A cross-regional virtual power plant collaborative scheduling method, device, medium and product

Abstract

The application discloses a kind of cross-regional virtual power plant collaborative scheduling method, equipment, medium and product, it is related to data processing field.The method includes collecting the real-time characteristic data such as time-space position, output / demand prediction of each distributed energy resource and load, determines each characteristic dimension dynamic weight based on the global optimization target and data of current scheduling period;According to dynamic weight and real-time data, generate dynamic resource cluster division instruction containing member list and collaborative constraint condition, and send initial cross-regional collaborative scheduling instruction containing net exchange power target value and the like and compensation price signal to each dynamic resource cluster local agent;After receiving the aggregated response boundary information returned by local agent, update instruction and signal, obtain target collaborative scheduling instruction, finally sent to each dynamic resource cluster execution, realize the effective control of cross-regional virtual power plant resource.The application can alleviate the problem that the adaptability of cross-regional virtual power plant collaborative scheduling instruction and actual resource capacity is not enough.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a cross-regional virtual power plant collaborative scheduling method, device, medium and product. BACKGROUND

[0002] A virtual power plant (VPP) is a smart energy system that aggregates and optimizes distributed resources such as distributed power generation, energy storage, and controllable loads through information communication technology and software systems. Its essence is to integrate fragmented energy resources into a "virtual" power plant with flexible adjustment capabilities to participate in the operation and market transactions of the power system, achieving efficient use of resources and safe and stable operation of the power grid.

[0003] Among them, cross-regional virtual power plant collaborative scheduling is a key means to address the contradiction between high proportion of renewable energy grid connection and uneven distribution of energy supply and demand in time and space. By breaking geographical boundaries, it integrates the complementary nature of wind and light resources in different climate zones, load characteristic differences, and cross-regional backup sharing capabilities, optimizes resource allocation on a larger spatial and temporal scale, effectively mitigates distributed energy volatility, relieves local grid congestion, and improves system overall resilience and economic efficiency.

[0004] However, in the collaborative scheduling of cross-regional virtual power plants, related technologies usually use fixed partitioning or type-based resource grouping mechanisms (such as dividing aggregated units according to administrative boundaries or resource types). This static aggregation method has a large contradiction with the dynamic and variable temporal and spatial characteristics, response capabilities, and cost structure of distributed resources, resulting in a state of "aggregation without integration" - that is, the physically forced aggregated resource group cannot efficiently collaborate due to the lack of target adaptability, causing insufficient adaptability of scheduling instructions to actual resource capabilities. SUMMARY

[0005] To solve the above technical problems and defects, the purpose of the present application is to provide a cross-regional virtual power plant collaborative scheduling method, device, medium and product, which can alleviate the problem of insufficient adaptability of cross-regional virtual power plant collaborative scheduling instructions to actual resource capabilities.

[0006] To achieve the above object, in a first aspect, the application provides a cross-regional virtual power plant collaborative scheduling method, comprising: collecting real-time characteristic data of each distributed energy resource and load, the real-time characteristic data at least including one of space-time position information, output / demand prediction curve, physical operation parameter, economic characteristic parameter and schedulable state; determining dynamic weights of each characteristic dimension based on a global optimization objective set in a current scheduling period and the real-time characteristic data; generating a dynamic resource cluster division instruction according to the dynamic weights and the real-time characteristic data, the dynamic resource cluster division instruction including a member list of a resource cluster and a collaborative constraint condition; sending an initial cross-regional collaborative scheduling instruction and a compensation price signal to each local agent corresponding to the dynamic resource cluster based on the dynamic resource cluster division instruction, the initial cross-regional collaborative scheduling instruction including a net exchange power target value and an auxiliary service demand; receiving aggregated response boundary information returned by each local agent based on the initial cross-regional collaborative scheduling instruction and the compensation price signal, the aggregated response boundary information being used to represent a feasible domain range of the dynamic resource cluster to the instruction; updating the collaborative scheduling instruction and the compensation price signal according to the aggregated response boundary information to obtain a target collaborative scheduling instruction; and sending the target collaborative scheduling instruction to each dynamic resource cluster for execution to control cross-regional virtual power plant resource operation.

[0007] The application adopts the above method steps, determines the dynamic weights of each characteristic dimension by collecting real-time characteristic data and combining with the global optimization objective, so that the resource division is more suitable for actual working conditions; when generating the dynamic resource cluster division instruction, the resource characteristics and physical constraints are comprehensively considered to ensure the executability of the division result. In the scheduling instruction generation and optimization process, the initial cross-regional collaborative scheduling instruction and the compensation price signal are first sent, and then the collaborative scheduling instruction and the compensation price signal are updated according to the aggregated response boundary information returned by the local agent to form a closed-loop feedback mechanism, the instruction is dynamically adjusted according to the actual feasible domain range of the resource, and the resource capacity is accurately matched. Finally, the target collaborative scheduling instruction is sent to control the resource operation, the scheduling instruction is highly adapted to the actual capacity of the resource, and the effectiveness and reliability of the cross-regional virtual power plant collaborative scheduling are improved.

[0008] Optionally, in some embodiments, the determination of the dynamic weights of each characteristic dimension based on the global optimization objective set in the current scheduling period and the real-time characteristic data comprises: determining a key characteristic dimension identifier according to the global optimization objective of the current scheduling period; calculating an influence weight coefficient of each dimension on the target achievement based on the key characteristic dimension identifier; and correcting the influence weight coefficient according to a resource state constraint in the real-time characteristic data to obtain the dynamic weight; the resource state constraint refers to a physical boundary condition limiting the resource scheduling feasible domain, including maximum / minimum power limit value, energy storage SOC safety interval, device start-stop state and network topology connection relationship.

[0009] By adopting the technical solutions in the above embodiments, the feature dimension weight is dynamically corrected by combining the current global optimization target and the resource state constraint, so that the feature weight used for resource cluster division can more accurately reflect the actual influence of the resource on the current scheduling target and the physical feasibility, and the effectiveness of subsequent resource aggregation and the rationality of the scheduling instruction are improved.

[0010] Optionally, in some embodiments, based on the key feature dimension identification, the influence weight coefficient of each dimension on the target achievement is calculated, including: querying a preset target feature mapping rule library to obtain a benchmark influence value of each key feature dimension on the optimization target; based on the benchmark influence value and a resource dynamic characteristic in the real-time feature data, an actual contribution correction factor of each dimension on the target achievement is calculated; the resource dynamic characteristic refers to the real-time change of the operating capacity parameter of the distributed energy resource, including the current output / load curve, response speed, climbing rate and cost fluctuation rate, and other time-varying attributes; the actual contribution correction factor is normalized to obtain the influence weight coefficient.

[0011] By querying the preset rule and combining the resource dynamic characteristic to correct the influence weight coefficient, the weight calculation is not only based on the static rule, but also can reflect the real-time change of the operating capacity and economy of the distributed resource, and the sensitivity of the weight to the current actual state of the resource is enhanced, thereby laying a foundation for more refined dynamic resource cluster division.

[0012] Optionally, in some embodiments, the dynamic resource cluster division instruction is generated according to the dynamic weight and the real-time feature data, including: a resource weighted feature vector is generated according to the dynamic weight and the real-time feature data; the resource weighted feature vector is processed by a constrained clustering algorithm to obtain a resource initial grouping scheme; the physical operability of the resource initial grouping scheme is verified to obtain a resource cluster entity that satisfies the cooperative constraint condition; and a dynamic resource cluster division instruction is generated based on a member list and a constraint boundary of each resource cluster entity.

[0013] By adopting the technical solutions in the above embodiments, the resource cluster is not only similar in features, but also physically feasible and satisfies the cooperative constraint by the constrained clustering algorithm based on the weighted feature vector and the verification of the physical operability, so that the resource cluster becomes an actually operable scheduling entity, and the practicability of the division scheme is improved.

[0014] Optionally, in some embodiments, the dividing and processing of the resource weighted feature vector by the band constrained clustering algorithm to obtain a resource initial grouping scheme comprises: constructing a resource similarity matrix based on the resource weighted feature vector; correcting the connection weight of the resource similarity matrix according to the power grid physical constraint condition to obtain a constraint corrected similarity matrix; performing spectral clustering feature decomposition on the constraint corrected similarity matrix to obtain a resource initial grouping scheme satisfying the constraint.

[0015] By considering and correcting the power grid physical constraint when constructing the resource similarity matrix, the technical scheme of the above embodiment directly integrates the power grid topology and capacity limit and other factors into the clustering process, ensures that the generated resource grouping scheme can better adapt to the power grid structure and operating conditions, and reduces the scheduling difficulty caused by network limitations.

[0016] Optionally, in some embodiments, the sending of the initial cross-regional collaborative scheduling instruction and the compensation price signal to the local agent corresponding to each dynamic resource cluster based on the dynamic resource cluster division instruction comprises: generating a power adjustment interval boundary of each dynamic resource cluster based on the topology structure in the dynamic resource cluster division instruction; constructing a distributed scheduling optimization model according to the power adjustment interval boundary; calculating an initial value of the optimal power instruction in the cluster based on the distributed scheduling optimization model to obtain an initial value matrix of the collaborative scheduling instruction; generating an initial cross-regional collaborative scheduling instruction and a compensation price signal according to the initial value matrix of the collaborative scheduling instruction, the marginal cost of power grid congestion and the adjustment cost; and sending the initial cross-regional collaborative scheduling instruction and the compensation price signal based on the mapping relationship between the dynamic resource cluster and the local agent.

[0017] By generating a power adjustment interval based on the topology of the dynamic resource cluster and constructing a distributed model, and by calculating an initial instruction and a compensation price in combination with the congestion and adjustment cost of the power grid, the technical scheme of the above embodiment provides a systematic and economically oriented way to generate a preliminary cross-regional collaborative scheduling instruction and an incentive signal, and starts a distributed collaborative optimization process.

