Load aggregation method and device based on network load graph

Abstract

According to the load aggregation method and device based on the network load graph, by constructing the dynamically updated network load graph and establishing the network load coupling relation model, the constraint of a power grid topological structure on load adjustment is accurately quantified, and matching of a regulation and control strategy and a real-time system state is ensured; by introducing a multi-dimensional comprehensive aggregation evaluation system, the differences of the load in the aspects of characteristics, economy, response speed and the like are fully considered, and refined portraying and accurate aggregation of load resources are realized; and finally, through combination of multi-time scale aggregation and a multi-objective optimization algorithm, system operation indexes which conflict with each other are effectively coordinated, and an optimal load aggregation scheme which can quickly respond to system changes and can comprehensively optimize the regulation and control effect, the economic cost and the response speed is generated.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system load regulation, and in particular to a load aggregation method and device based on a grid-load one-map. BACKGROUND

[0002] Load regulation can be represented as centralized removal or power limitation of loads in a related area by a power grid dispatching center when a power shortage or local overload occurs in the system.

[0003] In related technologies, in the operation scene of dynamic changes in grid topology and complex and multiple load characteristics, there is a technical problem of extensive load regulation strategy caused by dependence on static topology modeling and single-dimensional aggregation method. SUMMARY

[0004] The technical problem to be solved by the present application is that in the operation scene of dynamic changes in grid topology and complex and multiple load characteristics, there is a technical problem of extensive load regulation strategy caused by dependence on static topology modeling and single-dimensional aggregation method. The purpose is to provide a load aggregation method and device based on a grid-load one-map, which solves the technical problem of extensive load regulation strategy.

[0005] The present application is implemented by the following technical solutions:

[0006] In a first aspect, the present application provides a load aggregation method based on a grid-load one-map, comprising:

[0007] Obtaining real-time operation data and topology data of a power grid, and generating a dynamically updated grid-load one-map based on the topology data;

[0008] Based on the grid-load one-map, a grid-load coupling relationship model is established; wherein the grid-load coupling relationship model is used to quantify the constraint degree of the grid topology structure on load regulation;

[0009] Based on a plurality of preset dimension indicators, a comprehensive aggregation evaluation value of each load in the grid-load one-map is calculated;

[0010] Based on the comprehensive aggregation evaluation value and the response time characteristics of each load, multi-time-scale load hierarchical aggregation is performed to obtain an initial aggregation scheme set;

[0011] A multi-objective optimization algorithm is used to optimize and solve the initial aggregation scheme set to generate one or more target load aggregation schemes, wherein the multi-objective optimization algorithm is used to cooperatively optimize a plurality of system operation indicators in the initial aggregation scheme set that conflict with each other.

[0012] Further, the step of obtaining real-time operation data and topology data of a power grid and generating a dynamically updated grid-load one-map based on the topology data comprises:

[0013] Collect heterogeneous data from multiple sources, including data acquisition and monitoring control systems, equipment monitoring systems, and load management systems.

[0014] The multi-source heterogeneous data is standardized to form standardized time-series data; wherein, the standardization process includes unifying at least one of timestamps, data encoding formats, and units of measurement.

[0015] Based on standardized time-series data, an integrated power network topology model is constructed as a single network-load map; wherein, the integrated power network topology model is used to characterize the complete electrical connection relationship from power source to load, and it defines the entity types, topological relationships and electrical parameter attributes from power plant to end user;

[0016] Based on real-time switch status signals obtained from the monitoring and data acquisition system, the power network topology model is dynamically updated using a topology change detection algorithm to generate the dynamically updated network load map.

[0017] Furthermore, the step of constructing an integrated power network topology model as a single network-load map based on standardized time-series data includes:

[0018] Define the core entity types in the power network topology model, which include at least power plants, substations, transmission lines, distribution transformers, and end users;

[0019] Define the topological relationships between the core entity types, which include at least electrical connection relationships, ownership relationships, and power supply relationships;

[0020] Each of the core entity types is assigned at least one attribute parameter, which includes at least one of electrical parameters, equipment status parameters, and load characteristic parameters.

[0021] Furthermore, the step of dynamically updating the power network topology model based on real-time switch status signals obtained from the monitoring and data acquisition system, using a topology change detection algorithm, to generate the dynamically updated network-load map includes:

[0022] Real-time monitoring of status change events of switching equipment in the power grid;

[0023] Based on the state change event and the predefined power grid connection rules, determine whether a change in the power grid topology connection relationship is triggered;

[0024] In response to the determination that the change has been triggered, the reconstruction process of the integrated power network topology model is initiated. The connection relationship of all electrical nodes in the network is updated according to the changed switch status to generate a network load map reflecting the current real-time operating status and output the topology change log.

[0025] Furthermore, the step of calculating the comprehensive aggregate evaluation value of each load in a network load diagram based on preset multiple dimensional indicators includes:

[0026] Multiple dimensions are identified for load aggregation evaluation; wherein the multiple dimensions include at least a geographical dimension, an electrical dimension, a load characteristic dimension, and an economic dimension;

[0027] Based on the determined multiple dimensions, for each load in the network load diagram, its standardized eigenvalue under each of the defined dimensions is calculated.

[0028] Based on the current system operating status or control objectives, adaptive weights are assigned to each of the aforementioned dimensions.

[0029] Based on the standardized characteristic value and the corresponding adaptive weight of each load, the comprehensive aggregate evaluation value of the load is obtained by weighted calculation.

[0030] Furthermore, the step of calculating the standardized eigenvalue of each load in each dimension of the network load map based on the determined multiple dimensions includes:

[0031] Establish differentiated load characteristic models for different types of loads;

[0032] Based on the load characteristic model, at least one control parameter is determined to characterize the dynamic response behavior of the load; wherein the control parameter includes at least one of a load response time characteristic parameter and a load regulation depth parameter; the load response time characteristic parameter characterizes the relationship between the actual response time of the load and the power regulation command; the load regulation depth parameter characterizes the relationship between the actual adjustable capacity of the load and the regulation duration.

[0033] Based on the determined control parameters, the standardized characteristic value of the load under the load characteristic dimension is calculated.

[0034] Furthermore, the step of performing multi-timescale load hierarchical aggregation based on the comprehensive aggregated evaluation value and the response time characteristics of each load to obtain an initial aggregation scheme set includes:

[0035] Based on the response time characteristics of each load, the load is divided into at least two response time levels; wherein, the at least two response time levels include at least two of the following: second-level response level, minute-level response level, and hour-level response level;

[0036] For the load within each response time level, clustering is performed based on its comprehensive aggregate evaluation value to form the load aggregation sub-scheme corresponding to that level.

[0037] The initial aggregation scheme set is generated based on the load aggregation sub-schemes at all response time levels.

[0038] Furthermore, the step of using a multi-objective optimization algorithm to optimize the initial aggregation scheme set and generate one or more target load aggregation schemes includes:

[0039] Based on the initial aggregated solution set, an optimized candidate solution set is generated;

[0040] A multi-objective optimization model is constructed using the conflicting system operation indicators as optimization objectives.

[0041] A multi-objective optimization algorithm based on Pareto dominance is used to iteratively search the set of optimization candidate solutions to identify the non-dominated solution set, wherein the solutions in the non-dominated solution set do not dominate each other on the multiple system operating indicators.

[0042] The non-dominated solution set is determined as one or more target load aggregation schemes.

[0043] In a second aspect, the present invention provides an electronic device, comprising: a memory, and one or more processors communicatively connected to the memory; the memory stores instructions executable by the one or more processors, the instructions being executed by the one or more processors to cause the one or more processors to implement the method described above.

