A consensus control method, system, and storage medium for distributed marginal cost in microgrids based on intelligent leaders.

Abstract

This application discloses a consensus control method, system, and storage medium for distributed marginal cost in microgrids based on an intelligent leader. The method includes: establishing an economic dispatch model with the goal of minimizing total operating cost, deriving the optimal conditions under which the marginal costs of all distributed power sources must converge to the consensus target value; constructing a communication topology containing an intelligent leader and followers, designing a distributed consensus control law including two types of agents, with the intelligent leader employing a threshold saturation type or a neighbor feedback type control law; determining convergence parameter conditions and communication topology requirements based on Lyapunov theory; and having each agent execute the control law in parallel to drive the marginal cost to converge to the consensus target value. This application solves the problems of slow response and oscillation caused by a static leader, improving the convergence speed, stability, and operating efficiency of distributed economic dispatch in microgrids.

Description

Technical Field

[0001] This application relates to distributed control technology for microgrids, specifically to a distributed marginal cost consensus control method, system, and storage medium for microgrids based on intelligent leaders. Background Technology

[0002] Microgrids integrate distributed power sources, energy storage systems, flexible loads, and intelligent control units, serving as a crucial means to improve power supply reliability and promote the consumption of renewable energy. Economic dispatch is one of the core issues in microgrid operation, aiming to minimize total generation or operating costs while meeting power balance constraints. Compared to centralized dispatch relying on a central controller, distributed economic dispatch schemes, with their advantages of strong resistance to single points of failure, no need for a central node, and good scalability and robustness, have become a research hotspot for microgrids with multiple distributed power sources.

[0003] In distributed economic dispatch, the marginal cost consensus method derives the optimal solution condition based on Lagrange multiplier theory, namely, that the marginal costs of all distributed power sources are equal and converge to the common value corresponding to the global optimal solution. This method uses a distributed algorithm to allow each agent to gradually adjust its state by relying solely on neighbor information interaction, ultimately achieving the economic operation goal. Although existing research has made progress in improving convergence speed, handling communication constraints, and multi-objective optimization, the mainstream leader-follower consensus architecture still generally adopts a static leader mechanism with a fixed setpoint. This mechanism has significant limitations: the static leader cannot perceive changes in the consensus state of the follower group in real time, nor can it dynamically adjust its own instructions based on state differences. When the microgrid starts up or operating conditions change abruptly, the initial marginal cost state of the followers often deviates significantly from the leader's setpoint, leading to slow dynamic response of the system, prolonged convergence process, and significant overshoot or continuous oscillations in the transient phase. Such problems not only reduce dispatch efficiency but may also cause power quality fluctuations and even threaten system stability, seriously hindering further optimization of the microgrid's dynamic performance. Therefore, the static leader mechanism is unable to effectively perceive and guide the consensus process, becoming a key bottleneck restricting the practical application of distributed economic dispatch technology. Summary of the Invention

[0004] The purpose of this application is to provide a microgrid distributed marginal cost consensus control method, system, and computer-readable storage medium based on intelligent leaders, which can improve the convergence speed and stability of microgrid distributed marginal cost consensus control and reduce overshoot and oscillation phenomena.

[0005] This application provides a distributed marginal cost consensus control method for microgrids based on intelligent leaders, including the following steps: Step S1: Establish a microgrid economic dispatch model, taking the minimization of the total operating cost of the microgrid as the objective function. Considering power balance constraints and the quadratic convex function generation cost characteristics of each distributed power source, the first necessary condition for the optimal solution is derived through the Lagrange multiplier method: the marginal cost of all distributed power sources is equal and equal to the consensus target value. ; Step S2: Construct the communication topology and define a communication graph G consisting of intelligent leaders and followers. If follower i can receive information from the leader, then the leader's adjacency matrix elements are... ,otherwise ; Step S3: Design a distributed marginal cost consensus control law, which includes a follower agent control law or an intelligent leader control law. The follower agent control law enables each follower to update its own marginal cost state based on the difference between its marginal cost state and that of its neighboring followers, as well as the difference between its marginal cost state and that of the intelligent leader. The intelligent leader control law enables the intelligent leader's marginal cost state to be dynamically adjusted according to the system state, thereby guiding the marginal cost states of all followers to converge towards the consensus target value. Step S4: Based on Lyapunov theory, satisfy the parameter conditions for system convergence. and communication topology constraints, wherein the communication topology requires that the communication graph G contains a directed spanning tree rooted at the leader; Step S5: Each intelligent agent executes the distributed marginal cost consensus control law in parallel, driving the marginal cost of all followers. and the state of intelligent leaders Converging to the consensus target value .

[0006] Furthermore, in step S1, the consensus target value The generation cost characteristic, represented by the quadratic convex function, corresponds to the generation cost of distributed generation, where the grid electricity price or optimal dispatch value is the power price.

[0007] Furthermore, in step S2, the communication graph G is a directed or undirected graph corresponding to a wired or wireless network, and the leader adjacency matrix... N is the total number of followers.

[0008] Furthermore, in step S3, the control law of the follower agent is: , ; in, For the state of follower i, This is the current state. For the control input signal of the follower agent. The weight parameters are the communication topology between follower agents. Identification parameters for the communication topology between intelligent leaders and followers. To control the gain, Gather for the neighbors, For intelligent leaders at all times state, is the discrete time step.

[0009] Furthermore, in step S3, the intelligent leader's control law adopts a threshold saturation type control law: , ; in The state error of the intelligent leader. The gain of the threshold saturation control law. The saturation threshold, For discrete time steps, Control input signals for intelligent leaders; Alternatively, the intelligent leader's control law may employ a neighbor feedback-based control law: , ; in, A collection of neighboring followers for intelligent leaders. For neighbor feedback control laws, the weight parameters are... is the discrete time step.

[0010] Furthermore, in a threshold saturation control law, when | |> At that time, the intelligent leader approaches the consensus target value at a fixed rate. When | |≤ At that time, the intelligent leader adjusts the state in a linear manner.

[0011] Furthermore, the neighbor feedback control law dynamically adjusts the state differences between the intelligent leader and the neighbor followers, enabling the intelligent leader to proactively close the distance with the neighbor followers and accelerate the convergence of marginal cost consensus.

[0012] Furthermore, in step S4, the parameter conditions... The marginal cost used to drive all followers is derived through Lyapunov stability analysis. and the state of intelligent leaders Converging to the consensus target value .

[0013] Furthermore, in step S5, each agent exchanges local marginal cost information according to the topology of the communication graph G. The intelligent leader only interacts with its direct neighbor followers and does not need to obtain global information.

