduling method considering edge data center clusters. Specifically, the active distribution network optimization scheduling system considering edge data center clusters includes a construction module, an optimization module, and a scheduling module.

[0243] Among them, the building module constructs an edge data center cluster energy consumption model, including a load spatiotemporal migration model and an edge data center energy consumption model;

[0244] The optimization module constructs an active distribution network operation optimization model that takes into account the spatiotemporal flexibility of edge data center clusters, and constructs the constraints of the active distribution network operation optimization model based on the energy consumption model of edge data center clusters obtained from the construction module.

[0245] The scheduling module solves the active distribution network operation optimization model that takes into account the spatiotemporal flexibility of the edge data center cluster, obtained by the optimization module, and obtains the active distribution network optimization scheduling result considering the edge data center cluster.

[0246] In another embodiment of the present invention, a terminal device is provided, comprising a processor and a memory. The memory stores a computer program, the computer program including program instructions, and the processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions to implement corresponding method flows or corresponding functions. The processor described in this embodiment of the present invention can be used to consider the operation of an active distribution network optimization scheduling method for edge data center clusters, including:

[0247] An energy consumption model for an edge data center cluster, including a load spatiotemporal migration model and an edge data center energy consumption model, is constructed. An active distribution network operation optimization model considering the spatiotemporal flexibility of the edge data center cluster is constructed, and constraints for the active distribution network operation optimization model are constructed based on the edge data center cluster energy consumption model. The active distribution network operation optimization model considering the spatiotemporal flexibility of the edge data center cluster is solved to obtain the active distribution network optimization scheduling results considering the edge data center cluster.

[0248] In another embodiment of the present invention, a storage medium is also provided, specifically a computer-readable storage medium (memory). This computer-readable storage medium is a memory device in a terminal device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and extended storage media supported by the terminal device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device.

[0249] One or more instructions stored in a computer-readable storage medium can be loaded and executed by a processor to implement the corresponding steps of the active distribution network optimization scheduling method considering edge data center clusters in the above embodiments; one or more instructions in the computer-readable storage medium are loaded and executed by the processor in the following steps:

[0250] An energy consumption model for an edge data center cluster, including a load spatiotemporal migration model and an edge data center energy consumption model, is constructed. An active distribution network operation optimization model considering the spatiotemporal flexibility of the edge data center cluster is constructed, and constraints for the active distribution network operation optimization model are constructed based on the edge data center cluster energy consumption model. The active distribution network operation optimization model considering the spatiotemporal flexibility of the edge data center cluster is solved to obtain the active distribution network optimization scheduling results considering the edge data center cluster.

[0251] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0252] The test case is a modified IEEE-33 node system, as shown in the attached figure. Figure 2 As shown, general loads are not depicted. Detailed parameters of the IEEE-33 node system are shown in Tables 1 and 2. The edge data center cluster is located at nodes 21, 30, 10, and 16, wind turbines at nodes 9, 12, 15, and 29, and photovoltaic units at nodes 10, 16, 30, and 31. The link structure of the edge data center cluster is attached. Figure 3 As shown, the link bandwidth capacity is 5000 requests / s, and the migration cost coefficient for interactive tasks is 0.002 yuan / request / s. The edge data center clusters all have a PUE of 1.5, each with 2500 homogeneous servers. The peak power of a single server is 400W, the idle power is half of the peak power, and the average service rate is 20 requests / s. Typical scenarios for new energy generator output and interactive task arrival rates are attached. Figure 4As shown in Table 3, the allowed latency for interactive tasks is 100ms. Additionally, one batch processing task is offloaded to edge data centers 2, 3, and 4 respectively. To ensure the safe operation of the distribution network, the node voltage deviation does not exceed ±5%. The distribution network line capacity is shown in Table 4. The transmission capacity of lines closer to the root node is higher than that of lines farther from the root node, with a safety margin β = 0.8. The penalty coefficient for reducing new energy generation is 700 yuan / MWh, the line overload penalty coefficient is 300 yuan / MWh, and the electricity price is shown in Table 5.

