Distributed power distribution network key distribution method based on quantum communication

Abstract

The invention discloses a quantum communication-based distributed novel power distribution network key distribution method, which comprises the following steps of: predicting power flow fluctuation caused by distributed energy output change by establishing a prediction model so as to obtain power flow distribution data in real time; dynamically generating candidate nodes of the key distribution network based on the obtained power flow distribution data, and selecting a high-load fluctuation area as a quantum communication distribution center node so as to construct a distributed key distribution network topology; on the basis of the established network topology, quantum key distribution links are realized by generating quantum entanglement pairs. According to the method, the power flow fluctuation caused by the output change of the distributed energy can be acquired and predicted in real time, and then the key distribution path is dynamically adjusted to cope with the change of the running state of the power grid.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution network, and in particular to a distributed novel power distribution network key distribution method based on quantum communication. Background Art

[0002] In recent years, the scale of the distributed energy market in China has continued to grow. In 2023, the market size reached 107.718 billion yuan, becoming one of the important markets for distributed energy investment in the world. As of 2022, China's cumulative installed capacity of distributed energy is about 250.0049 million kilowatts, showing China's strong development momentum in the field of distributed energy. On March 21, 2024, the National Energy Administration issued the "Key Action Plan for Distribution Network Safety Risk Management", which mentioned that it is necessary to analyze the impact of large-scale access of new grid-connected entities such as distributed distributed energy power sources and energy storage electric vehicle charging piles on the distribution network, as well as to investigate and analyze the grid connection safety management of new grid-connected entities, so as to ensure the safety of the power grid.

[0003] One of the problems with the access of distributed energy sources to the distribution network is that the power flow distribution of the distribution network becomes more complicated. For example, Huang Haiquan, Huang Xiaowei, Jiang Wang, et al. A review of the research on distributed energy storage system solutions and configurations for new distribution networks [J]. Southern Energy Construction, 2024, 11(4) mentioned that in the new distribution network, the high proportion of distributed photovoltaic access has brought a heavy burden to the new distribution network: (1) The peak period of photovoltaic power generation is inconsistent with the peak period of electricity consumption, and the new distribution network has changed from a one-way power flow to a two-way power flow, causing power flow uncertainty and increasing network losses. Chinese patent CN105048451A provides an interval power flow calculation method based on the interval prediction of distributed energy generation, specifically using the ability of interval type II fuzzy logic to handle uncertainty problems, and obtaining an interval type II fuzzy prediction set of distributed energy generation, wherein the interval type II fuzzy prediction set clarifies the fluctuation range of power generation, which can be the initial iteration interval in the interval power flow calculation, avoiding the problems of non-convergence, too fast convergence, too slow convergence, etc. caused by artificially defining the initial interval, thereby improving the accuracy of the calculation. Peng Wei, Zheng Lianqing, Zheng Tianwen. Optimal configuration method of distributed photovoltaic energy storage system [J]. Sichuan Electric Power Technology, 2022, 45(1) A two-layer optimization model is used to minimize the fluctuations caused by the access of distributed photovoltaic to the distribution network: the outer optimization target is the access location, power and capacity of the energy storage system; the inner optimization target is the cost of the energy storage system. After the outer layer gives the determined energy storage parameters, the inner layer uses the particle swarm algorithm combined with the MATPOWER power flow calculation tool to optimize the operation strategy of the energy storage system so that the total cost of the energy storage system is minimized under this configuration; then, the inner optimization results are fed back to the outer layer for genetic algorithm selection, crossover and mutation operations, and the optimal configuration and operation strategy of energy storage in the feasible domain are determined through repeated iterative comparison. At the same time, in 2022, the main line of the 10kV Hengshan Line in Jiangbei, Ningbo, Zhejiang, achieved full coverage of quantum encryption smart switches. In 2023, the State Grid Wuhan Power Supply Company realized quantum encryption communication in the distribution automation terminal within the power supply ring network of Wuhan Economic and Technological Development Zone. Therefore, the issue that must be considered now is that quantum key distribution (QKD) relies on a stable communication channel to generate and distribute keys. However, fluctuations in distributed energy output will cause changes in the operating state of the power grid, affecting the generation and distribution of keys. Summary of the invention

[0004] In order to solve the stability problem of quantum key distribution in distributed energy distribution networks, the present invention provides a new distributed distribution network key distribution method based on quantum communication, which can obtain and predict the power flow fluctuations caused by changes in distributed energy output in real time, and then dynamically adjust the key distribution path to cope with changes in the operating status of the power grid.

