A low-voltage power distribution network resource aggregation and collaborative control method, system and medium

Abstract

The application discloses a low-voltage power distribution network resource aggregation and collaborative control method and system and a medium, which are applied to an edge node in a virtual power plant. The method comprises the following steps: acquiring state information of a plurality of distributed resources, and extracting feature vectors of the distributed resources; clustering the plurality of distributed resources according to the feature vectors, and obtaining a plurality of resource communities; determining an aggregated adjustment capacity according to the feature vectors of the distributed resources in each resource community, and generating an edge aggregation capacity curve corresponding to the edge node and uploading the edge aggregation capacity curve to a master station; receiving an adjustment target issued by the master station, determining a community adjustment task according to the adjustment target and the aggregated adjustment capacity, and generating and issuing control instructions corresponding to each distributed resource. The application can reduce the calculation pressure of the master station side and improve the matching between the control instructions and the actual adjustment capacity of the distributed resources.

Description

Technical Field

[0001] This application relates to the field of power system dispatch and control technology, and in particular to a method, system and medium for resource aggregation and collaborative control in low-voltage distribution networks. Background Technology

[0002] With the advancement of new power system construction, a large number of distributed resources, such as distributed photovoltaic systems, energy storage systems, electric vehicle charging stations, residential air conditioners, flexible building loads, and interruptible industrial and commercial loads, are being connected to the distribution network. While individual resources have small capacities, are geographically dispersed, and experience frequent changes in operating status, they can be aggregated through virtual power plants to form a flexible adjustment capability that can participate in grid dispatching, demand response, ancillary services, and market transactions. Existing virtual power plants mostly adopt a centralized cloud architecture, where a cloud master station collects data from underlying equipment, builds equipment models, and issues control commands.

[0003] When existing virtual power plants connect to a large number of heterogeneous distributed resources, edge nodes typically only undertake data acquisition, forwarding, or simple control execution functions. It is difficult to uniformly represent the real-time status, adjustment capabilities, and constraints of distributed resources and to aggregate them into clusters. As a result, the capability modeling and scheduling calculation of a large number of individual resources are still concentrated on the cloud master station, which not only increases the computing pressure on the cloud, but also makes it difficult for edge nodes to generate matching control commands based on the aggregation and adjustment capabilities of resource clusters. Summary of the Invention

[0004] This application provides a method, system, and medium for low-voltage distribution network resource aggregation and collaborative control, which aims to reduce cloud computing pressure and improve the matching between control commands and actual regulation capabilities of distributed resources when virtual power plants connect to a large number of heterogeneous distributed resources.

[0005] To achieve the above objectives, the first aspect of this application provides a low-voltage distribution network resource aggregation and collaborative control method, applied to edge nodes in a virtual power plant, wherein the distributed resource aggregation control method includes: Obtain the status information of multiple distributed resources corresponding to the edge node, and extract the feature vector corresponding to each distributed resource based on the status information; Based on the feature vectors corresponding to each of the distributed resources, multiple distributed resources are clustered to obtain multiple resource clusters. Based on the feature vectors of distributed resources in each resource community, determine the aggregation and adjustment capabilities corresponding to each resource community; Based on the aggregation and adjustment capabilities corresponding to multiple resource clusters, an edge aggregation capability curve corresponding to the edge node is generated, and the edge aggregation capability curve is uploaded to the main station so that the main station can determine the adjustment target corresponding to the edge node based on the edge aggregation capability curve. The system receives the adjustment target issued by the master station, determines the cluster adjustment task corresponding to each resource cluster based on the adjustment target and the aggregation adjustment capability corresponding to each resource cluster, and generates and issues control instructions corresponding to each distributed resource based on the cluster adjustment task.

[0006] By extracting feature vectors, dividing resource communities, determining aggregation and adjustment capabilities, and generating edge aggregation capability curves for multiple distributed resources at the edge node side, and generating control commands locally based on the adjustment targets issued by the master station, this application can reduce cloud computing pressure and improve the matching between control commands and the actual adjustment capabilities of distributed resources compared to the existing technology where the master station directly models and controls a large number of individual distributed resources.

[0007] Further, based on the state information, feature vectors corresponding to each of the distributed resources are extracted, including: The state information is preprocessed to obtain preprocessed state information corresponding to each of the distributed resources; Based on the preprocessed state information, determine one or more of the following characteristics for each of the distributed resources: real-time power characteristics, adjustable power characteristics, response time characteristics, adjustment cost characteristics, constraint characteristics, and reliability characteristics. Based on one or more of the real-time power characteristics, adjustable power characteristics, response time characteristics, adjustment cost characteristics, constraint characteristics, and reliability characteristics, a feature vector corresponding to each of the distributed resources is generated.

[0008] By preprocessing the state information of multi-source heterogeneous distributed resources and extracting one or more of the following features for different types of resources: real-time power features, adjustable power features, response time features, adjustment cost features, constraint features, and reliability features, this application can achieve unified feature modeling of multi-source heterogeneous resources and form a unified resource feature expression suitable for clustering and aggregation calculations, compared with the existing technology where the state of resources from different sources and of different types is difficult to express uniformly.

[0009] Furthermore, based on the feature vectors corresponding to each of the distributed resources, the distributed resources are clustered to obtain multiple resource clusters, including: Multiple candidate cluster numbers are obtained, and based on each candidate cluster number, the feature vectors of multiple distributed resources are pre-clustered to obtain the candidate clustering results corresponding to each candidate cluster number. The clustering results include multiple communities and their corresponding community centers. Based on the feature differences between the feature vectors of distributed resources and the corresponding community centers in each candidate clustering result, and the preset penalty coefficient, the clustering evaluation value corresponding to each candidate clustering number is determined. The number of candidate clusters corresponding to the minimum clustering evaluation value is determined as the target number of clusters; Based on the candidate clustering results corresponding to the target clustering number, multiple resource communities are obtained.

[0010] By setting multiple candidate cluster numbers and performing pre-clustering separately, determining the cluster evaluation value based on feature differences and preset penalty coefficients, and then determining the target cluster number using the minimum cluster evaluation value, this application achieves a balance between feature consistency and cluster number control within the community compared to the existing technology of fixing the number of clusters or grouping only by resource type, thereby improving the matching between resource community division results and resource regulation characteristics.

[0011] Further, based on the feature vectors of distributed resources in each resource community, the aggregation and adjustment capabilities corresponding to each resource community are determined, including: The adjustable power and availability coefficient of each distributed resource are obtained from the feature vector of each distributed resource in each resource community, and the adjustment capability of the resource community is determined based on the adjustable power and availability coefficient of each distributed resource in the resource community. Based on the aforementioned adjustment capabilities, determine the up-adjustment capability, down-adjustment capability, response time, duration, adjustment cost, and execution confidence level for each of the aforementioned resource clusters. Based on the adjustable capability, the adjustable capability, the response time, the duration, the adjustment cost, and the execution confidence, the aggregated adjustment capabilities are formed for each resource cluster.

[0012] By determining the adjustment capability of a resource community based on the adjustable power and availability coefficient of distributed resources within the community, and further determining the up-adjustment capability, down-adjustment capability, response time, duration, adjustment cost, and execution confidence, this application can more accurately characterize the comprehensive adjustment capability of a resource community in the current state compared to the prior art which only uses the rated capacity of individual resources or static parameters to characterize resource capabilities.

