Distributed resource dynamic heterogeneous aggregation and layered inertia support method, system and device and medium

Abstract

The invention discloses a distributed resource dynamic heterogeneous aggregation and layered inertia support method, system, device and medium, and belongs to the technical field of power system operation and control, and the method comprises the steps: constructing a heterogeneous resource unified state space model fusing stress and uncertainty, and mapping all resources into a three-dimensional space; defining and calculating complementary characteristic indexes among the distributed resources, and constructing a real-time dynamic complementation degree matrix; constructing a dynamic heterogeneous aggregation mechanism based on a complementation degree matrix, and dividing physically dispersed resources into virtual inertia aggregates with the maximum complementation degree; and three-layer cooperative inertia support control is established and implemented based on the virtual inertia aggregate generated by aggregation. According to the method, the communication calculation pressure of massive nodes is remarkably reduced while the timeliness of millisecond inertia response and the economic optimization of the whole network are guaranteed, and refined protection of energy storage life and robust response to uncertainty of a user side are realized.

Description

Technical Field

[0001] This invention relates to the field of advanced technology in power system operation and control, specifically to a method, system, device, and medium for dynamic heterogeneous aggregation and hierarchical inertia support of distributed resources. Background Technology

[0002] With the acceleration of the global energy transition, the power system is evolving from a traditional form dominated by synchronous generators to a new form dominated by new energy sources. Wind power, photovoltaic power, and other new energy power generation equipment are typically connected to the grid via power electronic converters. Their rotor speed is decoupled from the grid frequency, and they cannot provide natural rotational inertia like traditional synchronous generators. This characteristic of "high proportion of renewable energy and high proportion of power electronic equipment" leads to a significant decrease in the overall inertia level of the power grid. In low-inertia systems, when power disturbances occur, the rate of change of frequency (RoCoF) is extremely rapid, with a lower minimum frequency, easily triggering low-frequency load shedding and even causing large-scale power outages.

[0003] To address these issues, exploring the potential of "virtual inertia" in distributed resources has become an industry consensus. However, existing inertia scheduling and control technologies face severe challenges when dealing with massive amounts of distributed resources, and unresolved pain points remain in system-level coordination.

[0004] For the participation of massive distributed resources in power grid frequency regulation, current research and applications both domestically and internationally mainly focus on the following scattered technical directions. Although some progress has been made in their respective fields, they still face bottlenecks in system-level coordination: Lifetime loss and stress management issues in energy storage regulation: Most current energy storage frequency regulation control strategies are based on fixed droop curves or simple charge / discharge priorities, often ignoring the instantaneous impact of high-frequency power fluctuations on battery life. Although some literature has proposed economic dispatch models that consider State of Health (SOH), most treat battery loss as a long-term static cost, lacking a real-time stress control mechanism dynamically coupled with the Rate of Change in Frequency (RoCoF). This leads to situations where, in critical moments such as sharp frequency drops, energy storage systems may suffer excessive mechanical and thermal stress due to a lack of refined protection mechanisms, or premature depletion of System Capacity (SoC) due to over-response, failing to maintain continuous support.

[0005] Uncertainty and Conservative Control Issues of User-Side Resources: User-side resources, represented by electric vehicle (EV) clusters, exhibit high stochasticity. Existing control methods typically employ robust optimization or stochastic programming to address the uncertainty of user behavior (such as sudden grid disconnection due to travel demand). However, to ensure system safety, existing robust boundary settings are often overly conservative, sacrificing adjustment potential for most of the time period to cope with extremely low-probability severe scenarios. Furthermore, these resources are usually treated as passive controlled units, lacking mechanisms for real-time dynamic interaction with power generation-side resources (such as photovoltaics).

[0006] The issue of offline and singular resource response models: Existing technologies have established relatively complete transfer function models for the frequency response characteristics of different resources such as photovoltaics, energy storage, and temperature-controlled loads. However, these models are mainly used for offline simulation analysis or static parameter tuning, lacking the ability for online real-time control. In actual operation, the power grid's operating conditions are constantly changing, and models based on fixed parameters struggle to achieve adaptive matching for different inertia demand scenarios.

[0007] Communication and Convergence Bottlenecks in Distributed Cooperative Control Architectures: To achieve coordination of large-scale distributed resources, distributed control based on consensus algorithms is the current mainstream technical approach. However, traditional flat, fully distributed architectures face severe scalability challenges when dealing with the massive number of nodes in a distribution network. The increase in the number of nodes leads to an exponential increase in the burden on the communication network, slows down the algorithm's iteration and convergence speed, and makes it difficult to meet the stringent requirements of frequency security control for millisecond to second-level response speeds. In addition, existing resource aggregation is usually based on static partitioning by geographical location or physical type, ignoring the natural complementarity of different resources in the time, frequency, and energy domains.

