Cooperative control method, system, device and storage medium for grid connection of multiple microgrids

Abstract

The application relates to the field of grid-connected control technology and discloses a kind of multi-microgrid grid-connected cooperative control method, system, equipment and storage medium, including obtaining the multi-dimensional synchronous operation data of multi-microgrid cluster, and generating real-time dynamic cluster energy flow atlas;According to real-time dynamic cluster energy flow atlas, a multi-agent non-cooperative game model is established, the multi-agent non-cooperative game model takes each microgrid as a game subject, takes grid-connected power as a strategy space, and takes the minimization of cluster risk cost as an objective function;Solving the multi-agent non-cooperative game model obtains the cooperative grid-connected power target instruction;According to the cooperative grid-connected power target instruction, generate grid-connected control timing, and drive each microgrid to execute active grid-connected operation under the grid-connected control timing, to realize the cooperative grid-connected of multi-microgrid cluster to main grid.The application can realize smooth and efficient cooperative grid connection and power distribution optimization of multi-microgrid cluster, effectively improve the power quality and operation safety of power grid.

Description

Technical Field

[0001] This invention relates to the field of grid-connected control technology, and in particular to a collaborative control method, system, device and storage medium for multi-microgrid grid connection. Background Technology

[0002] Microgrids are small-scale power generation and distribution systems composed of distributed power sources, energy storage devices, energy conversion devices, loads, monitoring and protection devices, and are an important component of smart grids. With the widespread integration of distributed energy resources, multi-microgrid clusters, consisting of interconnected geographically adjacent microgrids, have become a common power grid configuration. Achieving safe and efficient coordinated grid connection between multi-microgrid clusters and the main power grid is a significant challenge in the field of power grid control.

[0003] In existing technologies, control methods for multi-microgrid grid interconnection are mainly divided into centralized control and distributed control. Centralized control typically relies on a central controller to collect information from all microgrids and make unified decisions. Although it can achieve global optimization, it has high communication requirements and is susceptible to single-point failures. Distributed control, on the other hand, decentralizes decision-making power to each microgrid, allowing each microgrid to make autonomous decisions based on local information and limited neighbor communication. While this improves flexibility, it often fails to achieve globally optimal control results. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a collaborative control method, system, device, and storage medium for multi-microgrid grid connection, enabling smooth and efficient collaborative grid connection and power allocation optimization of multi-microgrid clusters, thereby improving the safety and stability of grid operation.

[0005] In a first aspect, the present invention provides a coordinated control method for multiple microgrids connected to the grid, the method comprising: Acquire multidimensional synchronous operation data of multiple microgrid clusters, obtain the internal coupling relationship and power distribution status of the multiple microgrid clusters based on the multidimensional synchronous operation data, and generate a real-time dynamic cluster energy flow map; Based on the real-time dynamic cluster energy flow map, a multi-agent non-cooperative game model is established. The multi-agent non-cooperative game model takes each microgrid as the game agent, each game agent takes grid-connected power as the strategy space, and minimizes cluster risk cost as the objective function. Solving the multi-agent non-cooperative game model yields the coordinated grid-connected power target command; Based on the coordinated grid-connected power target instruction, a grid-connected control sequence is generated, and each microgrid is driven to perform active grid-connection operations under the grid-connected control sequence, so as to realize the coordinated grid connection of multiple microgrid clusters to the main grid.

[0006] Furthermore, the step of obtaining the internal coupling relationship and power allocation status of the multi-microgrid cluster based on the multi-dimensional synchronous operation data, and generating a real-time dynamic cluster energy flow map includes: Alignment processing of multi-dimensional synchronous operation data based on high-precision clock signals is performed to generate a synchronous data stream; Extract the real-time power status of each microgrid and the connection relationships between microgrids from the synchronous data stream to generate basic data of nodes and edges; Based on the basic data of the nodes and edges, a graph model is constructed with the microgrid as nodes and the power exchange path as edges; Based on the real-time power status, the node attributes and edge weights in the graph model are dynamically updated to generate a real-time dynamic cluster energy flow graph.

[0007] Furthermore, the step of establishing a multi-agent non-cooperative game model based on the real-time dynamic cluster energy flow map includes: Each microgrid in the real-time dynamic cluster energy flow map is taken as a game subject, and the grid-connected power is taken as the strategy space of each game subject. The grid-connected power includes grid-connected active power and grid-connected reactive power. The total grid connection risk cost under a specific strategy is taken as the cluster risk cost, and the objective function of each game subject is constructed by minimizing the cluster risk cost. The cluster risk cost includes the cluster penalty cost, the control effort cost, and the energy utilization efficiency term. The cluster penalty cost is obtained by performing a safety assessment of the power grid operation under different strategy combinations through the global topology view provided by the real-time dynamic cluster energy flow map; the control effort cost is obtained by calculating the physical stress generated by the internal resources of the microgrid in response to the grid connection command; and the energy utilization efficiency term is obtained by calculating the degree of absorption of distributed energy by the microgrid.

[0008] Further, the step of generating the grid-connected control timing sequence according to the coordinated grid-connected power target instruction includes: According to the coordinated grid-connected power target instruction, each microgrid is driven to perform power pre-adjustment to generate the status of the grid to be connected; Based on the real-time dynamic cluster energy flow graph, the optimal synchronization window that satisfies the synchronization conditions of voltage, frequency and phase angle is identified. Within the optimal synchronization window, a closing command for the grid-connected switch is generated, and together with the status of the station to be connected, they constitute the grid-connected control timing sequence.

[0009] Furthermore, after the step of driving each microgrid to perform active grid connection operation under the grid connection control timing, the method further includes: Collect transient data during grid connection operation and evaluate the execution deviation of each microgrid from the coordinated grid connection power target command; Based on the execution deviation, a control execution trust level is generated for each microgrid; The control execution trust level is used as a feedback parameter to dynamically correct the cluster risk cost of the multi-agent non-cooperative game model at the current moment. Based on the corrected cluster risk cost, the objective function of the multi-agent non-cooperative game model is generated for the next time step.