[0044] Thirdly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0045] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0046] This invention provides a load aggregation method based on a grid-load integrated map. By constructing a dynamically updated grid-load integrated map and establishing a grid-load coupling relationship model, it accurately quantifies the constraints of the power grid topology on load regulation, ensuring the matching of regulation strategies with real-time system states. By introducing a multi-dimensional comprehensive aggregation evaluation system, it fully considers the differences in load characteristics, economy, and response speed, achieving refined profiling and accurate aggregation of load resources. Finally, by combining multi-timescale aggregation with multi-objective optimization algorithms, it effectively coordinates conflicting system operation indicators, generating an optimal load aggregation scheme that can quickly respond to system changes and comprehensively optimize regulation effects, economic costs, and response speed. This fundamentally solves the problems of coarse regulation strategies and poor adaptability caused by traditional methods relying on static modeling and single-dimensional aggregation, significantly improving the accuracy, efficiency, and reliability of power grid load regulation. Attached Figure Description

[0047] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0048] Figure 1 This is one of the flowcharts for a load aggregation method based on a single network load map provided in the embodiments of this specification;

[0049] Figure 2 This is a second flowchart illustrating a load aggregation method based on a single network-load diagram, as provided in the embodiments of this specification.

[0050] Figure 3 This is a block diagram of an electronic device provided in the embodiments of this specification. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0052] In related technologies, a new power system based on new energy sources is being rapidly constructed. This system's operating environment exhibits two significant characteristics: First, on the grid side, the large-scale and high-proportion integration of distributed energy sources (such as wind and solar power) means that the grid topology is no longer static but frequently changes with the activation and deactivation of new energy sources and adjustments to grid operation modes. Second, on the load side, load types are becoming increasingly diversified. In addition to traditional industrial, commercial, and residential loads, interactive loads such as electric vehicle charging loads and distributed energy storage are rapidly developing, with significant differences in characteristics (such as response speed and regulation capacity). This complex environment of highly interactive "source-grid-load" relationships presents unprecedented challenges to the grid's refined control capabilities.

[0053] Against this backdrop, load aggregation, as an effective technical means to tap the load-side regulation potential and support the safe and stable operation of the power grid, aims to aggregate a large number of dispersed and heterogeneous load resources into several controllable aggregates according to certain rules, so that the power grid dispatch center can perform unified and efficient control. In related technologies, typical schemes for implementing load aggregation often follow this process: First, based on an offline-planned, relatively static power grid topology model (often only including information on substations, lines, and other backbone networks), preliminary geographical partitioning or voltage level stratification of loads is performed; then, aggregation and grouping are mainly based on the physical location of the loads or simple type labels; finally, relatively unified control commands are formulated and issued to the aggregation groups.

[0054] However, after careful analysis, it was found that the aforementioned technical solutions, when dealing with the dynamic and complex environment of new power systems, have inherent defects that make it difficult to overcome the fundamental problem of extensive control strategies.

[0055] Specifically, firstly, because the reliance on static topology modeling methods makes it impossible to perceive and reflect the actual connection status of the power grid in real time (e.g., the switching of tie switches in the distribution network), the "network boundary" upon which aggregation is based is severely disconnected from the "electrical boundary" of real-time operation. Control strategies formulated based on this disconnect may fail to effectively isolate faults or overloads electrically, and may even exacerbate the operational risks of local power grids. This is the root cause of the "mismatch between control strategies and the actual system operating state."

[0056] Secondly, because aggregation is performed using a single dimension (e.g., geographical partitioning), ignoring the significant differences in loads across key dimensions such as electrical distance, response characteristics, and regulation costs, the loads within the aggregation group exhibit poor "homogeneity." This "one-size-fits-all" aggregation approach makes it difficult for control commands to accurately match the actual regulation capacity of each load within the group: either the regulation is insufficient, failing to effectively alleviate grid pressure; or it is excessive, causing unnecessary disruption to user electricity consumption and economic losses. This directly leads to insufficient control precision and response speed, as the control system requires longer decision-making time and greater safety margins to cope with this uncertainty.

[0057] In summary, the root cause of the crude technical problems in the load control strategies of related technologies lies in the static nature of the underlying modeling methods and the single aggregation dimension, which cannot adapt to the dynamically changing power grid and the diversified load characteristics.

[0058] Based on this, the core inventive concept of this invention is as follows: by constructing a dynamically updated network-load map as a unified data base, a network-load coupling relationship model with quantitative topological constraints is established on this basis, and comprehensive indicators covering multiple dimensions such as geography, electrical, characteristics, and economy are used to accurately profile the load. Then, by combining multi-timescale aggregation and multi-objective optimization algorithms, a lean load aggregation scheme that can quickly respond to changes in system state and comprehensively optimize multiple performance objectives is generated, thereby fundamentally solving the above-mentioned technical problems.

[0059] like Figure 1 and Figure 2 As shown, this embodiment provides a load aggregation method based on a grid-load diagram. The execution entity of the method can be a terminal or a server. Specifically, the terminal can be a controller of a microgrid or a park energy management system. In a microgrid or park-level energy system consisting of a factory, commercial complex, or residential community, this method can be executed by its energy management system (EMS) or central controller. This entity is responsible for optimizing the operation of its internal distributed resources such as photovoltaics, energy storage, charging piles, and air conditioning. Through this method, it achieves precise aggregation and optimized control of internal loads to achieve optimal energy efficiency or participate in interaction with the upper-level power grid.

[0060] The server can be a server in a power grid dispatch center. Provincial or municipal power grid dispatch centers are responsible for the safe, stable operation and optimized control of the power grid in the entire region. The server deployed there can serve as the execution entity for this method.

[0061] The server can also be a cloud computing platform or a distributed computing node, and this method can be deployed on a cloud server. Power grid companies can upload data collected from various substations and distribution automation terminals to the cloud platform, which will then perform load aggregation calculations uniformly.

[0062] The method may include:

[0063] Step S12: Obtain real-time operation data and topology data of the power grid, and generate a dynamically updated grid-load map based on the topology data.

[0064] In this embodiment, the acquisition action can be represented as the executing entity collecting, receiving, or calling data from multiple data sources of the power grid operation monitoring system.

[0065] Specifically, the implementing entity can obtain real-time telemetry (e.g., power, current, voltage) and communication (e.g., switch open / closed status) data of equipment such as switches, circuit breakers, and transformers in the power grid by subscribing to or directly from the data acquisition and monitoring control system (SCADA) via the data bus.

[0066] The implementing entity can obtain the static topology model of the power grid and equipment parameters (such as line impedance and transformer capacity) from the distribution management system (DMS) or asset management system (AMS) through the interface.

[0067] The implementing entity can obtain load data from the user side from the electricity information collection system or the Advanced Metering System (AMI).

[0068] In this embodiment, the real-time operational data can be represented as measured values ​​or event information that directly reflect the dynamic and transient operational state and changes of the power grid at one or more moments. Specifically, it may include:

[0069] Switch status signals can originate from data acquisition and monitoring control systems or power distribution automation systems. They can be communication signals that reflect the real-time on / off position of equipment such as circuit breakers, disconnect switches, and load switches.

[0070] Electrical measurement data may include active power, reactive power, current values ​​transmitted by lines and transformers, and voltage amplitudes at bus nodes.

[0071] Load power data, which can be expressed as the measured power on the user side or the outgoing line side of the distribution transformer, can be obtained from the electricity information acquisition system, advanced measurement system (AMI) or dedicated transformer user monitoring system, reflecting the real-time size and changing trend of the load.

[0072] Equipment operating status can include transformer tap position, capacitor bank switching status, and output of distributed power sources.

[0073] The topology data can be represented as static model data describing various physical devices (components) in the power grid and their permanent or planned connections. It may include:

[0074] The connection relationships of power grid components define how components such as power plants, substations, transmission lines, distribution lines, busbars, transformers, and switches are connected together through electrical nodes to form the network topology.