[0005] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0006] A new distributed key distribution method for distribution network based on quantum communication includes:

[0007] By establishing a prediction model, the power flow fluctuations caused by changes in distributed energy output are predicted to obtain power flow distribution data in real time; based on the obtained power flow distribution data, candidate nodes of the key distribution network are dynamically generated, and high-load fluctuation areas are selected as quantum communication distribution center nodes, thereby constructing a distributed key distribution network topology; based on the established network topology, the quantum key distribution link is realized by generating quantum entangled pairs.

[0008] Further: The prediction model is established to predict the flow fluctuation caused by the change of distributed energy output, so as to obtain the flow distribution data in real time, including:

[0009] Integrate historical power flow data, real-time distributed energy output data, load data, node voltage, and meteorological conditions of each node in the power grid;

[0010] The spatiotemporal sequence data is converted into quantum feature vectors, and the quantum feature vectors are convolved through the quantum convolution layer to extract and enhance local features;

[0011] The data after quantum feature enhancement is input into the spatiotemporal convolutional neural network;

[0012] The spatiotemporal convolutional neural network includes a power grid spatiotemporal graph structure and a spatiotemporal convolutional layer;

[0013] The grid space-time graph structure includes treating each node in the grid as a node in the graph structure, representing the grid connection lines as edges in the graph, thereby constructing a grid space graph structure, and the data in the time dimension is used as a time series input of the graph;

[0014] The spatiotemporal convolution layer includes mapping the quantum feature vector after the convolution operation to the spatiotemporal relationship of each node through the spatiotemporal convolution operation, capturing the electrical coupling relationship between different nodes, and in the spatiotemporal convolution process, using the quantum feature vector as the data input of each node to gradually obtain the comprehensive influence of its neighboring nodes on the power flow;

[0015] Generate graph embeddings for each node. The graph embeddings of each node represent the distribution of power flow characteristics of the node at the current time and in the future, and reflect the influence of its neighboring nodes, so as to obtain the global distribution of power flow;

[0016] The nodes of the spatiotemporal convolutional layer are embedded into the input dynamic prediction layer, and the recurrent neural network is used for power flow prediction.

[0017] Further: the candidate nodes of the key distribution network are dynamically generated based on the acquired power flow distribution data, and the high load fluctuation area is selected as the quantum communication distribution center node, so as to construct the distributed key distribution network topology, including:

[0018] Obtain power flow prediction results and calibrate high load fluctuation nodes;

[0019] Generate a set of key distribution candidate nodes;

[0020] Select the central node;

[0021] Build a distributed key distribution network topology.

[0022] Further: the obtaining of power flow prediction results and calibration of high load fluctuation nodes includes:

[0023] Based on the power flow prediction data, the nodes with large load fluctuations, i.e., high load fluctuation nodes, are calibrated using the node load variance and power change rate.

[0024] Further: the generating of the key distribution candidate node set includes:

[0025] The high-load fluctuation nodes and their neighboring nodes are combined into a candidate node set. In the candidate node set, the network connectivity verification algorithm is used to confirm the connectivity of the set to ensure that all candidate nodes form a distribution network topology with multi-path redundancy.

[0026] Further: the selecting a central node includes:

[0027] By utilizing the load change rate of candidate nodes and the power flow in the node neighborhood and combining with graph theory algorithms, nodes that are in the load fluctuation area and have a high degree of connectivity with other candidate nodes are preferentially selected as quantum key distribution center nodes.