[0010] Furthermore, the step of generating the control execution trust level for each microgrid based on the execution deviation includes: The execution deviation of each microgrid is compared with a preset power deviation threshold. Based on the comparison results, the execution deviation is piecewise nonlinearly mapped to obtain the trust score at the current moment. The current trust score is weighted and summed with the control execution trust score from the previous time step to obtain the current control execution trust score.

[0011] Furthermore, the step of using the control execution trust level as a feedback parameter to dynamically correct the cluster risk cost in the multi-agent non-cooperative game model includes: Based on the control execution trust level, the weighting coefficient of the cluster penalty cost in the cluster risk cost of each microgrid is adjusted, and a dynamic feedback additional risk item is generated. The dynamic feedback additional risk item is merged into the cluster risk cost after the weight coefficient is adjusted to obtain the corrected cluster risk cost.

[0012] Secondly, the present invention provides a collaborative control system for multiple microgrids connected to the grid, the system comprising: The energy flow map construction module is used to acquire multi-dimensional synchronous operation data of multiple microgrid clusters, obtain the internal coupling relationship and power distribution status of the multiple microgrid clusters based on the multi-dimensional synchronous operation data, and generate a real-time dynamic cluster energy flow map. The game model construction module is used to establish a multi-agent non-cooperative game model based on the real-time dynamic cluster energy flow map. The multi-agent non-cooperative game model takes each microgrid as the game agent, each game agent takes grid-connected power as the strategy space, and minimizes cluster risk cost as the objective function. The grid connection command generation module is used to solve the multi-agent non-cooperative game model to obtain the coordinated grid connection power target command; The grid connection command execution module is used to generate a grid connection control sequence according to the coordinated grid connection power target command, and drive each microgrid to perform active grid connection operation under the grid connection control sequence, so as to realize the coordinated grid connection of multiple microgrid clusters to the main grid.

[0013] Thirdly, embodiments of the present invention also provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0014] Fourthly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.

[0015] This invention provides a collaborative control method, system, device, and storage medium for multi-microgrid grid interconnection. By constructing a real-time dynamic cluster energy flow map, this invention provides an accurate and consistent global system state view for subsequent decision-making, improving the accuracy and real-time performance of the collaborative control strategy. Through multi-agent non-cooperative game optimization, it achieves comprehensive optimization of security and stability within a distributed decision-making framework. By introducing an execution deviation assessment and control execution trust feedback mechanism, a closed-loop adaptive correction loop from physical execution to model decision-making is constructed, enabling dynamic adaptive adjustment of game model parameters and enhancing the model's adaptability and robustness to performance differences and environmental changes in various microgrids. This invention enables smooth and efficient collaborative grid interconnection and power allocation optimization of multi-microgrid clusters, effectively improving the power quality and operational safety of the power grid. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the collaborative control method for multi-microgrid grid connection in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of the collaborative control system for multiple microgrids connected to the grid in an embodiment of the present invention; Figure 3 This is an internal structural diagram of the computer device in an embodiment of the present invention.

[0017] Figure label: 10. Energy flow map construction module; 20. Game theory model construction module; 30. Grid connection instruction generation module; 40. Grid connection instruction execution module. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1The first embodiment of the present invention proposes a collaborative control method for multi-microgrid grid connection, including steps S10 to S40: Step S10: Obtain multi-dimensional synchronous operation data of the multi-microgrid cluster; based on the multi-dimensional synchronous operation data, obtain the internal coupling relationship and power distribution status of the multi-microgrid cluster, and generate a real-time dynamic cluster energy flow map. Step S20: Based on the real-time dynamic cluster energy flow map, establish a multi-agent non-cooperative game model. The multi-agent non-cooperative game model takes each microgrid as the game agent, each game agent takes grid-connected power as the strategy space, and minimizes cluster risk cost as the objective function. Step S30: Solve the multi-agent non-cooperative game model to obtain the collaborative grid-connected power target command; Step S40: Based on the coordinated grid-connected power target instruction, generate the grid-connected control timing sequence and drive each microgrid to perform active grid-connection operation under the grid-connected control timing sequence, so as to realize the coordinated grid connection of multiple microgrid clusters to the main grid.

[0020] This embodiment analyzes the multi-dimensional synchronous operation data of multiple microgrid clusters to construct a real-time dynamic cluster energy flow map. The real-time dynamic cluster energy flow map characterizes the internal coupling relationships and power distribution status of the multiple microgrid clusters. The specific construction steps include: Alignment processing of multi-dimensional synchronous operation data based on high-precision clock signals is performed to generate a synchronous data stream; Extract the real-time power status of each microgrid and the connection relationships between microgrids from the synchronous data stream to generate basic data of nodes and edges; Based on the basic data of the nodes and edges, a graph model is constructed with the microgrid as nodes and the power exchange path as edges; Based on the real-time power status, the node attributes and edge weights in the graph model are dynamically updated to generate a real-time dynamic cluster energy flow graph.

[0021] In this embodiment, the multi-dimensional synchronous operation data of the multi-microgrid cluster is first aligned to generate a synchronous data stream. The alignment steps include: receiving a high-precision clock signal, establishing a unified spatiotemporal reference within the cluster, synchronously collecting multi-dimensional synchronous operation parameters of grid connection points, distributed power sources and energy storage devices in each microgrid based on the unified spatiotemporal reference, aggregating the multi-dimensional synchronous operation parameters, and performing time-series alignment based on the unified spatiotemporal reference to generate a synchronous data stream.

