Power distribution network topology reconstruction and distributed resource optimization configuration method and system

Abstract

The invention discloses a power distribution network topology reconstruction and distributed resource optimization configuration method and system. The method comprises the following steps: representing a power distribution network topology structure and distributed resources as a graph structure; constructing a neural network model for processing the dynamic graph structure by adopting a graph attention network so as to extract features of topology and resource distribution of the power distribution network; converting a power distribution network topology reconstruction and distributed resource configuration problem into a reinforcement learning problem based on the extracted features of the power distribution network topology and resource distribution; and designing a reward function and carrying out training by using a deep reinforcement learning algorithm to realize collaborative optimization of topology reconstruction and distributed resource configuration of the power distribution network. The invention can overcome the isolation of the two problems in the traditional method and the limitation of the dynamic change of the operating environment of the power distribution network, and provides a more flexible and adaptive solution, thereby remarkably improving the operating efficiency and reliability of the power distribution network, and improving the reliability of the power distribution network. And a powerful technical support is provided for the development of the power distribution network towards the intelligent management direction.

Description

Technical Field

[0001] The present invention relates to the technical field of resource optimization configuration, and in particular to a method and system for power distribution network topology reconstruction and distributed resource optimization configuration. Background Art

[0002] With the development of social economy and technological progress, the demand for electricity continues to grow, especially the large-scale access of new loads such as distributed energy (such as solar photovoltaic panels, wind turbines) and electric vehicles, which has brought new challenges to the traditional distribution network. These new loads not only increase the load of the distribution network, but also change its load characteristics, making the operating environment of the distribution network more complex and changeable.

[0003] In traditional distribution network management methods, topology reconstruction and distributed resource optimization configuration are usually considered separately. Distribution network topology reconstruction mainly optimizes the network structure by changing the switch state to reduce network loss and improve power supply reliability, while distributed resource optimization configuration is based on the existing network structure to reasonably arrange the location and capacity of distributed power sources to achieve the best economic or environmental benefits. Although these two methods have improved the operating efficiency of the distribution network to a certain extent, they are carried out independently without considering the mutual influence and coupling relationship between the two, which makes it difficult to achieve the global optimal goal in practical applications. In addition, most of the existing optimization models and algorithms are based on static or semi-static data, and lack sufficient adaptability and flexibility for dynamic changes in the distribution network operation environment. For example, changes in weather conditions will affect the power generation of distributed energy, and the uncertainty of user electricity consumption behavior will also lead to deviations in load forecasting. These factors will affect the stable operation of the distribution network. Therefore, it is particularly important to develop a new method that can simultaneously consider distribution network topology reconstruction and distributed resource optimization configuration, and has self-learning and adaptive capabilities. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method and system for distribution network topology reconstruction and distributed resource optimization configuration, aiming to achieve coordinated optimization of distribution network topology reconstruction and distributed resource optimization configuration, fully consider the coupling relationship between the two to construct an optimization model that can adapt to the dynamic characteristics of the distribution network, and realize real-time optimization decision-making.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for distribution network topology reconstruction and distributed resource optimization configuration, comprising: representing the distribution network topology structure and distributed resources as a graph structure; using a graph attention network to construct a neural network model for processing dynamic graph structures to extract the characteristics of the distribution network topology and resource distribution; based on the extracted distribution network topology and resource distribution characteristics, converting the distribution network topology reconstruction and distributed resource configuration problems into reinforcement learning problems; designing a reward function and using a deep reinforcement learning algorithm for training to achieve coordinated optimization of distribution network topology reconstruction and distributed resource configuration.

[0008] As a preferred solution of the method for topology reconstruction of the distribution network and optimal configuration of distributed resources described in the present invention, wherein: the representation of the distribution network topology structure and distributed resources as a graph structure includes:

[0009] The distribution network topology and distributed resources are represented as an undirected graph G = (V, E), where V is a node set, E is an edge set, a node v∈V represents a bus or distributed resource, and an edge e∈E represents a line or connection relationship;

[0010] The node feature vector includes node type, voltage amplitude and phase angle, active power and reactive power, node capacity and limitation, and the edge feature vector includes line impedance, line capacity, and switch status.