[0018] Optionally, in some embodiments, the updating of the collaborative scheduling instruction and the compensation price signal according to the aggregated response boundary information to obtain a target collaborative scheduling instruction comprises: determining a power adjustment deviation matrix based on the initial cross-regional collaborative scheduling instruction and the aggregated response boundary information; updating the compensation price signal to obtain a compensation price update signal according to the power adjustment deviation matrix and real-time congestion data; and feeding back and correcting the initial cross-regional collaborative scheduling instruction based on the compensation price update signal to obtain a target collaborative scheduling instruction.

[0019] By means of the technical scheme in the above embodiment, the compensation price is iteratively updated and the instruction is corrected based on the deviation of the initial instruction and the aggregated response boundary and the real-time congestion data, a feedback loop is formed, the final scheduling instruction can continuously approach the actual response capability of the resource cluster and effectively cope with the real-time power grid congestion, and the adaptability and effectiveness of scheduling are significantly improved.

[0020] In a second aspect, an electronic device is provided, including: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is configured to store computer program codes, the computer program codes including computer instructions, and the one or more processors are configured to invoke the computer instructions to enable the electronic device to perform the method described in the first aspect or the second aspect, and any possible implementation manner of the first aspect or the second aspect.

[0021] In a third aspect, a computer-readable storage medium is provided, including instructions, when the instructions are executed on the electronic device, the electronic device is enabled to perform the method described in the first aspect or the second aspect, and any possible implementation manner of the first aspect or the second aspect.

[0022] In a fourth aspect, a computer program product is provided, including instructions, when the computer program product is executed on the electronic device, the electronic device is enabled to perform the method described in the first aspect or the second aspect, and any possible implementation manner of the first aspect or the second aspect.

[0023] It can be understood that the electronic device provided in the second aspect, the storage medium provided in the third aspect, and the computer program product provided in the fourth aspect are all used to perform the method provided by the present application. Therefore, the beneficial effects that can be achieved are referred to the beneficial effects in the corresponding method, which will not be repeated here. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 is a schematic diagram of a cross-regional virtual power plant collaborative scheduling system according to an embodiment of the present application;

[0025] Figure 2 is a flowchart of a cross-regional virtual power plant collaborative scheduling method according to an embodiment of the present application;

[0026] Figure 3 is a schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] The terminology used in the following description of the embodiments of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used in the description of the application, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It also will be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0028] The terms "first", "second", "third", "fourth", "fifth", "sixth", "seventh" and "eight" are used only to describe the purpose of the description and cannot be understood as indicating relative importance or implying the number of the technical features indicated. Therefore, the features defined with "first", "second", "third", "fourth", "fifth", "sixth", "seventh" and "eight" can include one or more of the features, and the meaning of "a plurality of" is two or more, unless otherwise specified.

[0029] It should also be noted that, unless otherwise explicitly specified and limited, the terms "set", "connected", and the like in the embodiments of the application should be understood in a broad sense. For example, "connected" can be fixed connection, or detachable connection, or integrally connected; can be mechanical connection, or electrical connection; can be direct connection, or indirect connection through an intermediate medium, can be the internal communication of two elements; can be wired communication connection, or wireless communication connection. For those skilled in the art, the specific meaning of the above terms in the application can be understood according to the specific circumstances. The embodiments of the application will be specifically described below.

[0030] In view of the problem in the related art that, in the cross-regional virtual power plant collaborative scheduling, due to the adoption of a fixed partition or a type-based resource grouping mechanism, the resource aggregation and dynamic characteristics are not matched, and the scheduling instruction and the actual resource capability are not matched well, the embodiments of the application provide a cross-regional virtual power plant collaborative scheduling technical solution. By introducing a dynamic weight determination mechanism based on real-time characteristic data and a global optimization target, and generating a dynamic resource cluster division instruction on this basis, the virtual power plant aggregation unit is flexibly and intelligently constructed according to the current power grid operation state, the scheduling target, and the real-time characteristics (including the space-time position, the output prediction, the operation parameter, the economic characteristics, and the schedulable state) of the distributed resources. This dynamic aggregation mode can ensure that the members in the same resource cluster have stronger target consistency and collaborative potential in the current scheduling period, thereby overcoming the contradiction that the resources are physically aggregated but difficult to efficiently cooperate in the static aggregation. Further, by sending an initial scheduling instruction and a compensation price signal to the dynamic resource cluster, and receiving the aggregated response boundary information fed back by the local agent, the information interaction and instruction iterative optimization between the upper-layer scheduling and the dynamic resource cluster are realized, so that the finally issued target collaborative scheduling instruction can more accurately match the actual aggregated response capability and the feasible region range of the dynamic resource cluster.

[0031] The method of this dynamic aggregation combined with feedback-based iterative optimization significantly improves the fine level and response efficiency of cross-regional virtual power plant resource scheduling, effectively suppresses the volatility of distributed energy, alleviates local grid congestion, improves the overall resilience and economy of the system, fundamentally solves the "gather but not integrate" problem caused by static aggregation, and realizes efficient collaborative use of resources.

[0032] Therefore, the embodiment of the application provides a cross-regional virtual power plant collaborative scheduling system, the architecture of which is as shown in Figure 1 .

[0033] The scheduling server at the top layer is the hub of the whole system, responsible for receiving the instructions or market signals of the power grid, and formulating detailed scheduling plans based on the global optimization goal; at the same time, the scheduling server also interacts with the power grid to report the overall operation state and schedulable capacity of the system.

[0034] Following it is the virtual power plant, which plays the role of an aggregator and logically integrates distributed distributed energy resources in different geographical locations or different types into a controllable whole. The virtual power plant receives the instructions of the scheduling server, and decomposes and optimizes them before issuing them to the next level of local agents.

[0035] The local agent is a regional-level management unit responsible for resource scheduling and control within its jurisdiction. Each local agent manages a number of dynamic resource clusters, which are flexible collections of resources formed according to actual needs and resource characteristics, such as photovoltaic arrays, wind turbines, energy storage systems, controllable loads, etc.

[0036] The dynamic resource cluster further refines the management instructions and finally acts on the distributed energy resources at the bottom layer.

[0037] The distributed energy resource is the physical unit that actually generates, stores or consumes energy in the system, which adjusts its operation according to the instructions of the local agent, such as changing the power generation, charging and discharging state or load start and stop.

[0038] The whole system realizes effective integration and unified scheduling of a large number of dispersed distributed energy resources through this hierarchical and collaborative way, improves energy utilization efficiency, enhances the stability and flexibility of the power grid, and can better participate in power market transactions. The information flow and control flow between each level ensure that the system can quickly respond to changes, realize optimal allocation and intelligent management of resources.

[0039] The following will be described in detail in conjunction with Figure 2 , a cross-regional virtual power plant collaborative scheduling method of the embodiment can be executed by a scheduling server, the method comprising the following steps:

[0040] Step 201, collect real-time characteristic data of each distributed energy resource and load.

[0041] The distributed energy resources include, but are not limited to, solar power generation equipment, wind power generation equipment, gas turbine and other distributed power generation units.

[0042] The real-time characteristic data at least includes one of spatial and temporal location information, output / demand prediction curve, physical operating parameter, economic characteristic parameter and dispatchable state.

[0043] The spatial and temporal location information refers to the specific location of the distributed energy resource or load in the geographical space and the attribute related to time.

[0044] The output / demand prediction curve refers to the predicted power generation power change trend of the distributed energy resource or the power demand change trend of the load in a future period of time.

[0045] The physical operating parameter refers to the physical quantity describing the current working state of the distributed energy equipment or load, such as voltage, current, frequency, temperature, state of charge (SOC) and the like.

[0046] The economic characteristic parameter refers to the cost or benefit information related to the participation of the distributed energy or load in the operation of the power system and the market transaction, such as generation cost, regulation cost, compensation price, market quotation and the like.

[0047] The dispatchable state refers to the state of whether and to what extent the distributed energy resource or load can accept and execute the dispatching instruction at the current time, such as whether the equipment is online, whether it is in a maintenance state, the adjustable power range and the like.

[0048] This step is the data collection process, which needs to establish a real-time data collection unit covering all the distributed energy resources (such as photovoltaic, wind power, energy storage, etc.) and controllable loads aggregated by the cross-regional virtual power plant. This real-time data collection unit can be connected to the local control unit or intelligent terminal of each distributed resource through various communication technologies (such as SCADA, Internet of Things communication, dedicated line network, etc.). The types of data collected include but are not limited to: GPS coordinates or other positioning information (spatiotemporal location information) of the device; output or demand curve (output / demand prediction curve) for a certain period in the future (for example, the next 24 hours, with a granularity of 15 minutes or 1 hour) generated by prediction algorithms based on historical data, weather forecasts, user behavior patterns, etc.; real-time voltage, current, active / reactive power, frequency, temperature, SOC of energy storage system, battery health status, etc. (physical operating parameters); economic-related data (economic characteristic parameters) such as generation cost, start-stop cost, maintenance cost, regulation cost, bidding for auxiliary services, market transaction price, etc. of the resource; and information reflecting whether the current device is available and the degree of availability (dispatchable state) such as the current operating mode (grid-connected / off-grid), maintenance plan, fault status, upper and lower limits of adjustable power, and climb rate limit. The collected data needs to be preprocessed, including data cleaning, format unification, missing value filling, etc., and then stored in a real-time database or transmitted to a dispatching server to provide accurate and timely input for subsequent dispatching decisions.

[0049] Step 202, based on the global optimization target set in the current dispatching period and the real-time feature data, determine the dynamic weight of each feature dimension.

[0050] Specifically, first, the global optimization target for the current dispatching period (for example, the next 15 minutes or 1 hour) needs to be determined. This target can be single, such as minimizing system operating cost, maximizing renewable energy consumption, or alleviating congestion on a specific transmission line; or it can be a combination of multiple objectives, such as considering economic efficiency, reliability, and environmental benefits.