[0028] Further: the construction of a distributed key distribution network topology includes:

[0029] Based on candidate nodes and central nodes, the topology of the distributed key distribution network is constructed using a graph structure generation algorithm.

[0030] Further: Based on the established network topology, realizing the quantum key distribution link by generating quantum entangled pairs includes:

[0031] Generate quantum entangled pairs between the selected distribution center node and its adjacent candidate nodes;

[0032] Based on the quantum key distribution protocol of entangled pair sharing, a key link is established between the distribution center node and the adjacent nodes.

[0033] Compared with the prior art, the present invention has the following technical advances:

[0034] The present invention acquires and predicts the power flow fluctuations caused by the changes in the output of new energy sources in real time by establishing a prediction model, and then dynamically adjusts the key distribution path to ensure the stability of key distribution. First, various real-time data involved in the operation of the power grid are collected, including historical power flow data, real-time distributed energy output data, load data, node voltage, meteorological conditions (such as wind speed, sunshine intensity, etc.) information, and power grid operation status, etc., and a comprehensive dynamic data set is formed through high-frequency data collection and transmission. Next, the advantages of graph neural networks are used to identify the correlation between different nodes and predict future power flow distribution.

[0035] After completing the power flow prediction, the present invention can identify potential high-load fluctuation areas, and by analyzing the characteristics of these areas, calibrate the nodes suitable as quantum key distribution centers. Using these calibrated nodes, the present invention can construct a preliminary set of candidate nodes, and on this basis, confirm the connectivity of the candidate nodes through graph theory algorithms, and finally form a distribution network topology with multi-path redundancy. Importantly, this topology is not static, but dynamically adjusted as real-time power flow data changes.

[0036] After monitoring the changes in renewable energy output and its impact on the grid load, the present invention applies the characteristics of quantum entanglement to adjust the key distribution path through the optimized network topology. This flexibility and adaptability ensures that quantum key distribution can remain efficient and stable even when renewable energy output fluctuates. Therefore, through the present invention, the key distribution network can respond to load fluctuations in real time in a constantly changing power environment, ensuring the continuity and security of key generation and distribution. DETAILED DESCRIPTION

[0037] The following specific embodiments may be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments. The embodiments of the present invention will be described below.

[0038] The present invention discloses a novel distributed key distribution method for power distribution network based on quantum communication, comprising the following steps:

[0039] Step 1: Predict the flow fluctuations caused by changes in distributed energy output by establishing a prediction model to obtain flow distribution data in real time.

[0040] Step 2: Dynamically generate candidate nodes for the key distribution network based on the acquired power flow distribution data, and select high-load fluctuation areas as quantum communication distribution center nodes to construct a distributed key distribution network topology.

[0041] Step 3: Based on the established network topology, the quantum key distribution link is realized by generating quantum entangled pairs.

[0042] Specifically, in step 1, in order to establish a prediction model, this embodiment combines quantum feature enhancement, spatiotemporal convolution layer and dynamic prediction layer to achieve accurate prediction of power flow distribution under changes in distributed energy output.

[0043] This model is built based on quantum feature enhancement technology, which is used to extract high-dimensional features of real-time data of distributed energy output, and uses spatiotemporal convolutional networks to model the spatiotemporal dynamics of power grid flows to generate short-term predictions of flow distribution. The ultimate goal of the model is to generate flow distribution prediction data at future moments, which is convenient for the subsequent optimization and adjustment of quantum key distribution paths. The specific implementation steps are as follows:

[0044] 1. The historical flow data of each node in the integrated power grid, real-time distributed energy output data (such as wind power and photovoltaic output), load data, node voltage and other information are uniformly processed into spatiotemporal sequence data. Then, the original data is converted into quantum feature vectors using the quantum state mapping method to more effectively capture complex nonlinear relationships. The quantum features are then convolved through the quantum convolution layer to achieve local feature extraction and enhancement, especially for node areas where distributed energy fluctuates frequently. This is to better adapt to the real-time volatility of distributed energy output. Finally, the data after quantum feature enhancement is input into the next step of the spatiotemporal convolutional neural network as the basic feature for dynamic modeling of power flow distribution.