[0014] This application also provides a microgrid distributed marginal cost consensus control system based on intelligent leaders, capable of running the aforementioned microgrid distributed marginal cost consensus control method based on intelligent leaders, including: The physical layer includes distributed power sources, energy storage devices, loads, and a point of common coupling (PCC). The physical layer is used to realize the production, storage, consumption, and connection to the main power grid. The communication network layer adopts a wired or wireless network of the communication graph G, and the communication network layer is used for information interaction between intelligent leaders and followers, and between followers; The control layer, which is the agent software deployed on the controller, includes an intelligent leader controller and a follower controller. The intelligent leader controller executes the intelligent leader control law, and the follower controller executes the follower agent control law. Monitoring and management layers are used to configure parameters. , , , The system also displays the operating status, which includes the marginal cost of each distributed power source, the status of the intelligent leader, and the consensus error index J.

[0015] Furthermore, the physical carrier of the intelligent leader controller is the energy router ER, which is responsible for power exchange between the microgrid and the main grid, and the follower controller is deployed in the local controller of each distributed power source; The monitoring and management system supports real-time parameter adjustment and can select the type of intelligent leader control law and configure the corresponding parameter values ​​according to the microgrid operation scenario, communication conditions, and performance requirements.

[0016] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described microgrid distributed marginal cost consensus control method based on intelligent leaders.

[0017] Compared with the prior art, this application has the following beneficial effects: This application provides a microgrid distributed marginal cost consensus control method, system, and computer-readable storage medium based on intelligent leader. By designing the control law of the intelligent leader, it drives all intelligent agents to converge to the consensus target value, which solves the problem of slow dynamic response and oscillation caused by the static leader mechanism. It has the advantages of improving the efficiency and reliability of distributed economic dispatch of microgrids. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a microgrid distributed marginal cost consensus control method based on intelligent leaders, provided in an embodiment of this application. Figure 2 This is a schematic diagram of an applicable 7-node grid-connected microgrid structure provided in the embodiments of this application; Figure 3 This is a schematic diagram of a typical communication topology between intelligent agents in a microgrid provided in the embodiments of this application; Figure 4 This is the marginal cost evolution curve of each agent when the intelligent leader control law provided in this application adopts a threshold saturation type control law; Figure 5 This is the control input evolution curve of the intelligent leader when the intelligent leader control law provided in this application adopts a threshold saturation type control law; Figure 6 This is the evolution curve of the system consensus error index when the intelligent leader control law provided in this application adopts a threshold saturation type control law; Figure 7 This is the marginal cost evolution curve of each intelligent agent when the intelligent leader control law provided in this application adopts the neighbor feedback type control law; Figure 8 This is the control input evolution curve of the intelligent leader when the intelligent leader control law provided in this application adopts the neighbor feedback type control law; Figure 9 This is the evolution curve of the system consensus error index when the intelligent leader control law provided in this application adopts a neighbor feedback control law; Figure 10 This is a schematic diagram of a microgrid distributed marginal cost consensus control system based on intelligent leaders, provided in an embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this application clearer, specific embodiments of this application will be described in further detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely for explaining this application and not for limiting it. It should also be noted that, for ease of description, only the parts relevant to this application are shown in the drawings, not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe operations (or steps) as being processed sequentially, many of these operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. A process can be terminated when its operation is completed, but it may also have additional steps not included in the drawings. A process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0020] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0021] For ease of understanding, the terminology used in this embodiment is explained below: Microgrid It integrates distributed power sources, energy storage systems, flexible loads, and intelligent control units, and is an important way to improve power supply reliability and promote the consumption of renewable energy.

[0022] Marginal Cost Consensus A mainstream distributed collaborative control method for achieving economic dispatch of microgrids is based on the idea of ​​using distributed algorithms to make the marginal costs of all distributed power sources in the system tend to be consistent and eventually converge to the common value (such as the grid electricity price) corresponding to the global optimal solution, thereby achieving economic operation.

[0023] Smart Leader The control concept proposed in this invention refers to an intelligent leader capable of dynamically adjusting its own output or control law based on the state information of its neighboring followers. The intelligent leader agent is associated with the grid connection point of the microgrid or a virtual reference, and its target marginal cost is the grid electricity price or the optimal scheduling value. Its marginal cost state... It is no longer a constant value, but is dynamically adjusted based on the state of its neighboring followers.

[0024] Static Leader In existing technologies, leaders (usually virtual commands or intelligent agents representing power grid interfaces) often employ fixed, unchanging settings. Such static leaders cannot dynamically adjust their commands based on the real-time consensus state of their followers.

[0025] consensus target value The first necessary condition for the optimal solution is derived through the Lagrange multiplier method, namely that the marginal costs of all distributed power sources are equal and equal to a common value (the optimal Lagrange multiplier). This common value is defined as the consensus target value (such as the grid electricity price).

[0026] Communication topology A directed or undirected graph of communication consisting of intelligent leader agents and follower agents. Each agent collaborates by exchanging its local marginal cost information with neighboring agents.

[0027] Laplace matrix A matrix used to describe the communication topology.

[0028] Leader's adjacency matrix A matrix representing the connection between leaders and followers. If the followers i If they can receive information from the leader, then ,otherwise .

[0029] In existing microgrid distributed economic dispatch schemes, the leader in the leader-follower marginal cost consensus scheme often uses a fixed setpoint. This static leader cannot dynamically adjust its own instructions according to the real-time consensus state of the follower group. When the initial state of the followers differs significantly from the leader's setpoint, it can lead to problems such as slow dynamic response, long convergence time, and overshoot or oscillation in the transient process, thus limiting further improvement in the dynamic performance of the system.

[0030] In this regard, such as Figure 1As shown, this application proposes a consensus control method for distributed marginal costs in microgrids based on intelligent leaders. This method aims to guide the marginal costs of each distributed power source in the microgrid to converge quickly and stably to a consensus target value through a dynamically adjusted intelligent leader, thereby achieving economical operation of the microgrid. The method includes the following steps: Step S1: Establish a microgrid economic dispatch model, taking the minimization of the total operating cost of the microgrid as the objective function. Considering power balance constraints and the quadratic convex function generation cost characteristics of each distributed power source, the first necessary condition for the optimal solution is derived through the Lagrange multiplier method: the marginal cost of all distributed power sources is equal and equal to the consensus target value. ; Step S2: Construct the communication topology and define a communication graph G consisting of intelligent leaders and followers. If follower i can receive information from the leader, then the leader's adjacency matrix elements are... ,otherwise ; Step S3: Design a distributed marginal cost consensus control law, which includes a follower agent control law or an intelligent leader control law. The follower agent control law enables each follower to update its own marginal cost state based on the difference between its marginal cost state and that of its neighboring followers, as well as the difference between its marginal cost state and that of the intelligent leader. The intelligent leader control law enables the intelligent leader's marginal cost state to be dynamically adjusted according to the system state, thereby guiding the marginal cost states of all followers to converge towards the consensus target value. Step S4: Based on Lyapunov theory, satisfy the parameter conditions for system convergence. and communication topology constraints, wherein the communication topology requires that the communication graph G contains a directed spanning tree rooted at the leader; Step S5: Each intelligent agent executes the distributed marginal cost consensus control law in parallel, driving the marginal cost of all followers. and the state of intelligent leaders Converging to the consensus target value .