[0075] Parameters of power grid components can include static electrical parameters such as line resistance, reactance, susceptance (R / X / B parameters), transformer turns ratio, short-circuit impedance, and rated capacity of generators and loads.

[0076] Power grid component attributes can include management information such as the device's name, serial number, type, affiliated substation, and geographical coordinates.

[0077] In this embodiment, the implementing entity can construct a unified, integrated, and time-evolving network topology and load distribution correlation model that can uniformly represent the entire network from power plants, transmission lines, substations, distribution lines, distribution transformers to end users; this is the dynamically updated network-load map. The dynamic update can be expressed as the model changing with the network operating state (especially switch changes), maintaining consistency with the physical network.

[0078] In a specific implementation plan, firstly, the implementing entity can standardize the acquired multi-source heterogeneous data. For example, all data can be unified to the same time base (Beijing timestamp), data encoding (equipment IDs follow the IECCIM standard) and units of measurement can be standardized, and missing or abnormal data can be cleaned, interpolated, or labeled. Next, based on the standardized data, the implementing entity can construct an initial "station-line-transformer-customer" hierarchical topology model. This model defines each entity type (e.g., power plant, substation, transmission line, distribution transformer, end user) and their connection relationships (e.g., electrical connection, affiliation). Finally, dynamic updates are achieved through an adaptive topology reconstruction algorithm. For example, real-time monitoring of switch status change events in the SCADA system is used to determine whether the topology has changed according to predefined grid connection rules. Once a change is confirmed, a topology analysis program is triggered to recalculate the electrical connection relationships of the entire network, update the "grid-load single map" model, and record the topology change log.

[0079] Step S14: Based on the grid-load diagram, establish a grid-load coupling relationship model; wherein, the grid-load coupling relationship model is used to quantify the degree of constraint of the power grid topology on load regulation.

[0080] In this embodiment, the grid-load coupling relationship model can be represented as an analytical model, which is used to quantitatively analyze the spatial and electrical constraints of the physical structure (topology, parameters) of the power grid on the load power regulation behavior. Its purpose is to transform the constraints of the physical power grid into computable constraints in an optimization problem.

[0081] In one specific implementation, the implementing entity can perform the calculation of electrical distance. Specifically, this calculation can be used to quantify the degree of electrical coupling between any two nodes in the power grid network (e.g., a load node and a power source node, or between two load nodes). For example, the implementing entity can analyze the unique or optimal electrical path connecting the two nodes, multiply the impedance parameter of each segment of the path by its physical length, and sum the products of all segments on the path. The sum obtained is the electrical distance between the two nodes. The smaller this value, the tighter the electrical connection between the two nodes, the smaller the electrical loss in power transmission, and the smaller the impact of load power transfer between the two nodes.

[0082] In a specific implementation plan, the implementing entity can perform a transmission capacity assessment. Specifically, this assessment can be used to determine the maximum active power that a line or transformer can carry under all safe operating constraints. For example, it can comprehensively consider the thermal stability power limit determined by the equipment's own heat generation, the voltage stability power limit for maintaining grid voltage stability, and the power angle stability power limit for ensuring the system's power angle does not become unstable. The final transmission capacity value can be the minimum of these three power limits, serving as the most stringent safe transmission capacity upper limit for that branch.

[0083] Ultimately, the output of the network-load coupling relationship model can be a comprehensive network-load coupling degree matrix. This matrix integrates electrical distance information across the entire network and transmission capacity information of each branch, thereby enabling a quantitative assessment of the feasibility and cost of load regulation operations between any parts of the network from both spatial and capacity dimensions.

[0084] Step S16: Based on preset multiple dimension indicators, calculate the comprehensive aggregate evaluation value of each load in the network load map.

[0085] In this embodiment, the multiple dimensional indicators may include:

[0086] The geographical dimension is used to characterize the spatial attributes of the load, which may include the physical latitude and longitude coordinates of the load node, the administrative division to which it belongs (e.g., province, city, district / county) or the power supply business area of ​​the power grid company, etc.

[0087] The electrical dimension measures the location and impact of a load within the electrical structure of the power grid. Specifically, it can include the voltage level to which the load is connected (e.g., 10 kV, 380 V), the electrical distance between the load node and key nodes in the power grid (e.g., power grid connection points, current or potential overloaded lines or transformers), and the importance or centrality of the node in the network topology.

[0088] The load characteristics dimension describes the physical characteristics and operating behavior patterns of the load itself. Specifically, it can include the industry classification attributes of the load (e.g., whether it belongs to industrial load, commercial load, residential load, or electric vehicle charging load), its historical electricity consumption data curve and predicted future electricity demand curve, the upper and lower limits of the power range that the load can adjust when responding to control commands, the maximum power change that can be achieved in a single control, the response time required from receiving the command to starting to execute the action, and the duration of a single control action, etc.

[0089] The economic dimension is used to assess the economic costs and benefits of load participation in aggregated regulation. Specifically, it can include the economic compensation standard required to regulate the load per unit power, the sensitivity of the load's electricity consumption behavior to changes in electricity prices, and the load's historical commitment to fulfilling its obligations in demand response projects, etc.

[0090] In this embodiment, the executing entity can transform the heterogeneous information of a load across multiple dimensions into a single comparable scalar value through weighted aggregation. Specifically, firstly, the executing entity can standardize the original index values ​​for each dimension, for example, by using normalization or Z-score standardization, to eliminate the influence of different indicators due to differences in units and numerical ranges. Then, the executing entity can assign a weight coefficient to each dimension. This weight coefficient can be pre-set to a fixed value based on expert experience, or it can employ an adaptive dynamic adjustment mechanism. For example, it can automatically and dynamically adjust the weight coefficients of each dimension based on the core control objectives of the current power grid system (whether to prioritize system safety and stability or to prioritize reducing economic costs) or the real-time system operating status, using optimization algorithms such as gradient descent (or other algorithms, not limited to this one). Finally, for each specific load in the power grid, its standardized feature value for each dimension is multiplied by the corresponding weight coefficient. These multiplication results across all dimensions are then summed to obtain the final value, which is the comprehensive aggregate evaluation value of the load.

[0091] Step S18: Based on the comprehensive aggregate evaluation value and the response time characteristics of each load, perform multi-time-scale load hierarchical aggregation to obtain an initial aggregation scheme set.

[0092] In this embodiment, the response time characteristic can be expressed as the time parameter required for the load to achieve the expected adjustment effect from receiving the control command. For example, second-level response (rapid load shedding, used for emergency frequency control), minute-level response (demand response of air conditioners and electric vehicle charging piles, used for power balancing), hour-level response (industrial load transfer, used for economic optimization), etc.

[0093] In this embodiment, the multi-timescale load hierarchical aggregation can be represented as dividing the load into different groups based on the response time, clustering within each group, and finally coordinating the different groups to form an overall solution.

[0094] In a specific implementation plan, firstly, the implementing entity can divide the load into different time levels (seconds, minutes, hours) based on response time characteristics. Then, for loads within the same time level, the implementing entity can group them using their comprehensive aggregate evaluation value as the primary characteristic and employ clustering algorithms (K-means algorithm, DBSCAN algorithm) to form several load aggregation sub-schemes with similar characteristics. Each sub-scheme represents a set of loads that are similar in space, electrical characteristics, economics, and have consistent response speeds. Finally, the implementing entity can coordinate load aggregation sub-schemes from different time levels. For example, by setting priorities or coordination weights for each level, the second-level, minute-level, and hour-level sub-schemes can be combined into a complete initial aggregation scheme set with hierarchical response capabilities.

[0095] Step S110: Use a multi-objective optimization algorithm to optimize the initial aggregation scheme set and generate one or more target load aggregation schemes. The multi-objective optimization algorithm is used to collaboratively optimize multiple conflicting system operation indicators in the initial aggregation scheme set.

[0096] In this embodiment, firstly, the executing entity can use the initial set of aggregation schemes obtained in the previous step as the basic set of candidate schemes for deep optimization.

