Multi-microgrid load frequency control method and device based on dynamic event triggering

Abstract

The invention provides a multi-microgrid load frequency control method based on dynamic event triggering, which can be applied to the technical field of intelligent microgrids. The method comprises the steps that a local sampling error of the micro-grid system is acquired, the local sampling error is obtained based on a global neighborhood error and a local neighborhood error of the micro-grid system, and the local neighborhood error is used for representing a state difference between the micro-grid system and a neighbor micro-grid system at a sampling moment; under the condition that the local sampling error meets a dynamic event triggering rule, updating an initial weight matrix of the action-evaluation neural network based on the local sampling error to obtain a target action-evaluation neural network; and determining a load frequency control strategy of the micro-grid system based on the system state of the micro-grid system at the sampling time and the target action-evaluation neural network. The invention further provides a multi-microgrid load frequency control device based on dynamic event triggering.

Description

Technical Field

[0001] The present invention relates to the technical field of smart microgrids, and more specifically, to a multi-microgrid load frequency control method and device based on dynamic event triggering. Background Art

[0002] The multi-microgrid system is a typical model of the new power system, which is a complex power system composed of multiple microgrids connected and coordinated. In the multi-microgrid system, load frequency control is to adjust the output power of the generator to match the load changes, so as to achieve adaptive regulation of the power load. Therefore, load frequency control plays a key role in maintaining the steady-state frequency at the nominal value during load disturbances.

[0003] Given the complex dynamics and nonlinearity of multi-microgrid systems, traditional methods based on linearized models can only be controlled at specific nominal operating points, and are difficult to cope with the various changes and uncertainties that the system may encounter in actual operation. In addition, traditional methods usually require certain system model information or structural knowledge to construct performance functions and optimization problems. Summary of the invention

[0004] In view of this, the present invention provides a multi-microgrid load frequency control method and device based on dynamic event triggering.

[0005] One aspect of the present invention provides a multi-microgrid load frequency control method based on dynamic event triggering, the control method comprising: obtaining a local sampling error of a microgrid system, wherein the local sampling error is obtained based on a global neighborhood error and a local neighborhood error of the microgrid system; when the local sampling error satisfies a dynamic event triggering rule, based on the local sampling error, updating an initial weight matrix of an action-evaluation neural network to obtain a target action-evaluation neural network; and determining a load frequency control strategy of the microgrid system based on the system state of the microgrid system at the sampling moment and the target action-evaluation neural network.

[0006] According to an embodiment of the present invention, the above-mentioned control method also includes: determining the load frequency control model of the above-mentioned microgrid system based on the respective control parameters of the speed regulator, turbine, power system and controller of the above-mentioned microgrid system; obtaining the local cost function of the above-mentioned microgrid system based on the load frequency control model of the above-mentioned microgrid system and the local neighborhood error of the above-mentioned microgrid system, wherein the above-mentioned local neighborhood error is determined based on the connection relationship between the above-mentioned microgrid system and its neighboring microgrid systems.

[0007] According to an embodiment of the present invention, the local cost function of the microgrid system is obtained based on the load frequency control model of the microgrid system and the local neighborhood error of the microgrid system, including: performing differential transformation on the load frequency control model of the microgrid system to obtain the state space model of the microgrid system; determining the global neighborhood error based on the system matrix, control matrix and the local neighborhood error of the microgrid system of the state space model of the microgrid system; and obtaining the local cost function of the microgrid system based on the global neighborhood error and the differential game reward function of the microgrid system.

[0008] According to an embodiment of the present invention, the microgrid system includes a local event trigger, and the control method further includes: determining a static event triggering rule of the local event trigger based on a local cost function of the microgrid system; updating the static event triggering rule using distributed dynamic variables to obtain the dynamic event triggering rule, wherein the distributed dynamic variables are used to indicate the degree of influence of the historical event triggering state on the event triggering state at the event triggering moment.

[0009] According to an embodiment of the present invention, when the local sampling error satisfies the dynamic event triggering rule, the initial weight matrix of the action-neural evaluation network is updated based on the local sampling error to obtain the target action-evaluation neural network, including: determining the dynamic triggering threshold based on the dynamic event triggering rule; when the local sampling error exceeds the dynamic triggering threshold, obtaining the sampling signal at the event triggering moment; determining the reward signal of the action-evaluation neural network based on the sampling signal at the event triggering moment and the Bellman residual of the microgrid system; and updating the initial weight matrix of the action-evaluation neural network based on the reward signal of the action-evaluation neural network to obtain the target action-evaluation neural network.

[0010] According to an embodiment of the present invention, the above-mentioned initial weight matrix of the action-evaluation neural network is updated based on the reward signal of the above-mentioned action-evaluation neural network to obtain the above-mentioned target action-evaluation neural network, including: processing the reward signal of the above-mentioned action-evaluation neural network by using the least squares method to obtain a weight update rule; based on the above-mentioned weight update rule, updating the above-mentioned initial weight matrix to obtain an updated weight matrix; and determining the above-mentioned target action-evaluation neural network according to the above-mentioned updated weight matrix.

[0011] According to an embodiment of the present invention, the above-mentioned initial weight matrix is ​​updated based on the above-mentioned weight update rule to obtain an updated weight matrix, including: based on the above-mentioned weight update rule, using the gradient descent method to calculate the weight matrix of the current iteration round; when the weight matrix of the above-mentioned current iteration round meets the iteration condition, the above-mentioned updated weight matrix is ​​obtained.

[0012] According to an embodiment of the present invention, the load frequency control strategy of the microgrid system is determined based on the system state of the microgrid system at the sampling moment and the target action-evaluation neural network, including: obtaining an event-triggered control strategy based on the activation function of the target action-evaluation neural network, the updated weight matrix and the sampling signal at the event triggering moment; and determining the load frequency control strategy of the microgrid system according to the event-triggered control strategy.

[0013] Another aspect of the present invention provides a multi-microgrid load frequency control device based on dynamic event triggering, the control device comprising: an acquisition module, used to acquire a local sampling error of a microgrid system, wherein the local sampling error is obtained based on a global neighborhood error and a local neighborhood error of the microgrid system; an update module, used to update an initial weight matrix of an action-neural evaluation network based on the local sampling error when the local sampling error satisfies a dynamic event triggering rule, so as to obtain a target action-evaluation neural network; a determination module, used to determine a load frequency control strategy of the microgrid system based on the system state of the microgrid system at the sampling moment and the target action-evaluation neural network.

[0014] Another aspect of the present invention provides an electronic device, comprising: one or more processors; a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0015] Another aspect of the present invention provides a computer-readable storage medium having a computer program or instruction stored thereon, wherein the computer program or instruction implements the steps of the above method when executed by a processor.