[0011] As a preferred solution of the method for topology reconstruction of distribution network and optimal configuration of distributed resources described in the present invention, the neural network model for processing dynamic graph structure using graph attention network includes:

[0012] For node i, calculate the attention coefficient between it and its neighbor node j;

[0013] Using the attention coefficient to perform weighted aggregation on the features of neighboring nodes;

[0014] A multi-head attention mechanism is used to characterize the features of the weighted aggregated neighbor nodes;

[0015] The features of distribution network topology and resource distribution are gradually extracted by stacking multiple layers of graph attention networks.

[0016] As a preferred solution of the method for topology reconstruction of distribution network and optimal configuration of distributed resources described in the present invention, the method of converting the problem of topology reconstruction of distribution network and distributed resource configuration into a reinforcement learning problem includes:

[0017] The state space S includes the distribution network topology, node voltage, line power flow, and distributed resource output. The action space A includes switch operations and distributed resource scheduling instructions. The transfer function P(s'|s,a) describes the probability that the system transfers from the current state s to the new state s' after executing action a. The reward function R(s,a) evaluates the immediate return of executing action a in state s, and the discount factor γ balances the immediate reward and long-term benefits.

[0018] As a preferred solution of the method for topology reconstruction of distribution network and optimal configuration of distributed resources described in the present invention, the design reward function includes:

[0019] Design a reward function that takes multiple objectives into consideration. The formula is:

[0020] R(s,a)=w1R loss +w2R voltage +w3R balance +w4R DER

[0021]

[0022] R voltage =-∑ i∈V (V i -V ref ) 2

[0023]

[0024] Among them, R loss represents the network loss reward term, R voltage Represents the voltage deviation bonus, R balance represents the load balancing reward, R DER represents the distributed resource utilization bonus item, w1, w2, w3, w4 are the weight coefficients of each item, r ij is the resistance of line (i, j), P ij and Q ij are the active and reactive power of the line, V i is the voltage amplitude of node i, V ref is the reference voltage, L i and are the actual load and maximum load capacity of load node i, std(·) represents the standard deviation, P i DER and P i DER,max are the actual output and maximum output of distributed resource i respectively.

[0025] As a preferred solution of the method for topology reconstruction of distribution network and optimal configuration of distributed resources described in the present invention, the training using deep reinforcement learning algorithm includes:

[0026] Initialize the policy network and value network parameters;

[0027] Use the current policy πθ to interact with the environment and collect trajectory data;

[0028] Calculate the advantage estimate A at each time step t ;

[0029] By maximizing the objective function L CLIP (θ) Update the policy network parameters and value network parameters to minimize the value estimation error;

[0030] Repeat the above process until convergence.

[0031] As a preferred solution of the method for distribution network topology reconstruction and distributed resource optimization configuration described in the present invention, the coordinated optimization of distribution network topology reconstruction and distributed resource configuration includes:

[0032] Obtain real-time operation data of the distribution network from the data acquisition and monitoring system;

[0033] A neural network model that processes dynamic graph structures using a graph attention network is used to estimate the current state of the distribution network.

[0034] Generate optimized actions based on the current state using the trained policy network;

[0035] Using the value network to evaluate the optimization actions and select the best actions;

[0036] The optimization decisions are sent to the execution system for field testing and performance evaluation, and feedback from the execution results is collected to continuously optimize and update the model.

[0037] In a second aspect, the present invention provides a distribution network topology reconstruction and distributed resource optimization configuration system, comprising:

[0038] A distribution network graph representation module is used to represent the distribution network topology and distributed resources as a graph structure;

[0039] The distribution network feature extraction module is used to construct a neural network model for processing dynamic graph structures using a graph attention network to extract the features of the distribution network topology and resource distribution;

[0040] A problem conversion module, for converting the distribution network topology reconstruction and distributed resource configuration problems into reinforcement learning problems based on the extracted distribution network topology and resource distribution features;

[0041] The collaborative optimization module is used to design the reward function and use the deep reinforcement learning algorithm for training to achieve collaborative optimization of distribution network topology reconstruction and distributed resource configuration.

