A method and system for optimizing routing in power communication networks

Abstract

This invention discloses a method and system for optimizing routing in power communication networks, belonging to the field of power communication network routing optimization technology. The specific steps of the method include: obtaining current network parameters and obtaining an edge weight matrix using a graph neural network model; wherein, the graph neural network model is obtained by learning the relationship between different transmission requirements and network scenarios and edge weights; solving for paths using a depth-first search algorithm based on the edge weight matrix; and selecting the optimal path for power communication network routing based on the solution results. This invention achieves highly reliable routing under different network scenarios and different service requirements, and can effectively select highly reliable paths to support service data transmission.

Description

Technical Field

[0001] This invention relates to the field of power communication network routing optimization technology, and in particular to a power communication network routing optimization method and system. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] As a dedicated communication network serving the power system, the power communication network aims to ensure the communication needs of various stations within the power system. Currently, the State Grid Corporation of China has built a large-scale power communication network across the country. The topology is mainly ring-shaped, with some extremely remote areas using chain-shaped networking, forming a highly complex optical fiber network, and using Synchronous Digital Hierarchy (SDH) technology to support data transmission.

[0004] In recent years, the stable and secure operation of power communication networks has faced significant challenges due to the increasing demand for electricity from residents. The requirements for transmission reliability, transmission latency, and load balancing between various stations in power communication networks are also becoming increasingly stringent. To ensure the timely and stable operation of power communication networks, new technologies are needed for route optimization to obtain better routing solutions.

[0005] However, most existing routing optimization schemes for power communication networks primarily address static topology problems, neglecting adaptability to network changes. For example, the addition or removal of sites or equipment, as well as the emergence and disappearance of links between sites, manifest as changes in the network topology of nodes and edges over time. Simultaneously, service demands within the power communication network also change over time, leading to route reconfiguration. Existing routing optimization algorithms for power communication networks typically struggle to quickly obtain reliable routing results under conditions of topology and demand changes. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the present invention aims to provide a routing optimization method and system for power communication networks. First, it predicts the weight parameters of edges based on a graph neural network. Then, it selects a path using a depth-first search approach based on the graph neural network prediction results. This invention achieves highly reliable routing under different network scenarios and service requirements, effectively selecting highly reliable paths to support service data transmission.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solution:

[0008] The first aspect of this invention provides a method for optimizing routing in a power communication network, comprising the following steps:

[0009] Obtain the current network parameters and use a graph neural network model to obtain the edge weight matrix; the graph neural network model is obtained by learning the relationship between different transmission requirements and network scenarios and edge weights.

[0010] The path is solved using a depth-first search algorithm based on the edge weight matrix;

[0011] The optimal path is selected for power communication network routing based on the solution results.

[0012] Furthermore, the specific steps to obtain the graph neural network model by learning the relationship between different transmission requirements and network scenarios and edge weights are as follows:

[0013] Historical network parameters are obtained as a training set, and the historical network parameters are organized to obtain the network topology, transmission requirements, and historical edge weight matrix.

[0014] A preliminary graph neural network model is established. Based on the network topology, transmission requirements, and historical edge weight matrix, the relationship between the values ​​of each edge weight and the network structure and transmission requirements is learned, thus obtaining the graph neural network model.

[0015] Furthermore, when learning the relationship between the values ​​of each edge weight and the network structure and transmission requirements based on the network topology, transmission requirements, and historical edge weight matrix, the historical edge weight matrix is ​​used for labeling, the network topology is used to extract network structure features through graph message passing network, and the information of transmission requirements is used for neural network learning and processing.

[0016] Furthermore, the path must satisfy the constraints determined by the depth-first search algorithm based on the edge weight matrix. These constraints include time delay constraints, link capacity constraints, flow conservation constraints, and path indivisibility constraints.

[0017] Furthermore, based on the edge weight matrix, the path is calculated using the edge weight routing algorithm, prioritizing high-weight service requirements; for low-weight service requirements, after the high-weight service requirements have selected a path and updated the network parameters, the path selection continues using the edge weight routing algorithm.