[0097] Next, the implementing entity can construct a multi-objective optimization function. This function can clearly define multiple specific system operation indicators that need to be optimized simultaneously. These indicators can conflict with each other; for example, they may include: indicators representing the quality of regulation effectiveness (the degree of elimination or mitigation of grid overload), indicators representing economic costs (the total compensation cost to load users), and indicators representing response speed (the overall time required from the issuance of the command to the load producing the expected regulation effect). The optimization task can be specifically defined as seeking a solution that simultaneously optimizes (i.e., minimizes or maximizes) the above multiple objective indicators.

[0098] Then, the implementing entity can use a specific multi-objective optimization algorithm based on Pareto dominance to solve the problem. Specifically, it can use non-dominated sorting genetic algorithm (NSGA-II) or multi-objective particle swarm optimization algorithm (MOPSO), etc. This algorithm repeatedly generates, evaluates, filters, and reorganizes the candidate solution set by simulating an iterative search process such as evolution or swarm intelligence. After multiple iterations and convergence, the algorithm will finally output a non-dominated solution set (i.e., a Pareto optimal solution set). In this solution set, any solution can satisfy the following condition: there is no other solution that is not inferior to it in all optimization objectives, and is strictly superior to it in at least one objective.

[0099] Finally, the non-dominated solution set obtained from the calculation is formally determined as one or more target load aggregation schemes that are ultimately output by this method.

[0100] This embodiment provides a load aggregation method based on a grid-load integrated map. By constructing a dynamically updated grid-load integrated map and establishing a grid-load coupling relationship model, it accurately quantifies the constraints of the power grid topology on load regulation, ensuring the matching of regulation strategies with real-time system states. By introducing a multi-dimensional comprehensive aggregation evaluation system, it fully considers the differences in load characteristics, economy, and response speed, achieving refined profiling and accurate aggregation of load resources. Finally, by combining multi-timescale aggregation with multi-objective optimization algorithms, it effectively coordinates conflicting system operation indicators, generating an optimal load aggregation scheme that can quickly respond to system changes and comprehensively optimize regulation effects, economic costs, and response speed. This fundamentally solves the problems of coarse regulation strategies and poor adaptability caused by traditional methods relying on static modeling and single-dimensional aggregation, significantly improving the accuracy, efficiency, and reliability of power grid load regulation.

[0101] In some implementations, the step of acquiring real-time operating data and topology data of the power grid, and generating a dynamically updated grid-load map based on the topology data, includes:

[0102] Step S122: Collect multi-source heterogeneous data from the data acquisition and monitoring control system, equipment monitoring system, and load management system.

[0103] In this embodiment, the data acquisition and monitoring control system can be deployed at the power grid dispatch center to collect real-time data from widely distributed nodes such as substations and power plants. The multi-source heterogeneous data provided by this data acquisition and monitoring control system can include communication data, which are status signals reflecting the real-time on / off positions of equipment such as circuit breakers and disconnect switches, and can also include telemetry data, which are continuously changing analog measurements such as active power, reactive power, current values, and bus voltage transmitted by lines and transformers.

[0104] In this embodiment, the equipment monitoring system can be an automated system focused on monitoring the operating status, health condition, and detailed parameters of one or more power devices. It is used for in-depth monitoring and condition assessment of important power equipment (transformers, reactors, capacitor banks, etc.). For example, for power transformers, the collected data may include winding temperature, oil temperature, oil level, dissolved gas analysis data, tap position, etc. For capacitor banks, the collected data may include switching status, reactive power output, etc.

[0105] The load management system can be defined as a dedicated system that collects electricity consumption information, analyzes electricity consumption behavior, and implements load control and management for electricity user-side loads. The data collected may include load power data, i.e., the active / reactive energy of the user or distribution transformer, and load profile information, such as user name, industry classification, power capacity, contact information, etc. It may also include load control status, such as the current switching status and controllability of controllable loads.

[0106] Step S124: Standardize the multi-source heterogeneous data to form standardized time-series data; wherein, the standardization process includes unifying at least one of timestamps, data encoding formats, and units of measurement.

[0107] In this embodiment, the processing of timestamp unification can be represented as the executing entity unifying the timestamps attached to data from different sources to the same time base and format.

[0108] In this embodiment, the processing of unified data encoding format can be represented as the executing entity mapping different codes or names for the same entity (such as a substation or a line) in different systems to a standard encoding system.

[0109] In this embodiment, the process of unifying the units of measurement can be represented as the executing entity unifying the data values ​​to a standard unit of measurement.

[0110] Step S126: Based on the standardized time-series data, construct an integrated power network topology model as a single network-load diagram; wherein, the integrated power network topology model is used to characterize the complete electrical connection relationship from power source to load, and defines the entity types, topological relationships and electrical parameter attributes from power plant to end user.

[0111] Step S128: Based on the real-time switch status signal obtained from the monitoring and data acquisition system, the power network topology model is dynamically updated through a topology change detection algorithm to generate the dynamically updated network load map.

[0112] In some implementations, the step of constructing an integrated power network topology model as a single network-load map based on standardized time-series data includes:

[0113] Step S1262: Define the core entity types in the power network topology model. The core entity types include at least power plants, substations, transmission lines, distribution transformers, and end users.

[0114] In this embodiment, the end user can be an industrial user, a commercial user, a residential user, or a specific electric vehicle charging station.

[0115] Step S1264: Define the topological relationships between the core entity types, wherein the topological relationships include at least electrical connection relationships, ownership relationships, and power supply relationships.

[0116] In this embodiment, the electrical connection relationship can be used to describe the physical connection relationship of the actual current flow path. For example, the entity of a transmission line establishes an electrical connection relationship with the entity of a power station and the entity of a substation through electrical nodes at its beginning and end, respectively.

[0117] In this implementation, affiliation can be used to describe the structural relationships of organizational management or asset ownership hierarchy. For example, an entity of a distribution transformer is administratively or asset-wise under the management of an entity of a substation. An entity of an end user is managed by an electricity service office for metering and billing.

[0118] In this embodiment, the power supply relationship can be used to describe the energy flow relationship in the final direction of electrical energy transmission.

[0119] Step S1266: Assign at least one attribute parameter to each of the core entity types, the attribute parameter including at least one of electrical parameters, equipment status parameters and load characteristic parameters.

[0120] In this embodiment, electrical parameters are parameters inherent to the entity and related to the electrical performance of the power grid. For example, impedance parameters such as resistance, reactance, conductance, and susceptance can be assigned to the entity of a transmission line.

[0121] In this embodiment, equipment status parameters can be represented as parameters indicating the operating status of an entity at a certain moment. For example, parameters indicating the open / closed state can be assigned to the entity of a circuit breaker in a substation.

[0122] In this embodiment, the load characteristic parameters can be represented as parameters used to describe the electricity consumption behavior patterns of end-user entities. For example, parameters such as historical daily load curves, monthly electricity consumption, typical electricity consumption periods, and interruptible load capacity can be assigned to end-user entities.

[0123] In some implementations, the step of dynamically updating the power network topology model based on real-time switch status signals obtained from a monitoring and data acquisition system using a topology change detection algorithm to generate the dynamically updated network-load map includes:

[0124] Step S1282: Real-time monitoring of status change events of switching equipment in the power grid.

[0125] In this embodiment, the executing entity can establish a real-time data interface with the SCADA system and continuously receive the data stream reported by the system, which includes communication signals of the status of all monitored devices.

[0126] In this embodiment, the state change event can be represented as a reversal of the remote signaling state value of the switching equipment (0 represents open, 1 represents closed). Once the change event is identified, the executing entity can encapsulate the event information. The encapsulated information may include at least: the unique identifier of the equipment that caused the change (equipment ID), the exact timestamp of the change, and the latest state value after the change, etc.