[0031] Specifically, the control method of this embodiment can be implemented in the following manner: In step S1, the microgrid economic dispatch model can be established using traditional optimization theory. For example, by manually setting or pre-setting a mathematical model, minimizing the total operating cost of the microgrid can be used as the objective function. When considering power balance constraints and the generation cost characteristics of each distributed power source, a simplified linear cost model or a piecewise linear cost model can be used. Subsequently, through analytical methods or numerical calculations, it is derived that under optimal operating conditions, the marginal costs of all distributed power sources should be equal and equal to a pre-set consensus target value. .

[0032] In step S2, the construction of the communication topology can be based on pre-planned physical connections. For example, a communication network can be formed by establishing fixed wired connections between the intelligent leader and some followers, as well as among the followers themselves. The definition of the communication graph G can be manually drawn or configured by the system designer according to the actual connection situation. Leader adjacency matrix elements The value of can be manually determined and set based on whether the follower i has a direct communication link with the intelligent leader.

[0033] In step S3, the design of the distributed marginal cost consensus control law can be based on traditional control theory. For example, proportional control or integral control can be used to formulate state update rules for both the follower agent and the intelligent leader. These control laws aim to gradually adjust their respective marginal costs through information interaction between agents, making them tend to be consistent. This distributed marginal cost consensus control law includes a follower agent control law and an intelligent leader control law.

[0034] The implementation of the control law for intelligent leaders can employ various strategies. For example, it can be based on the intelligent leader's own state and the consensus target value. The system can adjust its state based on deviations from the target value, or based on the state differences between the intelligent leader and its neighboring followers. Specifically, a threshold-saturation control strategy can be used, where the leader adjusts at a fixed rate when the deviation is large, and at a gentler pace when the deviation is small. Alternatively, a neighbor feedback control strategy can be used, where the intelligent leader dynamically adjusts its state based on the state information of its direct neighboring followers to guide the entire group towards the consensus target value. convergence.

[0035] In step S4, the parameter conditions that guarantee system convergence are determined. The determination of parameters can be based on system simulation or empirical adjustments. For example, by repeatedly running the simulation model and observing the convergence behavior of the system under different parameter combinations, a set of parameter values ​​that enable the system to converge stably can be selected. The communication topology requirement, that is, the communication graph G contains a directed spanning tree rooted at the leader, can serve as a basic principle in system design, ensuring that the leader can transmit information to all followers and ultimately guide the entire system to achieve consensus.

[0036] In step S5, each agent executes the distributed marginal cost consensus control law in parallel, meaning that each follower agent and the intelligent leader independently runs their control algorithm on their respective controllers. Through this parallel computation and information exchange, the marginal cost of all followers is... and the state of intelligent leaders It will be driven, gradually moving towards the preset consensus target value. By getting closer together, the distributed marginal cost consensus control of the entire microgrid can be achieved, thereby achieving the goal of economic operation.

[0037] This embodiment introduces an intelligent leader, which can dynamically adjust its own instructions based on the real-time consensus state of the follower group. This effectively solves the problems of slow dynamic response, long convergence time, and overshoot or oscillation in the transient process of traditional static leader schemes. As a result, the marginal cost of distributed power sources in the microgrid can converge to the consensus target value quickly and smoothly, improving the dynamic performance and operational efficiency of the microgrid's economic dispatch, and ensuring the stability and robustness of the system in complex and variable operating environments.

[0038] In some implementations, in step S1, the consensus target value This could be the grid electricity price or the optimal dispatch value. Consensus target value. This is the target value to which the marginal cost of all distributed power sources must ultimately converge. When a microgrid exchanges power with the main grid, the real-time electricity price of the main grid can be used as the consensus target value. This means that distributed power sources within a microgrid will adjust their marginal costs based on external electricity prices to achieve economic interaction with the main grid. For example, they might reduce their purchases from the main grid when prices are high and sell electricity to the main grid when prices are low. In scenarios where the microgrid operates independently or seeks optimal internal economics, the consensus target value... This can be pre-calculated using more sophisticated optimization algorithms (e.g., global optimization based on predicted load and generation), representing the marginal cost level that minimizes the overall operating cost of the microgrid under specific operating conditions. This approach ensures the optimal allocation and utilization of resources within the microgrid.

[0039] Furthermore, this application clarifies that the generation cost characteristic of the quadratic convex function corresponds to the generation cost C of distributed power sources. i C i =α i P Gi ²+β i P Gi +γ i , where α i β i γ i P is the cost coefficient. Gi Let P be the power output of distributed power source i. This cost model, in the form of a quadratic convex function, is a widely accepted approach that effectively reflects the actual power generation cost as a function of power P. Gi A changing mathematical model. Where α i β i γ iIt is determined based on factors such as the type of specific distributed power source (e.g., gas turbine, diesel generator), fuel cost, and maintenance cost, through historical data fitting or data provided by the manufacturer. α_i is usually positive to ensure the cost function is convex, thus guaranteeing a unique optimal solution to the economic scheduling problem. This specific functional form allows for a clear and operable marginal cost expression, dC, to be obtained in step S1 when deriving the first-order necessary conditions for the optimal solution using the Lagrange multiplier method. i / dP Gi =2α i P Gi + i .

[0040] The consensus target value is clearly defined through the above implementation methods. This can be either the grid electricity price or the optimal dispatch value. This application provides a flexible strategy for the economic dispatch of microgrids that is adaptable to different operating scenarios. When set to grid tariffs, microgrids can effectively interact economically with the main grid, optimize power purchase and sale strategies, and thus reduce the cost of interacting with the main grid. When set to the optimal scheduling value, it ensures that the microgrid achieves the best allocation and utilization of its internal resources under independent operation or specific objectives, further improving the overall economic efficiency of the microgrid. Simultaneously, the generation cost characteristics of each distributed power source are specifically modeled as a quadratic convex function C. i =α i P Gi ²+β i P Gi +γ i This application provides a precise mathematical foundation for the economic dispatch model of microgrids. This precise cost function form allows for the accurate derivation of the first necessary condition for the optimal solution in step S1 using the Lagrange multiplier method: the marginal costs of all distributed power sources are equal and equal to the consensus target value. This not only simplifies the calculation process for optimal power allocation, but also ensures that the obtained scheduling results can truly reflect the actual operating costs of each distributed power source, thereby minimizing the total operating cost of the microgrid and improving the accuracy and reliability of economic dispatch.