[0018] Furthermore, the specific steps for solving the path using the depth-first search algorithm based on the edge weight matrix are as follows:

[0019] Generate a set of feasible paths based on the edge weight matrix and business requirements;

[0020] The priority of path selection is set based on the weight value of the path;

[0021] Determine the type of current business requirement;

[0022] Select a path based on the current business requirement type, and update the graph neural network model parameters based on the selected path;

[0023] Calculate the reliability of the selected path and output the path.

[0024] Furthermore, the specific rules for selecting a path based on the current business requirement type are as follows:

[0025] For single-route service requirements, the path with the highest weight is selected;

[0026] For dual-route service requirements, the path with the highest weight is selected as the primary path.

[0027] For dual-route service requirements, the path with the highest combined reliability with the primary path is selected as the backup path.

[0028] A second aspect of the present invention provides a power communication network routing optimization system, comprising:

[0029] The model prediction module is configured to obtain the current network parameters and use a graph neural network model to obtain the edge weight matrix; wherein, the graph neural network model is obtained by learning the relationship between different transmission requirements and network scenarios and edge weights.

[0030] The path finding module is configured to solve paths using a depth-first search algorithm based on the edge weight matrix;

[0031] The communication network routing module is configured to select the optimal path for power communication network routing based on the solution results.

[0032] Furthermore, in the path solving module, the specific steps for solving the path using the depth-first search algorithm based on the edge weight matrix are as follows:

[0033] Generate a set of feasible paths based on the edge weight matrix and business requirements;

[0034] The priority of path selection is set based on the weight value of the path;

[0035] Determine the type of current business requirement;

[0036] Select a path based on the current business requirement type, and update the graph neural network model parameters based on the selected path;

[0037] Calculate the reliability of the selected path and output the path.

[0038] Furthermore, in the path finding module, the specific rules for selecting a path based on the current business requirement type are as follows:

[0039] For single-route service requirements, the path with the highest weight is selected;

[0040] For dual-route service requirements, the path with the highest weight is selected as the primary path.

[0041] For dual-route service requirements, the path with the highest combined reliability with the primary path is selected as the backup path.

[0042] The above one or more technical solutions have the following beneficial effects:

[0043] This invention discloses a routing optimization method and system for power communication networks. First, a graph neural network is established to learn the relationship between the values ​​of each edge weight and the network structure and transmission requirements. During the model usage phase, the edge weight matrix in the current network can be directly obtained based on the network structure and transmission requirements. Then, a depth-first search algorithm is designed to generate highly reliable paths based on the learned edge weight matrix. Compared with traditional optimization methods, this invention simplifies the calculation steps and avoids directly solving non-convex mixed-integer programming problems, resulting in higher overall reliability of path selection. Because the algorithm uses a graph neural network to predict the edge weight matrix, it achieves superior generalization under different topologies.

[0044] This invention analyzes the structure and transmission requirements of power communication networks. With delay and capacity as constraints and high-reliability transmission as the objective, it establishes a joint optimization problem for single and dual routing. A path selection algorithm integrating graph neural networks and depth-first search is proposed, achieving high-reliability routing under different network scenarios and service requirements. The proposed algorithm can effectively select high-reliability paths to support service data transmission.

[0045] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0046] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0047] Figure 1 This is a flowchart of the power communication network routing optimization method in Embodiment 1 of the present invention;

[0048] Figure 2 The diagram below shows the message passing network structure in Embodiment 1 of the present invention.

[0049] Figure 3 This is a diagram of the neural network model structure in Embodiment 1 of the present invention. Detailed Implementation

[0050] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0051] It should be noted that in the embodiments of this application, data related to network parameters and so on are involved. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0052] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0053] Example 1:

[0054] Dynamic network models are a feasible theoretical tool for addressing changes in the topology and service requirements of power communication networks. Dynamic networks often feature a large number of nodes and complex topologies, but by applying effective external input variables, it is possible to predict the states of nodes or links within the entire network, thereby simplifying problem-solving. Graph neural networks (GNNs) offer great flexibility and feasibility in handling dynamic network problems. They can perform relational reasoning on graph structures, extracting information from nodes and edges to learn feature information and analyze the relationships between nodes. Furthermore, GNNs can be combined with other algorithms to address issues where algorithms cannot be generalized to graph structures.