[0042] In a third aspect, the present invention provides a computing device, comprising:

[0043] Memory and processor;

[0044] The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the distribution network topology reconstruction and distributed resource optimization configuration method are implemented.

[0045] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the method for distribution network topology reconstruction and distributed resource optimization configuration.

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention provides a method and system for distribution network topology reconstruction and distributed resource optimization configuration. By representing the distribution network topology structure and distributed resources as a graph structure, and using a graph attention network to construct a neural network model for processing dynamic graph structures, the characteristics of the distribution network topology and resource distribution can be effectively extracted, and the distribution network topology reconstruction and distributed resource configuration problems can be transformed into reinforcement learning problems. Not only the intrinsic connection between the distribution network topology and resource distribution is fully considered, but also the coordinated optimization of the distribution network topology reconstruction and distributed resource configuration is achieved by designing a reasonable reward function and using a deep reinforcement learning algorithm for training. The present invention can overcome the isolation of the two problems in the traditional method, as well as the limitations in the face of dynamic changes in the distribution network operating environment, and provide a more flexible and adaptive solution, thereby significantly improving the operating efficiency and reliability of the distribution network, and providing strong technical support for the development of the distribution network towards intelligent management. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0048] Figure 1 A schematic diagram of the overall process logic of the method for topology reconstruction of a distribution network and optimal configuration of distributed resources according to an embodiment of the present invention;

[0049] Figure 2The present invention is a flowchart of the coordinated optimization of the distribution network topology reconstruction and distributed resource optimization configuration method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0051] Example 1

[0052] Reference Figure 1-Figure 2 In one embodiment of the present invention, a method for topology reconstruction of a distribution network and optimal configuration of distributed resources is provided. Figure 1 The specific steps shown include:

[0053] S100: Represent the distribution network topology and distributed resources as a graph structure;

[0054] S200: A neural network model for processing dynamic graph structures is constructed using a graph attention network to extract the characteristics of distribution network topology and resource distribution;

[0055] S300: Based on the extracted distribution network topology and resource distribution features, the distribution network topology reconstruction and distributed resource configuration problems are transformed into reinforcement learning problems;

[0056] S400: Design reward functions and use deep reinforcement learning algorithms for training to achieve coordinated optimization of distribution network topology reconstruction and distributed resource configuration.

[0057] It should be noted that the present invention provides a method and system for topological reconstruction of distribution network and optimal configuration of distributed resources. By representing the topological structure of distribution network and distributed resources as a graph structure, and using a graph attention network to construct a neural network model for processing dynamic graph structures, the characteristics of distribution network topology and resource distribution can be effectively extracted, and the distribution network topological reconstruction and distributed resource configuration problems can be transformed into reinforcement learning problems. Not only the intrinsic connection between distribution network topology and resource distribution is fully considered, but also the coordinated optimization of distribution network topological reconstruction and distributed resource configuration is achieved by designing a reasonable reward function and using a deep reinforcement learning algorithm for training. The present invention can overcome the isolation of the two problems in the traditional method, as well as the limitations in the face of dynamic changes in the distribution network operating environment, and provide a more flexible and adaptive solution, thereby significantly improving the operating efficiency and reliability of the distribution network, and providing strong technical support for the development of the distribution network towards intelligent management.

[0058] In the embodiment of the present application, the above step S100 represents the distribution network topology and distributed resources as a graph structure and includes the following sub-steps A1-A2:

[0059] In A1: collect and clean the historical operation data of the distribution network, where the historical operation data of the distribution network includes topology structure, load curve, distributed resource output, etc.;

[0060] In A2: the distribution network topology and distributed resources are represented as an undirected graph G = (V, E), where V is a node set, E is an edge set, a node v∈V represents a bus or distributed resource, and an edge e∈E represents a line or connection relationship;

[0061] Specifically, each node and edge has a corresponding feature vector to describe its attributes and status. The node feature vector includes node type (bus, load, distributed generation, energy storage, etc.), voltage amplitude and phase angle, active power and reactive power, node capacity and limitation. The edge feature vector includes line impedance, line capacity, and switch status.