[0022] Specifically, a high-precision clock signal is uniformly received by all microgrid controllers or data acquisition units. This high-precision clock signal typically originates from global navigation satellites such as GPS or BeiDou, or is distributed via network time protocols such as IEEE 1588PTP, achieving sub-microsecond accuracy. This signal establishes a unified spatiotemporal reference across the entire microgrid cluster's geographical area. This means that all devices within the cluster share the same high-precision time coordinate system, eliminating time asynchrony issues caused by differences or drift in local clocks of individual microgrids. Under the constraint of this unified spatiotemporal reference, measurement devices such as phase measurement units (PMUs) or intelligent electronic devices (IEDs) deployed at each microgrid grid connection point, distributed power sources such as photovoltaic arrays, wind turbines, and energy storage devices (e.g., battery terminals) are synchronously triggered, collecting multi-dimensional synchronous operating parameters at exactly the same time. These parameters constitute key information describing the microgrid's state, typically including voltage amplitude, voltage phase angle, frequency, active power, and reactive power. The multi-dimensional synchronous operating parameters with precise timestamps collected from each microgrid are then transmitted to a central processing node or distributed computing network for aggregation. At this stage, the aggregated parameters undergo rigorous time-series alignment based on a unified spatiotemporal reference. This involves merging data from different physical locations but with the same timestamp into the same data frame, thereby correcting minor time deviations caused by factors such as network transmission delays. This process ultimately generates a structured, time-consistent synchronous data stream. This synchronous data stream, in the form of a time series, accurately reproduces a snapshot of the entire microgrid cluster's global operation at every moment.

[0023] After obtaining the synchronous data stream, a real-time dynamic cluster energy flow graph is constructed by parsing the synchronous data stream. Specifically, from each time slice of the synchronous data stream, the real-time power state associated with each microgrid is extracted, mainly including the injection or absorption values ​​of active and reactive power, as well as the connection relationships between microgrids. These connection relationships are usually based on preset grid topology information, clarifying the direct physical line connections between microgrids. These extracted real-time power states and connection relationships together constitute the basic elements required to construct the graph model, namely the basic data of nodes and edges.

[0024] Based on the basic data of nodes and edges, the graph model is constructed as follows: , where the set of nodes Each node in Represents an independent microgrid entity, edge set Each edge in This indicates a node. With nodes There exists a power exchange path between them. i and j Both represent microgrid node indices.

[0025] To ensure the graph model reflects the state in real time, its attributes are dynamically assigned based on the continuously updated real-time power state from the synchronous data stream. Each node... The attributes of the node are updated to its current real-time power state, such as net exchange power. The edges connecting the two nodes... weight It is then dynamically updated to the actual active power exchanged between them, and its directionality can represent the direction of power flow, expressed by the following formula: ; in, For at any time Connecting to microgrids and microgrids The weight of the edge. For at any time From microgrids Flow to microgrid The active power is extracted directly from the data frame corresponding to the timestamp in the synchronous data stream.

[0026] Through this continuous, real-time data-based update mechanism, a static graphical model is transformed into a real-time dynamic cluster energy flow map that can intuitively characterize the direction, magnitude, and coupling relationships of power flow within a multi-microgrid cluster.

[0027] This embodiment establishes a multi-agent non-cooperative game model based on the global topology view provided by the real-time dynamic cluster energy flow graph. The specific steps include: Each microgrid in the real-time dynamic cluster energy flow map is taken as a game subject, and the grid-connected power is taken as the strategy space of each game subject. The grid-connected power includes grid-connected active power and grid-connected reactive power. The total grid connection risk cost under a specific strategy is taken as the cluster risk cost, and the objective function of each game subject is constructed by minimizing the cluster risk cost. The cluster risk cost includes the cluster penalty cost, the control effort cost, and the energy utilization efficiency term. The cluster penalty cost is obtained by performing a safety assessment of the power grid operation under different strategy combinations through the global topology view provided by the real-time dynamic cluster energy flow map; the control effort cost is obtained by calculating the physical stress generated by the internal resources of the microgrid in response to the grid connection command; and the energy utilization efficiency term is obtained by calculating the degree of absorption of distributed energy by the microgrid.

[0028] In this embodiment, each microgrid in the real-time dynamic cluster energy flow map is abstracted and defined as an independent game agent, i.e., a rational decision-maker. Each game agent... Define its policy space The strategy space is a set of parameters that the microgrid can autonomously adjust during grid-connected operation, representing the active power it plans to exchange with the main grid. and reactive power Therefore, the main players in the game strategy It can be represented as a two-dimensional vector. Its value range is constrained by the physical capabilities of the power generation equipment, energy storage units, and inverters within the microgrid.

[0029] Then, the objective function of the game subjects is constructed. This objective function is used to optimize the operational risk of the system. Therefore, this embodiment adopts the minimization of cluster risk cost to construct the objective function of each game subject. This objective function can comprehensively quantify the cluster penalty cost, control effort cost and energy utilization efficiency.

[0030] Specifically, for each game player Construct a cluster risk cost function This function is used to quantify the selection of a specific strategy. The cluster risk cost is calculated based on the overall stability of the cluster while considering the economic interests of individuals. This function is the core of the entire game theory model. To ensure the additivity of different physical property parameters during coordinated control, this embodiment introduces per-unit values ​​and a normalization function for the power system. The per-unit value of each physical quantity is calculated using its rated value or a system-defined benchmark value to establish a dimensionless cluster risk cost function. Its expression is: In the formula, As the main players in the game Grid-connected power strategy; All are preset dimensionless weighting coefficients; As the main players in the game The weighting coefficient of the cluster penalty cost at time t is used to reflect the safety margin requirement of the system for the control reliability of a specific microgrid. This coefficient can be dynamically adjusted according to the requirements at different times. The main players in the game Take a strategy at time t Meanwhile, all other game players adopt a combination of strategies. The total grid connection risk cost, i.e., cluster risk cost. Representative of the game players Take a strategy at time t The cost of control effort during the process is used to quantify the physical stress on microgrid resources in response to grid connection commands. Representative of the game players Take a strategy at time t Meanwhile, all other game players adopt a combination of strategies. The cluster penalty cost is calculated by evaluating the impact of strategy combinations on grid operation safety through a global topology view provided by a real-time dynamic cluster energy flow graph. Representative of the game players Take a strategy at time t The energy utilization efficiency term represents the degree of optimization of the microgrid for the absorption of distributed energy resources.