[0013] Further, based on the aggregation and adjustment capabilities corresponding to multiple resource clusters, an edge aggregation capability curve corresponding to the edge node is generated, and the edge aggregation capability curve is uploaded to the main station, including: The aggregation and adjustment capabilities corresponding to multiple resource clusters are summarized to determine the adjustable power boundary, duration, response time, adjustment cost, and execution confidence of the edge node within a preset scheduling period; Based on the adjustable power boundary, the duration, the response time, the adjustment cost, and the execution confidence, generate the edge aggregation capability curve corresponding to the edge node; Upload the edge aggregation capability curve to the main station.

[0014] By summarizing the aggregation and adjustment capabilities of multiple resource clusters, the adjustable power boundary, duration, response time, adjustment cost, and execution confidence of edge nodes within a preset scheduling period are determined, and an edge aggregation capability curve is generated and uploaded to the main station. Compared with the existing technology of uploading a large amount of raw operating data of individual resources, this application can characterize the overall adjustment capability of edge nodes in the form of capability curves, reducing the data processing complexity on the main station side.

[0015] Furthermore, the main station determines the adjustment target corresponding to the edge node based on the edge aggregation capability curve, including: The main station receives edge aggregation capability curves uploaded by multiple edge nodes, and determines the adjustable power boundary, duration, response time, adjustment cost, and execution confidence of each edge node based on each edge aggregation capability curve. The main station obtains the total regulation demand and scheduling constraint information corresponding to the virtual power plant; The master station determines the adjustment allocation result for each edge node based on the adjustable power boundary, duration, response time, adjustment cost, execution confidence, total adjustment requirement, and scheduling constraint information corresponding to each edge node. The master station generates the adjustment target corresponding to the edge node based on the adjustment allocation result, and sends the adjustment target to the edge node.

[0016] By having the main station determine the adjustment allocation results for each edge node and generate adjustment targets based on the edge aggregation capability curves, total adjustment requirements, and scheduling constraint information uploaded by multiple edge nodes, this application enables the main station to allocate adjustment targets based on the edge node-level aggregation capabilities, thereby reducing the dimensionality of the scheduling objects. Compared with the existing technology where the main station directly performs fine-grained scheduling on individual distributed resources, this application enables the main station to allocate adjustment targets based on the edge node-level aggregation capabilities.

[0017] Further, based on the adjustment target and the aggregation adjustment capability corresponding to each resource community, a community adjustment task corresponding to each resource community is determined, and control instructions corresponding to each distributed resource are generated and issued according to the community adjustment task, including: Based on the adjustment target and the aggregation adjustment capability corresponding to each resource community, the adjustment priority corresponding to each resource community is determined; Based on the adjustment priority corresponding to each resource community, the adjustment target is assigned to the corresponding resource community to obtain the community adjustment task corresponding to each resource community. Obtain the current state and constraints of each distributed resource within the resource cluster, and determine the device adjustment task corresponding to each distributed resource based on the cluster adjustment task corresponding to each resource cluster, the current state, and the constraints. Based on the device adjustment tasks corresponding to each of the distributed resources, control instructions for the corresponding distributed resources are generated and issued.

[0018] By determining the cluster adjustment task based on the adjustment target and the aggregation adjustment capability corresponding to each resource cluster, and determining the device adjustment task by combining the current state and constraints of each distributed resource within the resource cluster, this application enables the control command to match the resource cluster capability and the current constraints of the individual resource, thereby improving the pertinence of distributed resource control commands, compared to the existing technology of directly issuing a unified control strategy to individual devices.

[0019] Furthermore, after generating and issuing control instructions for the corresponding distributed resources based on the community adjustment task, the method further includes: Obtain the execution feedback information returned by each of the distributed resources based on the control command, and determine the adjustment deviation between the actual adjustment result and the adjustment target based on the execution feedback information; When the adjustment deviation exceeds the preset deviation threshold, the availability coefficient of each distributed resource is re-determined based on the execution feedback information, and the aggregation adjustment capability of the corresponding resource community is updated. Based on the updated aggregation adjustment capability, the unfinished adjustment amount is redistributed, and supplementary control instructions are generated. The edge aggregation capability curve is corrected based on the updated aggregation adjustment capability, and the corrected edge aggregation capability curve is uploaded to the main site.

[0020] By determining the adjustment deviation based on the execution feedback information of distributed resources, and updating the availability coefficient, aggregation adjustment capability, and edge aggregation capability curves when the adjustment deviation exceeds the preset deviation threshold, and by redistributing the unfinished adjustment amount, this application can dynamically correct the resource capability representation based on the actual execution situation, compared with the existing technology that lacks edge-side feedback correction after the control command is issued, and improve the matching between subsequent supplementary control commands and the actual availability status of resources.

[0021] Secondly, this application provides a low-voltage distribution network resource aggregation and collaborative control system, including edge nodes and a master station. The edge nodes include a feature extraction module, a community clustering module, an aggregation adjustment module, a curve uploading module, and an instruction issuing module. The feature extraction module is used to obtain the status information of multiple distributed resources corresponding to the edge node, and extract the feature vector corresponding to each distributed resource based on the status information. The community clustering module is used to cluster multiple distributed resources according to the feature vectors corresponding to each distributed resource to obtain multiple resource communities; The aggregation adjustment module is used to determine the aggregation adjustment capability corresponding to each resource community based on the feature vector of the distributed resources in each resource community. The curve upload module is used to generate an edge aggregation capability curve corresponding to the edge node based on the aggregation and adjustment capabilities corresponding to multiple resource clusters, and upload the edge aggregation capability curve to the main station so that the main station can determine the adjustment target corresponding to the edge node based on the edge aggregation capability curve. The instruction issuing module is used to receive the adjustment target issued by the master station, determine the cluster adjustment task corresponding to each resource cluster according to the adjustment target and the aggregation adjustment capability corresponding to each resource cluster, and generate and issue control instructions corresponding to each distributed resource according to the cluster adjustment task.

[0022] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs a low-voltage distribution network resource aggregation and collaborative control method as described in the above application. Attached Figure Description

[0023] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. The drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0024] Figure 1 This is a flowchart illustrating a low-voltage distribution network resource aggregation and collaborative control method according to an embodiment of this application; Figure 2 This is a schematic diagram of the structure of a low-voltage distribution network resource aggregation and collaborative control system provided in another embodiment of this application. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The purpose of providing these embodiments is to make the disclosure of this application more thorough and comprehensive. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0026] In the description of this application, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first," "second," "third," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0027] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections or electrical connections; they can refer to direct connections or indirect connections through an intermediate medium; and they can refer to the internal communication between two components. The terms "vertical," "horizontal," "left," "right," "upper," "lower," and similar expressions used herein are for illustrative purposes only and do not indicate or imply that the device or component referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as limiting this application. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0028] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terminology used in the specification of this application is for the purpose of describing specific embodiments only and is not intended to limit the application. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0029] To address the issues in existing technologies where virtual power plants connect to a large number of heterogeneous distributed resources, the cloud-based master station needs to directly perform capacity modeling and scheduling control on a large number of individual resources, resulting in high computational pressure and insufficient matching between control commands and the actual resource adjustment capabilities. For example... Figure 1 As shown in the figure, this application provides a detailed flowchart of a low-voltage distribution network resource aggregation and collaborative control method. This embodiment of the low-voltage distribution network resource aggregation and collaborative control method includes steps S101 to S105, which are described in detail below: Step S101: Obtain the status information of multiple distributed resources corresponding to the edge node, and extract the feature vector corresponding to each distributed resource based on the status information.

[0030] It should be noted that in this embodiment, the edge node can also be called the edge proxy node. Both are used to represent local aggregation control nodes deployed on the side of distribution radio areas, parks, buildings or charging stations.