[0127] Step S1284: Based on the state change event and the predefined power grid connection rules, determine whether a change in the power grid topology connection relationship is triggered.

[0128] In this embodiment, the predefined power grid connection rules can serve as the basis for the execution entity's judgment. These rules can be a knowledge base storing multiple possible power grid wiring methods and the consequences of switch actions. The rules can be pre-defined manually based on the power grid's main electrical wiring diagram and power grid operation procedures. For example, one rule could be defined as: when the only incoming switch on a busbar is open, that busbar and all its connected outgoing lines will lose power; another rule could be defined as: when a tie switch is closed, it will connect two previously independent power supply areas, forming a loop.

[0129] In this embodiment, after receiving a change event, the preset judgment algorithm in the execution entity can match the device ID with the rule base to locate the relevant power grid area affected by the switch action. Then, the judgment algorithm can simulate the execution of the switch action and, in conjunction with the rule base, deduce the impact of the action on the current electrical connection state. For example, the algorithm can infer: Will the opening of the switch lead to the formation of an electrical island (i.e., an isolated network)? Or will its closing connect two electrical islands? etc. The reasoning process can ultimately produce a Boolean (true / false) output. If the deduction conclusion is that the switch action changes the network connectivity (i.e., forms a new electrical island, merges existing electrical islands, or significantly changes the power flow path), it is judged as a triggered topology change; if the action does not affect global connectivity (e.g., the operation of a disconnector switch within the same loop), it is judged as not triggered.

[0130] Step S1286: In response to the determination that the change has been triggered, the reconstruction process of the integrated power network topology model is triggered, the connection relationship of all electrical nodes in the network is updated according to the changed switch status, so as to generate a network load map reflecting the current real-time operating status and output the topology change log.

[0131] In this embodiment, when the judgment result of step S1284 is true (i.e., a change is triggered), the execution entity can automatically initiate the topology reconfiguration process. This topology reconfiguration process first obtains the latest state of all switches at the current moment, serving as the input basis for reconfiguration. The preset reconfiguration core algorithm in the execution entity (e.g., depth-first search or breadth-first search algorithm, etc.) can use the power source point as the root node and, based on the connection state of the switches (closed switches are considered connected, and open switches are considered disconnected), traverse the entire network and recalculate the connection relationships of all electrical nodes. This process can identify all energized live nodes and de-energized dead nodes and update the connection relationships between them.

[0132] In this implementation, the executing entity can update the integrated power network topology model with the latest connection relationship results obtained from topology analysis. That is, the connection status, energization status, and other attributes of nodes and branches in the integrated power network topology model need to be updated to maintain consistency with the actual operation of the physical power grid.

[0133] Finally, the executing entity can fuse and visualize the updated topology model with the latest real-time measurement data (e.g., line power flow, load power), ultimately outputting a dynamically updated network-load map. This map reflects the topology and real-time operating conditions of the power grid under the current switching state. Throughout the process, the executing entity can synchronously generate a topology change log. This log records at least the following information in detail: the original change event that triggered the change, the time of the change, the affected power grid area (e.g., which lines or substations changed state), and a brief comparison of the topology before and after the change. This log is used for system auditing, incident tracing, and operational analysis.

[0134] In some implementations, the step of calculating the comprehensive aggregate evaluation value of each load in a network load diagram based on preset multiple dimensional indicators includes:

[0135] Step S162: Determine multiple dimensions for load aggregation evaluation; wherein the multiple dimensions include at least a geographical dimension, an electrical dimension, a load characteristic dimension, and an economic dimension.

[0136] In this embodiment, in addition to the dimensions described above, it may also include:

[0137] The environmental dimension can be used to assess the impact of load electricity consumption behavior and its regulation on the ecological environment.

[0138] The safety and reliability dimension can be used to assess the impact of load regulation behavior on the safe and stable operation of the power system and the reliability of power supply.

[0139] Step S164: Based on the determined multiple dimensions, calculate the standardized eigenvalue of each load in each dimension for each load in the network load diagram.

[0140] In this implementation, for directly quantifiable indicators, the implementing entity can standardize the data. For example, for directly measurable or statistical indicators in dimensions such as geography, electricity, and economy (e.g., distance, cost, voltage level), mathematical methods (min-max normalization or Z-score standardization) can be used to eliminate differences in dimensions and orders of magnitude, converting them into dimensionless, comparable values.

[0141] To characterize load response behavior, the executing entity can construct differentiated models. Specifically, a load response time characteristic model can be established to represent the relationship between the actual response time of the load and the magnitude of the power regulation command. For example, a negative exponential function can be used, where the actual response time of the load is proportional to its initial response time and decreases exponentially with increasing absolute value of the power regulation command. Using this model, characteristic parameters representing the speed of the load's response can be calculated.

[0142] A load regulation depth model can also be established, which can be used to characterize the relationship between the actual adjustable capacity of the load and the regulation duration.

[0143] Step S166: Based on the current system operating status or control target, assign adaptive weights to each dimension;

[0144] In this implementation, adaptive weights can be assigned to each dimension based on the current system operating status. For example, when a line is detected to have a risk of severe overload, the weights of the electrical and safety / reliability dimensions are automatically increased, guiding the aggregation strategy to prioritize loads that are electrically close and help to quickly alleviate the overload.

[0145] In this embodiment, adaptive weights can also be assigned to each dimension based on the control objective. For example, if the core objective of this aggregation is optimal economic efficiency, the weight of the economic dimension will be increased; if the objective is to maximize the absorption of new energy sources, the weight of the environmental dimension will be increased.

[0146] Step S168: Based on the standardized characteristic value of each load and the corresponding adaptive weight, the comprehensive aggregate evaluation value of the load is obtained by weighted calculation.

[0147] In this embodiment, for each load in the power grid, the standardized feature value obtained in step S164 for each dimension is multiplied by the adaptive weight currently corresponding to that dimension, determined in step S166, to obtain the weighted score of the load in that dimension. Subsequently, these weighted scores across all dimensions are summed. The final result is the comprehensive aggregate evaluation value of the load.

[0148] In some implementations, the step of calculating the standardized eigenvalue of each load in each of the determined multiple dimensions in a network load map includes:

[0149] Step S1642: Establish differentiated load characteristic models for different types of loads.

[0150] In this embodiment, the executing entity can establish a load response time characteristic model, which can be used to accurately quantify the dynamic relationship between the actual response time of the load and the magnitude of the external power regulation command.

[0151] In a specific implementation, a negative exponential function can be used to characterize this relationship. Specifically, the model can be configured such that the actual response time of the load is proportional to its initial response time and decreases exponentially with increasing absolute value of the power regulation command. This model can be expressed as:

[0152]

[0153] In the formula, For the initial response time, The response coefficient is used to characterize the sensitivity of this type of load response. This is the absolute value of the power adjustment command.

[0154] In this embodiment, the executing entity can also establish a load regulation depth model. This model can be used to accurately quantify the dynamic process of how the actual adjustable capacity of the load changes with the regulation duration.

[0155] In a specific implementation, a saturation exponential curve can be used to characterize this relationship. The model can be configured as follows:

[0156] The actual adjustment depth of the load starts from zero, increases with the duration of adjustment, and asymptotically approaches its maximum adjustable depth. This model can be expressed as:

[0157]

[0158] In the formula, To maximize the adjustable depth, The regulation coefficient is used to characterize the speed at which the load regulation capacity is released. To adjust the duration.

[0159] Step S1644: Based on the load characteristic model, determine at least one control parameter to characterize the dynamic response behavior of the load; wherein the control parameter includes at least one of a load response time characteristic parameter and a load regulation depth parameter; the load response time characteristic parameter characterizes the relationship between the actual response time of the load and the power regulation command; the load regulation depth parameter characterizes the relationship between the actual adjustable capacity of the load and the regulation duration.