[0041] In some implementations, the specific form of the communication graph G and the structure of the leader adjacency matrix B are described. Specifically, the communication graph G can be a directed graph or an undirected graph adapted to wired or wireless networks. Wired networks typically use physical media (e.g., optical fibers, Ethernet cables) for connection, offering high transmission rates, good stability, and strong anti-interference capabilities, making them suitable for scenarios with high communication reliability requirements. Wireless networks transmit information via radio waves (e.g., Wi-Fi, ZigBee, LoRa, 5G, etc.), offering more flexible deployment and relatively lower costs, making them particularly suitable for geographically dispersed distributed power sources. When the communication graph G is directed, it indicates that the information flow is unidirectional; for example, an intelligent leader can broadcast information to followers, but followers do not directly provide feedback to the leader, reflecting the unidirectional characteristics of some practical communication links. When the communication graph G is undirected, it indicates that the information flow is bidirectional, meaning that agents can send and receive information from each other, typically corresponding to a bidirectional communication link.

[0042] Meanwhile, the leader adjacency matrix B is defined as a diagonal matrix. , where N is the total number of followers. This means that the leader adjacency matrix B is an N×N matrix whose off-diagonal elements are all zero, and contains elements only on the diagonal. According to the definition in step S2 above, if follower i can receive information from the intelligent leader, then the corresponding diagonal element b i Greater than zero; otherwise, b i It equals zero. This diagonal matrix structure clearly represents the independent and direct information reception relationship between the intelligent leader and each follower, avoiding the modeling and analysis difficulties caused by complex off-diagonal elements.

[0043] Through the above implementation methods, this approach can be flexibly applied to various practical microgrid communication infrastructures, effectively constructing information exchange paths for both physically stable wired networks and flexibly deployed wireless networks. Furthermore, defining the leader adjacency matrix B as a diagonal matrix clearly represents the independent and direct information reception relationship between the intelligent leader and each follower, simplifying the mathematical model of the control system and enabling the follower agent's control law to more accurately utilize leader information for state adjustment. This explicit definition of the communication topology and leader connection matrix enhances the universality, robustness, and feasibility of the distributed marginal cost consensus control method, ensuring that the marginal cost of all followers remains consistent across different communication environments. and the state of intelligent leaders It can converge stably and efficiently to the consensus target value. .

[0044] In some implementations, in step S3, the control law of the follower agent is: , ; in, To control the gain, Gather for the neighbors, For intelligent leaders at all times state, is the discrete time step.

[0045] The execution of the control law of the follower agent can be carried out according to a preset discrete time step. Perform iterative updates. At each time step k, the state of follower i... It will be based on its current status Control input and time step Update the system. This includes control input. The calculation can be based on the state difference between follower i and its neighbor follower j. and the state of followers and intelligent leaders. Differences Control gain It can be an empirically set fixed positive value used to adjust the response speed of the control law. Let i be the set of its neighbors. The state of the intelligent leader at time k.

[0046] In step S3, the intelligent leader's control law adopts a threshold saturation type control law: , ; in This represents the state error of the intelligent leader at time k, which defines the current state of the intelligent leader. With consensus target value The deviation between these values ​​is the basis for the decision-making process of the control law. This is the gain of the threshold saturation control law; this gain parameter is used to adjust the strength of the control action. A larger gain... A higher value generally means a faster response time, but may also increase the risk of system oscillations; a smaller value... A value of 0 will make the system response smoother. This parameter defines an error range, serving as the saturation threshold. The discrete time step represents the time interval between state updates and control variable calculations performed by the control system. Within each time step, the state of the intelligent leader... Based on the current control quantity Update accordingly, thereby gradually moving towards the consensus target value. Approaching. The threshold saturation control law is passed through The function implements segmented control. When the intelligent leader has a state error... Larger (when | |> At that time, control quantity The size is limited to , direction from The decision is made to drive the intelligent leader toward the consensus target value at a constant maximum rate. By being close to the target, the system avoids the potential instability caused by an infinitely large increase in control input due to excessive error. When the state error... Smaller (when | |≤ At that time, the control quantity u0(k) is related to the error. Proportional relationship, that is = At this point, the intelligent leader makes fine-tuning adjustments to smoothly converge to the consensus target value. .

[0047] Through the above implementation method, the intelligent leader control law can adaptively adjust the control strategy according to the magnitude of its state error. When the intelligent leader's state deviates from the consensus target value... When significant deviations exist, the threshold-saturation control law provides a robust yet constrained control effect, enabling the intelligent leader to converge to the target value quickly and effectively shortening the initial adjustment time of the system. This holds true when the intelligent leader's state approaches the consensus target value. At this time, the control law switches to a fine proportional control mode, avoiding oscillations or overshoot caused by excessive control, and ensuring that the intelligent leader can smoothly and accurately converge to the consensus target value. This segmented control strategy effectively balances convergence speed and system stability, improving the robustness and performance of distributed marginal cost consensus control in microgrids, and enabling it to exhibit good dynamic response characteristics when faced with errors of different magnitudes.

[0048] In some implementations, in the threshold saturation control law, when the state error of the intelligent leader... The absolute value is greater than the saturation threshold. At that time, the intelligent leader approaches the consensus target value at a fixed rate. When the state error of the intelligent leader The absolute value is less than or equal to the saturation threshold. In this case, the intelligent leader adjusts the state in a linear manner to avoid impacting the system.

[0049] Specifically, when the state error of the intelligent leader The absolute value is greater than the preset saturation threshold. At that time, in the threshold saturation control law The function will output At this point, the intelligent leader's control... Will be identified as This implies the state of intelligent leaders. It will proceed at a fixed rate Towards consensus target value Adjustment. This fixed-rate adjustment strategy aims to ensure that when the state of the intelligent leader deviates significantly from the consensus target value, it can move quickly and effectively towards the target value, thereby accelerating the overall convergence process of the system.

[0050] Furthermore, when the state error of the intelligent leader... The absolute value is less than or equal to the saturation threshold. At that time, in the threshold saturation control law The function will output At this point, the intelligent leader's control... Will be identified as ,Right now This implies the state of intelligent leaders. Will be with an error Proportional linear rate Towards consensus target value Adjustment. This linear adjustment method allows the adjustment rate to decrease smoothly as the error decreases when the state of the intelligent leader approaches the consensus target value, thus achieving fine-grained control.

[0051] Through the above implementation, the threshold-saturation control law of the intelligent leader is designed as a segmented adjustment strategy. When there is a large deviation between the state of the intelligent leader and the consensus target value, a fixed rate is used to quickly approach the target, ensuring that the system can respond rapidly and converge efficiently in the initial stage or when subjected to large disturbances. When the error decreases to within the saturation threshold, a linear adjustment mode is switched, making the adjustment rate proportional to the error magnitude, thereby achieving smooth and gradual convergence. This dual-mode adjustment mechanism effectively avoids oscillations or overshoots that may be caused by fixed-rate adjustments near the target value, ensuring that the state of the intelligent leader can converge quickly and smoothly to the consensus target value, thereby improving the stability and accuracy of the distributed marginal cost consensus control of the microgrid, while reducing the impact on the operation of the microgrid.

