Method, system and equipment for determining reliability path of power dual-mode remote communication network

Abstract

The invention discloses a method, a system and equipment for determining a reliability path of a power dual-mode remote communication network. The method comprises the following steps: determining a simulation topology model of a target power dual-mode remote communication network; constructing a global operation state matrix according to communication nodes and channels of the target power dual-mode remote communication network; performing coding processing on the simulation topology model and the global operation state matrix in sequence to obtain a first feature vector and a second feature vector; inputting the first feature vector and the second feature vector into a pre-constructed graph neural network processing model to obtain node features and edge features; calculating the node features and the edge features to obtain a path feature vector, and inputting the path feature vector into a path performance calculation formula to obtain a real-time path performance result of the predicted path; and determining an optimal path based on the real-time path performance result. According to the method provided by the embodiment of the invention, the optimal path can be accurately determined in a complex network dynamic change process.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power, and in particular, to a method, system and device for determining a reliability path of a power dual-mode remote communication network. Background Art

[0002] With the rapid development of modern power systems and the wide application of smart grids, power dual-mode remote communication technology has played a crucial role in ensuring the stable operation of power systems and improving energy scheduling efficiency.

[0003] Currently, most power dual-mode remote communication networks use single-mode or simple dual-mode communication modes for data transmission. The redundant path mechanism of single-mode or dual-mode communication modes cannot accurately determine the optimal path when the network dynamically changes complexly, thus affecting the stability of communication.

[0004] Therefore, how to accurately determine the optimal path during the complex network dynamic change process has become a technical problem that needs to be urgently solved by those skilled in the art. Summary of the Invention

[0005] The present invention provides a method, system and device for determining a reliability path of a power dual-mode remote communication network to solve the technical problem that the path of the current power dual-mode remote communication network cannot be accurately determined, so as to meet the actual requirements of the power dual-mode remote communication network in a complex environment.

[0006] To solve the above technical problem, an embodiment of the present invention provides a method for determining a reliability path of a power dual-mode remote communication network, including:

[0007] Determine a simulation topology model of a target power dual-mode remote communication network;

[0008] Construct a global operating state matrix of the target power dual-mode remote communication network according to communication nodes and channels of the target power dual-mode remote communication network;

[0009] Encode the simulation topology model and the global operating state matrix in sequence to obtain a first feature vector of the communication nodes and a second feature vector of the channels;

[0010] Input the first feature vector and the second feature vector into a pre-constructed graph neural network processing model to obtain node features corresponding to the first feature vector and edge features corresponding to the second feature vector;

[0011] Calculate the path feature vector from the node features and the edge features, and input the path feature vector into the path performance calculation formula constructed by the path prediction packet loss rate, the path prediction bandwidth utilization rate, and the path delay to obtain the real-time path performance result of the predicted path;

[0012] Based on the real-time path performance result, screen all alternative paths in the target power dual-mode remote communication network to determine the optimal path.

[0013] As one of the preferred solutions, after determining the simulation topology model of the target power dual-mode remote communication network, the method for determining the reliable path of the power dual-mode remote communication network further includes:

[0014] Calculate the correlation coefficient for the edges between any two communication nodes in the simulation topology model to obtain the correlation coefficient;

[0015] Cluster all communication nodes according to the spatial clustering algorithm to obtain each clustering center, where the center point of each clustering center is determined by the correlation coefficient;

[0016] Update the simulation topology model according to the calculated compactness of the clustering center.

[0017] As one of the preferred solutions, the process of updating the simulation topology model includes:

[0018] If the compactness of the clustering center is lower than the preset threshold, perform the splitting, merging, and / or node migration operations of the clustering center.

[0019] As one of the preferred solutions, constructing the global operation state matrix of the target power dual-mode remote communication network according to the communication nodes and channels of the target power dual-mode remote communication network includes:

[0020] Collect the operation state parameters of each communication node in the target power dual-mode remote communication network, and use the

[0021] operation state parameters of the communication node to construct a node state vector;

[0022] Collect the operation state parameters of each channel in the target power dual-mode remote communication network, and use the operation state parameters of the channel to construct a channel state matrix;

[0023] Perform normalization processing on the node state vector and the channel state matrix, and use the normalized node state vector and channel state matrix to construct the global operation state matrix of the target power dual-mode remote communication network.

[0024] As one of the preferred solutions, inputting the first feature vector and the second feature vector into a pre-constructed graph neural network processing model includes:

[0025] During the processing of the graph neural network processing model, aiming at minimizing the path performance prediction error, optimize and train the constructed path performance prediction model by constructing a loss function, and the loss function is expressed as:

[0026]

[0027] where L represents the loss function of the path performance prediction model, represents the true performance value of path k, represents the predicted performance value of path k, N represents the total number of training paths, and η 1 is the weight of the compactness optimization term of the clustering center, and 1 - Γ(C m ) represents the penalty term for low-compactness clustering centers.