[0031] Specifically, edge agent nodes can be deployed in distribution substations, building energy management systems, charging station energy management systems, park energy gateways, or community energy storage controllers. Edge agent nodes connect to underlying distributed resources via IoT communication protocols, dedicated power networks, 5G, Wi-Fi, PLCs, or industrial buses to collect status information from each distributed resource. This status information can include real-time power, voltage, current, switch status, state of charge, temperature, user authorization status, device operating mode, and communication quality data.

[0032] It should be noted that, due to the different operating characteristics and constraints of different types of distributed resources, the focus of edge agent nodes on data collection for different resources may vary. For example, for air conditioning resources, the focus is on collecting indoor temperature, set temperature, and operating power; for electric vehicle charging piles, the focus is on collecting current power consumption, target power consumption, estimated departure time, and user priority; for energy storage resources, the focus is on collecting state of charge, maximum charging and discharging power, and health status; and for photovoltaic resources, the focus is on collecting current output, inverter status, and short-term predicted output.

[0033] Furthermore, after obtaining the above status information, the edge proxy node performs anomaly removal, missing information completion, time synchronization, and normalization on the status information to obtain the preprocessed status information corresponding to each distributed resource.

[0034] In some preferred embodiments, based on the state information, feature vectors corresponding to each of the distributed resources are extracted, including: The state information is preprocessed to obtain preprocessed state information corresponding to each of the distributed resources; Based on the preprocessed state information, determine one or more of the following characteristics for each of the distributed resources: real-time power characteristics, adjustable power characteristics, response time characteristics, adjustment cost characteristics, constraint characteristics, and reliability characteristics. Based on one or more of the real-time power characteristics, adjustable power characteristics, response time characteristics, adjustment cost characteristics, constraint characteristics, and reliability characteristics, a feature vector corresponding to each of the distributed resources is generated.

[0035] Specifically, after completing the state information preprocessing, the edge proxy node extracts device-level multidimensional feature vectors corresponding to each distributed resource based on the operational characteristics of different types of distributed resources. These device-level multidimensional feature vectors include at least power state, adjustability, response time, adjustment cost, user constraints, and historical reliability, and can be expressed as: in, Used to characterize the real-time power of the i-th distributed resource at time t; adjustable capability Used to characterize the adjustable power that the distributed resource can provide under the current operating state and constraints; response time Used to characterize the time required for the distributed resource to complete a regulation response from receiving a control command; regulation cost. Used to characterize the user compensation cost, equipment depreciation cost, or operating economic cost incurred by calling upon the distributed resource to participate in regulation; User constraints Used to characterize user comfort, charging needs, or process operating limitations; historical reliability This is used to characterize the success rate or deviation of the distributed resource in historical adjustment tasks.

[0036] It should be noted that for air conditioning resources, user constraints can include the deviation between indoor temperature and the user's comfort temperature range; for electric vehicle charging piles, user constraints can include target power, estimated departure time, and user priority; for energy storage resources, user constraints can include state of charge range, maximum charging and discharging power, and equipment safety limits. Through these device-level multidimensional feature vectors, different types of resources such as air conditioning, charging piles, energy storage, photovoltaics, and building flexible loads can be converted into a unified data representation, facilitating subsequent resource clustering and aggregation adjustment capability calculations.

[0037] Therefore, by preprocessing the state information of multi-source heterogeneous distributed resources and extracting one or more of the following features for different types of resources: real-time power features, adjustable power features, response time features, adjustment cost features, constraint features, and reliability features, this application can achieve unified feature modeling of multi-source heterogeneous resources and form a unified resource feature expression suitable for clustering and aggregation calculations, compared with the existing technology where the state of resources from different sources and of different types is difficult to express uniformly.

[0038] Step S102: Based on the feature vectors corresponding to each distributed resource, cluster the multiple distributed resources to obtain multiple resource clusters.

[0039] In some preferred embodiments, based on the feature vectors corresponding to each of the distributed resources, multiple distributed resources are clustered to obtain multiple resource clusters, including: Multiple candidate cluster numbers are obtained, and based on each candidate cluster number, the feature vectors of multiple distributed resources are pre-clustered to obtain the candidate clustering results corresponding to each candidate cluster number. The clustering results include multiple communities and their corresponding community centers. Based on the feature differences between the feature vectors of distributed resources and the corresponding community centers in each candidate clustering result, and the preset penalty coefficient, the clustering evaluation value corresponding to each candidate clustering number is determined. The number of candidate clusters corresponding to the minimum clustering evaluation value is determined as the target number of clusters; Based on the candidate clustering results corresponding to the target clustering number, multiple resource communities are obtained.

[0040] Specifically, after obtaining the device-level multidimensional feature vectors corresponding to each distributed resource, the edge proxy node dynamically clusters the multiple distributed resources based on these feature vectors to obtain multiple resource clusters. This dynamic clustering is not simply grouping by device type, but rather determining whether each distributed resource is suitable for participation in regulation based on its current adjustment capability, response speed, adjustment cost, user constraints, and historical reliability. Distributed resources with similar response and constraint characteristics are grouped into the same resource cluster. For example, although two air conditioners are both air conditioning resources, if one has a significant margin of safety in its indoor temperature range while the other is close to the user's comfort temperature limit, their current peak-shaving capabilities differ, and they can be grouped into different resource clusters. Similarly, although two electric vehicle charging stations have the same current charging power, if the expected departure time and target battery level of the corresponding vehicles are different, their delayed charging capabilities differ, and they can also be grouped into different resource clusters.

[0041] Furthermore, each candidate clustering result includes multiple candidate clusters and the corresponding cluster center for each candidate cluster. Subsequently, the edge proxy node determines the clustering evaluation value corresponding to each candidate cluster number based on the feature difference between the device-level multidimensional feature vector of each distributed resource and the corresponding cluster center, as well as the preset penalty coefficient used to constrain the number of clusters. The candidate cluster number with the smallest clustering evaluation value is determined as the target cluster number, thereby obtaining multiple resource clusters.

[0042] In this embodiment, the dynamic clustering objective can be simplified as follows: Where J represents the value of the dynamic clustering objective function; K represents the number of resource communities; and k represents the index of the resource community. This represents the k-th resource community; This represents the device-level multidimensional feature vector corresponding to the i-th distributed resource; This represents the community center corresponding to the k-th resource community; λ represents the feature difference between the device-level multidimensional feature vector of the i-th distributed resource and the center of its resource community; λ represents the preset penalty coefficient, which is used to constrain the number of resource communities and avoid increasing the computational burden on edge nodes due to an excessive number of resource communities.

[0043] It should be noted that edge proxy nodes can implement the above dynamic clustering using dynamic K-means, density clustering, online Gaussian mixture models, or lightweight neural network clustering methods. The clustering results can be updated on a rolling basis as the distributed resource status changes; at the same time, to avoid control instability caused by frequent switching of resource clusters, edge proxy nodes can set a minimum hold time and a cluster change threshold, updating the resource cluster partitioning results only when the resource status change reaches a preset level.

[0044] Therefore, by setting multiple candidate cluster numbers and performing pre-clustering separately, determining the cluster evaluation value based on feature differences and preset penalty coefficients, and then determining the target cluster number with the minimum cluster evaluation value, compared with the existing technology of fixing the number of clusters or grouping only according to resource type, this application can achieve a balance between feature consistency within the community and community number control, and improve the matching between resource community division results and resource regulation characteristics.

[0045] Step S103: Determine the aggregation and adjustment capability corresponding to each resource community based on the feature vectors of the distributed resources in each resource community.