[0051] Then, according to this global optimization target, identify the key feature dimensions most relevant to it. For example, if the goal is to alleviate congestion, the geographical location (spatiotemporal location information) and adjustable power range (dispatchable state) of the resource become very important; if the goal is to minimize operating cost, the economic characteristic parameters (such as generation cost, regulation cost) and output / demand prediction curve of the resource become more critical.

[0052] Next, based on these key feature dimensions, combined with a pre-set rule base or machine learning model, calculate the influence weight coefficient of each dimension on achieving the current optimization target. This calculation process may need to consider the mutual influence between different dimensions and the dynamic characteristics of the resource itself.

[0053] Finally, according to the resource state constraints (such as physical limits of devices, safe operating range, etc.) reflected in the real-time feature data, the preliminary calculated influence weight coefficients are corrected to ensure that the weights reflect the true importance of each dimension under the current actual constraints, thereby obtaining the dynamic weights of each feature dimension for subsequent dynamic resource cluster division. This process is dynamic, meaning that the weights of the feature dimensions will be adjusted accordingly at different scheduling periods or when the grid state changes.

[0054] Step 203, generating a dynamic resource cluster division instruction according to the dynamic weights and the real-time feature data, the dynamic resource cluster division instruction including a member list of the resource cluster and a coordination constraint condition.

[0055] Among them, the resource cluster refers to a logical set formed by clustering algorithm according to the real-time feature data of distributed energy resources and loads and the dynamic weights determined based on the global optimization target within the current scheduling period. The physical meaning of the resource cluster is that the resources within this set have similar characteristics (such as similar response ability, economic characteristics, geographical location, or contribution potential to the current scheduling target) at the current time, and physically satisfy certain coordination constraints (such as grid topology, capacity limit, etc.), so they can participate in scheduling and power market transactions as a whole, achieving stronger regulation ability and better overall performance than individual resources. The resource cluster is not a fixed physical entity, but a virtual aggregation unit that changes dynamically according to real-time conditions, aiming to overcome the mismatch between traditional static aggregation methods and dynamic characteristics of resources, and achieve "aggregation and integration".

[0056] The implementation process of this step first weights the real-time feature data (including spatiotemporal location, prediction curve, physical parameters, economic characteristics, and schedulable state) of each distributed resource according to the dynamic weights to form a weighted feature vector, which comprehensively reflects the comprehensive characteristics of the resource under the current scheduling target.

[0057] Then, a clustering algorithm (such as K-Means, hierarchical clustering, DBSCAN, etc.) is used to cluster the weighted feature vectors. In the clustering process, some basic coordination constraints need to be considered, such as setting maximum cluster member number, minimum cluster capacity, etc. or introducing a geographical location penalty term in distance calculation to make resources with similar geographical locations more likely to be divided into the same cluster.

[0058] The clustering algorithm divides the resources into several clusters. For each formed cluster, a resource cluster division instruction is generated, which contains the member list of the cluster (i.e., which distributed resources belong to this cluster) and some basic coordination constraint conditions, such as the upper and lower limits of the total power that the cluster as a whole needs to meet within the current scheduling period.

[0059] Step 204, based on the dynamic resource cluster division instruction, send the initial cross-regional cooperative scheduling instruction and the compensation price signal to the corresponding local agent of each dynamic resource cluster.

[0060] Wherein, the initial cross-regional cooperative scheduling instruction includes the net exchange power target value and the auxiliary service demand. The net exchange power target value refers to the target value of the dynamic resource cluster to exchange power with the external power grid (or other regions / resource clusters) in the current scheduling period (i.e. the total power generation minus the total power consumption or the power output / input target). The auxiliary service demand refers to the specific auxiliary service type and quantity required by the dynamic resource cluster to provide to the power grid in the current scheduling period, such as frequency modulation, voltage regulation or backup capacity, etc.

[0061] This step is to convert the dynamic resource cluster division instruction into the initial scheduling instruction and the compensation price signal, and send it to the local agent. First, according to the resource cluster division instruction generated in step 203, determine which resources each dynamic resource cluster contains. Then, based on the global optimization target set in the current scheduling period (for example, system total cost minimization, renewable energy consumption maximization, etc.), a preliminary scheduling optimization model is constructed. This model takes each dynamic resource cluster as the optimization object, considers the preliminary aggregation capacity of each cluster (for example, the simple superposition of the predicted output / demand curve based on the cluster members) and some basic grid constraints. By solving this optimization model, the initial net exchange power target value of each dynamic resource cluster (i.e. the power that the cluster needs to inject or absorb from the grid) and the preliminary auxiliary service demand (the dynamic power support service provided by the energy storage cluster through rapid charging and discharging, such as frequency modulation, voltage stabilization, black start, etc., whose core value lies in compensating the real-time power imbalance of the grid through cross-regional cooperative scheduling) can be obtained.

[0062] At the same time, according to the preliminary scheduling result and the real-time operation state of the grid, the preliminary compensation price signal can be calculated to guide the resource cluster to adjust. Finally, these initial net exchange power target values, auxiliary service demands and compensation price signals are packaged and sent to the local agent corresponding to each dynamic resource cluster through the communication network.

[0063] Step 205, receiving the aggregated response boundary information returned by each local agent based on the initial cross-regional cooperative scheduling instruction and the compensation price signal, the aggregated response boundary information being used to represent the feasible domain range of the dynamic resource cluster to the instruction.

[0064] Wherein, the aggregated response boundary information is the multi-dimensional feasibility commitment of each dynamic resource cluster to the scheduling instruction, and its core is to dynamically describe the physical limit and economic response boundary of the cluster under the cooperative scheduling constraint through a mathematical model, which specifically includes three dimensions:

[0065] Power feasible region (the maximum charge-discharge power range that the cluster can provide under the current SOC and device state, such as -50MW~+80MW, with a ramp rate limit such as ±20MW / min);

[0066] Energy capability boundary (a sustainable output duration curve derived based on the SOC safety window, such as 25%-85%, for example, if discharging at 80MW, the SOC can only be sustained for 0.9 hours from 85% to 25%);

[0067] Service value mapping (a cost-response sensitivity matrix generated according to the compensation price signal, for example, when the frequency regulation compensation is greater than 15 yuan / MWh, the response capability is increased by 40%), used to quantify the range of elastic expansion of the feasible region under economic incentives.

[0068] The local agent is a software entity deployed in the regional layer or the virtual power plant management platform, responsible for receiving instructions and price signals sent by the dispatching server, managing specific distributed resources within the dynamic resource cluster under its jurisdiction, and feeding back the aggregated capability information of the resource cluster to the upper layer. It can be regarded as the representative or manager of the resource cluster locally, responsible for coordinating the operation of the resources within the cluster to respond to the instructions from the upper layer.

[0069] This step is for the local agent to calculate and feed back the aggregated response boundary information of the resource cluster to the upper layer according to the received initial cross-regional coordinated dispatching instructions and compensation price signals, combined with the real-time state and characteristics of the resources within the dynamic resource cluster managed by it. After receiving the instructions (including the net exchange power target value and auxiliary service demand) and the compensation price signal, the local agent will immediately start to evaluate the overall capability of the resource cluster under its jurisdiction.

[0070] First, the dispatch server calculates the power feasible region through the local agent. The local agent iterates through all the member resources within the resource cluster, obtaining their real-time running status (e.g., whether the generator is online, the current state of charge SOC of the energy storage device, the current consumption power of the interruptible load), physical constraints (such as the maximum / minimum output or consumption power of each resource, the ramp rate limit), and device health status. Then, the local agent aggregates the capabilities and constraints of individual resources, while considering the coordination constraints within the resource cluster (e.g., certain resources need to coordinate actions to achieve specific cluster-wide functions). Through an internal aggregation algorithm or model, the maximum and minimum charge-discharge power that the dynamic resource cluster as a whole can provide at the current moment (i.e., the maximum power injected or absorbed from the grid) is calculated, forming the power feasible region interval (e.g., -50MW ~ +80MW). At the same time, the ramp rate limit of the cluster as a whole (e.g., ±20MW / min) is also calculated, which is the maximum rate of power change. These calculations reflect the instantaneous power regulation capability limit of the resource cluster under the current physical state.

[0071] Second, the energy capability boundary is calculated through the local agent. This is mainly for the case where the resource cluster contains energy storage resources. The local agent obtains the real-time SOC of the energy storage resources, and combines it with the pre-set SOC safety window (e.g., 25%-85%). Based on the current SOC level and the power feasible region, the local agent can deduce how long the resource cluster can continuously discharge or charge at different power levels. For example, if the resource cluster currently has sufficient total energy storage capacity and is at a high SOC level, and the power feasible region allows discharging at 80MW, the local agent will calculate how long it can continuously discharge at 80MW while keeping the SOC within the safety window (e.g., 0.9 hours). By calculating the sustainable time length corresponding to different discharge / charge power levels, a sustainable output time curve is constructed, which is the energy capability boundary. It reflects the energy throughput capability of the resource cluster within a period of time.

[0072] Finally, the local agent calculates the service value mapping. The local agent analyzes the potential impact of the received compensation price signal on the willingness and ability of each member resource in the resource cluster to respond. Resources can have different cost functions or response strategies to price signals. The local agent evaluates how the number and ability of resources in the resource cluster willing to participate in scheduling or provide auxiliary services will change at different compensation price levels. For example, if the frequency regulation compensation price is low, only part of the low-cost resources may be willing to provide frequency regulation services; if the price increases to more than 15 yuan / MWh, more high-cost or incentivized resources may join, thereby increasing the frequency regulation response ability of the resource cluster by 40%. The local agent quantifies the relationship between this price signal and the elastic expansion range of the feasible region of the resource cluster, forming a cost-response sensitivity matrix or similar mapping relationship. This reflects the impact of economic incentives on the scheduling flexibility of the resource cluster.