[0062] It should be noted that the above step S100, by representing the distribution network topology and distributed resources as a graph structure, can intuitively and comprehensively capture the relationship between each node and edge in the distribution network, including the connection method between different devices and their attribute information. This is not only conducive to clearly displaying the overall architecture of the distribution network, but also lays a solid foundation for the subsequent use of advanced algorithms such as graph attention networks to process dynamic graph structures, enabling the system to more accurately identify and analyze complex interaction patterns within the distribution network, thereby providing strong support for achieving efficient distribution network topology reconstruction and distributed resource configuration.

[0063] In the embodiment of the present application, the above step S200 uses a graph attention network to construct a neural network model for processing a dynamic graph structure to extract the characteristics of the distribution network topology and resource distribution, including:

[0064] The calculation process of the graph attention network layer is:

[0065] For node i, calculate the attention coefficient between it and its neighbor node j:

[0066] α ij =soft max j (Leaky Re LU(a T [Wh i ||Wh j ]))

[0067] Among them, h i and h jare the feature vectors of nodes i and j respectively, W is a weight matrix used to transform the node feature vector, a is a learnable vector used to calculate the linear combination in the attention mechanism, and || means concatenating the two transformed feature vectors together;

[0068] Use the attention coefficient to perform weighted aggregation on the features of neighbor nodes:

[0069] h′ i =σ(∑ j α ij W j )

[0070] Among them, σ is the activation function (such as ReLU), which is used to introduce nonlinear characteristics, α ij is the attention coefficient between node i and node j, which is used to weight the feature vector of neighbor node j;

[0071] A multi-head attention mechanism is used to characterize the features of weighted aggregated neighbor nodes:

[0072]

[0073] in, is the attention coefficient between node i and node j in the kth attention head, W k is the weight matrix in the kth attention head, used to transform the node feature vector, || k It means concatenating the output feature vectors of different attention heads k;

[0074] By stacking multiple layers of graph attention networks, the features of distribution network topology and resource distribution are gradually extracted. The output of the last layer of graph attention network is used as the global representation of the distribution network for subsequent decision-making processes.

[0075] It should be noted that the above step S200 adopts a graph attention network to construct a neural network model for processing dynamic graph structures, which can efficiently extract key topological and resource distribution features from the graph structure of the distribution network. This model is particularly good at capturing complex and nonlinear relationships between nodes, and can maintain good adaptability even when the network structure changes, thereby providing high-quality data support for the subsequent transformation of distribution network topology reconstruction and distributed resource allocation problems into reinforcement learning problems, ensuring that the actual structural characteristics of the distribution network can be fully utilized during the optimization process, thereby improving the accuracy and practicality of the optimization results.

[0076] In the embodiment of the present application, the above step S300 converts the distribution network topology reconstruction and distributed resource configuration problems into reinforcement learning problems based on the extracted distribution network topology and resource distribution features, including:

[0077] The distribution network topology reconstruction and distributed resource optimization configuration problems are transformed into a Markov decision process (MDP), which is defined as follows:

[0078] State space S: includes distribution network topology, node voltage, line power flow, and distributed resource output;

[0079] Action space A: includes switch operations and distributed resource scheduling instructions;

[0080] Transition function P(s'|s,a): describes the probability that the system will transition from the current state s to the new state s' after executing action a;

[0081] Reward function R(s,a): evaluates the immediate return of performing action a in state s;

[0082] Discount factor γ: Balances immediate rewards and long-term gains.

[0083] It should be noted that the above step S300 can not only formalize complex optimization objectives and constraints into mathematical models, but also effectively integrate the static and dynamic information of the distribution network through the reinforcement learning framework, so that the optimization algorithm can learn the optimal decision-making strategy through continuous trial and error, thereby realizing the coordinated optimization of distribution network topology reconstruction and distributed resource allocation, improving the overall performance and adaptability of the system, and ensuring that the distribution network can still maintain efficient and reliable operation in the face of various changes in the operating environment.

[0084] In the embodiment of the present application, the above step S400 includes the following sub-steps D1-D3;

[0085] In D1: Design a reward function that takes multiple objectives into account;

[0086] In D2: the model is trained using the proximal strategy optimization algorithm;

[0087] In D3: Based on the trained model, the coordinated optimization of distribution network topology reconstruction and distributed resource configuration is achieved.