[0046] In some preferred embodiments, the aggregation and adjustment capability corresponding to each resource community is determined based on the feature vectors of distributed resources in each resource community, including: The adjustable power and availability coefficient of each distributed resource are obtained from the feature vector of each distributed resource in each resource community, and the adjustment capability of the resource community is determined based on the adjustable power and availability coefficient of each distributed resource in the resource community. Based on the aforementioned adjustment capabilities, determine the up-adjustment capability, down-adjustment capability, response time, duration, adjustment cost, and execution confidence level for each of the aforementioned resource clusters. Based on the adjustable capability, the adjustable capability, the response time, the duration, the adjustment cost, and the execution confidence, the aggregated adjustment capabilities are formed for each resource cluster.

[0047] Specifically, after completing dynamic clustering, the edge proxy node treats each resource cluster as an equivalent virtual adjustable unit and determines the aggregated adjustment capability corresponding to that resource cluster based on the feature vectors of each distributed resource within the cluster. Specifically, the edge proxy node can obtain the corresponding adjustable power and availability coefficient from the feature vectors of each distributed resource, and determine the adjustment capability of the resource cluster at the current time based on the adjustable power and availability coefficient of each distributed resource within the cluster. For the k-th resource cluster, its equivalent adjustment capability can be obtained by aggregating the adjustable power and availability coefficient of each distributed resource within the cluster, expressed as: Among them, the available coefficient This is used to characterize the availability of the corresponding distributed resource in the current moment. If the distributed resource is online, has obtained user authorization, has normal communication status, and has high historical execution reliability, then the availability coefficient of the distributed resource is high; if the distributed resource is offline, has refused to respond to the user, has poor communication quality, or has large historical execution deviation, then the availability coefficient of the distributed resource is low. Used to characterize the adjustable power of the corresponding distributed resource participating in adjustment at the current moment; This represents the k-th resource community.

[0048] Furthermore, after determining the adjustability of the resource cluster, the edge proxy node determines multiple equivalent metrics corresponding to the resource cluster based on the adjustable power, current state, constraints, and historical execution data of each distributed resource within the cluster. These equivalent metrics may include one or more of the following: adjustable capacity, adjustable capacity, response time, duration, adjustment cost, execution confidence, and security boundary.

[0049] Among them, the up-adjustable capacity represents the ability of a resource cluster to reduce load, increase output, or release regulation capacity to the grid; the down-adjustable capacity represents the ability of a resource cluster to increase load, reduce output, or absorb surplus energy. Edge agent nodes can determine the up-adjustable and down-adjustable capacities corresponding to a resource cluster based on the current power, allowable operating power range, and availability coefficient of each distributed resource within the resource cluster, which can be expressed as: in, Let represent the upscalability of the j-th resource cluster at time t; This represents the downscaling capability of the j-th resource cluster at time t; This represents the security constraint reduction factor for the j-th resource cluster at time t. Let represent the j-th resource cluster; i represents the i-th distributed resource within the j-th resource cluster. This represents the availability coefficient of the i-th distributed resource at time t; Let represent the equivalent net load power of the i-th distributed resource at time t; This represents the minimum allowed operating power of the i-th distributed resource at time t; This represents the maximum allowed operating power of the i-th distributed resource at time t.

[0050] It should be noted that when different types of resources adopt different power positive direction conventions, the edge proxy node can first convert the power of each distributed resource into an equivalent net load power, and then calculate the up-adjustable and down-adjustable capabilities.

[0051] Furthermore, response time is used to represent the equivalent time required for a resource cluster to complete a regulation response from receiving a regulation task. Edge proxy nodes can weight the response time of the resource cluster based on the individual response time and participation power of each distributed resource within the cluster, which can be expressed as: in, ε represents the equivalent response time of the j-th resource community at time t, which is used to characterize the time required for the resource community to complete the adjustment response from receiving the community adjustment task; ε represents a preset minimum positive number to avoid the denominator being zero.

[0052] It should be noted that, in order to improve the conservatism of scheduling, the preset quantile values ​​of the response times of multiple distributed resources within a resource cluster can also be used as the equivalent response time of the resource cluster.

[0053] Furthermore, duration is used to represent the time during which a resource community can continuously provide regulation capability under a given regulation amount. Edge agent nodes can determine the duration of a resource community based on the equivalent energy currently available for regulation of each distributed resource within the community and the community's regulation amount, which can be expressed as: in, Indicates the first The equivalent energy that a resource can currently be used for regulation. This represents the adjustment amount allocated to the j-th resource community.

[0054] It should be noted that, for energy storage resources, the equivalent energy can be determined by the state of charge margin; for air conditioning resources, the equivalent energy can be determined by the temperature comfort margin; and for electric vehicle charging resources, the equivalent energy can be determined by the remaining dwell time and the amount of charge that can be delayed.

[0055] Furthermore, the adjustment cost represents the unit cost or total cost incurred in calling upon the resource cluster to participate in adjustment. Edge agent nodes can determine the adjustment cost of a resource cluster based on the user compensation cost, equipment depreciation cost, operational economic cost, or comfort impact cost of each distributed resource within the cluster. This cost can be expressed as: in, Indicates the first Distributed resources at time The corresponding unit adjustment costs may include user compensation costs, equipment depreciation costs, battery life loss costs, comfort impact costs, or process impact costs.

[0056] It should be noted that when determining the total cost of a single adjustment task, the edge proxy node can also combine the adjustment amount and duration of the resource community to determine the total adjustment cost corresponding to that resource community, which can be expressed as: in, This represents the total adjustment cost corresponding to the j-th resource community performing one adjustment task; This represents the equivalent unit adjustment cost of the j-th resource cluster at time t; This represents the adjustment amount allocated to the j-th resource community; This represents the duration for the j-th resource community to perform the adjustment task. By multiplying the equivalent unit adjustment cost, the adjustment amount, and the duration, the total cost required for the resource community to complete one adjustment task can be obtained.

[0057] Furthermore, execution confidence is used to represent the degree of certainty that a resource community completes the actual adjustment according to the community adjustment task. Edge agent nodes can first determine the execution confidence of an individual distributed resource based on its online status, user authorization status, communication quality, historical execution reliability, prediction accuracy, and current constraint margin. Specifically, the execution confidence of an individual resource can be expressed as: in, Indicates the device's online status; online means... Offline ; Indicates the user's authorization status; authorization is... Unauthorized ; Indicates communication quality; Indicates historical execution reliability; Indicates the accuracy of the prediction or the reliability of the status; Indicates the current constraint margin of the equipment; Let be the weight coefficient, and satisfy... .

[0058] Then, by aggregating the contributions of each distributed resource, the execution confidence corresponding to the resource cluster is obtained, which can be expressed as: in, The closer this value is to 1, the more likely the resource cluster is to adjust according to the targets issued by the cloud or edge proxy nodes.

[0059] The higher the execution confidence level, the more likely the resource community is to complete the corresponding adjustment amount according to the community adjustment task assigned by the edge node. After the control execution is completed, the edge agent node can also perform rolling correction on the historical execution reliability based on the deviation between the actual completed adjustment amount and the commanded adjustment amount, specifically: in, To achieve the actual adjustment amount, β is the instruction adjustment amount, and β∈(0,1) is the update step size.