[0045] 2. Construct the spatiotemporal graph structure of the power grid, regard each node in the power grid as a node in the graph structure, and represent the power grid connection lines as edges in the graph, so as to construct the spatial graph structure of the power grid, and the data in the time dimension is used as the time series input of the graph. Then the spatiotemporal convolution layer is performed. Through the spatiotemporal convolution operation, the quantum feature data is mapped to the spatiotemporal relationship of each node to capture the electrical coupling relationship between different nodes. In the spatiotemporal convolution process, the quantum feature is used as the data input of each node, and the comprehensive influence of its neighboring nodes on the power flow is gradually obtained. Quantum parallel computing is introduced in the spatiotemporal convolution layer to enhance the speed and computational efficiency of data processing, and generate graph embeddings for each node. Each node embedding represents the power flow feature distribution of the node at the current moment and the future moment, and reflects the influence of its neighboring nodes, so as to obtain the global distribution of the power grid power flow. The output of the spatiotemporal convolution layer is the power flow feature embedding of each node, which provides feature data with spatiotemporal relationship for the next step of dynamic power flow prediction.

[0046] 3. The nodes of the spatiotemporal convolutional layer are embedded into the input dynamic prediction layer, and a recurrent neural network (such as GRU) is used to perform short-term power flow prediction. This layer predicts the changes in power flow distribution at several moments in the future based on the current power flow characteristics. In order to improve the real-time accuracy of the prediction, the quantum attention mechanism gives higher weights to nodes with greater influence in the prediction process to adapt to local fluctuations in the power flow in real time. The attention mechanism adjusts the weights through quantum computing to enhance attention to important nodes, thereby improving the prediction accuracy of the model when the output of distributed energy sources fluctuates violently. The result of the dynamic prediction layer combined with the quantum attention mechanism is the power flow prediction at future moments, which can reflect the real-time power flow distribution under the fluctuation of distributed energy output.

[0047] 4. To ensure the accuracy and training efficiency of the model under high-dimensional data, quantum gradient descent can be used to optimize the model parameters to improve its convergence speed and prediction performance under dynamic data. The optimized model is applied to the power distribution prediction, and the power distribution at future moments is output in the form of a graph, indicating the predicted power, node voltage, power flow direction and size of each node. These prediction results are used for dynamic adjustment of the key distribution path.

[0048] Specifically, in step 2, the goal is to dynamically generate candidate nodes of the key distribution network based on the predicted power flow distribution with the support of the power flow prediction results in step 1, and set the high-load fluctuation area as the main quantum communication distribution center node through the graph theory algorithm to build a distributed key distribution network topology. The following are the specific implementation steps:

[0049] 1. Obtain power flow prediction results and calibrate high load fluctuation nodes

[0050] Based on the power flow prediction data, the node load fluctuation analysis method is used to calibrate the nodes with large load fluctuations; these nodes may have high load changes due to the fluctuation of distributed energy output, and need to be prioritized as candidates for quantum key distribution center nodes. In order to identify and calibrate high load fluctuation nodes in the quantum key distribution network, so as to use them as potential candidates for quantum key distribution center nodes, including:

[0051] Obtain the power flow forecast data for the future time period from step 1. The data content includes the load, power flow, node voltage and historical data of each node. Clean the collected forecast data, remove outliers and noise, and use the load volatility analysis method to define the load fluctuation index of each node, including load variance and power change rate. The calculation formula is as follows:

[0052] Load variance calculation formula:

[0053]

[0054] in, represents the load variance of node i, L i,t represents the load value of node i at time t, μ i is the average load of node i, and N is the total number of time points.

[0055] Power change rate calculation formula:

[0056]

[0057] Among them, R i represents the power change rate of node i at time t.