[0008] In summary, existing technologies typically treat various resources as "islands" or employ rigid aggregation methods. In reality, distributed resources possess highly valuable "complementary characteristics": temporal complementarity—the volatility of photovoltaic output and the randomness of load changes often have a hedging effect over time. Frequency complementarity—power-type resources (such as supercapacitors) are suitable for handling high-frequency disturbances, while energy-type resources (such as EVs) are suitable for handling persistent low-frequency shortages. Energy complementarity—high-SoC energy storage and low-SoC EVs can form a closed loop in energy flow, reducing the impact on the main grid. Summary of the Invention

[0009] In view of the above-mentioned problems, the present invention is proposed.

[0010] Therefore, this invention aims to overcome the limitations of existing technologies, such as single resource utilization and slow convergence due to flat control architecture, and proposes a dynamic heterogeneous aggregation method based on complementary characteristics. By quantifying the complementarity between resources, physically dispersed resources are dynamically reorganized into virtual units with self-balancing capabilities, and hierarchical control is implemented. This achieves global optimization of network-wide inertia support while ensuring low communication costs and high response speed.

[0011] To address the aforementioned technical problems, this invention provides the following technical solution: a method for dynamic heterogeneous aggregation and hierarchical inertia support of distributed resources, comprising, A unified state-space model for heterogeneous resources integrating stress and uncertainty is constructed, mapping all resources into a three-dimensional space; complementary characteristic indices among distributed resources are defined and calculated, and a real-time dynamic complementarity matrix is ​​constructed; a dynamic heterogeneous aggregation mechanism based on the complementarity matrix is ​​constructed, dividing physically dispersed resources into virtual inertia aggregates with maximized complementarity; based on the virtual inertia aggregates generated by aggregation, a three-layer collaborative inertia support control is established and implemented.

[0012] As a preferred embodiment of the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method described in this invention, the heterogeneous resource unified state space model includes establishing a high-fidelity model that includes physical constraints, operating costs and dynamic characteristics. The high-fidelity model integrates the MPPT unloaded potential energy model of distributed photovoltaics, the dynamic inertia stress model of energy storage systems, and the probabilistic robust capacity model of electric vehicle clusters. Map the state of all resource models to a unified three-dimensional feature space.

[0013] As a preferred embodiment of the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method described in this invention, the construction of the real-time dynamic complementarity matrix includes defining complementary characteristic indicators that describe the collaborative potential between different resources based on the established three-dimensional feature space. Based on the complementary characteristic index, the comprehensive complementarity between any two nodes in the entire network is calculated, and a real-time dynamic complementarity matrix is ​​constructed.

[0014] As a preferred embodiment of the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method described in this invention, the dynamic heterogeneous aggregation mechanism includes: using a dynamic complementarity matrix as a similarity matrix, dynamically dividing the distributed resources in the physical network into separate virtual inertia aggregates, and performing aggregation principles. Set reconstruction trigger conditions to trigger the update of the complementarity matrix and spectral clustering reconstruction.

[0015] As a preferred embodiment of the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method described in this invention, the dynamic heterogeneous aggregation mechanism further includes using a dynamic complementarity matrix and a dynamic graph segmentation algorithm to reconstruct the physical network into a virtual inertia aggregate. Construct a graph Laplace matrix, treating the power grid as a graph. The weight of the edge is Calculate the degree matrix ,in Calculate the normalized Laplace matrix: ; Perform spectral embedding, for Perform eigenvalue decomposition and solve the characteristic equation. Select the eigenvectors corresponding to the K smallest non-zero eigenvalues. Constructing a matrix Each row of U is considered as the coordinate of the corresponding node in the K-dimensional feature space; Virtual inertia aggregates are generated using K-means clustering. K-means clustering is performed on the coordinates of N nodes in the feature space, dividing the entire network into K clusters. Each cluster is a virtual inertia aggregate; Setting the trigger conditions for dynamic refactoring includes timed triggering and event triggering; The timed trigger is every [time]. Recalculate the aggregation; The event is triggered when the power imbalance within a virtual inertia aggregate exceeds 80% of the corresponding adjustment capability, or when the system frequency deviation... At that time, a forced refactoring is triggered to find new complementary resources; in, In this context, V represents the set of nodes and E represents the set of edges. Let be the edge weight, representing the complementarity between node m and node n; For degree matrix, These are the diagonal elements of the degree matrix. For the normalized Laplace matrix, It is the identity matrix; This is the complementarity matrix. The time interval for triggering at regular intervals. This represents the absolute value of the system frequency deviation.