[0028] As one of the preferred solutions, screening all alternative paths in the target power dual-mode remote communication network based on the real-time path performance results, and the screening process includes:

[0029] Rank all alternative paths in the target power dual-mode remote communication network according to the real-time path performance results to obtain path priorities, and determine the optimal path according to the path priorities.

[0030] Another embodiment of the present invention provides a reliability path determination system for a power dual-mode remote communication network, including:

[0031] A determination module, used to determine the simulation topology model of the target power dual-mode remote communication network;

[0032] A construction module, used to construct the global operating state matrix of the target power dual-mode remote communication network according to the communication nodes and channels of the target power dual-mode remote communication network;

[0033] An encoding processing module, used to sequentially perform encoding processing on the simulation topology model and the global operating state matrix to obtain the first feature vector of the communication node and the second feature vector of the channel;

[0034] A processing module, used to input the first feature vector and the second feature vector into a pre-constructed graph neural network processing model to obtain the node features corresponding to the first feature vector and the edge features corresponding to the second feature vector;

[0035] A calculation module, configured to calculate the node features and the edge features to obtain a path feature vector, and input the path feature vector into a path performance calculation formula constructed by a path prediction packet loss rate, a path prediction bandwidth utilization rate, and a path delay, so as to obtain a real-time path performance result of the predicted path;

[0036] A screening module, configured to screen all alternative paths in the target power dual-mode remote communication network based on the real-time path performance result, and determine an optimal path.

[0037] As one preferred solution, after determining a simulation topology model of the target power dual-mode remote communication network, the reliability path determination system of the power dual-mode remote communication network further includes:

[0038] An association calculation module, configured to calculate an association degree coefficient for an edge between any two communication nodes in the simulation topology model to obtain an association degree coefficient;

[0039] A clustering module, configured to cluster all communication nodes according to a spatial clustering algorithm to obtain respective clustering centers, wherein the center point of each clustering center is determined by the association degree coefficient;

[0040] An update module, configured to update the simulation topology model according to the calculated compactness of the clustering centers.

[0041] Another embodiment of the present invention provides a reliability path determination device for a power dual-mode remote communication network, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the reliability path determination method for the power dual-mode remote communication network as described above is implemented.

[0042] Another embodiment of the present invention provides a computer-readable storage medium storing a computer program. When a device where the computer-readable storage medium is located executes the computer program, the reliability path determination method for the power dual-mode remote communication network as described above is implemented.

[0043] Compared with the prior art, the beneficial effects of the embodiments of the present invention are at least one of the following:

[0044] After determining the simulation topology model of the target power dual-mode remote communication network and constructing the global operating state matrix, the real-time states of communication nodes and channels within the network can be accurately captured. By encoding and processing the simulation topology model and the global operating state matrix, the eigenvectors of communication nodes and channels are obtained. This step simplifies the processing difficulty of complex network information and retains key information for the processing of graph neural networks. Using the structural characteristics of the graph neural network processing model and the feature information of nodes and edges, node features and edge features are obtained. Calculating the node features and edge features yields the path feature vector, which provides strong support for the accurate construction of path features. The generation of the path feature vector synthesizes the information of nodes and edges. Inputting the path feature vector into the path performance calculation formula can obtain the real-time performance results of the predicted path, thereby accurately evaluating the actual performance of the path. Based on these real-time performance results, the optimal path is determined. This process not only ensures the stability and reliability of the power dual-mode remote communication network but also improves the efficiency and quality of data transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a schematic flowchart of a method for determining a reliable path in a power dual-mode remote communication network according to one embodiment of the present invention;

[0046] Figure 2 is a schematic structural diagram of a system for determining a reliable path in a power dual-mode remote communication network according to one embodiment of the present invention;

[0047] Figure 3 is a schematic structural diagram of a device for determining a reliable path in a power dual-mode remote communication network according to one embodiment of the present invention.

[0048] REFERENCE SIGNS:

[0049] Among them, 11, determination module; 12, construction module; 13, encoding and processing module; 14, processing module; 15, calculation module; 16, screening module; 21, processor; 22, memory. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0051] In the description of this application, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second", "third", etc. may explicitly or implicitly include one or more of such features. In the description of this application, unless otherwise stated, the meaning of "a plurality" is two or more.

[0052] In the description of this application, it should be noted that unless otherwise clearly defined and limited, the terms "installed", "connected", "connected to" shall be construed in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the communication inside two components. The terms "vertical", "horizontal", "left", "right", "up", "down" and similar expressions used herein are only for the purpose of illustration and do not indicate or imply that the indicated device or component must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be construed as a limitation on the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0053] In the description of this application, it should be noted that unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by those skilled in the technical field to which this technology belongs. The terms used in the description of this invention in the specification are only for the purpose of describing specific embodiments and are not intended to limit this invention. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0054] With the rapid development of modern power systems and the wide application of smart grids, power remote communication technology has played a crucial role in ensuring the stable operation of power systems and improving energy dispatching efficiency. In a complex communication environment, power dual-mode remote communication networks face multiple challenges such as signal interference, equipment failures, and network congestion, posing higher requirements for their communication reliability.