[0052] In some implementations, the intelligent leader control law employs a neighbor feedback control law: , ; in, A collection of neighboring followers for intelligent leaders. For neighbor feedback control laws, the weight parameters are... is the discrete time step.

[0053] The neighbor feedback control law This is a control strategy that dynamically adjusts based on the state differences between the intelligent leader and its direct neighbor followers. The control law calculates the state of the intelligent leader itself. Gathering with its neighbors and followers The state of each follower j The difference between them, multiplied by the weighting parameter. Accumulate these inputs to generate the control input for the intelligent leader. This approach enables intelligent leaders to proactively perceive and respond to the state of followers within their local communication range, rather than relying solely on their own errors or preset saturation thresholds.

[0054] The state update formula Describes the intelligent leader in discrete time steps The internal state update mechanism. The state of the intelligent leader at time k+1. It is determined by its state at time k Add control input Multiply by time step This means that the state of the intelligent leader is iteratively updated based on the input signals calculated by its control law, thereby gradually moving towards the consensus target value. near.

[0055] The Let be the set of neighboring followers of the intelligent leader, representing the set of follower agents that directly interact with the intelligent leader. The intelligent leader only exchanges information with these followers directly connected in the communication topology, without needing to obtain the state information of all followers in the entire microgrid. This local information interaction mechanism reduces the communication burden and improves the scalability and robustness of the system.

[0056] The weight parameter for the neighbor feedback control law is a positive value used to adjust the intensity of the intelligent leader's response to the state differences of its neighbor followers. The larger the value, the more sensitive the intelligent leader is to differences in the state of its neighbors, and the greater the magnitude of its state adjustment, which may accelerate convergence but may increase oscillations. A smaller value results in a smoother response, which may lead to a slower convergence speed but better system stability. The selection of this parameter requires comprehensive consideration of the system's convergence speed, stability, and robustness to noise.

[0057] Discrete time step refers to the time interval between system state updates. In digital control systems, continuous-time processes are discretized into a series of state updates at specific points in time. The choice of [specific parameter] affects the system's real-time performance and computational load. Smaller [specific parameter] It can simulate the behavior of continuous systems more accurately, but it increases the computational burden; larger Conversely, in practical applications, The choice requires a balance between control precision and computing resources.

[0058] By employing a neighbor feedback control law, the intelligent leader can proactively perceive and utilize the state information of its direct neighbor followers. This dynamic adjustment mechanism based on local differences allows the intelligent leader to actively close the state distance with its neighbor followers, rather than passively waiting for them to converge. This not only enhances the coupling between the intelligent leader and its followers and improves the efficiency of information exchange, but also significantly accelerates the marginal cost reduction for all followers. and the state of intelligent leaders Towards consensus target value The convergence process improves the overall performance and response speed of the distributed marginal cost consensus control of microgrids.

[0059] In some implementations, the neighbor feedback control law is dynamically adjusted based on the state differences between the intelligent leader and its neighbors, enabling the intelligent leader to proactively close the distance with its neighbors and accelerate marginal cost consensus convergence. Specifically, the intelligent leader uses its neighbor feedback control law... Continuously monitor and calculate its own status The state of its direct neighbor follower j The differences between them. These differences are the basis for driving the intelligent leader's state updates, ensuring that the leader can perceive inconsistencies within the local network. Dynamic adjustment means that this calculation and update are performed in real time, continuously correcting the leader's control output as the system state changes. This is to respond to the real-time fluctuations in the marginal cost of distributed generation in microgrids. Here, "actively closing the gap" means that the intelligent leader does not merely passively respond to the states of its neighbors, but rather, through the design of its control law, generates a clear trend that brings its own state closer to the average or target state of its neighbors. When there is a deviation between the intelligent leader's state and the states of its neighbors, the control law generates a control variable that causes the intelligent leader's state update direction to reduce this deviation, thereby "closing the gap" in the state space. This reflects the leader's active guiding role in the consensus process, rather than merely being an information transmitter. "Accelerating marginal cost consensus convergence" means that through the aforementioned active adjustment mechanism of the intelligent leader, the marginal costs of all followers and the intelligent leader can be significantly reduced. and Achieve consensus target value The time required. In distributed control systems, convergence speed is one of the key indicators for measuring control performance. An intelligent leader actively closing the distance with its neighboring followers means it can integrate and disseminate local information more quickly, while simultaneously moving towards the consensus goal faster. This drives the entire system to eliminate state differences more quickly, achieving rapid consensus at marginal cost.

[0060] Through the above implementation, the intelligent leader can utilize its neighbor feedback control law to dynamically adjust based on the state differences with its neighbors, actively closing the distance between itself and its followers. This proactive adjustment mechanism significantly enhances the intelligent leader's guiding role in the consensus process, enabling it to move beyond passively responding to local information and actively drive its own state towards that of its neighbors. This effectively aligns the consensus target value λ... It enables faster transmission to neighboring followers and drives down the marginal cost of all distributed power sources throughout the microgrid system. and the state of intelligent leaders Fast and efficient convergence. Therefore, this application improves the real-time performance and response speed of microgrid economic dispatch, ensuring that the system can quickly reach the optimal economic operating state under dynamically changing operating environments, and avoiding economic losses or system instability caused by slow convergence.

[0061] In some implementations, in step S4, the parameter conditions The marginal cost used to drive all followers is derived through Lyapunov stability analysis. and the state of intelligent leaders Converging to the consensus target value .

[0062] Specifically, parameter conditions These are key regulating factors in the distributed marginal cost consensus control law, directly affecting the speed, stability, and convergence characteristics of the system's dynamic response. Among them, The control gain of the control law for the follower agent. The gain of the threshold saturation control law. These are the weighting parameters for the neighbor feedback control law. The appropriate selection of these parameters is crucial for achieving the desired system performance. Lyapunov stability analysis is a rigorous mathematical method used to evaluate and guarantee the stability of dynamic systems. By constructing a suitable Lyapunov function and analyzing its rate of change along the system trajectory, sufficient conditions for the system to remain stable or asymptotically converge to an equilibrium point can be derived. In this application, this method is used to systematically determine the range or specific values ​​of the control parameters to ensure the overall stability of the microgrid distributed marginal cost consensus control system. This guarantees all... and It can converge quickly and smoothly to Fast convergence refers to the system state reaching the consensus target value within a relatively short period of time. This is of great significance for the real-time economic dispatch of microgrids. Smooth convergence means that the system state will not experience significant oscillations or overshoots during the convergence process, thus avoiding adverse effects on the stable operation of the microgrid. The parameter conditions determined through Lyapunov stability analysis aim to simultaneously satisfy these two performance indicators and optimize the system's dynamic response.