[0060] Furthermore, the safety boundary is used to indicate whether the distribution network operation constraints are met when resource communities participate in regulation. For the regulation amount allocated to the j-th resource community... The edge agent node can perform a safety check on the regulation based on operating parameters such as transformer load, feeder current, node voltage, and reverse power flow, and define a safety boundary function: The safety boundary can be expressed as: When a resource community performs an adjustment ΔPj, and simultaneously meets the following constraints, it indicates that the resource community has not reached the distribution network safety boundary under the current adjustment: At this point, the safety boundary function is set to 1, indicating that the resource cluster is allowed to participate in regulation according to this adjustment amount. If any of the above constraints are not met, it means that this adjustment amount may lead to the transformer capacity exceeding the limit, feeder current exceeding the limit, node voltage exceeding the limit, or reverse power flow exceeding the limit. In this case, the safety boundary function is set to 0, i.e.: in, This represents the safety boundary function for the j-th resource cluster under the adjustment amount ΔPj; Indicates the load of the transformer in the distribution area. Indicates the maximum allowable load of the transformer in the distribution area; Indicates the feeder current. Indicates the maximum allowable current for the feeder; Indicates node voltage. and These represent the lower and upper limits of the allowable node voltage, respectively; Indicates the reverse power flow. This indicates the maximum permissible reverse power flow.

[0061] It should be noted that in engineering implementation, to avoid abrupt changes in resource community capabilities due to binary judgment, the aforementioned safety boundary function can be transformed into a continuous safety reduction coefficient. The continuous safety reduction coefficient can be expressed as: When the transformer capacity, feeder current, voltage exceed limits, or reverse power flow constraints approach the boundary... This automatically reduces the adjustability and execution confidence of the resource community in its external reporting.

[0062] Therefore, by determining the adjustment capability of a resource community based on the adjustable power and availability coefficient of distributed resources within the resource community, and further determining the up-adjustment capability, down-adjustment capability, response time, duration, adjustment cost, and execution confidence, this application can more accurately characterize the comprehensive adjustment capability of a resource community in the current state compared to the prior art which only uses the rated capacity of individual resources or static parameters to characterize resource capabilities.

[0063] Step S104: Based on the aggregation and adjustment capabilities corresponding to the multiple resource clusters, generate the edge aggregation capability curve corresponding to the edge node, and upload the edge aggregation capability curve to the main station so that the main station can determine the adjustment target corresponding to the edge node based on the edge aggregation capability curve.

[0064] In some preferred embodiments, based on the aggregation and adjustment capabilities corresponding to multiple resource communities, an edge aggregation capability curve corresponding to the edge node is generated, and the edge aggregation capability curve is uploaded to the main station, including: The aggregation and adjustment capabilities corresponding to multiple resource clusters are summarized to determine the adjustable power boundary, duration, response time, adjustment cost, and execution confidence of the edge node within a preset scheduling period; Based on the adjustable power boundary, the duration, the response time, the adjustment cost, and the execution confidence, generate the edge aggregation capability curve corresponding to the edge node; Upload the edge aggregation capability curve to the main station.

[0065] Specifically, after determining the aggregation and adjustment capabilities corresponding to each resource cluster, the edge agent node summarizes the aggregation and adjustment capabilities of the multiple resource clusters under its jurisdiction, generating an edge aggregation capability curve corresponding to the edge node. The edge aggregation capability curve is used to characterize the overall adjustment capability that the edge node can provide to the master station within a preset scheduling period, so that the master station does not need to directly obtain the original operating data of each individual distributed resource, but can allocate adjustment targets based on the edge node-level capability information.

[0066] Specifically, edge proxy nodes can determine the upper and lower limits of adjustable power, duration, response time, adjustment cost, and execution confidence of the edge node at time t based on the adjustable power, adjustable power, duration, response time, adjustment cost, and execution confidence of each resource cluster, and generate an edge aggregation capability curve. The edge aggregation capability curve can be represented as: in, and This indicates the upper and lower boundaries of the adjustable power that the edge node can provide. Indicates the duration. Indicates response time. Indicates equivalent adjustment costs, Indicates the execution confidence level.

[0067] Specifically, the upper and lower limits of adjustable power can be obtained by summing the adjustable and lower limits of each resource cluster; the duration can be determined comprehensively based on the duration of each resource cluster under the corresponding adjustment amount; the response time can be determined by weighting the equivalent response time of each resource cluster and the amount of adjustment involved; the equivalent adjustment cost can be determined based on the adjustment cost and adjustment contribution of each resource cluster; and the execution confidence can be determined comprehensively based on the execution confidence and availability of each resource cluster. The edge aggregation capability curve can be generated on a rolling basis according to preset scheduling cycles such as 5 minutes, 15 minutes, or 1 hour. When the status of the resources under the jurisdiction of the edge node changes little and the execution confidence is stable, the edge agent node can upload the edge aggregation capability curve to the master station at a fixed period; when the online status of resources, user authorization status, distribution network safety boundary, or execution confidence changes significantly, the edge agent node can also trigger event reporting and upload the updated edge aggregation capability curve to the master station.

[0068] In this method, the data uploaded by the edge proxy node to the main station is the aggregated capability curve, rather than the raw status data of all the distributed resources under its jurisdiction. Compared to the traditional method, which requires uploading data of "number of devices × single device status item", this application only requires uploading data of "number of edge nodes × aggregated indicator item". When a single edge node manages thousands to tens of thousands of distributed resources, it can significantly reduce the amount of cloud-edge communication data and reduce the data processing pressure on the main station.

[0069] Therefore, by summarizing the aggregation and adjustment capabilities of multiple resource clusters, the adjustable power boundary, duration, response time, adjustment cost, and execution confidence of edge nodes within a preset scheduling period are determined, and an edge aggregation capability curve is generated and uploaded to the main station. Compared with the existing technology of uploading a large amount of raw operating data of individual resources, this application can characterize the overall adjustment capability of edge nodes in the form of capability curves, reducing the data processing complexity on the main station side.

[0070] In some preferred embodiments, the master station determines the adjustment target corresponding to the edge node based on the edge aggregation capability curve, including: The main station receives edge aggregation capability curves uploaded by multiple edge nodes, and determines the adjustable power boundary, duration, response time, adjustment cost, and execution confidence of each edge node based on each edge aggregation capability curve. The main station obtains the total regulation demand and scheduling constraint information corresponding to the virtual power plant; The master station determines the adjustment allocation result for each edge node based on the adjustable power boundary, duration, response time, adjustment cost, execution confidence, total adjustment requirement, and scheduling constraint information corresponding to each edge node. The master station generates the adjustment target corresponding to the edge node based on the adjustment allocation result, and sends the adjustment target to the edge node.

[0071] Specifically, the master station acquires the total regulation demand and dispatch constraints corresponding to the virtual power plant. The total regulation demand may include peak shaving demand, valley filling demand, renewable energy consumption demand, or ancillary service demand; the dispatch constraints may include grid security constraints, market price constraints, response time requirements, duration requirements, and regulation cost constraints. Based on the total regulation demand, dispatch constraints, and the edge aggregation capability curves uploaded by each edge node, the master station determines the regulation allocation result for each edge node.

[0072] Under the premise of meeting the overall regulation requirements and safety constraints, edge nodes with short response times, low regulation costs, high execution confidence, and minimal impact on users can be prioritized. Excessive regulation targets should be avoided for edge nodes with insufficient duration, low execution confidence, or those close to the distribution network safety boundary. Subsequently, the master station generates regulation targets for each edge node based on the regulation allocation results and distributes these targets to the corresponding edge nodes.

[0073] For example, when a virtual power plant needs to provide 30MW of peak shaving capacity, the master station can determine, based on the edge aggregation capacity curves uploaded by multiple edge nodes, that the residential park air conditioning and energy storage edge nodes can respond quickly and provide 7MW of peak shaving capacity, the electric vehicle charging station edge nodes can provide 8MW of peak shaving capacity by reducing charging power, and the commercial building edge nodes can provide 5MW of peak shaving capacity by adjusting chiller units, fresh air systems, or lighting systems. The remaining adjustment is handled by other edge nodes or backup resources. Thus, the master station can generate adjustment targets corresponding to 7MW, 8MW, 5MW, and other adjustment amounts, and distribute them to the corresponding edge nodes.