[0055] Currently, most power dual-mode remote communication networks use single-mode or simple dual-mode communication modes for data transmission. In the single-mode mode, due to the lack of alternative paths and redundancy mechanisms, the network is prone to data loss or communication interruption when signal interference or node failures occur. In the simple dual-mode communication mode, mode switching and path selection are usually carried out manually or by fixed strategies, making it difficult to cope with complex network dynamic changes, resulting in low resource utilization efficiency and large communication delays, and it is difficult to meet the requirements of power systems for high real-time performance and high reliability.

[0056] In recent years, with the application of intelligent algorithms and big data technologies, some power dual-mode remote communication networks have begun to introduce intelligent path selection and resource allocation algorithms. For example, by predicting path performance through a statistical-based model or prioritizing paths through simple rules to improve communication reliability and efficiency. However, the following main problems have emerged in the practical application of the existing technologies: First, the existing methods usually cannot comprehensively consider the node status, channel characteristics, and network dynamic changes in the power dual-mode remote communication network, resulting in insufficient accuracy in predicting path performance. Second, there is a lack of comprehensive analysis of complex network topologies and multi-path performance during the path selection and resource allocation process, making it difficult to fully utilize network resources.

[0057] In summary, how to accurately determine the optimal path during the complex network dynamic change process has become a technical problem that needs to be urgently solved by those skilled in the art.

[0058] To address this problem, an embodiment of the present invention provides a method for determining a reliable path in a power dual-mode remote communication network. Specifically, please refer to Figure 1 , Figure 1 which shows a schematic flowchart of the method for determining a reliable path in a power dual-mode remote communication network in one embodiment of the present invention. The method includes:

[0059] S1: Determine the simulation topology model of the target power dual-mode remote communication network;

[0060] S2: Construct the global operating state matrix of the target power dual-mode remote communication network according to the communication nodes and channels of the target power dual-mode remote communication network;

[0061] S3: Perform encoding processing on the simulation topology model and the global operating state matrix in sequence to obtain the first feature vector of the communication nodes and the second feature vector of the channels;

[0062] S4: Input the first feature vector and the second feature vector into a pre-constructed graph neural network processing model to obtain the node features corresponding to the first feature vector and the edge features corresponding to the second feature vector;

[0063] S5: Calculate the path feature vector from the node features and the edge features, and input the path feature vector into a path performance calculation formula constructed by path prediction packet loss rate, path prediction bandwidth utilization rate, and path delay to obtain the real-time path performance result of the predicted path;

[0064] S6: Screen all alternative paths in the target power dual-mode remote communication network based on the real-time path performance result to determine the optimal path.

[0065] In step S1, a simulation topology model of the target power dual-mode remote communication network can be established based on the physical structure data and communication data of the power system. After determining the simulation topology model of the target power dual-mode remote communication network, the correlation coefficient between any two communication nodes in the simulation topology model is calculated to obtain the correlation coefficient. All communication nodes are clustered according to the spatial clustering algorithm to obtain each clustering center, where the center point of each clustering center is determined by the correlation coefficient. According to the calculated compactness of the clustering center, the simulation topology model is updated.

[0066] Among them, the correlation coefficient A ij is used to characterize the comprehensive correlation relationship between two communication nodes based on network state parameters such as bandwidth, delay, and packet loss rate, and can be expressed as:

[0067]

[0068] Among them, B ij represents the bandwidth between communication node v i and communication node v j , D ij represents the delay between communication node v i and communication node v j , L ij represents the packet loss rate between communication node v i and communication node v j , the value range is [0, 1), α and β are influence weight coefficients for adjusting the bandwidth and packet loss rate, and θ is an adjustment coefficient for controlling the steepness of the correlation coefficient curve. When the value of the adjustment coefficient θ increases, the correlation coefficient is more sensitive to changes in bandwidth, delay, and packet loss rate.

[0069] For each clustering center C m calculate the compactness index Γ(C m ), which is used to measure the average correlation degree between internal communication nodes of this clustering center:

[0070]

[0071] Among them, |C m | represents the number of communication nodes in clustering center C m , represents the cumulative value of the correlation coefficients between all pairs of internal nodes of the clustering center.

[0072] If the compactness index Γ(C m) is lower than the first preset threshold, the cluster center splitting, merging and / or node migration operations are performed to optimize the internal communication node correlation of the cluster center, wherein splitting is to split the large cluster center into smaller cluster centers to improve the internal correlation; merging is to merge the cluster centers with low correlation into the cluster centers with higher correlation; node migration refers to migrating certain nodes from the original cluster center to other cluster centers with higher correlation.

[0073] The main process is to analyze the density of cluster centers, identify clusters with too low or too high density, consider increasing the connection between nodes or optimizing the communication path for clusters with too low density, and consider dispersing nodes or adjusting resource allocation to reduce communication pressure for clusters with too high density. The simulation topology model is adjusted according to the update strategy, and the correlation coefficient and clustering are recalculated to verify the update effect. The whole process is an iterative optimization process. By continuously calculating the correlation coefficient, performing spatial clustering and updating the simulation topology model, the network structure can be gradually optimized to improve communication efficiency and resource utilization.