[0063] The parameter conditions are systematically derived and determined using Lyapunov stability analysis. This application provides a solid theoretical foundation for parameter configuration of the distributed marginal cost consensus control law. This method avoids the limitations of traditional empirical or trial-and-error parameter adjustments, thus enabling precise design of control parameters to ensure that the marginal cost of all followers in the entire microgrid distributed control system remains constant when facing dynamic changes. and the state of intelligent leaders It can converge to the consensus target value with the expected speed and stability. This not only significantly improves the system's convergence efficiency and shortens the time required to reach the optimal economic dispatch state, but also enhances the stability and reliability of microgrid operation by effectively suppressing oscillations and overshoots during the convergence process, thereby achieving a more optimized economic operation effect.

[0064] In some implementations, in step S5, each agent exchanges local marginal cost information according to the topology of the communication graph G, and the intelligent leader only interacts with its direct neighbor followers without needing to obtain global information.

[0065] Specifically, each agent, including the leader and followers, exchanges local marginal cost information based on the connections defined in a pre-constructed communication graph G. The communication graph G can be directed or undirected, and its topology determines which agents can directly exchange information. This information exchange forms the basis of distributed consensus control, ensuring that each agent can perceive the state of its neighbors and adjust its own marginal cost accordingly. For example, each follower i receives information from its set of neighbors. Marginal cost information of other followers j in the process and from intelligent leaders Information (if b) i >0). Based on this, when an intelligent leader executes its control law, its information interaction scope is limited to the set of its directly connected neighbor followers. Within this localized information exchange model, the intelligent leader does not need to establish communication links with all its followers, nor does it need to receive information from non-direct neighbor followers. This localized information exchange model significantly reduces the communication complexity and computational burden of the intelligent leader. For example, the intelligent leader can periodically send its current state to its direct neighbor followers. And receive marginal cost information from these direct neighbor followers. Furthermore, the intelligent leader does not need to acquire marginal cost information from all followers in the microgrid or any other global system state information throughout the control process. Its control decisions rely solely on its own state and local information interactions with its immediate neighbor followers. This design avoids the single point of failure risk and large-scale data processing challenges common in centralized control, making the system more scalable and robust.

[0066] Through the above implementation, the intelligent leader interacts only with its direct neighbor followers and does not need to acquire global information, thus significantly reducing the communication burden and computational complexity of the intelligent leader. This localized information interaction mode makes the microgrid distributed marginal cost consensus control method more scalable when facing large-scale microgrid systems. Since the intelligent leader no longer needs to process and store large amounts of global information, the system's real-time response capability is improved, while reducing the demand for communication bandwidth. In addition, this design enhances the system's robustness; even if some non-neighbor followers experience communication failures, it will not directly affect the intelligent leader's decisions, thereby ensuring the stability and efficiency of the consensus control process.

[0067] The above embodiments are illustrated below with specific examples; see details below. Figures 2-10 .

[0068] refer to Figure 2This is a 7-bus microgrid system comprising four distributed power sources, several loads, and connected to the main grid via a common coupling point. The grid electricity price is set to a constant. Currency unit / MWh.

[0069] refer to Figure 3 Design a communication topology for this microgrid. Assume the energy router is the intelligent leader (node ​​0), and the four DG controllers (DG, Distributed Generator) are followers (nodes 1-4). The communication links are as follows: Figure 3 As shown, a directed spanning tree is formed with the leader as the root, and the leader can receive information from node 1 and node 2 (i.e., ).

[0070] The first intelligent leader control law (threshold saturation type) is adopted. Parameter settings: control gain. , saturation threshold Discrete time step Initial condition: Leader followers , , , Run a distributed control algorithm. Each agent exchanges information according to the communication topology. The value is determined, and its own state is updated based on the aforementioned control law.

[0071] Simulation results are as follows Figures 4-6 As shown. Figure 4 The display shows that all marginal costs converge to the target value of 50 within approximately 10 seconds, and the dynamic process is smooth. Figure 5 Leader control input was displayed. The saturation characteristics. Figure 6 Display consensus error index Rapid decline.

[0072] In the same microgrid system ( Figure 2 ), communication topology ( Figure 3 Under the given conditions and initial conditions, the second intelligent leader control law (neighbor feedback type) is adopted. Parameter settings are as follows: , , The leader can receive information from both node 1 and node 2, therefore... Run the control algorithm.

[0073] Simulation results are as follows Figure 7-9 As shown. Figure 7 The results show that all marginal costs converge to the target value rapidly within approximately 3 seconds, a convergence speed significantly faster than in Example 1. Figure 8 Leader control input was displayed. The dynamic changes. Figure 9 Display consensus error index It decays at a faster rate.

[0074] The table below compares the system consensus convergence performance under two different control schemes provided in this application: threshold saturation control law and neighbor feedback control law. The performance comparison is shown in the table below, with consensus error as the metric. The time required to drop to a certain threshold Compared to the percentage acceleration relative to the static leader, both intelligent leader schemes can effectively accelerate consensus, with the second scheme being particularly effective.

[0075] It should be noted that the method of this embodiment can be executed by a single device, such as a computer or server. The method of this embodiment can also be applied to a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method of this embodiment, and the multiple devices will interact with each other to complete the method described.

[0076] It should be noted that the above description describes some embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0077] like Figure 10 As shown, based on the same inventive concept and corresponding to any of the above embodiments, this application also discloses a microgrid distributed marginal cost consensus control system based on intelligent leaders. This system is capable of executing the above-described microgrid distributed marginal cost consensus control method based on intelligent leaders. The system includes a physical layer, a communication network layer, a control layer, and a monitoring and management layer.

[0078] The physical layer is the foundation of the microgrid system, responsible for the actual flow of electrical energy. This layer includes distributed power sources, such as photovoltaics, wind turbines, and fuel cells, for generating electricity; energy storage devices, such as battery storage systems and supercapacitors, for storing and releasing electrical energy to balance supply and demand or provide ancillary services; loads, i.e., electrical devices within the microgrid, for consuming electrical energy; and the Point of Common Coupling (PCC), which serves as the interface between the microgrid and the external main grid, enabling bidirectional energy exchange. This physical layer ensures the microgrid's energy self-sufficiency and interconnectivity with the main grid.