[0016] Another aspect of the present invention provides a computer program product, comprising a computer program or instructions, which implement the steps of the above method when executed by a processor.

[0017] According to an embodiment of the present invention, a multi-microgrid system is regarded as a multi-agent system, and each agent decides whether to trigger an event and perform sampling based on its own local sampling error and global neighborhood error. Through a dynamic triggering mechanism, the agent can dynamically adjust the triggering rules according to the real-time state of the system, thereby reducing unnecessary data transmission and control updates to improve the overall performance and reliability of the system. In addition, based on the connection relationship between the multiple agents obtained by sampling, a reinforcement learning method is used to learn and train the optimal control strategy by observing the state and rewards of the multi-agent system to avoid model dependence. This method is applicable to a variety of complex multi-microgrid systems and can improve the flexibility and control efficiency of the control method. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and other objects, features and advantages of the present invention will become more apparent through the following description of the embodiments of the present invention with reference to the accompanying drawings.

[0019] Figure 1 An exemplary system architecture of a multi-microgrid load frequency control method and device based on dynamic event triggering according to an embodiment of the present invention is shown.

[0020] Figure 2 A flow chart of a multi-microgrid load frequency control method based on dynamic event triggering according to an embodiment of the present invention is shown.

[0021] Figure 3 A schematic diagram of the communication structure of a multi-microgrid system including three microgrids according to a specific embodiment of the present invention is shown.

[0022] Figure 4 A state diagram of neighborhood errors of a multi-microgrid system according to a specific embodiment of the present invention is shown.

[0023] Figure 5 A schematic diagram of event triggering results of a multi-microgrid system based on a dynamic event triggering mechanism according to a specific embodiment of the present invention is shown.

[0024] Figure 6 A block diagram of a multi-microgrid load frequency control device based on dynamic event triggering according to an embodiment of the present invention is shown.

[0025] Figure 7 A block diagram of an electronic device suitable for implementing a multi-microgrid load frequency control method based on dynamic event triggering according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0026] Below, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present invention. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of embodiments of the present invention. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of concepts of the present invention.

[0027] The terms used herein are only for describing specific embodiments and are not intended to limit the present invention. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0028] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0029] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0030] The multi-microgrid (hereinafter referred to as multi-microgrid) system is a typical model of the new power system. The multi-microgrid system is a complex power system composed of multiple microgrids connected and coordinated. In the multi-microgrid system, load frequency control is to adjust the output power of the generator to match the load changes, thereby achieving adaptive regulation of the power load. Therefore, load frequency control plays a key role in maintaining the steady-state frequency at the nominal value during load disturbances.

[0031] Given the complex dynamics and nonlinearity of multi-microgrid systems, traditional linearized model-based methods can only perform control at specific nominal operating points, making it difficult to cope with the various changes and uncertainties that the system may encounter in actual operation. For example, parameters such as load demand and generator output power may change over time, and measurement data may also be affected by noise, missing data, or delays, resulting in low accuracy or stability in actual control strategies. In addition, traditional methods usually require certain system model information or structural knowledge to construct performance functions and optimization problems.

[0032] In view of this, an embodiment of the present invention proposes a model-free multi-microgrid load frequency control method triggered by dynamic events, which regards the multi-microgrid system as a multi-agent system. Each agent decides whether to trigger an event and perform sampling based on its own local sampling error and global neighborhood error. Through the dynamic triggering mechanism, the agent can dynamically adjust the triggering rules according to the real-time state of the system, thereby reducing unnecessary data transmission and control updates to improve the overall performance and reliability of the system. In addition, based on the connection relationship between the multiple agents obtained by sampling, a reinforcement learning method is used to learn and train the optimal control strategy by observing the state and rewards of the multi-agent system to avoid model dependence. This method is applicable to a variety of complex multi-microgrid systems and can improve the flexibility and control efficiency of the control method.

[0033] Specifically, an embodiment of the present invention provides a multi-microgrid load frequency control method based on dynamic event triggering, including: obtaining a local sampling error of a microgrid system; when the local sampling error satisfies a dynamic event triggering rule, based on the local sampling error, updating an initial weight matrix of an action-neural evaluation network to obtain a target action-evaluation neural network; and determining a load frequency control strategy of the microgrid system based on the system state of the microgrid system at the sampling moment and the target action-evaluation neural network.

[0034] It should be noted that the multi-microgrid load frequency control method and device based on dynamic event triggering determined in the embodiment of the present invention can be used in the field of smart microgrid technology. The multi-microgrid load frequency control method and device based on dynamic event triggering determined in the embodiment of the present invention can also be used in any field other than the field of smart microgrid technology, such as artificial intelligence, reinforcement learning, etc. The application field of the multi-microgrid load frequency control method and device based on dynamic event triggering determined in the embodiment of the present invention is not limited.

[0035] In the embodiments of the present invention, the collection, updating, analysis, processing, use, transmission, provision, invention, storage and other aspects of the data involved (for example, including but not limited to user personal information) are in compliance with the provisions of relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. In particular, necessary measures are taken for user personal information to prevent illegal access to user personal information data and maintain the security of user personal information, network security and national security.

[0036] In the embodiment of the present invention, before obtaining or collecting the user's personal information, the user's authorization or consent is obtained.

[0037] Figure 1 The exemplary system architecture of the multi-microgrid load frequency control method and device based on dynamic event triggering according to an embodiment of the present invention is shown. It should be noted that: Figure 1The examples shown are merely examples of system architectures to which the embodiments of the present invention may be applied, to help those skilled in the art understand the technical content of the present invention, but do not mean that the embodiments of the present invention may not be used in other devices, systems, environments or scenarios.

[0038] like Figure 1 As shown, the system architecture according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0039] The user can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, and / or social platform software, etc. (only for example).

[0040] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0041] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0042] It should be noted that the multi-microgrid load frequency control method based on dynamic event triggering provided in the embodiment of the present invention can generally be executed by the server 105. Accordingly, the multi-microgrid load frequency control device based on dynamic event triggering provided in the embodiment of the present invention can generally be set in the server 105. The multi-microgrid load frequency control method based on dynamic event triggering provided in the embodiment of the present invention can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the multi-microgrid load frequency control device based on dynamic event triggering provided in the embodiment of the present invention can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Alternatively, the multi-microgrid load frequency control method based on dynamic event triggering provided in the embodiment of the present invention may also be executed by the first terminal device 101, the second terminal device 102, and the third terminal device 103, or may also be executed by other terminal devices different from the first terminal device 101, the second terminal device 102, and the third terminal device 103. Accordingly, the multi-microgrid load frequency control device based on dynamic event triggering provided in the embodiment of the present invention may also be set in the terminal device 101, 102, or 103, or in other terminal devices different from the first terminal device 101, the second terminal device 102, and the third terminal device 103.