[0074] Specifically, in the cluster center splitting operation, the cluster center C m Subdivide and divide the cluster center C m Split into sub-cluster centers C m1 and C m2 , calculate the ratio of the mean internal correlation of the sub-cluster centers to the mean external correlation between the sub-cluster centers, denoted as the difference coefficient Δ:

[0075]

[0076] in, and Represent the sub-cluster centers C m1 and C m2 The sum of the correlations between internal nodes, Represents the sum of the correlations between sub-cluster centers.

[0077] When the coefficient of difference Δ(C m1 ,C m2 ) is greater than 1, indicating that the internal correlation of the sub-cluster centers is significantly higher than the correlation between the sub-cluster centers, the split operation is valid and the split result is retained, otherwise it falls back to the cluster center division state before the split.

[0078] For the merging and node migration between different cluster centers, calculate the cluster center C p and cluster center C q The merger index M(C p ,C q ) and the single node migration index Ψ(v i ,C q ):

[0079]

[0080] Among them, represents the total correlation degree of all node pairs between the clustering center C p and the clustering center C q .

[0081] When the merging index M(C p , C q ) is greater than the merging threshold, it is considered that the merging of two clustering centers can enhance the overall communication reliability. If the correlation degree ratio Ψ(v i in its original belonging clustering center C p and the clustering center C q ) is greater than 1, then this node is migrated from C i to the clustering center C q ; p to the clustering center C q ;

[0082] For the merging between different clustering centers and node migration, calculate the merging index M(C p and the clustering center C q ) and the single node migration index Ψ(v p , C q ): i , C q ):

[0083]

[0084]

[0085] Among them, represents the total correlation degree of all node pairs between the clustering center C p and the clustering center C q .

[0086] When the merging index M(C p , C q ) is greater than the merging threshold, it is considered that the merging of two clustering centers can enhance the overall communication reliability. If the correlation degree ratio Ψ(v i in its original belonging clustering center C p and the clustering center C q ) is greater than 1, then this node is migrated from C i to the clustering center C q ; p to the clustering center C q .

[0087] After completing the operations of splitting, merging, or node migration of the clustering centers, update the edge weight set in the power dual-mode remote communication network topology model. For each edge perform re-assignment:

[0088]

[0089] where δ m represents the retention coefficient within the clustering center. When nodes are in the same clustering center, the edge weight within the clustering center is increased using the correlation coefficient; otherwise, the edge weight between clustering centers is decreased. Update the simulation topology model based on the updated edge weight set.

[0090] In this step, high-correlation nodes in the network are divided into a tight clustering center structure through the dynamic clustering center division algorithm. In a high-load scenario, the clustering center division result can effectively reduce the resource occupancy of low-priority paths in the network, thereby optimizing the bandwidth and channel allocation of high-priority paths.

[0091] In communication data, communication nodes and channels are two crucial concepts. A communication node refers to a connection point in a telecommunications network, which can be a redistribution point or a communication endpoint. These nodes can receive or forward information and are the basic elements that make up a communication network. In data communication, the specific form of a node can be diverse, including but not limited to:

[0092] Data circuit-terminating equipment: such as bridges, switches, modems, and hubs, etc. These devices play roles such as data forwarding, switching, and modulation / demodulation in data communication to ensure the correct transmission of data in the network;

[0093] Data terminal equipment: such as hosts, digital mobile phones, and printers, etc. These devices are the producers and receivers of data and exchange data with other devices through the communication network.

[0094] A channel is the path for data transmission, which provides a path for data to be transmitted in a communication network.

[0095] In step S2, construct the global operating state matrix of the target power dual-mode remote communication network according to the communication nodes and channels of the target power dual-mode remote communication network, specifically including:

[0096] Collect the operating state parameters of each communication node in the target power dual-mode remote communication network. The operating state parameters of each communication node include signal strength, node processing capacity, node load, and node online status. Use the operating state parameters of the communication nodes to construct a node state vector V i , and the node state vector is expressed as:

[0097] Vi = [P i , C i , L i , S i

[0098] Among them, P i represents the received signal strength of node v i , which is used to characterize the communication stability of the node. C i represents the computing power of node v i , which is used to measure the data processing ability of the node. L i represents the load ratio of node v i , and its value range is [0, 1], which is used to characterize the resource occupancy of the node. S i represents the online status of node v i , where S i = 1 indicates that the node is online, and S i = 0 indicates that the node is offline;

[0099] Collect the operating status parameters of each channel in the target power dual-mode remote communication network. The operating status parameters of each channel include channel bandwidth, channel delay, channel packet loss rate, and channel interference intensity. Use the operating status parameters of the channel to construct a channel status matrix E ij , and the channel status matrix is expressed as:

[0100]