[0031] It should be noted that, during calculation, all physical components in the above formula can be mapped to... Dimensionless scales within the interval are used to address the calculation deviation problem caused by inconsistent dimensions of the original parameters.

[0032] In a preferred embodiment, the control effort cost is characterized by the power change rate, which is calculated using the following formula: in, Let be the active power that microgrid i intends to perform at time t. Let i be the actual power of microgrid i at time t-1. This represents the rated capacity of microgrid i.

[0033] The above formula characterizes the inverter switching losses and thermal effects by using the square of the power change rate. Its goal is not to quantify the losses, but to suppress drastic power fluctuations from a physical perspective.

[0034] The cluster penalty cost establishes a mapping relationship between power strategy combinations and bus voltage deviation through the node admittance matrix in the graph, thereby quantifying the impact of strategy combinations on grid operation safety. Its expression is as follows: In the formula, As the main players in the game Take a strategy at time t Meanwhile, all other game players adopt a combination of strategies. Key Nodes of Time k The voltage amplitude, which is calculated through the real-time dynamic cluster energy flow map, reflects the voltage state of the cluster under a specific strategy combination; It is a set of key nodes, including the grid connection points of each microgrid, the interconnection nodes between microgrids, and the hub nodes connected to the main grid, which are electrical nodes that have an important impact on the stability of the cluster. This is a preset reference voltage amplitude, which is determined by the bus voltage.

[0035] The cluster penalty cost quantifies the risk of voltage deviation in critical nodes caused by grid connection behavior. When the voltage deviates from the safe range, this factor increases significantly, thereby forcibly constraining the behavior of the subjects in the game logic to ensure physical steady state.

[0036] The energy utilization efficiency term is calculated based on the output value of distributed power sources within the microgrid, and its formula is as follows: in, This refers to the real-time predicted maximum output of distributed power sources within a microgrid at time t. Let be the active power that microgrid i intends to execute at time t.

[0037] This item serves as a guiding principle, encouraging game participants to maximize the use of green energy and improve overall energy management efficiency while meeting stability constraints.

[0038] Finally, by integrating the strategy spaces of all game participants with the cluster risk-cost function, a multi-agent non-cooperative game model describing the interactive behavior of the entire microgrid system was formed.

[0039] Solving a multi-agent non-cooperative game model to obtain a Nash equilibrium solution under the constraint of cluster operational stability includes the following steps: calculating the optimal response strategy for each player based on the cluster risk cost function using an iterative optimization algorithm; determining whether the combination of optimal response strategies for all players converges to a stable state and generating a convergence judgment result; and determining the stable state as a Nash equilibrium solution when the convergence judgment result is yes.

[0040] Specifically, based on the cluster risk cost function, an iterative optimization algorithm is used to simulate the process by which each player continuously adjusts its strategy in dynamic interactions to seek the minimum cost. At each step of the iteration process, based on the strategy combinations of all players in the previous round, the optimal response strategy under the current conditions is calculated individually for each player. For each player… It assumes all other game players strategy It is fixed and unchanging, and then in its own strategy space. Internal search for a risk-cost function that enables its cluster Minimization strategy This search process is essentially an optimization problem, with the objective function being: in, The main players in the game The optimal response strategy Function representation: finding the function that makes the function Strategies to reach the minimum value .

[0041] Once the optimal response strategies for all players have been calculated, these strategies are combined into a new global strategy. Then, a crucial decision step is performed to determine whether the combination of optimal response strategies for all players has converged to a stable state. This decision is made by comparing the difference between the global strategy combinations obtained from two consecutive iterations. If this difference is less than a preset convergence threshold, a positive convergence result is generated.

[0042] When the convergence test result is positive, it indicates that an equilibrium point has been reached, meaning that no single player can gain a higher payoff by unilaterally changing their strategy. At this point, the strategy combination in this stable state is determined as the Nash equilibrium solution of the multi-player non-cooperative game model. This solution represents a cooperative operating point that balances the individual interests of all microgrid entities with the overall stability of the cluster.

[0043] Then, the Nash equilibrium solution of the multi-agent non-cooperative game model is converted into a coordinated grid-connection power target command. Execution of this command achieves grid connection between the multiple microgrids and the main grid. In traditional grid connection command execution, the triggering timing of the grid connection operation often relies on simple electrical threshold judgments, lacking dynamic predictability and easily resulting in large power surges at the moment of grid connection. To address this issue, this embodiment generates a grid connection control timing sequence to achieve stable and smooth coordinated grid connection. The specific steps for generating the grid connection control timing sequence include: According to the coordinated grid-connected power target instruction, each microgrid is driven to perform power pre-adjustment to generate the status of the grid to be connected; Based on the real-time dynamic cluster energy flow graph, the optimal synchronization window that satisfies the synchronization conditions of voltage, frequency and phase angle is identified. Within the optimal synchronization window, a closing command for the grid-connected switch is generated, and together with the status of the station to be connected, they constitute the grid-connected control timing sequence.

[0044] In this embodiment, the local controller of each microgrid drives its internal controllable resources to perform power pre-adjustment based on the received coordinated grid-connection power target command. That is, the controller actively adjusts the output power of the distributed power sources and the charging and discharging state of the energy storage system, so that the active and reactive power output of the microgrid at the grid connection point precisely approaches the target value set in the command. After completing this adjustment, the microgrid enters a stable operating state where the power output meets the coordinated target, ready for grid connection.

[0045] While maintaining the grid connection status, the microgrid continuously and frequently monitors the real-time dynamic cluster energy flow map and its underlying synchronization data flow to accurately capture an ideal grid connection opportunity, i.e., the optimal synchronization window. The optimal synchronization window is identified based on preset electrical synchronization conditions, namely, comparing three key electrical quantities—voltage, frequency, and phase angle—between the microgrid's grid connection point and the main grid. The optimal synchronization window is determined to have been entered only when the differences between these three quantities simultaneously meet preset threshold ranges. Once the synchronization conditions are detected, a grid connection switch closing command is immediately generated within the optimal synchronization window. This closing command, combined with the previous grid connection status, constitutes a complete grid connection control sequence, driving the physical switch to close precisely.