[0160] In this embodiment, for an established load response time characteristic model (e.g., a model based on a negative exponential function), key parameters in the model, such as initial response time and response coefficient, can be determined by analyzing historical data or experimental tests. The combination of these parameters (i.e., the model itself) constitutes the control parameters characterizing the response time characteristics of this type of load. It clearly describes the quantitative relationship of how the actual response time dynamically changes with power regulation commands.

[0161] In this embodiment, for an established load regulation depth model (e.g., a model based on a saturation exponential curve), its key parameters, such as the maximum regulation depth and regulation coefficient, can also be determined through parameter identification. The combination of these parameters constitutes the control parameters characterizing the load regulation depth characteristics of this type of load.

[0162] Step S1646: Based on the determined control parameters, calculate the standardized characteristic value of the load in the load characteristic dimension.

[0163] In this embodiment, the control parameters characterizing the dynamic response of the load can be directly used as standardized feature values. For example, the values ​​of the response coefficient and the regulation coefficient can be dimensionlessly processed and used as feature values ​​of the load in terms of response speed and regulation capability.

[0164] In this embodiment, the control parameters described above can also be used to calculate their characteristic values ​​under a certain standard scenario. For example, a load response time characteristic model can be used to calculate the expected response time under a standard unit power regulation command, and this time value can be used as a standardized characteristic value. Similarly, a load regulation depth model can be used to calculate the regulation depth that can be achieved within a standard duration, and this depth value can be used as a standardized characteristic value.

[0165] In some implementations, the step of performing multi-timescale load hierarchical aggregation based on the comprehensive aggregated evaluation value and the response time characteristics of each load to obtain an initial aggregation scheme set includes:

[0166] Step S182: Based on the response time characteristics of each load, divide the load into at least two response time levels; wherein, the at least two response time levels include at least two of the following: second-level response level, minute-level response level, and hour-level response level.

[0167] In this embodiment, the second-level response layer is used to aggregate load resources with extremely fast response times. Its application scenario can be emergency safety control of the power grid, such as rapidly shedding loads to suppress frequency collapse and prevent the escalation of accidents. Loads included in this layer can have response capabilities ranging from milliseconds to seconds.

[0168] The minute-level response tier is used to aggregate load resources with medium response speeds. Its core application scenarios include medium-term regulation and demand response, such as balancing renewable energy power fluctuations and participating in peak shaving. Loads included in this tier can have minute-level response capabilities.

[0169] The hourly response level is used to aggregate load resources with slower response times but greater adjustment potential and longer durations. Examples include high-energy-consuming industrial production lines, large thermal / cold storage devices, and manufacturing enterprises that adjust production plans.

[0170] Step S184: For the load within each response time level, cluster and group it based on its comprehensive aggregate evaluation value to form the load aggregation sub-scheme corresponding to that level.

[0171] In this embodiment, for all loads that have been classified into the same response level (e.g., minute-level response level), their comprehensive aggregated evaluation value is used as the core feature vector. Clustering algorithms can be used to group the loads.

[0172] Specifically, this algorithm can automatically group loads with similar comprehensive aggregate evaluation values ​​into the same cluster. Each clustering result constitutes a load aggregation sub-scheme. This sub-scheme can specify which specific loads make up a controllable aggregation unit. For example, at the minute-level response level, multiple sub-schemes can be formed, such as "Area A commercial air conditioning cluster" and "Area B electric vehicle charging cluster".

[0173] Step S186: Generate the initial aggregation scheme set based on the load aggregation sub-schemes of all response time levels.

[0174] In this embodiment, a time-scale coordination algorithm can be used for integration to generate the initial aggregation scheme set.

[0175] Specifically, the following formula can be used for collaborative optimization:

[0176]

[0177] In the formula, Let be the weight of the i-th time scale in the overall optimization objective. This represents the aggregation result (i.e., sub-scheme) at the i-th time scale (seconds, minutes, hours). This weight... It can also be dynamically adjusted based on the real-time operating status of the power grid and the control objectives. For example, when an emergency fault occurs in the system, the weight of the second-level response level is significantly increased, and when the system needs to smooth out fluctuations, the weight of the minute-level response level is increased.

[0178] In this implementation, the coordination algorithm ultimately outputs an initial set of aggregation schemes. This set of schemes is a hierarchical and partitioned coordination strategy. It explicitly includes complete guidance on how to coordinate and invoke second-level, minute-level, and hour-level load aggregation sub-schemes under what system states, providing a high-quality decision-making basis for subsequent multi-objective optimization.

[0179] In some implementations, the step of using a multi-objective optimization algorithm to optimize the initial aggregation scheme set and generate one or more target load aggregation schemes includes:

[0180] Step S1102: Generate an optimized candidate solution set based on the initial aggregation solution set.

[0181] In this embodiment, clustering algorithms can be used to expand the initial aggregation scheme. Specifically, machine learning methods such as K-means or DBSCAN can be used to perform cluster analysis on the load aggregation groups in the initial scheme set. By adjusting the parameters of the clustering algorithm (e.g., the number of cluster centers K in K-means, or the neighborhood radius eps in DBSCAN), multiple candidate grouping methods with subtle differences in grouping granularity, intra-group similarity, and inter-group differences can be generated. Each grouping method constitutes a candidate load aggregation scheme. By combining these candidate schemes generated from different clustering results, a richer and more diverse set of optimized candidate schemes is formed.

[0182] Step S1104: Using the conflicting system operation indicators as optimization objectives, construct a multi-objective optimization model.

[0183] In this embodiment, the objective of the multi-objective optimization model may include:

[0184] The objective function f1(x) for the control effect is used to quantify the degree to which the load aggregation scheme improves the safe operation of the power grid. For example, it can be defined as the degree to which the power margin of heavily overloaded lines or equipment in the power grid is improved after the implementation of the scheme, or the degree to which the voltage deviation of voltage-over-limit nodes is reduced. The smaller the objective value, the better the control effect.

[0185] The economic cost objective f2(x) is used to quantify the total economic cost of implementing the load aggregation scheme. For example, it can be defined as the sum of compensation fees payable to all users participating in the load. The smaller this objective value, the better the economic efficiency.

[0186] The response speed objective f3(x) is used to quantify the overall response speed of the solution. For example, it can be defined as the weighted average response time of all invoked loads in the solution. The smaller the objective value, the faster the response speed.

[0187] The final optimization model can be expressed as finding the decision variable x such that the objective function F(x) = [f1(x), f2(x), f3(x)] reaches its optimum (i.e. minimizes).

[0188] Step S1106: Use a multi-objective optimization algorithm based on Pareto dominance to iteratively search the set of optimization candidate solutions and identify the non-dominated solution set, wherein the solutions in the non-dominated solution set do not dominate each other in the multiple system operating indicators.

[0189] In this embodiment, the non-dominated sorting genetic algorithm (NSGA-II) can be used to solve the problem.

[0190] Specifically, it may include:

[0191] 1. The set of optimized candidate solutions generated in step S1102 can be used as the initial population to start the iterative search;

[0192] 2. In each generation of the population, the algorithm can evaluate all solutions based on the Pareto dominance relation. If a solution is no worse than another solution in all objectives, and is strictly better in at least one objective, then the former is said to dominate the latter. Based on this relation, the solutions in the population are divided into different non-dominated ranks, where the solution in Rank 0 is not dominated by any other solution and is called a non-dominated solution.

[0193] 3. To maintain population diversity, the algorithm calculates the crowding degree of each solution in the target space, prioritizing solutions located in sparse regions. Tournament selection, combining non-dominated sorting and crowding degree, selects superior individuals for the next generation.

[0194] 4. Perform crossover and mutation operations on the selected individuals to generate new offspring populations, thereby exploring new possible solutions.