[0074] Therefore, by having the main station determine the adjustment allocation results and generate adjustment targets for each edge node based on the edge aggregation capability curves, total adjustment requirements, and scheduling constraint information uploaded by multiple edge nodes, compared with the existing technology where the main station directly performs fine-grained scheduling on individual distributed resources, this application enables the main station to allocate adjustment targets based on the edge node-level aggregation capabilities, reducing the dimension of scheduling objects.

[0075] Step S105: Receive the adjustment target issued by the master station, determine the cluster adjustment task corresponding to each resource cluster based on the adjustment target and the aggregation adjustment capability corresponding to each resource cluster, and generate and issue control instructions corresponding to each distributed resource based on the cluster adjustment task.

[0076] In some preferred embodiments, based on the adjustment target and the aggregation adjustment capability corresponding to each resource community, a community adjustment task corresponding to each resource community is determined, and control instructions corresponding to each distributed resource are generated and issued according to the community adjustment task, including: Based on the adjustment target and the aggregation adjustment capability corresponding to each resource community, the adjustment priority corresponding to each resource community is determined; Based on the adjustment priority corresponding to each resource community, the adjustment target is assigned to the corresponding resource community to obtain the community adjustment task corresponding to each resource community. Obtain the current state and constraints of each distributed resource within the resource cluster, and determine the device adjustment task corresponding to each distributed resource based on the cluster adjustment task corresponding to each resource cluster, the current state, and the constraints. Based on the device adjustment tasks corresponding to each of the distributed resources, control instructions for the corresponding distributed resources are generated and issued.

[0077] Specifically, after receiving the adjustment target from the master station, the edge node, based on the adjustment target and the aggregation adjustment capability corresponding to each resource cluster, completes the adjustment task allocation locally and generates control instructions for the corresponding distributed resources. In this embodiment, the edge node can adopt a two-level task allocation method, including task allocation between resource clusters and task allocation within resource clusters.

[0078] Preferably, the first level is task allocation among resource clusters. Edge nodes determine the adjustment priority for each resource cluster based on its adjustability, response time, duration, adjustment cost, user constraints, and execution confidence. Then, according to the adjustment priority, the adjustment targets issued by the main station are allocated to the corresponding resource clusters, resulting in cluster adjustment tasks for each resource cluster. The adjustment priority can be determined based on principles such as low cost, fast response, sufficient duration to meet adjustment requirements, high execution confidence, and minimal user impact.

[0079] Preferably, the second level involves task allocation for devices within a resource cluster. After determining the cluster adjustment task corresponding to each resource cluster, the edge node obtains the current state and constraints of each distributed resource within the resource cluster, and determines the device adjustment task corresponding to each distributed resource based on the cluster adjustment task, current state, and constraints. Subsequently, the edge node generates and issues control commands to the corresponding distributed resources based on each device adjustment task.

[0080] For example, for an air conditioning resource cluster, edge nodes can moderately increase the set temperature of some air conditioners or reduce the operating frequency of the compressor; for an electric vehicle charging resource cluster, edge nodes can reduce the charging power of some vehicles or delay charging; for an energy storage resource cluster, edge nodes can control the discharge of energy storage devices or adjust the charging power; for a building flexible load cluster, edge nodes can adjust the operating strategies of chillers, fresh air systems, lighting systems or other controllable loads.

[0081] As a preferred approach, edge nodes can employ one or more of the following methods to complete the adjustment task allocation: local optimization control, consensus collaborative control, and multi-agent reinforcement learning control. Specifically, local optimization control refers to edge nodes centrally calculating the adjustment allocation results for each resource cluster and each distributed resource; consensus collaborative control refers to device agents within a resource cluster negotiating adjustment tasks through local communication to ensure the resource cluster as a whole meets the cluster's adjustment task requirements; and multi-agent reinforcement learning control refers to multiple device agents autonomously generating control actions based on historical experience and current state in scenarios with complex user behavior and high model uncertainty.

[0082] For ease of engineering implementation, the edge-side scheduling objective can be simplified as follows: Among them, the adjustment cost is used to characterize the user compensation cost, equipment wear cost, or operating economic cost incurred by calling resource clusters or distributed resources to perform adjustment tasks; the target deviation penalty is used to constrain the actual adjustment amount of the edge node to be close to the adjustment target issued by the master station; the constraint violation penalty is used to prevent the violation of user comfort, charging demand, energy storage state of charge, equipment safety limits, and distribution network safety boundaries. In this way, the edge node can transform the macro-adjustment target issued by the master station into resource cluster-level tasks and equipment-level control commands locally, improving the matching between control commands and the actual adjustability of distributed resources.

[0083] Therefore, by determining the cluster adjustment task based on the adjustment target and the aggregation adjustment capability corresponding to each resource cluster, and by determining the device adjustment task in combination with the current state and constraints of each distributed resource within the resource cluster, compared with the existing technology of directly issuing a unified control strategy to individual devices, this application can match the control command with the resource cluster capability and the current constraints of the individual resource, thereby improving the pertinence of the distributed resource control command.

[0084] In some preferred embodiments, after generating and issuing control instructions for the corresponding distributed resources based on the community adjustment task, the method further includes: Obtain the execution feedback information returned by each of the distributed resources based on the control command, and determine the adjustment deviation between the actual adjustment result and the adjustment target based on the execution feedback information; When the adjustment deviation exceeds the preset deviation threshold, the availability coefficient of each distributed resource is re-determined based on the execution feedback information, and the aggregation adjustment capability of the corresponding resource community is updated. Based on the updated aggregation adjustment capability, the unfinished adjustment amount is redistributed, and supplementary control instructions are generated. The edge aggregation capability curve is corrected based on the updated aggregation adjustment capability, and the corrected edge aggregation capability curve is uploaded to the main site.

[0085] Therefore, by determining the adjustment deviation based on the execution feedback information of distributed resources, and updating the availability coefficient, aggregation adjustment capability, and edge aggregation capability curve when the adjustment deviation exceeds the preset deviation threshold, and simultaneously redistributing the uncompleted adjustment amount, compared with the existing technology that lacks edge-side feedback correction after the control command is issued, this application can dynamically correct the resource capability representation according to the actual execution situation and improve the matching between subsequent supplementary control commands and the actual availability status of resources.

[0086] Specifically, after issuing control commands to each distributed resource, the edge node continuously collects the execution feedback information returned by each distributed resource, and determines the adjustment deviation between the actual adjustment result and the adjustment target based on the execution feedback information. The execution feedback information may include actual power change, control command execution status, user exit response information, communication status change information, and device constraint change information, etc.

[0087] When the adjustment deviation is within the preset deviation threshold range, it indicates that the actual adjustment amount completed by the edge node can meet the adjustment target issued by the master station, and the edge node can maintain the current control strategy. When the adjustment deviation exceeds the preset deviation threshold, it indicates that some distributed resources may be offline, have user logout response, communication anomalies, large prediction errors, or changes in device constraints, and the edge node triggers a rolling correction process.

[0088] Preferably, the rolling correction process includes: reassessing the availability coefficients of each distributed resource; updating the aggregation adjustment capability of the corresponding resource community based on the reassessed availability coefficients; redistributing the unfinished adjustment amount based on the updated aggregation adjustment capability and generating supplementary control instructions; correcting the edge aggregation capability curve based on the updated aggregation adjustment capability and uploading the corrected edge aggregation capability curve to the main station; and reducing the historical reliability or execution confidence of unstable resources to reduce their call ratio in subsequent adjustment tasks.