[0073] The local agent integrates the calculated power feasible region (including power interval and ramp rate), energy capacity boundary (sustainable output duration curve), and service value mapping (cost-response sensitivity matrix) into aggregated response boundary information, and sends it back to the scheduling server through the communication network. These information are multi-dimensional commitments of dynamic resource clusters to the feasibility of their current scheduling instructions, providing key real-time feedback for subsequent collaborative scheduling optimization and instruction correction by the scheduling server.

[0074] Step 206, updating the collaborative scheduling instruction and the compensation price signal according to the aggregated response boundary information to obtain a target collaborative scheduling instruction.

[0075] Specifically, after the scheduling server receives the aggregated response boundary information feedback from all dynamic resource clusters, these information contain the multi-dimensional constraints and response characteristics of each resource cluster under the current actual state, which are more accurate and closer to reality than the predicted capacity used in the initial scheduling model. These information include:

[0076] Power feasible region: the actual maximum / minimum charging / discharging power range that each resource cluster can provide at the current time, as well as the ramp rate limit.

[0077] Energy capacity boundary: mainly for clusters with energy storage, describes the duration of sustainable output or charging at different power levels, reflecting the energy throughput capacity of the cluster over a period of time.

[0078] Service value mapping: describes the relationship between economic compensation price and the response ability of the resource cluster (such as adjustable power range, willingness to participate in auxiliary services).

[0079] The dispatch server utilizes these feedback information to build an updated co-scheduling optimization model. This new model is similar to the initial model, but its constraints are no longer based on the predicted ideal capabilities, but directly use the actual aggregated response boundary information feedback from each resource cluster.

[0080] In particular:

[0081] Update of power constraints: The power output or absorption constraints for each resource cluster in the optimization model will directly adopt the power feasible region interval from its feedback. For example, if a cluster feedbacks its power feasible region as [-50MW, +80MW], then in the optimization model, the power variable for this cluster will be limited within this range. Meanwhile, the feedbacked ramp rate limit will also be considered.

[0082] Update of energy constraints: For resource clusters with energy storage, the optimization model will incorporate its feedbacked energy capability boundary. This can be achieved by introducing additional energy balance equations or constraints, ensuring that the energy dispatch for a resource cluster within a dispatch period meets the actual energy limit reflected by its sustainable output duration curve. For example, if a sustained high power discharge is required in a certain period, the model will check whether the energy capability boundary of the cluster supports such sustained discharge.

[0083] Consideration of economic incentives: The optimization model will utilize the feedbacked service value mapping. This can be embodied by introducing terms related to the compensation price in the objective function, or by taking the service value mapping as an additional constraint or parameter. For example, the model can predict and utilize the additional response capability that a resource cluster can provide according to the current set compensation price, so as to effectively utilize economic levers to guide resource behavior while meeting the demand of the power grid.

[0084] The dispatch server recalculates the optimal dispatch target for each dynamic resource cluster within the current dispatch period, i.e. the target co-scheduling instruction, by solving this updated optimization model (usually a large-scale mathematical programming problem), taking into account the actual aggregated capability constraints of all resource clusters and the overall operation requirements of the power grid (such as load balancing, power flow constraints, voltage stability, etc.) and global optimization objectives (such as minimum total operation cost, maximum renewable energy consumption). These target instructions are based on the real capabilities of resource clusters, so they are more feasible and robust.

[0085] Meanwhile, the optimization model may also output new compensation price signals during the solving process. These price signals reflect the current system's demand and value for different types of regulation capabilities. For example, if the system has a high demand for frequency regulation services in a certain period, the optimization model may calculate a higher frequency regulation compensation price to encourage resource clusters with frequency regulation capabilities to provide more services. These updated price signals are also included in the target coordinated scheduling instructions and sent to the local agents for further guidance of optimization and response within the resource clusters.

[0086] The scheduling server uses the real capability information fed back by the resource clusters to correct the optimization model, re-solves to obtain more realistic and optimized scheduling instructions and price signals, thereby improving the effectiveness and reliability of the coordinated scheduling of the cross-regional virtual power plant.

[0087] Step 207, sending the target coordinated scheduling instructions to each dynamic resource cluster for execution to control the operation of the cross-regional virtual power plant resources.

[0088] Specifically, the scheduling server sends the determined target net exchange power target value, auxiliary service demand, and final compensation price signal of each dynamic resource cluster to the corresponding local agent through the communication network. After receiving the target coordinated scheduling instructions, the local agent decomposes them into detailed control instructions for each specific distributed resource within its jurisdiction. For example, if the target instructions require a certain resource cluster to inject 50 MW of power into the grid in a certain period, the local agent will determine which generators to increase power output, which energy storage devices to discharge, or which interruptible loads to reduce consumption based on the real-time state, capability, and priority of each resource within the cluster, as well as the specific power output / consumption values.

[0089] The local agent sends these detailed control instructions to each distributed resource controller or intelligent terminal within the cluster, which directly controls the operation of the devices (such as adjusting generator power output, controlling energy storage charging and discharging, controlling load start and stop, etc.).

[0090] In this way, the global optimization instructions of the scheduling server are coordinated and executed by the local agent, ultimately achieving effective control and coordinated operation of the cross-regional virtual power plant resources.

[0091] This embodiment improves the matching degree of scheduling instructions and the actual capacity of large-scale distributed resources by constructing a dynamic, feedback-based cross-regional virtual power plant coordinated scheduling framework. At the same time, the dynamic resource cluster division mechanism is introduced, effectively solving the problem of "clustering without integration" caused by static aggregation, and improving the efficiency and resource utilization of the coordinated scheduling of the cross-regional virtual power plant.

[0092] Specifically, the embodiment first collects real-time characteristic data of each distributed energy resource and load, which comprehensively reflects the space-time characteristics, operating state, economy and dispatchable capacity of the resource. Unlike fixed partitioning or type grouping, the embodiment dynamically determines the weight of each characteristic dimension based on the current global optimization target and real-time characteristic data, which means that the importance of different types and locations of resources can be flexibly adjusted according to real-time operating requirements and resource conditions.

[0093] Subsequently, dynamic resource cluster partitioning instructions are generated according to the dynamic weights and real-time data, which explicitly indicate which members are included in each resource cluster and the coordination constraints therebetween. This dynamic partitioning method enables the composition of the resource cluster to closely match the current scheduling target and resource capacity, avoiding the inadaptability problem caused by static grouping.

[0094] Then, initial cross-regional coordinated scheduling instructions and compensation price signals containing net exchange power target values and ancillary service requirements are sent to local agents corresponding to the dynamic resource clusters, and preliminary scheduling is started. The local agents return aggregated response boundary information representing the feasible domain range of the resource cluster to the instructions according to the actual response capacity of the resource cluster managed thereby. The scheduling server iteratively updates the coordinated scheduling instructions and compensation price signals according to the feedback information, and obtains target coordinated scheduling instructions that are more in line with the actual situation.

[0095] Finally, the target coordinated scheduling instructions are sent to each dynamic resource cluster for execution, realizing accurate control and coordinated operation of the cross-regional virtual power plant resources.

[0096] The embodiment can fully utilize the flexibility and complementarity of distributed resources through the dynamic and real-time resource cluster partitioning and feedback-based iterative optimization process, overcome the limitations of static aggregation, and make the scheduling instructions highly adaptable to the actual resource capacity, thereby realizing efficient coordinated scheduling of cross-regional distributed energy resources on a larger time and space scale, effectively smoothing fluctuations, relieving grid congestion, improving the overall resilience and economy of the system, and alleviating the problem of insufficient adaptability of scheduling instructions to actual resource capacity.

[0097] The embodiment also provides a more specific cross-regional virtual power plant coordinated scheduling method, including the following steps:

[0098] S301, collecting real-time characteristic data of each distributed energy resource and load.

[0099] The description of this step can be referred to the foregoing embodiments, which will not be repeated here.

[0100] S302, determining a key characteristic dimension identifier according to a global optimization target of a current scheduling period.

[0101] Specifically, according to the global optimization target of the current scheduling period, such as the target of achieving economic optimal scheduling of the cross-regional power system, key feature dimensions can be selected from a plurality of factors related to the target. For example, the spatio-temporal distribution characteristics of distributed energy (power generation output conditions in different regions and different time periods), load characteristics (power consumption rules of various loads, peak and valley periods, etc.), and power grid transmission constraints (capacity limitations of different transmission lines and characteristics of power flow distribution), etc. By analyzing the potential impact of these factors on the achievement of the global optimization target, it is determined which dimensions are key and are identified as the basis for subsequent calculations to ensure that subsequent work around the factors that have the greatest impact on achieving the target is carried out, such as weight calculation, so that the entire scheduling decision is more focused and more in line with the target requirements.

[0102] S303, based on the key feature dimension identification, calculating the influence weight coefficient of each dimension on the target.

[0103] Specifically, after determining the key feature dimension identification, a suitable mathematical method is used to calculate the influence weight coefficient of each dimension on the target, such as regression analysis based on historical data, expert experience valuation combined with analytic hierarchy process, etc.

[0104] Taking regression analysis as an example, a large number of corresponding samples of different key feature dimension data and global optimization target achievement conditions (such as economic cost, power balance degree, etc.) in past scheduling periods are collected, a regression model is established, and the weight proportion of each key feature dimension in the process of influencing the target achievement is calculated through model fitting, that is, the influence weight coefficient, so as to quantify the importance of each dimension to achieve the global optimization target and provide an initial basis for subsequent dynamic weight adjustment.

[0105] In some embodiments, the present step can specifically include:

[0106] S3031, querying a preset target feature mapping rule library to obtain a baseline influence value of each key feature dimension on the optimization target.

[0107] In the present embodiment, the preset target feature mapping rule library is a pre-constructed database or knowledge base, which stores the mapping relationship between different key feature dimensions and optimization targets. These rules are usually set in advance based on historical operation data, expert experience or simulation analysis, for example, for the "economic scheduling target", the rule library may explicitly specify that the baseline influence value of "distributed power output prediction accuracy" is 0.3 and the baseline influence value of "load peak valley difference" is 0.25, etc.