[0253] Table 1 Branch impedance data of IEEE-33 node distribution network system

[0254]

[0255]

[0256] Table 2 Node load data of IEEE-33 node distribution network system

[0257]

[0258] Table 3 Batch Processing Task Parameters

[0259]

[0260] Table 4 Power Distribution System Line Capacity

[0261]

[0262] To verify the effectiveness of edge data center clusters in alleviating distribution network line overload, improving renewable energy absorption, and reducing distribution network line losses, four different scenarios were set up in the case study to investigate the impact of spatial migration of interactive tasks and temporal migration of batch processing tasks on distribution network operation:

[0263] Scenario 1 assumes that the spatial migration of interactive tasks between edge data center clusters is not considered, while batch processing tasks are processed with maximum computing power.

[0264] Scenario 2 considers the spatial migration of interactive tasks between edge data center clusters, but does not consider the temporal migration of batch processing tasks.

[0265] Scenario 3 does not consider the spatial migration of interactive tasks between edge data center clusters, but considers the migration of batch processing tasks in the time dimension.

[0266] Scenario 4 considers the spatiotemporal migration of two types of computing loads simultaneously.

[0267] According to the appendix Figure 5It can be seen that the cumulative overload of power exceeding the safety threshold for the entire day in Scenario 1 reached 9.02 MWh. Compared to Scenario 1, the cumulative overload of power for the entire day in Scenarios 2, 3, and 4 were 5.25 MWh, 3.80 MWh, and 1.57 MWh, respectively, representing reductions of 66.15%, 19.41%, and 77.06%. Interactive tasks adjust the computing load distribution of the edge data center cluster through spatial migration, flexibly adjusting the load level of the edge data center cluster according to the output of new energy power generation, mitigating line overload caused by large-scale new energy generation and excessive load. Batch processing tasks achieve flexible transfer of edge data center load in the time dimension through time migration, mitigating line overload while responding to changes in electricity prices, reducing system energy consumption costs, and improving operational economy. It is evident that the spatiotemporal migration of computing load can significantly reduce the degree of line overload in the distribution network.

[0268] According to the appendix Figure 6 As can be seen, the changes in renewable energy reduction are not significant in Scenario 1 and Scenario 3, and the time migration of batch processing tasks has no significant effect on reducing renewable energy reduction. The renewable energy reduction in Scenario 2 and Scenario 4 is much lower than that in Scenario 1, indicating that the spatial migration of interactive tasks has a significant effect on improving the active distribution network's ability to accept renewable energy.

[0269] Table 6 Distribution Network Operation Indicators under Various Scenarios

[0270]

[0271] Table 6 shows the distribution network operation indicators for each scenario. It can be seen that, compared with scenario one, the total operating cost of the distribution network in scenarios two, three and four has been reduced to varying degrees.

[0272] The above examples illustrate that the present invention can reduce the overload of distribution network lines and improve the ability of active distribution networks to accept new energy sources.

[0273] In summary, the present invention provides an active distribution network optimization scheduling method and system that considers edge data center clusters, which can fully leverage the spatiotemporal flexibility of energy consumption of edge data center clusters, reduce distribution network line overload, and improve the active distribution network's ability to accept new energy sources.

[0274] The above content is only for illustrating the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the scope of protection of the claims of this invention.