[0089] Furthermore, when communication between the cloud master station and the edge node is briefly interrupted, the edge node can enter an autonomous operation mode. In autonomous operation mode, the edge node performs conservative control based on the most recently received adjustment target, local load forecast results, local distribution network security constraints, and current resource status. This avoids transformer capacity exceeding limits, feeder current exceeding limits, node voltage exceeding limits, or reverse power flow exceeding limits in the local system, and also avoids excessive use of user resources.

[0090] Furthermore, after cloud-edge communication is restored, the edge nodes upload the execution results, adjustment deviations, resource status changes, and corrected edge aggregation capability curves from the communication interruption period to the master station, enabling the master station to update subsequent scheduling plans based on the supplementary information. Through the above feedback correction and autonomous operation mechanism, edge nodes can maintain local adjustment capabilities when resource status changes or cloud-edge communication anomalies occur, improving the robustness and continuity of the virtual power plant aggregation control process.

[0091] In some specific application scenarios, the above-described method embodiments can be applied to virtual power plant regulation scenarios such as peak shaving for air conditioning in residential parks, orderly charging of electric vehicle charging stations, and coordinated consumption of distributed photovoltaic and energy storage, so as to demonstrate the technical effect of edge nodes in aggregating and controlling massive distributed resources.

[0092] In a residential park's air conditioning peak shaving scenario, a residential park connects 10,000 household air conditioners, with edge nodes deployed in the park's power distribution room. At 18:00 in summer, the main station sends a 3MW peak shaving target to the corresponding edge nodes in the park, requiring a response within 5 minutes and a duration of 60 minutes. The edge nodes collect real-time power, indoor temperature, set temperature, user comfort range, and online status of each air conditioner. Based on temperature margin, response speed, and user comfort risk, they divide the air conditioning resources into a priority adjustment group with a large temperature margin and low comfort risk, an auxiliary adjustment group with a medium temperature margin and short-term participation in peak shaving, and a low-call group close to the upper limit of comfort temperature or with stricter user preferences. Subsequently, the edge nodes prioritize allocating peak shaving tasks to the priority and auxiliary adjustment groups and generate corresponding control commands for the air conditioners, such as moderately increasing the set temperature of some air conditioners or reducing the compressor operating frequency. Therefore, the main station does not need to directly model and issue control commands to each of the 10,000 air conditioners. Instead, it can complete the local cluster adjustment task allocation through edge nodes, which can reduce the amount of cloud-edge communication data and the computing pressure on the main station. At the same time, since the edge nodes consider temperature margin and comfort risk when allocating tasks, it can reduce the calls to users with high comfort risk, and improve the real-time performance and user experience of peak shaving control.

[0093] In an orderly charging scenario at an electric vehicle charging station, a centralized charging station connects 2000 charging piles with a current total charging power of 18MW. During the evening peak, the main station requests the corresponding edge nodes of this charging station to reduce their load by 8MW for one hour. The edge nodes collect data on each vehicle's current battery level, target battery level, estimated departure time, current charging power, and user priority. Based on this information, they categorize vehicles into groups with delayed charging, moderately adjustable charging, and non-delayed charging. For vehicles with later departure times, higher current battery levels, or lower user priority, the edge nodes prioritize reducing their charging power or delaying charging. For vehicles about to leave or whose target battery level has not yet been met, the edge nodes maintain their charging power as much as possible. Thus, the edge nodes can translate the 8MW load reduction target issued by the main station into locally executable device-level control commands for the charging station, providing peak-shaving capabilities for the virtual power plant while meeting users' basic charging needs. Compared to the main station directly controlling each charging pile or uniformly reducing the power of all charging piles, this scenario reduces communication pressure, improves the precision of orderly charging control, and minimizes the impact on vehicles about to leave and high-priority users.

[0094] In a scenario where distributed photovoltaic (PV) and energy storage are used in synergistic absorption, a residential area experiences high PV output at midday, leading to a risk of reverse power flow in some areas. The main station requests the corresponding edge nodes in this area to increase their local absorption capacity. The edge nodes generate an edge aggregation capacity curve to absorb excess PV based on the energy storage's state of charge, the pre-charging capability of electric vehicles, and the pre-cooling capability of air conditioners. They then coordinate various resources locally to execute control actions. For example, the edge nodes can control the charging of some energy storage devices, coordinate the pre-charging of some electric vehicles, and pre-cool some air conditioners. Thus, when a reverse power flow risk occurs in a localized area, the edge nodes can quickly utilize local absorbable resources, improving the local absorption capacity of excess PV. Compared to the main station's centralized judgment and individual issuance of device control commands, this scenario shortens the local absorption control response time, reduces the risk of reverse power flow exceeding limits, and improves the synergistic utilization efficiency among distributed PV, energy storage, electric vehicles, and flexible loads.

[0095] By implementing the embodiments of this application, feature vector extraction, resource community division, aggregation and adjustment capability determination, and edge aggregation capability curve generation are performed on multiple distributed resources at the edge node side. Control commands are generated locally based on the adjustment targets issued by the master station. Compared with the prior art, where the master station directly models and controls a large number of individual distributed resources, this application can reduce the computing pressure on the cloud and improve the matching between control commands and the actual adjustment capabilities of distributed resources.

[0096] Based on the above method embodiments, this application provides corresponding system embodiments.

[0097] like Figure 2 As shown, another embodiment of this application provides a low-voltage distribution network resource aggregation and collaborative control system, including edge nodes and a master station. The edge nodes include a feature extraction module, a community clustering module, an aggregation adjustment module, a curve uploading module, and an instruction issuing module. The feature extraction module is used to obtain the status information of multiple distributed resources corresponding to the edge node, and extract the feature vector corresponding to each distributed resource based on the status information. The community clustering module is used to cluster multiple distributed resources according to the feature vectors corresponding to each distributed resource to obtain multiple resource communities; The aggregation adjustment module is used to determine the aggregation adjustment capability corresponding to each resource community based on the feature vector of the distributed resources in each resource community. The curve upload module is used to generate an edge aggregation capability curve corresponding to the edge node based on the aggregation and adjustment capabilities corresponding to multiple resource clusters, and upload the edge aggregation capability curve to the main station so that the main station can determine the adjustment target corresponding to the edge node based on the edge aggregation capability curve. The instruction issuing module is used to receive the adjustment target issued by the master station, determine the cluster adjustment task corresponding to each resource cluster according to the adjustment target and the aggregation adjustment capability corresponding to each resource cluster, and generate and issue control instructions corresponding to each distributed resource according to the cluster adjustment task.

[0098] It is understood that the above system embodiments correspond to the method embodiments of this application, and can implement a low-voltage distribution network resource aggregation and collaborative control method provided by any of the above method embodiments of this application.

[0099] It should be noted that the system embodiments described above are merely illustrative, and some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided in this application, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0100] Based on the above-described method embodiments, another embodiment of this application provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute a low-voltage distribution network resource aggregation and collaborative control method as described in any of the above-described method embodiments of this application.

[0101] The modules / units integrated in the device / terminal equipment, if implemented as software functional units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0102] Based on the above-described method embodiments, another embodiment of this application provides a computer program product, including a computer program or instructions, which, when executed by a communication device, implements a low-voltage distribution network resource aggregation and collaborative control method.