[0055] Embodiment 1 of the present invention provides a routing optimization method for power communication networks, such as... Figure 1 As shown, it includes the following steps:

[0056] Step 1: Obtain the current network parameters and use the graph neural network model to obtain the edge weight matrix.

[0057] Step 2: Solve the path using the depth-first search algorithm based on the edge weight matrix.

[0058] Step 3: Select the optimal path for power communication network routing based on the solution results.

[0059] In step 1, a graph neural network model is obtained by learning the relationship between different transmission requirements and network scenarios and edge weights. The relaxed solution is obtained for the known transmission requirements and network parameters, and this solution is used as the label for the graph neural network. By using the graph neural network to construct the relationship between different transmission requirements and network scenarios and edge weights, the trained model can quickly predict edge weight values.

[0060] The specific steps are as follows:

[0061] (1) Obtain historical network parameters as a training set, and organize the historical network parameters to obtain network topology, transmission requirements and historical edge weight matrix;

[0062] (2) Establish a preliminary graph neural network model, such as Figure 3 As shown, the structure includes a graph message-passing network (MPNN) layer connected to two fully connected neural network layers. MPNN is a typical application of graph neural networks, implementing the connection between graph structures and neural networks. MPNN can extract topological features, collecting information from each node and edge in the entire network. MPNN can collect graph features for graph embedding under different network topologies, allowing the trained model to generalize to handle different topologies. Its structure diagram is shown below. Figure 2 As shown.

[0063] Graph messaging networks mainly consist of three steps: First, generating characteristic information of surrounding nodes or links; second, aggregating the generated characteristic information onto the node; and third, updating the node based on the aggregated characteristic information and its original characteristic information. The formula is as follows:

[0064]

[0065]

[0066] Where m is the message generation function, A is the message aggregation function, and U is the message update function. i is the number of iterations, N(v) is the number of neighboring nodes of node v, and e uv This represents the feature information of the edges connected to node v. Let represent the i-th iteration, and u represent the node.

[0067] (3) Based on the network topology, transmission requirements and historical edge weight matrix, learn the relationship between the value of each edge weight and the network structure and transmission requirements. Use the historical edge weight matrix for labeling. Use the network topology to extract network structure features through graph message passing network. Use the information of transmission requirements for neural network learning and processing to obtain graph neural network model.

[0068] Specifically, the neural network's input consists of network topology, transmission requirements, and edge weights, with the edge weight matrix only used as label input during the training phase. After processing the input data, the two fully connected layers of the neural network ultimately output the predicted edge weight matrix. The network topology, also known as graph information, is extracted using a graph message passing network for network structure feature extraction. The transmission requirements information, to facilitate neural network learning and processing, calculates which links have been traversed using the shortest path algorithm, assigning each traversed link an initial value of 1 / h. n , where h n This represents the number of hops. During the training phase, the optimal solution of the objective function after 0-1 variable relaxation is used as the label of the edge weight matrix. This part is not required during the model deployment phase.

[0069] In step 2, the obtained graph neural network model outputs a predicted edge weight matrix by inputting network parameters. The edge weight represents the importance of each link to the current transmission demand. A larger weight indicates a greater gain for the selected path in terms of transmission demand; that is, selecting a path with a larger weight improves the objective value of the optimization problem. Priorities are assigned based on the calculated weights of each path; paths with higher weights are considered better for the current demand and should be selected first. After selecting the optimal path, the network resources it occupies, such as capacity, are calculated, and the network parameters are updated. Based on the edge weight matrix, a depth-first search algorithm is used to solve for the constraints that the path must satisfy. These constraints include delay constraints, link capacity constraints, flow conservation constraints, and path indivisibility constraints, specifically:

[0070]

[0071] stC1: Delay Constraint

[0072] C2: Link capacity constraint

[0073] C3: Flow conservation constraint

[0074] C4: Indivisible path constraint

[0075] Among them, constraint C1 means that the transmission path for single and dual routing service requirements cannot exceed the latency limit of the service; C2 means that the capacity required for the service requirements carried by each link cannot exceed the capacity limit; C3 means that the service transmission requirements start from the source node, pass through several relays, and reach the destination node; C4 means that the transmission path is indivisible.

[0076] If the selected path does not meet the constraints, the next highest weighted path is selected, and so on, until a path that meets the constraints is found.