[0046] In a preferred embodiment, after the grid connection operation is completed, this embodiment performs adaptive correction on the model based on execution feedback. The specific steps include: Collect transient data during grid connection operation and evaluate the execution deviation of each microgrid from the coordinated grid connection power target command; Based on the execution deviation, a control execution trust level is generated for each microgrid; The control execution trust level is used as a feedback parameter to dynamically correct the cluster risk cost of the multi-agent non-cooperative game model at the current moment. Based on the corrected cluster risk cost, the objective function of the multi-agent non-cooperative game model is generated for the next time step.

[0047] In this embodiment, transient data during the grid connection operation is acquired at high frequency within a very short time after the grid-connected switch is closed. This transient data consists of actual measured values ​​describing the instantaneous changes in the electrical characteristics of the grid connection point, such as active and reactive power at the millisecond level. These measured power values ​​are compared one by one with the coordinated grid connection power target command issued to each microgrid to evaluate the execution deviation of each microgrid from the command. The execution deviation is expressed as: in, Microgrid The execution deviation at time t Microgrids extracted from transient data The actual output power after grid connection at time t It is the power value at time t in the coordinated grid-connected power target command derived from the Nash equilibrium solution.

[0048] Based on the calculated execution deviation, a quantization model is used to generate a control execution confidence score for each microgrid. This is a numerical metric used to evaluate the reliability and accuracy of the microgrid's control system. Generally, the smaller the execution deviation, the higher the generated control execution confidence score, and vice versa. This quantization process abstracts the physical execution performance into parameters that can be used in the model. For example, a linear function can be used to characterize the negative correlation between execution deviation and control execution confidence score.

[0049] In a preferred embodiment, a piecewise nonlinear mapping method is used to generate the control execution trust level, and the specific steps include: The execution deviation of each microgrid is compared with a preset power deviation threshold. Based on the comparison results, the execution deviation is piecewise nonlinearly mapped to obtain the trust score at the current moment. The current trust score is weighted and summed with the control execution trust score from the previous time step to obtain the current control execution trust score.

[0050] In this embodiment, each microgrid Execution deviation The absolute value of the power deviation threshold is different from the preset power deviation threshold. A comparison is made. This power deviation threshold is preset according to safe operating procedures and control accuracy requirements, defining an acceptable error range. The output is a comparison result reflecting the severity of the current execution deviation.

[0051] Based on the comparison results, a confidence score characterizing the current control performance is generated. This confidence score is a normalized value, typically between 0 and 1, used to reflect the control accuracy of this grid connection operation in real time. For example, when the execution deviation is zero, the confidence score is 1; when the execution deviation equals the power deviation threshold, the score is an intermediate value; and when the execution deviation far exceeds the threshold, the score approaches 0. Therefore, this embodiment employs a piecewise mapping function. f To calculate the trust score, this function aims to non-linearly map physical execution deviations into a control performance evaluation scale, as expressed below: In the formula, Assess the trust level of microgrid i at time t. Microgrid The absolute value of the execution deviation at time t. This is a preset power deviation threshold used to define the acceptable error range.

[0052] The piecewise mapping function in this embodiment employs a quadratic penalty mechanism. When the execution deviation is within a threshold, the score decays rapidly as the deviation increases, sensitively capturing minute fluctuations in control performance. Conversely, when the deviation exceeds the threshold, the score is reset to zero, indicating that the current control state of the game entity is unreliable. This mapping method transforms transient execution deviations at the physical level into dimensionless control performance evaluation indicators, providing standardized parameters for the subsequent dynamic correction of the game model.

[0053] Furthermore, to obtain a smoother indicator that reflects long-term performance, the current control performance confidence score is weighted and fused with the historical control execution confidence score of the microgrid, thereby updating and outputting the final control execution confidence score. The historical control execution confidence score is the control execution confidence score calculated at the previous time step; therefore, the control execution confidence score can be expressed as: in, As a weighting factor, Let i be the control execution confidence level of microgrid i at time t. Let i be the confidence level of control execution at time t-1. Assign a trust score to microgrid i at time t. Weighting factor. A value between 0 and 1, used to adjust the weighting of current performance relative to historical performance. A larger value indicates a lower weighting. The value indicates that the control execution trust level is more sensitive to the latest control performance, while a smaller value indicates a lower level of trust. The value indicates a greater emphasis on long-term stable performance.

[0054] Then, the control execution trust level is used as a feedback parameter to dynamically correct the cluster risk cost in the multi-agent non-cooperative game model. The specific steps include: Based on the control execution trust level, the weighting coefficient of the cluster penalty cost in the cluster risk cost of each microgrid is adjusted, and a dynamic feedback additional risk item is generated. The dynamic feedback additional risk item is merged into the cluster risk cost after the weight coefficient is adjusted to obtain the corrected cluster risk cost.

[0055] In this embodiment, firstly, based on each microgrid Control execution trust For its cluster risk cost function The weighting coefficients of the cluster penalty cost are adjusted. Its weighting coefficients are used to quantify the impact of individual strategies on cluster stability. This is used to adjust its importance in the total cost function. The control execution confidence level of a microgrid affects cluster stability; for example, a lower control execution confidence level directly impacts cluster stability. Based on the negative correlation between the two, the dynamic adjustment process of their weighting coefficients can be expressed as: In the formula, Microgrid The weighting coefficient of the cluster penalty cost at the next time step, i.e., time t+1. The base weight value.