[0101] Among them, B ij represents the effective bandwidth of channel e ij , which is used to measure the data transmission ability of the channel. D ij represents the communication delay of channel e ij , which is used to characterize the response speed of the channel. L ij represents the data packet loss rate of channel e ij , and its value range is [0, 1], which is used to describe the reliability of the channel. I ij represents the interference intensity of channel e ij , which is used to measure the influence of external interference on the channel;

[0102] Perform normalization processing on the node state vector and the channel state matrix. Use the normalized node state vector and the channel state matrix to construct the global operating status matrix of the target power dual-mode remote communication network. The global operating status matrix is expressed as:

[0103]

[0104] ​In step S3, the simulation topology model and the global operating state matrix are encoded in sequence to obtain the first feature vector of the communication node and the second feature vector of the channel. During the encoding process, the node feature encoding function is used to convert the communication node state vector into a high-dimensional feature vector, i.e., the first feature vector; the channel feature encoding function is used to convert the channel state vector into a high-dimensional feature vector, i.e., the second feature vector.

[0105] After obtaining the feature vectors, the tightness index of the cluster center to which the node belongs is incorporated into the node data and channel data in combination with the result of dynamic cluster center partitioning.

[0106]

[0107] Among them, represents the first feature vector of node v i , φ 1 is the node feature encoding function, κ is the weight coefficient of the cluster center tightness, represents the second feature vector of channel e ij , φ 2 is the channel feature encoding function.

[0108] Incorporating the tightness into the node features can more comprehensively reflect the overall characteristics and correlation structure of the cluster center to which the node belongs. The tightness index measures the average correlation degree of the nodes inside the cluster center. Taking it as a part of the features helps the model better understand the importance of the node in the cluster center and the impact of the overall cluster center on the communication performance. Compared with the method without combining the tightness index, incorporating the tightness into the node features can improve the sensitivity of the model to the network topology structure and the characteristics of the cluster center, so as to more accurately optimize the communication path and predict the network performance.

[0109] The power dual-mode remote communication network is trained using the graph neural network processing model to obtain the node features corresponding to the first feature vector and the edge features corresponding to the second feature vector;

[0110] The calculation method of the node features is:

[0111]

[0112] Among them, N(i)∩C m represents the set of neighbor nodes of node v i inside its belonging cluster center C m , N(i)\C m represents the set of neighbor nodes of node v i in other cluster centers, and are the learnable weight matrices of the l-th layer, b (l)is the bias term, and σ is the activation function;

[0113] The calculation method of the edge feature combines the difference coefficient between the cluster centers, which is expressed as:

[0114]

[0115] Specifically, in step S4, inputting the first feature vector and the second feature vector into a pre-constructed graph neural network processing model, the process further includes:

[0116] During the processing of the graph neural network processing model, with the goal of minimizing the path performance prediction error, optimize and train the constructed path performance prediction model by constructing a loss function, and the loss function is expressed as:

[0117]

[0118] where L represents the loss function of the path performance prediction model, represents the true performance value of path k, represents the predicted performance value of path k, N represents the total number of training paths, and η 1 is the weight of the compactness optimization term of the cluster center, and 1 - Γ(C m ) represents the penalty term for low-compactness cluster centers.

[0119] In step S5, calculate the node feature and the edge feature to obtain the path feature vector P k , and the path feature vector P k is expressed as:

[0120]

[0121] where P k represents the comprehensive performance index of path k, λ 1 is the weight coefficient of the cluster center compactness on the path performance, and ω 1 , ω 2 , ω 3 and ω 4 are the weight coefficients of bandwidth, delay, packet loss rate, and interference intensity on the path performance respectively, represents the normalized bandwidth of channel e ij in the path, represents the normalized delay of channel e ij in the path, represents the normalized packet loss rate of channel e ij in the path, represents the normalized interference intensity of channel e ij in the path.

[0122] Construct a path delay prediction model by combining path feature vectors. The prediction of path delay D k is:

[0123]

[0124] where D k represents the predicted delay of path k, represents the normalized delay of channel e ij on the path, represents the normalized interference intensity of channel e ij , and β 1 and β 2 are the weight coefficients of delay and interference on path delay respectively.

[0125] Construct a path bandwidth utilization prediction model by combining path feature vectors. The prediction of path bandwidth utilization B k is:

[0126]

[0127] where B k represents the predicted bandwidth utilization of path k, represents the normalized bandwidth of channel e ij on the path, represents the normalized packet loss rate of channel e ij , and γ 1 and γ 2 are the weight coefficients of bandwidth and packet loss rate on path bandwidth utilization respectively.

[0128] Construct a path packet loss rate prediction model by combining path feature vectors. The prediction of path packet loss rate L k is:

[0129]

[0130] where L k represents the predicted packet loss rate of path k, |P k | represents the number of edges in path k, and δ 1 and δ 2 are the weight coefficients of packet loss rate and interference intensity on path packet loss rate respectively.

[0131] Update the path performance result by using the path feature vector as the input of the path performance index, and obtain the real-time path performance result P of the predicted path through weighted combination k,new :

[0132] P k,new = ω 4 σB k - ω 5 ·D k - ω6 ·L k

[0133] Among them, ω 4 , ω 5 and ω 6 are the weights of bandwidth utilization, latency, and packet loss rate on path performance respectively.