[0077] Step 3: For each business requirement, the neural network model generates an edge weight matrix for that business requirement. Due to the different weights α for different business requirements... n and α k Different and competing for link resources, high-weight services are more important in power communication networks, and their paths are calculated based on edge-weighted routing algorithms. Among them, α n The weight α represents the single-route transmission demand. k This represents the weight of dual-route transmission requirements. After the higher-weight service requirements select a path and update network parameters, lower-weight service requirements continue to select a path based on an edge-weighted routing algorithm. Finally, all paths are selected sequentially according to their weights, and the overall reliability is calculated. The overall reliability is the sum of the reliability of each individual path multiplied by its corresponding service requirement weight.

[0078] The specific steps for solving the path using the depth-first search algorithm based on the edge weight matrix are as follows:

[0079] 1) Generate a set of feasible paths based on the edge weight matrix and business requirements;

[0080] 2) Set the priority of path selection based on the weight value of the path;

[0081] 3) Determine the type of current business requirement;

[0082] 4) Select the path based on the current business requirement type;

[0083] 5) Determine if the selected path meets the constraints. If not, delete the selected path from the set of feasible paths and return to step 3) to select a new path.

[0084] 6) Update the graph neural network model parameters based on the selected path;

[0085] 7) Calculate the reliability of the selected path and output the path. The reliability value represents the probability of normal operation of different links. The reliability of each path is calculated from the links it traverses. Specifically, the reliability of a path is equal to the product of the reliability of each sub-path it traverses. If a certain requirement traverses three links A, B, and C, then the reliability of that path is P. A *P B *P C Among them, P A P B P C These are the reliability values ​​for links A, B, and C, respectively.

[0086] The specific rules for selecting a path based on the current business requirement type are as follows:

[0087] For single-route service requirements, the path with the highest weight is selected;

[0088] For dual-route service requirements, the path with the highest weight is selected as the primary path.

[0089] For dual-route service requirements, the path with the highest combined reliability with the primary path is selected as the backup path.

[0090] This invention analyzes the structure and transmission requirements of power communication networks. With delay and capacity as constraints and high-reliability transmission as the objective, it establishes a joint optimization problem for single and dual routing. A path selection algorithm integrating graph neural networks and depth-first search is proposed, achieving high-reliability routing under different network scenarios and service requirements. The proposed algorithm can effectively select high-reliability paths to support service data transmission.

[0091] The specific embodiments described in this invention are merely illustrative of the spirit of the invention. Those skilled in the art can make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

[0092] Example 2:

[0093] Embodiment 2 of the present invention provides a power communication network routing optimization system, comprising:

[0094] The model prediction module is configured to obtain the current network parameters and use a graph neural network model to obtain the edge weight matrix; wherein, the graph neural network model is obtained by learning the relationship between different transmission requirements and network scenarios and edge weights.

[0095] The path finding module is configured to solve paths using a depth-first search algorithm based on the edge weight matrix;

[0096] The communication network routing module is configured to select the optimal path for power communication network routing based on the solution results.

[0097] In the path finding module, the specific steps for finding the path using the depth-first search algorithm based on the edge weight matrix are as follows:

[0098] Generate a set of feasible paths based on the edge weight matrix and business requirements;

[0099] The priority of path selection is set based on the weight value of the path;

[0100] Determine the type of current business requirement;

[0101] Select a path based on the current business requirement type, and update the graph neural network model parameters based on the selected path;

[0102] Calculate the reliability of the selected path and output the path.

[0103] The specific rules for selecting a path based on the current business requirement type are as follows:

[0104] For single-route service requirements, the path with the highest weight is selected;

[0105] For dual-route service requirements, the path with the highest weight is selected as the primary path.

[0106] For dual-route service requirements, the path with the highest combined reliability with the primary path is selected as the backup path.

[0107] The steps and methods involved in the above embodiment two correspond to those in embodiment one. For specific implementation details, please refer to the relevant description section of embodiment one.