[0056] The corrected game theory model effectively constrains the behavior of unreliable agents. When the microgrid's control execution trust level is extremely low, it indicates poor control accuracy and insufficient reliability. Amplifying the weight of the cluster penalty cost allows the microgrid to pay extremely high costs for "strategies that endanger cluster stability" during the game process, thereby forcing it to choose a more conservative grid connection strategy and avoid threatening cluster stability. Through extreme weighted penalties, a negative incentive is created for the low-trust microgrid, driving it to optimize control performance and improve execution accuracy, ultimately improving the overall robustness of cluster control.

[0057] It's important to note that in the dynamic correction formula for the weighting coefficients, the control execution trust score, as the denominator, may sometimes be zero. However, because this embodiment incorporates historical control execution trust scores during the calculation, even if a single execution deviation exceeds the threshold, resulting in a current trust score of zero, the current control execution trust score will not directly reach zero due to the weighting effect of historical control execution trust scores; it will only gradually decrease. Only when the microgrid experiences multiple consecutive severe execution deviations will the control execution trust score approach zero. This severe execution deviation can be understood as a risk of failure in the microgrid. Therefore, when the calculated control execution trust score is less than the preset safety threshold, a fault warning will be issued, and the grid connection operation of the microgrid will be suspended until feedback is received and the warning is lifted. Then, the game optimization will be performed again according to the basic weight values, thereby ensuring the safety and stability of the grid connection process.

[0058] To further enhance this feedback effect, this embodiment also introduces an additional risk term negatively correlated with the control execution trust level into the cluster risk cost function of the microgrid. This additional risk term is used to directly penalize behaviors with low trust levels, and its calculation formula is as follows: In the formula, This represents the additional risk term for microgrid i at time t+1. For risk correction operators.

[0059] Due to the control execution trust level for To maintain dimensional consistency, the dimensionless scales between them... The value of is usually determined based on the system's worst-case operating condition, i.e. The tolerance limit setting at that time is preferred. The range of values ​​is This ensures that the additional risk item is on the same order of magnitude as the cluster instability risk, thereby playing a significant incentive and constraint role without disrupting the game balance.

[0060] This additional risk item is directly included in the total cost function as an independent item. When the control execution confidence level of a microgrid is not 1 at the current moment, its cluster risk cost at the next moment will increase by this additional risk item, and its expression is as follows: When a microgrid with low control execution confidence exists, the presence of this additional risk term will suppress its net gains under any strategy, thus incentivizing it to improve control performance to restore confidence. Once confidence is restored, this additional risk term will be zero in the cluster risk cost at subsequent time points, meaning it disappears.

[0061] This embodiment achieves dynamic correction of the cluster risk cost function by adjusting the weight coefficient of the cluster penalty cost and introducing additional risk terms, thereby strengthening the synergistic effect between the real-time dynamic cluster energy flow map and the multi-agent non-cooperative game model.

[0062] To verify the feasibility of this invention in practice, it was applied to the dispatch center of a regional power grid. This regional power grid administers multiple renewable energy microgrids, including a photovoltaic power generation microgrid (MG1), a wind power generation microgrid (MG2), and a corporate park microgrid equipped with energy storage (MG3).

[0063] Currently, these microgrids operate independently and lack coordination when connecting to the main grid, often causing voltage fluctuations, frequency shifts, and power surges in the main grid at the moment of connection, severely affecting grid stability and power quality. The regional grid dispatch center hopes to use the method of this invention to achieve proactive coordinated control of multiple microgrids during the grid connection process, ensuring the safe and economical operation of the grid.

[0064] In this embodiment, the regional power grid dispatch center deploys the collaborative control system of this invention. First, by receiving a high-precision clock signal provided by the BeiDou Navigation Satellite System (BDS), a sub-microsecond unified spatiotemporal reference is established among MG1, MG2, and MG3. Based on this reference, phase measurement units (PMUs) deployed at each microgrid grid connection point synchronously collect data such as voltage amplitude, phase angle, frequency, active power, and reactive power at exactly the same time, and aggregate them into a synchronous data stream with strictly aligned timing. This data stream is analyzed to construct a real-time dynamic cluster energy flow graph with MG1, MG2, and MG3 as nodes and the physical lines between them as edges. The edge weights of the graph are based on the real-time active power exchange value. Dynamically updated.

[0065] Subsequently, MG1, MG2, and MG3 were defined as independent game agents, and a multi-agent non-cooperative game model was established based on the energy flow graph. The strategy space of each microgrid is the active power it plans to exchange with the main grid. and reactive power Its cluster risk cost function This approach integrates electricity sales revenue, control and regulation costs, and a key cluster penalty term. Based on energy flow graph analysis, this penalty term quantifies the potential negative impact of individual microgrid grid connection strategies on the overall voltage stability of the cluster. An iterative optimization algorithm is used to solve the Nash equilibrium of this game theory model, yielding an optimal coordinated grid connection power target command that balances the interests of all parties with grid stability. After this command is issued, each microgrid performs power pre-regulation while continuously monitoring the energy flow graph. Once the optimal synchronization window is found where the microgrid's grid connection point and the main grid's voltage, frequency, and phase angle meet synchronization conditions, a closing command is issued to complete the grid connection.

[0066] To verify the beneficial effects of this embodiment, a planned grid connection process of three microgrids was tested at a dispatch center in a certain location, and compared with historical data using the traditional independent grid connection method. Key performance indicators before and after coordinated grid connection were recorded during the experiment.

[0067] During the grid connection decision-making phase, a Nash equilibrium solution was obtained, setting active power targets of 2.5MW, 3.0MW, and 1.8MW for the three microgrids MG1, MG2, and MG3, respectively. After grid connection was implemented, transient data showed that the actual output power of MG1 and MG3 deviated very little from the target values, at 2.51MW and 1.82MW, respectively. However, due to the internal wind turbine response delay, MG2's actual output power was 2.65MW, resulting in a larger execution deviation of -0.35MW.

[0068] Based on this execution deviation, the control execution trust scores of each microgrid were updated. MG1 and MG3 had higher trust scores, and after weighted fusion with historical trust scores, their final control execution trust scores were updated to 0.95 and 0.93, respectively. However, MG2 had a larger deviation, resulting in a lower current trust score and a decrease in its final control execution trust score from 0.90 in the previous period to 0.82.