[0103] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of this application. It should be understood that the above descriptions are merely specific embodiments of this application and are not intended to limit the scope of protection of this application. In particular, it should be noted that any modifications, equivalent substitutions, or improvements made by those skilled in the art within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for resource aggregation and collaborative control in low-voltage distribution networks, characterized in that, The distributed resource aggregation control method, applied to edge nodes in a virtual power plant, includes: Obtain the status information of multiple distributed resources corresponding to the edge node, and extract the feature vector corresponding to each distributed resource based on the status information; Based on the feature vectors corresponding to each of the distributed resources, multiple distributed resources are clustered to obtain multiple resource clusters. Based on the feature vectors of distributed resources in each resource community, determine the aggregation and adjustment capabilities corresponding to each resource community; Based on the aggregation and adjustment capabilities corresponding to multiple resource clusters, an edge aggregation capability curve corresponding to the edge node is generated, and the edge aggregation capability curve is uploaded to the main station so that the main station can determine the adjustment target corresponding to the edge node based on the edge aggregation capability curve. The system receives the adjustment target issued by the master station, determines the cluster adjustment task corresponding to each resource cluster based on the adjustment target and the aggregation adjustment capability corresponding to each resource cluster, and generates and issues control instructions corresponding to each distributed resource based on the cluster adjustment task.

2. The method for low-voltage distribution network resource aggregation and collaborative control according to claim 1, characterized in that, The step of extracting feature vectors corresponding to each of the distributed resources based on the status information includes: The state information is preprocessed to obtain preprocessed state information corresponding to each of the distributed resources; Based on the preprocessed state information, determine one or more of the following characteristics for each of the distributed resources: real-time power characteristics, adjustable power characteristics, response time characteristics, adjustment cost characteristics, constraint characteristics, and reliability characteristics. Based on one or more of the real-time power characteristics, adjustable power characteristics, response time characteristics, adjustment cost characteristics, constraint characteristics, and reliability characteristics, a feature vector corresponding to each of the distributed resources is generated.

3. The method for resource aggregation and collaborative control of a low-voltage distribution network according to claim 1, characterized in that, The step involves clustering multiple distributed resources based on their corresponding feature vectors to obtain multiple resource clusters, including: Multiple candidate cluster numbers are obtained, and based on each candidate cluster number, the feature vectors of multiple distributed resources are pre-clustered to obtain the candidate clustering results corresponding to each candidate cluster number. The clustering results include multiple communities and their corresponding community centers. Based on the feature differences between the feature vectors of distributed resources and the corresponding community centers in each candidate clustering result, and the preset penalty coefficient, the clustering evaluation value corresponding to each candidate clustering number is determined. The number of candidate clusters corresponding to the minimum clustering evaluation value is determined as the target number of clusters; Based on the candidate clustering results corresponding to the target clustering number, multiple resource communities are obtained.

4. The method for resource aggregation and collaborative control of a low-voltage distribution network according to claim 1, characterized in that, The step of determining the aggregation and adjustment capability corresponding to each resource community based on the feature vectors of distributed resources in each resource community includes: The adjustable power and availability coefficient of each distributed resource are obtained from the feature vectors of the distributed resources in each resource community, and the adjustment capability of the resource community is determined based on the adjustable power and availability coefficient of each distributed resource in the resource community. Based on the aforementioned adjustment capabilities, determine the up-adjustment capability, down-adjustment capability, response time, duration, adjustment cost, and execution confidence level for each of the aforementioned resource clusters. Based on the adjustable capability, the adjustable capability, the response time, the duration, the adjustment cost, and the execution confidence, the aggregated adjustment capabilities are formed for each resource cluster.

5. The method for low-voltage distribution network resource aggregation and collaborative control according to claim 4, characterized in that, The step of generating an edge aggregation capability curve corresponding to the edge node based on the aggregation adjustment capabilities corresponding to multiple resource clusters, and uploading the edge aggregation capability curve to the main station, includes: The aggregation and adjustment capabilities corresponding to multiple resource clusters are summarized to determine the adjustable power boundary, duration, response time, adjustment cost, and execution confidence of the edge node within a preset scheduling period; Based on the adjustable power boundary, the duration, the response time, the adjustment cost, and the execution confidence, generate the edge aggregation capability curve corresponding to the edge node; Upload the edge aggregation capability curve to the main station.

6. The method for resource aggregation and collaborative control of a low-voltage distribution network according to claim 1, characterized in that, The main station determines the adjustment target corresponding to the edge node based on the edge aggregation capability curve, including: The main station receives edge aggregation capability curves uploaded by multiple edge nodes, and determines the adjustable power boundary, duration, response time, adjustment cost, and execution confidence of each edge node based on each edge aggregation capability curve. The main station obtains the total regulation demand and scheduling constraint information corresponding to the virtual power plant; The master station determines the adjustment allocation result for each edge node based on the adjustable power boundary, duration, response time, adjustment cost, execution confidence, total adjustment requirement, and scheduling constraint information corresponding to each edge node. The master station generates the adjustment target corresponding to the edge node based on the adjustment allocation result, and sends the adjustment target to the edge node.

7. The method for resource aggregation and collaborative control of a low-voltage distribution network according to claim 1, characterized in that, The step of determining the community adjustment task corresponding to each resource community based on the adjustment target and the aggregation adjustment capability corresponding to each resource community, and generating and issuing control instructions corresponding to each distributed resource based on the community adjustment task, includes: Based on the adjustment target and the aggregation adjustment capability corresponding to each resource community, the adjustment priority corresponding to each resource community is determined; Based on the adjustment priority corresponding to each resource community, the adjustment target is assigned to the corresponding resource community to obtain the community adjustment task corresponding to each resource community. Obtain the current state and constraints of each distributed resource within the resource cluster, and determine the device adjustment task corresponding to each distributed resource based on the cluster adjustment task corresponding to each resource cluster, the current state, and the constraints. Based on the device adjustment tasks corresponding to each of the distributed resources, control instructions for the corresponding distributed resources are generated and issued.

8. A method for resource aggregation and collaborative control in a low-voltage distribution network according to claims 1 to 7, characterized in that, After generating and issuing control instructions for the corresponding distributed resources based on the community adjustment task, the method further includes: Obtain the execution feedback information returned by each of the distributed resources based on the control command, and determine the adjustment deviation between the actual adjustment result and the adjustment target based on the execution feedback information; When the adjustment deviation exceeds the preset deviation threshold, the availability coefficient of each distributed resource is re-determined based on the execution feedback information, and the aggregation adjustment capability of the corresponding resource community is updated. Based on the updated aggregation adjustment capability, the unfinished adjustment amount is redistributed, and supplementary control instructions are generated. The edge aggregation capability curve is corrected based on the updated aggregation adjustment capability, and the corrected edge aggregation capability curve is uploaded to the main site.

9. A low-voltage distribution network resource aggregation and collaborative control system, characterized in that, It includes edge nodes and a main station. The edge nodes include a feature extraction module, a community clustering module, an aggregation adjustment module, a curve uploading module, and an instruction issuing module. The feature extraction module is used to obtain the status information of multiple distributed resources corresponding to the edge node, and extract the feature vector corresponding to each distributed resource based on the status information. The community clustering module is used to cluster multiple distributed resources according to the feature vectors corresponding to each distributed resource to obtain multiple resource communities; The aggregation adjustment module is used to determine the aggregation adjustment capability corresponding to each resource community based on the feature vector of the distributed resources in each resource community. The curve upload module is used to generate an edge aggregation capability curve corresponding to the edge node based on the aggregation and adjustment capabilities corresponding to multiple resource clusters, and upload the edge aggregation capability curve to the main station so that the main station can determine the adjustment target corresponding to the edge node based on the edge aggregation capability curve. The instruction issuing module is used to receive the adjustment target issued by the master station, determine the cluster adjustment task corresponding to each resource cluster according to the adjustment target and the aggregation adjustment capability corresponding to each resource cluster, and generate and issue control instructions corresponding to each distributed resource according to the cluster adjustment task.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements a low-voltage distribution network resource aggregation and collaborative control method as described in any one of claims 1 to 8.

Images