[0058] Based on the method defined above, the load variance and power change rate of all nodes are calculated in turn to evaluate the degree of load fluctuation of the nodes during the predicted time period. A load fluctuation threshold is set according to historical data analysis or industry standards to distinguish high-load fluctuation nodes from normal nodes. The calculated load fluctuation index is compared with the set threshold, and the nodes whose load fluctuation exceeds the set threshold are calibrated to obtain a group of high-load fluctuation nodes. These nodes may have large load changes due to fluctuations in distributed energy output and need to be given priority as candidates for the distribution center nodes of quantum keys.

[0059] 2. Generate a set of candidate nodes for key distribution

[0060] The high-load fluctuation nodes and their neighboring nodes are combined into a preliminary candidate node set to improve the coverage of the fluctuation area and ensure that the key distribution network can more efficiently adapt to possible load changes in the future. In the candidate node set, the network connectivity verification algorithm is used to confirm the connectivity of the set to ensure that all candidate nodes form a distribution network topology with multi-path redundancy, specifically:

[0061] First, a neighborhood radius R is set, which is used to define the range of neighboring nodes of each high-load fluctuation node. For each calibrated high-load fluctuation node i, according to the set neighborhood radius R, the neighboring node set N is identified by the following formula: i :

[0062] N i ={j|d(i,j)≤R}

[0063] Where d(i,j) represents the distance between node i and node j. If the distance is less than or equal to the neighborhood radius R, node j is considered to be a neighboring node of node i.

[0064] Finally, we get the neighboring node set N corresponding to each high load fluctuation node. i .

[0065] All high-load fluctuation nodes and their corresponding neighboring nodes are merged to form a preliminary candidate node set C. The preliminary candidate node set C is regarded as a graph model, in which nodes represent candidate nodes and edges represent the connection relationship between nodes. The graph is represented by the adjacency matrix A, in which A[i][j]=1 indicates that there is a connection between node i and node j, otherwise it is 0.

[0066] Use the breadth-first search (BFS) algorithm to start from any candidate node and verify the connectivity of the candidate node set:

[0067] Starting from any node s in the set, mark it as visited and add it to the queue;

[0068] Iterate each node in the queue, check all its adjacent nodes, add unvisited adjacent nodes to the queue, and mark them as visited;

[0069] When the queue is empty, if the number of visited nodes is equal to the size of the candidate node set C, then the set is connected, otherwise it is not connected.

[0070] If the connectivity verification passes, the preliminary candidate node set C can be confirmed as the final candidate node set C f , which serves as the basis for the subsequent construction of the key distribution network topology, generates a set of key distribution candidate nodes with connectivity and coverage.

[0071] 3. Central node selection based on graph theory

[0072] Using the load fluctuation intensity indicators in the candidate node set, including the load change rate and the power flow in the node neighborhood, and combining with the graph theory algorithm, the nodes in the load fluctuation area with high connectivity with other candidate nodes are preferentially selected as the quantum key distribution center nodes, specifically:

[0073] From the candidate node set, extract the load fluctuation intensity index, which includes the load change rate R of the node. i and the power flow F in the node neighborhood i , where the power flow F in the node neighborhood is i It is expressed as:

[0074]

[0075] Among them, P i,j represents the flow from node i to neighboring node j, N i is the set of neighboring nodes of node i.

[0076] For each node i in the candidate node set, calculate its load fluctuation intensity index H i , which can be expressed as follows:

[0077] H i =αR i +βF i

[0078] Among them, α and β are weight coefficients, indicating the importance of load change rate and flow rate.

[0079] The candidate node set is regarded as a graph, where nodes represent candidate nodes and edges represent the connection relationship between nodes. The graph is described by the adjacency matrix A, where A[i][j]=1 indicates that there is a connection between node i and node j.

[0080] Use PageRank algorithm to calculate the centrality C of candidate nodes i , expressed by the following formula:

[0081]

[0082] Where d is the damping factor, N(i) is the set of neighboring nodes of node i, and |N(j)| is the number of connections of neighboring node j.