Claims

1. A method for active power distribution network optimization scheduling considering edge data center clusters, characterized in that, Includes the following steps: S1. Construct an energy consumption model for an edge data center cluster, including a load spatiotemporal migration model and an edge data center energy consumption model; S2. Construct an active distribution network operation optimization model that considers the spatiotemporal flexibility of edge data center clusters. Based on the energy consumption model of the edge data center cluster obtained in step S1, construct the constraints of the active distribution network operation optimization model. The active distribution network operation optimization model takes minimizing the total operating cost as its optimization objective, specifically: in, s Number the scene. For scene collection, for s The probability of the scene occurring for s Energy costs for edge data center clusters in this scenario for s Cost of migrating interactive tasks in different scenarios for s Distribution network line loss costs in this scenario for s Reduce penalties for new energy vehicles in this scenario. for s Overload penalty cost for power distribution lines in this scenario; The constraints of the active distribution network operation optimization model include constraints related to the edge data center cluster, active and reactive power output constraints of new energy generating units, and power flow constraints of the distribution network. The constraints related to the edge data center cluster include: Interactive task constraints: in, For the scene s Mid-term t Edge data centers j The number of active servers processing interactive tasks. For the scene s middle t Time-edge data centers j The workload of interactive tasks it carries. For edge data centers j The number of customers served per unit of time. The maximum allowable latency for interactive tasks, A collection of edge data centers or a collection of nodes connecting edge data centers. For a set of values ​​for a given time period, For scene collection, For the scene s middle t Time period from edge data center i Migration to edge data center j Interactive task load, For the scene s middle t Time-based offloading to edge data centers j Interactive task arrival rate For the scene s middle t Time-edge data centers i The workload of interactive tasks it carries. For edge data centers i and j The maximum bandwidth limit between links; Batch processing task constraints: in, To offload to edge data centers j Batch processing tasks l Maximum allowable delay For the scene s After migration t Time-edge data centers j Batch processing tasks l Load capacity For the scene s Offloading to edge data centers j Batch processing tasks l Load capacity For the scene s After migration t Time-edge data centers j The load of all batch processing tasks it supports. For the scene s middle t Time-edge data centers j The number of active servers hosting batch processing tasks. For edge data centers j The number of customers served per unit of time; Edge data center computing capacity constraints: Power consumption constraints for edge data centers: in, For the scene s Mid-term t Edge data centers j The load power, For edge data centers j The efficiency of electrical energy utilization For edge data centers j Server idle power consumption For the scene s Mid-term t Edge data centers j The number of active servers processing interactive tasks. For the scene s middle t Time-edge data centers j The number of active servers hosting batch processing tasks. For edge data centers j The peak power consumption of the server For the scene s middle t Time-edge data centers j The workload of the interactive tasks it carries; The specific constraints on the active and reactive power output of new energy generator sets are as follows: in, They are respectively s In the scene t Periodic new energy generator sets j Efforts made, whether productive or not. for s In the scene t Periodic new energy generator sets j The recent forecast has been effective. for t Periodic new energy generator sets j The upper limit of reactive power output, For scene collection, A set of nodes connecting new energy generator sets; Distribution network power flow constraints include: Nodal active power balance constraints in, Represented by node j Let the set of receiving nodes be the sending end. for s In the scene t Time period flow line (i, j) active power, for s In the scenario where the root node injects active power during time period t, For the line (i, j) The resistance, for s In the scene t Time period flow line (i, j) The square of the current amplitude, for s In the scene t Time-edge data centers j The load power, for s In the scene t Time period nodes j The load power, N A set of node numbers; Nodal reactive power balance constraints in, for s In the scene t Time period flow line (i, j) reactive power, for s In the scene t Reactive power is injected into the root node of the time period. For the line (i, j) Reactance, for s In the scene t Time period nodes j The load power; Line voltage drop constraints: in, for s In the scene t Time period flow line (i , j) The square of the current amplitude, for s In the scene t Time period nodes j The square of the voltage amplitude; Constraints on the relationship between line power, current and node voltage: Node voltage amplitude constraints: in, They are nodes j Permissible upper / lower limits of voltage amplitude; Line current constraints: Line power constraints: in, For the line (i , j) The maximum amount of active power that can be transmitted; S3. Solve the active distribution network operation optimization model that takes into account the spatiotemporal flexibility of the edge data center cluster obtained in step S2 to obtain the active distribution network optimization scheduling result considering the edge data center cluster.