[0108] In implementation, the dispatch server retrieves the benchmark impact value corresponding to the key feature dimension (such as the spatio-temporal position, the output prediction curve) from the rule base according to the global optimization target (such as the economy, the reliability or the low-carbon target) of the current dispatch cycle, as the initial quantified basis for the contribution of each dimension to the target.

[0109] The construction process of the target feature mapping rule base includes: first, collecting a large amount of associated data of key feature dimensions (such as the spatio-temporal position of distributed energy, the output prediction curve, etc.) and optimization targets (such as economy, reliability) under different dispatch scenarios; second, organizing domain experts to evaluate and score the impact of each dimension on different targets, while using simulation models to simulate the actual contribution of feature dimensions to the target under various working conditions; and finally, systematically combing and standardizing these data, expert evaluation and simulation results to form the mapping relationship between different optimization targets and key feature dimensions, and storing them in the rule base to provide a basis for subsequent query of benchmark impact values.

[0110] S3032, based on the benchmark impact value and the resource dynamic characteristics in the real-time feature data, calculating a correction factor of the actual contribution of each dimension to the target achievement.

[0111] The resource dynamic characteristics refer to the real-time change of the operating ability parameters of the distributed energy resource, including the current output / load curve, response speed, ramp rate and cost fluctuation rate, etc. time-varying attributes.

[0112] The resource dynamic characteristics change with time and directly affect the actual ability of the resource to contribute to the global optimization target at the current time. For example, whether the current output curve of photovoltaic deviates from the predicted value due to cloud cover, whether the response speed of energy storage meets the dispatch instruction, whether the wind turbine ramp rate meets the grid regulation requirements, whether the price fluctuation leads to cost changes, etc.

[0113] In implementation, the benchmark impact value of each key feature dimension is coupled with the corresponding real-time dynamic characteristic data for analysis: for example, if the current output of a certain wind farm is 20% lower than the predicted value due to sudden wind speed drop, the benchmark impact value of its "output prediction curve" dimension needs to be multiplied by a correction factor (such as 0.8); if the response speed of a certain energy storage device is 10% slower than the rated value, the correction factor of its "response speed" dimension may be 0.9.

[0114] Through such dynamic matching, the correction factor of the actual contribution of each dimension to the target achievement is obtained to reflect the adjustment of the real-time working condition to the benchmark impact

[0115] S3033, normalizing the actual contribution correction factor to obtain the impact weight coefficient.

[0116] Since the actual contribution correction factors of each dimension may be in different dimensions or numerical ranges (e.g., the output curve correction factor is 0.7-1.2, and the response speed correction factor is 0.8-1.1), normalization processing is needed to convert them into comparable weight coefficients.

[0117] In implementation, linear normalization methods (such as min-max normalization) or standardization methods (such as Z-score standardization) are usually used to map all correction factors to the [0, 1] interval, and ensure that the sum of the weight coefficients of each dimension is 1. For example, if the correction factors of the three key dimensions in a scheduling period are 0.9, 1.1, and 0.8, respectively, the weight coefficients become 0.3, 0.367, and 0.267 after normalization, thereby forming a dynamic weight system reflecting the real-time resource characteristics, providing a quantitative basis for subsequent resource cluster division and scheduling instruction generation.

[0118] S304, according to the resource state constraint in the real-time feature data, the influence weight coefficient is corrected to obtain the dynamic weight.

[0119] The resource state constraint refers to the physical boundary condition that limits the resource scheduling feasible region, which is an insurmountable constraint range determined by the physical characteristics of the device itself and the structure of the power grid, and can specifically include maximum / minimum power limit, energy storage SOC safety interval, device start-stop state, and network topology connection relationship formed by the connection of elements in the power grid, etc.

[0120] After obtaining the resource state constraint information in the real-time feature data, the previously calculated influence weight coefficients are corrected in combination with these rigid constraints. For example, when a distributed power source is in a shutdown state due to the device start-stop state constraint, the influence weight coefficient calculated by the space-time distribution characteristic dimension of the power source needs to be adjusted downward according to the actual situation that the power source cannot participate in scheduling; if the energy storage SOC is at the edge of the safety interval, the weight coefficient of the related feature dimension involving energy storage also needs to be adjusted to ensure the safety of the device.

[0121] In this way, the physical boundary conditions of the resource scheduling feasible region (maximum / minimum power limit, energy storage SOC safety interval, etc.) are integrated, and the weight coefficient can dynamically reflect the influence of each dimension on the target under the current actual resource constraint, so that the subsequent scheduling decision based on the dynamic weight is more consistent with the actual operation scenario, and the scientificity and feasibility of the scheduling are enhanced

[0122] S305, according to the dynamic weight and the real-time feature data, a resource weighted feature vector is generated.

[0123] Specifically, the scheduling server first extracts the multi-dimensional features (such as spatial and temporal position coordinates, output prediction curve values, response speed parameters, etc.) of each distributed energy resource from real-time feature data, then multiplies each feature dimension with the corresponding dynamic weight to obtain the weighted feature value. For example, the weight of the "spatial and temporal position" dimension of a certain photovoltaic power station is 0.25, and the standardized geographical position coordinates are (0.6, 0.3), so the weighted coordinates are (0.15, 0.075); the weight of the "output prediction curve" dimension is 0.3, and the current prediction error rate is 8%, so the weighted value is 0.024.

[0124] Then, the weighted feature values of all dimensions are combined in order to form the resource weighted feature vector of the resource point, so that each component in the vector reflects the influence weight of the corresponding feature on the global optimization target.

[0125] S306, the resource weighted feature vector is divided and processed by a clustering algorithm with constraints to obtain a resource initial grouping scheme.

[0126] Specifically, the scheduling server can use a clustering algorithm that incorporates physical constraints (such as spectral clustering with power upper limit constraints or constrained K-means) to group the resource weighted feature vectors. When calculating the similarity between resources (such as Euclidean distance or cosine similarity), the algorithm simultaneously embeds conditions such as power grid topology constraints (such as transmission line capacity limitations), device operation constraints (such as storage SOC safety intervals, wind turbine climb rate upper limits), etc.

[0127] For example, during the clustering process, if two resource points are similar in features but belong to different electrical islands (subject to network topology constraints), the algorithm forces them into different groups; if the total regulation power of resources in a group exceeds the line transmission capacity, the grouping is adjusted again.

[0128] Through iterative optimization of clustering centers and grouping boundaries, a resource initial grouping scheme that meets the constraint conditions is finally generated. The resource initial grouping scheme is a preliminary resource grouping result that considers the similarity of multi-dimensional resource features while embedding physical conditions such as power grid topology constraints, device operating parameter limitations, etc. It groups distributed energy, energy storage, and loads into several groups based on functional complementarity and constraint compatibility. Although it is a preliminary grouping, it has initially met some basic constraints and needs to be further optimized through physical operability verification to form a resource cluster entity that can be actually executed.

[0129] In some embodiments, the present step can specifically include:

[0130] S3061, constructing a resource similarity matrix based on the resource weighted feature vector.

[0131] The purpose of this step is to quantify the similarity of any two distributed energy resources or loads in the weighted feature space. In implementation, for each pair of resources, the distance or similarity measure between their respective weighted feature vectors is calculated, for example using Euclidean distance, cosine similarity or other suitable measure method for the feature type. The smaller the Euclidean distance or the greater the cosine similarity, the more similar the two resources are in the weighted features. The similarity values between all resource pairs are constructed into a symmetric matrix, i.e. a resource similarity matrix, where each element of the resource similarity matrix represents the similarity of the corresponding two resources.

[0132] Specifically, the scheduling server compares the weighted feature vectors of all distributed energy resources pairwise, calculates the similarity between the resources using the Euclidean distance or cosine similarity algorithm, and forms an N x N matrix (N is the total number of resources). Each element S(i,j) in the matrix represents the feature similarity of resource i and resource j, and the larger the value, the higher the matching degree of the two in the time and space position, output characteristics and other dimensions. For example, the weighted feature vector of a certain photovoltaic power station and energy storage device has a similarity of 0.8 in the "spatial complementarity" dimension and 0.6 in the "response speed" dimension. The comprehensive similarity after normalization is 0.75, which is filled into the corresponding position of the matrix to construct a complete resource similarity matrix.

[0133] S3062, according to the grid physical constraint condition, the connection weight of the resource similarity matrix is corrected to obtain a constraint corrected similarity matrix.

[0134] The grid physical constraint condition refers to the rigid restriction determined by the grid topology structure, device operation characteristics, etc., including transmission line transmission capacity, electrical island distribution, device power limit, etc., which directly affects the feasibility of resource clustering and scheduling.

[0135] The connection weight is a value in the resource similarity matrix that represents the matching degree of different resources, to reflect the possibility and closeness of actual cooperative scheduling of resources.

[0136] The correction method is usually achieved by adjusting the value of the corresponding element in the similarity matrix: for resource pairs that must be divided into the same group, the similarity between them can be significantly increased (or the distance can be reduced), so that they are more inclined to be divided together in clustering; for resource pairs that must be divided into different groups, the similarity between them can be significantly reduced (or the distance can be increased), so that they are more inclined to be divided into different groups. The corrected matrix is called a constraint corrected similarity matrix.

[0137] Specifically, the dispatch server can call the power grid topology model and real-time operation data to constrain and correct the initial similarity matrix: if there is congestion (transmission capacity is lower than a preset threshold) or the resources i and j belong to different electrical islands (no physical connection) between the transmission line of the resources i and j, the S(i, j) in the matrix is multiplied by a penalty coefficient (such as 0.3) or directly set to 0; if the maximum output limit of the resource i and the load demand of the resource j may cause local power grid overload after clustering, the connection weight of the two is reduced. For example, although a certain wind farm and industrial load are similar in characteristics, the current transmission capacity between them is only 20% of the rated value, and the similarity weight of the two is corrected from 0.6 to 0.6*0.2=0.12, forming a corrected similarity matrix reflecting the physical constraints of the power grid.