[0088] In a feasible implementation, the reward function formula that comprehensively considers multiple objectives is expressed as:

[0089] R(s,a)=w1R loss +w2R voltage +w3R balance +w4R DER

[0090]

[0091] R voltage =-Σ i∈V (Vi -V ref ) 2

[0092]

[0093] Among them, R loss represents the network loss reward term, R voltage Represents the voltage deviation bonus, R balance represents the load balancing reward, R DER represents the distributed resource utilization bonus item, w1, w2, w3, w4 are the weight coefficients of each item, r ij is the resistance of line (i, j), P ij and Q ij are the active and reactive power of the line, V i is the voltage amplitude of node i, V ref is the reference voltage, L i and are the actual load and maximum load capacity of load node i, std(·) represents the standard deviation, P i DER and P i DER,max are the actual output and maximum output of distributed resource i respectively.

[0094] In a feasible implementation, the model is trained using a proximal policy optimization algorithm. The proximal policy optimization algorithm introduces a truncated objective function to ensure monotonous performance improvement while avoiding instability caused by excessive policy updates. The objective function of the proximal policy optimization algorithm is calculated as:

[0095] L CLIP (θ)=E[min(r t (θ)A t ,clip(r t (θ), 1-∈, 1+∈)A t ]

[0096] in, represents the probability ratio of the new and old strategies, A t is the advantage function, ε is the cutoff parameter;

[0097] The specific steps for training the model using the proximal strategy optimization algorithm are:

[0098] Initialize the policy network and value network parameters;

[0099] Use the current policy πθ to interact with the environment and collect trajectory data;

[0100] Calculate the advantage estimate A at each time step t;

[0101] By maximizing the objective function L CLIP (θ) Update the policy network parameters and value network parameters to minimize the value estimation error;

[0102] Repeat the above process until convergence.

[0103] In a feasible implementation, achieving the coordinated optimization of distribution network topology reconstruction and distributed resource configuration includes:

[0104] Obtain real-time operation data of the distribution network from the data acquisition and monitoring system;

[0105] A neural network model that processes dynamic graph structures using a graph attention network is used to estimate the current state of the distribution network.

[0106] Generate optimized actions based on the current state using the trained policy network;

[0107] Use the value network to evaluate optimization actions and select the best actions;

[0108] The optimization decisions are sent to the execution system for field testing and performance evaluation, and feedback from the execution results is collected to continuously optimize and update the model.

[0109] It should be noted that the above step S400 can automatically adjust the optimization strategy according to the actual operation of the distribution network, and find the optimal solution through continuous iterative learning, which not only improves the flexibility and adaptability of the optimization process, but also can effectively deal with the uncertainty and dynamic changes in the distribution network operation environment, ensuring that the optimization results meet the current operation needs and have long-term sustainability, thereby significantly improving the operation efficiency and reliability of the distribution network.

[0110] For example, Figure 2 The figure shows the specific implementation process of the distribution network topology reconstruction and distributed resource optimization configuration method based on graph neural network proposed in this embodiment, and the specific steps include:

[0111] (1) Data preprocessing: Collect and clean the historical operation data of the distribution network, including topology, load curve, and distributed resource output;

[0112] (2) Graph representation construction: Map the distribution network into a graph structure and define the feature vectors of nodes and edges;

[0113] (3) Graph neural network model design: Construct a GAT-based neural network model to extract high-level features of the distribution network;

[0114] (4) Construction of reinforcement learning environment: define the state space, action space and reward function to realize the distribution network simulation environment;

[0115] (5) PPO algorithm implementation: Develop a deep reinforcement learning algorithm based on PPO, including a policy network and a value network;

[0116] (6) Model training:

[0117] a. Initialize the policy network and value network parameters;

[0118] b. Use the current strategy to interact with the environment and collect trajectory data;

[0119] c. Calculate the advantage estimate;

[0120] d. Update the policy network and value network parameters;

[0121] e. Repeat steps bd until convergence;

[0122] (7) Model evaluation: Evaluate the performance of the trained model on the test set;