[0043] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0044] Figure 2 A flow chart of a multi-microgrid load frequency control method based on dynamic event triggering according to an embodiment of the present invention is shown.

[0045] like Figure 2 As shown, the method includes operations S210 to S230.

[0046] In operation S210, a local sampling error of the microgrid system is obtained.

[0047] In operation S220, when the local sampling error satisfies the dynamic event triggering rule, the initial weight matrix of the action-neural evaluation network is updated based on the local sampling error to obtain the target action-evaluation neural network.

[0048] In operation S230, a load frequency control strategy of the microgrid system is determined based on the system state of the microgrid system at the sampling moment and the target action-evaluation neural network.

[0049] According to an embodiment of the present invention, for a distributed microgrid system composed of multiple microgrids, each microgrid system can be regarded as an independent intelligent agent to perceive its own state, interact with other intelligent agents, and make decisions and actions based on this information.

[0050] According to an embodiment of the present invention, the local neighborhood error is used to represent the state difference between any microgrid system and its neighboring microgrid system at a specific sampling time. The local neighborhood error can be used to reflect the consistency of the state of any microgrid system and its neighboring microgrid at the sampling time. The local sampling state can be used to represent the local state of any microgrid system in the system at the sampling time.

[0051] According to an embodiment of the present invention, the global neighborhood error is used to represent the state difference between a microgrid system and its neighboring microgrid systems at any time within a time window related to a sampling moment, and the global system state can be used to characterize the state of the entire microgrid system perceived or measured by the microgrid system at any time within the time window related to the sampling moment, wherein the global system state may include the overall frequency, total output power, load conditions, etc. of the system, and the global system state may change in real time, reflecting the dynamic behavior of the system at the current moment.

[0052] According to an embodiment of the present invention, under a distributed trigger setting, each local event trigger may calculate a local sampling error based on a global neighborhood error in a global system state and a local neighborhood error in a local sampling state.

[0053] According to an embodiment of the present invention, the local sampling error can be used to determine the sampling triggering time of each local event trigger. The event triggering rule of each local event trigger corresponds to an adaptive triggering threshold. When the local sampling error exceeds the triggering threshold of the event triggering rule, the local event trigger will generate an event triggering signal, and the local event trigger will obtain new sampling information and calculate a new local sampling error to form an event triggering cycle. At the non-triggering moment, the current value will remain unchanged.

[0054] According to an embodiment of the present invention, in a distributed multi-microgrid system, in order to improve the efficiency and flexibility of event triggering, a dynamic event triggering mechanism can be introduced based on static event triggering rules. This mechanism not only relies on static triggering conditions, but also can dynamically adjust the triggering rules according to the real-time status of the system.

[0055] According to an embodiment of the present invention, when the local sampling error satisfies the dynamic event triggering rule, the system enters the reinforcement learning stage. The new sampling information obtained based on the local event trigger is used as the current state of the microgrid system and as the input of the action-evaluation neural network. The initial weight matrix of the action-evaluation neural network is updated by reinforcement learning and other methods until the loss function of the action-evaluation neural network converges or reaches a preset number of iterations, and an updated target action-evaluation neural network can be obtained. The load frequency control strategy of the microgrid system based on dynamic event triggering is determined based on the updated target action-evaluation neural network.

[0056] Based on this, an embodiment of the present invention proposes a model-free multi-microgrid load frequency control method triggered by dynamic events, which regards the multi-microgrid system as a multi-agent system. Each agent decides whether to trigger an event and perform sampling based on its own local sampling error and global neighborhood error. Through the dynamic triggering mechanism, the agent can dynamically adjust the triggering rules according to the real-time state of the system, thereby reducing unnecessary data transmission and control updates to improve the overall performance and reliability of the system; in addition, based on the connection relationship between the multiple agents obtained by sampling, a reinforcement learning method is used to learn and train the optimal control strategy by observing the state and rewards of the multi-agent system to avoid model dependence. This method is applicable to a variety of complex multi-microgrid systems and can improve the flexibility and control efficiency of the control method.

[0057] According to an embodiment of the present invention, the control method also includes: determining a load frequency control model of the microgrid system based on the respective control parameters of the microgrid system's speed regulator, turbine, power system and controller; and obtaining a local cost function of the microgrid system based on the load frequency control model of the microgrid system and the local neighborhood error of the microgrid system.

[0058] According to an embodiment of the present invention, since the load fluctuation of the power system under normal operation is small, its dynamic behavior can be approximated by a linear model around the working point. Therefore, a load frequency control model of the microgrid system can be constructed based on the state variables of the governor, turbine, power system and controller. In one embodiment, the load frequency control model of each intelligent agent obtained mainly includes four state variables: frequency deviation, increment of generator output power, increment of governor valve position and increment of integral control. Specifically, the load frequency control model of the microgrid system is as follows:

[0059] (1);

[0060] In the formula, Indicates control signal; Indicates the frequency deviation, represents the increment of generator output power, Indicates the increment of the governor valve position, represents the increment of integral control, Represents the load change and is considered as a disturbance signal. represents the time constant of the speed regulator, represents the time constant of the turbine, represents the time constant of the generator, represents the gain, Indicates the speed adjustment coefficient caused by the speed regulator. Represents the integral control gain.

[0061] According to an embodiment of the present invention, the local neighborhood error of each microgrid system can be determined based on the connection relationship between the microgrid system and its neighboring microgrid systems. The local neighborhood error can be used to represent the weighted sum of the state differences between the i-th microgrid system and the neighboring microgrid system, plus the weighted difference between its state and the virtual leader. This definition helps capture the interaction between microgrid systems and their association with the virtual leader. Specifically, the local neighborhood error can be expressed as:

[0062] (2);

[0063] In the formula, represents the local neighborhood error of the i-th microgrid system, represents the connection weight in the adjacency matrix A, which represents the connection relationship between the i-th microgrid system and the j-th microgrid system, x i and x j They represent the state variables of the i-th microgrid system and the j-th microgrid system, x0 represents the state variable of the virtual leader; N i represents the neighbor set of the i-th microgrid system, represents the pinning gain, which is used to represent the connection strength between the i-th microgrid system and the j-th microgrid system. If the agent can obtain information from the leader, then ;otherwise, .

[0064] According to an embodiment of the present invention, the local neighborhood error and the consistency error The convergence is the same, by making the neighborhood error asymptotically stable, that is, , the goal of synchronizing the optimal states of each agent can be achieved.