[0195] 5. After multiple iterations and convergence, the algorithm finally outputs a non-dominated solution set. Any solution in this set satisfies the following condition: any improvement on any objective will lead to a deterioration of at least one other objective. These solutions form a Pareto front in the objective function space.

[0196] Step S1108: Determine the non-dominated solution set as one or more target load aggregation schemes.

[0197] In this implementation, the non-dominated solution set output by the algorithm is itself defined as the final target load aggregation scheme set.

[0198] In one specific implementation plan, a hierarchical and partitioned load aggregation and heavy overload control scheme based on a single network load map is provided.

[0199] Specifically, with the construction of new power systems, power loads are becoming increasingly diversified and complex, making traditional extensive load management models inadequate for the needs of refined regulation. Currently, there are three main problems: First, the accuracy of load regulation strategies is insufficient. Traditional methods often employ regional, large-scale load shedding measures, lacking refined differentiation and targeted regulation strategies for different types of loads. This results in low regulation accuracy, long response times, and a lack of multi-timescale coordination, failing to comprehensively consider the coordinated cooperation of second-level rapid response, minute-level medium-term adjustment, and hour-level long-term optimization. Second, the ability to model grid-load topology relationships is insufficient. Traditional grid modeling is mostly based on static topology structures, making it difficult to adapt to the new power system operating environment of large-scale distributed energy access and dynamic network topology changes. It lacks in-depth modeling of the coupling relationship between grid topology and load distribution, making it impossible to achieve load optimization and dynamic regulation based on grid structure characteristics. Third, hierarchical and regional load aggregation technology is immature. Existing load aggregation methods often use simple geographical partitioning or voltage level hierarchies, failing to fully consider the comprehensive aggregation of multi-dimensional factors such as load characteristics, response speed, and regulation capacity. This lack of adaptability to dynamic changes in external factors prevents dynamic adjustment of aggregation strategies based on system operating status and load characteristic changes. Especially in scenarios of heavy overload in the main and distribution networks, existing technologies cannot achieve accurate load location, intelligent aggregation decision-making, and rapid hierarchical response, which seriously restricts the safe and stable operation of the power system and the improvement of lean management level.

[0200] The hierarchical and regional load aggregation and heavy overload control scheme based on a single network load map may include:

[0201] Step 1: First, integrate heterogeneous data from multiple sources such as SCADA system, equipment monitoring system, and load management system. Standardize the timestamps, coordinate system, and coding specifications, establish a unified data model, and map the original records into comparable feature vectors. Mark missing and abnormal data for subsequent processing.

[0202] An innovative data quality assessment formula is adopted:

[0203]

[0204] In the formula, This represents the data quality weight, which is the quality weight score of the i-th data segment. Let i be the proportion of missing data points in the i-th data segment out of the total expected data points. Let represent the severity of noise in the i-th data segment. and To adjust parameters, methods such as window alignment, interpolation, and denoising can be used to form time-series data fragments and quality weight matrices, providing quantifiable weighting basis for subsequent construction of a single network-load map.

[0205] Step 2: Construct an integrated power network topology model of "station-line-transformer-customer", define core entity types covering power plants, substations, transmission lines, transformer equipment, end users, etc., clarify semantic relationships such as power grid topology association, equipment structure connection, load distribution relationship, and electrical distance dependence, and assign rich attributes such as load characteristics, electrical parameters, equipment status, and transmission capacity to each entity and relationship.

[0206] Step 3: In the presence of missing data, noise interference, inconsistent formats, or semantic ambiguity, the adaptive topology reconstruction algorithm is activated. Based on real-time SCADA data and topology change detection algorithm, and through technical means such as switch status monitoring, line switching detection, and equipment start-up and shutdown identification, the dynamic perception and automatic updating of the power grid topology structure are realized, and the updated grid load map G and its topology change information are output.

[0207] Step 4: Model the network-load coupling relationship based on a single network-load diagram G, using the electrical distance calculation formula:

[0208]

[0209] In the formula, This represents the electrical distance between node i and node j, used to quantify the degree of electrical coupling or proximity between the two nodes. Let be the impedance of the k-th segment of the line along the path. Let k be the length of the k-th segment of the path.

[0210] The transmission capacity assessment formula is used as follows:

[0211]

[0212] In the formula, , , The thermal stability limit, voltage stability limit, and power angle stability limit are respectively used to establish a correlation model between the power grid topology and load distribution characteristics, output the grid-load coupling degree matrix C, and quantify the degree of constraint and support capability of the network structure on load regulation.

[0213] Step 5: Employ a multi-dimensional aggregation index calculation algorithm to establish a comprehensive aggregation index system covering geographical dimensions (spatial distance, power supply area), electrical dimensions (voltage level, electrical distance, network topology), characteristic dimensions (load type, regulation capacity, response speed), and economic dimensions (regulation cost, electricity price sensitivity). The comprehensive aggregation index calculation formula is as follows:

[0214]

[0215] In the formula, This is the comprehensive aggregate evaluation value of the i-th load. Let j be the weight of the j-th dimension. For the standardized feature value of the i-th load in the j-th dimension, an adaptive adjustment mechanism based on index weights is used:

[0216]

[0217] In the formula, Let be the weight of the j-th dimension at time t. For the loss function L, the weights The gradient indicates the direction of weight adjustment. For learning rate, The loss function is used to dynamically optimize the aggregation strategy based on the system's operating status and control objectives, and output the aggregation index matrix A to provide a quantitative basis for subsequent aggregation decisions.

[0218] Step 6: Construct a heterogeneous load characteristic modeling method. Establish differentiated characteristic models for different types of loads, such as industrial loads, commercial loads, residential loads, and electric vehicle charging loads, using the load response characteristic formula:

[0219]

[0220] In the formula, For the initial response time, For the response coefficient, The adjustment depth formula is used to represent the power change:

[0221]

[0222] In the formula, To maximize the adjustable depth, For adjustment coefficients, To adjust the time, including key parameters such as load response time constant, adjustment depth, duration, and recovery characteristics, a technical foundation is provided for precise aggregation.

[0223] Step 7: Employ a multi-timescale aggregation algorithm to design a hierarchical aggregation mechanism comprising second-level fast response aggregation (emergency load shedding), minute-level medium-term adjustment aggregation (demand response), and hour-level long-term optimization aggregation (load transfer), using the following timescale coordination formula:

[0224]

[0225] In the formula, The weight for the i-th time scale (seconds / minutes / hours) For the aggregation result at the i-th time scale, the time scale coordination algorithm is used to achieve collaborative optimization of the aggregation strategy under different time dimensions, and output the hierarchical and partitioned load aggregation result L.

[0226] Step 8: Construct an intelligent aggregation decision engine. Based on reinforcement learning and multi-objective optimization algorithms, use clustering algorithms (K-means, DBSCAN) to generate multiple aggregation candidate schemes, and establish an optimization model that includes multiple objectives such as regulation effect, economic cost, and response speed. The multi-objective optimization function is as follows:

[0227]

[0228] In the formula, To achieve the desired regulatory effect, For the economic cost objective, To achieve the target response speed, the Non-Dominated Sorting Genetic Algorithm (NSGA-II) is used to solve for the Pareto optimal solution set. The system dynamically learns the optimal aggregation strategy based on information such as historical regulation effects, real-time system status, and predicted load demand, and outputs the optimal aggregation scheme P.

[0229] Step 9: Activate the intelligent identification method for heavy overload bottlenecks, based on real-time power flow calculation and safety constraint analysis, and adopt the overload risk assessment formula:

[0230]

[0231] In the formula, This is the actual power. Rated power, For duration, At the critical time, through methods such as N-1 safety verification, short-circuit current analysis, and voltage stability assessment, the system automatically identifies overloaded lines, transformers, and weak nodes in the main and distribution networks, predicts potential overload risks and evolution trends, and outputs the heavy overload bottleneck identification result B and its risk level.