[0069] In the next calibration of the grid-connected game model, based on the reduced trust level of MG2, the weight coefficient of the cluster penalty cost in its cluster risk cost function will be automatically adjusted. Increased from the baseline value of 1.1 to And introduced additional risk items. This means that in the new game, any high-power output strategy that might lead to instability in MG2 will face a higher economic penalty. The model ultimately assigned it a more conservative grid-connected power target of 2.8MW, thereby reducing its potential risk.

[0070] From the perspective of grid stability data, during the coordinated grid connection process, the voltage deviation rate of the key bus in the main grid was controlled within 0.8%, and the frequency deviation was less than 0.01Hz. In contrast, historical data shows that under similar scale independent grid connection conditions, the voltage deviation rate can reach up to 4.5%, and the frequency deviation can reach 0.08Hz. This embodiment suppresses nearly 80% of the transient disturbances caused by grid connection. It can be seen that the effect of adopting this embodiment is significant. This embodiment can achieve smooth and efficient coordinated grid connection, thereby improving the power quality and security of the grid.

[0071] This embodiment provides a collaborative control method for multi-microgrid grid connection. By constructing a real-time dynamic cluster energy flow map based on synchronous data flow, it provides an accurate and consistent global system state view for subsequent decision-making, solving the problems of model distortion and control inaccuracy caused by data asynchrony, and improving the accuracy and real-time performance of the collaborative control strategy. Through multi-agent non-cooperative game optimization based on multi-microgrid collaborative grid connection, it achieves overall optimization of security and stability under a distributed decision-making framework. By introducing an execution deviation evaluation and control execution trust feedback mechanism, a closed-loop adaptive correction loop from physical execution to model decision-making is constructed, realizing dynamic adaptive adjustment of game model parameters, enhancing the model's adaptability and robustness to performance differences and environmental changes in various microgrids, and further improving the power quality and security of the grid.

[0072] Please see Figure 2 Based on the same inventive concept, the second embodiment of this invention proposes a collaborative control system for multi-microgrid grid connection, comprising: The energy flow map construction module 10 is used to acquire multi-dimensional synchronous operation data of the multi-microgrid cluster, obtain the internal coupling relationship and power distribution status of the multi-microgrid cluster based on the multi-dimensional synchronous operation data, and generate a real-time dynamic cluster energy flow map. The game model construction module 20 is used to establish a multi-agent non-cooperative game model based on the real-time dynamic cluster energy flow map. The multi-agent non-cooperative game model takes each microgrid as the game agent, each game agent takes grid-connected power as the strategy space, and minimizes cluster risk cost as the objective function. The grid connection instruction generation module 30 is used to solve the multi-agent non-cooperative game model to obtain the coordinated grid connection power target instruction; The grid connection instruction execution module 40 is used to generate a grid connection control sequence according to the coordinated grid connection power target instruction, and drive each microgrid to perform active grid connection operation under the grid connection control sequence, so as to realize the coordinated grid connection of multiple microgrid clusters to the main grid.

[0073] The technical features and effects of the multi-microgrid grid-connected collaborative control system proposed in this invention are the same as those of the method proposed in this invention, and will not be repeated here. Each module in the above-mentioned multi-microgrid grid-connected collaborative control system can be implemented entirely or partially through software, hardware, or a combination thereof. Each module can be embedded in or independent of the processor in a computer device in hardware form, or it can be stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0074] Furthermore, embodiments of the present invention also propose a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0075] Please see Figure 3The diagram illustrates the internal structure of a computer device in one embodiment. This computer device can specifically be a terminal or a server. The computer device includes a processor, memory, network interface, display, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a collaborative control method for multi-microgrid grid connection. The display screen of the computer device can be a liquid crystal display (LCD) or an e-ink display. The input devices of the computer device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse, etc.

[0076] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computing devices may include more or fewer components than those shown in the figure, or combine certain components, or have the same component arrangement.

[0077] Furthermore, embodiments of the present invention also propose a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described method.

[0078] In summary, the embodiments of this invention propose a collaborative control method, system, device, and storage medium for multi-microgrid grid connection. The method acquires multi-dimensional synchronous operation data of a multi-microgrid cluster, obtains the internal coupling relationship and power allocation status of the cluster based on this data, and generates a real-time dynamic cluster energy flow map. Based on the real-time dynamic cluster energy flow map, a multi-agent non-cooperative game model is established, with each microgrid as a player, each player using grid-connected power as its strategy space, and minimizing cluster risk cost as the objective function. The multi-agent non-cooperative game model is solved to obtain a collaborative grid-connected power target instruction. Based on the collaborative grid-connected power target instruction, a grid-connected control sequence is generated, and each microgrid is driven to perform active grid-connection operations under the grid-connected control sequence, thereby achieving collaborative grid connection of the multi-microgrid cluster to the main grid. This invention constructs a real-time dynamic cluster energy flow map based on synchronous data streams, providing an accurate and consistent global system state view for subsequent decision-making, thus improving the accuracy and real-time performance of collaborative control strategies. Through multi-agent non-cooperative game optimization based on multi-microgrid collaborative grid connection, it achieves comprehensive optimization of security and stability within a distributed decision-making framework. By introducing execution deviation evaluation and control execution trust feedback mechanisms, it realizes dynamic adaptive adjustment of game model parameters, enhancing the model's adaptability and robustness to performance differences and environmental changes among microgrids. This invention enables smooth and efficient collaborative grid connection and power allocation optimization of multi-microgrid clusters, effectively improving the power quality and operational safety of the power grid.

[0079] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0080] The embodiments described above are merely preferred embodiments of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the technical principles of this invention, and these improvements and substitutions should also be considered within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the scope of the claims.