2. The active distribution network optimization scheduling method considering edge data center clusters according to claim 1, characterized in that, In step S1, the constraints of the load spatiotemporal migration model are as follows: Load balancing constraints that should be satisfied during interactive task migration: in, E For a set of nodes connecting edge data centers, T For a set of values ​​for a given time period, express t Time period from edge data center j Migration to edge data center i Interactive task load, This indicates an interactive task that has not been migrated. for t Time-based offloading to edge data centers j Interactive task arrival rate; The amount of interactive task migration cannot exceed the link bandwidth limit. Bandwidth constraints: in, For edge data centers i and j The maximum bandwidth limit between links; Edge data centers j Total interactive tasks processed per unit of time: in, For the post-migration t Time-edge data centers j The workload of the interactive tasks it carries; Batch processing task workload constraints: in, To offload to edge data centers j Batch processing tasks l Maximum allowable delay To offload to edge data centers j Batch processing tasks l Load capacity After time migration, the first j An edge data center in i Batch processing tasks carried by the time period l Load; Edge data centers j Total batch processing tasks processed per unit time: in, For edge data centers j The total number of batch processing tasks processed per unit of time.

3. The active distribution network optimization scheduling method considering edge data center clusters according to claim 1, characterized in that, In step S1, the energy consumption model for the edge data center is as follows: in, This represents the total energy consumption of the edge data center. For the power utilization efficiency of edge data centers, This represents the server's idle power consumption. The number of active servers for handling interactive tasks. The number of active servers that can handle batch processing tasks. This represents the server's peak power consumption. The amount of interactive task load it can handle. The load of batch processing tasks it can handle. This refers to the number of customers served per unit of time.

4. The active distribution network optimization scheduling method considering edge data center clusters according to claim 1, characterized in that, In step S3, the scheduling results include: The operational status and energy consumption of edge data centers, the optimal scheduling and operation mode of the power distribution network, and the economic indicators of power distribution network operation including edge data centers.

5. An active power distribution network optimization scheduling system considering edge data center clusters, characterized in that, include: Build modules to construct an edge data center cluster energy consumption model, including a load spatiotemporal migration model and an edge data center energy consumption model; The optimization module constructs an active distribution network operation optimization model that considers the spatiotemporal flexibility of edge data center clusters. Based on the energy consumption model of the edge data center cluster obtained from the construction module, it constructs the constraints of the active distribution network operation optimization model. The active distribution network operation optimization model takes minimizing the total operating cost as its optimization objective, specifically: in, s Number the scene. For scene collection, for s The probability of the scene occurring for s Energy costs for edge data center clusters in this scenario for s Cost of migrating interactive tasks in different scenarios for s Distribution network line loss costs in this scenario for s Reduce penalties for new energy vehicles in this scenario. for s Overload penalty cost for power distribution lines in this scenario; The constraints of the active distribution network operation optimization model include constraints related to the edge data center cluster, active and reactive power output constraints of new energy generating units, and power flow constraints of the distribution network. The constraints related to the edge data center cluster include: Interactive task constraints: in, For the scene s Mid-term t Edge data centers j The number of active servers processing interactive tasks. For the scene s middle t Time-edge data centers j The workload of interactive tasks it carries. For edge data centers j The number of customers served per unit of time. The maximum allowable latency for interactive tasks, A collection of edge data centers or a collection of nodes connecting edge data centers. For a set of values ​​for a given time period, For scene collection, For the scene s middle t Time period from edge data center i Migration to edge data center j Interactive task load, For the scene s middle t Time-based offloading to edge data centers j Interactive task arrival rate For the scene s middle t Time-edge data centers i The workload of interactive tasks it carries. For edge data centers i and j The maximum bandwidth limit between links; Batch processing task constraints: in, To offload to edge data centers j Batch processing tasks l Maximum allowable delay For the scene s After migration t Time-edge data centers j Batch processing tasks l Load capacity For the scene s Offloading to edge data centers j Batch processing tasks l Load capacity For the scene s After migration t Time-edge data centers j The load of all batch processing tasks it supports. For the scene s middle t Time-edge data centers j The number of active servers hosting batch processing tasks. For edge data centers j The number of customers served per unit of time; Edge data center computing capacity constraints: Power consumption constraints for edge data centers: in, For the scene s Mid-term t Edge data centers j The load power, For edge data centers j The efficiency of electrical energy utilization For edge data centers j Server idle power consumption For the scene s Mid-term t Edge data centers j The number of active servers processing interactive tasks. For the scene s middle t Time-edge data centers j The number of active servers hosting batch processing tasks. For edge data centers j The peak power consumption of the server For the scene s middle t Time-edge data centers j The workload of the interactive tasks it carries; The specific constraints on the active and reactive power output of new energy generator sets are as follows: in, They are respectively s In the scene t Periodic new energy generator sets j Efforts made, whether productive or not. for s In the scene t Periodic new energy generator sets j The recent forecast has been effective. for t Periodic new energy generator sets j The upper limit of reactive power output, For scene collection, A set of nodes connecting new energy generator sets; Distribution network power flow constraints include: Nodal active power balance constraints in, Represented by node j Let the set of receiving nodes be the sending end. for s In the scene t Time period flow line (i, j) active power, for s In the scenario where the root node injects active power during time period t, For the line (i, j) The resistance, for s In the scene t Time period flow line (i, j) The square of the current amplitude, for s In the scene t Time-edge data centers j The load power, for s In the scene t Time period nodes j The load power, N A set of node numbers; Nodal reactive power balance constraints in, for s In the scene t Time period flow line (i, j) reactive power, for s In the scene t Reactive power is injected into the root node of the time period. For the line (i, j) Reactance, for s In the scene t Time period nodes j The load power; Line voltage drop constraints: in, for s In the scene t Time period flow line (i , j) The square of the current amplitude, for s In the scene t Time period nodes j The square of the voltage amplitude; Constraints on the relationship between line power, current and node voltage: Node voltage amplitude constraints: in, They are nodes j Permissible upper / lower limits of voltage amplitude; Line current constraints: Line power constraints: in, For the line (i , j) The maximum amount of active power that can be transmitted; The scheduling module solves the active distribution network operation optimization model that takes into account the spatiotemporal flexibility of the edge data center cluster, obtained by the optimization module, and obtains the active distribution network optimization scheduling result considering the edge data center cluster.