[0123] (8) Decision support system development: integrating trained models to achieve real-time optimization of decision making;

[0124] (9) System deployment and testing: Deploy the decision support system to the actual distribution network for field testing and performance evaluation;

[0125] (10) Continuous optimization: Continuously optimize and update the model based on actual operation data and feedback;

[0126] Therefore, the present invention provides a method and system for distribution network topology reconstruction and distributed resource optimization configuration. By representing the distribution network topology structure and distributed resources as a graph structure, and using a graph attention network to construct a neural network model for processing dynamic graph structures, the characteristics of the distribution network topology and resource distribution can be effectively extracted, and the distribution network topology reconstruction and distributed resource configuration problems can be transformed into reinforcement learning problems. Not only the intrinsic connection between the distribution network topology and resource distribution is fully considered, but also the coordinated optimization of the distribution network topology reconstruction and distributed resource configuration is achieved by designing a reasonable reward function and using a deep reinforcement learning algorithm for training. The present invention can overcome the isolation of the two problems in the traditional method, as well as the limitations in the face of dynamic changes in the distribution network operating environment, and provide a more flexible and adaptive solution, thereby significantly improving the operating efficiency and reliability of the distribution network, and providing strong technical support for the development of the distribution network towards intelligent management.

[0127] Example 2

[0128] This embodiment provides a distribution network topology reconstruction and distributed resource optimization configuration system, including a distribution network diagram representation module, a distribution network feature extraction module, a problem conversion module and a collaborative optimization module;

[0129] Specifically, the distribution network graph representation module is used to represent the distribution network topology and distributed resources as a graph structure;

[0130] Specifically, the distribution network feature extraction module is used to construct a neural network model for processing dynamic graph structures using a graph attention network to extract the features of the distribution network topology and resource distribution;

[0131] Specifically, the problem conversion module is used to convert the distribution network topology reconstruction and distributed resource configuration problems into reinforcement learning problems based on the extracted distribution network topology and resource distribution features;

[0132] Specifically, the collaborative optimization module is used to design the reward function and use the deep reinforcement learning algorithm for training to achieve collaborative optimization of distribution network topology reconstruction and distributed resource configuration.

[0133] It should be noted that the technical solution of the system for distribution network topology reconstruction and distributed resource optimization configuration belongs to the same concept as the technical solution of the above-mentioned distribution network topology reconstruction and distributed resource optimization configuration method. The details not described in detail in the technical solution of the distribution network topology reconstruction and distributed resource optimization configuration system in this embodiment can be found in the description of the technical solution of the above-mentioned distribution network topology reconstruction and distributed resource optimization configuration method.

[0134] The above-mentioned unit modules may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above-mentioned modules.

[0135] This embodiment also provides a computing device, which includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for topological reconstruction of a distribution network and optimal configuration of distributed resources is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, a trackball or a touchpad set on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0136] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method proposed in the above embodiment is implemented.

[0137] The storage medium proposed in this embodiment and the method proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0138] Through the above description of the implementation methods, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ReadOnly, Memory, ROM), random access memory (RandomAccess Memory, RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform the methods of various embodiments of the present invention.

[0139] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0140] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0141] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0142] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0143] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0144] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0145] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for topology reconstruction of a distribution network and optimal configuration of distributed resources, characterized in that: include: Represent the distribution network topology and distributed resources as a graph structure; A graph attention network is used to build a neural network model that processes dynamic graph structures to extract the characteristics of distribution network topology and resource distribution. Based on the extracted distribution network topology and resource distribution features, the distribution network topology reconstruction and distributed resource configuration problems are transformed into reinforcement learning problems; Design a reward function and use deep reinforcement learning algorithm for training to achieve coordinated optimization of distribution network topology reconstruction and distributed resource configuration.

2. The method for topology reconstruction and distributed resource optimization configuration of the distribution network according to claim 1, characterized in that: The method of representing the distribution network topology and distributed resources as a graph structure includes: The distribution network topology and distributed resources are represented as an undirected graph G = (V, E), where V is a node set, E is an edge set, a node v∈V represents a bus or distributed resource, and an edge e∈E represents a line or connection relationship; The node feature vector includes node type, voltage amplitude and phase angle, active power and reactive power, node capacity and limitation, and the edge feature vector includes line impedance, line capacity, and switch status.