Claims

1. A collaborative control method for multi-microgrid grid connection, characterized in that, include: Acquire multidimensional synchronous operation data of multiple microgrid clusters, obtain the internal coupling relationship and power distribution status of the multiple microgrid clusters based on the multidimensional synchronous operation data, and generate a real-time dynamic cluster energy flow map; Based on the real-time dynamic cluster energy flow map, a multi-agent non-cooperative game model is established. This model treats each microgrid as a player, with grid-connected power as the strategy space, and minimizes cluster risk cost as the objective function. The cluster risk cost includes cluster penalty cost, control effort cost, and energy utilization efficiency. The cluster penalty cost is the voltage deviation rate of key grid nodes under different strategy combinations obtained from the real-time dynamic cluster energy flow map; the control effort cost is the rate of change of grid-connected power of the microgrid; and the energy utilization efficiency is obtained by calculating the degree of absorption of distributed energy resources by the microgrid. Solving the multi-agent non-cooperative game model yields the coordinated grid-connected power target command; Based on the coordinated grid-connected power target instruction, a grid-connected control sequence is generated, and each microgrid is driven to perform active grid-connection operations under the grid-connected control sequence, so as to realize the coordinated grid connection of multiple microgrid clusters to the main grid.

2. The coordinated control method for multi-microgrid grid connection according to claim 1, characterized in that, The steps of obtaining the internal coupling relationship and power distribution status of the multi-microgrid cluster based on the multi-dimensional synchronous operation data, and generating a real-time dynamic cluster energy flow map, include: Alignment processing of multi-dimensional synchronous operation data based on high-precision clock signals is performed to generate a synchronous data stream; Extract the real-time power status of each microgrid and the connection relationships between microgrids from the synchronous data stream to generate basic data of nodes and edges; Based on the basic data of the nodes and edges, a graph model is constructed with the microgrid as nodes and the power exchange path as edges; Based on the real-time power status, the node attributes and edge weights in the graph model are dynamically updated to generate a real-time dynamic cluster energy flow graph.

3. The coordinated control method for multiple microgrids connected to the grid according to claim 1, characterized in that, The steps for establishing a multi-agent non-cooperative game model based on the real-time dynamic cluster energy flow map include: Each microgrid in the real-time dynamic cluster energy flow map is taken as a game subject, and the grid-connected power is taken as the strategy space of each game subject. The grid-connected power includes grid-connected active power and grid-connected reactive power. The total grid connection risk cost under a specific strategy is taken as the cluster risk cost, and the objective function of each game subject is constructed by minimizing the cluster risk cost.

4. The coordinated control method for multi-microgrid grid connection according to claim 1, characterized in that, The step of generating the grid connection control timing sequence according to the coordinated grid connection power target instruction includes: According to the coordinated grid-connected power target instruction, each microgrid is driven to perform power pre-adjustment to generate the status of the grid to be connected; Based on the real-time dynamic cluster energy flow graph, the optimal synchronization window that satisfies the synchronization conditions of voltage, frequency and phase angle is identified. Within the optimal synchronization window, a closing command for the grid-connected switch is generated, and together with the status of the station to be connected, they constitute the grid-connected control timing sequence.

5. The coordinated control method for multiple microgrids connected to the grid according to claim 3, characterized in that, After the step of driving each microgrid to perform active grid connection operation under the grid connection control sequence, the method further includes: Collect transient data during grid connection operation and evaluate the execution deviation of each microgrid from the coordinated grid connection power target command; Based on the execution deviation, a control execution trust level is generated for each microgrid; The control execution trust level is used as a feedback parameter to dynamically correct the cluster risk cost of the multi-agent non-cooperative game model at the current moment. Based on the corrected cluster risk cost, the objective function of the multi-agent non-cooperative game model is generated for the next time step.

6. The coordinated control method for multiple microgrids connected to the grid according to claim 5, characterized in that, The step of generating the control execution trust level for each microgrid based on the execution deviation includes: The execution deviation of each microgrid is compared with a preset power deviation threshold. Based on the comparison results, the execution deviation is piecewise nonlinearly mapped to obtain the trust score at the current moment. The current trust score is weighted and summed with the control execution trust score from the previous time step to obtain the current control execution trust score.

7. The coordinated control method for multiple microgrids connected to the grid according to claim 5, characterized in that, The step of using the control execution trust level as a feedback parameter to dynamically correct the cluster risk cost in the multi-agent non-cooperative game model includes: Based on the control execution trust level, the weighting coefficient of the cluster penalty cost in the cluster risk cost of each microgrid is adjusted, and a dynamic feedback additional risk item is generated. The dynamic feedback additional risk item is merged into the cluster risk cost after the weight coefficient is adjusted to obtain the corrected cluster risk cost.

8. A collaborative control system for multiple microgrids connected to the grid, characterized in that, include: The energy flow map construction module is used to acquire multi-dimensional synchronous operation data of multiple microgrid clusters, obtain the internal coupling relationship and power distribution status of the multiple microgrid clusters based on the multi-dimensional synchronous operation data, and generate a real-time dynamic cluster energy flow map. The game model construction module is used to establish a multi-agent non-cooperative game model based on the real-time dynamic cluster energy flow map. The multi-agent non-cooperative game model uses each microgrid as a player, with grid-connected power as the strategy space, and minimizes cluster risk cost as the objective function. The cluster risk cost includes cluster penalty cost, control effort cost, and energy utilization efficiency. The cluster penalty cost is the voltage deviation rate of key nodes in the grid under different strategy combinations obtained from the real-time dynamic cluster energy flow map; the control effort cost is the rate of change of grid-connected power of the microgrid; and the energy utilization efficiency is obtained by calculating the degree of absorption of distributed energy resources by the microgrid. The grid connection command generation module is used to solve the multi-agent non-cooperative game model to obtain the coordinated grid connection power target command; The grid connection command execution module is used to generate a grid connection control sequence according to the coordinated grid connection power target command, and drive each microgrid to perform active grid connection operation under the grid connection control sequence, so as to realize the coordinated grid connection of multiple microgrid clusters to the main grid.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the 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 a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images