[0138] S3063, performing spectral clustering feature decomposition on the constraint-corrected similarity matrix to obtain an initial resource grouping scheme that satisfies the constraints.

[0139] Among them, spectral clustering is a clustering method based on graph theory, which regards data points as vertices of a graph and the similarity between points as the weight of edges.

[0140] In this step, the constraint-corrected similarity matrix is regarded as the adjacency matrix of a similarity graph. The dispatch server converts the constraint-corrected similarity matrix into a Laplacian matrix (L=D-S, where D is a diagonal matrix and the diagonal elements are the sum of each row similarity), performs eigenvalue decomposition on the Laplacian matrix, extracts the eigenvectors corresponding to the first k smallest eigenvalues (k is a preset cluster number), and forms a new matrix with these eigenvectors, each row representing a k-dimensional feature representation of a resource. Then, the matrix is clustered by K-means algorithm or the like to obtain an initial resource grouping scheme. For example, after extracting the first three eigenvectors, each resource is mapped to a three-dimensional space and clustered into three groups, ensuring that resources in the same group are highly similar in characteristics under the premise of satisfying the power grid constraints, and finally generating an initial resource grouping that meets the topology constraints and device operation restrictions.

[0141] This embodiment can effectively consider these restrictions in the clustering process by encoding the constraint information in the similarity matrix.

[0142] S307, verifying the physical operability of the initial resource grouping scheme to obtain a resource cluster entity that satisfies the cooperative constraint condition.

[0143] Specifically, the dispatch server evaluates whether the aggregated characteristics of all resources in the group meet the physical constraints of grid operation, such as total power margin, voltage stability, power flow limit, etc. Secondly, it checks whether the communication, control and physical connection between resources in the group support collaborative operation. Finally, it verifies whether the operation constraints of each resource itself (such as energy storage SOC range, device minimum operation time, etc.) can be met under the collaborative control of the group. Only after passing all these physical level verifications, can the group be confirmed as a collaborative resource cluster entity to participate in subsequent scheduling.

[0144] The multi-dimensional physical constraint verification for each resource cluster in the initial grouping scheme can include: ① power balance verification, calculating the real-time difference between the output of power supply and the demand of load in the cluster, to ensure that it does not exceed the power limit of the cluster and the interconnected line; ② device state verification, checking whether the SOC of energy storage in the cluster is within the safe range (such as 20%-80%), and whether the generator set is in the allowed start-stop state; ③ topology connectivity verification, confirming whether the resources in the cluster are connected through physical lines by the grid topology model, to avoid forming an island that cannot be scheduled; ④ response capability verification, evaluating whether the comprehensive response speed of resources in the cluster meets the requirements of scheduling instructions (such as the response of seconds level required by frequency modulation).

[0145] If any of the verification items fails, adjust the grouping parameters to re-cluster until all resource clusters meet the collaborative constraint conditions.

[0146] S308, based on the member list and constraint boundary of each resource cluster entity, generate dynamic resource cluster division instructions.

[0147] Specifically, first, arrange the member list by cluster, and clearly identify the unique identification (such as device ID, geographic location code) and parameters (such as rated power, response speed) of each cluster containing distributed power supply, energy storage, controllable load and other devices; secondly, determine the cluster-level constraint boundary, including the upper and lower limits of net exchange power, the maximum value of ramp rate, the SOC range of energy storage, etc. operation parameters; finally, add collaborative scheduling rules, such as complementary scheduling strategy of power supply and energy storage in the cluster (energy storage discharges when photovoltaic output is at a low point), load adjustment priority (important load is given priority to guarantee).

[0148] The dynamic resource cluster division instructions are packaged in a standardized data format (such as JSON or XML), including version number, effective time, etc. meta information, and are issued to the corresponding local agent for execution through the dispatch communication network.

[0149] For each resource cluster entity, the dynamic resource cluster partitioning instruction needs to explicitly specify the unique identifiers of the specific resource units it contains (member list) so that these resources can be uniformly managed and controlled during scheduling execution. At the same time, the instruction also needs to give the equivalent operating constraint boundary that the resource cluster presents to the outside as an aggregate entity, which may include the aggregate total power / load range, the aggregate response speed, the aggregate ramp rate, the aggregate cost function, etc. These information constitute the core content of the dynamic resource cluster partitioning instruction, guiding the upper-level scheduling decision maker to optimize the resource cluster as a whole.

[0150] S309, generating power regulation interval boundaries of each dynamic resource cluster based on the topology structure in the dynamic resource cluster partitioning instruction.

[0151] Wherein, the topology structure refers to the network layout composed of nodes (such as power sources, loads, energy storage device nodes) and their physical connection relationships (such as transmission lines, transformers, etc.) in the power grid, including electrical connection modes between nodes, line transmission capacity, electrical island distribution, etc. physical attribute information, used to clarify the spatial distribution of resource clusters and power transmission path constraints.

[0152] The dispatching server parses the power grid topology data (such as node connection relationship, line parameter) in the dynamic resource cluster partitioning instruction, combines the physical operating characteristics and constraints of all resources within the resource cluster, such as the maximum and minimum power of the generator, the adjustable range of the load, the capacity and charge / discharge power limit of the energy storage, and the ramp rate of each resource itself. By superimposing and comprehensively considering the constraints of these individual resources, the total power regulation range that the resource cluster as a whole can provide at the current time or in the future period of time is obtained, i.e. the aggregated power regulation interval boundary. This boundary represents the flexibility and capability range of the resource cluster as a dispatching unit.

[0153] For example, a resource cluster contains 2 wind turbines (each rated power 1.5MW), 1 energy storage (charge / discharge power ±2MW), and the maximum transmission capacity of the transmission line in the cluster is 5MW, then the total power regulation interval of the cluster is (-2MW, 5MW), where the negative value represents energy storage discharge or load reduction, and the positive value represents power output or energy storage charging, and the boundary value is constrained by both line capacity and device limit.

[0154] S310, constructing a distributed scheduling optimization model according to the power regulation interval boundary.

[0155] Wherein, the distributed scheduling optimization model can contain a global objective function (such as minimizing the total operating cost of the system, maximizing social welfare, etc.) and system-level operating constraints (such as total network power balance, transmission line power flow limit, etc.).

[0156] In this embodiment, the key of the distributed scheduling optimization model is to take the aggregated power regulation interval boundary of the resource cluster as a constraint at the cluster level. By using a distributed optimization framework such as decomposition coordination method (such as Lagrange relaxation, ADMM, etc.), the complex global scheduling problem is decomposed into multiple relatively independent intra-cluster sub-problems and a coordinator problem, so that each resource cluster can approach the global optimal solution step by step through interaction with the coordinator under the premise of meeting its own boundary constraints.

[0157] When using a distributed optimization framework (such as ADMM, Lagrange relaxation, etc.), the global scheduling problem is decomposed into multiple intra-cluster sub-problems and a coordinator problem; each intra-cluster sub-problem takes the aggregated power regulation interval boundary of the resource cluster as the core constraint, and defines the feasible power regulation range of the cluster under the current coordination signal; the coordinator problem is responsible for handling global constraints such as global balance constraints and network flow constraints, and through iterative interaction, coordinates the local decisions of each cluster, so that it converges to a global optimal scheduling scheme that meets the global constraints while meeting the power regulation interval boundary of each cluster.

[0158] S311, based on the distributed scheduling optimization model, calculate the intra-cluster optimal power instruction initial value to obtain the collaborative scheduling instruction initial value matrix.

[0159] The optimal power instruction initial value is the initial power regulation optimal value of each dynamic resource cluster obtained by solving the constraints such as power regulation interval boundary and tie-line transmission capacity, which provides basic execution parameters for cross-regional collaborative scheduling.

[0160] After the distributed scheduling optimization model is constructed, the scheduling server starts the solving process of the model. In the initial iteration stage, the coordination signal (such as shadow price or target power) is generated according to the preliminary information and sent to the local agent corresponding to each resource cluster. After receiving the coordination signal, each local agent solves the sub-problem of the resource cluster according to the aggregated power regulation interval boundary of the resource cluster and the specific characteristics of the member resources in the resource cluster, and calculates the optimal power response or the optimal output / load instruction of the internal resources under the current coordination signal. The initial optimal power instructions (aggregated values or decomposed values of internal resources) of each resource cluster obtained by calculation are collected to form a collaborative scheduling instruction initial value matrix, which serves as the basis for subsequent iterations or actual instructions.

[0161] Specifically, the scheduling server calls an optimization solver (such as Gurobi, CPLEX) to solve the distributed scheduling optimization model, and obtains the optimal power regulation instruction initial value of each resource cluster. For example, the current output of a certain cluster is 3 MW, the load demand is 5 MW, and the SOC of the energy storage is 50%. The model calculates that the energy storage needs to discharge 1.5 MW, and at the same time, 0.5 MW is purchased from the adjacent cluster through the tie line to meet the power balance and minimize the cost. The 1.5 MW discharge instruction is used as the power regulation initial value of the cluster. The initial values of all clusters are organized in matrix form to form a cooperative scheduling instruction initial value matrix, and the matrix elements correspond to the active power regulation amount (unit: MW) of each cluster.

[0162] In some embodiments, the initial cross-regional cooperative scheduling instruction can also include scheduling period, priority, and other information, which are packaged in a standardized format.

[0163] The grid congestion marginal cost refers to the additional economic cost generated by increasing the transmission of each unit of power when the transmission of power in a specific region or time period is blocked due to the capacity limitation of the transmission line; and the regulation cost refers to the operating cost or compensation fee generated by regulating (such as charging and discharging, starting and stopping) the distributed energy, energy storage, or load to achieve power balance in the grid.