[0108] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0109] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A routing optimization method for power communication networks, characterized in that, Includes the following steps: Obtain the current network parameters and use a graph neural network model to obtain the edge weight matrix. Specifically, the graph neural network model is obtained by learning the relationship between different transmission requirements and network scenarios and the edge weights. The specific steps are as follows: Historical network parameters are obtained as a training set, and the historical network parameters are organized to obtain the network topology, transmission requirements, and historical edge weight matrix. A preliminary graph neural network model is established. Based on the network topology, transmission requirements, and historical edge weight matrix, the relationship between the values ​​of each edge weight and the network structure and transmission requirements is learned to obtain the graph neural network model. The path is solved using a depth-first search algorithm based on the edge weight matrix. The specific steps are as follows: generate a set of feasible paths based on the edge weight matrix and business requirements; set the priority of path selection based on the weight values ​​of the paths; determine the current business requirement type; select a path based on the current business requirement type and update the graph neural network model parameters based on the selected path; calculate the reliability of the selected path and output the path. The specific rules for selecting a path based on the current business requirement type are as follows: For single-route service requirements, the path with the highest weight is selected; For dual-route service requirements, the path with the highest weight is selected as the primary path. For dual-route service requirements, the path with the highest combined reliability with the primary path is selected as the backup path. The optimal path is selected for power communication network routing based on the solution results.

2. The power communication network routing optimization method as described in claim 1, characterized in that, When learning the relationship between the values ​​of each edge weight and the network structure and transmission requirements based on the network topology, transmission requirements and historical edge weight matrix, the historical edge weight matrix is ​​used for labeling, and the network topology is used to extract network structure features through a graph message passing network. The information from the transmission requirements is used for neural network learning and processing.

3. The power communication network routing optimization method as described in claim 1, characterized in that, The path must satisfy the constraints determined by the depth-first search algorithm based on the edge weight matrix. These constraints include time delay constraints, link capacity constraints, flow conservation constraints, and path indivisibility constraints.

4. The power communication network routing optimization method as described in claim 1, characterized in that, Based on the edge weight matrix, the path is calculated using the edge weight routing algorithm, prioritizing the business requirements with higher weights. For business requirements with lower weights, after the path is selected by the high-weight business requirements and the network parameters are updated, the path selection continues using the edge weight routing algorithm.

5. A power communication network routing optimization system, characterized in that, include: The model prediction module is configured to acquire the current network parameters and obtain the edge weight matrix using a graph neural network model. Specifically, the graph neural network model is obtained by learning the relationship between different transmission requirements and network scenarios and the edge weights. The specific steps are as follows: Historical network parameters are obtained as a training set, and the historical network parameters are organized to obtain the network topology, transmission requirements, and historical edge weight matrix. A preliminary graph neural network model is established. Based on the network topology, transmission requirements, and historical edge weight matrix, the relationship between the values ​​of each edge weight and the network structure and transmission requirements is learned to obtain the graph neural network model. The path finding module is configured to solve paths using a depth-first search algorithm based on the edge weight matrix. The specific steps are as follows: generate a set of feasible paths based on the edge weight matrix and business requirements; set the priority of path selection based on the weight values ​​of the paths; determine the current business requirement type; select a path based on the current business requirement type and update the graph neural network model parameters based on the selected path; calculate the reliability of the selected path and output the path. The specific rules for selecting a path based on the current business requirement type are as follows: For single-route service requirements, the path with the highest weight is selected; For dual-route service requirements, the path with the highest weight is selected as the primary path. For dual-route service requirements, the path with the highest combined reliability with the primary path is selected as the backup path. The communication network routing module is configured to select the optimal path for power communication network routing based on the solution results.

6. The power communication network routing optimization system as described in claim 5, characterized in that, In the path finding module, the specific steps for finding the path using the depth-first search algorithm based on the edge weight matrix are as follows: Generate a set of feasible paths based on the edge weight matrix and business requirements; The priority of path selection is set based on the weight value of the path; Determine the type of current business requirement; Select a path based on the current business requirement type, and update the graph neural network model parameters based on the selected path; Calculate the reliability of the selected path and output the path.

7. The power communication network routing optimization system as described in claim 6, characterized in that, In the path finding module, the specific rules for selecting a path based on the current business requirement type are as follows: For single-route service requirements, the path with the highest weight is selected; For dual-route service requirements, the path with the highest weight is selected as the primary path. For dual-route service requirements, the path with the highest combined reliability with the primary path is selected as the backup path.

Images