[0134] Perform priority sorting on all alternative paths in the target power dual-mode remote communication network according to the real-time path performance results to obtain path priorities, and determine the optimal path according to the path priorities.

[0135] Specifically, perform priority sorting on all alternative paths in the target power dual-mode remote communication network according to the real-time path performance results to obtain path priorities, and sort the path priorities from high to low. Paths with higher priorities have better transmission performance in the communication environment.

[0136] Select the optimal path in the current environment based on the path priority sorting results. The selection rule for the optimal path is that the path priority is the highest and satisfies the capacity constraint C k ≥C min , C k represents the capacity of path k, which is equal to the minimum value of the capacities of all channels on the path, and C min is the minimum capacity threshold set by the system to ensure that the path transmission performance meets the current load requirements. If the path with the highest priority cannot meet the capacity constraint, select the sub-optimal path until the condition is satisfied.

[0137] While selecting the optimal path, determine the set of alternative paths. The selection rule for the set of alternative paths P backup is:

[0138] P backup ={P k,new ∣R k >R threshold ∧C k ≥C backup}

[0139] Among them, R threshold is the threshold of path priority, and C backup is the lower capacity limit of the alternative path, ensuring that the alternative path has sufficient transmission capacity when the main path fails.

[0140] Dynamically monitor the optimal path and alternative paths, and update the path performance metrics P k,new and path priorities in real time. When it is detected that the performance of the current optimal path deteriorates or fails, re-select from the alternative paths according to the priority

[0141] ^

[0142] Select an alternative path P from the set best 。

[0143] An efficient alternative path switching mechanism is designed through a path priority sorting algorithm and a dynamic adjustment strategy for the alternative path set. This mechanism uses path priority thresholds and capacity constraints to screen the alternative path set, and ensures that when the primary path fails, it can quickly switch to the alternative path with the optimal performance by dynamically monitoring the performance metrics of the optimal path, thus avoiding data transmission interruption.

[0144] In the above steps, the present invention constructs a path performance calculation formula based on a graph neural network combined with the results of dynamic clustering center division, deeply fuses multi-dimensional information such as node features, channel characteristics, and clustering center tightness, and can comprehensively analyze the path performance through a comprehensive prediction formula for delay, bandwidth utilization, and packet loss rate, so as to achieve the precision of path priority sorting.

[0145] Another embodiment of the present invention provides a reliability path determination system for a power dual-mode remote communication network. Specifically, please refer to Figure 2 , Figure 2 which shows a schematic structural diagram of a reliability path determination system for a power dual-mode remote communication network in one embodiment of the present invention, including:

[0146] A determination module 11, configured to determine a simulation topology model of a target power dual-mode remote communication network;

[0147] A construction module 12, configured to construct a global operating state matrix of the target power dual-mode remote communication network according to communication nodes and channels of the target power dual-mode remote communication network;

[0148] An encoding processing module 13, configured to sequentially perform encoding processing on the simulation topology model and the global operating state matrix to obtain a first feature vector of the communication nodes and a second feature vector of the channels;

[0149] A processing module 14, configured to input the first feature vector and the second feature vector into a pre-constructed graph neural network processing model to obtain node features corresponding to the first feature vector and edge features corresponding to the second feature vector;

[0150] A calculation module 15, configured to calculate path feature vectors from the node features and the edge features, and input the path feature vectors into a path performance calculation formula constructed by path predicted packet loss rate, path predicted bandwidth utilization, and path delay to obtain real-time path performance results of the predicted paths;

[0151] A screening module 16, configured to screen all alternative paths in the target power dual-mode remote communication network based on the real-time path performance results, and determine the optimal path.

[0152] The present invention provides an embodiment:

[0153] In a power dispatching center of a certain province, the master station is responsible for monitoring and dispatching the operating status of multiple distributed energy stations within the province. At 14:30, the master station monitoring system detected an abnormal signal in the communication link between a certain sub-station (numbered Node-23) and the master station. The channel status showed that the bandwidth of the main wired link between Node-23 and the master station dropped suddenly from the original 50 Mbps to 5 Mbps, the latency increased sharply from 20 ms to 200 ms, and the packet loss rate was as high as 15%. System analysis showed that this link was strongly interfered with, possibly due to a regional equipment failure.

[0154] In the traditional communication mode, the master station relies on a fixed path switching strategy. After the wired link fails, it tries to switch to the wireless mode. However, at this time, the signal strength of the wireless channel is also affected by the overload of neighboring base stations, and only a bandwidth of less than 10 Mbps can be maintained. It is difficult for the system to select a suitable path in a short time, resulting in the interruption of real-time data transmission of Node-23 and affecting the load dispatching decision of the master station.