[0065] According to an embodiment of the present invention, a local cost function of a microgrid system is obtained based on a load frequency control model of the microgrid system and a local neighborhood error of the microgrid system, including: performing a differential transformation on the load frequency control model of the microgrid system to obtain a state space model of the microgrid system; determining a global neighborhood error based on a system matrix, a control matrix and a local neighborhood error of the microgrid system of the state space model of the microgrid system; and obtaining a local cost function of the microgrid system based on the global neighborhood error and a differential game gain function of the microgrid system.

[0066] According to an embodiment of the present invention, the idea of ​​non-zero-sum differential game can be adopted, and the strategic interaction between all game agents is modeled using a communication network to describe which objects can obtain the decision information of the game agents. Specifically, the control input of each microgrid, that is, the agent, is regarded as the strategy of game decision, and a multi-agent collaborative load frequency control model based on graphical differential game can be obtained. For example, the load frequency control model of the microgrid system can be differentially transformed to obtain the state space model of the control model as follows:

[0067] (3);

[0068] In the formula, represents the state space expression of the load frequency control model of the microgrid system, x i (t) represents the state vector of the i-th microgrid system, u i (t) represents the control input vector of the i-th microgrid system, represents the system matrix, represents the control matrix.

[0069] According to an embodiment of the present invention, a global neighborhood error is determined based on a system matrix of a state space model of a microgrid system, a control matrix, and a local neighborhood error of the microgrid system. By stabilizing the global neighborhood error to zero, coordinated control of the load frequency of multiple microgrids can be achieved. Specifically, the global neighborhood error of the system is represented as follows:

[0070] (4);

[0071] In the formula, represents the global neighborhood error of the system, represents the in-degree of the i-th microgrid system, which is the in-degree matrix The diagonal elements of i and u j They represent the control input vectors of the i-th microgrid system and the j-th microgrid system respectively.

[0072] In a specific embodiment of the present invention, since the control goal is to seek the optimal control strategy for each game agent, so that the global neighborhood error of the system is asymptotically stable, and the local cost function of each game agent needs to be optimized. In order to achieve this goal, based on the global neighborhood error of the system and the differential game reward function of the microgrid system, the local cost function of the microgrid system can be constructed. Specifically, the expression of the local cost function of the microgrid system is as follows:

[0073] (5);

[0074] In the formula, represents the local cost function of the i-th microgrid system, Q i and R i is a positive definite weight matrix, represents the control input vector of other microgrid systems except the i-th microgrid system.

[0075] In this specific embodiment, by optimizing the local cost function shown in formula (5), the optimal cooperative control strategy of each intelligent agent can be obtained. The optimal cooperative control strategy of each microgrid system is expressed as follows:

[0076] (6);

[0077] In the formula, represents the optimal coordinated control strategy of the i-th microgrid system, Represents the optimal value function of the local cost function.

[0078] Based on this, the embodiment of the present invention constructs a load frequency control model of the microgrid system, uses the local neighborhood error to capture the interaction between microgrid systems and the association with the virtual leader, and uses the local neighborhood error to derive the local cost function, thereby realizing the coordinated load frequency control of multiple microgrid systems. Therefore, this method provides a new idea and method for the coordinated control of microgrid systems. At the same time, by considering the control input of each microgrid system as a game decision strategy and using the graphical differential game theory for coordinated control, the overall stability of the microgrid system is effectively improved.

[0079] In addition, since the multi-agent collaborative control scheme does not need to rely on the precise model of the system, it enables each agent to adaptively respond to local load requirements and collaborate with neighboring agents to achieve global regulation goals. Therefore, the embodiment of the present invention formulates the optimal strategy for each agent through information exchange between neighbors, so that each agent gradually reaches the desired state, thereby achieving coordination of the entire system.

[0080] According to an embodiment of the present invention, the control method also includes: determining a static event triggering rule of a local event trigger based on a local cost function of the microgrid system; updating the static event triggering rule using distributed dynamic variables to obtain a dynamic event triggering rule, wherein the distributed dynamic variables are used to represent the degree of influence of the historical event triggering state on the event triggering state at the event triggering moment.

[0081] According to an embodiment of the present invention, before determining the event triggering rule of the local event trigger, the initial values ​​of the local neighborhood error and the global neighborhood error are determined, and the local sampling error can be calculated within the triggering interval. Specifically, the calculation formula of the local sampling error is as follows:

[0082] (7);

[0083] In the formula, represents the local sampling error at a certain time t, represents the kth triggering moment, represents the k+1th triggering moment, Represents the local neighborhood error at the kth triggering moment.

[0084] According to an embodiment of the present invention, at the time of event triggering , the trigger error can be reset to zero and the control signal u can be updated i In order to convert the discrete control signal into a continuous control signal, a zero-order holder is used in the trigger interval to keep the control signal unchanged until the next trigger moment arrives. In this way, although the control signal is discrete, the system can approximately consider that the control signal is continuous in the trigger interval.

[0085] According to an embodiment of the present invention, after obtaining the local sampling error, the distributed static event triggering rule can be determined according to the local cost function of each game agent. In a specific embodiment, considering conservative design and stability design, the static event triggering rule can be designed according to the local cost function and conservative parameters to obtain the static event triggering rule as follows:

[0086] (8);

[0087] In the formula, represents the conservative parameter, and is a positive real number.

[0088] According to the embodiment of the present invention, it should be noted that the rule actually corresponds to an adaptive static trigger threshold. If the local sampling error exceeds the static trigger threshold, a new event will be triggered. The corresponding trigger condition can be obtained, which can be as follows:

[0089] (9);

[0090] In the formula, Indicates the static trigger threshold.

[0091] According to an embodiment of the present invention, on the basis of the static event triggering rules, in order to improve the triggering efficiency, a suitable filter coefficient can also be selected for each local event trigger. Specifically, the filter coefficient and the distributed internal dynamic variables can be used to construct a dynamic event triggering rule. Among them, the filter coefficient can be used to measure the influence of the triggering information on the dynamic variables, and the distributed internal dynamic variables can be used to track the current state of the microgrid system and update these states according to past events or trigger data. For example, in a distributed microgrid system, each microgrid system may need to maintain an internal variable representing its data synchronization state, which will be dynamically adjusted according to the data update events of other microgrid systems. Among them, the expression of the distributed internal dynamic variables is as follows:

[0092] (10);

[0093] In the formula, represents distributed internal dynamic variables, represents the filter coefficient, represents the first-order form of the distributed internal dynamic variable. Among them, the initial value of the distributed internal dynamic variable .