6. The active power distribution network optimization and scheduling system considering edge data center clusters according to claim 5, characterized in that, In the module setup, the constraints of the load spatiotemporal migration model are as follows: Load balancing constraints that should be satisfied during interactive task migration: in, E For a set of nodes connecting edge data centers, T For a set of values ​​for a given time period, express t Time period from edge data center j Migration to edge data center i Interactive task load, This indicates an interactive task that has not been migrated. for t Time-based offloading to edge data centers j Interactive task arrival rate; The amount of interactive task migration cannot exceed the link bandwidth limit. Bandwidth constraints: in, For edge data centers i and j The maximum bandwidth limit between links; Edge data centers j Total interactive tasks processed per unit of time: in, For the post-migration t Time-edge data centers j The workload of the interactive tasks it carries; Batch processing task workload constraints: in, To offload to edge data centers j Batch processing tasks l Maximum allowable delay To offload to edge data centers j Batch processing tasks l Load capacity After time migration, the first j An edge data center in i Batch processing tasks carried by the time period l Load; Edge data centers j Total batch processing tasks processed per unit time: in, For edge data centers j The total number of batch processing tasks processed per unit of time.

7. The active power distribution network optimization and scheduling system considering edge data center clusters according to claim 5, characterized in that, In the construction module, the energy consumption model for the edge data center is as follows: in, This represents the total energy consumption of the edge data center. For the power utilization efficiency of edge data centers, This represents the server's idle power consumption. The number of active servers for handling interactive tasks. The number of active servers that can handle batch processing tasks. This represents the server's peak power consumption. The amount of interactive task load it can handle. The load of batch processing tasks it can handle. This refers to the number of customers served per unit of time.

8. The active power distribution network optimization and scheduling system considering edge data center clusters according to claim 5, characterized in that, In the scheduling module, the scheduling results include: The operational status and energy consumption of edge data centers, the optimal scheduling and operation mode of the power distribution network, and the economic indicators of power distribution network operation including edge data centers.

Images