[0083] Calculate the degree D of the candidate node i :

[0084] D i =|N(i)|

[0085] Nodes with higher connectivity are selected as potential hub nodes.

[0086] Combined with the load fluctuation intensity index H i and the centrality index C i or D i , set the threshold, and select the node with the highest comprehensive score as the central node N of quantum key distribution c .

[0087] 4. Build a distributed key distribution network topology

[0088] Based on candidate nodes and central nodes, the initial topology of the distributed key distribution network is constructed using a graph structure generation algorithm to ensure high connectivity and redundancy between nodes in the network.

[0089] First, it is necessary to clearly define the goals of building a distributed key distribution network, including high connectivity, redundancy, and fault tolerance, so that the network can still maintain its functionality in the event of node failure or load changes.

[0090] Then set the required topological parameters, such as the maximum connection distance between nodes, the connection weight threshold (set according to node load and connection stability), and the required number of redundant paths. Finally, determine the network requirements and topological parameters, laying the foundation for subsequent topological map generation.

[0091] Based on the candidate node set C f and the selected central node N c , construct the adjacency matrix A, where A[i][j]=1 indicates that there is a valid connection between node i and node j. The validity of the connection can be judged based on the node load and the connection weight. If the connection weight W ij Above the set threshold W threshold , the connection is valid.

[0092] Define the connection weight W between nodes ij The calculation formula can be used for comprehensive evaluation based on load fluctuation intensity and geographical distance factors:

[0093]

[0094] Where d(i,j) is the physical distance between node i and node j, H i and H j are the load fluctuation intensity indicators of node i and node j respectively.

[0095] Finally, the constructed adjacency matrix A is obtained, and the result generated by the minimum spanning tree algorithm forms the initial topological structure G init , G init There is at least one path between all nodes to ensure the connectivity of the network.

[0096] In order to optimize the topology structure to enhance redundancy, on the basis of the initial topology structure, for each pair of nodes (i, j), additional paths are found to enhance network redundancy, ensuring that when a connection fails, communication can still be carried out through other paths. In this embodiment, the path finding algorithm in graph theory is used (finding multiple paths between nodes, and selecting according to the weight of the path, preferentially selecting the path with lower weight as the redundant path, adding the redundant path to the network topology, and forming the optimized topology structure G opt , that is, the distributed key distribution network topology.

[0097] Specifically, in step 2, the final optimized topology G has been generated. opt This structure includes high-load fluctuation nodes, adjacent candidate nodes, and multi-path redundancy, which is the basic topology of the quantum key distribution network. Therefore, in step 3, this embodiment uses the quantum entanglement characteristics and G optThe multi-path redundancy provided establishes a key distribution link with high security and anti-interference capability for the distributed key distribution network.

[0098] First, from G opt Select the quantum entanglement link in G opt In the process, according to the connection strength and load fluctuation strength of each node, node pairs suitable for quantum key distribution paths are selected to ensure that these links can cover the high load fluctuation area and its neighboring nodes to improve the robustness of the key distribution network. opt The weight information (i.e., load fluctuation intensity, transmission distance) contained in the adjacency matrix is ​​used to evaluate the transmission reliability R of the quantum entangled pair on each link. ij , the formula is:

[0099]

[0100] Where d(i,j) is the distance between node i and node j, H i and H j is the load fluctuation intensity index, and α and β are adjustment parameters.

[0101] Then, quantum entanglement-based key distribution is constructed on the selected link. opt Entangled state pairs are generated in turn on the links selected in the network, and the entangled state pairs are distributed to the communication nodes through these links. To ensure the stability and anti-interference of key distribution, the generation and distribution of quantum bits are completed using the QKD protocol.

[0102] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein with equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the scope of protection of the claims of the present invention.

Claims

1. A new distributed distribution network key distribution method based on quantum communication, including: By establishing a prediction model, the power flow fluctuation caused by the change of distributed energy output is predicted to obtain power flow distribution data in real time; Based on the acquired power flow distribution data, candidate nodes of the key distribution network are dynamically generated, and high-load fluctuation areas are selected as quantum communication distribution center nodes to build a distributed key distribution network topology; based on the established network topology, the quantum key distribution link is realized by generating quantum entangled pairs.