[0155] After applying the method of the present invention, the dispatching center system automatically starts the system. The system first collects all communication link status parameters between Node-23 and the master station through the global operating status matrix in real time, and updates the node and channel characteristics. The operating status of Node-23 shows that the signal strength has dropped to -90 dBm, the node load has increased to 85%, and the current processing capacity is close to saturation. The channel status indicates that the interference intensity of the main wired link reaches -70 dBm, and the interference intensity of the wireless link is -85 dBm, which cannot meet the large data volume transmission requirements.

[0156] Through the dynamic clustering center division algorithm, the system divides Node-23 and its surrounding nodes into a high-density clustering center structure, and combines the graph neural network model to re-evaluate the current path performance. At 14:31, the system generates a priority ranking of 5 available paths. The path with the highest priority is the multi-hop link of the master station - Node-17 - Node-23, and the predicted path performance indicators are a bandwidth of 40 Mbps, a latency of 50 ms, and a packet loss rate of 2.5%.

[0157] At 14:32, the system determines that the main station - Node - 17 - Node - 23 is the optimal path according to the backup path selection rule. At the same time, the main station - Node - 18 - Node - 23 and the main station - Node - 19 - Node - 23 are selected as backup paths, and the path performance monitoring is updated in real time. The main station immediately switches the data traffic from the failed wired link to the newly selected optimal path, and the data transmission at Node - 23 resumes normal.

[0158] At 14:34, the main station detects again that the channel interference intensity of the current optimal path rises from -75 dBm to -60 dBm, resulting in a decline in path performance. The system quickly switches the data stream to the backup path main station - Node - 18 - Node - 23, and at the same time adjusts the load distribution strategy to reduce the pressure on the main path.

[0159] During the whole process, the switching time of the method of the present invention at the key node is only 35 ms, which is significantly better than 120 ms of the traditional method.

[0160] In summary, this embodiment demonstrates the superior performance of the method of the present invention in terms of communication reliability and resource optimization by simulating communication link failures in a complex environment, successfully solves the problems of lag in path selection and high switching delay in the traditional method, and provides technical support for the stable operation of the power dual - mode remote communication network.

[0161] Another embodiment of the present invention provides a reliability path determination device for a power dual - mode remote communication network, including a processor 21, a memory 22, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the reliability path determination method for the power dual - mode remote communication network as described above.

[0162] See Figure 3 , which is the structural block diagram of the reliability path determination device for the power dual - mode remote communication network provided by the embodiment of the present invention. The reliability path determination device for the power dual - mode remote communication network provided by the embodiment of the present invention includes a processor 21, a memory 22, and a computer program stored in the memory 22 and configured to be executed by the processor 21. When the processor 21 executes the computer program, it implements the steps in the reliability path determination method embodiment of the power dual - mode remote communication network as described above, such as Figure 1 the steps S1 - S6 described in ; or, when the processor 21 executes the computer program, it implements the functions of each module in the above - mentioned device embodiments, such as the determination module 11.

[0163] Accordingly, an embodiment of the present invention provides a computer-readable storage medium, which includes a stored computer program. When the computer program runs, it controls the device where the computer-readable storage medium is located to execute the steps in the method for determining the reliability path of the power dual-mode remote communication network in the above-mentioned embodiment, such as Figure 1 the steps S1 to S6 described therein.

[0164] Compared with the prior art, the beneficial effects of the embodiments of the present invention are at least one of the following:

[0165] After determining the simulation topology model of the target power dual-mode remote communication network and constructing the global operation state matrix, it is possible to accurately capture the real-time states of communication nodes and channels in the network. By encoding and processing the simulation topology model and the global operation state matrix, the feature vectors of communication nodes and channels are obtained. This step simplifies the processing difficulty of complex network information and also retains key information for the processing of graph neural networks; using the structural characteristics of the graph neural network processing model and the feature information of nodes and edges, node features and edge features are obtained, and path feature vectors are obtained by calculating the node features and edge features, which provides strong support for the accurate construction of path features. The generation of path feature vectors synthesizes the information of nodes and edges; inputting the path feature vectors into the path performance calculation formula, the real-time performance results of the predicted path can be obtained, so as to accurately evaluate the actual performance of the path; based on these real-time performance results, the optimal path is determined. This process not only ensures the stability and reliability of the power dual-mode remote communication network, but also improves the efficiency and quality of data transmission.

[0166] In summary, this process gives full play to the advantages of the technology from global state capture, feature vector extraction, to deep feature mining, path feature construction, and then to performance prediction and path screening, providing a solid guarantee for the stable operation and performance improvement of the power dual-mode remote communication network.

[0167] The above-described embodiments only represent several implementation manners of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention should be subject to the appended claims.