[0016] As a preferred embodiment of the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method described in this invention, the three-layer collaborative inertia support control includes constructing a hierarchical control architecture, with different layers executing different control logic and time scales: The underlying layer is device-level, the time scale is millisecond-level, each distributed resource executes local adaptive droop control, introduces dynamic participation factors, and fine-tunes based on local real-time status; The middle layer is at the aggregate level, with a time scale of seconds. Each virtual inertia aggregate has an aggregate leader to collect the status of its members, solve the power allocation problem based on internal complementarity, and utilize the complementarity of heterogeneous resources to prioritize the use of low-cost, high-health resources to smooth out internal fluctuations. The upper layer is system-level, with a time scale of seconds to minutes. It adopts an improved hierarchical consensus algorithm to interact with the marginal inertia cost and net power deficit of all aggregates. Through iteration, it eliminates the power imbalance across the entire network and achieves convergence of marginal costs among all aggregates, thus achieving global economic optimization.

[0017] As a preferred embodiment of the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method described in this invention, the underlying layer includes device-level adaptive droop control, where each device i executes a control law: Among them, dynamic parameters and It is issued in real time by the middle layer controller: in, The command is for the real-time power reference value of device i. To provide initial foundation power for device i, This is the dynamic droop coefficient. For dynamic virtual inertia coefficient, As a participating factor, This refers to the rated adjustment coefficient of the equipment. The middle layer includes aggregate-level complementary optimization, where each virtual inertia aggregate selects an aggregate leader node. The aggregate leader collects information about members within the cluster and solves the optimization problem. Using the Lagrange multiplier method, the aggregate leader calculates the current uniform incremental rate of the aggregate: in, A collection of aggregates Let i be the cost function of device i. The amount of power adjustment shared by device i This represents the total power deficit that needs to be balanced within the k-th polymer. / Let i be the minimum / maximum adjustable power constraint for device i at the current moment. The uniform incremental rate of the k-th aggregate; The upper layer includes system-level hierarchical consistency coordination, where the leaders of each virtual inertia aggregate constitute an upper-layer sparse communication network to execute an improved consensus algorithm. in, Let this be the marginal cost state of aggregate k at the next iteration. Let k be the set of neighboring aggregates that are directly connected to aggregate k in the upper-layer communication network. The communication weighting coefficient between aggregates k and j. This represents the iteration step size of the algorithm. / The total power generation and total load power within polymer k are given. The power requirements are to support the macroscopic inertia allocated to the aggregate.

[0018] Another objective of this invention is to provide a distributed resource dynamic heterogeneous aggregation and hierarchical inertia support system.

[0019] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a distributed resource dynamic heterogeneous aggregation and hierarchical inertia support system, comprising: a model building module, a complementary module, an aggregation module, a support module, a multi-type distributed resource controller, and a hierarchical communication network; The model building module constructs a unified state space model of heterogeneous resources that integrates stress and uncertainty, mapping all resources into a three-dimensional space. The complementary module defines and calculates the complementary characteristic index between distributed resources, and constructs a real-time dynamic complementarity matrix. The aggregation module constructs a dynamic heterogeneous aggregation mechanism based on the complementarity matrix, which divides physically dispersed resources into virtual inertia aggregates with maximized complementarity. The support module establishes and implements a three-layer collaborative inertia support control based on the virtual inertia aggregate generated by aggregation. The multi-type distributed resource controller includes PV inverters, PCS, charging pile controllers, and an aggregate manager configured on the edge computing node; The hierarchical communication network is a hierarchical communication network connecting controllers at each level, and is configured to automatically execute complementary resource evaluation, dynamic aggregation, and hierarchical collaborative control commands.

[0020] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method.

[0021] The present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method.

[0022] The beneficial effects of this invention are as follows: By deeply exploring the multi-dimensional complementary characteristics of heterogeneous resources, this invention constructs a dynamic aggregation mechanism and a three-layer collaborative architecture. While ensuring millisecond-level inertial response timeliness and optimal network economy, it significantly reduces the communication and computing pressure of massive nodes and achieves refined protection of energy storage lifespan and robust response to user-side uncertainties. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart illustrating a distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method according to an embodiment of the present invention. Detailed Implementation