[0094] According to an embodiment of the present invention, based on the static event triggering rule shown in formula (8) and the distributed internal dynamic variable shown in formula (10), a bridge parameter can be used to establish a connection between the static triggering rule and the dynamic triggering rule, thereby obtaining a dynamic event triggering rule. Specifically, the expression of the dynamic event triggering rule is as follows:

[0095] (11);

[0096] In the formula, Indicates dynamic event triggering rules. Represents the bridge parameters, where .

[0097] Therefore, the dynamic triggering condition of each game agent can be expressed as:

[0098] (12);

[0099] In the formula, Indicates the dynamic trigger threshold.

[0100] According to an embodiment of the present invention, the control method further includes: determining a dynamic trigger threshold based on a dynamic event trigger rule; and acquiring a sampling signal at the event triggering moment when a local sampling error exceeds the dynamic trigger threshold.

[0101] According to an embodiment of the present invention, by continuously monitoring the local sampling error, the local sampling error is compared with the dynamic trigger condition. Once the local sampling error exceeds the dynamic trigger threshold, a new event is triggered and an event trigger signal is generated at the triggering time. At this time, the event trigger obtains a new local neighborhood error at the triggering time, and uses the local neighborhood error obtained by the new sampling to calculate a new local sampling error. Update the control strategy as shown in formula (6) At the non-triggering moment, the current control strategy remains unchanged. The updated control strategy is as follows:

[0102] (13);

[0103] In the formula, Represents the updated control strategy.

[0104] Based on this, the above steps constitute a complete event-triggered control loop, and each intelligent agent independently updates its strategy, thereby realizing distributed control of multi-microgrid load frequency. This improves the scalability and flexibility of the system, making it easier for the system to adapt to the needs of large-scale microgrid systems. Among them, the static event trigger rule provides basic trigger conditions to ensure the stability and conservatism of the system, while the introduction of distributed dynamic variables and dynamic event trigger rules can improve the trigger efficiency, enabling the system to automatically adjust the trigger conditions according to the current state and historical data, thereby responding more flexibly to changes in local and global states.

[0105] According to an embodiment of the present invention, when a local sampling error satisfies a dynamic event triggering rule, an initial weight matrix of an action-neural evaluation network is updated based on the local sampling error to obtain a target action-evaluation neural network, including: determining a reward signal of the action-evaluation neural network based on a sampling signal at the event triggering moment and a Bellman residual of a microgrid system; and updating an initial weight matrix of the action-evaluation neural network based on the reward signal of the action-evaluation neural network to obtain a target action-evaluation neural network.

[0106] According to an embodiment of the present invention, in order to approximately solve the differential game problem, a multi-agent cooperative load frequency control model based on a graphical differential game as shown in formula (3) can be used to construct an action-evaluation neural network. Specifically, a three-layer feedforward neural network including an input layer, a hidden layer, and an output layer can be used to construct an action-evaluation neural network, wherein the input layer receives state information, the hidden layer performs nonlinear transformation, and the output layer outputs an approximate value function or control strategy.

[0107] According to an embodiment of the present invention, the action-evaluation neural network may include an evaluation network and an action network. The evaluation network may be used to approximate the value function, i.e., the long-term benefit under a given state and strategy. The action network may be used to approximate the control strategy, i.e., the optimal control input under a given state. Its output may be expressed as:

[0108] (14);

[0109] (15);

[0110] In the formula, represents the approximated evaluation network, represents the approximated action network, where and represents the basis function vector, and Represent the initial weight matrices of the evaluation network and action network respectively, and Represents the number of neurons in the hidden layer.

[0111] According to an embodiment of the present invention, for each game agent, considering the system operation cost and the player control cost, a different strategy integral reinforcement learning algorithm can be used to train the action-evaluation neural network to seek an approximate optimal solution.

[0112] Specifically, the process of implementing the heterostrategy integral reinforcement learning algorithm based on the action-evaluation neural network includes two stages: the online measurement stage and the offline learning stage. Among them, the online measurement stage is used to collect system data. In this stage, the agent collects system data under the specified control input, and these system data will be used in the subsequent offline learning stage. The offline learning stage uses the collected system data and the iterative process to update the weight matrix and converge to the optimal solution. Among them, the degree of approximation of the current value function can be evaluated by constructing the Bellman residual, and the reward signal is calculated based on the sampled signal at the event triggering moment and the Bellman residual of the microgrid system. Because the heterostrategy Bellman equation does not explicitly contain the dynamics represented by A and B. Instead, the knowledge of the system model is implicitly reflected through the available measurement data, especially and . Therefore, a prerequisite for iteration is an online measurement phase to collect data.

[0113] In addition, by evaluating the value function using system data generated by any control strategy, the exploration capability of the learning process can be enhanced. Each game agent calculates the reward signal for reinforcement learning based on the system operation cost and control cost.

[0114] In some specific embodiments, the constructed Bellman residual is as follows:

[0115] (16);

[0116] (17);

[0117] (18);

[0118] (19);

[0119] (20);

[0120] In the formula, represents the Bellman residual, represents the weight matrix of the agent, represents the weight matrix of the agent's evaluation network, Represents the weight matrix of the agent's action network, vector , and represents the reinforcement learning signal to be calculated, where and Represent the basis function vectors of the evaluation network and the action network respectively, , is the weight parameter.

[0121] According to the Bellman residual, as shown in formula (16), the reward signal of the evaluation network and the action network can be determined. The reward signal is the reinforcement accumulation in a fixed integral interval, and the expression is as follows:

[0122] (twenty one);

[0123] (twenty two);

[0124] Where: q represents the number of groups of collected data, and It is composed of multiple groups of reinforced cumulative signals and The matrix composed of.

[0125] According to an embodiment of the present invention, based on the reward signal of the action-evaluation neural network, the initial weight matrix of the action-evaluation neural network is updated to obtain the target action-evaluation neural network, including: using the least squares method to process the reward signal of the action-evaluation neural network to obtain a weight update rule; based on the weight update rule, the initial weight matrix is ​​updated to obtain an updated weight matrix; according to the updated weight matrix, the target action-evaluation neural network is determined.

[0126] According to an embodiment of the present invention, the reward signal of the action-evaluation neural network is processed using the least squares method to construct a network weight update rule, wherein the network weight update rule can be as follows:

[0127] (twenty three);

[0128] In the formula, Represents the weight matrix of the agent.

[0129] According to an embodiment of the present invention, the initial weight matrices of the evaluation network and the action network in formula (14) and formula (15) can be updated based on formula (23) to determine the target action-evaluation neural network. The intelligent agent will use the weights of the updated target action-evaluation neural network to determine the control strategy based on event triggering.

[0130] According to an embodiment of the present invention, based on the weight update rule, the initial weight matrix is ​​updated to obtain an updated weight matrix, including: based on the weight update rule, using the gradient descent method to calculate the weight matrix of the current iteration round; when the weight matrix of the current iteration round meets the iteration condition, obtaining the updated weight matrix.