[0164] The initial cross-regional cooperative scheduling instruction is a refined result based on the initial value matrix and considering the current system operation requirements and constraints, which clearly defines the target power or regulation task of each resource cluster in the next scheduling period. The compensation price signal is usually calculated based on the coordination signal (such as the dual variable corresponding to the system-level constraint) issued by the coordinator in the distributed optimization process and the regulation cost of the resource itself, which is used to encourage the resource cluster to adjust according to the scheduling instruction and compensate for the service or cost generated.

[0165] Specifically, the scheduling server calculates the power flow of the tie line between clusters according to the cooperative scheduling instruction initial value matrix, and determines the economic compensation of cross-regional scheduling in combination with the real-time congestion marginal cost (LMP) of the grid. For example, the LMP difference of a certain tie line is 0.15 yuan / kWh, and a certain cluster transmits 1 MW of power to the adjacent cluster through the tie line for 1 hour, with a congestion compensation of 150 yuan. At the same time, considering the regulation cost of the resources in the cluster (such as the charging and discharging loss cost of the energy storage of 0.08 yuan / kWh), the initial cross-regional cooperative scheduling instruction (such as "discharge 1 MW for 1 hour") and the compensation price signal (such as the compensation unit price of 0.23 yuan / kWh) of the cluster are generated.

[0166] In some embodiments, the initial cross-regional cooperative scheduling instruction can also include scheduling period, priority, and other information, which are packaged in a standardized format.

[0167] S313, send the initial cross-regional coordinated dispatching instruction and the compensation price signal based on the mapping relationship between the dynamic resource cluster and the local agent.

[0168] The dispatching server sends the initial cross-regional coordinated dispatching instruction and the compensation price signal to each local agent through the SCADA or EMS communication network according to the predefined mapping table of the dynamic resource cluster and the local agent (such as the correspondence between the cluster ID and the regional controller IP address).

[0169] For example, cluster A corresponds to the North China regional agent, and cluster B corresponds to the East China regional agent. The instruction is transmitted through a special communication protocol (such as IEC 61850). After the local agent receives the instruction, it parses the instruction, converts it into a control signal of a local device (such as a power setting value of a energy storage converter), and performs settlement preparation according to the compensation price signal.

[0170] After the local agent receives the instruction and the signal, it will be responsible for the decomposition and execution of the instruction within the cluster, coordinating and managing its member resources, so that they adjust the power according to the instruction, and perform corresponding economic settlement or response according to the compensation price signal. The mapping relationship between the dynamic resource cluster and the local agent ensures that the dispatching instruction can be accurately transmitted to the entity responsible for execution.

[0171] S314, receive the aggregated response boundary information returned by each local agent based on the initial cross-regional coordinated dispatching instruction and the compensation price signal.

[0172] This step can refer to the description of the foregoing embodiments, which will not be repeated here.

[0173] S315, determine the power adjustment deviation matrix based on the initial cross-regional coordinated dispatching instruction and the aggregated response boundary information.

[0174] This step is to compare the initial cross-regional coordinated dispatching instruction sent to each dynamic resource cluster in the last round with the aggregated response boundary information received from each cluster. The aggregated response boundary information can reflect the actual power adjustment range or optimal response that the resource cluster can provide in the current state. By comparing the target power or adjustment amount in the instruction with the range allowed by the response boundary or the actual response value, the deviation of each resource cluster in power adjustment can be calculated. This deviation can be that the instruction value exceeds the range of the cluster's ability, or the difference between the optimal response of the cluster after receiving the compensation price signal and the instruction value. Arrange these deviation values according to the resource cluster to form a power adjustment deviation matrix.

[0175] For example, the initial dispatch instruction of a cluster requires discharging 2 MW, but the actual adjustable range of the cluster is 1.5-2.5 MW (aggregated response boundary), so the power regulation deviation is 0.5 MW; if the initial plan of a tie line is to transmit 3 MW, but its current actual available capacity is only 2.8 MW, then the deviation is 0.2 MW. The deviation values of all clusters and tie lines are arranged according to a preset index to form a power regulation deviation matrix, and the elements of the matrix are positive, indicating that the instruction exceeds the response capability, or negative, indicating that the response capability has redundancy.

[0176] S316, updating the compensation price signal according to the power regulation deviation matrix and real-time congestion data to obtain a compensation price update signal.

[0177] In the framework based on distributed optimization, the compensation price signal is usually associated with the marginal cost of the power grid or the dual variable of the constraint. When there is a large deviation in the power regulation of a certain resource cluster, or when congestion occurs in a certain area of the power grid, the corresponding compensation price signal can be adjusted to encourage the resource cluster to change its behavior. For example, if transmission congestion occurs in a certain area, the compensation price of the resource cluster related to the area can be adjusted to encourage them to make adjustments that help to alleviate the congestion. This updating process is usually iterative, aiming to guide the local optimal decision of each cluster to converge to the global optimal solution through the adjustment of the price signal, thereby obtaining a compensation price update signal.

[0178] Specifically, the dispatch server dynamically adjusts the compensation price based on the power regulation deviation matrix and in combination with real-time congestion data of the power grid (such as line LMP difference, congestion duration, etc.). If the deviation leads to the aggravation of line congestion in a certain area (such as the deviation causing the load rate of a certain line to rise from 80% to 90%), the unit price of the congestion compensation of the area is increased (such as from 0.1 yuan / kWh to 0.15 yuan / kWh); if the response capability of a certain cluster is insufficient to cause the adjustment of the dispatch plan, the cluster needs to transfer the power regulation task to other clusters, and the cluster receiving the transfer is paid compensation for the regulation service (such as 0.08 yuan / kWh). The updated compensation price signal contains the latest economic incentive parameters of each area and each period.

[0179] S317, feeding back and correcting the initial cross-area collaborative dispatch instruction based on the compensation price update signal to obtain a target collaborative dispatch instruction.

[0180] The dispatch server feeds back the compensation price update signal to the local agent corresponding to each dynamic resource cluster. After receiving the updated compensation price signal, the local agent will recalculate the optimal power response within its resource cluster or perform internal resource scheduling optimization according to the signal. This new optimal response or scheduling result reflects the decision of the resource cluster after considering the latest compensation price signal.

[0181] The dispatch server collects these new response information (or the information implied by the price signal itself) and revises the initial cross-regional coordinated dispatch instruction based on these feedbacks. The revised instruction is closer to the global optimal solution considering the actual capacity of each resource cluster and the grid constraints in the current state. After several iterations of such revision processes (in practical applications, it may be through a few iterations or generated once based on prediction information), the final converged or obtained instruction is the target coordinated dispatch instruction, which is used to guide the actual operation of each resource cluster.

[0182] Specifically, the dispatch server updates the compensation price as an economic constraint, re-solves the distributed dispatch optimization model, and revises the dispatch instruction to minimize the total cost (including congestion cost and regulation cost). For example, due to the increase in compensation price, the originally planned 2MW power regulation task is partially transferred to the adjacent region (such as 1.2MW), and the storage charging and discharging strategy is adjusted (such as increasing 0.3MW discharging) to balance the deviation. The revised instruction contains the new power regulation amount, regulation period and priority of each cluster, forming the final target coordinated dispatch instruction, ensuring economic optimal dispatch under the premise of meeting physical constraints.

[0183] S318, sends the target coordinated dispatch instruction to each dynamic resource cluster for execution to control the operation of the cross-regional virtual power plant resources.

[0184] This step can refer to the description of the foregoing embodiments, which will not be repeated here.

[0185] The method of the embodiment can improve the dispatch efficiency, flexibility and economy of the power system in complex operating environment by introducing dynamic resource cluster division and cross-regional coordinated dispatch method based on distributed optimization. The core technical effects are reflected in the following aspects:

[0186] Firstly, the dispatch is more refined and flexible. Traditional dispatch often takes a single unit or substation as a unit, which is inefficient when facing a large number of distributed resources and flexible loads. This scheme dynamically divides resources with similar characteristics or close geographical locations into resource clusters, achieving the aggregation management of large-scale heterogeneous resources. This aggregation not only simplifies the decision-making burden of the dispatch center, but also more accurately grasps the overall regulation capacity of regional resources, thereby realizing more refined dispatch control and improving the system's ability to cope with fluctuations.

[0187] Secondly, the cross-regional resource coordination optimization capability is enhanced. By constructing a distributed scheduling optimization model, the scheme can effectively coordinate different regions and different types of resource clusters to participate in system regulation. The distributed optimization framework allows each resource cluster to perform local optimization under the control of the local agent, while achieving global target coordination through information interaction with the coordinator (such as compensation price signals). This breaks down regional barriers, enabling cross-regional resources to more effectively participate in grid balancing and congestion management, improving the overall efficiency and economy of the system.

[0188] Thirdly, the economy of scheduling is improved. The scheme considers the marginal cost of grid congestion and the regulation cost, and guides the resource clusters to regulate through the compensation price signal. The compensation price mechanism can reflect the scarcity of resources and the value of regulation, encouraging resources with lower costs and stronger regulation capabilities to participate in scheduling first, thereby reducing the total operating cost of the system. Especially in dealing with grid congestion, the compensation price can effectively guide resources to perform regulation that is beneficial to alleviating congestion, avoiding high-cost curtailment of wind or solar power or load shedding.

[0189] Fourthly, the robustness and adaptability of the system are improved. Dynamic resource cluster division can flexibly adjust the composition of resource clusters according to system operating state, resource availability, etc., so that the scheduling scheme can better adapt to the dynamic changes of grid structure and the uncertainty of new energy output. The distributed scheduling architecture decentralizes part of the decision-making power to the local agent, reducing the computational pressure and communication burden of the central node, improving the anti-interference ability and stability of the system.