[0079] The communication network layer provides a channel for information exchange among the various agents in the microgrid. This layer adopts a wired or wireless network structure defined by the communication graph G, such as Ethernet, fiber optic networks, Wi-Fi, Zigbee, 4G / 5G, etc. The main function of this communication network layer is to realize information interaction between intelligent leaders and followers, as well as among follower agents, including but not limited to local marginal cost information of each distributed power source, state information of the intelligent leader, control commands, etc. Reliable and efficient communication is the key to the implementation of distributed consensus control algorithms.

[0080] The control layer is the core intelligent component of the distributed marginal cost consensus control in a microgrid. This layer is deployed in the intelligent agent software within the controller, specifically including an intelligent leader controller and follower controllers. The intelligent leader controller is responsible for executing the intelligent leader control law, adjusting its own state according to a preset threshold saturation control law or a neighbor feedback control law, guiding the entire system to converge towards the consensus target value. The follower controller executes the follower intelligent agent control law, adjusting its marginal cost based on its own state, neighbor information, and intelligent leader information to achieve consensus with the intelligent leader and other followers. This layer implements the computation and decision-making of the distributed control algorithm through software logic.

[0081] The monitoring and management layer provides a human-machine interface and advanced management functions for the operation of the microgrid. This layer is used to configure key parameters required for system operation, such as consensus target values. Control gain Gain of threshold saturation control law The saturation threshold ϵ and the weight parameters of the neighbor feedback control law. In addition, this layer is also responsible for displaying the real-time operating status of the system, including the marginal cost of each distributed power source, the status of the intelligent leader, and the consensus error index J, so that operators can have a comprehensive understanding of the microgrid's operating status, perform performance evaluation, and diagnose faults.

[0082] Through the aforementioned system architecture, this application provides a comprehensive and operable implementation platform for a microgrid distributed marginal cost consensus control method based on intelligent leaders. The physical layer provides the foundation for the production, storage, and consumption of electrical energy; the communication network layer ensures the information flow required for distributed control; the control layer transforms abstract control algorithms into actual execution instructions; and the monitoring and management layer provides the ability to configure, monitor, and optimize the entire system. This layered and integrated design effectively addresses the challenges faced by microgrid distributed control methods in practical applications, such as physical support, information interaction, algorithm execution, and operation management. It ensures that the microgrid can efficiently and stably achieve economic dispatch objectives and can be flexibly adjusted and optimized according to actual operational needs.

[0083] In some implementations, the physical carrier of the intelligent leader controller is an energy router (ER), which is responsible for power exchange between the microgrid and the main grid, and the follower controller is deployed in the local controller of each distributed power source.

[0084] Specifically, the Energy Router (ER) is an intelligent device integrating power electronic conversion, energy management, and information communication functions. Its core function is to achieve flexible energy scheduling and management within the microgrid and between the microgrid and the external power grid. Integrating the intelligent leader controller into the Energy Router (ER) means that the intelligent leader's control logic can run directly on physical hardware with powerful computing and communication capabilities, thus providing a stable execution platform for complex distributed marginal cost consensus control algorithms. This integration allows control commands to act more directly and quickly on the energy flow of the microgrid, improving the real-time performance and effectiveness of control. Simultaneously, as a key interface between the microgrid and the main grid, the Energy Router (ER) undertakes the management responsibility for bidirectional power flow. It can flexibly adjust the power transmitted from the microgrid to the main grid or obtained from the main grid based on the microgrid's operating status, the main grid's electricity price signal, and the instructions of the intelligent leader controller. This power exchange function is an important guarantee for the economical operation and stability of the microgrid. Through the coordination of the Energy Router (ER), the overall operating efficiency of the microgrid can be optimized, and the needs of the main grid can be responded to. Furthermore, the local controller of a distributed power source is a control unit directly connected to the distributed power source (such as a photovoltaic inverter or energy storage converter), responsible for real-time monitoring and adjustment of parameters such as output power, voltage, and frequency. Deploying follower controllers within these local controllers allows each distributed power source to independently execute its marginal cost consensus control law and update its state based on local information and information received from its neighbors. This distributed deployment reduces dependence on a central controller, improves system robustness and scalability, and enables rapid and precise control of distributed power sources.

[0085] Through the above implementation method, the intelligent leader controller is integrated into the energy router (ER), which is responsible for power exchange between the microgrid and the main grid, achieving a tight integration of control logic and physical execution. This enables the intelligent leader to directly and efficiently coordinate energy flow between the microgrid and the main grid, thereby optimizing the overall economic operation of the microgrid and its interaction with the main grid while performing marginal cost consensus control. Simultaneously, deploying follower controllers in the local controllers of each distributed power source allows each distributed power source to independently adjust its marginal cost based on local and neighbor information, enhancing the system's distributed characteristics and response speed. This physically-level distributed deployment not only improves the real-time performance and reliability of the control system but also reduces communication latency through proximity control, ensuring the marginal cost of all followers. and the state of intelligent leaders It can converge to the consensus target value quickly and smoothly. This effectively improves the efficiency and stability of consensus control for distributed marginal costs in microgrids.

[0086] In some implementations, a microgrid distributed marginal cost consensus control system based on intelligent leaders is proposed. This system includes a monitoring and management layer capable of configuring parameters. , , , , It displays the system's operating status and includes a control layer with intelligent leader and follower controllers deployed to execute the intelligent leader control law and follower agent control law. The aforementioned monitoring and management layer supports real-time parameter adjustment, allowing the selection of the intelligent leader control law type and configuration of corresponding parameter values ​​based on the microgrid's operating scenario, communication conditions, and performance requirements.

[0087] Specifically, the monitoring and management layer, serving as the upper-level interface of the system, is designed to dynamically modify system operating parameters. This means that during the operation of the microgrid system, operators or automation programs can instantly modify and update control parameters through this interface without interrupting the entire control process or redeploying the system. This real-time adjustment capability can be achieved through a graphical user interface, command-line interface, or by interacting with a higher-level energy management system.

[0088] "Microgrid operating scenarios" refer to the specific operating conditions of the microgrid, such as sudden increases or decreases in load, significant fluctuations in the output of distributed power sources (e.g., solar and wind power), and the microgrid switching from grid-connected mode to islanded mode or vice versa. Different operating scenarios may have different requirements for the response speed and stability of the control system. "Communication conditions" refer to the quality of the internal communication network of the microgrid, including communication latency, packet loss rate, and bandwidth limitations. Poor communication conditions may lead to untimely or incomplete information transmission, thus affecting the performance of the control algorithm. "Performance requirements" refer to the desired effect of the control system, such as requiring the system to converge quickly to the consensus target value. To maintain high control accuracy or good robustness under specific disturbances.