[0131] According to an embodiment of the present invention, based on the system state of the microgrid system at the sampling moment and the target action-evaluation neural network, a load frequency control strategy of the microgrid system is determined, including: obtaining an event-triggered control strategy based on the activation function of the target action-evaluation neural network, the updated weight matrix and the sampling signal at the event triggering moment; and determining the load frequency control strategy of the microgrid system according to the event-triggered control strategy.

[0132] According to an embodiment of the present invention, once the optimal value is reached, the agent can obtain an event-triggered model-free control strategy based on the target action-evaluation neural network and deploy it in a multi-microgrid system or other practical applications. Specifically, based on the activation function of the target action-evaluation neural network, the updated weight matrix and the sampled signal at the event triggering moment, the event-triggered control strategy can be determined as follows:

[0133] (twenty four);

[0134] In the formula, Represents a control strategy based on event triggering, represents the sampling state of the global neighborhood error of the ith agent at the corresponding k triggering time, Represents the activation function of the action network.

[0135] According to an embodiment of the present invention, in order to ensure the convergence of the action-evaluation neural network, detection noise can be introduced into the action strategy to meet the persistent excitation (PE) assumption. Specifically, the control input can be configured as ,in is the initial admissible control, and It is the detection noise. Introducing detection noise can not only enhance the diversity of data, so that the system generates more state-action pairs at runtime, thereby increasing the diversity of data, but also ensure a sufficient data set size. By continuously introducing noise, it can ensure that a sufficient number of data points are collected during the training process.

[0136] According to an embodiment of the present invention, when calculating the matrix When the inverse of Has full column rank. If the matrix is ​​not full column rank, then its inverse matrix will not exist, which will cause the algorithm to fail to execute correctly. In order to meet this condition, the number of collected data points should satisfy , where h ci and h ai are the number of neurons in the hidden layer. This condition ensures the richness and diversity of the data set, which is conducive to the training of the neural network.

[0137] Reference below Figure 3~Figure 5 , the multi-microgrid load frequency control method based on dynamic event triggering is further explained in combination with specific embodiments.

[0138] Figure 3 A schematic diagram of the communication structure of a multi-microgrid system including three microgrids according to a specific embodiment of the present invention is shown.

[0139] Figure 4 A state diagram of neighborhood errors of a multi-microgrid system according to a specific embodiment of the present invention is shown.

[0140] Figure 5 A schematic diagram of event triggering results of a multi-microgrid system based on a dynamic event triggering mechanism according to a specific embodiment of the present invention is shown.

[0141] In a specific embodiment of the present invention, the parameters of the multi-microgrid system are selected as follows: the time constant of the generator , Gain , the time constant of the turbine , time constant of the speed regulator , speed adjustment coefficient caused by the speed regulator .

[0142] like Figure 3 As shown, Figure 3 The communication structure of a multi-microgrid system including three microgrids is shown. The positive weight matrices in the local cost functions of the three microgrids are configured as , , and , , .

[0143] In this specific embodiment, the distributed internal dynamic variables can be selected as a first-order filtering form, and the relevant parameters in the dynamic event triggering rule shown in formula (11) are set to positive real numbers. , conservative parameters , bridge parameters .

[0144] In this specific embodiment, the basis function vector of the evaluation network can be configured as , the basis function vector of the action network can be configured as , the integration time is , the initial value of the distributed internal dynamic variable is set to , the algorithm iteration threshold is 10 -6 In the online measurement phase, the detection noise The initial acceptable strategy is arrive Applied to multi-microgrid system. In the offline learning phase, weight iteration is performed by reusing the collected data.

[0145] Based on the above embodiments, Figure 4 As shown, the neighborhood error is 1 , , Represents the frequency deviation signals of regions 1, 2, and 3 respectively, and the neighborhood error 2 , , The incremental deviation signals of the generator output power in regions 1, 2, and 3 are respectively represented. , , The incremental deviation signals of the governor valve position in regions 1, 2, and 3 are respectively represented. , , They represent the incremental signals of the integral control of areas 1, 2, and 3. It can be seen that the error is adjusted to zero within about 5 seconds after data acquisition in the online measurement stage, and the multi-microgrid system achieves coordinated control.

[0146] like Figure 5 As shown in Figure 2, the sampling process and data communication process based on dynamic event triggering are non-periodic. It is also worth noting that due to the dynamic event triggering mechanism, the controllers of each microgrid system only need to perform 82, 81, and 83 control calculations respectively, which greatly reduces the control calculation and communication resource consumption compared to 300 times in the traditional method.

[0147] It should be noted that, unless it is explicitly stated that there is a sequence of execution between different operations shown in the flowchart in the embodiments of the present invention, or different operations have a sequence of execution in technical implementation, otherwise, the execution order of multiple operations may not be particular, and multiple operations may also be executed simultaneously.

[0148] Figure 6 A block diagram of a multi-microgrid load frequency control device based on dynamic event triggering according to an embodiment of the present invention is shown.

[0149] like Figure 6 As shown, the multi-microgrid load frequency control device based on dynamic event triggering includes an acquisition module 610, an update module 620, and a determination module 630.

[0150] The acquisition module 610 is used to obtain the local sampling error of the microgrid system, wherein the local sampling error is obtained based on the global neighborhood error and the local neighborhood error of the microgrid system, wherein the local neighborhood error is used to represent the state difference between the microgrid system and its neighboring microgrid systems at the sampling moment, and the global neighborhood error is used to represent the state difference between the microgrid system and its neighboring microgrid systems within a time window related to the sampling moment.

[0151] The updating module 620 is used to update the initial weight matrix of the action-neural evaluation network based on the local sampling error when the local sampling error meets the dynamic event triggering rule to obtain the target action-evaluation neural network.

[0152] The determination module 630 is used to determine the load frequency control strategy of the microgrid system based on the system state of the microgrid system at the sampling moment and the target action-evaluation neural network.

[0153] According to an embodiment of the present invention, the multi-microgrid load frequency control device based on dynamic event triggering further includes a model determination module and a local cost function determination module.

[0154] The model determination module is used to determine the load frequency control model of the microgrid system based on the respective control parameters of the speed governor, turbine, power system and controller of the microgrid system.

[0155] The local cost function determination module is used to obtain the local cost function of the microgrid system based on the load frequency control model of the microgrid system and the local neighborhood error of the microgrid system, wherein the local neighborhood error is determined based on the connection relationship between the microgrid system and its neighboring microgrid systems.

[0156] According to an embodiment of the present invention, the local cost function determination module includes a state space model determination submodule, a global neighborhood error determination submodule and a local cost function determination submodule.