[0025] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0026] Example 1, referring to Figure 1 This is one embodiment of the present invention, which provides a method for dynamic heterogeneous aggregation and hierarchical inertia support of distributed resources, including: S100. Construct a unified state-space model for heterogeneous resources that integrates stress and uncertainty, and map all resources into three-dimensional space. S200. Define and calculate the complementarity index between distributed resources, and construct a real-time dynamic complementarity matrix. S300. Construct a dynamic heterogeneous aggregation mechanism based on the complementarity matrix to divide physically dispersed resources into virtual inertia aggregates with maximized complementarity. S400: Based on the virtual inertia aggregate generated by aggregation, establish and implement a three-layer collaborative inertia support control; It should be noted that existing technologies suffer from drawbacks such as single resource utilization, insufficient complementarity mining, slow convergence of flattened communication, and rigid aggregation methods. This invention, through dynamic aggregation, enables the "expensive" adjustment requirements to be absorbed at low cost within the aggregate through complementary resources, maximizing the benefits of resource complementarity. Through a layered architecture, the bottom layer guarantees millisecond-level response, while the upper layer achieves network-wide economic optimization, solving the problem of large-scale node computing. By utilizing the self-balancing capability of heterogeneous aggregates, the system's dependence on the communication network is reduced, and its anti-disturbance capability is enhanced.

[0027] Therefore, addressing the aforementioned problems, through steps S100-S400, this invention relates to a systematic method for supporting grid frequency inertia by utilizing the inherent complementary characteristics of massive, heterogeneous, and dispersed distributed resources (including distributed photovoltaics, electrochemical energy storage, electric vehicle clusters, temperature-controlled loads, etc.) in the time, frequency, and energy domains, through a dynamic graph theory aggregation algorithm and a hierarchical collaborative architecture. This method is applicable to frequency security defense and optimized scheduling of distribution networks, microgrid clusters, virtual power plants (VPPs), and regional integrated energy systems.

[0028] Example 2, refer to Figure 1 This is one embodiment of the present invention, which provides a method for dynamic heterogeneous aggregation and hierarchical inertia support of distributed resources, including: In this embodiment of the invention, S100 constructs a unified state-space model for heterogeneous resources that integrates stress and uncertainty, mapping all resources to a three-dimensional space, including the following steps S101-S102: S101. Establish a high-fidelity model that includes physical constraints, operating costs, and dynamic characteristics; The high-fidelity model integrates the MPPT unloaded potential energy model of distributed photovoltaics, the dynamic inertia stress model of energy storage systems, and the probabilistic robust capacity model of electric vehicle clusters. Specifically, the MPPT unloading potential energy model for the distributed photovoltaic system is as follows: Traditional photovoltaic systems operate in MPPT mode and lack inertia. This invention adopts a combination of reserved backup and virtual inertia of DC capacitors. Define the output power of photovoltaic unit i for: in, This represents the maximum power point power under the current irradiance. To reserve a margin (e.g., 10%), which is used to provide an upward adjustment margin for inertia; The power supplied by inertia consists of two parts: the virtual inertia of the DC capacitor and the reserve release during load shedding. The virtual inertia of the DC capacitor is derived from the DC-side capacitor of the Boost converter. voltage Waves release energy; energy change for: in, This is the reference voltage value for the DC-side capacitor of the Boost converter.

[0029] The load shedding and reserve release is achieved by adjusting the duty cycle to move the operating point toward the MPPT point, thereby releasing reserve power. (Photovoltaic inertia cost function) The primary consideration is the opportunity cost of curtailing solar power: This function is a strongly convex function, where... , This is a coefficient related to current electricity prices and subsidies.

[0030] Specifically, the dynamic inertia stress model of the energy storage system is a stress-lifetime coupled model. The operation of energy storage is not only about energy exchange, but also about the consumption of battery life.

[0031] Generalized cost function of energy storage unit j Defined as: Among them, dynamic inertia stress coefficient Core innovation points: in, The charging and discharging power of energy storage unit j For system frequency, The frequency change rate is denoted by SoC, which represents the state of charge. For SoC adjustment factor, Here, BaseCost represents the real-time state of charge of energy storage unit j, and BaseCost represents the basic regulation cost of the energy storage. This is the stress sensitivity coefficient. The loss factor is taken into account the battery health status.

[0032] When the system frequency change rate When the value is large, the coefficient grows exponentially, which means that the economic cost of calling up the battery for high-rate discharge is artificially increased by the algorithm at the moment when the frequency drops sharply. This forces the system to reduce the high-rate impact on the battery in non-emergency situations, or to prioritize calling up batteries with better health.

[0033] SOH factor The lower the battery's health level, the higher its unit adjustment cost, thus it is naturally protected in the algorithm.

[0034] Specifically, the probabilistic robust capacity model for an electric vehicle cluster (EV cluster) determines the maximum adjustable capacity for an EV aggregator k. It is a random variable; based on chance constraint transformation, a robust available capacity boundary is defined. : in, This represents the expected available cluster capacity at the current moment, predicted based on historical data. The standard deviation of the forecast reflects the degree of uncertainty in user behavior; The confidence level represents the probability of default allowed by the system.