[0089] The intelligent leader control law can be implemented in various forms, such as threshold saturation control law or neighbor feedback control law. Depending on the different operating scenarios and performance requirements, the control law type most suitable for the current situation is selected. For example, when a fast response is required and the error tolerance is high, one control law can be chosen; while when a smooth transition is required and the system is sensitive to shocks, another control law is chosen. This selection can be automated through a pre-set policy library, a rule-based expert system, or a machine learning algorithm. For the selected intelligent leader control law type, specific control parameters need to be set, such as the gain in the threshold saturation control law. And the saturation threshold ϵ, and the weight parameters in the neighbor feedback control law. The configuration of these parameters directly affects the response characteristics and convergence behavior of the control law. Parameter configuration can be based on empirical values, offline optimization results, or dynamic optimization through online adaptive algorithms.

[0090] Through the above implementation methods, the monitoring and management layers are endowed with the ability to adjust parameters in real time and select the type of intelligent leader control law. This means that the system is no longer limited to a fixed set of control strategies and parameters, but can dynamically optimize its control behavior based on the constantly changing operating scenarios, communication conditions, and specific performance requirements of the microgrid in actual operation. For example, when the microgrid load fluctuates drastically or the output of renewable energy changes significantly, the system can adjust the control gain in real time. , Alternatively, the system can switch to a more robust control law type to ensure rapid and stable convergence of marginal costs. Similarly, when communication networks experience congestion or increased packet loss, the system can adjust parameters to reduce sensitivity to communication quality, thereby maintaining control effectiveness. This dynamic adaptability significantly enhances the flexibility, robustness, and efficiency of distributed marginal cost consensus control in microgrids, ensuring continuous optimization of economic dispatch under various complex operating conditions and avoiding performance degradation or system instability caused by static configuration.

[0091] In practical applications of distributed marginal cost consensus control methods for microgrids, an efficient, flexible, and repeatable approach is needed to automate complex control logic and computational processes. Relying solely on manual operation or fixed hardware circuits makes it difficult to adapt to the dynamic changes in the microgrid operating environment and to guarantee the accuracy and real-time performance of the control strategy.

[0092] For ease of description, the above apparatus is described in terms of its functions, divided into various modules. Of course, in implementing this disclosure, the functions of each module can be implemented in one or more software and / or hardware.

[0093] The apparatus described above is used to implement the corresponding microgrid distributed marginal cost consensus control method based on intelligent leader in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0094] Based on the same inventive concept, corresponding to any of the above embodiments, this application also proposes a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, can implement the steps of the above-described microgrid distributed marginal cost consensus control method based on intelligent leaders.

[0095] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0096] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the microgrid distributed marginal cost consensus control method based on intelligent leader as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0097] The above description is merely a preferred embodiment and the technical principles employed in this application. This application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions that can be made by those skilled in the art will not depart from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of this application. The scope of this application is determined by the scope of the claims.

Claims

1. A consensus control method for distributed marginal cost in microgrids based on intelligent leaders, characterized in that, Includes the following steps: Step S1: Establish a microgrid economic dispatch model; Step S2: Construct the communication topology and define a communication graph G consisting of intelligent leaders and followers. If follower i can receive information from the leader, then the leader's adjacency matrix elements are... ,otherwise ; Step S3: Design a distributed marginal cost consensus control law; Step S4: Based on Lyapunov theory, satisfy the parameter conditions for system convergence. and communication topology constraints, wherein the communication topology requires that the communication graph G contains a directed spanning tree rooted at the leader; Step S5: Each intelligent agent executes the distributed marginal cost consensus control law in parallel, driving the marginal cost of all followers. and the state of intelligent leaders Converging to the consensus target value .

2. The microgrid distributed marginal cost consensus control method based on intelligent leader as described in claim 1, characterized in that, In step S1, the consensus target value The power grid price or optimal dispatch value represents the generation cost characteristic of the quadratic convex function, which corresponds to the generation cost of distributed power sources.

3. The microgrid distributed marginal cost consensus control method based on intelligent leader as described in claim 1, characterized in that, In step S2, the communication graph G is a directed or undirected graph corresponding to a wired or wireless network, and the leader adjacency matrix... N is the total number of followers.

4. The microgrid distributed marginal cost consensus control method based on intelligent leader as described in claim 1, characterized in that, In step S3, the control law of the follower agent is: ， ； in, For the state of follower i, This is the current state. For the control input signal of the follower agent. The weight parameters are the communication topology between follower agents. Identification parameters for the communication topology between intelligent leaders and followers. To control the gain, Gather for the neighbors, For intelligent leaders at all times state, is the discrete time step.

5. The microgrid distributed marginal cost consensus control method based on intelligent leader as described in claim 1, characterized in that, In step S3, The intelligent leader control law adopts a threshold saturation type control law: ， ； in The state error of the intelligent leader. The gain of the threshold saturation control law. The saturation threshold, For discrete time steps, Control input signals for intelligent leaders; In the threshold saturation control law, when | |> At that time, the intelligent leader approaches the consensus target value at a fixed rate. When | |≤ At that time, the intelligent leader adjusts the state in a linear manner.

6. The microgrid distributed marginal cost consensus control method based on intelligent leader as described in claim 1, characterized in that, In step S3, The intelligent leader control law adopts a neighbor feedback type control law: ， ； in, A collection of neighboring followers for intelligent leaders. For neighbor feedback control laws, the weight parameters are... is the discrete time step.

7. The microgrid distributed marginal cost consensus control method based on intelligent leader as described in claim 1, characterized in that, In step S4, the parameter conditions The marginal cost used to drive all followers is derived through Lyapunov stability analysis. and the state of intelligent leaders Converging to the consensus target value .

8. A microgrid distributed marginal cost consensus control system based on intelligent leader, capable of executing the microgrid distributed marginal cost consensus control method based on intelligent leader as described in any one of claims 1-7, characterized in that, include: The physical layer includes distributed power sources, energy storage devices, loads, and a point of common coupling (PCC). The physical layer is used to realize the production, storage, consumption, and connection to the main power grid. The communication network layer adopts a wired or wireless network of the communication graph G, and the communication network layer is used for information interaction between intelligent leaders and followers, and between followers; The control layer, which is the agent software deployed on the controller, includes an intelligent leader controller and a follower controller. The intelligent leader controller executes the intelligent leader control law, and the follower controller executes the follower agent control law. Monitoring and management layers are used to configure parameters. , , , The system also displays the operating status, which includes the marginal cost of each distributed power source, the status of the intelligent leader, and the consensus error index J.

9. The microgrid distributed marginal cost consensus control system based on intelligent leader as described in claim 8, characterized in that, The physical carrier of the intelligent leader controller is the energy router ER, which is responsible for power exchange between the microgrid and the main grid. The follower controller is deployed in the local controller of each distributed power source. The monitoring and management system supports real-time parameter adjustment and can select the type of intelligent leader control law and configure the corresponding parameter values ​​according to the microgrid operation scenario, communication conditions, and performance requirements.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the microgrid distributed marginal cost consensus control method based on intelligent leaders as described in any one of claims 1-7.

Images