[0157] The state space model determination submodule is used to perform differential transformation on the load frequency control model of the microgrid system to obtain the state space model of the microgrid system.

[0158] The global neighborhood error determination submodule is used to determine the global neighborhood error based on the system matrix of the state space model of the microgrid system, the control matrix and the local neighborhood error of the microgrid system.

[0159] The local cost function determination submodule is used to obtain the local cost function of the microgrid system based on the global neighborhood error and the differential game reward function of the microgrid system.

[0160] According to an embodiment of the present invention, the control device further includes a static event triggering rule determination module and a dynamic event triggering rule determination module.

[0161] The static event triggering rule determination module is used to determine the static event triggering rule of the local event trigger based on the local cost function of the microgrid system.

[0162] The dynamic event trigger rule determination module is used to update the static event trigger rule using distributed dynamic variables to obtain the dynamic event trigger rule, wherein the distributed dynamic variables are used to indicate the influence of the historical event trigger state on the event trigger state at the event triggering moment.

[0163] According to an embodiment of the present invention, the updating module 620 includes a dynamic trigger threshold determination submodule, a sampling signal determination submodule, a reward signal determination submodule, and a target neural network determination submodule.

[0164] The dynamic trigger threshold determination submodule is used to determine the dynamic trigger threshold based on the dynamic event trigger rule.

[0165] The sampling signal determination submodule is used to obtain the sampling signal at the event triggering moment when the local sampling error exceeds the dynamic triggering threshold.

[0166] The reward signal determination submodule is used to determine the reward signal of the action-evaluation neural network based on the sampled signal at the event triggering moment and the Bellman residual of the microgrid system.

[0167] The target neural network determination submodule updates the initial weight matrix of the action-evaluation neural network based on the reward signal of the action-evaluation neural network to obtain the target action-evaluation neural network.

[0168] According to an embodiment of the present invention, the target neural network determination submodule includes a processing unit, an updating unit and a target neural network determination unit.

[0169] The processing unit is used to process the reward signal of the action-evaluation neural network by using the least square method to obtain the weight update rule.

[0170] The updating unit is used to update the initial weight matrix based on the weight updating rule to obtain an updated weight matrix.

[0171] The target neural network determination unit is used to determine the target action-evaluation neural network according to the updated weight matrix.

[0172] According to an embodiment of the present invention, the updating unit includes a calculating subunit and a determining subunit.

[0173] The calculation subunit is used to calculate the weight matrix of the current iteration round based on the weight update rule using the gradient descent method.

[0174] The determination subunit is used to obtain an updated weight matrix when the weight matrix of the current iteration round meets the iteration condition.

[0175] According to an embodiment of the present invention, the determination module 630 includes a control strategy determination submodule and a target control strategy determination submodule.

[0176] The control strategy determination submodule is used to obtain a control strategy based on event triggering based on the activation function of the target action-evaluation neural network, the updated weight matrix and the sampling signal at the event triggering moment.

[0177] The target control strategy determination submodule is used to determine the load frequency control strategy of the microgrid system according to the control strategy based on event triggering.

[0178] According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits, or at least part of the functions of any one of them can be implemented in one module. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be split into multiple modules for implementation. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be at least partially implemented as hardware circuits, such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems on chips, systems on substrates, systems on packages, application specific integrated circuits (ASICs), or can be implemented by hardware or firmware in any other reasonable way of integrating or packaging the circuit, or by any one of the three implementation methods of software, hardware, and firmware, or by a proper combination of any of them. Alternatively, according to the embodiments of the present invention, one or more of the modules, submodules, units, and subunits can be at least partially implemented as computer program modules, and when the computer program modules are run, the corresponding functions can be executed.

[0179] For example, any multiple of the acquisition module 610, the update module 620, and the determination module 630 can be combined in one module / unit / sub-unit for implementation, or any one of the modules / units / sub-units can be split into multiple modules / units / sub-units. Alternatively, at least part of the functions of one or more of these modules / units / sub-units can be combined with at least part of the functions of other modules / units / sub-units and implemented in one module / unit / sub-unit. According to an embodiment of the present invention, at least one of the acquisition module 610, the update module 620, and the determination module 630 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, at least one of the acquisition module 610, the update module 620, and the determination module 630 may be at least partially implemented as a computer program module, and when the computer program module is executed, the corresponding function may be executed.

[0180] It should be noted that the multi-microgrid load frequency control device part based on dynamic event triggering in the embodiment of the present invention corresponds to the multi-microgrid load frequency control method part based on dynamic event triggering in the embodiment of the present invention. The description of the multi-microgrid load frequency control device part based on dynamic event triggering specifically refers to the multi-microgrid load frequency control method part based on dynamic event triggering, which will not be repeated here.

[0181] Figure 7 A block diagram of an electronic device suitable for implementing a multi-microgrid load frequency control method based on dynamic event triggering according to an embodiment of the present invention is shown. Figure 7 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0182] like Figure 7 As shown, the computer electronic device according to an embodiment of the present invention includes a processor 701, which can perform various appropriate actions and processes according to the program stored in the read-only memory ROM 702 or the program loaded from the storage part 708 to the random access memory RAM 703. The processor 701 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 701 may also include an on-board memory for caching purposes. The processor 701 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0183] In RAM 703, various programs and data required for the operation of the electronic device are stored. The processor 701, ROM 702 and RAM 703 are connected to each other via a bus 704. The processor 701 performs various operations of the method flow according to the embodiment of the present invention by executing the programs in ROM 702 and / or RAM 703. It should be noted that the program can also be stored in one or more memories other than ROM 702 and RAM 703. The processor 701 can also perform various operations of the method flow according to the embodiment of the present invention by executing the programs stored in the one or more memories.

[0184] According to an embodiment of the present invention, the electronic device may further include an input / output (I / O) interface 705, which is also connected to the bus 704. The electronic device may further include one or more of the following components connected to the input / output (I / O) interface 705: an input portion 706 including a keyboard, a mouse, etc.; an output portion 707 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 708 including a hard disk, etc.; and a communication portion 709 including a network interface card such as a LAN card, a modem, etc. The communication portion 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the input / output (I / O) interface 705 as needed. A removable medium 711, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 710 as needed, so that a computer program read therefrom is installed into the storage portion 708 as needed.

[0185] According to an embodiment of the present invention, the method flow according to an embodiment of the present invention can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable storage medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 709, and / or installed from the removable medium 711. When the computer program is executed by the processor 701, the above-mentioned functions defined in the system of the embodiment of the present invention are executed. According to an embodiment of the present invention, the system, equipment, device, module, unit, etc. described above can be implemented by a computer program module.