[0035] EV Inertia Cost Function : in, This is the battery loss cost coefficient; For the power adjustment of electric vehicle clusters; n is the total number of electric vehicles in the cluster; n is the electric vehicle number index. The current moment; Costs of user anxiety, as user time spent away With the approach of [a specific date], this [item / condition] increased dramatically, limiting the availability of vehicles about to leave the network.

[0036] S102. Map the state of all resource models to a unified three-dimensional feature space.

[0037] In this embodiment of the invention, step S200 defines and calculates the complementarity characteristic index between distributed resources and constructs a real-time dynamic complementarity matrix, including the following steps S201-S202: This is the key difference between this invention and the prior art: it is necessary to mathematically quantify why combining photovoltaics and EVs is better than combining two photovoltaics together. Define the entire network node set as For any two nodes Define complementarity .

[0038] S201. Based on the established three-dimensional feature space, complementary characteristic indicators describing the synergistic potential between different resources are defined, including time-domain power complementarity, frequency-domain response complementarity, and energy-domain complementarity. Specifically, the time-domain power complementarity This indicator measures the ability of two resources to hedge against natural power fluctuations. If an increase in the output of one resource corresponds exactly to a decrease in the output (or an increase in the load) of the other resource, then the two resources are highly complementary and can achieve internal balance. in, Let be the output power of node m at time t. Let n be the output power of node n at time t; The Pearson correlation coefficient has a value range of [-1, 1]. like If it is a perfect negative correlation, then The complementarity is the strongest; for example, photovoltaic power output is mainly during the day, while some nighttime charging loads have low complementarity with it; however, the air conditioning load of office buildings is positively correlated with photovoltaic output (high light intensity), and if it is a comparison between photovoltaic and load reduction capacity, then the opposite needs to be considered.

[0039] In this invention, it is even more important to explain the complementarity of adjustment capabilities, that is, when node m is in the "dead zone" or "bottleneck" of adjustment capability, whether node n is in the "abundant zone" of adjustment capability.

[0040] The frequency domain response complementarity This indicator measures the responsiveness of a resource across different frequency bands. Using a frequency response model, the frequency domain transfer function of each resource can be obtained. : If node m excels in high-frequency response (e.g., supercapacitors, with large bandwidth), while node n excels in low-frequency response (e.g., EVs, with small bandwidth but large capacity), then their amplitude-frequency characteristics differ significantly, resulting in large integral values ​​and strong complementarity. By combining them, a virtual unit with excellent full-band response can be constructed.

[0041] in, This is the weighting function used in frequency domain response calculation. Let n be the frequency domain transfer function of node n.

[0042] The energy domain complementarity This measures the degree of matching between the two in terms of energy buffer pools, i.e.: in, For type indicator functions, Let m be the charge state of node m. The charge state of node n.

[0043] Physical meaning: A SoC with 90% energy storage and a SoC with 10% energy storage can be combined to form a combination with strong bidirectional regulation capabilities, one responsible for discharging and the other for charging.

[0044] S202. Based on the complementary characteristic index, calculate the comprehensive complementarity between any two nodes in the network and construct a real-time dynamic complementarity matrix. Considering the above indicators, and to avoid increased network loss due to the aggregation of nodes that are too physically far apart, an electrical distance constraint is introduced: in, For electrical distance, The weighting coefficients include , , , This is the electrical distance normalization coefficient.

[0045] In an embodiment of the present invention, a dynamic heterogeneous aggregation mechanism based on the complementarity matrix is ​​constructed in S300 to divide physically dispersed resources into virtual inertia aggregates with maximized complementarity, including the following steps S301-S302: S301. Using the dynamic complementarity matrix as the similarity matrix, the distributed resources in the physical network are dynamically divided into separate virtual inertia aggregates (VIA), and the aggregation principle is applied. Specifically, by using a dynamic complementarity matrix and a dynamic graph segmentation algorithm, the physical network is reconstructed into a virtual inertia aggregate. Construct a graph Laplace matrix, treating the power grid as a graph. The weight of the edge is Calculate the degree matrix ,in Calculate the normalized Laplace matrix: Among them, among them, In this context, V represents the set of nodes (representing each distributed resource) and E represents the set of edges (representing the logical connections between resources). Let be the edge weight, representing the complementarity between node m and node n; The degree matrix is ​​a diagonal matrix that reflects the overall complementary strength of the nodes; represents the diagonal elements of the degree matrix, and its value is the sum of the complementary degrees of all edges connected to node i; This is the normalized Laplacian matrix, used for subsequent feature extraction; It is the identity matrix; It is a complementarity matrix, which is composed of the comprehensive complementarity between any two nodes in the entire network; The time interval for triggering at regular intervals is used to recalculate the aggregation relationship at fixed intervals. This represents the absolute value of the system frequency deviation.