[0190] Finally, an effective scheduling tool is provided for new power systems. With the high proportion of new energy, distributed power and flexible load, the power system presents high randomness, high volatility and strong coupling characteristics. The dynamic aggregation and distributed collaborative scheduling method provided by the technical scheme can effectively manage and utilize these new resources, providing key technical support for building a new power system dominated by new energy.

[0191] In summary, the technical scheme realizes effective aggregation management and cross-regional optimization coordination of large-scale heterogeneous resources through dynamic resource cluster division and distributed collaborative scheduling, improving the refinement, flexibility, economy and robustness of power system scheduling.

[0192] The method provided by the above embodiments is executed by a scheduling server, and it can be understood that the scheduling server is an electronic device. The electronic device in the embodiments of the present application is described from the perspective of hardware processing. Please refer to Figure 3 , which is an entity device structure diagram of the electronic device in the embodiments of the present application.

[0193] It should be noted that Figure 3The structure of the electronic device shown is merely an example and should not impose any limitation on the functions and the range of use of the embodiments of the present application.

[0194] As shown in Figure 3 The electronic device includes a Central Processing Unit (CPU) 401 which can perform various appropriate actions and processes, such as executing the methods described in the above embodiments, according to programs stored in a Read-Only Memory (ROM) 402 or programs loaded from a storage section 408 into a Random Access Memory (RAM) 403. Various programs and data required for system operation are also stored in the Random Access Memory (RAM) 403. The Central Processing Unit (CPU) 401, the Read-Only Memory (ROM) 402, and the Random Access Memory (RAM) 403 are connected to each other through a bus 404. An Input / Output (I / O) interface 405 is also connected to the bus 404.

[0195] The following components are connected to the Input / Output (I / O) interface 405: an input section 406 including an audio input device, a button switch, and the like; an output section 407 including a display, an audio output device, an indicator, and the like; a storage section 408 including a hard disk and the like; and a communication section 409 including a network interface card such as a LAN (Local Area Network) card, a modem, and the like. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the Input / Output (I / O) interface 405 as necessary. A removable medium 411 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, and the like is mounted on the drive 410 as necessary so that a computer program read therefrom is installed in the storage section 408 as necessary.

[0196] In particular, the processes described above with reference to the flowcharts can be implemented as a computer software program in accordance with embodiments of the present application. For example, embodiments of the present application include a computer program product comprising a computer program carried on a computer readable medium, the computer program comprising computer instructions for performing the methods illustrated by the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via the communication section 409, and / or installed from the removable media 411. When the computer program is executed by the central processing unit (CPU) 401, various functions defined in the present application are performed.

[0197] Note that specific examples of computer readable storage media can include without limitation: an electrical connection having one or more wires; a portable computer diskette; a hard disk; a random access memory (RAM); a read-only memory (ROM); an erasable programmable read only memory (EPROM); a flash memory; an optical fiber; a portable compact disc read-only memory (CD-ROM); an optical storage device; a magnetic storage device; or any suitable combination of the foregoing. In the present application, a computer readable storage medium can be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

[0198] The flowcharts and block diagrams in the drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowcharts or block diagrams can represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks can sometimes be executed in the reverse order, depending upon the functionality involved.

[0199] In particular, the electronic device of the present embodiment includes a processor and a memory coupled to the one or more processors, the memory storing computer program code comprising computer instructions to be invoked by the one or more processors to cause the electronic device to perform the method provided by the above-described embodiments.

[0200] As another aspect, the present application also provides a computer readable storage medium, which can be included in the electronic device described in the above embodiments, or can exist separately without being assembled into the electronic device. The storage medium carries one or more computer programs, which, when executed by a processor of the electronic device, cause the electronic device to implement the method provided in the above embodiments.

[0201] The above described and above embodiments are only used to illustrate the technical solutions of the present application, rather than limiting them; even though the present application has been described in detail with reference to the foregoing embodiments, those ordinarily skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some technical features thereof; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0202] In the above embodiments, according to the context, the term "when" can be interpreted as meaning "if" or "after" or "in response to determining" or "in response to detecting". Similarly, according to the context, the phrase "upon determining" or "if detecting (the stated condition or event)" can be interpreted as meaning "if determining" or "in response to determining" or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)".

[0203] Those ordinarily skilled in the art can understand that all or part of the processes in the above method embodiments can be implemented by a computer program instructing relevant hardware, which can be stored in a computer readable storage medium, and when executed, can include the processes of the above method embodiments. The foregoing storage medium includes ROM or random access memory (RAM), magnetic or optical disks, and various other storage media that can store program codes.

Claims

1. A method for cross-regional virtual power plant collaborative scheduling, characterized in that, The method comprises the following steps: Collecting real-time characteristic data of each distributed energy resource and load, wherein the real-time characteristic data at least includes one of spatio-temporal location information, output / demand prediction curve, physical operation parameter, economic characteristic parameter and dispatchable state; Based on the global optimization target set in the current dispatching cycle and the real-time characteristic data, the dynamic weight of each characteristic dimension is determined; According to the dynamic weight and the real-time characteristic data, a dynamic resource cluster division instruction is generated, wherein the dynamic resource cluster division instruction includes a member list of a resource cluster and a cooperative constraint condition; Based on the dynamic resource cluster division instruction, an initial cross-regional cooperative dispatching instruction and a compensation price signal are sent to each local agent corresponding to the dynamic resource cluster, wherein the initial cross-regional cooperative dispatching instruction includes a net exchange power target value and an auxiliary service demand; Receiving aggregated response boundary information returned by each local agent based on the initial cross-regional cooperative dispatching instruction and the compensation price signal, wherein the aggregated response boundary information is used to represent the feasible domain range of the dynamic resource cluster to the instruction; According to the aggregated response boundary information, the cooperative dispatching instruction and the compensation price signal are updated to obtain a target cooperative dispatching instruction, including: based on the initial cross-regional cooperative dispatching instruction and the aggregated response boundary information, a power adjustment deviation matrix is determined; according to the power adjustment deviation matrix and real-time congestion data, the compensation price signal is updated to obtain a compensation price update signal; based on the compensation price update signal, the initial cross-regional cooperative dispatching instruction is corrected to obtain the target cooperative dispatching instruction; The target cooperative dispatching instruction is sent to each dynamic resource cluster for execution to control the operation of the cross-regional virtual power plant resource.

2. The method of claim 1, wherein, The dynamic weight of each characteristic dimension is determined based on the global optimization target set in the current dispatching cycle and the real-time characteristic data, including: According to the global optimization target of the current dispatching cycle, a key characteristic dimension identifier is determined; Based on the key characteristic dimension identifier, an influence weight coefficient of each dimension on the target achievement is calculated; According to the resource state constraint in the real-time characteristic data, the influence weight coefficient is corrected to obtain the dynamic weight; the resource state constraint refers to the physical boundary condition limiting the resource dispatching feasible domain, including maximum / minimum power limit, energy storage SOC safety interval, device start-stop state and network topology connection relationship.

3. The method of claim 2, wherein, Based on the key characteristic dimension identifier, an influence weight coefficient of each dimension on the target achievement is calculated, including: Querying a preset target characteristic mapping rule library to obtain a reference influence value of each key characteristic dimension on the optimization target; Based on the reference influence value and the resource dynamic characteristic in the real-time characteristic data, an actual contribution correction factor of each dimension on the target achievement is calculated; the resource dynamic characteristic refers to the real-time changeable operation capability parameter of the distributed energy resource, including the current output / load curve, response speed, ramp rate and cost fluctuation rate; The actual contribution correction factor is normalized to obtain the influence weight coefficient.

4. The method of claim 1, wherein, The dynamic resource cluster division instruction is generated according to the dynamic weight and the real-time characteristic data, including: generate a resource-weighted feature vector according to the dynamic weight and the real-time feature data; perform division processing on the resource-weighted feature vector through a constraint clustering algorithm to obtain a resource initial grouping scheme; verify physical operability of the resource initial grouping scheme to obtain resource cluster entities satisfying a collaborative constraint condition; generate dynamic resource cluster division instructions based on a member list and a constraint boundary of each resource cluster entity.

5. The method of claim 4, wherein, The division processing on the resource-weighted feature vector through the constraint clustering algorithm to obtain the resource initial grouping scheme comprises: construct a resource similarity matrix based on the resource-weighted feature vector; correct connection weights of the resource similarity matrix according to a power grid physical constraint condition to obtain a constraint-corrected similarity matrix; perform spectral clustering feature decomposition on the constraint-corrected similarity matrix to obtain the resource initial grouping scheme satisfying the constraint.

6. The method of claim 1, wherein, The sending of the initial cross-regional collaborative scheduling instructions and the compensation price signals to local agents corresponding to each dynamic resource cluster based on the dynamic resource cluster division instructions comprises: generate power adjustment interval boundaries of each dynamic resource cluster based on a topological structure in the dynamic resource cluster division instructions; construct a distributed scheduling optimization model according to the power adjustment interval boundaries; calculate in-cluster optimal power instruction initial values based on the distributed scheduling optimization model to obtain a collaborative scheduling instruction initial value matrix; generate initial cross-regional collaborative scheduling instructions and compensation price signals according to the collaborative scheduling instruction initial value matrix, a power grid congestion marginal cost and an adjustment cost; send the initial cross-regional collaborative scheduling instructions and the compensation price signals based on a mapping relationship between the dynamic resource clusters and the local agents.

7. An electronic device, comprising: comprise one or more processors and a memory; The memory is coupled to the one or more processors, and the memory is configured to store computer program codes, the computer program codes comprising computer instructions, and the one or more processors are configured to invoke the computer instructions to enable the electronic device to perform the method according to any one of claims 1-6.

8. A computer readable storage medium storing computer instructions, characterized in that, The computer instructions enable the electronic device to perform the method according to any one of claims 1-6 when the computer instructions run on the electronic device.

9. A computer program product, characterised in that, The computer program product enables the electronic device to perform the method according to any one of claims 1-6 when the computer program product runs on the electronic device.

Images