2. The distributed new distribution network key distribution method based on quantum communication according to claim 1 is characterized in that: The prediction model is established to predict the flow fluctuation caused by the change of distributed energy output, so as to obtain the flow distribution data in real time, including: The historical power flow data, real-time distributed energy output data, load data, node voltage, and meteorological condition information of each node in the integrated power grid are processed into spatiotemporal sequence data; The spatiotemporal sequence data is converted into quantum feature vectors, and the quantum feature vectors are convolved through the quantum convolution layer to extract and enhance local features; The data after quantum feature enhancement is input into the spatiotemporal convolutional neural network; The spatiotemporal convolutional neural network includes a power grid spatiotemporal graph structure and a spatiotemporal convolutional layer; The grid space-time graph structure includes treating each node in the grid as a node in the graph structure, representing the grid connection lines as edges in the graph, thereby constructing a grid space graph structure, and the data in the time dimension is used as a time series input of the graph; The spatiotemporal convolution layer includes mapping the quantum feature vector after the convolution operation to the spatiotemporal relationship of each node through the spatiotemporal convolution operation, capturing the electrical coupling relationship between different nodes, and in the spatiotemporal convolution process, using the quantum feature vector as the data input of each node to gradually obtain the comprehensive influence of its neighboring nodes on the power flow; Generate graph embeddings for each node. The graph embeddings of each node represent the distribution of power flow characteristics of the node at the current time and in the future, and reflect the influence of its neighboring nodes, so as to obtain the global distribution of power flow; The nodes of the spatiotemporal convolutional layer are embedded into the input dynamic prediction layer, and the recurrent neural network is used for power flow prediction.

3. The distributed new distribution network key distribution method based on quantum communication according to claim 1 is characterized in that: The method of dynamically generating candidate nodes of the key distribution network based on the acquired power flow distribution data and selecting the high load fluctuation area as the quantum communication distribution center node to construct the distributed key distribution network topology includes: Obtain power flow prediction results and calibrate high load fluctuation nodes; Generate a set of key distribution candidate nodes; Select the central node; Build a distributed key distribution network topology.

4. The distributed novel distribution network key distribution method based on quantum communication according to claim 1, 2 or 3, characterized in that: The obtaining of power flow prediction results and calibration of high load fluctuation nodes comprises: Based on the power flow prediction data, the nodes with large load fluctuations, i.e., high load fluctuation nodes, are calibrated using the node load variance and power change rate.

5. The distributed novel distribution network key distribution method based on quantum communication according to claim 3 is characterized in that: The generating of the key distribution candidate node set comprises: The high-load fluctuation nodes and their neighboring nodes are combined into a candidate node set. In the candidate node set, the network connectivity verification algorithm is used to confirm the connectivity of the set to ensure that all candidate nodes form a distribution network topology with multi-path redundancy.

6. The distributed novel distribution network key distribution method based on quantum communication according to claim 5 is characterized in that: The selecting central node comprises: By utilizing the load change rate of candidate nodes and the power flow in the node neighborhood and combining with graph theory algorithms, nodes that are in the load fluctuation area and have a high degree of connectivity with other candidate nodes are preferentially selected as quantum key distribution center nodes.

7. The distributed novel distribution network key distribution method based on quantum communication according to claim 6 is characterized in that: The construction of a distributed key distribution network topology includes: Based on candidate nodes and central nodes, the topology of the distributed key distribution network is constructed using a graph structure generation algorithm.

8. The distributed novel distribution network key distribution method based on quantum communication according to claim 7 is characterized in that: The method of realizing a quantum key distribution link by generating a quantum entangled pair based on the established network topology includes: Generate quantum entangled pairs between the selected distribution center node and its adjacent candidate nodes; Based on the quantum key distribution protocol of entangled pair sharing, a key link is established between the distribution center node and the adjacent nodes.

Images