[0186] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiment; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present invention is implemented.

[0187] According to an embodiment of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium. For example, it may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device.

[0188] For example, according to an embodiment of the present invention, the computer-readable storage medium may include the ROM 702 and / or the RAM 703 described above and / or one or more memories other than the ROM 702 and the RAM 703 .

[0189] An embodiment of the present invention also includes a computer program product, which includes a computer program, which contains program code for executing the method provided by the embodiment of the present invention. When the computer program product runs on an electronic device, the program code is used to enable the electronic device to implement the multi-microgrid load frequency control method based on dynamic event triggering provided by the embodiment of the present invention.

[0190] When the computer program is executed by the processor 701, the above functions defined in the system / device of the embodiment of the present invention are executed. According to the embodiment of the present invention, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0191] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 709, and / or installed from the removable medium 711. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0192] According to an embodiment of the present invention, the program code for executing the computer program provided by the embodiment of the present invention can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0193] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. In this regard, the boxes in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the boxes may also occur in an order different from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they may sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that the boxes in the block diagram or flowchart, and the combination of the boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions. It will be understood by those skilled in the art that the features recorded in the various embodiments of the present invention can be combined and / or combined in various ways, even if such a combination or combination is not explicitly recorded in the present invention. In particular, without departing from the spirit and teachings of the present invention, the features recorded in the various embodiments of the present invention can be combined and / or combined in various ways. All these combinations and / or combinations fall within the scope of the present invention.

[0194] The embodiments of the present invention are described above. However, these embodiments are only for the purpose of illustration, and are not intended to limit the scope of the present invention. Although each embodiment is described above, it does not mean that the measures in each embodiment cannot be used in combination advantageously. Without departing from the scope of the present invention, those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present invention.

Claims

1. A multi-microgrid load frequency control method based on dynamic event triggering, characterized in that: The control method includes: Obtaining a local sampling error of the microgrid system, wherein the local sampling error is obtained based on a global neighborhood error and a local neighborhood error of the microgrid system, wherein the local neighborhood error is used to represent a state difference between the microgrid system and its neighboring microgrid systems at a sampling moment, and the global neighborhood error is used to represent a state difference between the microgrid system and its neighboring microgrid systems within a time window associated with the sampling moment; When the local sampling error satisfies the dynamic event triggering rule, the initial weight matrix of the action-evaluation neural network is updated based on the local sampling error to obtain a target action-evaluation neural network; Based on the system state of the microgrid system at the sampling moment and the target action-evaluation neural network, a load frequency control strategy of the microgrid system is determined.

2. The control method according to claim 1, characterized in that: The control method further includes: Determining a load frequency control model of the microgrid system based on respective control parameters of a speed regulator, a turbine, a power system, and a controller of the microgrid system; A local cost function of the microgrid system is obtained based on a load frequency control model of the microgrid system and a local neighborhood error of the microgrid system, wherein the local neighborhood error is determined based on a connection relationship between the microgrid system and its neighboring microgrid systems.

3. The control method according to claim 2, characterized in that: The obtaining of a local cost function of the microgrid system based on the load frequency control model of the microgrid system and the local neighborhood error of the microgrid system includes: Performing differential transformation on the load frequency control model of the microgrid system to obtain a state space model of the microgrid system; Determining the global neighborhood error based on a system matrix and a control matrix of a state-space model of the microgrid system and a local neighborhood error of the microgrid system; Based on the global neighborhood error and the differential game reward function of the microgrid system, a local cost function of the microgrid system is obtained.

4. The control method according to claim 3, characterized in that: The microgrid system includes a local event trigger, and the control method further includes: Determining a static event triggering rule of the local event trigger based on a local cost function of the microgrid system; The static event triggering rule is updated using distributed dynamic variables to obtain the dynamic event triggering rule, wherein the distributed dynamic variables are used to represent the degree of influence of the historical event triggering state on the event triggering state at the event triggering moment.

5. The control method according to claim 1, characterized in that: When the local sampling error satisfies the dynamic event triggering rule, the initial weight matrix of the action-neural evaluation network is updated based on the local sampling error to obtain the target action-evaluation neural network, including: Determining a dynamic trigger threshold based on the dynamic event triggering rule; When the local sampling error exceeds the dynamic trigger threshold, acquiring a sampling signal at the event triggering moment; Determining a reward signal of the action-criteria neural network based on the sampled signal at the event triggering moment and the Bellman residual of the microgrid system; Based on the reward signal of the action-evaluation neural network, the initial weight matrix of the action-evaluation neural network is updated to obtain the target action-evaluation neural network.

6. The control method according to claim 5, characterized in that: The updating of the initial weight matrix of the action-evaluation neural network based on the reward signal of the action-evaluation neural network to obtain the target action-evaluation neural network includes: Processing the reward signal of the action-evaluation neural network using the least squares method to obtain a weight update rule; Based on the weight update rule, the initial weight matrix is ​​updated to obtain an updated weight matrix; The target action-evaluation neural network is determined according to the updated weight matrix.

7. The control method according to claim 6, characterized in that: The updating of the initial weight matrix based on the weight updating rule to obtain an updated weight matrix includes: Based on the weight update rule, the weight matrix of the current iteration round is calculated using the gradient descent method; When the weight matrix of the current iteration round meets the iteration condition, the updated weight matrix is ​​obtained.

8. The control method according to claim 6, characterized in that: The determining of a load frequency control strategy of the microgrid system based on the system state of the microgrid system at the sampling moment and the target action-evaluation neural network includes: Obtaining an event-triggered control strategy based on the activation function of the target action-evaluation neural network, the updated weight matrix, and the sampled signal at the event triggering moment; According to the event-triggered control strategy, a load frequency control strategy of the microgrid system is determined.

9. A multi-microgrid load frequency control device based on dynamic event triggering, characterized in that: The control device comprises: an acquisition module, configured to acquire a local sampling error of the microgrid system, wherein the local sampling error is obtained based on a global neighborhood error and a local neighborhood error of the microgrid system, wherein the local neighborhood error is used to represent a state difference between the microgrid system and its neighboring microgrid systems at a sampling moment, and the global neighborhood error is used to represent a state difference between the microgrid system and its neighboring microgrid systems within a time window associated with the sampling moment; An updating module, configured to update an initial weight matrix of the action-neural evaluation network based on the local sampling error to obtain a target action-evaluation neural network when the local sampling error satisfies a dynamic event triggering rule; A determination module is configured to determine a load frequency control strategy of the microgrid system based on the system state of the microgrid system at the sampling moment and the target action-evaluation neural network.

10. An electronic device comprising: one or more processors; a storage device for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 8.

Images