[0046] Perform spectral embedding, for Perform eigenvalue decomposition and solve the characteristic equation. Select the eigenvectors corresponding to the K smallest non-zero eigenvalues. Constructing a matrix Each row of U is considered as the coordinates of the corresponding node in the K-dimensional feature space; where v is the feature vector. for, For the Kth eigenvector, It is the set of real numbers; Virtual inertia aggregates are generated using K-means clustering. K-means clustering is performed on the coordinates of N nodes in the feature space, dividing the entire network into K clusters. Each cluster is a virtual inertia aggregate; The principle is as follows: the objective function of spectral clustering is equivalent to minimizing the normalized cut graph, that is, minimizing the connection weights between clusters (low complementarity) and maximizing the connection weights within clusters (high complementarity). This ensures that each VIA has the strongest complementary self-balancing ability.

[0047] S302. Set the reconstruction trigger conditions to trigger the update of the complementarity matrix and spectral clustering reconstruction; Specifically, setting the trigger conditions for dynamic refactoring includes timed triggering and event triggering; The timed trigger is every [time]. (e.g., every 15 minutes) Recalculate the aggregation; The event is triggered when the power imbalance within a virtual inertia aggregate exceeds 80% of the corresponding adjustment capability, or when the system frequency deviation... At that time, a forced refactoring is triggered to find new complementary resources.

[0048] In an embodiment of the present invention, S400 establishes and implements a three-layer cooperative inertia support control based on the virtual inertia aggregate generated by aggregation, including the following steps S401-S403: S401. Construct a hierarchical control architecture, with different levels executing different control logic and time scales: The underlying layer is device-level, the time scale is millisecond-level, each distributed resource executes local adaptive droop control, introduces dynamic participation factors, and fine-tunes based on local real-time status; Specifically, the underlying layer is device-level adaptive droop control with millisecond-level response, ensuring system survivability and providing the fastest response. Each device i executes a control law: Dynamic parameters and The key innovation is that these two parameters are not fixed, but are issued in real time by the middle-level controller.

[0049] in, This is the command for the real-time power reference value of device i; Provide initial foundational force for device i; The dynamic droop factor determines the magnitude of power adjustment as the frequency deviation changes; The dynamic virtual inertia coefficient determines the magnitude of the power adjustment as the frequency changes. This refers to the rated adjustment coefficient of the equipment. These are participating factors, related to the health, cost, and current state of the equipment; for example, when energy storage stress is excessive. It will decrease automatically.

[0050] S402, the middle layer is at the aggregate level, the time scale is at the second level, each virtual inertia aggregate has an aggregate leader, collects the status of internal members, solves the power allocation problem based on internal complementarity, and utilizes the complementarity of heterogeneous resources to prioritize the use of low-cost, high-health resources to smooth out internal fluctuations. Specifically, the middle layer features aggregate-level complementary optimization and second-level coordination, achieving power balance within the VIA and minimizing total internal cost. Each VIA selects a cluster leader node. The leader collects information about the cluster members and solves the following optimization problem: The solution strategy utilizes the Lagrange multiplier method, and Leader calculates the uniform incremental rate (i.e., the marginal cost of the aggregate) for the aggregate. in, A collection of aggregates; Let i be the cost function of device i; The amount of power adjustment shared by device i; This represents the total power deficit that needs to be balanced within the k-th aggregate. / The minimum / maximum adjustable power constraint for device i at the current moment; The uniform incremental rate of the k-th aggregate; Due to the various resource cost functions The parameters have been uniformly defined in step S100 (including stress, opportunity cost, etc.). This step automatically realizes the economic allocation of heterogeneous resources. For example, when the frequency fluctuates slightly, the low-cost MPPT load shedding is mainly called; when the frequency fluctuates drastically, the high-cost energy storage is called.

[0051] S403, the upper layer is system-level, with a time scale of seconds to minutes. It adopts an improved hierarchical consensus algorithm, interacts the marginal inertia cost and net power deficit of all aggregates, and eliminates the power imbalance across the entire network through iteration, and realizes the convergence of marginal costs among all aggregates, achieving global economic optimization. Specifically, the upper layer employs system-level hierarchical consistency and coordination, operating at the second-to-minute level; it eliminates marginal cost differences between VIAs to achieve network-wide optimization, with the Leaders of each VIA forming the upper-layer sparse communication network; and it executes an improved consensus algorithm. in, Let this be the marginal cost state of aggregate k at the next iteration. Let k be the set of neighboring aggregates that are directly connected to aggregate k in the upper-layer communication network. The communication weighting coefficient between aggregates k and j; This represents the iteration step size of the algorithm. / The total power generation and total load power within polymer k; The power requirements are to support the macroscopic inertia allocated to this aggregate; The variable here is the macroscopic state of the "aggregate," not the state of a single device. This allows the number of nodes participating in the iteration to increase from... (Equipment level) reduced to (At the aggregate level), it greatly improves the convergence speed and meets the timeliness requirements of inertia support.