Claims

1. A reliability path determination method for a power dual-mode remote communication network, characterized in that: include: Determine the simulation topology model of the target power dual-mode telecommunication network; Constructing a global operation state matrix of the target electric power dual-mode remote communication network according to the communication nodes and channels of the target electric power dual-mode remote communication network; Encoding the simulation topology model and the global operation state matrix in sequence to obtain a first eigenvector of the communication node and a second eigenvector of the channel; Inputting the first feature vector and the second feature vector into a pre-built graph neural network processing model to obtain node features corresponding to the first feature vector and edge features corresponding to the second feature vector; Calculating the node features and the edge features to obtain a path feature vector, inputting the path feature vector into a path performance calculation formula constructed by a path prediction packet loss rate, a path prediction bandwidth utilization rate, and a path delay, to obtain a real-time path performance result of the predicted path; All candidate paths in the target electric power dual-mode telecommunication network are screened based on the real-time path performance results to determine an optimal path.

2. The reliability path determination method of the electric power dual-mode remote communication network according to claim 1, characterized in that: After determining the simulation topology model of the target electric power dual-mode remote communication network, the reliability path determination method of the electric power dual-mode remote communication network further includes: Calculating the correlation coefficient of the edges between any two communication nodes in the simulation topology model to obtain the correlation coefficient; Clustering all communication nodes according to a spatial clustering algorithm to obtain the compactness of each cluster center, wherein the center point of each cluster center is determined by the correlation coefficient; The simulation topology model is updated according to the calculated compactness of the cluster centers.

3. The reliability path determination method of the electric power dual-mode remote communication network according to claim 2, characterized in that: The process of updating the simulation topology model includes: If the compactness of the cluster centers is lower than a preset threshold, a partitioning method of splitting, merging and / or migrating nodes of the cluster centers is executed.

4. The reliability path determination method of the electric power dual-mode remote communication network according to claim 1, characterized in that: The step of constructing a global operation state matrix of the target electric power dual-mode remote communication network according to the communication nodes and channels of the target electric power dual-mode remote communication network includes: Collecting the operating state parameters of each communication node in the target electric power dual-mode remote communication network, and constructing a node state vector using the operating state parameters of the communication node; Collecting the operating state parameters of each channel in the target electric power dual-mode remote communication network, and constructing a channel state matrix using the operating state parameters of the channels; The node state vector and the channel state matrix are normalized, and the global operation state matrix of the target electric power dual-mode remote communication network is constructed using the normalized node state vector and the channel state matrix.

5. The reliability path determination method of the electric power dual-mode remote communication network according to claim 2, characterized in that: The step of inputting the first feature vector and the second feature vector into a pre-built graph neural network processing model includes: In the process of processing the graph neural network processing model, the goal is to minimize the path performance prediction error, and the constructed path performance prediction model is optimized and trained by constructing a loss function. The loss function is expressed as: Wherein, L represents the loss function of the path performance prediction model, represents the actual performance value of path k, represents the prediction performance value of path k, N represents the total number of training paths, η1 is the weight of the compactness optimization term of the cluster center, 1-Γ(C m ) represents the penalty term for low-density cluster centers.

6. The reliability path determination method of the electric power dual-mode remote communication network according to claim 1, characterized in that: The screening of all candidate paths in the target electric power dual-mode telecommunication network based on the real-time path performance result includes: All candidate paths in the target electric power dual-mode telecommunication network are prioritized according to the real-time path performance results to obtain path priorities, and an optimal path is determined according to the path priorities.

7. A reliability path determination system for a power dual-mode telecommunication network, characterized in that: include: A determination module, used to determine a simulation topology model of a target electric power dual-mode telecommunication network; A construction module, used to construct a global operation state matrix of the target electric power dual-mode remote communication network according to the communication nodes and channels of the target electric power dual-mode remote communication network; A coding processing module, used for sequentially coding the simulation topology model and the global operation state matrix to obtain a first eigenvector of the communication node and a second eigenvector of the channel; A processing module, used to input the first feature vector and the second feature vector into a pre-built graph neural network processing model to obtain node features corresponding to the first feature vector and edge features corresponding to the second feature vector; A calculation module, used to calculate the node features and the edge features to obtain a path feature vector, and input the path feature vector into a path performance calculation formula constructed by a path prediction packet loss rate, a path prediction bandwidth utilization rate, and a path delay to obtain a real-time path performance result of the predicted path; The screening module is used to screen all candidate paths in the target electric power dual-mode telecommunication network based on the real-time path performance result to determine the optimal path.

8. The reliability path determination system of the electric power dual-mode telecommunication network according to claim 7, characterized in that: After determining the simulation topology model of the target electric power dual-mode remote communication network, the reliability path determination system of the electric power dual-mode remote communication network further includes: An association calculation module is used to calculate the association coefficient of the edge between any two communication nodes in the simulation topology model to obtain the association coefficient; A clustering module, used for clustering all communication nodes according to a spatial clustering algorithm to obtain the compactness of each cluster center, wherein the center point of each cluster center is determined by the correlation coefficient; An updating module is used to update the simulation topology model according to the calculated compactness of the cluster centers.

9. A reliability path determination device for a power dual-mode remote communication network, characterized in that: It comprises a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, it implements the reliability path determination method of the power dual-mode telecommunication network as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein when the device where the computer-readable storage medium is located executes the computer program, the reliability path determination method of the power dual-mode remote communication network as described in any one of claims 1 to 6 is implemented.

Images