3. The method for topology reconstruction of the distribution network and optimal configuration of distributed resources according to claim 2, characterized in that: The neural network model constructed by using the graph attention network to process the dynamic graph structure includes: For node i, calculate the attention coefficient between it and its neighbor node j; Using the attention coefficient to perform weighted aggregation on the features of neighboring nodes; A multi-head attention mechanism is used to characterize the features of the weighted aggregated neighbor nodes; The features of distribution network topology and resource distribution are gradually extracted by stacking multiple layers of graph attention networks.

4. The method for topology reconstruction of the distribution network and optimal configuration of distributed resources according to claim 3, characterized in that: The transformation of the distribution network topology reconstruction and distributed resource configuration problems into reinforcement learning problems includes: The state space S includes the distribution network topology, node voltage, line power flow, and distributed resource output. The action space A includes switch operations and distributed resource scheduling instructions. The transfer function P(s'|s,a) describes the probability that the system transfers from the current state s to the new state s' after executing action a. The reward function R(s,a) evaluates the immediate return of executing action a in state s, and the discount factor γ balances the immediate reward and long-term benefits.

5. The method for topology reconstruction of distribution network and optimal configuration of distributed resources according to claim 4, characterized in that: The design reward function includes: Design a reward function that takes multiple objectives into consideration. The formula is: R(s,a)=w1R loss +w2R voltage +w3R balance +w4R DER R voltage =-Σ i∈V (V i -V ref ) 2 Among them, R loss represents the network loss reward term, R voltage Represents the voltage deviation bonus, R balance represents the load balancing reward, R DER represents the distributed resource utilization bonus item, w1, w2, w3, w4 are the weight coefficients of each item, r ij is the resistance of line (i, j), P ij and Q ij are the active and reactive power of the line, V i is the voltage amplitude of node i, V ref is the reference voltage, L i and L i max are the actual load and maximum load capacity of load node i, respectively, std(·) represents the standard deviation, and are the actual output and maximum output of distributed resource i respectively.

6. The method for topology reconstruction of the distribution network and optimal configuration of distributed resources according to claim 5, characterized in that: The training using the deep reinforcement learning algorithm includes: Initialize the policy network and value network parameters; Use the current policy πθ to interact with the environment and collect trajectory data; Calculate the advantage estimate A at each time step t ; By maximizing the objective function L CLIP (θ) Update the policy network parameters and value network parameters to minimize the value estimation error; Repeat the above process until convergence.

7. The method for topology reconstruction of the distribution network and optimal configuration of distributed resources according to claim 6, characterized in that: The method of realizing the coordinated optimization of distribution network topology reconstruction and distributed resource configuration includes: Obtain real-time operation data of the distribution network from the data acquisition and monitoring system; A neural network model that processes dynamic graph structures using a graph attention network is used to estimate the current state of the distribution network. Generate optimized actions based on the current state using the trained policy network; Using the value network to evaluate the optimization actions and select the best actions; The optimization decisions are sent to the execution system for field testing and performance evaluation, and feedback from the execution results is collected to continuously optimize and update the model.

8. A system using the method for topology reconstruction of a distribution network and optimal configuration of distributed resources as claimed in any one of claims 1 to 7, characterized in that: include: A distribution network graph representation module is used to represent the distribution network topology and distributed resources as a graph structure; The distribution network feature extraction module is used to construct a neural network model for processing dynamic graph structures using a graph attention network to extract the features of the distribution network topology and resource distribution; A problem conversion module, for converting the distribution network topology reconstruction and distributed resource configuration problems into reinforcement learning problems based on the extracted distribution network topology and resource distribution features; The collaborative optimization module is used to design the reward function and use the deep reinforcement learning algorithm for training to achieve collaborative optimization of distribution network topology reconstruction and distributed resource configuration.

9. An electronic device, comprising: Memory and processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the distribution network topology reconstruction and distributed resource optimization configuration method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the method for topology reconstruction of a distribution network and optimal configuration of distributed resources as claimed in any one of claims 1 to 7.

Images