[0052] Example 3 is an embodiment of the present invention, illustrating a distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method. It should be noted that the technical solution of a distributed resource dynamic heterogeneous aggregation and hierarchical inertia support system and the technical solution of the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method described above belong to the same concept. Details not described in detail in the technical solution of the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support system in this embodiment can be found in the description of the technical solution of the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method described above.

[0053] This embodiment provides a distributed resource dynamic heterogeneous aggregation and hierarchical inertia support system, including: a model building module, a complementary module, an aggregation module, a support module, a multi-type distributed resource controller, and a hierarchical communication network; The model building module constructs a unified state-space model of heterogeneous resources that integrates stress and uncertainty, mapping all resources into a three-dimensional space. The complementarity module defines and calculates the complementarity characteristic index between distributed resources and constructs a real-time dynamic complementarity matrix. The aggregation module constructs a dynamic heterogeneous aggregation mechanism based on the complementarity matrix, which divides physically dispersed resources into virtual inertia aggregates that maximize complementarity. The support module establishes and implements a three-layer collaborative inertia support control based on the virtual inertia aggregate generated by aggregation. The multi-type distributed resource controllers include PV inverters, PCS, charging pile controllers, and aggregate managers configured on edge computing nodes; The hierarchical communication network is a hierarchical communication network connecting controllers at various levels. It is configured to automatically execute complementary resource evaluation, dynamic aggregation, and hierarchical collaborative control commands.

[0054] This embodiment also provides an electronic device applicable to a distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method proposed in the above embodiment.

[0055] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements a distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method as proposed in the above embodiments.

[0056] The storage medium proposed in this embodiment and the method for implementing a distributed resource dynamic heterogeneous aggregation and hierarchical inertia support proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0057] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0058] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for dynamic heterogeneous aggregation and hierarchical inertia support of distributed resources, characterized in that: include, Construct a unified state-space model for heterogeneous resources that integrates stress and uncertainty, and map all resources into a three-dimensional space; Define and calculate the complementarity index between distributed resources, and construct a real-time dynamic complementarity matrix; A dynamic heterogeneous aggregation mechanism based on the complementarity matrix is ​​constructed to divide physically dispersed resources into virtual inertia aggregates with maximized complementarity. Based on the virtual inertia aggregate generated by aggregation, a three-layer collaborative inertia support control is established and implemented.

2. The method for dynamic heterogeneous aggregation and hierarchical inertia support of distributed resources as described in claim 1, characterized in that: The heterogeneous resource unified state space model includes establishing a high-fidelity model that incorporates physical constraints, operating costs, and dynamic characteristics. The high-fidelity model integrates the MPPT unloaded potential energy model of distributed photovoltaics, the dynamic inertia stress model of energy storage systems, and the probabilistic robust capacity model of electric vehicle clusters. Map the state of all resource models to a unified three-dimensional feature space.

3. The method for dynamic heterogeneous aggregation and hierarchical inertia support of distributed resources as described in claim 2, characterized in that: The construction of the real-time dynamic complementarity matrix includes defining complementary characteristic indicators that describe the synergistic potential between different resources based on the established three-dimensional feature space. Based on the complementary characteristic index, the comprehensive complementarity between any two nodes in the entire network is calculated, and a real-time dynamic complementarity matrix is ​​constructed.

4. The method for dynamic heterogeneous aggregation and hierarchical inertia support of distributed resources as described in claim 3, characterized in that: The dynamic heterogeneous aggregation mechanism includes using a dynamic complementarity matrix as a similarity matrix to dynamically divide distributed resources in the physical network into separate virtual inertia aggregates and to implement aggregation principles. Set reconstruction trigger conditions to trigger the update of the complementarity matrix and spectral clustering reconstruction.