[0232] Step 10: Employ a load migration optimization algorithm with electrical distance constraints, establish a load migration optimization model based on electrical distance and transmission constraints, and adopt the load migration optimization objective function:

[0233]

[0234] In the formula, Let i be the electrical distance of the i-th segment of the migration path. The weight coefficients for this path. The additional line loss cost of the power grid after load relocation. For loss weight, To provide economic compensation to load users for the inconvenience caused by control measures, To compensate for the weighting, the load composition, transmission margin, and regulation potential of the overloaded lines are analyzed. Based on a network-load map, a target area with transmission margin is searched. An improved Dijkstra algorithm is used to calculate the shortest electrical distance path from the overloaded area to the target area. Under the premise of satisfying network security constraints, network loss and regulation cost are minimized, and the optimal migration path set P is output.

[0235] Step 11: First, construct a graded response strategy design. Based on factors such as overload severity, duration, and impact range, the overload situation is divided into three levels: mild, moderate, and severe. Finally, a multi-objective collaborative optimization algorithm is adopted to design a collaborative optimization model that considers multiple objectives such as control effect, economic cost, user satisfaction, and system security. The multi-objective collaborative optimization function is as follows:

[0236]

[0237] in + + + =1, through Pareto optimal solution set solving, dynamic weight adjustment, constraint relaxation and other techniques, to achieve balance optimization among multiple objectives, ensure the comprehensive optimality of the control strategy, and output multi-objective collaborative control scheme M, the entire process is completed in seconds.

[0238] According to an embodiment of the present invention, an electronic device is provided; please refer to... Figure 3 The electronic device in this embodiment may include one or more of the following components: a processor, a network interface, memory, non-volatile memory, and one or more application programs, wherein the one or more application programs may be stored in non-volatile memory and configured to be executed by one or more processors, and the one or more programs are configured to perform the methods as described in the foregoing method embodiments.

[0239] According to embodiments of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a computer, causes the computer to perform the method described in any of the above embodiments.

[0240] According to embodiments of the present invention, a computer program product comprising instructions is also provided, which, when executed by a computer, cause the computer to perform a method in any of the above embodiments.

[0241] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A load aggregation method based on a single network load map, characterized in that, include: Acquire real-time operation data and topology data of the power grid, and generate a dynamically updated grid-load map based on the topology data; Based on the aforementioned grid-load diagram, a grid-load coupling relationship model is established; wherein, the grid-load coupling relationship model is used to quantify the degree of constraint of the power grid topology on load regulation; Based on multiple preset indicators, the comprehensive aggregate evaluation value of each load in the network load diagram is calculated. Based on the comprehensive aggregate evaluation value and the response time characteristics of each load, load hierarchical aggregation is performed at multiple time scales to obtain an initial aggregation scheme set; A multi-objective optimization algorithm is used to optimize the initial aggregation scheme set to generate one or more target load aggregation schemes. The multi-objective optimization algorithm is used to collaboratively optimize multiple conflicting system operation indicators in the initial aggregation scheme set.

2. The loading polymerization method according to claim 1, characterized in that, The step of acquiring real-time operation data and topology data of the power grid, and generating a dynamically updated grid-load map based on the topology data, includes: Collect heterogeneous data from multiple sources, including data acquisition and monitoring control systems, equipment monitoring systems, and load management systems. The multi-source heterogeneous data is standardized to form standardized time-series data; wherein, the standardization process includes unifying at least one of timestamps, data encoding formats, and units of measurement. Based on standardized time-series data, an integrated power network topology model is constructed as a single network-load map; wherein, the integrated power network topology model is used to characterize the complete electrical connection relationship from power source to load, and it defines the entity types, topological relationships and electrical parameter attributes from power plant to end user; Based on real-time switch status signals obtained from the monitoring and data acquisition system, the power network topology model is dynamically updated using a topology change detection algorithm to generate the dynamically updated network load map.

3. The loading polymerization method according to claim 2, characterized in that, The step of constructing an integrated power network topology model as a single network-load map based on standardized time-series data includes: Define the core entity types in the power network topology model, which include at least power plants, substations, transmission lines, distribution transformers, and end users; Define the topological relationships between the core entity types, which include at least electrical connection relationships, ownership relationships, and power supply relationships; Each of the core entity types is assigned at least one attribute parameter, which includes at least one of electrical parameters, equipment status parameters, and load characteristic parameters.

4. The loading polymerization method according to claim 2, characterized in that, The step of dynamically updating the power network topology model based on real-time switch status signals obtained from the monitoring and data acquisition system, and generating the dynamically updated network-load map, includes: Real-time monitoring of status change events of switching equipment in the power grid; Based on the state change event and the predefined power grid connection rules, determine whether a change in the power grid topology connection relationship is triggered; In response to the determination that the change has been triggered, the reconstruction process of the integrated power network topology model is initiated. The connection relationship of all electrical nodes in the network is updated according to the changed switch status to generate a network load map reflecting the current real-time operating status and output the topology change log.

5. The loading polymerization method according to claim 1, characterized in that, The step of calculating the comprehensive aggregate evaluation value of each load in a network load diagram based on preset multiple dimension indicators includes: Multiple dimensions are identified for load aggregation evaluation; wherein the multiple dimensions include at least a geographical dimension, an electrical dimension, a load characteristic dimension, and an economic dimension; Based on the determined multiple dimensions, for each load in the network load diagram, its standardized eigenvalue under each of the defined dimensions is calculated. Based on the current system operating status or control objectives, adaptive weights are assigned to each of the aforementioned dimensions. Based on the standardized characteristic value and the corresponding adaptive weight of each load, the comprehensive aggregate evaluation value of the load is obtained by weighted calculation.

6. The loading polymerization method according to claim 5, characterized in that, The step of calculating the standardized eigenvalue of each load in each dimension of the network load map based on the determined multiple dimensions includes: Establish differentiated load characteristic models for different types of loads; Based on the load characteristic model, at least one control parameter is determined to characterize the dynamic response behavior of the load; wherein, the control parameter includes at least one of a load response time characteristic parameter and a load regulation depth parameter; the load response time characteristic parameter characterizes the relationship between the actual response time of the load and the power regulation command; the load regulation depth parameter characterizes the relationship between the actual adjustable capacity of the load and the regulation duration. Based on the determined control parameters, the standardized characteristic value of the load under the load characteristic dimension is calculated.

7. The loading polymerization method according to claim 1, characterized in that, The step of performing multi-time-scale load hierarchical aggregation based on the comprehensive aggregated evaluation value and the response time characteristics of each load to obtain an initial aggregation scheme set includes: Based on the response time characteristics of each load, the load is divided into at least two response time levels; wherein, the at least two response time levels include at least two of the following: second-level response level, minute-level response level, and hour-level response level; For the load within each response time level, clustering is performed based on its comprehensive aggregate evaluation value to form the load aggregation sub-scheme corresponding to that level. The initial aggregation scheme set is generated based on the load aggregation sub-schemes at all response time levels.

8. The loading polymerization method according to claim 1, characterized in that, The step of using a multi-objective optimization algorithm to optimize the initial aggregation scheme set and generate one or more target load aggregation schemes includes: Based on the initial aggregated solution set, an optimized candidate solution set is generated; A multi-objective optimization model is constructed using the conflicting system operation indicators as optimization objectives. A multi-objective optimization algorithm based on Pareto dominance is used to iteratively search the set of optimization candidate solutions to identify the non-dominated solution set, wherein the solutions in the non-dominated solution set do not dominate each other on the multiple system operating indicators. The non-dominated solution set is determined as one or more target load aggregation schemes.

9. An electronic device, characterized in that, include: A memory, and one or more processors communicatively connected to the memory; The memory stores instructions that can be executed by the one or more processors to cause the one or more processors to implement the method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 8.

Images