5. The method for dynamic heterogeneous aggregation and hierarchical inertia support of distributed resources as described in claim 4, characterized in that: The dynamic heterogeneous aggregation mechanism also includes using a dynamic complementarity matrix and a dynamic graph segmentation algorithm to reconstruct the physical network into a virtual inertia aggregate. Construct a graph Laplace matrix, treating the power grid as a graph. The weight of the edge is Calculate the degree matrix ,in Calculate the normalized Laplace matrix: ; Perform spectral embedding, for Perform eigenvalue decomposition and solve the characteristic equation. Select the eigenvectors corresponding to the K smallest non-zero eigenvalues. Constructing a matrix Each row of U is considered as the coordinate of the corresponding node in the K-dimensional feature space; Virtual inertia aggregates are generated using K-means clustering. K-means clustering is performed on the coordinates of N nodes in the feature space, dividing the entire network into K clusters. Each cluster is a virtual inertia aggregate; Setting the trigger conditions for dynamic refactoring includes timed triggering and event triggering; The timed trigger is every [time]. Recalculate the aggregation; The event is triggered when the power imbalance within a virtual inertia aggregate exceeds 80% of the corresponding adjustment capability, or when the system frequency deviation... At that time, a forced refactoring is triggered to find new complementary resources; in, In this context, V represents the set of nodes and E represents the set of edges. Let be the edge weight, representing the complementarity between node m and node n; For degree matrix, These are the diagonal elements of the degree matrix. For the normalized Laplace matrix, It is the identity matrix; This is the complementarity matrix. The time interval for triggering at regular intervals. This represents the absolute value of the system frequency deviation.

6. The method for dynamic heterogeneous aggregation and hierarchical inertia support of distributed resources as described in claim 5, characterized in that: The three-layer collaborative inertia support control includes constructing a hierarchical control architecture, with different layers executing different control logic and time scales: The underlying layer is device-level, the time scale is millisecond-level, each distributed resource executes local adaptive droop control, introduces dynamic participation factors, and fine-tunes based on local real-time status; The middle layer is at the aggregate level, with a time scale of seconds. Each virtual inertia aggregate has an aggregate leader to collect the status of its members, solve the power allocation problem based on internal complementarity, and utilize the complementarity of heterogeneous resources to prioritize the use of low-cost, high-health resources to smooth out internal fluctuations. The upper layer is system-level, with a time scale of seconds to minutes. It adopts an improved hierarchical consensus algorithm to interact with the marginal inertia cost and net power deficit of all aggregates. Through iteration, it eliminates the power imbalance across the entire network and achieves convergence of marginal costs among all aggregates, thus achieving global economic optimization.

7. The method for dynamic heterogeneous aggregation and hierarchical inertia support of distributed resources as described in claim 6, characterized in that: The underlying layer includes device-level adaptive droop control, where each device i executes a control law: Among them, dynamic parameters and It is issued in real time by the middle layer controller: in, The command is for the real-time power reference value of device i. To provide initial foundation power for device i, This is the dynamic droop coefficient. For dynamic virtual inertia coefficient, As a participating factor, This refers to the rated adjustment coefficient of the equipment. The middle layer includes aggregate-level complementary optimization, where each virtual inertia aggregate selects an aggregate leader node. The aggregate leader collects information about members within the cluster and solves the optimization problem. Using the Lagrange multiplier method, the aggregate leader calculates the current uniform incremental rate of the aggregate: in, A collection of aggregates Let i be the cost function of device i. The amount of power adjustment shared by device i This represents the total power deficit that needs to be balanced within the k-th polymer. / Let i be the minimum / maximum adjustable power constraint for device i at the current moment. The uniform incremental rate of the k-th aggregate; The upper layer includes system-level hierarchical consistency coordination, where the leaders of each virtual inertia aggregate constitute an upper-layer sparse communication network to execute an improved consensus algorithm. in, Let this be the marginal cost state of aggregate k at the next iteration. Let k be the set of neighboring aggregates that are directly connected to aggregate k in the upper-layer communication network. The communication weighting coefficient between aggregates k and j. This represents the iteration step size of the algorithm. / The total power generation and total load power within polymer k are given. The power requirements are to support the macroscopic inertia allocated to the aggregate.

8. A distributed resource dynamic heterogeneous aggregation and hierarchical inertia support system, applying the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method as described in any one of claims 1 to 7, characterized in that, include: Model building module, complementary module, aggregation module, support module, multi-type distributed resource controller, and hierarchical communication network; The model building module constructs a unified state space model of heterogeneous resources that integrates stress and uncertainty, mapping all resources into a three-dimensional space. The complementary module defines and calculates the complementary characteristic index between distributed resources, and constructs a real-time dynamic complementarity matrix. The aggregation module constructs a dynamic heterogeneous aggregation mechanism based on the complementarity matrix, which divides physically dispersed resources into virtual inertia aggregates with maximized complementarity. The support module establishes and implements a three-layer collaborative inertia support control based on the virtual inertia aggregate generated by aggregation. The multi-type distributed resource controller includes PV inverters, PCS, charging pile controllers, and an aggregate manager configured on the edge computing node; The hierarchical communication network is a hierarchical communication network connecting controllers at each level, and is configured to automatically execute complementary resource evaluation, dynamic aggregation, and hierarchical collaborative control commands.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the distributed resource dynamic heterogeneous aggregation and hierarchical inertia support method according to any one of claims 1 to 7.

Images